The
NEW ENGLA ND JOURNAL
of
MEDICINE
Perspective november 10, 2011
Making Good on ACOs’ Promise — The Final Rule for the Medicare Shared Savings Program Donald M. Berwick, M.D.
D
uring my career as a practicing pediatrician, my patients and I benefited from being part of a well-managed system of care, coordinated and financed to support seamlessness and patient-
centeredness. We had an electronic health record — always available and up to the minute — which reminded me when a test or follow-up was due. For children with severe asthma, I worked as part of a team, with a home health nurse to teach skills and anticipate needs, an allergist instantly available as a coach, pharmacists to help plan care and detect errors, and advanced practice nurses to ensure 24/7 access. As a result, my patients stayed out of emergency rooms and hospital beds, remaining at home and in school, where they belonged. Function improved and costs fell. The dedicated professionals in
the U.S. health care system work to deliver the highest-quality health care they can. But as any health care provider will tell you, our system is full of roadblocks, red tape, and frustrations that keep them from practicing the type of medicine that most clinicians envisioned when they chose their noble field. Physicians, nurses, and other health care professionals want the support required to work with engaged patients to make the clinical decisions most appropriate to their circumstances; to collaborate with colleagues to provide a safe, seamless experience; and to be paid for keeping people well. Instead, the status quo — with
inadequate dissemination of usable clinical information, misaligned financial incentives, and in many cases, inertia — is rife with barriers to the coordinated care that patients want, providers want to give, and our unsustainable system so desperately needs. To be sure, exactly this type of medicine is practiced every day in hundreds of places throughout the country. Innovative entrepreneurs and dedicated clinicians have found ways to break down barriers and redesign care to better help their patients and communities. But bringing the best of our system to every community in the country is the health care challenge of our time. Eighteen months after President Barack Obama signed the Affordable Care Act, the Department of Health and Human Services (DHHS) has created a broad array of pathways for health care
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1753
PERSPE C T I V E
Final Rule for Medicare Shared Savings Program
Proposed Rule vs. Final Rule for Accountable Care Organizations (ACOs) in the Medicare Shared Savings Program. Topic
Proposed Rule
Modifications in Final Rule
Transition to risk in Track 1
ACOs could choose from two tracks, each entailing a 3-year agreement. Track 1 would comprise 2 years of one-sided shared savings with a mandatory transition in year 3 to performancebased risk under a two-sided model of shared savings and losses. Track 2 would comprise 3 years all under the two-sided model.
Remove two-sided risk from Track 1. Two tracks would still be offered for ACOs at different levels of readiness, with one providing higher sharing rates for ACOs willing to also share in losses.
Prospective vs. retrospective
Retrospective assignment based on utilization of primary care services, with prospective identification of a benchmark population.
A preliminary prospective-assignment method with beneficiaries identified quarterly; final reconciliation after each performance year, made on the basis of patients served by the ACO.
Proposed measures to assess quality
65 measures in 5 domains, including patient experience of care, utilization claims–based measures, and measures assessing process and outcomes. Pay for full and accurate reporting first year, pay for performance in subsequent years. Alignment of proposed measures with existing quality programs and private-sector initiatives.
33 measures in 4 domains. (Note: Claims-based measures not finalized to be used for ACO-monitoring purposes.) Longer phase-in of measures over course of agreement: first year, pay for reporting; second and third years, pay for reporting and performance. Finalize as proposed.
Sharing savings
One-sided risk model: sharing beginning at savShare on first dollar for all ACOs in both models once minimum ings of 2%, with some exceptions for small, savings rate has been achieved. physician-only, and rural ACOs. Two-sided risk model: sharing from first dollar.
Sharing beneficiary identification claims data
Claims data shared only for patients seen by ACO The ACO may contact beneficiaries from provided quarterly lists primary care physician during performance to notify them of data sharing and opportunity to decline. year; beneficiaries given opportunity to decline at the point of care.
Eligible entities
The four groups specified by the Affordable Care In addition to groups included in the proposed rule, Federally Act, as well as critical access hospitals paid Qualified Health Centers and Rural Health Clinics are also through Method II, are eligible to form an ACO. eligible to both form and participate in an ACO. In order for ACOs can be established with broad collaborabeneficiaries to be assigned on the basis of utilization of prition beyond these providers. mary care services, these organizations must provide a list of practitioners who directly render primary care services in their facilities so that beneficiaries can be assigned on the basis of utilization of their services.
Start date
Agreement for 3 years with uniform annual start Program established by January 1, 2012; first round of applicadate; performance years based on calendar years. tions are due in early 2012. First ACO agreements start April 1, 2012, and July 1, 2012. ACOs will have agreements with a first performance “year” of 18 or 21 months. ACOs starting April 1, 2012, or July 1, 2012, have option of an interim payment if they report calendar year (CY) 2012 quality measures. ACO must report quality measures for CY 2013 to qualify for first-performance-year shared savings.
Aggregate reports and preliminary prospective list
Reports will be provided at the beginning of each performance year and include: name, date of birth, sex, and health insurance claim number.
Additional reports will be provided quarterly.
Electronic health record (EHR) use
Aligning ACO requirements with EHR requirements, 50% of primary care physicians must be defined as meaningful users by start of second performance year.
No longer a condition of participation. Retained EHR as quality measure but weighted higher than any other measure for quality-scoring purposes.
Assignment process
One-step assignment process: beneficiaries assigned on the basis of a plurality of allowed charges for primary care services rendered by primary care physicians (internal medicine, general practice, family practice, and geriatric medicine).
Two-step assignment process: Step 1: for beneficiaries who have received at least one primary care service from a physician, use plurality of allowed charges for primary care services rendered by primary care physicians. Step 2: for beneficiaries who have not received any primary care services from a physician, use plurality of allowed charges for primary care services rendered by any other ACO professional.
Marketing guidelines
All marketing materials must be approved by the Centers for Medicare and Medicaid Services (CMS).
“File and use” 5 days after submission and after certifying compliance with marketing guidelines; CMS to provide approved language.
1754
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
PERSPECTIVE
Final Rule for Medicare Shared Savings Program
providers to begin — or in many cases, accelerate — their careimprovement journey in partnership with the Medicare and Medicaid programs and in synergy with the private sector. Today, the DHHS is taking its next major step by finalizing the rules for the establishment of accountable care organizations (ACOs) under the Medicare Shared Savings Program created by Section 3022 of the health care reform law. ACOs are voluntary groups of physicians, hospitals, and other health care providers that are willing to assume responsibility for the care of a clearly defined population of Medicare beneficiaries attributed to them on the basis of patients’ use of primary care services. If an ACO succeeds in both delivering high-quality care or improving care and reducing the cost of that care below what would otherwise have been expected, it will share in the savings it achieves for Medicare. Under the ACO model, Medicare beneficiaries are still free to seek care from any Medicare provider they wish. Indeed, Medicare beneficiaries should find their care experience enhanced by a program that supports providers in engaging with their patients to deliver on the three-part aim: better care for individuals, better health for populations, and lower cost growth through improvements in care. The DHHS proposed its initial set of guidelines for ACOs on March 31, 2011, and sought widespread comment on both the direction and the details of this important new program for Medicare. We at the Centers for Medicare and Medicaid Services (CMS) received more than 1200 formal comments from throughout the health care community, supple-
mented by feedback at dozens of informal listening sessions. The vast majority of the comments we received were supportive of the vision of the Shared Savings Program and optimistic about the potential for ACOs to be a force for change in our broken health care system. However, numerous suggestions were also offered for improvements to the proposed rule that would lead to a larger, more pluralistic set of ACO participants without compromising patient outcomes or choice. In partic ular, commenters asked CMS to reduce barriers to entry by streamlining governance and reporting burdens on potential ACOs; improve the potential financial return for ACOs willing to make the necessary, and often substantial, investments to improve care; and ensure beneficiary protections. In response, CMS is making several significant changes in its final rule to strengthen the ACO program for providers and beneficiaries alike (see table). Major changes include providing better, and more timely, information to ACOs at the outset of the performance year through preliminary prospective alignment of beneficiaries (while retaining a retrospective reconciliation to ensure that ACOs are measured on the basis of the patients they actually care for during the year); retaining a strong monitoring and qualitymeasurement mechanism while streamlining the metrics to focus on what matters most, including reducing the total number of quality measures by about half; allowing start-up ACOs to choose a “savings only” track without financial risk during their initial contract period; sharing savings with successful ACOs on a “first dollar” basis when the ACO achieves meaningful savings for
the Medicare program and improves care or provides highquality care; and creating a pathway for full participation of federally qualified health centers and rural health clinics that provide a primary care safety net for Medicare beneficiaries in underserved areas. Taken together, these changes and numerous others create a more feasible and attractive on-ramp for a diverse set of providers and organizations to participate as ACOs. In addition, the Center for Medicare and Medicaid Innovation is announcing today an advanced payment initiative that will allow small physician practices and rural community hospitals that face particular challenges in forming ACOs to receive up-front access to needed capital. For established organizations with a track record of providing robust coordinated care, the CMS innovation center is offering a pioneer ACO program designed to encourage and support the next wave of innovation from vanguard organizations that are positioned to help realize the full potential of the ACO model. And for organizations and clinicians not yet prepared to make the transition to ACOs, the DHHS is offering a menu of alternative options — including a comprehensive primary care program, bundled payments for care improvement, and a community-based transitional care program — that all seek to provide the incentives and supports necessary to move the mainstream of U.S. health care toward accountable care. Whether provided through ACOs or an alternative innovation opportunity, coordinated care is meant to allow providers to break away from the tyranny of the 15-minute visit, instill a re-
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1755
PERSPE C T I V E
Final Rule for Medicare Shared Savings Program
newed sense of collegiality, and return to the type of medicine that patients and families want. For patients, coordinated care means more “quality time” with their physician and care team (a patient’s advocate in an increasingly complex medical system) and more collaboration in leading a healthy life. And for Medicare, coordinated care represents the
most promising path toward financial sustainability and away from alternatives that shift costs onto patients, providers, and private purchasers. We believe that today’s ACO rule is the next step in our shared commitment to a better, more lasting health care system. We look forward to being a trusted partner in our nation’s journey
toward patient-centered, coordinated care. Disclosure forms provided by the author are available with the full text of this article at NEJM.org. Dr. Berwick is the administrator of the Centers for Medicare and Medicaid Services, Baltimore. This article (10.1056/NEJMp1111671) was published on October 20, 2011, at NEJM.org. Copyright © 2011 Massachusetts Medical Society.
Getting Moving on Patient Safety — Harnessing Electronic Data for Safer Care Ashish K. Jha, M.D., M.P.H., and David C. Classen, M.D.
M
ore than a decade ago, the Institute of Medicine released its famous report To Err Is Human, which set an ambitious agenda for the United States to reduce the number of Americans who were hurt or killed by medical errors and adverse events. In response, a series of new initiatives was launched, including the funding of new research on ways of making care safer and encouragement of programs shielding health care providers from liability if they reported adverse events. Federal agencies set up patientsafety organizations and established ambitious patient-safety goals; accrediting organizations set aggressive patient-safety standards; and providers hired patient-safety officers and implemented numerous patient-safety initiatives. So what are the fruits of these efforts? Recently, we have received some deeply disappointing news: three studies have called into question whether we’ve made any progress at all. Landrigan et al. found that rates of injury due to medical error had remained es-
1756
sentially unchanged between 2000 and 2008 at 10 North Carolina hospitals.1 A report from the Inspector General of the Department of Health and Human Services (DHHS) revealed that Medicare patients experienced substantial harm in U.S. hospitals as recently as 2008.2 Finally, Classen and colleagues found that almost one in three patients are harmed during their hospital stay and that traditional approaches to measuring adverse events, whether using voluntary reporting or patient-safety indicators, substantially underestimate the events’ frequency.3 If the United States has made progress in patient safety, it has been inadequate. The primary reason for insufficient progress is the lack of a robust measurement program: there are still no nationally agreed-on methods for systematically identifying, tracking, and reporting adverse events. Here, the patientsafety movement can learn from the quality-improvement efforts that predate it. In the 1990s, emerging evidence suggested that providers were inconsistent in
their adherence to evidence-based treatments such as the use of aspirin for patients with acute myocardial infarction. Efforts by the Joint Commission for the Accreditation of Healthcare Organizations to systematically measure performance and give feedback to hospitals, coupled with subsequent efforts to publicly report performance on these measures, led to dramatic improvements in compliance.4 In the few areas of patient safety that have seen demonstrable improvement (e.g., catheter-related bloodstream infections), the changes are due, at least in part, to robust measurement programs, such as those run by the Centers for Disease Control and Prevention. In other areas, inadequate measures have hindered progress, and patients continue to suffer from the consequences of unsafe care. Although there is a shortage of good patient-safety metrics, poorquality measures are plentiful. The best known among these are patient-safety indicators, which use billing data to identify potential complications during a hos-
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
PERSPE C T I V E
Final Rule for Medicare Shared Savings Program
newed sense of collegiality, and return to the type of medicine that patients and families want. For patients, coordinated care means more “quality time” with their physician and care team (a patient’s advocate in an increasingly complex medical system) and more collaboration in leading a healthy life. And for Medicare, coordinated care represents the
most promising path toward financial sustainability and away from alternatives that shift costs onto patients, providers, and private purchasers. We believe that today’s ACO rule is the next step in our shared commitment to a better, more lasting health care system. We look forward to being a trusted partner in our nation’s journey
toward patient-centered, coordinated care. Disclosure forms provided by the author are available with the full text of this article at NEJM.org. Dr. Berwick is the administrator of the Centers for Medicare and Medicaid Services, Baltimore. This article (10.1056/NEJMp1111671) was published on October 20, 2011, at NEJM.org. Copyright © 2011 Massachusetts Medical Society.
Getting Moving on Patient Safety — Harnessing Electronic Data for Safer Care Ashish K. Jha, M.D., M.P.H., and David C. Classen, M.D.
M
ore than a decade ago, the Institute of Medicine released its famous report To Err Is Human, which set an ambitious agenda for the United States to reduce the number of Americans who were hurt or killed by medical errors and adverse events. In response, a series of new initiatives was launched, including the funding of new research on ways of making care safer and encouragement of programs shielding health care providers from liability if they reported adverse events. Federal agencies set up patientsafety organizations and established ambitious patient-safety goals; accrediting organizations set aggressive patient-safety standards; and providers hired patient-safety officers and implemented numerous patient-safety initiatives. So what are the fruits of these efforts? Recently, we have received some deeply disappointing news: three studies have called into question whether we’ve made any progress at all. Landrigan et al. found that rates of injury due to medical error had remained es-
1756
sentially unchanged between 2000 and 2008 at 10 North Carolina hospitals.1 A report from the Inspector General of the Department of Health and Human Services (DHHS) revealed that Medicare patients experienced substantial harm in U.S. hospitals as recently as 2008.2 Finally, Classen and colleagues found that almost one in three patients are harmed during their hospital stay and that traditional approaches to measuring adverse events, whether using voluntary reporting or patient-safety indicators, substantially underestimate the events’ frequency.3 If the United States has made progress in patient safety, it has been inadequate. The primary reason for insufficient progress is the lack of a robust measurement program: there are still no nationally agreed-on methods for systematically identifying, tracking, and reporting adverse events. Here, the patientsafety movement can learn from the quality-improvement efforts that predate it. In the 1990s, emerging evidence suggested that providers were inconsistent in
their adherence to evidence-based treatments such as the use of aspirin for patients with acute myocardial infarction. Efforts by the Joint Commission for the Accreditation of Healthcare Organizations to systematically measure performance and give feedback to hospitals, coupled with subsequent efforts to publicly report performance on these measures, led to dramatic improvements in compliance.4 In the few areas of patient safety that have seen demonstrable improvement (e.g., catheter-related bloodstream infections), the changes are due, at least in part, to robust measurement programs, such as those run by the Centers for Disease Control and Prevention. In other areas, inadequate measures have hindered progress, and patients continue to suffer from the consequences of unsafe care. Although there is a shortage of good patient-safety metrics, poorquality measures are plentiful. The best known among these are patient-safety indicators, which use billing data to identify potential complications during a hos-
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
PERSPECTIVE
Harnessing Electronic Data for Safer Care
pitalization. They generally have poor sensitivity and specificity, and their utility varies with hospitals’ billing practices.3 And because data on them are collected in a post hoc fashion, they fail to engage clinicians at the time of care delivery — and data are generally unavailable to providers for review until years after the care is delivered. In an attempt to make patient-safety measures more visible, the Centers for Medicare and Medicaid Services (CMS) now makes hospital performance data on certain patient-safety indicators publicly available.5 However, these inadequate measures of safety, even if delivered to hospitals more quickly, are unlikely to engage front-line clinicians in activities that will make care safer. Another approach to safety measurement has been the use of voluntary (or occasionally, mandatory) reporting of adverse events. These efforts sporadically yield important insights — but generally have very low sensitivity (most adverse events are never reported), which makes it difficult for provider organizations to know whether they’re making progress. Finally, some have used the “trigger tools” method, which, though not extensively validated, appears to be sensitive in detecting adverse events.3 However, this approach, used primarily in research, is resource-intensive to implement and has gained little traction among providers as an ongoing approach to monitoring safety. Despite these challenges, we currently have an opportunity to turn the tide. The Health Information Technology for Economic and Clinical Health (HITECH) Act of 2009 provides, through CMS, financial incentives for physicians and hospitals to become “mean-
ingful users” of electronic health records (EHRs). EHRs can systematically measure patient safety, turning a laborious, manually intensive, and expensive process of sifting through medical records to identify adverse events into an automated one that is efficient, consistent, and affordable. Although the technology is already available, most EHRs today are not built with this capability in mind — and it won’t be easy (or cheap) to retrofit EHR systems later. Without an explicit commitment on the part of EHR vendors to build systems that can systematically track adverse events, most EHRs will fail to do it adequately, if at all. However, the federal government, which creates the meaningful use criteria, could include the ability to effectively measure and report adverse event rates as a “core” requirement of meaningful use. By staking out this ground, CMS can signal to vendors the importance of including such a capability in every EHR sold in the United States. If CMS chooses to use EHR-derived safety measures for public reporting or pay for performance, these metrics will need further validation, a lengthy process, but the agency can expedite the activities needed to ensure that we have validated measures quickly. Even without these additional validation efforts, simply providing better-quality EHR-derived safety data to physicians and hospitals can have a profound effect on patient-safety activities throughout the country. We recognize that EHR vendors face competing demands, and many advocacy groups are clamoring to have particular functions included in meaningful use. But the $30 billion in
taxpayer subsidies for EHR adoption was sold to the U.S. public principally as a way of making health care safer. The current EHR systems, if implemented well, may have a modest effect on safety. Requiring the presence and use of a safety-measurement module for identifying and tracking adverse events would provide a critical signal to providers that monitoring adverse events is essential. Such systems would provide information to hospitals on their performance relative to their peers and their progress toward the goal of causing no harm. Most important, it would allow them to track the effects of their interventions and determine which efforts worked and which ones didn’t. Data from such EHR surveillance systems could have patient identifiers removed and be pooled across many sites, which would increase the depth and breadth of the possible analyses and lead to new insight into delivering safer care. The U.S. health care system is at a crossroads when it comes to ensuring patient safety and earning the trust of the public. Our inadequate progress since To Err Is Human is disheartening, but we have an opportunity to right the ship. By making systematic measurement of adverse events a requisite function of the EHRs that are eligible for financial incentives, the federal government can change the way safety is measured and improved throughout the health care system. Without these data, we are likely to repeat our recent experience: good intentions, a lot of effort, and little demonstrable benefit. According to IOM estimates, as many as a million Americans may have died owing to adverse events in U.S. hospitals
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1757
PERSPE C T I V E
Harnessing Electronic Data for Safer Care
over the past decade. We must do better over the next decade. EHRs can improve the safety and culture of U.S. health care, but only if the federal government, as the nation’s largest health care payer, demonstrates that it is serious about improving patient safety. Disclosure forms provided by the authors are available with the full text of this article at NEJM.org. From the Department of Health Policy and Management, Harvard School of Public
Health, and the VA Boston Healthcare System — both in Boston (A.K.J.); and Computer Sciences Corporation and the University of Utah — both in Salt Lake City (D.C.C.). 1. Landrigan CP, Parry GJ, Bones CB, Hackbarth AD, Goldmann DA, Sharek PJ. Temporal trends in rates of patient harm resulting from medical care. N Engl J Med 2010;363: 2124-34. [Erratum, N Engl J Med 2010;363: 2573.] 2. Levinson D. Adverse events in hospitals: national incidence among Medicare beneficiaries. Washington, DC: Office of the Inspector General, Department of Health and Human Services, 2010.
3. Classen DC, Resar R, Griffin F, et al. ‘Global trigger tool’ shows that adverse events in hospitals may be ten times greater than previously measured. Health Aff (Millwood) 2011;30:581-9. [Erratum, Health Aff (Millwood) 2011;30:1217.] 4. Chassin MR, Loeb JM, Schmaltz SP, Wachter RM. Accountability measures — using measurement to promote quality improvement. N Engl J Med 2010;363:683-8. 5. Hospital quality initiatives: outcome measures. Baltimore: Centers for Medicare & Medicaid Services, 2011 (https://www .cms.gov/HospitalQualityInits/ 20_OutcomeMeasures.asp). Copyright © 2011 Massachusetts Medical Society.
Evidence-Based Medicine in the EMR Era Jennifer Frankovich, M.D., Christopher A. Longhurst, M.D., and Scott M. Sutherland, M.D.
M
any physicians take great pride in the practice of evidence-based medicine. Modern medical education emphasizes the value of the randomized, controlled trial, and we learn early on not to rely on anecdotal evidence. But the application of such superior evidence, however admirable the ambition, can be constrained by trials’ strict inclusion and exclusion criteria — or the complete absence of a relevant trial. For those of us practicing pediatric medicine, this reality is all too familiar. In such situations, we are used to relying on evidence at Levels III through V — expert opinion — or resorting to anecdotal evidence. What should we do, though, when there aren’t even meager data available and we don’t have a single anecdote on which to draw? We recently found ourselves in such a situation as we admitted to our service a 13-year-old girl with systemic lupus erythematosus (SLE). Our patient’s presentation was complicated by nephroticrange proteinuria, antiphospholipid antibodies, and pancreatitis. Al-
1758
though anticoagulation is not standard practice for children with SLE even when they’re critically ill, these additional factors put our patient at potential risk for thrombosis, and we considered anticoagulation. However, we were unable to find studies pertaining to anticoagulation in our patient’s situation and were therefore reluctant to pursue that course, given the risk of bleeding. A survey of our pediatric rheumatology colleagues — a review of our collective Level V evidence, so to speak — was equally fruitless and failed to produce a consensus. Without clear evidence to guide us and needing to make a decision swiftly, we turned to a new approach, using the data captured in our institution’s electronic medical record (EMR) and an innovative research data warehouse. The platform, called the Stanford Translational Research Integrated Database Environment (STRIDE), acquires and stores all patient data contained in the EMR at our hospital and provides immediate advanced text searching ca-
pability.1 Through STRIDE, we could rapidly review data on an SLE cohort that included pediatric patients with SLE cared for by clinicians in our division between October 2004 and July 2009. This “electronic cohort” was originally created for use in studying complications associated with pediatric SLE and exists under a protocol approved by our institutional review board. Of the 98 patients in our pediatric lupus cohort, 10 patients developed thrombosis, documented in the EMR, while they were acutely ill. The prevalence was higher among patients who had persistent nephrotic-range proteinuria and pancreatitis (see table). As compared with our patients with lupus who did not have these risk factors, the risk of thrombosis was 14.7 (95% confidence interval [CI], 3.3 to 96) among patients with persistent nephrosis and 11.8 (95% CI, 3.8 to 27) among those with pancreatitis. This automated cohort review was conducted in less than 4 hours by a single clinician. On the basis of this real-time, informatics-
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
PERSPE C T I V E
Harnessing Electronic Data for Safer Care
over the past decade. We must do better over the next decade. EHRs can improve the safety and culture of U.S. health care, but only if the federal government, as the nation’s largest health care payer, demonstrates that it is serious about improving patient safety. Disclosure forms provided by the authors are available with the full text of this article at NEJM.org. From the Department of Health Policy and Management, Harvard School of Public
Health, and the VA Boston Healthcare System — both in Boston (A.K.J.); and Computer Sciences Corporation and the University of Utah — both in Salt Lake City (D.C.C.). 1. Landrigan CP, Parry GJ, Bones CB, Hackbarth AD, Goldmann DA, Sharek PJ. Temporal trends in rates of patient harm resulting from medical care. N Engl J Med 2010;363: 2124-34. [Erratum, N Engl J Med 2010;363: 2573.] 2. Levinson D. Adverse events in hospitals: national incidence among Medicare beneficiaries. Washington, DC: Office of the Inspector General, Department of Health and Human Services, 2010.
3. Classen DC, Resar R, Griffin F, et al. ‘Global trigger tool’ shows that adverse events in hospitals may be ten times greater than previously measured. Health Aff (Millwood) 2011;30:581-9. [Erratum, Health Aff (Millwood) 2011;30:1217.] 4. Chassin MR, Loeb JM, Schmaltz SP, Wachter RM. Accountability measures — using measurement to promote quality improvement. N Engl J Med 2010;363:683-8. 5. Hospital quality initiatives: outcome measures. Baltimore: Centers for Medicare & Medicaid Services, 2011 (https://www .cms.gov/HospitalQualityInits/ 20_OutcomeMeasures.asp). Copyright © 2011 Massachusetts Medical Society.
Evidence-Based Medicine in the EMR Era Jennifer Frankovich, M.D., Christopher A. Longhurst, M.D., and Scott M. Sutherland, M.D.
M
any physicians take great pride in the practice of evidence-based medicine. Modern medical education emphasizes the value of the randomized, controlled trial, and we learn early on not to rely on anecdotal evidence. But the application of such superior evidence, however admirable the ambition, can be constrained by trials’ strict inclusion and exclusion criteria — or the complete absence of a relevant trial. For those of us practicing pediatric medicine, this reality is all too familiar. In such situations, we are used to relying on evidence at Levels III through V — expert opinion — or resorting to anecdotal evidence. What should we do, though, when there aren’t even meager data available and we don’t have a single anecdote on which to draw? We recently found ourselves in such a situation as we admitted to our service a 13-year-old girl with systemic lupus erythematosus (SLE). Our patient’s presentation was complicated by nephroticrange proteinuria, antiphospholipid antibodies, and pancreatitis. Al-
1758
though anticoagulation is not standard practice for children with SLE even when they’re critically ill, these additional factors put our patient at potential risk for thrombosis, and we considered anticoagulation. However, we were unable to find studies pertaining to anticoagulation in our patient’s situation and were therefore reluctant to pursue that course, given the risk of bleeding. A survey of our pediatric rheumatology colleagues — a review of our collective Level V evidence, so to speak — was equally fruitless and failed to produce a consensus. Without clear evidence to guide us and needing to make a decision swiftly, we turned to a new approach, using the data captured in our institution’s electronic medical record (EMR) and an innovative research data warehouse. The platform, called the Stanford Translational Research Integrated Database Environment (STRIDE), acquires and stores all patient data contained in the EMR at our hospital and provides immediate advanced text searching ca-
pability.1 Through STRIDE, we could rapidly review data on an SLE cohort that included pediatric patients with SLE cared for by clinicians in our division between October 2004 and July 2009. This “electronic cohort” was originally created for use in studying complications associated with pediatric SLE and exists under a protocol approved by our institutional review board. Of the 98 patients in our pediatric lupus cohort, 10 patients developed thrombosis, documented in the EMR, while they were acutely ill. The prevalence was higher among patients who had persistent nephrotic-range proteinuria and pancreatitis (see table). As compared with our patients with lupus who did not have these risk factors, the risk of thrombosis was 14.7 (95% confidence interval [CI], 3.3 to 96) among patients with persistent nephrosis and 11.8 (95% CI, 3.8 to 27) among those with pancreatitis. This automated cohort review was conducted in less than 4 hours by a single clinician. On the basis of this real-time, informatics-
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
PERSPECTIVE
Evidence-Based Medicine in the EMR Era
Results of Electronic Search of Patient Medical Records (for a Cohort of 98 Pediatric Patients with Lupus) Focused on Risk Factors for Thrombosis Relevant to Our 13-Year-Old Patient with Systemic Lupus Erythematosus.* Keywords Used to Conduct xpedited Electronic Search E
Outcome or Risk Factor
Prevalence of Thrombosis
Relative Risk (95% CI)
no./total no (%) Outcome — thrombosis
“Thrombus,” “Thrombosis,” “Blood clot”
10/98 (10)
Not applicable
Present at any time
“Nephrosis,” “Nephrotic,” “Proteinuria”
8/36 (22)
7.8 (1.7–50)
Present >60 days
“Urine protein”
7/23 (30)
14.7 (3.3–96)
Pancreatitis
“Pancreatitis,” “Lipase”
5/8 (63)
11.8 (3.8–27)
Antiphospholipid antibodies
“Aspirin”
6/51 (12)
1.0 (0.3–3.7)
Thrombosis risk factor Heavy proteinuria (>2.5 g per deciliter)
* In all cases, the sentences surrounding the keywords were manually reviewed to determine their relevance to our patient. Pancre atitis was defined as an elevated lipase level (twice the upper limit of normal) coexisting with abdominal pain. We used the word “aspirin” as a proxy for antiphospholipid antibodies, since it is standard practice at our institution to give all patients with these antibodies aspirin; if “aspirin” was found in the chart, than antiphospholipid-antibody status was confirmed by investigating the laboratory results.
enabled data analysis, we made the decision to give our patient anticoagulants within 24 hours after admission. Our case is but one example of a situation in which the existing literature is insufficient to guide the clinical care of a patient. But it illustrates a novel process that is likely to become much more standard with the widespread adoption of EMRs and more sophisticated informatics tools. Although many other groups have highlighted the secondary use of EMR data for clinical research,2,3 we have now seen how the same approach can be used to guide real-time clinical decisions. The rapid electronic chart review and analysis were not only feasible, but also more helpful and accurate than physician recollection and pooled colleague opinion. Such real-time availability of data to guide deci-
sion making has already transformed other industries,4 and the growing prevalence of EMRs along with the development of sophisticated tools for real-time analysis of deidentified data sets will no doubt advance the use of this datadriven approach to health care delivery. We look forward to a future in which health information systems help physicians learn from every patient at every visit and close the feedback loop for clinical decision making in real time. Did we make the correct decision for our patient? Thrombosis did not develop, and the patient did not have any sequelae related to her anticoagulation; truthfully, though, we may never really know. We will, however, know that we made the decision on the basis of the best data available — acting, as the fictional detective Nero Wolfe would say, “in the light of experience as guided
by intelligence.”5 In the practice of medicine, one can’t do better than that. Disclosure forms provided by the authors are available with the full text of this article at NEJM.org. From the Division of Rheumatology (J.F.), the Division of Systems Medicine (C.A.L.), and the Division of Nephrology (S.M.S.), Department of Pediatrics, Stanford University School of Medicine, Palo Alto, CA. This article (10.1056/NEJMp1108726) was published on November 2, 2011, at NEJM.org. 1. Lowe HJ, Ferris TA, Hernandez PM, Weber SC. STRIDE — an integrated standards-based translational research informatics platform. AMIA Annu Symp Proc 2009;Nov 14:391-5. 2. Prokosch HU, Ganslandt T. Perspectives for medical informatics: reusing the electronic medical record for clinical research. Methods Inf Med 2009;48:38-44. 3. Gunn PW, Hansen ML, Kaelber DC. Underdiagnosis of pediatric hypertension — an example of a new era of clinical research enabled by electronic medical records. AMIA Annu Symp Proc 2007;October 11:966. 4. Halevy A, Norvig P, Pereira F. The Unreasonable Effectiveness of Data. IEEE Intelligent Systems, March/April 2009:8-12. 5. Stout R. In the best families. New York: Viking Press, 1950:71. Copyright © 2011 Massachusetts Medical Society.
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1759
PERSPE C T I V E
Cost-Effectiveness and U.S. Vaccination Policy
The Role of Cost-Effectiveness in U.S. Vaccination Policy Jane J. Kim, Ph.D.
V
accination policy is driven by several factors, including vaccine safety and efficacy, avertable disease burden, acceptability, and societal value. One measure of value is an intervention’s costeffectiveness, defined as the additional cost required per additional unit of health benefit produced as compared with the next-mosteffective alternative. It is important to differentiate cost-effectiveness (value for money) from affordability (financial resources required); indeed, interventions with high value may not always be affordable. Although information on the cost-effectiveness of health interventions is increasingly being used in health policy globally, the extent to which this information influences decisions varies by country. For example, the governments in Britain and Australia explicitly and routinely incorporate findings from costeffectiveness analyses into coverage and reimbursement decisions; in contrast, in the United States, it has been essentially taboo for anyone in the public sector to refer explicitly to cost as a factor in health decisions. One exception is the Advisory Committee on Immunization Practices (ACIP), an independent expert advisory board that formally includes cost-effectiveness among the types of evidence it considers when making vaccinepolicy recommendations to the Centers for Disease Control and Prevention (CDC). The ACIP strives to be transparent and balanced, inviting perspectives from stakeholders ranging from scientists to patient groups, and tries to harmonize its recommendations with those of professional
1760
organizations, such as the American Academy of Family Physicians and the American Academy of Pediatrics. Historically, ACIP recommendations have influenced coverage decisions by both private and public insurers. Through a separate process, the ACIP also determines what vaccines are to be covered by the federal Vaccines for Children (VFC) program, which covers children who are Medicaid-eligible, uninsured or underinsured, or American Indians or Alaska natives up to the age of 18. With nearly 50% of U.S. children eligible for VFC coverage,1 the ACIP faces dual pressures: it must maximize underserved children’s access to vaccines while selecting vaccines that provide the most bang for the buck. This pressure will increase with the rollout of the Affordable Care Act, which mandates coverage of all ACIPrecommended childhood immunizations. With low cost and high efficacy, many vaccines are estimated to be cost-saving — the up-front expenditure for vaccination is entirely offset by costs averted through disease prevention. However, newly licensed and expensive vaccines, such as those against human papillomavirus (HPV, the virus causally linked to cervical cancer) and meningococcal disease, are being considered for use in ways that raise questions regarding their overall public health value as estimated in cost-effectiveness analyses. In late October, the ACIP is expected to vote on routine HPV vaccination in boys and young men and to discuss meningococcal vaccination in infants, includ-
ing its cost-effectiveness. Since 2007, routine HPV vaccination has been recommended for girls 11 to 12 years of age (and as early as 9 years), with “catch-up” vaccination recommended up to the age of 26, despite evidence of rapidly diminishing marginal returns and decreasing cost-effectiveness after 21 years of age.2 After the Food and Drug Administration (FDA) approved the quadrivalent HPV (HPV4) vaccine for males in 2009, the ACIP voted for “permissive” — but not routine — use of it in boys and men 9 to 26 years of age for prevention of genital warts. Despite this less enthusiastic stance, the ACIP voted in favor of VFC coverage for eligible males 9 to 18 years of age. The committee was persuaded not to recommend routine male HPV vaccination in part by evidence that it may not be costeffective, especially if vaccine uptake in girls and young women is high, given the sexual transmission of HPV infections and expected herd-immunity benefits through female-only vaccination. Recent data on uptake among adolescent girls, however, show less than 50% completion of the three-dose series, suggesting that HPV vaccination of boys may be cost-effective at this time. Furthermore, since the 2009 guidelines were issued, the indications for HPV4 have expanded to include prevention of anal cancers. Routine male HPV vaccination, especially if targeted at an early age, when the vaccines are expected to have highest benefit, would maximize protection for men who have sex with men, a group at high risk for HPV-related cancers that would receive little
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
PERSPECTIVE
Cost-Effectiveness and U.S. Vaccination Policy
herd-immunity protection from female-only vaccination. With respect to meningococcal vaccination, in October 2010, the ACIP decided in a narrow vote to recommend a single booster dose of the quadrivalent meningo coccal conjugate vaccine (MCV4) at the age of 16 despite evidence that routine adolescent MCV4 vaccination does not provide good value for money, largely because of low disease incidence rates and relatively high vaccine cost. Since then, the FDA has approved the licensure of one meningococcal vaccine for use in infants and is reviewing the licensing application for another. In considering expanding use to infants, the ACIP will need to contend with evidence that MCV4 vaccination at such young ages, which requires at least two doses, is even less cost-effective than adolescent vaccination.3 The cost-effectiveness of vaccines is influenced by several factors, including vaccine efficacy and durability, severity of disease burden, vaccine price, and deliveryprogram costs. The meningococcal and HPV vaccines are among the most expensive vaccines on the market, with costs of $82 and $109 per dose, respectively, in the public sector (private-sector costs are 20 to 30% higher).4 With the relatively high costs of new vaccines, the U.S. immunization program is placing an increasing financial strain on the health system. Today, the schedule of recommended routine child and adolescent vaccines includes more than 30 doses against 16 diseases — more than double the number in 1980. The public-sector cost of fully vaccinating one person as recommended through adulthood (not including annual influenza vaccines) is roughly $1,450 for males and $1,800 for females, of
which the HPV and meningococcal vaccinations alone account for more than 25% at current prices. Cost-effectiveness analysis provides information on whether the health gain associated with each new vaccine is worth the cost, as compared with other options for health spending. For example, the VFC program must weigh the cost of covering expensive vaccines against an alternative use of those dollars, such as outreach to improve uptake of other routine vaccines in the eligible population. Indeed, a recent CDC analysis showed that it would be more cost-effective to spend up to the purchase price of the HPV vaccine on improving vaccine uptake among girls than it would be to extend the program to boys.5 As the use of cost-effectiveness information increases, we should consider some important limitations of current analyses. The tendency to evaluate single diseases or interventions in isolation is restrictive. Individual vaccines may appear cost-effective, but the overall U.S. vaccination program may be unaffordable or provide less value than other bundled preventive health services targeting the same age group. Real-world obstacles should also be integrated into analyses; for example, the lack of organized vaccine-delivery mech anisms for older age groups can affect vaccine-uptake rates among adolescents and adults, and shortages in vaccine supply (as experienced with influenza vaccines) can influence cost-effectiveness results. To make cost-effectiveness analysis a more practical tool, analysts should evaluate investments across multiple diseases and interventions and include the influences of nonmonetary constraints. As we confront sobering proposals to cut more than $300 bil-
lion in federal health spending over the next decade, public health decision makers will increasingly have to make explicit choices among health investments while keeping a vigilant eye on total expenditures. Identification of highvalue health interventions through comparative effectiveness analysis has been prioritized by the new Patient-Centered Outcomes Research Institute. Evidence of cost-effectiveness, if provided in a transparent, standardized, and comprehensive manner, can help to highlight important tradeoffs and contribute to policy recommendations for vaccinations and other health interventions. Disclosure forms provided by the author are available with the full text of this article at NEJM.org. From the Center for Health Decision Science, Department of Health Policy and Management, Harvard School of Public Health, Boston. This article (10.1056/NEJMp1110539) was published on October 19, 2011, at NEJM.org. Editor’s Note: On October 25, the ACIP voted to recommend that boys 11 to 12 years of age be routinely vaccinated against HPV, indicating that the vaccine series can be started as early as age 9 and that men up to age 21 who have not yet received the vaccine should be vaccinated. 1. Smith PJ, Lindley MC, Rodewald LE. Vaccination coverage among U.S. children aged 19-35 months entitled by the Vaccines for Children program, 2009. Public Health Rep 2011;126:Suppl 2:109-23. 2. Kim JJ, Brisson M, Edmunds WJ, Goldie SJ. Modeling cervical cancer prevention in developed countries. Vaccine 2008;26:Suppl 10:K76-K86. 3. Shepard CW, Ortega-Sanchez IR, Scott RD II, Rosenstein NE. Cost-effectiveness of conjugate meningococcal vaccination strategies in the United States. Pediatrics 2005; 115:1220-32. 4. CDC vaccine price list. Atlanta: Centers for Disease Control and Prevention (http:// www.cdc.gov/vaccines/programs/vfc/ cdc-vac-price-list.htm). 5. Chesson HW, Ekwueme DU, Saraiya M, Dunne EF, Markowitz LE. The cost-effectiveness of male HPV vaccination in the United States. Vaccine 2011;29:8443-50. Copyright © 2011 Massachusetts Medical Society.
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1761
new england journal of medicine The
established in 1812
november 10, 2011
vol. 365 no. 19
German Outbreak of Escherichia coli O104:H4 Associated with Sprouts Udo Buchholz, M.D., M.P.H., Helen Bernard, M.D., Dirk Werber, D.V.M., Merle M. Böhmer, Cornelius Remschmidt, M.D., Hendrik Wilking, D.V.M., Yvonne Deleré, M.D., Matthias an der Heiden, Ph.D., Cornelia Adlhoch, D.V.M., Johannes Dreesman, Ph.D., Joachim Ehlers, D.V.M., Steen Ethelberg, Ph.D., Mirko Faber, M.D., Christina Frank, Ph.D., Gerd Fricke, Ph.D., Matthias Greiner, D.V.M., Ph.D., Michael Höhle, Ph.D., Sofie Ivarsson, M.Sc., Uwe Jark, D.V.M., Markus Kirchner, M.D., M.P.H., Judith Koch, M.D., Gérard Krause, M.D., Ph.D., Petra Luber, Ph.D., Bettina Rosner, Ph.D., M.P.H., Klaus Stark, M.D., Ph.D., and Michael Kühne, D.V.M., Ph.D.
A bs t r ac t Background
A large outbreak of the hemolytic–uremic syndrome caused by Shiga-toxin–producing Escherichia coli O104:H4 occurred in Germany in May 2011. The source of infection was undetermined. Methods
We conducted a matched case–control study and a recipe-based restaurant cohort study, along with environmental, trace-back, and trace-forward investigations, to determine the source of infection. Results
The case–control study included 26 case subjects with the hemolytic–uremic syndrome and 81 control subjects. The outbreak of illness was associated with sprout consumption in univariable analysis (matched odds ratio, 5.8; 95% confidence interval [CI], 1.2 to 29) and with sprout and cucumber consumption in multivariable analysis. Among case subjects, 25% reported having eaten sprouts, and 88% reported having eaten cucumbers. The recipe-based study among 10 groups of visitors to restaurant K included 152 persons, among whom bloody diarrhea or diarrhea confirmed to be associated with Shiga-toxin–producing E. coli developed in 31 (20%). Visitors who were served sprouts were significantly more likely to become ill (relative risk, 14.2; 95% CI, 2.6 to ∞). Sprout consumption explained 100% of cases. Trace-back investigation of sprouts from the distributor that supplied restaurant K led to producer A. All 41 case clusters with known trading connections could be explained by producer A. The outbreak strain could not be identified on seeds from the implicated lot. Conclusions
Our investigations identified sprouts as the most likely outbreak vehicle, underlining the need to take into account food items that may be overlooked during subjects’ recall of consumption.
From the Department of Infectious Disease Epidemiology (U.B., H.B., D.W., M.M.B., C.R., H.W., Y.D., M. Heiden, C.A., M.F., C.F., M. Höhle, J.K., G.K., B.R., K.S.) and the Post graduate Training for Applied Epidemiology Program (H.W., C.A.), Robert Koch Insti tute; the Federal Office of Consumer Protec tion and Food Safety (G.F., P.L.); and the Federal Institute for Risk Assessment (M.G.) — all in Berlin; the Governmental Institute of Public Health of Lower Saxony, Hannover (J.D., M. Kirchner), the Lower Saxony State Office for Consumer Protection and Food Safety, Oldenburg (J.E., U.J., M. Kühne), and the Veterinary University, Hannover (M.G.) — all in Germany; the Swedish Institute for Communicable Disease Control, Solna, Swe den (S.I.); and the Department of Epidemi ology, Statens Serum Institut, Copenhagen (S.E.). Address reprint requests to Dr. Buch holz at the Department of Infectious Dis ease Epidemiology, Robert Koch Institute, DGZ-Ring 1, 13086 Berlin, Germany, or at
[email protected]. Members of the Task Force EHEC (Entero hemorrhagic Escherichia coli) at the Fed eral Office of Consumer Protection and Food Safety in Germany, the Danish HUS (Hemolytic–Uremic Syndrome) Investiga tion Team, the Governmental Institute of Public Health of Lower Saxony HUS Investi gation Team, the Lower Saxony State Of fice for Consumer Protection and Food Safety HUS Investigation Team, the Robert Koch Institute HUS Investigation Team, and the Swedish HUS Investigation Team are listed in the Supplementary Appendix, available at NEJM.org. This article (10.1056/NEJMoa1106482) was published on October 26, 2011, at NEJM.org. N Engl J Med 2011;365:1763-70. Copyright © 2011 Massachusetts Medical Society.
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1763
The
n e w e ng l a n d j o u r na l
H
uman infection with Shiga-toxin– producing Escherichia coli is a major cause of postdiarrheal hemolytic–uremic syndrome. This life-threatening disorder, which is characterized by acute renal failure, hemolytic anemia, and thrombocytopenia, typically affects children under the age of 5 years. Shiga-toxin–producing E. coli O157 is the serogroup that is most frequently isolated from patients with the hemolytic–uremic syndrome worldwide.1 In May 2011, a large outbreak of the hemolytic– uremic syndrome associated with the rare E. coli serotype O104:H4 occurred in Germany.2-5 The main epidemiologic features were that the peak of the epidemic was reached on May 21 and May 22 4,5 and that the vast majority of case subjects either resided or had traveled in northern Germany. Almost all patients from other European countries or from North America had recently returned from northern Germany.2,6,7 Of the affected case subjects, 90% were adults, and more than two thirds of case subjects with the hemolytic–uremic syndrome were female.4 Early studies in Hamburg suggested that infections were probably community-acquired and were not related to food consumption in a particular restaurant. A first case–control study that was conducted on May 23 and 24 suggested that raw food items, such as tomatoes, cucumbers, or leaf salad,3 were the source of infection. The consumption of sprouts, which was previously implicated in outbreaks of Shiga-toxin–producing E. coli in the United States8 and Japan,9 was mentioned by only 25% of case subjects in exploratory interviews, so consumption of sprouts was not tested analytically. This report describes the investigations that were conducted by the federal agencies under the auspices of the German Ministry of Health and the Ministry of Food, Agriculture, and Consumer Protection, as well as by the respective state agencies, to identify the vehicle of infection of this international outbreak.
Me thods Study Design
Three types of parallel studies were conducted: one case–control study, one recipe-based restaurant cohort study, and combined trace-back and trace-forward investigations. The main results of
1764
of
m e dic i n e
all three studies became available between June 2 and June 9, 2011. Results, even if preliminary, were communicated among the investigation groups and the Task Force EHEC (Enterohemorrhagic Escherichia coli) at the Federal Office of Consumer Protection and Food Safety in Berlin. If the results were judged to have appropriate validity, they were communicated as soon as possible to the public. The case–control and cohort studies were conducted within the framework of the Communicable Diseases Law Reform Act of Germany. Mandatory regulations were observed, and review by an ethics committee was not required. Case–Control Study
From May 29 to June 4, we conducted a case–control study to further specify the type of raw vegetables associated with illness in this outbreak. A case was defined as clinically diagnosed hemolytic–uremic syndrome in an adult who was hospitalized in one of three hospitals in northern Germany, located in the cities of Bremen, Bremerhaven, and Lübeck. Control subjects were individually matched with case subjects on the basis of age group and neighborhood. Case and control subjects were predominantly asked about consumption of fruit and vegetable items, including sprouts, during the 14 days before the onset of illness (for case subjects) or before the interview date (for control subjects) (for details, see the Supplementary Appendix, available with the full text of this article at NEJM.org). Recipe-Based Restaurant Cohort Study
Since the earlier studies had not identified a single source of infection, we conducted a cohort study at restaurant K in Lübeck, Schleswig–Holstein. Preliminary information revealed that several visitor groups with subsequent cases of gastroenteric disease had eaten in restaurant K between May 12 and May 16, 2011, which was defined as the outbreak period in this study. Using the booking notes, we identified cohorts that had eaten in the restaurant during this period and asked all members about the menu items they had consumed. We interviewed the chef of the restaurant about the ingredients and their quantities used to prepare the menu items offered in the restaurant. A case was defined as an illness in a member of any of the cohorts that was associated with bloody diarrhea, self-reported laboratory-confirmed Shiga-toxin–producing E. coli O104 infection, or
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
Escherichia coli O104:H4 Associated with Sprouts
the hemolytic–uremic syndrome with an onset of diarrhea within 2 weeks after visiting restaurant K. Non–case subjects were those who remained healthy after visiting restaurant K. We excluded from the analysis all subjects who had diarrhea that was not bloody, who did not have laboratory-confirmed Shiga-toxin–producing E. coli O104 infection, or whose disease onset was later than 14 days after visiting restaurant K. For the analysis, we evaluated only data on ingredients that were used in the dishes that were served to the restaurant guests (i.e., data that were solely based on the information obtained from the chef) (for details, see the Supplementary Appendix). To estimate the total number of cases accrued at restaurant K during the study period, we collected information on the total number of main dishes purchased in this period from billing data. Using the attack rate among persons who were served sprouts from the cohort study, we calculated the number of ill persons as the number of main dishes served containing sprouts times the attack rate among sprout eaters.
Table 1. Vegetables or Fruits Evaluated in a Case–Control Study in the German Outbreak.*
Food Item
Case Subjects Control Subjects Exposed Exposed
Matched Odds Ratio (95% CI)
P Value
no./total no. (%) Sprouts
6/24 (25)
7/80 (9)
4.35 (1.05–18.0)
0.04
Cucumbers
22/25 (88)
52/79 (66)
3.53 (0.96–12.9)
0.06
Apples
22/24 (92)
57/81 (70)
3.91 (0.86–17.7)
0.08
Peppers
16/24 (67)
35/80 (44)
2.66 (0.90–7.9)
0.08
Strawberries
19/26 (73)
43/81 (53)
2.33 (0.90–6.0)
0.08
* P>0.10 for raw onions, tomatoes, leaf salad, asparagus, carrots, and basil.
sprouts from suspect producer A to outbreak clusters, a study that was conducted by the food-safety authorities of the respective counties, compiled by the pertinent federal food-safety authorities, and analyzed by the task force. The task force also ini tiated a trace-back investigation from producer A.
R e sult s
Environmental, Trace-Back, and TraceForward Investigations
Case–Control Study
From physicians, patients, county and regional health departments, and foreign national public health institutes, we received information on clusters or apparently sporadic cases of illness that occurred in the context of the overall epidemic among persons who had probable exposure at only one location or venue (see the Supplementary Appendix). Information was continuously reported to national and local food-safety authorities and to the task force. On the basis of findings in the early studies, food-safety authorities initially concentrated their investigations on tomatoes, cucumbers, and leaf salads, as well as on other vegetables eaten raw and salad ingredients, including toppings. Local and state food-safety authorities assessed distribution channels of raw food products connected with clusters or single case subjects with single exposures. In addition, both epidemiologically suspected and other raw food items were sampled and, after specific enrichment procedures, were tested by means of immunoassay for Shiga toxin and polymerase-chain-reaction assay for the Shiga-toxin stx2 prophage gene cluster and for genetic markers of the O104:HA strains. The task force initiated a trace-forward investigation for
We included 26 case subjects (9 male and 17 female) and 81 control subjects in the study. On univariable analysis, the only significant variable was sprouts (Table 1). Other food items, such as raw minced beef and milk and other dairy products, were not significantly associated with illness. The sequential addition and removal of other variables resulted in a multivariable model containing only sprouts (matched odds ratio, 5.8; 95% confidence interval [CI], 1.2 to 29.0) and cucumbers (matched odds ratio, 6.0; 95% CI, 1.1 to 31.0). Before the date on which the public was advised not to consume sprouts (June 10, 2011), only 6 of 24 case subjects (25%) remembered having consumed them (Table 1, and the Supplementary Appendix). After that date, we wished to ascertain the possible degree of false recall among the case subjects. We tried to recontact all case and control subjects who had not reported sprout consumption previously. Of 8 case subjects who could be reached, 3 (38%) remembered having eaten sprouts in the 14 days before the onset of illness. By contrast, all 37 control subjects who had not reported sprout consumption in previous interviews continued to report that they had not eaten sprouts.
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1765
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
Table 2. Characteristics of 10 Cohorts of Subjects Who Visited Restaurant K between May 12 and May 16, 2011.*
Cohort Subjects No. Size with Data no. 1
37
Male Sex
Median Age (IQR)
%
yr
37
5
49 (45–57)
Subjects with LaboratoryConfirmed STEC Infection
Attack Rate for Subjects Subjects Subjects Subjects Evaluated Fulfilling Fulfilling with for Case Case Case HUS Definition† Definition Definition
Subjects with Diarrhea
Subjects with Bloody Diarrhea
10
9
10
4
no.
no. (%)
no.
%
34 (92)
9
26
2
2
2
50
49 (46–52)
1
1
2
0
1
50
3
31
31
55
57 (45–68)
11
4
5
1
25 (81)
5
20
4
11
10
0
54 (53–55)
2
0
1
0
9 (90)
1
11
5
12
9
0
46 (43–48)
1
1
2
0
9 (100)
6
19
19
37
32 (15–49)
1
1
1
0
7
10
10
50
40 (40–42)
0
0
0
8
17
14
57
68 (65–73)
3
3
4
9
25
25
40
74 (69–75)
9
9
10
13
11
36
24 (21–48)
3
1
177
168
32
53 (42–67)
41
29
Total
2 (100)
1
11
17 (89)
1
6
0
9 (90)
0
0
1
14 (100)
3
21
6
1
25 (100)
9
36
2
1
8 (73)
1
13
33
8
152 (90)
31
20
* HUS denotes hemolytic–uremic syndrome, IQR interquartile range, and STEC Shiga-toxin–producing Escherichia coli. † The case definition was the presence of bloody diarrhea, laboratory-confirmed infection with Shiga-toxin–producing E. coli, or the hemolytic– uremic syndrome with an onset of disease within 2 weeks after visiting restaurant K. According to the case definition, 16 subjects were ex cluded either because they had diarrhea that was not bloody or because the date of the onset of diarrhea was more than 14 days after visit ing restaurant K or was unknown.
Recipe-Based Restaurant Cohort Study
We identified 10 cohorts with a total of 177 persons who had eaten at restaurant K (Table 2). Of these persons, we interviewed 168 (95%), including 161 who were interviewed directly and 7 for whom information was obtained from a proxy. Among the 152 persons who could be evaluated for the case definition, 31 (20%) had an illness that fulfilled the case definition. Among these subjects, the hemolytic–uremic syndrome developed in 8 (26%) (see the Supplementary Appendix). In univariable analysis of all raw food items, only visitors who had been served sprouts were significantly more likely to become ill (Table 3). The P value for the risk ratio for all other items was greater than 0.15. Of 115 persons who had been served sprouts, 31 (27%; 95% CI, 19 to 36) became case subjects, whereas none of 37 persons who had not been served sprouts reported having gastrointestinal symptoms that fulfilled the case definition. Thus, all 31 case subjects had been served menu items containing sprouts. Nearly half the menu items served in the restaurant contained raw sprouts as a garnish or were served with
1766
a side salad containing raw sprouts. Side salads contained radicchio, Chinese cabbage, lettuce, cucumber, tomato, and sprouts. No menu item contained cooked sprouts. During this period, the restaurant used only one type of sprout assortment, which was received from a distributor in Schleswig-Holstein and contained four types of sprouts: lentil sprouts, alfalfa sprouts, fenugreek sprouts, and adzuki bean sprouts. During the outbreak period, 884 main dishes containing sprouts were served to the guests of the restaurant. Applying the attack rate of 27% among sprout eaters, we extrapolated that a tot al of 239 cases of bloody diarrhea (95% CI, 168 to 318) occurred among customers of this restaurant. Environmental, Trace-Back, and TraceForward Investigations
The task force identified 41 clusters or cases with single exposures. One of the clusters was a hotel in lower Saxony in which Swedish citizens were affected (cluster 1) (Fig. 1). A “spicy sprout mixture” that was sampled from the distributor (distributor 1) of the hotel’s restaurant on June 2, 2011,
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
Escherichia coli O104:H4 Associated with Sprouts
Table 3. Relative Risk of Infection Associated with Sprouts and Other Raw Food Items in Univariable Analysis.
Food Item
Total Subjects Evaluated
Subjects Exposed (Percent of Cohort)
Cases among Subjects Exposed (Attack Rate)
no.
Subjects Not Exposed (Percent of Cohort)
Cases among Subjects Not Exposed (Attack Rate)
Relative Risk (95% CI)
P Value
no. (%)
Sprouts
152
115 (76)
31 (27)
37 (24)
Tomatoes
152
50 (33)
14 (28)
102 (67)
17 (17)
1.68 (0.77–3.62)
0.18
Cucumbers
152
50 (33)
14 (28)
102 (67)
17 (17)
1.68 (0.77–3.62)
0.18
Chinese cabbage
152
45 (30)
13 (29)
107 (70)
18 (17)
1.72 (0.77–3.71)
0.17
Radicchio
152
45 (30)
13 (29)
107 (70)
18 (17)
1.72 (0.77–3.71)
0.17
Lettuce
152
45 (30)
13 (29)
107 (70)
18 (17)
1.72 (0.77–3.71)
0.17
tested positive for Shiga toxin on a commercial immunoassay. Although confirmatory testing later proved to be negative, immediate trace back led to producer A in Lower Saxony, at which a total of 452 water, seed, sprout, and surface samples tested negative for Shiga-toxin–producing E. coli O104:HA. Producer A was licensed as a horticultural farm and produced 18 different sorts of sprouts at the time. Protective measures consisted of regularly instructing employees on the application of proper hygiene necessary for the production of sprouts and the frequent testing of sprouts for salmonella, according to European Union regulations, as well as for coliforms. All employees of producer A were interviewed, and 5 of 15 had become ill in May 2011 or tested positive for O104:H4. Employees frequently ate sprouts produced at their company. Preferred types were fenugreek, broccoli, and garlic sprouts. Tracing forward from producer A led to four distributors (Fig. 1). Distributor 1 was connected not only to the restaurant in which cluster 1 had occurred but also to restaurant K. Distributor 4 delivered food to a caterer in Frankfurt that was linked to a cafeteria outbreak that occurred early in the epidemic.3 Subsequently, 22 more distributors (for a total of 26) that obtained sprouts from producer A were identified. Distributors were located in 7 of the 16 federal states. Each of the 41 case clusters could be linked with at least one of the identified sprout distributors (Fig. 2).10 Fenugreek or lentil sprouts were suspected as the outbreak vehicle because these types of sprouts were the common ingredients in two different sprout mixtures that were packaged for distribution by producer A and had been supplied to most of the 41 clusters.
0
14.23 (2.55–∞)
0.001
In mid-June, investigations on the origin of sprouts that were consumed by additional case subjects revealed that two case subjects from Lüneburg, Lower Saxony, had eaten a homegrown sprout mix that included fenugreek sprouts. The seeds for these sprouts had been purchased at a retail store that had the same supplier of seeds (supplier X) as producer A (Fig. 1).
Discussion We report evidence from epidemiologic, microbiologic, and food trace-back and trace-forward investigations that incriminates sprouts as the vehicle of infection in this large outbreak of the hemolytic–uremic syndrome associated with Shigatoxin–producing E. coli. Although definitive molecular evidence is lacking, the argument that sprouts were responsible for this outbreak is strong on the basis of the following five factors: both epidemiologic studies implicated sprouts, the restaurant study showed that 100% of cases of illness could be explained by the consumption of sprouts, no other food ingredient consumed at restaurant K was associated with the risk of illness, all 41 clusters or cases of single exposure could be linked to sprout producer A and its distribution channels, and several employees of sprout producer A who frequently consumed sprouts at the company became symptomatically ill early in the outbreak period or tested positive for Shiga-toxin–producing E. coli O104:H4. Information obtained during the outbreak investigation in Germany already hinted at an outbreak source before producer A in the seed and sprout distribution chain. Seeds that were used by producer A and by the two case subjects from
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1767
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
Cluster 1 Hotel and restaurant, District Lüneburg, Lower Saxony
At least 18 cases
Cluster 2 Restaurant K, Lübeck, Schleswig– Holstein
At least 50 cases
Distributor 2, Lower Saxony
Cluster 3 Restaurant, District Rotenburg/ Wümme, Lower Saxony
At least 4 cases
Distributor 3, North Rhine– Westphalia
Cluster 4 Cafeteria, Bochum, North Rhine–Westphalia
At least 6 cases
Distributor 1, Schleswig–Holstein
Sprout producer A
Cluster 5 Cafeteria, Frankfurt, Hesse
At least 20 cases
Several distributors, Germany
Homegrown sprouts, District Lüneburg, Lower Saxony
At least 2 cases
Distributors, United Kingdom, France
Homegrown sprouts, Bordeaux, France
15 cases
Distributor 4, Hesse
Sprout seeds
Supplier X
Distributor D
Figure 1. Anatomy of the German Outbreak. Shown are the trading connections from supplier X to sprout producer A and through four distributors to five outbreak clusters, as well as to two other distributor groups leading to a cluster of illnesses caused by Shiga-toxin–producing E. coli in Lüneburg and an outbreak in France unrelated to producer A. Known case subjects at restaurant K included those who were part of the cohort study and others who were not.
Lower Saxony who grew their own sprouts originated from supplier X. In June 2011, an outbreak of the hemolytic–uremic syndrome associated with Shiga-toxin–producing E. coli O104:HA occurred in Bordeaux, France.11 The E. coli responsible for the outbreak was genetically related to that in the German outbreak, and there was an epidemio1768
logic association with consumption of homegrown fenugreek sprouts. These findings spurred trace-back investigations by a task force (set up by the European Food Safety Authority (EFSA), which consisted of experts from the European Commission, relevant European Union member states, the European Center for Disease Preven-
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
Escherichia coli O104:H4 Associated with Sprouts
tion and Control, the World Health Organization, the Food and Agriculture Organization of the United Nations, and EFSA staff members. It concluded that a certain lot of fenugreek seeds that was imported in late 2009 from Egypt was the most likely common link for the outbreaks in Germany and France.12 With respect to the point of contamination, it is possible that it occurred at the site where seeds were produced, during transportation, or at the importer. By August 2011, this question had not been resolved. The account of the outbreak was dramatic: 4321 outbreak cases, including 3469 cases of Shigatoxin–producing E. coli and 852 cases of the hemolytic–uremic syndrome, had been reported by July 26, 2011, when the outbreak was declared to be over.13 By that time, 50 patients had died. After epidemiologic and food-safety investigations had concurred in identifying sprouts as the outbreak vehicle, the public had been advised on June 10 to abstain from the consumption of raw sprouts, to eliminate raw sprouts in possession at that time, and to remove any sprouts stemming from producer A.14 In addition, producer A was temporarily closed. Cases of illness still occurred until the end of July 2011, partially as a consequence of secondary transmission,15 but the number of cases dropped substantially. Early in the outbreak investigation, raw food products other than sprouts had been suspected as the vehicle. The three studies that we present here built on these findings and complement one another. The early epidemiologic findings helped food-safety authorities to streamline their investigations, which led them to turn their attention to producer A. Since the case–control study was conducted before sprouts came into focus, the significant association between sprouts and illness is therefore revealing and important. Nevertheless, the findings received strong support only through the restaurant study and the food traceback and trace-forward investigations. The restaurant study provided an idea of why tomatoes, cucumbers, and leaf salad had been suspected early on. The one dish that frequently exposed guests to sprouts was the side salad, which contained tomatoes, cucumbers, three sorts of leaf salads, and sprouts. Sprouts may have been the ingredient that visitors recalled least in such a mixed salad. This hypothesis is also suggested by the results of the repeated interviews in the case– control study. Because the earlier studies had attempted to find a vehicle that explained the n engl j med 365;19
Figure 2. Trading Network Leading to the German Outbreak. The trading network for the contaminated sprouts led from producer A in Lower Saxony (yellow circle) to 26 sprout distributors (black dots) and 41 identified out break clusters (red squares), established by combined back and forward tracing.
majority of cases, sprouts were missed. Although international guidelines16 generally recommend otherwise, this experience suggests that food items or ingredients that are deemed to be hard to remember should be included in analytical studies, even if such items are mentioned by less than 50% of those surveyed. Producer A was licensed as a horticultural company. Although hygienic measures were satisfactory and local food-safety authorities had inspected the company routinely under the same conditions as a food-processing company, it became apparent that European legislation has important deficits regarding Shiga-toxin–producing E. coli. Production of food that is vulnerable to contamination with this pathogen, such as sprouts or sprout seeds, should be monitored for this organism so that hygienic measures prevent amplification. In addition, both incoming seeds and outgoing food products should be tested for Shiga-toxin–producing E. coli. In general, focused restaurant studies provide a favorable situation to identify the vehicle, even in large, geographically dispersed outbreaks, because place and time of exposure are known and
nejm.org
november 10, 2011
The New England Journal of Medicine
1769
Escherichia coli O104:H4 Associated with Sprouts
menu cards can be used to spur visitors’ memories. If one or few specific menu items are identified, ingredients can be further analyzed.17 If not, the reason may be that a common ingredient is contained in many menu items, resulting in the necessity to collect detailed ingredient information of the whole or a large part of the menu,18,19 information that needs to be obtained from those preparing the food. In conclusion, we have presented investigative results regarding an outbreak of the hemolytic– uremic syndrome associated with Shiga-toxin– producing E. coli. Under favorable circumstances, the recipe-based restaurant cohort study proved to be a quick method for detecting suspected food
ingredients with high reliability in a complicated setting of exposures. Recommendations regarding sprout use and consumption may need to be strengthened or adjusted as a consequence of this outbreak. Disclosure forms provided by the authors are available with the full text of this article at NEJM.org. We thank the senior and junior owner and particularly the chef of restaurant K in Lübeck for their support and patience; physicians, patients, and colleagues in the hospitals in Bremen (Klinikum Bremen Mitte), Bremerhaven (Klinikum Bremerhaven), and Lübeck (Universitätsklinikum Schleswig-Holstein, Campus Lübeck), as well as county health departments of Bremen, Bremerhaven, Cuxhaven, and Lübeck for their support in conducting the case–control study; and Johan Giesecke, Stephen Palmer, Kåre Mølbak, Henriette de Valk, Roland Salmon, Lyle Petersen, Robert Tauxe, and Hubert Deluyker for their input into the interpretation of study and surveillance data and the management of the outbreak.
References 1. Tarr PI, Gordon CA, Chandler WL.
Shiga-toxin-producing Escherichia coli and haemolytic uraemic syndrome. Lancet 2005; 365:1073-86. 2. Askar M, Faber MS, Frank C, et al. Update on the ongoing outbreak of haemolytic uraemic syndrome due to Shiga toxin-producing Escherichia coli (STEC) serotype O104, Germany, May 2011. Euro Surveill 2011;16:19883. 3. Frank C, Faber MS, Askar M, et al. Large and ongoing outbreak of haemolytic uraemic syndrome, Germany, May 2011. Euro Surveill 2011;16:19878. 4. Frank C, Werber D, Cramer JP, et al. Epidemic profile of Shiga-toxin–producing Escherichia coli O104:H4 outbreak in Germany. N Engl J Med 2011. DOI: 10.1056/ NEJMoa1106483. 5. Wadl M, Rieck T, Nachtnebel M, et al. Enhanced surveillance during a large outbreak of bloody diarrhoea and haemolytic uraemic syndrome caused by Shiga toxin/ verotoxin-producing Escherichia coli in Germany, May to June 2011. Euro Surveill 2011;16:19893. 6. European Centre for Disease Prevention and Control. ECDC rapid risk assessment: outbreak of Shiga toxin-producing E. coli (STEC) in Germany. June 14, 2011 (http:// www.ecdc.europa.eu/en/publications/ Publications/1106_TER_Risk_Assessment _EColi_revised.pdf). 7. World Health Organization. International Health Regulations: EHEC outbreak in Germany (http://www.euro.who.int/en/ what-we-do/health-topics/emergencies/
1770
international-health-regulations/ehecoutbreak-in-germany). 8. Taormina PJ, Beuchat LR, Slutsker L. Infections associated with eating seed sprouts: an international concern. Emerg Infect Dis 1999;5:626-34. 9. Michino H, Araki K, Minami S, et al. Massive outbreak of Escherichia coli O157:H7 infection in schoolchildren in Sakai City, Japan, associated with consumption of white radish sprouts. Am J Epidemiol 1999;150:787-96. 10. Bundesinstitut für Risikobewertung (Bf R). Bedeutung von Sprossen und Keimlingen sowie Samen zur Sprossenherstellung im EHEC O104:H4 Ausbruchs geschehen im Mai und Juni 2011 (Stellung nahme Nr. 023/2011 des BfR. July 5, 2011) (http://www.bfr.bund.de/cm/343/bedeutung _von_sprossen_und_keimlingen_sowie _samen_zur_sprossenherstellung_im _ehec_o104_h4_ausbruchsgeschehen _im_mai_und_juni_2011.pdf ). 11. Gault G, Weill FX, Mariani-Kurkdjian P, et al. Outbreak of haemolytic uraemic syndrome and bloody diarrhoea due to Escherichia coli O104:H4, south-west France, June 2011. Euro Surveill 2011;16:19905. 12. European Food Safety Authority. Tracing seeds, in particular fenugreek (Trigonella foenum-graecum) seeds, in relation to the Shiga toxin-producing E. coli (STEC) O104:H4 2011 outbreaks in Germany and France. July 5, 2011 (http://www.efsa .europa.eu/en/supporting/doc/176e.pdf). 13. Robert Koch-Institute. Informationen zum EHEC-/HUS-Ausbruchsgeschehen von
Mai bis Juli 2011 in Deutschland — Ende des Ausbruchs. Epidemiol Bull 2011;31:295-6 (http://www.rki.de/cln_160/nn_2030884/ DE/Content/Infekt/EpidBull/Archiv/2011/ 31__11,templateId=raw,property= publicationFile.pdf/31_11.pdf ). 14. Idem. Neue Erkenntnisse zum EHECAusbruch: Gemeinsame Pressemitteilung von BfR, BVL und RKI. June 10, 2011 (http://www.rki.de/cln_169/nn_467482/DE/ Content/Service/Presse/Pressemitteilungen/ 2011/08__2011.html). 15. Hauri A, Gotsch U, Strotmann I, et al. Secondary transmissions during the outbreak of Shiga toxin-producing Escherichia coli O104 in Hesse, Germany, 2011. Euro Surveill 2011;16:19937. 16. Foodborne disease outbreaks: guidelines for investigation and control. Geneva: World Health Organization, 2008. 17. Mody RK, Greene SA, Gaul L, et al. National outbreak of Salmonella serotype saintpaul infections: importance of Texas restaurant investigations in implicating jalapeño peppers. PLoS One 2011;6(2): e16579. 18. Buchholz U, Mermin J, Rios R, et al. An outbreak of food-borne illness associated with methomyl-contaminated salt. JAMA 2002;288:604-10. 19. Wheeler C, Vogt TM, Armstrong GL, et al. An outbreak of hepatitis A associated with green onions. N Engl J Med 2005; 353:890-7. Copyright © 2011 Massachusetts Medical Society.
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
original article
Epidemic Profile of Shiga-Toxin–Producing Escherichia coli O104:H4 Outbreak in Germany Christina Frank, Ph.D., Dirk Werber, D.V.M., Jakob P. Cramer, M.D., Mona Askar, M.D., Mirko Faber, M.D., Matthias an der Heiden, Ph.D., Helen Bernard, M.D., Angelika Fruth, Ph.D., Rita Prager, Ph.D., Anke Spode, M.D., Maria Wadl, D.V.M., Alexander Zoufaly, M.D., Sabine Jordan, M.D., Markus J. Kemper, M.D., Per Follin, M.D., Ph.D., Luise Müller, M.Sc., Lisa A. King, M.P.H., Bettina Rosner, Ph.D., Udo Buchholz, M.D., M.P.H., Klaus Stark, M.D., Ph.D., and Gérard Krause, M.D., Ph.D., for the HUS Investigation Team*
ABSTRACT Background
We describe an outbreak of gastroenteritis and the hemolytic–uremic syndrome caused by Shiga-toxin–producing Escherichia coli in Germany in May, June, and July, 2011. The consumption of sprouts was identified as the most likely vehicle of infection. Methods
We analyzed data from reports in Germany of Shiga-toxin–producing E. coli gastroenteritis and the hemolytic–uremic syndrome and clinical information on patients presenting to Hamburg University Medical Center (HUMC). An outbreak case was defined as a reported case of the hemolytic–uremic syndrome or of gastroenteritis in a patient infected by Shiga-toxin–producing E. coli, serogroup O104 or serogroup unknown, with an onset of disease during the period from May 1 through July 4, 2011, in Germany. Results
A total of 3816 cases (including 54 deaths) were reported in Germany, 845 of which (22%) involved the hemolytic–uremic syndrome. The outbreak was centered in northern Germany and peaked around May 21 to 22. Most of the patients in whom the hemolytic–uremic syndrome developed were adults (88%; median age, 42 years), and women were overrepresented (68%). The estimated median incubation period was 8 days, with a median of 5 days from the onset of diarrhea to the development of the hemolytic–uremic syndrome. Among 59 patients prospectively followed at HUMC, the hemolytic–uremic syndrome developed in 12 (20%), with no significant differences according to sex or reported initial symptoms and signs. The outbreak strain was typed as an enteroaggregative Shiga-toxin–producing E. coli O104:H4, producing extended-spectrum beta-lactamase.
From the Departments of Infectious Disease Epidemiology (C.F., D.W., M.F., M.H., H.B., M.W., B.R., U.B., K.S., G.K.) and Infectious Diseases (A.F., R.P.) and the Postgraduate Training for Applied Epidemiology Program (M.A.), Robert Koch Institute, Berlin; the Department of Internal Medicine, University Medical Center Hamburg–Eppendorf (J.P.C., A.Z., S.J., M.J.K.) and the Health Department of the Hamburg Northern District (A.S.) — both in Hamburg, Germany; the Department of Communicable Disease Control and Prevention, Västra Götaland Region, Gothenburg, Sweden (P.F.); the Statens Serum Institut, Copenhagen (L.M.); and the Institut de Veille Sanitaire, Saint-Maurice, France (L.A.K.). Address reprint requests to Dr. Werber at
[email protected]. Drs. Frank and Werber contributed equally to this article. *The members of the Hemolytic–Uremic Syndrome (HUS) Investigation Team are listed in the Supplementary Appendix, available at NEJM.org. A preliminary version of this article was published on June 22, 2011, and updated on June 24, 2011. This final version (10.1056/NEJMoa1106483) was published on October 26, 2011, at NEJM.org. N Engl J Med 2011;365:1771-80.
Conclusions
In this outbreak, caused by an unusual E. coli strain, cases of the hemolytic–uremic syndrome occurred predominantly in adults, with a preponderance of cases occurring in women. The hemolytic–uremic syndrome developed in more than 20% of the identified cases. n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
Copyright © 2011 Massachusetts Medical Society.
1771
The
n e w e ng l a n d j o u r na l
O
n May 19, 2011, the Robert Koch Institute, Germany’s national-level public health authority, was informed about a cluster of three cases of the hemolytic–uremic syndrome in children admitted on the same day to the university hospital in the city of Hamburg. On May 20, a team from the Robert Koch Institute arrived in Hamburg to assist with the public health investigation. It quickly became clear that the case numbers were continuing to rise, that there were also cases in adults, and that other areas of Germany, especially northern Germany, were also affected. An investigation of the outbreak involving all levels of public-health and food-safety authorities was initiated to identify the causative agent and the vehicle of infection in order to prevent further cases of disease. Sprouts were eventually identified as the most likely vehicle of infection.1 The hemolytic–uremic syndrome, which was first described in children in the 1950s,2 is characterized by the triad of acute renal failure, hemolytic anemia, and thrombocytopenia. Diarrheaassociated hemolytic–uremic syndrome occurs primarily in children, and a precipitating infection with Shiga-toxin–producing Escherichia coli, mainly of serotype O157:H7, is the primary cause.3 The usual reservoir for these bacteria is ruminants, particularly cattle. Human infection with Shigatoxin–producing E. coli occurs through the inadvertent ingestion of fecal matter — for example, through contaminated food or water or through contact with animals or their farm environment or, secondarily, through contact with infected humans. In contrast, in adults, the hemolytic–uremic syndrome with prodromal diarrhea, indicating an infectious cause, is a rare event. For example, from 1989 through 2006, only 21 of the 322 adults (7%) listed in the Oklahoma registry as having thrombotic thrombocytopenic purpura or the hemolytic– uremic syndrome presented with bloody diarrhea.4 Earlier, we presented descriptive epidemiologic, clinical, and microbiologic information on the unusual outbreak in Germany in a preliminary report (available at NEJM.org). This report updates and finalizes this information.
Methods German Surveillance System
of
m e dic i n e
(Stx) in E. coli isolates or of its encoding gene (stx) in stool enrichment culture or isolates must, by law, be reported by diagnosing laboratories to local health departments. This reporting process allows the identification of Shiga-toxin–producing E. coli infection independently of serogroup (serotyping information is requested but not required). The German case definition of Shiga-toxin–producing E. coli gastroenteritis (without the hemolytic–uremic syndrome) requires, besides laboratory confirmation, the presence of at least one of the following symptoms: diarrhea (three or more loose stools in a 24-hour period), abdominal cramps, or vomiting. In addition, physicians are required to report clinical symptoms compatible with diarrhea-associated hemolytic–uremic syndrome in a patient. The German case definition of the hemolytic– uremic syndrome comprises thrombocytopenia (platelet count of <150,000 per cubic millimeter), hemolytic anemia, and acute renal dysfunction. The third criterion is met if at least one of the following findings is present: an increase in the serum creatinine level (unspecified), oliguria, anuria, proteinuria, or hematuria. Reported cases of the hemolytic–uremic syndrome or Shiga-toxin–producing E. coli infection were investigated and recorded by the local health department, and the reports were forwarded electronically, without identifying information, through the state to the federal level. Disease onset was defined as the onset of diarrhea, regardless of whether the hemolytic–uremic syndrome developed at a later date. An outbreak case was defined as a reported case of the hemolytic–uremic syndrome or a reported case of gastroenteritis in a patient infected by Shiga-toxin–producing E. coli, of serogroup O104 or unknown serogroup, with a disease onset during the period from May 1 through July 4, 2011, in Germany. We describe here data from the national reporting database on infectious diseases as of September 19, 2011. The descriptive analysis focuses primarily on reported cases of the hemolytic–uremic syndrome as indicators for the entire outbreak. To show the outbreak area, a map of the incidence of the disease according to county was generated (Fig. 1). Cases were attributed to a particular county if that county was the probable place of infection.
According to the German Protection against In- Clinical Information fection Act of 2001, the detection of a Shiga toxin We analyzed clinical data from two groups of pa-
1772
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
Shiga-Toxin–Producing E. coli Outbreak in Germany
Denmark Baltic Sea North Sea
Schleswig–Holstein Mecklenburg–Vorpommern Bremen Hamburg Netherlands
Poland
Lower Saxony
Berlin
Brandenburg Saxony–Anhalt
North Rhine– Westphalia
Saxony
Restaurant clusters
Thuringia Hesse Rhineland– Palatinate
Czech Republic
Saarland Bavaria France
N
Baden–Württemberg Incidence per 100,000 population
100 km
Austria
Switzerland
0.00 0.01–0.70 0.71–2.50 2.51–30.00
Figure 1. Incidence of the Hemolytic–Uremic Syndrome According to County in Germany. The incidence shown is per 100,000 population. A total of 845 cases were detected in this outbreak. Cases are attributed to a particular county if that county was the probable site of infection.
COLOR FIGURE
Version 5
Author Fig # Title ME
n engl j med 365;19
nejm.org
november 10, 2011
The New England Journal of Medicine
DE Artist
10/13/11
Frank Werber 1 E coli DMoskowitz LB LAM AUTHOR PLEASE NOTE:
1773
Figure has been redrawn and type has been reset Please check carefully
The
n e w e ng l a n d j o u r na l
tients at the Hamburg University Medical Center (HUMC): patients who were positive for stx at their first presentation to the HUMC during the period from May 19 through June 1 (a cross-sectional analysis of data extracted from electronic medical records) and a prospectively assembled hospital-based cohort of adults who were seen from May 25 through June 6 at a special unit that was set up during the outbreak. The study protocol for the cohort study was approved by the ethics committee of the Hamburg Chamber of Physicians. Patients were enrolled in the study if they presented with bloody diarrhea or if they had any diarrhea after contact with a patient who had Shiga-toxin–producing E. coli infection. All patients provided written informed consent. Patients were followed for at least 14 days and were tested for the outbreak strain according to the protocol of the National Consulting Laboratory on Hemolytic–Uremic Syndrome.5 Only data from patients infected by the outbreak strain were included in the analysis. The proportion of patients with the hemolytic–uremic syndrome among all patients who were positive for Shigatoxin–producing E. coli was calculated. Platelet counts and creatinine and lactate dehydrogenase levels were monitored daily. To estimate the proportion of nonbloody diarrhea, bloody diarrhea (without the hemolytic– uremic syndrome), and the hemolytic–uremic syndrome among patients who were thought to be infected with E. coli O104, we pooled the data from six cohorts that were investigated by public health authorities during the course of the outbreak — originally with the intention of identifying the most likely vehicle of infection. We describe the frequencies and proportions of selfreported (bloody) diarrhea and clinically diagnosed hemolytic–uremic syndrome irrespective of the results of microbiologic diagnosis. Microbiologic Analysis
Shiga-toxin–producing E. coli infection was diagnosed by private microbiologic laboratories either by screening for Stx with the use of one of several commercially available enzyme immunoassays or by detection of stx with the use of polymerase chain reaction (PCR). The National Reference Center for Salmonella and Other Bacterial Enteric Pathogens confirmed the presence of
1774
of
m e dic i n e
Shiga-toxin–producing E. coli, cultured the isolates, and characterized the Shiga-toxin–producing E. coli in samples from local or regional laboratories that were positive for Stx or stx. Chromogenic agar media for Enterobacteriaceae that were positive for extended-spectrum beta-lacta mase (ESBL) were used for isolation of the strain. Biochemical characterization of the strain was performed with the use of various commercially available tests (VITEK, bioMérieux; MicroPlate GN, BIOLOG; and API, bioMérieux). Shiga-toxin– producing E. coli virulence-factor genes (stx1, stx2, eae, and ehx) were detected by established PCR methods.6,7 The presence of virulence-factor genes that are typical of enteroaggregative E. coli, such as aatA, aggR, aap, aggA and aggC, were detected according to established PCR protocols.8 Antimicrobial susceptibility was tested by means of microdilution assays with the use of minimal inhibitory concentrations according to the guidelines of the European Committee on Antimicrobial Susceptibility Testing. Serotyping of Shiga-toxin– producing E. coli followed standard protocols.9 One-enzyme (XbaI) pulsed-field gel electrophoresis was performed on Shiga-toxin–producing E. coli O104:H4 isolates.10 Given the strain’s properties, a shortened protocol was recommended by the National Consulting Laboratory on Hemolytic–Uremic Syndrome5 and was used by the National Reference Center and HUMC for confirmation of the outbreak strain. Statistical Analysis
For statistical comparisons, the z test was used for proportions, and the Mann–Whitney U test for age distribution. The incubation period was estimated on the basis of data from selected patients with the hemolytic–uremic syndrome or Shiga-toxin– producing E. coli gastroenteritis for whom the date of onset of diarrhea was known. They either had stayed in northern Germany for no more than 48 hours or were part of disease clusters with known date and place of exposure.1 The interval between the date of onset of diarrhea and the date of diagnosis of the hemolytic–uremic syndrome was calculated with the use of information from the clinician’s notification form, which was sent without identifying information to the Robert Koch Institute.
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
Shiga-Toxin–Producing E. coli Outbreak in Germany
R e sult s Outbreak Cases
A total of 3816 cases reported to public health authorities in Germany with onset dates during the period from May 1 through July 4, 2011, were attributed to the outbreak: 845 cases of the hemolytic–uremic syndrome, including 36 fatal cases (4.2%; 95% confidence interval [CI], 3.0 to 5.8), and 2971 additional cases of Shiga-toxin– producing E. coli gastroenteritis (all laboratoryconfirmed), including 18 fatal cases (0.6%; 95% CI, 0.4 to 1.0). Thus, the hemolytic–uremic syndrome developed in 22% of the patients ascertained in this outbreak. The number of cases of hemolytic-uremic syndrome during the outbreak period was almost 70 times the number that had occurred during this period in previous years.
The outbreak grew dramatically starting on May 8; cases of the hemolytic–uremic syndrome peaked on May 21, and cases of Shiga-toxin– producing E. coli gastroenteritis peaked on May 22 (Fig. 2), with a median date of hospitalization for the hemolytic–uremic syndrome of May 24. Among patients who died, death occurred a median of 10 days after the onset of the disease. The first patients who had laboratory-confirmed infection with the outbreak strain became ill on May 8 (for earlier outbreak cases, information on serogroup was not available). Cases of the hemolytic–uremic syndrome were reported from all 16 states in Germany. The highest incidences were reported from the northern states of Hamburg (10.0 cases per 100,000 population), Schleswig–Holstein (6.9 cases per 100,000), Bremen (2.7 cases per 100,000), Meck-
250 STEC gastroenteritis cases in males STEC gastroenteritis cases in females HUS cases in males
200
HUS cases in females
No. of Cases
150
100
50
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 1 2 3 4
May
June
July
Date of Disease Onset
Figure 2. Epidemiologic Curve of the Outbreak. RETAKE: 1st Shown are the numbers of cases of the hemolytic–uremic syndrome (HUS) and of Shiga-toxin–producing Escherichia coli (STEC) gastroAUTHOR: Frank (Werber) 2nd845 cases of the hemolytic–uremic synenteritis, according to sex. Only cases with a known date of onset are included here — 802 of FIGURE: 1 of 3 3rd drome and 2700 of 2971 cases of Shiga-toxin–producing E. coli diarrhea. Revised ARTIST: ts TYPE:
Line
Combo
4-C
H/T
SIZE 7 col
AUTHOR, PLEASE NOTE: Figure has been redrawn and type has been reset. Please check carefully. JOB: 36519 ISSUE: 11-10-11 n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1775
The
n e w e ng l a n d j o u r na l
lenburg–Vorpommern (2.3 cases per 100,000), and Lower Saxony (1.8 per 100,000) — the “northern Germany outbreak area,” where the outbreak started almost simultaneously in the various affected states. Most of the cases from other states can be linked to travel-related exposures in the northern Germany outbreak area. Figure 2 shows the incidence of the disease according to county of infection. Aside from two satellite clusters linked to restaurants in eastern North Rhine–Westphalia and southern Hesse, the area with high incidences (2.5 to 30 reported cases per 100,000 population) was centered around the city of Hamburg. A total of 88% of the case patients with the hemolytic–uremic syndrome in this outbreak were adults (i.e., persons older than 17 years of age). Among case patients 17 years of age or younger, the median age was 11 years. Only 2% of the case patients with the hemolytic–uremic syndrome were younger than 5 years of age, as compared with 69% of case patients with the hemolytic–uremic syndrome reported in Germany from 2001 through 2010.11 The median age of all patients with the hemolytic–uremic syndrome in the outbreak was 42 years. The median age of case patients with the hemolytic–uremic syndrome who died was 74 years (range, 20 to 91, except for one 2-year old boy), and the median age of patients with Shiga-toxin–producing E. coli gastroenteritis who died was 82 years (range, 38 to 89). Among women, the incidence of the hemolytic–uremic syndrome peaked in the age group of 30 to 34 years, and among men in the age group of 25 to 29 years (Fig. 1 in the Supplementary Appendix, available with the full text of this article at NEJM.org). A total of 68% of the case patients with the hemolytic–uremic syndrome and 58% of the case patients with Shiga-toxin–producing E. coli gastroenteritis were female; among the patients in the two groups who died, 75% and 50%, respectively, were female. The proportion of male case patients with the hemolytic–uremic syndrome rose from 30% among the patients with a disease onset in May to 46% among the patients with disease onset in June (P = 0.02). On the basis of data from 91 case patients, we estimated that the median incubation period for this pathogen in this outbreak was 8 days (interquartile range, 6 to 10), without an apparent difference between cases of Shiga-toxin–producing E. coli gastroenteritis and cases of the 1776
of
m e dic i n e
hemolytic–uremic syndrome. Among 98 case patients for whom data on both the date of onset of diarrhea and the date of onset of the hemolytic–uremic syndrome were known, the interval from the onset of diarrhea to the diagnosis of the hemolytic–uremic syndrome was 5 days (interquartile range, 4 to 7). Clinical Information
Data on 166 patients, obtained at their first presentation to HUMC, were analyzed; 143 of the patients (86%) were adults; 62% were female. No patient had a fever (defined as a temperature of at least 38.5°C) at the first presentation. Bloody diarrhea was reported less often in children than in adults (64% [14 of 22 children] vs. 91% [126 of 138 adults], P<0.001), whereas abdominal pain was a very common symptom in both children and adults, occurring in 93% of the children (13 of 14) and in 88% of the adults (121 of 137). Vomiting occurred more often in children than in adults (72% [13 of 18 children] vs. 18% [20 of 114 adults], P <0.001). Most patients did not have significantly elevated leukocyte levels (most were within the normal range; in some cases, counts were approximately 13,000 per cubic millimeter) or C-reactive protein levels (typically about 15 to 35 mg per liter [normal level, <5 mg per liter]). A total of 40 patients (24%) already met the criteria for the hemolytic–uremic syndrome at the time of presentation. Clinical and laboratory values in adults and children, stratified according to the presence or absence of the hemolytic–uremic syndrome, are summarized in Table 1. Among the 135 patients who were followed in the hospital-based cohort, the outbreak strain was detected in 59 (44%), and the hemolytic– uremic syndrome developed in 12 of these patients (20%; 95% CI, 11 to 33). Demographic and clinical characteristics at presentation did not differ significantly between patients with diarrhea in whom the hemolytic–uremic syndrome developed and those in whom it did not develop (Table 2). An examination of the platelet counts and creatinine and lactate dehydrogenase levels 5 days before through 2 days after the onset of the syndrome in 22 patients with the hemolytic– uremic syndrome (Fig. 3) indicates that the development of the hemolytic–uremic syndrome was sudden. In six closed cohorts, 127 of 416 persons reported having had diarrhea, of whom 27 (21%) were clinically diagnosed with the hemolytic–
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
Shiga-Toxin–Producing E. coli Outbreak in Germany
Table 1. Demographic and Clinical Characteristics and Laboratory Test Values of Patients Positive for Shiga-Toxin–Producing Escherichia coli at First Presentation.* Total (N = 166)
Variable
Adults (N = 142) Without HUS (N = 119)
Children (N = 23)
P Value†
With HUS (N = 23)
Age — yr
Without HUS (N = 6)
P Value†
With HUS (N = 17)
0.12
0.42
Median
38
37
38
12
10
Range
18–87
20–84
18–87
1–17
1–15
Male sex — no. (%)
63 (38)
45 (38)
4 (17)
0.06
4 (67)
10 (59)
0.74
Bloody diarrhea — no./total no. (%)
141/161 (88)
106/116 (91)
20/22 (91)
0.94
3/5 (60)
11/17 (65)
0.85
Abdominal pain — no./total no. (%)
134/152 (88)
102/115 (89)
19/22 (86)
0.76
5/5 (100)
8/9 (89)
0.44
Nausea — no./total no. (%)
38/115 (33)
23/84 (27)
9/19 (47)
0.09
3/4 (75)
3/7 (43)
0.30
Vomiting — no./total no. (%)
33/133 (25)
14/94 (15)
6/20 (30)
0.11
4/5 (80)
9/13 (69)
0.65
Temperature — °C
36.7±0.5
36.6±0.5
36.8±0.5
0.31
36.9±0.5
37.0±0.6
0.80
Hemoglobin — g/dl
13.4±2.2
14.2±1.3
11.8±2.8
<0.001
13.7±1.6
10.1±1.9
<0.001
Leukocytes — ×10−9/liter
11.3±4.1
11.0±3.5
12.4±5.8
0.11
12.2±5.3
12.1±5.0
0.98
209.3±90.0
245.2±51.3
111.7±91.9
<0.001
295.7±35.5
63.6±56.8
<0.001
Creatinine — mg/dl
1.4±1.9
0.8±0.2
2.0±1.7
<0.001
0.7±0.3
4.7±4.3
0.04
Bilirubin — mg/dl
0.9±0.7
0.8±0.5
1.8±1.0
<0.001
0.8±0.5
1.3±0.8
0.24
Platelets — ×10−9/liter
Lactate dehydrogenase — U/liter
424±575
193±92
671±464
<0.001
235±39
1707±860
<0.001
C-reactive protein — mg/liter
18.1±28.8
14.4±26.0
29.9±32.6
0.02
39.0±61.2
18.5±18.9
0.22
* Plus–minus values are means ±SD. Data are for patients who presented to the Hamburg University Medical Center between May 19 and June 11, 2011. To convert the values for creatinine to micromoles per liter, multiply by 88.4. To convert the values for bilirubin to micromoles per liter, multiply by 17.1. HUS denotes the hemolytic–uremic syndrome. † P values are for the presence versus the absence of the hemolytic–uremic syndrome in children and adults.
uremic syndrome (Table 1 in the Supplementary this outbreak clone. A total of 120 of these were Appendix). analyzed by pulsed-field gel electrophoresis; all had indistinguishable patterns. Microbiologic Features
The serotype of the E. coli outbreak strain is O104:H4. The strain ferments sorbitol within 24 hours and is positive for lactose and beta-glucuronidase. The pathogen possesses genes typical of enteroaggregative E. coli, such as attA, aggR, aap, aggA, and aggC, located on a virulence plasmid. In addition, the strain carries the gene for a Shiga-toxin 2 variant (stx2a). Other typical Shigatoxin–producing E. coli genes such as stx1, eae, and ehx are missing. All isolates classified as the outbreak strain are resistant to beta-lactam antibiotics (e.g., ampicillin) and third-generation cephalosporins and are partially resistant to fluoroquinolones (nalidixic acid). The strain is sensitive to carbapenems and ciprofloxacin. The outbreak strain produces an ESBL complex (CTX-M15) and beta-lactamase TEM-1. The National Reference Center typed 1023 isolates of
Discussion We describe the epidemiologic characteristics of an outbreak of infection with Shiga-toxin–producing E. coli O104:H4. There were more than 800 incident cases of the hemolytic–uremic syndrome in this outbreak during the period from May 1 through July 4, 2011, and altogether more than 3800 cases of disease. In addition, as many as 15 other countries, including the United States, reported cases occurring among people who had traveled to northern Germany: 51 cases of the hemolytic–uremic syndrome (including 2 deaths) and 89 cases of Shiga-toxin–producing E. coli gastroenteritis.12 The outbreak probably began on May 8; however, because not every case was laboratory-confirmed and serotyped, we cannot be sure that there were not earlier cases.
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1777
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
Table 2. Demographic and Clinical Characteristics of Patients Positive for Shiga-Toxin–Producing Escherichia coli (STEC) Who Were Followed Prospectively. Characteristic Age — yr Male sex — no. (%)
Total (N = 59)
Without HUS (N = 47)
With HUS (N = 12)
38.6±14.0
38.5±13.3
38.9±16.8
0.93
23 (39)
18 (38)
5 (42)
0.76
P Value†
Reported fever — no./total no. (%)
4/55 (7)
4/46 (9)
0/9 (0)
0.12
Bloody diarrhea — no./total no. (%)
48/58 (83)
37/46 (80)
10/12 (83)
0.87
Interval between onset of diarrhea and first presentation in STEC unit — days
4.1±4.7
4.0±5.0
4.1±3.3
0.98
Stool frequency — no. of stools/day
9.4±8.9
9.9±9.6
7.4±5.4
0.47
Abdominal pain — no. (%)
46 (78)
35 (74)
11 (92)
0.24
Vomiting — no. (%)
11 (19)
7 (15)
4 (33)
0.15
Previous contact with other patients with STEC — no. (%)
13 (22)
11 (23)
2 (17)
0.80
* Plus–minus values are means ±SD. Data are for patients who were prospectively followed at the STEC unit of the Hamburg University Medical Center between May 25 and June 6, 2011. † P values are for the presence versus the absence of the hemolytic–uremic syndrome.
There are important differences between this outbreak and previous outbreaks of Shiga-toxin– producing E. coli infection,13-18 such as the one that occurred in Japan in 1996, in which there were 121 cases of the hemolytic–uremic syndrome — all in children.14 First, the hemolytic– uremic syndrome represents more than 20% of the ascertained cases, which is a much larger percentage than in other outbreaks. Second, the majority of the cases of the hemolytic–uremic syndrome (88%) occurred in adults rather than in children, with the majority occurring in women. Third, the causative agent was a Shiga-toxin–producing E. coli strain of serotype O104:H4. The outbreak strain combines the virulence properties of two different diarrhea-causing E. coli pathotypes: typical enteroaggregative E. coli and Shiga-toxin–producing E. coli. The outbreak strain carries the chromosomal backbone of a typical enteroaggregative E. coli strain.19,20 It is likely that it has acquired the bacteriophage encoding stx2a and other genetic elements.20-22 Similar but not identical enteroaggregative Shiga-toxin–producing E. coli — even of serotype O104:H4 — have been isolated previously, albeit rarely, from patients with the hemolytic–uremic syndrome23,24 (e.g., in 2001 from two siblings in Germany in whom the hemolytic–uremic syndrome had developed25). Since typical enteroaggregative E. coli are isolated primarily from humans,26 the origin of this outbreak may not have been zoonotic. Two observations suggest that the pathogen in this outbreak is exceptionally virulent. First, 1778
the hemolytic–uremic syndrome developed in a large proportion of patients — 22% of case patients ascertained in the German surveillance system for infectious diseases, 20% of prospectively observed patients with Shiga-toxin–producing E. coli diarrhea at a hospital in Hamburg, and 21% of persons in closed cohorts investigated during the course of the outbreak. These proportions are consistently higher than those in previous outbreaks13-18 and higher than the proportion (6%) ascertained through active surveillance of Shiga-toxin–producing E. coli O157:H7, the virulent prototype of Shiga-toxin–producing E. coli, in the United States.27 Second, the outbreak strain caused the hemolytic–uremic syndrome in 101 children even though the strain lacked the intestinal adherence factor intimin (encoded by the gene eae); eae-negative strains have previously been isolated from adults with the hemolytic–uremic syndrome28 but rarely from children. For example, 97% of Shiga-toxin–producing E. coli isolated from children with the hemolytic–uremic syndrome in Germany and Austria carried the eae gene.29 Another unique feature of this outbreak, probably attributable to the pathogen, was the estimated median incubation period of 8 days, which was longer than the 3-day to 4-day incubation period reported for Shiga-toxin–producing E. coli O157:H7.13,30 The patients affected in this outbreak were mainly adults, and even among pediatric case patients with the hemolytic–uremic syndrome, the median age was considerably higher than
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
Shiga-Toxin–Producing E. coli Outbreak in Germany
n engl j med 365;19
A Platelets (×10−9/liter)
300 200 100 0
−5 (N= 5)
−4 −3 −2 −1 0 1 2 (N= 8) (N= 14) (N= 9) (N= 12) (N= 22) (N= 20) (N= 21)
Days in Relation to HUS Onset
B Creatinine (mg/dl)
3.0 2.5 2.0 1.5 1.0 0.5 0.0
−5 (N= 5)
−4 −3 −2 −1 0 1 2 (N= 8) (N= 14) (N= 8) (N= 12) (N= 22) (N= 20) (N= 21)
Days in Relation to HUS Onset
C Lactate Dehydrogenase (U/liter)
that of pediatric case patients in other years; the median age of 11 years in the outbreak surveillance data was supported by a median age of 11.8 years among 33 outbreak case patients at HUMC’s pediatric nephrology unit. It is unclear whether this atypical age distribution of cases primarily reflects patterns of sprout consumption or is attributable to the specific properties of this outbreak strain — or both. Furthermore, it remains to be elucidated why women were overrepresented among the cases of the hemolytic– uremic syndrome. The predominance of women among the case patients may be driven by the food vehicle if women are more health conscious and thus more likely to eat sprouts. Toward the end of the outbreak, when secondary household transmission probably contributed a larger proportion of cases, there was a shift to a more even sex distribution. No sex difference was observed with respect to the risk of development of the hemolytic–uremic syndrome among a limited sample of patients with diarrhea who were prospectively followed in the hospital-based cohort. The most common clinical sign in adults was bloody diarrhea accompanied by abdominal cramps. The clinical presentation in adults differed from that in children. Bloody diarrhea occurred significantly more often in adults — irrespective of the presence or absence of the hemolytic–uremic syndrome — whereas vomiting was reported more frequently in children. Clinical symptoms such as abdominal pain, bloody diarrhea, and the frequency of loose stools did not differ between patients in whom the hemolytic–uremic syndrome developed and those in whom it did not. Changes in laboratory values, indicating renal failure and hemolysis, occurred quickly, often within 24 hours (Fig. 3). Daily laboratory testing of platelet counts and creatinine and lactate dehydrogenase levels appeared to be pivotal for the early diagnosis of the hemolytic–uremic syndrome, and these laboratory tests were more sensitive than were patientreported symptoms and the physical examination. Indeed, several patients reported that they had begun to recover from bloody diarrhea several days after the initial presentation, at the same time as the onset of the hemolytic–uremic syndrome. For many reported cases, information on exact symptoms (e.g., diarrhea or bloody diarrhea) and additional microbiologic information were not available. Consequently, although the clinical picture of the hemolytic–uremic syndrome in adults ap-
1500 1000 500 0
−5 (N= 5)
−4 −3 −2 −1 0 1 2 (N= 8) (N= 14) (N= 8) (N= 12) (N= 22) (N= 20) (N= 21)
Days in Relation to HUS Onset
Figure 3. Selected Laboratory Values. Shown are selected laboratory values from up to 22 patients followed prospectively from their first presentation, relative to the onset of the hemolytic– uremic syndrome (HUS) (indicated by the vertical red line). Solid lines represent medians, and dashed lines interquartile ranges. To convert values for creatinine to micromoles per liter, multiply by 88.4.
peared to be very specific for this outbreak, among the cases of Shiga-toxin–producing E. coli diarrhea, those unrelated to this outbreak could not be efficiently filtered out owing to the lack of serotype information in many reported cases. In summary, this outbreak exemplifies the threat posed by foodborne pathogens with their propensity to cause large common-source outbreaks. These outbreaks may have unusual disease patterns and sometimes affect new population subgroups.
nejm.org
november 10, 2011
The New England Journal of Medicine
1779
Shiga-Toxin–Producing E. coli Outbreak in Germany Disclosure forms provided by the authors are available with the full text of this article at NEJM.org. We thank the physicians and laboratory personnel, who were working under intense strain and yet kept up their notification
requirements; local and state health departments for quickly passing on case data to the Robert Koch Institute; and epidemiologists in other countries for providing detailed travel and disease information regarding patients with travel-associated cases.
References 1. Buchholz U, Bernard H, Werber D, et
al. German outbreak of Escherichia coli O104:H4 associated with sprouts. N Engl J Med 2011. DOI: 10.1056/NEJMoa1106482. 2. Gasser C, Gautier E, Steck A, Siebenmann RE, Oechslin R. Hemolytic-uremic syndrome: bilateral necrosis of the renal cortex in acute acquired hemolytic anemia. Schweiz Med Wochenschr 1955;85: 905-9. (In German.) 3. Tarr PI, Gordon CA, Chandler WL. Shiga-toxin-producing Escherichia coli and haemolytic uraemic syndrome. Lancet 2005;365:1073-86. 4. Karpac CA, Li X, Terrell DR, et al. Sporadic bloody diarrhoea-associated thrombotic thrombocytopenic purpurahaemolytic uraemic syndrome: an adult and paediatric comparison. Br J Haematol 2008;141:696-707. 5. Laboratory information regarding the STEC outbreak strain (as of 1 June, 2011). Muenster, Germany: Universitatsklinikum Münster, 2011 (http://www.ehec.org/pdf/ Laborinfo_01062011.pdf). 6. Cebula TA, Payne WL, Feng P. Simultaneous identification of strains of Escherichia coli serotype O157:H7 and their Shiga-like toxin type by mismatch amplification mutation assay-multiplex PCR. J Clin Microbiol 1995;33:248-50. [Erratum, J Clin Microbiol 1995;33:1048.] 7. Friedrich AW, Bielaszewska M, Zhang WL, et al. Escherichia coli harboring Shiga toxin 2 gene variants: frequency and association with clinical symptoms. J Infect Dis 2002;185:74-84. 8. Jenkins C, Chart H, Willshaw GA, Cheasty T, Smith HR. Genotyping of enteroaggregative Escherichia coli and identification of target genes for the detection of both typical and atypical strains. Diagn Microbiol Infect Dis 2006;55:13-9. 9. Prager R, Strutz U, Fruth A, Tschäpe H. Subtyping of pathogenic Escherichia coli strains using flagellar (H)-antigens: serotyping versus fliC polymorphisms. Int J Med Microbiol 2003;292:477-86. 10. Ribot EM, Fair MA, Gautom R, et al. Standardization of pulsed-field gel electrophoresis protocols for the subtyping of Escherichia coli O157:H7, Salmonella, and Shigella for PulseNet. Foodborne Pathog Dis 2006;3:59-67. 11. German national infectious disease
1780
notification data base. Berlin: Robert Koch Institute. 12. Outbreaks of E. coli O104:H4 infection: update 30. Geneva: World Health Organization, 2011 (http://www.euro.who.int/en/ what-we-do/health-topics/emergencies/ international-health-regulations/news/ news/2011/07/outbreaks-of-e.-coli-o104h4infection-update-30). 13. Bell BP, Goldoft M, Griffin PM, et al. A multistate outbreak of Escherichia coli O157:H7-associated bloody diarrhea and hemolytic uremic syndrome from hamburgers: the Washington experience. JAMA 1994;272:1349-53. 14. Fukushima H, Hashizume T, Morita Y, et al. Clinical experiences in Sakai City Hospital during the massive outbreak of enterohemorrhagic Escherichia coli O157 infections in Sakai City, 1996. Pediatr Int 1999;41:213-7. 15. Guh A, Phan Q, Nelson R, et al. Outbreak of Escherichia coli O157 associated with raw milk, Connecticut, 2008. Clin Infect Dis 2010;51:1411-7. 16. Matsell DG, White CT. An outbreak of diarrhea-associated childhood hemolytic uremic syndrome: the Walkerton epidemic. Kidney Int Suppl 2009;Feb(112):S35-S37. 17. Söderström A, Osterberg P, Lindqvist A, et al. A large Escherichia coli O157 outbreak in Sweden associated with locally produced lettuce. Foodborne Pathog Dis 2008;5:339-49. 18. Wendel AM, Johnson DH, Sharapov U, et al. Multistate outbreak of Escherichia coli O157:H7 infection associated with consumption of packaged spinach, August-September 2006: the Wisconsin investigation. Clin Infect Dis 2009;48: 1079-86. 19. Mellmann A, Harmsen D, Cummings CA, et al. Prospective genomic characterization of the German enterohemorrhagic Escherichia coli O104:H4 outbreak by rapid next generation sequencing technology. PLoS ONE 2011;6(7): e22751. 20. Rohde H, Qin J, Cui Y, et al. Opensource genomic analysis of Shiga-toxin– producing E. coli O104:H4. N Engl J Med 2011;365:718-24. 21. Rasko DA, Webster DR, Sahl JW, et al. Origins of the E. coli strain causing an outbreak of hemolytic–uremic syndrome
in Germany. N Engl J Med 2011;365:70917. 22. Scheutz F, Møller Nielsen E, FrimodtMøller J, et al. Characteristics of the enteroaggregative Shiga toxin/verotoxin-producing Escherichia coli O104:H4 strain causing the outbreak of haemolytic uraemic syndrome in Germany, May to June 2011. Euro Surveill 2011;16:pii:19889. 23. European Food Safety Authority. Technical report: Shiga toxin/verotoxinproducing Escherichia coli in humans, food and animals in the EU/EEA, with special reference to the German outbreak strain STEC O104. Stockholm: European Center for Disease Prevention and Control, 2011. 24. Scavia G, Morabito S, Tozzoli R, et al. Similarity of Shiga toxin–producing Escherichia coli O104:H4 strains from Italy and Germany. Emerg Infect Dis 2011;17: 1957-8. 25. Prager R, Fruth A, Tschape H. Virulence factors of enteroaggregative E. coli (EAEC) from humans. Presented at the EHEC-Workshop 2007, Wildbad Kreuth, Germany, May 9–11, 2007. 26. Uber AP, Trabulsi LR, Irino K, et al. Enteroaggregative Escherichia coli from humans and animals differ in major phenotypical traits and virulence genes. FEMS Microbiol Lett 2006;256:251-7. 27. Gould LH, Demma L, Jones TF, et al. Hemolytic uremic syndrome and death in persons with Escherichia coli O157:H7 infection, Foodborne Diseases Active Surveillance Network sites, 2000-2006. Clin Infect Dis 2009;49:1480-5. 28. Bielaszewska M, Friedrich AW, Aldick T, Schurk-Bulgrin R, Karch H. Shiga toxin activatable by intestinal mucus in Escherichia coli isolated from humans: predictor for a severe clinical outcome. Clin Infect Dis 2006;43:1160-7. 29. Gerber A, Karch H, Allerberger F, Verweyen HM, Zimmerhackl LB. Clinical course and the role of Shiga toxin-producing Escherichia coli infection in the hemolytic-uremic syndrome in pediatric patients, 1997-2000, in Germany and Austria: a prospective study. J Infect Dis 2002; 186:493-500. 30. Mead PS, Griffin PM. Escherichia coli O157:H7. Lancet 1998;352:1207-12. Copyright © 2011 Massachusetts Medical Society.
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
original article
Glucocorticoids plus N-Acetylcysteine in Severe Alcoholic Hepatitis Eric Nguyen-Khac, M.D., Ph.D., Thierry Thevenot, M.D., Marie-Astrid Piquet, M.D., Ph.D., Saïd Benferhat, M.D., Odile Goria, M.D., Denis Chatelain, M.D., Ph.D., Blaise Tramier, M.D., François Dewaele, M.D., Salah Ghrib, M.D., Marika Rudler, M.D., Nicolas Carbonell, M.D., Hervé Tossou, M.D., Abdeslam Bental, M.D., Brigitte Bernard-Chabert, M.D., and Jean-Louis Dupas, M.D., for the AAH-NAC Study Group*
A bs t r ac t Background
Mortality among patients with severe acute alcoholic hepatitis is high, even among those treated with glucocorticoids. We investigated whether combination therapy with glucocorticoids plus N-acetylcysteine would improve survival. Methods
We randomly assigned 174 patients to receive prednisolone plus N-acetylcysteine (85 patients) or only prednisolone (89 patients). All patients received 4 weeks of prednisolone. The prednisolone–N-acetylcysteine group received intravenous N-acetylcysteine on day 1 (at a dose of 150, 50, and 100 mg per kilogram of body weight in 250, 500, and 1000 ml of 5% glucose solution over a period of 30 minutes, 4 hours, and 16 hours, respectively) and on days 2 through 5 (100 mg per kilogram per day in 1000 ml of 5% glucose solution). The prednisolone-only group received an infusion in 1000 ml of 5% glucose solution per day on days 1 through 5. The primary outcome was 6-month survival. Secondary outcomes included survival at 1 and 3 months, hepatitis complications, adverse events related to N-acetylcysteine use, and changes in bilirubin levels on days 7 and 14. Results
Mortality was not significantly lower in the prednisolone–N-acetylcysteine group than in the prednisolone-only group at 6 months (27% vs. 38%, P = 0.07). Mortality was significantly lower at 1 month (8% vs. 24%, P = 0.006) but not at 3 months (22% vs. 34%, P = 0.06). Death due to the hepatorenal syndrome was less frequent in the prednisolone– N-acetylcysteine group than in the prednisolone-only group at 6 months (9% vs. 22%, P = 0.02). In a multivariate analysis, factors associated with 6-month survival were a younger age (P<0.001), a shorter prothrombin time (P<0.001), a lower level of bilirubin at baseline (P<0.001), and a decrease in bilirubin on day 14 (P<0.001). Infections were less frequent in the prednisolone–N-acetylcysteine group than in the prednisolone-only group (P = 0.001); other side effects were similar in the two groups. Conclusions
Although combination therapy with prednisolone plus N-acetylcysteine increased 1-month survival among patients with severe acute alcoholic hepatitis, 6-month survival, the primary outcome, was not improved. (Funded by Programme Hospitalier de Recherche Clinique; AAH-NAC ClinicalTrials.gov number, NCT00863785.)
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
From Service d’Hépato-Gastroentérologie (E.N.-K., F.D., J.-L.D.) and Pathology Service (D.C.), Amiens University Hospital, and Equipe Région INSERM 24, University of Picardy (E.N.-K.) — both in Amiens; Service d’Hépatologie, Besançon University Hospital, Besançon (T.T.); Service d’Hépato-Gastroentérologie, Caen University Hospital, Caen (M.-A.P.); Service d’Hépato-Gastroentérologie, SaintQuentin General Hospital, Saint-Quentin (S.B.); Service d’Hépato-Gastroentérologie, Rouen University Hospital, Rouen (O.G.); Biostatistics Department, Aubagne General Hospital, Aubagne (B.T.); Service d’Hépato-Gastroentérologie, Cambrai General Hospital, Cambrai (S.G.); Service d’Hépato-Gastroentérologie, Pitié– Salpêtrière University Hospital (M.R.) and Service d’Hépatologie, Saint-Antoine University Hospital (N.C.) — both in Paris; Service d’Hépato-Gastroentérologie, Beauvais General Hospital, Beauvais (H.T.); Service d’Hépato-Gastroentérologie, Abbeville General Hospital, Abbeville (A.B.); and Service d’Hépato-Gastroentérologie, Reims University Hospital, Reims (B.B.-C.) — all in France. Address reprint requests to Dr. Nguyen-Khac at the Hepato-Gastroenterology Service, Amiens University Hospital, Pl. Victor Pauchet, F-80054 Amiens CEDEX 01 France, or at nguyen-khac
[email protected]. *The members of the Acute Alcoholic Hepatitis–N-Acetylcysteine (AAH-NAC) Study Group are listed in the Supplementary Appendix, available at NEJM.org. N Engl J Med 2011;365:1781-9. Copyright © 2011 Massachusetts Medical Society.
1781
The
n e w e ng l a n d j o u r na l
S
evere acute alcoholic hepatitis is a life-threatening alcoholic liver disease.1 Although glucocorticoid treatment is recommended2,3 and improves survival,4-11 mortality remains high, with 35% of patients dying within 6 months.1 Long-term alcohol consumption increases intestinal permeability, worsens endotoxemia,12 stimulates Kupffer cells,13 and thus increases production of proinflammatory cytokines.14 High levels of tumor necrosis factor α (TNF-α) activate cell-death pathways and induce the production of reactive oxygen species, notably superoxide anions, by the hepatocyte mitochondria, leading to cell death. This situation is accompanied by severe mitochondrial depletion of glutathione,15 the primary antioxidant in cells. Furthermore, hepatocytes are much more sensitive to TNF-α when their antioxidant reserves are low.16 Combination therapy with an antioxidant and glucocorticoids would have the advantage of both acting on the inflammatory process and reconstituting cellular glutathione reserves. N-acetylcysteine could have value as an antioxidant in the treatment of acute alcoholic hepatitis, because the thiol group in N-acetylcysteine is able to reduce levels of free radicals. Administration of N-acetylcysteine might reconstitute the glutathione stocks of the hepatocytes. At present, N-acetylcysteine is used in the treatment of acetaminopheninduced hepatitis. We conducted a trial to evaluate the efficacy of glucocorticoids plus N-acetylcysteine, as compared with glucocorticoids alone, in patients with severe acute alcoholic hepatitis.
Me thods Patients
The complete study protocol is available (in French) with the full text of this article at NEJM.org. From 2004 through 2009, patients hospitalized for severe acute alcoholic hepatitis at 11 French university hospitals were evaluated for eligibility. The inclusion criteria were an age of 18 years or older, an average alcohol intake of more than 50 g per day during the 3 months before enrollment, a Maddrey’s discriminant function of 32 or more, and liver histologic findings consistent with alcoholic hepatitis (Mallory bodies surrounded by polymorphonuclear neutrophils). Maddrey’s discriminant function is calculated as [4.6 × (patient’s prothrom-
1782
of
m e dic i n e
bin time − control prothrombin time, in seconds)] + serum bilirubin level, in milligrams per deciliter. A value of 32 or more clearly identifies patients with a high risk of early death.7 Key exclusion criteria were the hepatorenal syndrome, hepatocellular carcinoma, uncontrolled bacterial infection or gastrointestinal hemorrhage in the previous 4 days, infection with hepatitis C virus (HCV), hepatitis B virus (HBV), human immunodeficiency virus (HIV) infection, autoimmune hepatitis, hemochromatosis, Wilson’s disease, alpha1-antitrypsin deficiency, aceta minophen-induced hepatitis, cancer, N-acetylcysteine allergy, and serious cardiac, respiratory, or neurologic disease. Study Design
We performed a multicenter, randomized, controlled trial. Patients who met the eligibility criteria were randomly assigned to receive either prednisolone plus N-acetylcysteine or only prednisolone. Randomization was performed centrally in blocks of four by means of a computerized procedure, with stratification according to center. The treatment assignments were not concealed from the investigators or the patients. The study was conducted in compliance with the protocol. The initial evaluation included transjugular or percutaneous liver biopsy, ultrasonography of the liver, and esophageal endoscopy. The clinical examination included the recording of cardiac frequency, blood pressure, temperature, and assessment for hepatic encephalopathy, ascites, gastrointestinal hemorrhage, and jaundice. Alcohol use was evaluated with the Alcohol Use Disorders Identification Test (AUDIT)17 and the CAGE questionnaire.18 AUDIT scores, which range from 0 to 40, are determined by 10 standardized questions on the use of alcoholic beverages during the past year; a score of more than 8 indicates hazardous and harmful alcohol use. CAGE scores range from 0 to 4, and each of the letters in the acronym refers to one of the four questions; a score of 2 to 4 suggests alcohol abuse. Laboratory tests measured prothrombin time; levels of bilirubin, aspartate aminotransferase, γ-glutamyltransferase, alkaline phosphatase, albumin, creatinine, sodium, potassium, phosphorus, hemoglobin, iron, transferrin, ferritin, alpha1antitrypsin, and ceruloplasmin; platelet, white-cell, and polymorphonuclear-neutrophil counts; and
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
Glucocorticoids plus n-Acetylcysteine in Alcoholic Hepatitis
antinuclear, anti–smooth muscle, antimitochondrial, and anti–liver–kidney microsomal antibodies. Patients were tested for HBV, HCV, and HIV. Screening for bacterial infections included urine, ascites, and blood cultures, as well as chest radiography. The Child–Pugh score (which ranks the severity of cirrhosis) and Maddrey’s discriminant function were calculated. The Child–Pugh scoring system assigns 1 to 3 points for each of five variables (prothrombin time, albumin level, bilirubin level, ascites, and hepatic encephalopathy), with 3 points indicating the most severe derangement. A Child–Pugh score of 5 or 6 indicates class A disease (the least severe), 7 to 9 points class B (moderately severe), and 10 to 15 points class C (the most severe). Each investigator determined the duration of hospitalization. Patients were monitored weekly during the first month and then monthly until month 6. Each visit included a clinical examination, screening for complications (gastrointestinal hemorrhage, the hepatorenal syndrome, hepatic encephalopathy, spontaneous bacterial peritonitis, and other infections), evaluation of compliance with treatment and abstinence from alcohol consumption, laboratory tests (prothrombin time; levels of bilirubin, albumin, aspartate aminotransferase, γ-glutamyltransferase, alkaline phosphatase, and creatinine; and white-cell and polymorphonuclear-neutrophil counts), and calculation of the Child–Pugh score and Maddrey’s discriminant function. All patients were followed for 6 months or until death. The status (alive or dead) of patients lost to follow-up was assessed by telephoning a family member or by contacting the death registry at the patient’s birthplace. Study Treatments
Both groups received 40 mg of oral prednisolone per day for 28 days. For the first 5 days, patients in the prednisolone–N-acetylcysteine group received intravenous infusions of N-acetylcysteine (Fluimucil, Zambon Group). On day 1, they received 150 mg per kilogram of body weight in 250 ml of 5% glucose solution over a period of 30 minutes, 50 mg per kilogram in 500 ml of glucose solution over a period of 4 hours, and 100 mg per kilogram in 1000 ml of glucose solution over a period of 16 hours. On days 2 through 5, they received 100 mg per kilogram per day in 1000 ml of glucose solution. The patients in the
prednisolone-only group received an infusion in 1000 ml of 5% glucose solution per day on days 1 through 5. The treatment of ascites with diuretics, albumin, and sodium restriction was allowed, as was the use of beta-blockers for portal hypertension. Management of alcohol addiction was left to the individual center. The use of acetaminophen, pentoxifylline, or anti–TNF-α treatments was prohibited. All patients received normal hospital nutrition (1800 to 2000 kcal per day). Study Outcomes
The primary outcome was survival at 6 months. Prognostic factors for 6-month mortality were examined. The secondary outcomes were survival at 1 and 3 months, changes in bilirubin levels after 7 and 14 days of treatment, occurrence of hepatitis complications, and adverse events related to N-acetylcysteine use. Liver transplantation or use of the molecular adsorbent recirculating system (MARS) during the trial was treated as a mortality end point in survival analyses. Study Oversight
The study was approved by the institutional review board at Amiens University Hospital and was conducted in compliance with the French Huriet– Sérusclat legislation on medical research and the Declaration of Helsinki. All patients provided written informed consent before enrollment. (For patients with hepatic encephalopathy, informed consent was obtained from the next of kin.) Statistical Analysis
Primary and secondary outcomes were compared between the two treatment groups. Quantitative variables, expressed as means ±SD, were compared with the use of the Wilcoxon test, Kruskal–Wallis test, or Student’s t-test, as appropriate. Qualitative variables, expressed as percentages, were compared with the use of a chi-square test or Fisher’s exact test. Kaplan–Meier survival curves were plotted for up to 180 days and compared with the use of a logrank test. Factors that were significantly predictive of mortality in a univariate analysis (P<0.05) were included in a multivariate Cox logistic-regression analysis with stepwise elimination. A bilirubin decrease on day 7 or 14 was defined as a lower absolute value than on day 0. In secondary analyses that were not prespecified in the study protocol,
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1783
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
430 Patients were assessed for eligibility
250 Were excluded 161 Did not meet inclusion criteria 37 Declined to participate 52 Had other reasons
180 Underwent randomization
89 Were assigned to receive prednisolone plus N-acetylcysteine
91 Were assigned to receive prednisolone only
4 Did not receive study drug 2 Met exclusion criteria 1 Had hepatocarcinoma 1 Had lung cancer 2 Had other reasons
2 Did not receive study drug 1 Met exclusion criteria (hepatocarcinoma) 1 Had other reason
85 Received study drug and were included in the intention-to-treat population
89 Received study drug and were included in the intention-to-treat population
8 Discontinued glucocorticoids 7 Died within the first month 1 Had spontaneous bacterial peritonitis 3 Discontinued study drug after the first dose owing to a rash 11 Were lost to follow-up 2 After 14 days 9 After a median of 90 days
23 Discontinued glucocorticoids 20 Died within the first month 1 Was treated with MARS on day 27 1 Had spontaneous bacterial peritonitis 1 Had a lung infection 13 Were lost to follow-up 1 After 7 days 4 After 14 days 1 After 21 days 7 After a median of 60 days
85 Were included in the analysis
89 Were included in the analysis
Figure 1. Enrollment and Outcomes. Data on mortality were available for all patients in the intention-to-treat population, including those lost to follow-up. MARS denotes molecular adsorbent recirculating system.
we compared causes of death between the two groups and examined the effect of change in bilirubin levels over time on survival. An intermediate safety analysis was performed after 50% of the enrolled patients had completed 6 months of follow-up (with P<0.05 chosen as the threshold for statistical significance). For the final analysis, a P value of less than 0.025 was used. All the statistical analyses were performed in the modified 1784
intention-to-treat population. All reported P values are two-sided. The required sample size was calculated on the assumption that the survival rate would be 67% at 6 months in the prednisolone-only group.1 With an alpha error of 0.05, a beta error of 0.10, and a hypothetical improvement in survival of 20% at month 6 for the prednisolone–N-acetylcysteine group, the required sample size was 174 patients.
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
Glucocorticoids plus n-Acetylcysteine in Alcoholic Hepatitis
Table 1. Baseline Characteristics of the Patients.* Reference Range
Characteristic Age — yr
Prednisolone Only (N = 89)
Prednisolone– N-Acetylcysteine (N = 85)
P Value
52.2±8.5
52.8±8.7
0.66
Male sex — no. (%)
50 (56)
55 (65)
0.25
Time between hospital admission and start of treatment in trial — days
7.9±6.7
7.2±6.6
0.45 0.86
Alcohol intake — g/day
108.8±55.3
107.2±58.0
AUDIT score†
22.6±7.4
22.5±7.4
0.91
CAGE score‡
3.2±0.9
3.3±0.9
0.28 0.35
Child–Pugh score§
11.3±1.3
11.0±1.5
Maddrey’s discriminant function¶
58.2±20.0
54.0±18.2
0.11
Hepatic encephalopathy — no. (%)
39 (44)
38 (45)
0.46
Ascites — no. (%) Prothrombin time — % of normal Bilirubin — μmol/liter Albumin — g/liter Aspartate aminotransferase — U/liter γ-Glutamyltransferase — U/liter Alkaline phosphatase — U/liter Creatinine — μmol/liter White-cell count — per mm3
62 (70)
59 (69)
1.00
70–100
37.5±9.9
39.9±10.9
0.12
0–19
260.1±171.0
238.1±136.0
0.35 0.42
37–45
24.9±5.4
24.3±5.9
<35–45
109.9±49.1
127.5±83.9
0.10
0–85
223.0±240.0
308.9±289.0
0.05
0–240
176.8±161.5
174.8±83.8
0.90
45–120
73.1±21.0
73.3±29.9
0.96
10,901±5640
10,528±6534
0.69
8179±5205
7530±6254
0.48
4000–10,000
Polymorphonuclear-neutrophil count — per
mm3
1400–7500
* Plus–minus values are means ±SD. Reference ranges are from the lowest reference value used by any of the centers to the highest value used by any center. † The Alcohol Use Disorders Identification Test (AUDIT), developed by the World Health Organization, assesses excessive drinking with a standardized interview that includes 10 questions on the use of alcoholic beverages during the previous year. A score of more than 8 indicates hazardous and harmful alcohol use (score range, 0 to 40). ‡ The CAGE questionnaire includes 4 questions on use of alcoholic beverages. A score of 2 or more indicates alcohol abuse (score range, 0 to 4). § The Child–Pugh score ranks the severity of cirrhosis by assigning 1 to 3 points for each of five variables (prothrombin time, albumin level, bilirubin level, presence or absence of ascites, and presence or absence of hepatic encephalopathy), with 3 points indicating the most severe derangement. A Child–Pugh score of 5 or 6 indicates class A disease (the least severe), 7 to 9 points class B (moderately severe), and 10 to 15 points class C (the most severe). ¶ Maddrey’s discriminant function is calculated as [4.6 × (patient’s prothrombin time – control prothrombin time, in seconds)] + serum bilirubin level, in milligrams per deciliter. A value of 32 indicates severe acute alcoholic hepatitis with a high risk of early death.
R e sult s Study Population
A total of 430 patients were evaluated for eligibility (Fig. 1). After exclusion of 250 patients who did not meet the inclusion criteria or for other reasons, 180 underwent randomization. The final analysis was performed with 174 patients (73 from Amiens, 39 from Besançon, 20 from Caen, 15 from SaintQuentin, 11 from Rouen, 7 from Cambrai, 2 from Saint-Antoine, 2 from Pitié-Salpêtrière, 2 from Abbeville, 2 from Beauvais, and 1 from Reims). The
baseline characteristics of the patients did not differ significantly between the two groups, with the exception of a slightly higher γ-glutamyltrans ferase level in the prednisolone–N-acetylcysteine group (P = 0.05) (Table 1). The results of the scheduled interim analysis have been published previously.19 Mortality
In regard to the primary outcome, 57 patients had died by 6 months. The mortality rate was 38% in the prednisolone-only group (34 of 89) and 27% in
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1785
The
n e w e ng l a n d j o u r na l
1.0 0.9 0.8
Prednisolone–N-acetylcysteine
Proportion Surviving
0.7 0.6
Prednisolone only
0.5 P=0.07 by log-rank test 0.4 0.3 0.2 0.1 0.0
0
30
60
90
120
150
180
56 63
55 63
46 48
Days No. at Risk Prednisolone only 89 Prednisolone– 85 N-acetylcysteine
69 78
61 73
60 66
of
m e dic i n e
infections accounted for 9% of patients in the prednisolone-only group (8 of 89) and 8% of those in the prednisolone–N-acetylcysteine group (7 of 85) (odds ratio, 0.91; 95% CI, 0.28 to 2.93; P = 0.85), with a mean time to death of 31±22 days and 56±45 days, respectively (P = 0.05). In the prednisolone-only group, 4 patients died from septic shock, 3 from lung infections, and 1 from spontaneous bacterial peritonitis; in the prednisolone–N-acetylcysteine group, 2 patients died from septic shock, 2 from lung infections, and 1 each from subphrenic abscess, spontaneous bacterial peritonitis, and pyelonephritis. The other causes of death in the prednisolone-only group were esophageal variceal hemorrhage (1 patient), esophageal ulcer hemorrhage (1), hemorrhagic stroke (1), torsade de pointes (1), and terminal liver failure (1); in addition, 1 patient was treated with MARS on day 27. In the prednisolone–N-acetylcysteine group, the other causes of death were esophageal variceal hemorrhage (5 patients) (P = 0.11), hemorrhagic stroke (1), and terminal liver failure (1); in addition, 1 patient underwent liver transplantation on day 170. Adverse Events
Figure 2. Kaplan–Meier Curves for 6-Month Survival in the Intention-toTreat Population.
At 6 months, the rate of the hepatorenal syndrome was 25% in the prednisolone-only group (22 of 89 patients) and 12% in the prednisolone– N-acetylcysteine group (10 of 85) (odds ratio with combination therapy, 0.41; 95% CI, 0.17 to 0.98; P = 0.02) (Table 2). The overall rate of infection was 42% in the prednisolone-only group (37 of 89 patients) and 19% in the prednisolone–N-acetylcysteine group (16 of 85) (odds ratio, 0.33; 95% CI, 0.15 to 0.68; P = 0.001). The two groups did not differ significantly with respect to other complications. Among patients with relapse of alcohol use after 1 month, 13% died in the prednisolone-only group (2 of 15) versus 7% in the prednisolone–Nacetylcysteine group (1 of 15) (P = 1.00).
the prednisolone–N-acetylcysteine group (23 of 85) (hazard ratio with combination therapy, 0.62; 95% confidence interval [CI], 0.37 to 1.06; P = 0.07) (Fig. 2). The mean time to death was 40±35 days (median, 27; range, 3 to 149) in the prednisolone-only group and 54±37 days (median, 50.5; range, 5 to 149) in the prednisolone–N-acetylcysteine group (P = 0.14). In regard to the secondary outcomes, the respective mortality rates in the prednisoloneonly and prednisolone–N-acetylcysteine groups were 24% (21 of 89) and 8% (7 of 85) at 1 month (hazard ratio, 0.58; 95% CI, 0.14 to 0.76; P = 0.006) Predictive Factors for Death and 34% (30 of 89) and 22% (19 of 85) at 3 months At 6 months, nine factors were significantly associ(hazard ratio, 0.33; 95% CI, 0.33 to 1.04; P = 0.06). ated with mortality in a univariate analysis (Table 3). Age, hepatic encephalopathy, prothrombin time, Causes of Death baseline bilirubin level, baseline creatinine level, At 6 months, 22% of the patients in the prednis- Maddrey’s discriminant function, Child–Pugh olone-only group (20 of 89) had died of the hepa- score, change from baseline in the bilirubin level on torenal syndrome, versus 9% of the patients in the day 7, change from baseline in the bilirubin level prednisolone–N-acetylcysteine group (8 of 85) on day 14, and treatment group were all included in (odds ratio, 2.79; 95% CI, 1.08 to 7.42; P = 0.02); a Cox model for multivariate analysis. Variables the mean time to death was 36.5±28 days and independently associated with increased mortality 66±33 days, respectively (P = 0.30). Deaths due to were older age (odds ratio, 1.07; 95% CI, 1.03 to 1786
n engl j med 365;19
nejm.org
november 10, 2011
The New England Journal of Medicine
Glucocorticoids plus n-Acetylcysteine in Alcoholic Hepatitis
1.11; P<0.001), prolonged prothrombin time (odds ratio, 0.93; 95% CI, 0.91 to 0.97; P<0.001), higher baseline bilirubin level (odds ratio, 1.007; 95% CI, 1.005 to 1.009; P<0.001), and the absence of a decrease in the bilirubin level on day 14 (odds ratio, 0.23; 95% CI, 0.12 to 0.45; P<0.001). Change over Time in Bilirubin Level and Effect on Survival
The change in the median bilirubin level in each group over the course of the study is shown in Figure 1S in the Supplementary Appendix, available at NEJM.org. The difference was significant only on day 14 (160 μmol per liter in the prednisolone-only group vs. 115 μmol per liter in the prednisolone–N-acetylcysteine group, P = 0.003) (Table 1S in the Supplementary Appendix). For the whole study population, data on the bilirubin level on day 7 were available for 159 patients (2 patients had died, and data were missing for 13 other patients). A bilirubin decrease on day 7 was observed in 128 patients (81%). The 6-month survival rate was 90% among patients with a bilirubin decrease on day 7 and 10% among patients without a decrease (odds ratio, 78; 95% CI, 29 to 210; P<0.001). On day 14, data on the bilirubin level were available for 143 patients (7 patients had died, and data were missing for 24 other patients). A bilirubin decrease on day 14 was observed in 80% of the patients. The 6-month survival rate was 89% among patients with a bilirubin decrease on day 14 and 11% among patients without a decrease (odds ratio, 61; 95% CI, 23 to 166; P<0.001). On day 7, 78% of patients in the prednisolone-only group had a bilirubin decrease, versus 81% in the prednisolone–N-acetylcysteine group (odds ratio, 1.38; 95% CI, 0.59 to 3.29; P = 0.41). On day 14, 74% of patients in the prednisolone-only group had a bilirubin decrease, versus 87% in the prednisolone– N-acetylcysteine group (odds ratio, 2.47; 95% CI, 0.96 to 6.49; P = 0.04).
Discussion In patients with severe acute alcoholic hepatitis, the combination of N-acetylcysteine and prednisolone did not significantly improve 6-month survival, as compared with prednisolone only. The rationale for the use of antioxidants in the treatment of acute alcoholic hepatitis is based on the pivotal role of oxidative stress in the disorder. Liver protection by N-acetylcysteine has been shown in mouse models of acute and chronic alcoholic hepatitis.20-22 How-
Table 2. Adverse Events.
Event
Prednisolone Prednisolone– Only N-Acetylcysteine (N = 89) (N = 85) no. (%)
Hepatorenal syndrome*
22 (25)
10 (12)
All infections†
37 (42)
16 (19)
Spontaneous bacterial peritonitis
9 (10)
5 (6)
Lung infection
8 (9)
3 (4)
Urinary system infection
7 (8)
4 (5)
Erysipelas
3 (3)
0
Staphylococcal septicemia
2 (2)
2 (2)
Gram-negative septicemia
2 (2)
2 (2)
Esophageal candidiasis
2 (2)
0
Otitis externa
1 (1)
0
Infection of unknown cause Esophageal variceal hemorrhage
3 (3) 8 (9)
0 10 (12)
Esophageal ulcer hemorrhage
1 (1)
0
Gastroduodenal ulcer hemorrhage
0
3 (4)
Umbilical variceal hemorrhage
0
1 (1)
Thigh hematoma
0
1 (1)
Hepatic encephalopathy
4 (4)
3 (4)
Hepatic hydrothorax‡
0
1 (1)
Functional kidney failure
0
1 (1)
Inguinal hernia occlusion
1 (1)
0
Lumbar vertebral collapse
1 (1)
1 (1)
Cerebral hemorrhage
2 (2)
1 (1)
Cancer of the pharynx
1 (1)
0
Uterine fibroid hemorrhage
1 (1)
0
Calf hematoma
1 (1)
0
Torsade de pointes
1 (1)
0
Rash after first injection of N-acetylcysteine
0
3 (4)
MARS use§
1 (1)
0
Liver transplantation§
0
1 (1)
Relapse of alcohol use after 1 month
15 (17)
15 (18)
* P = 0.02 for the between-group comparison. † P = 0.001 for the between-group comparison. ‡ Hepatic hydrothorax was related to portal hypertension. § Liver transplantation or use of the molecular adsorbent recirculating system (MARS) was considered a primary-outcome failure.
ever, in patients with severe acute alcoholic hepatitis, the benefits of antioxidants have not been shown. In a randomized trial, a combination of antioxidants that included N-acetylcysteine was significantly worse than prednisolone with respect to survival.23 Similarly, in a study with a complex de-
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1787
n e w e ng l a n d j o u r na l
The
of
m e dic i n e
Table 3. Factors Associated with Mortality at 6 Months (Univariate Analysis).* Factor
Status at 6 Mo
Age — yr Prothrombin time — % of normal Bilirubin — μmol/liter
P Value
Dead
Alive
55.6±6.9
50.9±8.9
<0.001
35.2±10.5
40.3±10.1
0.003
330.4±182.1
209.9±122.8
<0.001
Creatinine — μmol/liter
80.8±24.6
69.4±25.4
0.005
Hepatic encephalopathy — no./total no. (%)
18/57 (32)
19/116 (16)
0.03
Maddrey’s discriminant function
66.3±17.8
51.2±17.9
<0.001
Child–Pugh score
11.5±1.3
10.9±1.4
0.01
Bilirubin decrease on day 7 — no./total no. (%)
31/51 (61)
97/108 (90)
<0.001
Bilirubin decrease on day 14 — no./total no. (%)
29/46 (63)
86/97 (89)
0.001
* Plus–minus values are means ±SD.
sign, another N-acetylcysteine-containing antioxidant regimen (with or without glucocorticoids) did not improve 6-month survival.24 Finally, a randomized trial showed that use of N-acetylcysteine for 14 days did not confer any survival benefit, as compared with oral nutritional support.25 Although there was no significant difference in survival at 6 months between our study groups, there was a short-term survival benefit at 1 month with prednisolone–N-acetylcysteine as compared with prednisolone only. We used N-acetylcysteine because it has antioxidant properties,26 decreases levels of free radicals, increases glutathione levels,27 and represses the expression of nuclear factor κB and TNF-α.28 The dose, duration, and administration route used were the same as those used for the treatment of drug intoxication29,30 and the hepatorenal syndrome.31 At 3 and 6 months, we observed a lower mortality rate in the prednisolone–N-acetylcysteine group than in the prednisolone-only group, but the differences were not significant. These findings may be related to a lack of power. It is also possible that 5 days of N-acetylcysteine was not enough. A longer period of intravenous N-acetylcysteine combined with prednisolone could perhaps be considered, with subsequent oral administration of N-acetylcysteine until 1 month. The improvement in short-term survival that we observed in our study could be linked, at least in part, to a reduced risk of the hepatorenal syndrome in the prednisolone–N-acetylcysteine group. In a study involving 12 patients with the hepato renal syndrome, the survival rate at 1 month after 1788
N-acetylcysteine infusion was unexpectedly high (67%).31 In regard to morbidity, the prednisolone– N-acetylcysteine group in our study had significantly fewer infectious complications than the prednisolone-only group. It has been shown that patients with severe alcoholic hepatitis who do not have a response to treatment have significantly more bacterial infections than patients who have a response.32 Alternatively, N-acetylcysteine could have beneficial effects by increasing blood flow to the liver, improving liver function, increasing the cardiac index,33 and decreasing the intrahepatic lactate levels seen in patients with septic shock.34 Our study provides prospective validation of the finding that a decrease in the bilirubin level after 7 days of treatment is associated with a favorable prognosis35 and also shows that a decrease on day 14 is associated with increased survival. However, in a multivariate analysis, only a decrease in the bilirubin level on day 14 remained significant, and the decrease was more frequent in the prednisolone–N-acetylcysteine group than in the prednisolone-only group. Thus, for a 5-day course of prednisolone–N-acetylcysteine therapy, therapeutic efficacy is better evaluated on day 14 than on day 7. In conclusion, we observed improved survival at 1 month among patients with severe acute alcoholic hepatitis who received combination therapy with prednisolone and N-acetylcysteine, as compared with those who received prednisolone only, but 6-month mortality, our primary outcome, was not improved with combination therapy.
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
Glucocorticoids plus n-Acetylcysteine in Alcoholic Hepatitis Supported by Programme Hospitalier de Recherche Clinique, a program of the French government that supports hospital-based clinical research. No potential conflict of interest relevant to this article was reported.
Disclosure forms provided by the authors are available with the full text of this article at NEJM.org. We thank Prof. Thierry Poynard for helpful comments on the manuscript.
References 1. Mathurin P, Mendenhall CL, Carithers
RL Jr, et al. Corticosteroids improve shortterm survival in patients with severe alcoholic hepatitis (AH): individual data analysis of the last three randomized placebo controlled double blind trials of corticosteroids in severe AH. J Hepatol 2002;36: 480-7. 2. Imperiale TF, O’Connor JB, McCullough AJ. Corticosteroids are effective in patients with severe alcoholic hepatitis. Am J Gastroenterol 1999;94:3066-8. 3. O’Shea RS, Dasarathy S, McCullough AJ. Alcoholic liver disease. Hepatology 2010;51:307-28. 4. Helman RA, Temko MH, Nye SW, Fallon HJ. Alcoholic hepatitis: natural history and evaluation of prednisolone therapy. Ann Intern Med 1971;74:311-21. 5. Maddrey WC, Boitnott JK, Bedine MS, Weber FL Jr, Mezey E, White RI Jr. Corticosteroid therapy of alcoholic hepatitis. Gastroenterology 1978;75:193-9. 6. Mendenhall CL, Anderson S, GarciaPont P, et al. Short-term and long-term survival in patients with alcoholic hepatitis treated with oxandrolone and prednisolone. N Engl J Med 1984;311:1464-70. 7. Carithers RL Jr, Herlong HF, Diehl AM, et al. Methylprednisolone therapy in patients with severe alcoholic hepatitis: a randomized multicenter trial. Ann Intern Med 1989;110:685-90. 8. Ramond MJ, Poynard T, Rueff B, et al. A randomized trial of prednisolone in patients with severe alcoholic hepatitis. N Engl J Med 1992;326:507-12. 9. Imperiale TF, McCullough AJ. Do corticosteroids reduce mortality from alcoholic hepatitis? A meta-analysis of the randomized trials. Ann Intern Med 1990; 113:299-307. 10. Daures JP, Peray P, Bories P, et al. Corticoid therapy in the treatment of acute alcoholic hepatitis: results of a meta-analysis. Gastroenterol Clin Biol 1991;15:2238. (In Spanish.) 11. Rambaldi A, Saconato HH, Christensen E, Thorlund K, Wetterslev J, Gluud C. Systematic review: glucocorticosteroids for alcoholic hepatitis — a Cochrane Hepato-Biliary Group systematic review with meta-analyses and trial sequential analyses of randomized clinical trials. Aliment Pharmacol Ther 2008;27:1167-78. 12. Mathurin P, Deng QG, Keshavarzian A, Choudhary S, Holmes EW, Tsukamoto H. Exacerbation of alcoholic liver injury by enteral endotoxin in rats. Hepatology 2000;32:1008-17. 13. Enomoto N, Ikejima K, Bradford B, et
al. Alcohol causes both tolerance and sensitization of rat Kupffer cells via mechanisms dependent on endotoxin. Gastroenterology 1998;115:443-51. 14. Bird GL, Sheron N, Goka AK, Alexander GJ, Williams RS. Increased plasma tumor necrosis factor in severe alcoholic hepatitis. Ann Intern Med 1990;112:917-20. 15. Hirano T, Kaplowitz N, Tsukamoto H, Kamimura S, Fernandez-Checa JC. Hepatic mitochondrial glutathione depletion and progression of experimental alcoholic liver disease in rats. Hepatology 1992; 16:1423-7. 16. Colell A, García-Ruiz C, Miranda M, et al. Selective glutathione depletion of mitochondria by ethanol sensitizes hepatocytes to tumor necrosis factor. Gastroenterology 1998;115:1541-51. 17. Reinert DF, Allen JP. The Alcohol Use Disorders Identification Test: an update of research findings. Alcohol Clin Exp Res 2007;31:185-99. 18. Dhalla S, Kopec JA. The CAGE questionnaire for alcohol misuse: a review of reliability and validity studies. Clin Invest Med 2007;30:33-41. 19. Nguyen-Khac E, Thevenot T, Piquet MA, et al. Treatment of severe acute alcoholic hepatitis (AAH) with corticoids plus n-acetyl cysteine (C-NAC) versus corticoids (C): planed interim analysis of a multicentre, controlled, randomized trial. J Hepatol 2008;48:Suppl 2:S17. 20. Wang AL, Wang JP, Wang H, et al. A dual effect of N-acetylcysteine on acute ethanol-induced liver damage in mice. Hepatol Res 2006;34:199-206. 21. Ronis MJ, Butura A, Sampey BP, et al. Effects of N-acetylcysteine on ethanolinduced hepatotoxicity in rats fed via total enteral nutrition. Free Radic Biol Med 2005;39:619-30. 22. Ozaras R, Tahan V, Aydin S, Uzun H, Kaya S, Senturk H. N-acetylcysteine attenuates alcohol-induced oxidative stress in rats. World J Gastroenterol 2003;9:791-4. 23. Phillips M, Curtis H, Portmann B, Donaldson N, Bomford A, O’Grady J. Antioxidants versus corticosteroids in the treatment of severe alcoholic hepatitis — a randomised clinical trial. J Hepatol 2006;44:784-90. 24. Stewart S, Prince M, Bassendine M, et al. A randomized trial of antioxidant therapy alone or with corticosteroids in acute alcoholic hepatitis. J Hepatol 2007;47:27783. 25. Moreno C, Langlet P, Hittelet A, et al. Enteral nutrition with or without N-acetyl cysteine in the treatment of severe acute
alcoholic hepatitis: a randomized multicenter controlled trial. J Hepatol 2010;53: 1117-22. 26. Aruoma OI, Halliwell B, Hoey BM, Butler J. The antioxidant action of N-acetyl cysteine: its reaction with hydrogen peroxide, hydroxyl radical, superoxide, and hypochlorous acid. Free Radic Biol Med 1989;6:593-7. 27. Neuschwander-Tetri BA, Bellezzo JM, Britton RS, Bacon BR, Fox ES. Thiol regulation of endotoxin-induced release of tumour necrosis factor alpha from isolated rat Kupffer cells. Biochem J 1996; 320:1005-10. 28. Verhasselt V, Vanden Berghe W, Vanderheyde N, Willems F, Haegeman G, Goldman M. N-acetyl-L-cysteine inhibits primary human T cell responses at the dendritic cell level: association with NFkappaB inhibition. J Immunol 1999;162: 2569-74. 29. Prescott LF, Park J, Ballantyne A, Adriaenssens P, Proudfoot AT. Treatment of paracetamol (acetaminophen) poisoning with N-acetylcysteine. Lancet 1977; 2:432-4. 30. Lee WM, Hynan LS, Rossaro L, et al. Intravenous N-acetylcysteine improves transplant-free survival in early stage non-acetaminophen acute liver failure. Gastroenterology 2009;137:856-64. 31. Holt S, Goodier D, Marley R, et al. Improvement in renal function in hepatorenal syndrome with N-acetylcysteine. Lancet 1999;353:294-5. 32. Louvet A, Wartel F, Castel H, et al. Infection in patients with severe alcoholic hepatitis treated with steroids: early response to therapy is the key factor. Gastroenterology 2009;137:541-8. 33. Rank N, Michel C, Haertel C, et al. N-acetylcysteine increases liver blood flow and improves liver function in septic shock patients: results of a prospective, randomized, double-blind study. Crit Care Med 2000;28:3799-807. 34. Hein OV, Ohring R, Schilling A, et al. N-acetylcysteine decreases lactate signal intensities in liver tissue and improves liver function in septic shock patients, as shown by magnetic resonance spectroscopy: extended case report. Crit Care 2004;8:R66-R71. 35. Mathurin P, Abdelnour M, Ramond MJ, et al. Early change in bilirubin levels is an important prognostic factor in severe alcoholic hepatitis treated with prednisolone. Hepatology 2003;38:1363-9. Copyright © 2011 Massachusetts Medical Society.
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1789
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
original article
Early Liver Transplantation for Severe Alcoholic Hepatitis Philippe Mathurin, M.D., Ph.D., Christophe Moreno, M.D., Ph.D., Didier Samuel, M.D., Ph.D., Jérôme Dumortier, M.D., Ph.D., Julia Salleron, M.S., François Durand, M.D., Ph.D., Hélène Castel, M.D., Alain Duhamel, M.D., Ph.D., Georges-Philippe Pageaux, M.D., Ph.D., Vincent Leroy, M.D., Ph.D., Sébastien Dharancy, M.D., Ph.D., Alexandre Louvet, M.D., Ph.D., Emmanuel Boleslawski, M.D., Ph.D., Valerio Lucidi, M.D., Thierry Gustot, M.D., Ph.D., Claire Francoz, M.D., Christian Letoublon, M.D., Denis Castaing, M.D., Jacques Belghiti, M.D., Vincent Donckier, M.D., Ph.D., François-René Pruvot, M.D., and Jean-Charles Duclos-Vallée, M.D., Ph.D.
A BS T R AC T Background From Hôpital Claude Huriez, Services Maladies de l’Appareil Digestif and INSERM Unité 995 (P.M., H.C., S.D., A.L.), Services de Chirurgie Digestive et de Transplantation (E.B., F.-R.P.), and the Department of Biostatistics and INSERM EA2694 (J.S., A.D.), Centre Hospitalier Universitaire (CHU) de Lille and Université Nord de France, Lille; Assistance Publique– Hôpitaux de Paris Hôpital Paul Brousse, Centre Hépato-Biliaire and INSERM Unité 785, Université Paris-Sud — both in Villejuif (D.S., D.C., J.-C.D.-V.); Hôpital Edouard Herriot, Unité de Transplantation Hépatique, Lyon (J.D.); Hôpital Beaujon, Hepatology and Liver Intensive Care, INSERM Unité 773 (F.D., C.F.), and Hepatobiliary and Pancreatic Surgery (J.B.), University Denis Diderot–Paris VII, Clichy; Hôpital Saint Eloi, Service d’HépatoGastroentérologie et Transplantation Hépatique, Montpellier (G.-P.P.); and Hôpital Albert Michallon, Service d’Hépatogas troentérologie and INSERM Unité 823 (V. Leroy), and Service de Transplantation Hépatique (C.L.), CHU de Grenoble, Gre noble — all in France; and the Departments of Gastroenterology, Hepatopancreatology, and Digestive Oncology (C.M., T.G.), and Hepatobiliary Surgery and Liver Transplantation (V. Lucidi, V.D.), Erasme Hospital, Université Libre de Bruxelles, Brussels. Address reprint requests to Dr. Mathurin at Hôpital Claude Huriez, Services Maladies de l’Appa reil Digestif et INSERM Unité 995 CHU, Université Lille Nord de France, F-59000 Lille, France, or at
[email protected]. N Engl J Med 2011;365:1790-1800. Copyright © 2011 Massachusetts Medical Society.
1790
A 6-month abstinence from alcohol is usually required before patients with severe alcoholic hepatitis are considered for liver transplantation. Patients whose hepatitis is not responding to medical therapy have a 6-month survival rate of approximately 30%. Since most alcoholic hepatitis deaths occur within 2 months, early liver transplantation is attractive but controversial. Methods
We selected patients from seven centers for early liver transplantation. The patients had no prior episodes of alcoholic hepatitis and had scores of 0.45 or higher according to the Lille model (which calculates scores ranging from 0 to 1, with a score ≥0.45 indicating nonresponse to medical therapy and an increased risk of death in the absence of transplantation) or rapid worsening of liver function despite medical therapy. Selected patients also had supportive family members, no severe coexisting conditions, and a commitment to alcohol abstinence. Survival was compared between patients who underwent early liver transplantation and matched patients who did not. Results
In all, 26 patients with severe alcoholic hepatitis at high risk of death (median Lille score, 0.88) were selected and placed on the list for a liver transplant within a median of 13 days after nonresponse to medical therapy. Fewer than 2% of patients admitted for an episode of severe alcoholic hepatitis were selected. The centers used 2.9% of available grafts for this indication. The cumulative 6-month survival rate (±SE) was higher among patients who received early transplantation than among those who did not (77±8% vs. 23±8%, P<0.001). This benefit of early transplantation was maintained through 2 years of follow-up (hazard ratio, 6.08; P = 0.004). Three patients resumed drinking alcohol: one at 720 days, one at 740 days, and one at 1140 days after transplantation. Conclusions
Early liver transplantation can improve survival in patients with a first episode of severe alcoholic hepatitis not responding to medical therapy. (Funded by Société Nationale Française de Gastroentérologie.) n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
Early Liver Tr ansplantation for Alcoholic Hepatitis
L
iver transplantation for alcoholic liver disease has a favorable outcome but remains controversial.1,2 Reluctance to perform transplantation in patients with alcoholism is often based on the view that they are responsible for their illness and are likely to resume alcohol use after transplantation.3 To select the most appropriate patients with severe forms of alcoholic liver disease for transplantation, most programs require a 6-month abstinence period before patients can be considered. Nevertheless, data regarding the 6-month rule as a predictor of long-term sobriety are controversial.4 Despite the frequent use of the rule, the United Network for Organ Sharing and the French Consensus Conference5 do not consider it to be a formal guideline. Glucocorticoids have been recommended by recent U.S. guidelines to treat severe alcoholic hepatitis.6 The Lille model enables early identification of patients unlikely to respond to medical treatment.7,8 Strict application of the rule requiring 6 months of sobriety 9,10 may be disadvantageous to such patients, 70 to 80% of whom die within that period. Consequently, early liver transplantation should be evaluated in pilot studies restricted to carefully selected patients, as recommended by the latest French consensus.5 The aims of this study were to determine whether early liver transplantation improves the 6-month survival rate among patients whose severe alcoholic hepatitis is unresponsive to medical management, to evaluate the rate of alcohol relapse after transplantation in patients selected without applying the 6-month rule, and to evaluate the burden of early transplantation on the overall transplantation activity of participating centers.
Me thods Study Conduct
Scientific committees from the Association Française d’Etude du Foie, the Association Française de Chirurgie Hépato-Biliaire et de Transplantation, and the Agence de Biomédecine (the French government agency in charge of graft allocation) approved the study. All selected patients provided written informed consent for transplantation surgery. Selection of the Study Patients
Seven transplant centers agreed to perform early liver transplantation in patients with severe alco-
holic hepatitis not responding to medical therapy. Alcoholic hepatitis was considered to be severe if the Maddrey’s discriminant function was greater than 32, calculated as follows: 4.6 × (patient’s prothrombin time in seconds − matched control’s prothrombin time in seconds) + patient’s serum bilirubin level in milligrams per deciliter. A Maddrey’s discriminant function of greater than 32 is the threshold for initiating glucocorticoid treatment.6 Nonresponse to medical therapy was defined according to the Lille model as a score of 0.45 or more after 7 days of medical therapy or a continuous increase in the Model for End-Stage Liver Disease (MELD) score,11 reflecting an early worsening of liver function. Medical therapy consisted of standard medical care for severe liver insufficiency and use of glucocorticoids (40 mg per day of prednisolone for at least 7 days). Nonresponse to medical therapy is associated with 6-month survival of approximately 30%.7 We selected patients with severe alcoholic hepatitis who were considered to be candidates for early transplantation, according to the following criteria: nonresponse to medical therapy (as defined above), severe alcoholic hepatitis as the first liver-decompensating event, presence of close supportive family members, absence of severe coexisting or psychiatric disorders, and agreement by patients (with support from family members) to adhere to lifelong total alcohol abstinence. The selection process consisted of several meetings between four medical team circles, the patient’s family, and the patient. The team circles were as follows: first, the inner circle, closest to the patient, comprising nurses, one resident, and one fellow; second, a specialist in addiction; third, senior hepatologists; and fourth, the outermost circle, consisting of an anesthetist and surgeons. The four team circles had to reach complete consensus on selection. This selection process was performed at all seven participating centers, which provided data on all early transplantations performed until September 1, 2010. The Lille and Brussels centers started the program in November 2005; the others began later. Two study patients did not receive glucocorticoids because their physicians considered the benefits to be negligible. Additional details about the selection process are provided in the Supplementary Appendix, available with the full text of this article at NEJM.org.
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1791
n e w e ng l a n d j o u r na l
The
of
m e dic i n e
Assessment of Alcohol Use after Transplantation
after transplantation), we recorded data on all infections and subsequent treatment as well as After transplantation, alcohol use was assessed at any alcohol relapses, including those occurring short intervals during informal interviews of pa- more than 6 months after transplantation. tients and their families performed according to the design of previous studies.10,12,13 (See the Sup- Burden of Early Liver Transplantation plementary Appendix for additional details on as- The total number of transplantations and the sessment of alcohol use after transplantation.) number of transplantations for alcoholic liver disease were reported annually by all centers, Data Collection starting with the first early transplantation and Development of biologic features of abnormal until September 1, 2010. Only the Lille and Brusliver or kidney function was ascertained at least sels centers had prospective databases of patients weekly from the first day of medical therapy until with biopsy-proven severe alcoholic hepatitis. the patient was placed on the transplantation These databases systematically recorded data for list. During the data collection period (i.e., the all patients with severe alcoholic liver disease first day of medical therapy through 24 months meeting the criteria for nonresponse to medical Table 1. Characteristics of the 26 Study Patients with Severe Alcoholic Hepatitis.* Characteristic
Value
Male sex — no. (%)
15 (58)
Age — yr Median
47.4
Range
34.9–60.5
Glucocorticoid therapy — no. (%)
24 (92)
Duration of glucocorticoid therapy — days Median
12
Range
7–31
First day of medical therapy Prothrombin time — sec Median
23.2
Range
14.3–82.0
International normalized ratio Median
2.5
Range
1.3–5.8
Bilirubin — mg/dl Median
27.7
Range
10.0–61.3
Serum creatinine — mg/dl Median
0.82
Range
0.47–6.10
Albumin — g/liter Median
25
Range
14–33
Maddrey’s discriminant function† Median
76
Range
36–165
MELD score‡
1792
Median
30.1
Range
22.0–47.3
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
Early Liver Tr ansplantation for Alcoholic Hepatitis
Table 1. (Continued.) Characteristic
Value
After 7 days of medical therapy Prothrombin time — sec Median
33.5
Range
11.0–56.0
International normalized ratio Median
2.2
Range
1.3–4.7
Bilirubin — mg/dl Median
28.4
Range
9.6–48.7
Serum creatinine — mg/dl Median
0.96
Range
0.48–4.95
Albumin — g/liter Median
27
Range
18–37
Lille score‡ Median
0.880
Range
0.260–0.996
MELD score§ Median
28.5
Range
23.0–52.4
* To convert the values for serum creatinine to micromoles per liter, multiply by 88.4. To convert the values for bilirubin to micromoles per liter, multiply by 17.1. † Maddrey’s discriminant function is a measure of severity of alcoholic hepatitis and is calculated as follows: 4.6 × (patient’s prothrombin time in seconds − matched control’s prothrombin time in seconds) + patient’s serum bilirubin level in milligrams per deciliter. A score of more than 32 indicates severe alcoholic hepatitis and is the threshold for initiating glucocorticoid treatment. ‡ A Lille score of 0.45 or higher indicates nonresponse to medical therapy. The Lille score ranges from 0 to 1 and is calculated (www.lillemodel.com) with the use of the following formula: Exp (−R)/(1 + Exp [−R]), where R = (3.19 − 0.101 × age in years) + (0.147 × albumin on day 0 in grams per liter) + ([0.0165 × change in bilirubin between day 0 and day 7 of medical therapy, in micromoles per liter] − [0.206 × renal insufficiency {rated as 0 if absent and 1 if present}] − [0.0065 × bilirubin level on day 0 in micromoles per liter] − [0.0096 × prothrombin time in seconds]). In patients who received albumin infusions, the last available albumin value before infusion was used. § The Model for End-Stage Liver Disease (MELD) score is calculated (www.mayoclinic.org/meld/mayomodel7.html) as follows: (9.57 × log creatinine in milligrams per deciliter) + (3.78 × log bilirubin in milligrams per deciliter) + (11.20 × log international normalized ratio) + 6.43. Scores range from 6 to 42, with higher scores indicating a worse prognosis.
therapy and, for those not selected for early trans- discriminant function, and Lille score (see the plantation, the primary reason for exclusion. Supplementary Appendix). Second, we randomly sampled control patients Case–Control Study from a set of patients with severe alcoholic hepaTwo patients whose disease was not responding titis who were listed in a combined prospective to medical therapy — a case patient who under- database of the Lille center. To avoid the risk of went early transplantation and a control patient selecting the same control in the two matching who did not — were matched with the use of two procedures (nonrandom and random), we excluded matching-selection processes. First, we nonran- all the matched controls who were already sedomly selected the control patient who was the lected with the nonrandom matching procedure. best fit for each case patient who underwent The final combined database contained a total transplantation, according to age, sex, Maddrey’s of 651 potential control patients. The random
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1793
The
n e w e ng l a n d j o u r na l
selection was performed by means of the global optimal algorithm14 (SAS software, version 9.2; SAS Institute) with the following preestablished ranges or values: age, ±10 years of the case patient’s age; sex, same as the case patient; Maddrey’s discriminant function, same category as the case patient’s (<60, 60 to 90, or >90); and Lille score, ±0.15 of the case patient’s score (see the Supplementary Appendix). We also examined whether patients who underwent early liver transplantation owing to nonresponse to medical therapy had outcomes similar to patients whose disease had responded to medical therapy (i.e., those with a Lille score <0.45). To address this question, control patients with Lille scores less than 0.45 were matched to case patients on the basis of the global optimal algorithm, with terms for age, sex, and Maddrey’s discriminant function (but not Lille score). Statistical Analysis
Assuming 6-month survival rates of 70% among patients who underwent transplantation and 30% among the matched controls,15,16 we calculated that at least 18 patients and 18 controls would have to be included for the study to have a statistical power of at least 80% to show a significant difference in the survival rate between the two groups. Variables were compared between the two groups with the use of chi-square tests and t-tests. The follow-up time was defined as the period from the first day of medical therapy to the last follow-up visit. In the case–control study, we estimated patients’ rate of survival (expressed as a percentage ±SE) by means of the Kaplan–Meier method and compared survival between the two groups by using the log-rank test. The first day of medical therapy was defined as the first day of glucocorticoid administration or, for the two patients not treated with glucocorticoids, the first day of admission. The rate of 6-month survival — the primary end point — was measured from the first day of medical therapy to the date of death from any cause. In addition, we performed an analysis over an extended follow-up period of 2 years. Data for patients without events of interest were censored at the date of the last follow-up visit. A Cox proportional-hazards analysis, after adjustment for the MELD score and center (Lille vs. other centers), was also performed. All P values are two-tailed. 1794
of
m e dic i n e
R e sult s Characteristics of the Study Population
Twenty-six patients with severe alcoholic hepatitis that had failed to respond to medical management underwent liver transplantation at the seven centers (Table 1). Alcoholic hepatitis was proved by biopsy specimens obtained by the transjugular route in 23 of the 26 case patients (88%) and confirmed in all case patients by means of histologic analysis of explants. The first transplantation was performed under this selection process on August 24, 2006, and the final one was conducted on June 16, 2010. Nonresponse to medical therapy was defined as a Lille score of 0.45 or more in 25 patients and early worsening of liver function in 1 patient whose MELD score increased from 23 at day 7 to 36 at day 21, the day of listing (albeit whose Lille score was 0.26). A total of 24 of 26 patients were treated with glucocorticoids for 12 days (95% confidence interval [CI], 7 to 18). The remaining 2 patients did not receive glucocorticoids before referral to the transplantation unit. One had type 1 hepatorenal syndrome, and both had high Lille scores (0.88 and 0.66) within 7 days after hospitalization and before transfer to the transplantation center. For these 2 patients, the transplantation physicians believed that glucocorticoid therapy would have a negligible effect.17 MELD Scores and Transplantation
The median MELD score at the time of listing as a candidate for transplantation was 34 (Table 2). The decision to list patients for transplantation was made soon after ascertainment of nonresponse to medical therapy (median, 13 days), and transplantation was performed soon after listing (median, 9 days afterward). Nonresponse to medical therapy was confirmed by the worsening of MELD scores until the time of listing. Substantial improvement before the time of listing (i.e., >10% decrease in the MELD score) was observed in three patients who underwent transplantation, after treatment with hepatic or renal support, which decreased bilirubin and creatinine levels. Survival
Liver grafts were obtained from cadaveric donors (see the Supplementary Appendix). Patients who underwent transplantation and controls were well-matched (Table 3). The 6-month survival rate
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
Early Liver Tr ansplantation for Alcoholic Hepatitis
was significantly higher among patients undergoing transplantation (77±8%) than among matched controls (23±8%, P<0.001) (Fig. 1). The benefit of performing early transplantation was supported by the fact that 90% of deaths (18 of 20) among control patients occurred within 2 months after identification of nonresponse to medical therapy (Fig. 1). After transplantation, five of six deaths were related to infection occurring within 2 weeks after surgery; the infection was invasive aspergillus infection in four cases. Cerebral aspergillosis developed in one patient who had undergone transplantation and survived; it was rapidly treated with voriconazole. The patients undergoing transplantation who died did not differ significantly from those who remained alive, although the duration of glucocorticoid therapy was longer (see the Supplementary Appendix). Additional analysis with extension of follow-up to 2 years showed that transplantation remained associated with survival in univariate analyses (71±9% vs. 23±8%, P<0.001) (Fig. 1) and multivariate analyses (hazard ratio, 6.08; 95% CI, 1.77 to 20.88; P = 0.004) adjusted for the center (Lille vs. other centers) and the MELD score. Use of the global optimal algorithm resulted in the random selection of 69 additional matched controls who were nonresponders to medical therapy and 92 additional matched controls who were responders. The 69 randomly selected matched controls whose disease was nonresponsive to medical therapy were not significantly different from the 26 patients who underwent transplantation in terms of male sex (59.4%, P = 0.90), median age (control group, 51.8 years; 95% CI, 47.0 to 53.8; P = 0.19 for comparison with case patients), median Maddrey’s discriminant function (control group, 70.8; 95% CI, 64.2 to 82.3; P = 0.79), and Lille score (control group, 0.87; 95% CI, 0.80 to 0.93; P = 0.66). The 92 randomly selected matched controls whose disease did respond to medical therapy were not significantly different from the 26 case patients in terms of male sex (57.7%, P = 0.60), median age (control group, 46.2 years; 95% CI, 45.0 to 48.1; P = 0.37), and median Maddrey’s discriminant function (control group, 68.2; 95% CI, 65.4 to 76.0; P = 0.33), but as expected, the responder control group had lower Lille scores (0.18; 95% CI, 0.10 to 0.21; P<0.001) than the case-patient group. The 6-month survival rate was higher among the 26 case patients who underwent transplantation
Table 2. Clinical and Biologic Events between Study Enrollment and Listing for Liver Transplantation in the 26 Study Patients with Severe Alcoholic Hepatitis. Event
Value
Bacterial infection — no. (%)*
18 (69)
Ascites
6 (23)
Pulmonary
5 (19)
Urinary
4 (15)
Bacteremia
5 (19)
Other
2 (8)
Gram-negative bacteria
9 (35)
Gram-positive cocci
6 (23)
Pneumocystis carinii
2 (8)
Fungal infection
3 (12)
Spontaneous bacterial peritonitis without identifiable bacteria
2 (8)
Hepatorenal syndrome — no. (%)
15 (58)
Renal support or molecular adsorbent recirculating system — no. (%)
10 (38)
Gastrointestinal bleeding — no. (%)
4 (15)
Mechanical ventilation — no. (%)
4 (15)
Time from end of glucocorticoid treatment to listing — days Median (95% CI) Range
13 (6 to 17) 1 to 46
MELD at time of listing Median (95% CI) Range
34 (29 to 37) 25 to 41
Difference in MELD between nonresponse and listing† Median (95% CI) Range
−1.9 (−6.0 to 0.0) −14.2 to 14.4
Time from listing to liver transplantation — days Median (95% CI) Range
9 (3 to 11) 1 to 37
* Infection developed at multiple sites in four patients. † Nonresponse to medical therapy was determined after 7 days of medical therapy. All improvements were related to the use of a molecular adsorbent recirculating system or renal-replacement therapy.
than among the nonresponder controls (77±8% vs. 30±6%, P<0.001) but not significantly different than the rate among the responder controls (77±8% vs. 85±4%, P = 0.33) (Fig. 2). Burden of Early Liver Transplantation
During the study period, 891 transplantations were performed at the seven centers; 315 of these transplantations were for alcoholic liver disease. A total of 26 of all 891 procedures (2.9%) and 26 of the 315 procedures for alcoholic liver disease (8.3%) were early transplantations.
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1795
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
Table 3. Results of Case–Control Comparisons of the 26 Study Patients and 26 Matched Controls.* Characteristic
Patients Who Underwent Transplantation
Matched Controls
P Value
Living with a partner — no. (%)
17 (65)
13 (50)
0.19
Employed — no. (%)
17 (65)
16 (62)
0.92
Male sex — no. (%)
15 (58)
15 (58)
Age — yr Median (range) 95% CI
1.00 0.34
47.4 (34.9–60.5)
50.6 (34.5–60.6)
42.6–52.4
46.5–52.6
76.0 (35.6–165.0)
80.6 (48.4–202.0)
61.2–91.0
66.4–97.5
First day of medical therapy Maddrey’s discriminant function Median (range) 95% CI
0.56
Lille score Median (range) 95% CI
0.28 0.880 (0.260–0.996)
0.827 (0.250–0.999)
0.760–0.950
0.690–0.874
30.1 (22.0–47.3)
29.1 (19.1–40.0)
27.1–33.4
25.6–32.4
23.2 (14.3–82.0)
24.1 (16.7–51.5)
18.9–27.8
20.7–26.6
MELD score Median (range) 95% CI
0.27
Prothrombin time — sec Median (range) 95% CI
0.63
Bilirubin — mg/dl Median (range) 95% CI
0.61 27.7 (10.0–61.3)
26.2 (4.7–65.0)
19.1–31.8
17.4–34.7
0.8 (0.5–6.1)
1.1 (0.6–3.8)
0.6–1.1
0.8–1.4
25 (14–33)
24 (11–36)
23–29
21–26
28.5 (23.0–52.4)
29.4 (19.4–60.2)
26.2–33.7
25.6–31.8
22.2 (14.2–59.0)
23.5 (14.7–70.0)
20.7–24.5
20.3–26.3
28.4 (9.6–48.7)
27.7 (4.3–66.8)
24.7–34.1
20.1–34.6
Serum creatinine — mg/dl Median (range) 95% CI
0.11
Albumin — g/liter Median (range) 95% CI
0.42
After 7 days of medical therapy MELD score Median (range) 95% CI
0.47
Prothrombin time — sec Median (range) 95% CI
0.63
Bilirubin — mg/dl Median (range) 95% CI
0.58
Creatinine — mg/dl Median (range) 95% CI
0.26 1.0 (0.5–5.0)
1.2 (0.6–5.2)
0.6–1.2
0.9–1.4
27 (18–37)
27 (21–51)
24–30
24–30
Albumin — g/liter Median (range) 95% CI
0.52
* To convert the values for serum creatinine to micromoles per liter, multiply by 88.4. To convert values for bilirubin to micromoles per liter, multiply by 17.1.
1796
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
Early Liver Tr ansplantation for Alcoholic Hepatitis
100
77±8%
Patients undergoing transplantation
Survival (%)
75
71±9%
P<0.001 at 6 mo (primary end point)
50
Matched controls
23±8%
25
0
P<0.001 at 24 mo (extended follow-up)
0
6
12
23±8%
18
24
Months No. at Risk Patients undergoing transplantation Matched controls
26
20
15
14
13
26
6
6
5
4
Figure 1. Kaplan–Meier Estimates of Survival in the 26 Study Patients and the 26 Best-Fit Matched Controls.
In all, 233 patients were admitted for severe alcoholic hepatitis at the Brussels center (52 patients) and the Lille center (181 patients). A total of 18 of the 233 underwent transplantation (see Table 1 in the Supplementary Appendix), 14 of whom had been referred by community hospitals. The remaining 4 were directly selected by the Brussels and Lille centers through their own recruitment of patients with severe alcoholic hepatitis (representing 1.8% of the 219 patients who had not undergone transplantation or referral by community hospitals). The reason for exclusion from early transplantation was a predisposition to addiction or unfavorable social or familial profiles in approximately 90% of nonresponders with severe alcoholic liver disease.
deciliter (32.5 μmol per liter) (95% CI, 1.5 to 3.0 mg per deciliter [25.6 to 51.3 μmol per liter]) at 1 month to 0.7 mg per deciliter (12.0 μmol per liter) (95% CI, 0.6 to 1.1 mg per deciliter [10.3 to 18.8 μmol per liter]) at 6 months (P<0.001). After transplantation, patients were followed at short intervals, with a median of 11 visits (95% CI, 9 to 13) during the 6-month period. Since that time, follow-up is ongoing, with a median of 11 visits (95% CI, 5 to 14) at a median interval of 1.8 months (95% CI, 0.7 to 2.3). No alcoholic relapse was observed within the initial 6-month follow-up period. Three of 26 patients later resumed drinking alcohol, one at 720 days, one at 740 days, and one at 1140 days after transplantation. Despite counseling by an addiction specialist, 2 patients remained daily consumers (30 g per day and >50 g per day), Follow-up and Assessment of Alcohol Relapse whereas 1 drank occasionally (approximately 10 g In case patients who were alive at 6 months, per week). None of them has had graft dysfunction. liver tests and creatinine levels returned to normal ranges within the first month after transDiscussion plantation. There was no significant difference between the 1-month and 6-month values of the The high risk of early death7,17 in patients with aminotransferases, γ-glutamyltransferase, cre- severe alcoholic hepatitis not responding to medatinine, or international normalized ratio (see ical therapy11,18,19 makes it necessary to consider the Supplementary Appendix), although the me- all available treatment options, including transdian total bilirubin level declined from 1.9 mg per plantation, in targeted patients.20 Choosing the n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1797
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
100 Responder controls 85±4% P=0.33 Patients undergoing transplantation
Survival (%)
75
77±8%
P<0.001
50 Nonresponder controls 30±6%
25
0
0
2
4
6
Months No. at Risk Responder controls Patients undergoing transplantation Nonresponder controls
92 26
77 21
75 21
71 20
69
21
21
19
Figure 2. Kaplan–Meier Estimates of Survival among the 26 Study Patients and Randomly Selected Matched Controls.
appropriate moment for transplantation is crucial, to avoid adverse effects of intervening too soon or too late.21 This prospective multicenter study showed that early transplantation clearly improves the probability of 6-month survival among patients in whom medical therapy failed. We selected control patients by using two different processes of matching so as to reduce investigator bias and limit period and cohort effects. Team members requested stringent selection for at least two reasons. First, donor grafts remain in short supply. Second, in the allocation system based on MELD scores, patients with severe alcoholic hepatitis who are not having a response to medical therapy are likely to rank at the top of the transplant waiting list. Therefore, team members believed that patients who were unaware of their underlying liver disease constituted the most urgent problem. However, the exclusion of patients with previous episodes of liver decompensation raises an ethical question, since there is no agreement about the rationale for excluding alcoholic patients from transplantation.22 Studies evaluating the usefulness of fungal prophylaxis before transplantation, duration of glucocorticoid use before and after transplan1798
tation, and the tailoring of immunosuppressive regimens are warranted in light of the deaths due to aspergillosis. Our findings challenge both the notion of a prescribed abstinence period as the only alcoholism-related criterion for transplant eligibility23 and the opinion of experts that alcoholic hepatitis is a contraindication for transplantation.24 However, the stringency of our selection process resulted in our selecting a very small number of patients with alcoholic hepatitis for early transplantation. Although numerous studies have lent support to the validity of a sobriety period, they also have observed that the enforcement of this period alone delays listing for transplantation a considerable number of candidates with a low probability of relapse.10,25-30 Indeed, the duration of abstinence before transplantation is a poor predictor of relapse of alcoholism.31 Organization of the medical staff into four team circles should have ensured a balanced selection process. No relapse of alcoholism was observed during the 6 months following transplantation, although three patients had a relapse later. This low rate of relapse was probably related to our stringent selection of candidates for transplantation; for
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
Early Liver Tr ansplantation for Alcoholic Hepatitis
instance, physicians from Lille and Brussels selected fewer than 2% of the patients with severe alcoholic hepatitis recruited at their centers. Regarding organ shortage,32 early liver transplantation accounted for only 2.9% of grafts used during the study period. Nevertheless, modifications in guidelines for liver transplantation in patients with alcoholism may conflict with public preferences for liver-transplant allotment33,34 and may decrease willingness to donate.3 However, this has not occurred in response to transplantation being offered to patients with fulminant hepatic failure due to voluntary acetaminophen poisoning, nor to intravenous-drug users with acute hepatitis B virus infection. The limited supply of donor organs frequently biases what should be equal access to potential medical benefits for all patients.35 Development of an international database collecting information on survival and addiction is warranted to provide more facts and less conjecture in future discussions of the role of early transplantation in the treatment of severe alcoholic hepatitis.36 Although our data are encouraging, the study did have limitations. The design did not allow for rigorous assessment of long-term outcomes. Matched controls may not have been comparable to patients in terms of support of family members, intentions of patients to remain alcoholfree, or availability of counseling in the event of a return to drinking. Future studies elucidating long-term outcomes will require a control group of patients with alcoholism undergoing transplantation after abstaining from alcohol for a 6-month period who are matched to patients on the basis of social and familial characteristics. Previous studies of patients with alcoholism who underwent transplantation suggest that the rate of relapse over the long term may be approximately 25 to 35%.10,12,13 Finally, reproducibility of the selection process needs to be attempted. In summary, early liver transplantation may be an appropriate rescue option for selected patients whose first episode of severe alcoholic hepatitis is not responsive to medical therapy, after careful assessment of their addiction profile. Our encouraging results must be confirmed by other groups.
Supported by a grant from the Société Nationale Française de Gastroentérologie. Dr. Dharancy reports receiving consulting fees from Roche and lecture fees from Astellas and serving as a board member of Novartis. Dr. Dumortier reports receiving consulting fees, lecture fees, and grant support from Novartis and serving as a board member of Novartis; receiving lecture fees from JanssenCilag; receiving grant support from and serving as a board member of Astellas; receiving grant support from and serving as a board member of Roche; receiving grant support from ScheringPlough; and serving as a board member of GlaxoSmithKline. Dr. Durand reports receiving consulting fees from Astellas and consulting fees and grant support from Novartis. Dr. Leroy reports receiving lecture fees and grant support from and serving as a board member of Roche; receiving lecture fees and grant support from and serving as a board member of Bristol-Myers Squibb; receiving lecture fees and grant support from and serving as a board member of Gilead; receiving lecture fees from and serving as a board member of Merck; and serving as a board member of Janssen-Cilag. Dr. Louvet reports receiving consulting fees from Schering-Plough and Bristol-Myers Squibb. Dr. Mathu rin reports receiving lecture fees and grant support from and serving as a board member of Roche; receiving lecture fees from and serving as a board member of Schering-Plough; receiving lecture fees from and serving as a board member of BristolMyers Squibb; receiving lecture fees from and serving as a board member of Gilead; receiving lecture fees from Bayer Healthcare; receiving grant support from and serving as a board member of Janssen-Cilag; and serving as a board member of Norgine. Dr. Pageaux reports receiving lecture fees, grant support, and reimbursement for travel expenses from Roche; receiving lecture fees from Bayer HealthCare and Gilead; receiving grant support from and serving as a board member of Astellas; and receiving grant support from Novartis. Dr. Pruvot reports receiving lecture fees and reimbursement for travel expenses from Roche. Dr. Samuel reports receiving consulting fees, lecture fees, and grant support from Roche; consulting fees and lecture fees from Schering-Plough; consulting fees from Merck; consulting fees, lecture fees, and grant support from Astellas; consulting fees, lecture fees, and grant support from Novartis; consulting fees from Janssen-Cilag; consulting fees and lecture fees from Bristol-Myers Squibb; consulting fees and lecture fees from Gilead; and consulting fees and lecture fees from Laboratoire Français du Fractionnement et des Biotechnologies. No other potential conf licts were reported. Disclosure forms provided by the authors are available with the full text of this article at NEJM.org. We thank Nicole Declerck, Olivier Cottencin, Hamid Aftisse, Benjamin Rolland, and Christelle Bouquignaud (Hôpital Huriez, Lille, France); Corinne Antoine (Agence de la Biomédecine, Saint-Denis La Plaine, France); Delphine Degré, Nathalie Boon, Nadine Bourgeois, Michael Adler, Alexis Buggenhout, Nicolas Clumeck, and Brigitte Ickx (Hôpital Erasme, Brussels); René Adam, Daniel Azoulay, Chady Salloum, Eric Vibert, Faouzi Saliba, Philippe Ichai, Teresa Antonini, Audrey Coilly, Rodolphe Sobesky, Amine Benyamina, Lisa Blecha Magali Belnard, Colette Danet, Elisabeth Pasdeloup, Catherine Tanguy, and Gael Berthelot (Hôpital Paul-Brousse, Villejuif, France); Sophie Kalamarides, Pauline Houssel, and Vanessa Esnault (Hôpital Beaujon, Clichy, France); and all the nurses of the transplant teams and hepatology units of all participating centers for their contributions to this study.
References 1. Burra P, Senzolo M, Adam R, et al.
Liver transplantation for alcoholic liver disease in Europe: a study from the ELTR (European Liver Transplant Registry). Am J Transplant 2010;10:138-48.
2. Burroughs AK, Sabin CA, Rolles K, et
al. 3-Month and 12-month mortality after first liver transplant in adults in Europe: predictive models for outcome. Lancet 2006;367:225-32.
3. Shawcross DL, O’Grady JG. The
6-month abstinence rule in liver transplantation. Lancet 2010;376:216-7. 4. Beresford TP, Everson GT. Liver transplantation for alcoholic liver disease: bias,
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1799
Early Liver Tr ansplantation for Alcoholic Hepatitis beliefs, 6-month rule, and relapse — but where are the data? Liver Transpl 2000; 6:777-8. 5. Consensus conference: Indications for Liver Transplantation, January 19 and 20, 2005, Lyon-Palais Des Congrès: text of recommendations (long version). Liver Transpl 2006;12:998-1011. 6. O’Shea RS, Dasarathy S, McCullough AJ. Alcoholic liver disease. Hepatology 2010; 51:307-28. 7. Louvet A, Naveau S, Abdelnour M, et al. The Lille model: a new tool for therapeutic strategy in patients with severe alcoholic hepatitis treated with steroids. Hepatology 2007;45:1348-54. 8. Louvet A, Wartel F, Castel H, et al. Infection in patients with severe alcoholic hepatitis treated with steroids: early response to therapy is the key factor. Gastroenterology 2009;137:541-8. 9. Everhart JE, Beresford TP. Liver transplantation for alcoholic liver disease: a survey of transplantation programs in the United States. Liver Transpl Surg 1997;3: 220-6. 10. Foster PF, Fabrega F, Karademir S, Sankary HN, Mital D, Williams JW. Prediction of abstinence from ethanol in alcoholic recipients following liver transplantation. Hepatology 1997;25:146977. 11. Dunn W, Jamil LH, Brown LS, et al. MELD accurately predicts mortality in patients with alcoholic hepatitis. Hepatology 2005;41:353-8. 12. Pageaux GP, Bismuth M, Perney P, et al. Alcohol relapse after liver transplantation for alcoholic liver disease: does it matter? J Hepatol 2003;38:629-34. 13. Pageaux GP, Michel J, Coste V, et al. Alcoholic cirrhosis is a good indication for liver transplantation, even for cases of recidivism. Gut 1999;45:421-6. 14. Rosenbaum PR, Rubin DB. Constructing a control group using multivariate matched sampling methods that incorporate the propensity score. Am Stat 1985;39: 33-8. 15. Poynard T, Barthelemy P, Fratte S, et al. Evaluation of liver transplantation in alcoholic cirrhosis by a case-control study
1800
and simulated controls. Lancet 1994;344: 502-7. 16. Poynard T, Naveau S, Doffoel M, et al. Evaluation of efficacy of liver transplantation in alcoholic cirrhosis using matched and simulated controls: 5-year survival. J Hepatol 1999;30:1130-7. 17. Mathurin P, O’Grady J, Carithers RL, et al. Corticosteroids improve short-term survival in patients with severe alcoholic hepatitis: meta-analysis of individual patient data. Gut 2011;60:255-60. 18. Forrest EH, Evans CD, Stewart S, et al. Analysis of factors predictive of mortality in alcoholic hepatitis and derivation and validation of the Glasgow alcoholic hepatitis score. Gut 2005;54:1174-9. 19. Srikureja W, Kyulo NL, Runyon BA, Hu KQ. MELD score is a better prognostic model than Child-Turcotte-Pugh score or discriminant function score in patients with alcoholic hepatitis. J Hepatol 2005;42: 700-6. 20. Lucey MR, Mathurin P, Morgan TR. Alcoholic hepatitis. N Engl J Med 2009; 360:2758-69. 21. Olthoff KM, Brown RS Jr, Delmonico FL, et al. Summary report of a national conference: evolving concepts in liver allocation in the MELD and PELD era. Liver Transpl 2004;10:Suppl 2:A6-A22. 22. Benjamin M, Turcotte JG. Ethics, alcoholism and liver transplantation. In: Lucey MR, Merion RM, Beresford TP. Liver transplantation & the alcoholic patient. New York: Cambridge University Press, 1994: 113-30. 23. Lucey MR, Brown KA, Everson GT, et al. Minimal criteria for placement of adults on the liver transplant waiting list: a report of a national conference organized by the American Society of Transplant Physicians and the American Association for the Study of Liver Diseases. Liver Transpl Surg 1997;3:62837. 24. Bathgate AJ. Recommendations for alcohol-related liver disease. Lancet 2006; 367:2045-6. 25. Miguet M, Monnet E, Vanlemmens C, et al. Predictive factors of alcohol relapse after orthotopic liver transplantation for
alcoholic liver disease. Gastroenterol Clin Biol 2004;28:845-51. 26. Bird GL, O’Grady JG, Harvey FA, Calne RY, Williams R. Liver transplantation in patients with alcoholic cirrhosis: selection criteria and rates of survival and relapse. BMJ 1990;301:15-7. 27. Bravata DM, Olkin I, Barnato AE, Keeffe EB, Owens DK. Employment and alcohol use after liver transplantation for alcoholic and nonalcoholic liver disease: a systematic review. Liver Transpl 2001;7: 191-203. 28. Kumar S, Stauber RE, Gavaler JS, et al. Orthotopic liver transplantation for alcohol ic liver disease. Hepatology 1990;11:159-64. 29. Osorio RW, Ascher NL, Avery M, Bachetti P, Roberts JP, Lake JR. Predicting recidivism after orthotopic liver transplantation for alcoholic liver disease. Hepatology 1994;20:105-10. 30. Yates WR, Martin M, LaBrecque D, Hillebrand D, Voigt M, Pfab D. A model to examine the validity of the 6-month abstinence criterion for liver transplantation. Alcohol Clin Exp Res 1998;22:513-7. 31. DiMartini A, Day N, Dew MA, et al. Alcohol consumption patterns and predictors of use following liver transplantation for alcoholic liver disease. Liver Transpl 2006;12:813-20. 32. McMaster P. Transplantation for alcoholic liver disease in an era of organ shortage. Lancet 2000;355:424-5. 33. Neuberger J. Transplantation for alcoholic liver disease: a perspective from Europe. Liver Transpl Surg 1998;4:Suppl 1: S51-S57. 34. Ubel PA, Jepson C, Baron J, Mohr T, McMorrow S, Asch DA. Allocation of transplantable organs: do people want to punish patients for causing their illness? Liver Transpl 2001;7:600-7. 35. Surman OS, Cosimi AB, DiMartini A. Psychiatric care of patients undergoing organ transplantation. Transplantation 2009;87:1753-61. 36. Dureja P, Lucey MR. The place of liver transplantation in the treatment of severe alcoholic hepatitis. J Hepatol 2010;52:75964. Copyright © 2011 Massachusetts Medical Society.
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
Review article current concepts
Sudden, Unexpected Death in Epilepsy Orrin Devinsky, M.D.
E
pilepsy is characterized by both recurrent seizures and clinical uncertainty. Paroxysmal symptoms unpredictably punctuate life. Although most people with epilepsy live full and productive lives, doctors may too readily assure patients that seizures “never hurt the brain” and “are never fatal.” Over time, convulsive seizures can progressively impair cognition and behavior and alter brain structure.1 In rare cases, seizures can be fatal. Sudden, unexpected death in epilepsy refers to a death in a patient with epilepsy that is not due to trauma, drowning, status epilepticus, or other known causes but for which there is often evidence of an associated seizure. The event may be unwitnessed, and the person is often found dead in bed. A finding of sudden, unexpected death in epilepsy is definite when clinical criteria are met and autopsy reveals no alternative cause of death, such as stroke, myocardial infarction, or drug intoxication, although there may be evidence of a seizure (e.g., a bitten tongue or pulmonary edema).2 Sudden, unexpected death in epilepsy is probable when clinical criteria are met but there is no autopsy, and it is possible when there is an alternative cause of death or when clinical data are lacking.3 The incidence of sudden, unexpected death in epilepsy is probably underestimated because records and databases are incomplete and because it often goes unrecognized by coroners, medical examiners, and physicians who are unaware of the disorder or its diagnostic criteria. The incidence per 1000 patient-years varies with the sample population, increasing from 0.09 to 2.65 in community samples to 1.2 to 5.9 in tertiary care epilepsy centers to 6.0 to 9.3 among patients evaluated for or treated with surgery or vagus-nerve stimulation for epilepsy (Fig. 1).3-13 The rate of sudden, unexpected death increases with the duration and severity of epilepsy. It is much less common in children than in adults5,9,14; among children with epilepsy who are younger than 14 years of age, cases of sudden, unexpected death are rare and are largely restricted to children who have a major neurologic impairment or a history of a major neurologic insult.5,9,14 Like adults, most children in whom sudden, unexpected death in epilepsy is diagnosed die in bed and have a history of generalized tonic–clonic seizure.5 Sudden, unexpected death contributes to the increased mortality among children with disorders associated with both treatment-resistant epilepsy and developmental delay, such as autism, the Dravet syndrome, tuberous sclerosis complex, and chromosome 15q11-13 duplication.9,15,16 The magnitude of the problem of sudden, unexpected death in epilepsy is unrecognized in the medical and lay communities. In a population-based cohort of children with epilepsy who were followed for 40 years, sudden, unexpected death occurred in 9% of patients and accounted for 38% of all deaths.9 Although the epilepsy began in childhood, almost all the deaths occurred in adulthood. Among high-risk patients with major neurologic disorders (e.g., mental retardation or cerebral palsy) or treatment-resistant epilepsy, rates of sudden, unexpected death can exceed 10% per decade.4,9 By comparison, the risk of death from surgery for epilepsy n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
From the Department of Neurology, NYU Langone School of Medicine, New York. Address reprint requests to Dr. Devinsky at the Department of Neurology, NYU Langone School of Medicine, 223 E. 34th St., New York, NY 10016, or at
[email protected]. N Engl J Med 2011;365:1801-11. Copyright © 2011 Massachusetts Medical Society.
1801
The
n e w e ng l a n d j o u r na l
12
Incidence (per 1000 person-yr)
10
8
6
4
2
tio
la
Po
pu
Ch
ild
re
n
(3 ) nB Te Co as rti ho ed ar rt y(4 Ca ) re Co Ch ho Lo ild rt ng ho (6 -T od ) er O m n Fo se Pa llo t w tie w- it nt up h sw (1 ) i Re th Pa M t ar e tie da nt nt tio al sw n ith (3 ) Re Pa fra E tie pi ct lep or n Re ts R sy y ce ef (4 ) iv err in e g d Su fo rg r o er r y( 2)
0
Figure 1. Incidence of Sudden, Unexpected Death in Epilepsy According to Population Type. Data are based on the pooled results of 23 studies, cited in Sillanpää and Shinnar9 and Tomson et al.11 Numbers in parentheses indicate the number of studies for each population. I bars denote 95% confidence intervals.
is approximately 1 death per 1500 procedures; the monthly risk of death for patients at very high risk for sudden, unexpected death is approximately 1 in 2000.
R isk Fac t or s a nd Mech a nisms Seizure
Evidence from epidemiologic, observational, clinical, and pathological studies strongly suggests that in most cases, sudden, unexpected death in epilepsy occurs after a seizure, usually a tonic– clonic seizure (Table 1). In case–control studies, the most consistent risk factor is an increased frequency or recent history of seizure, especially tonic–clonic seizure (Table 1).11,12,17,19,34,35 Among patients who have had more than three tonic– clonic seizures in the preceding year, the risk of sudden, unexpected death is increased by a factor of more than 8.13 Additional risk factors include lack of treatment with antiepileptic drugs or subtherapeutic levels of such drugs, antiepileptic-drug 1802
of
m e dic i n e
polytherapy, frequent changes in antiepileptic drugs, early adulthood, epilepsy of long duration, and mental retardation (Table 1).11,12,18-20 Among 15 cases of sudden, unexpected death in epilepsy that were witnessed in the community, the death was preceded by a tonic–clonic seizure in 12 persons and was postictal in 2; 1 person shouted, “I’m going to have a seizure,” before collapsing.27 Difficulty breathing was observed in 80% of these cases,27 with 70% of patients found in a prone position,18 suggesting that suffocation contributed to their deaths. Half of witnessed cases in children were preceded by a tonic–clonic seizure, whereas the other half were preceded by sudden loss of consciousness.5 Seizure can cause apnea or arrhythmia without convulsive activity. In six infants, there were 23 life-threatening events in which hypoxemia and apnea followed seizure recorded on electroencephalography (EEG) and in which other clinical features were subtle or absent.36 The only cardiac change was tachycardia, which supports the theory that a seizureinduced respiratory mechanism may be involved in pediatric cases of sudden, unexpected death in epilepsy.36 There is no definite documentation of an unexpected death that was not preceded by a seizure. Although witnesses have reported deaths that were not observed to be preceded by seizure, such an event has not been captured in the millions of hours of video and ambulatory EEG recordings obtained each year from patients with epilepsy. In separate case reports of 13 patients in epilepsy-monitoring units, 8 died suddenly, and 5 were resuscitated after a life-threatening episode. These cases may not accurately reflect the course of sudden, unexpected death in the community, since most of the patients in the study had treatment-resistant epilepsy and underwent rapid withdrawal of medication. However, recorded cases do provide valuable information on the mechanisms of these deaths. Each of the 13 patients had seizures just before death: 12 had tonic–clonic seizures and 1 had complex partial seizures11,21,27-33,37; 10 of the patients had two or more seizures the day before they died, and most of the patients died during sleep. Patients who were successfully resuscitat ed were younger than those who died (mean age, 29 years vs. 43 years); older patients may be more vulnerable to seizure-induced cardiopulmonary or brain dysfunction. Electrocardiography (ECG) was usually performed during video EEG, with the patient’s respiratory effort assessed through visual
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
current concepts
Table 1. Risk Factors for and Mechanisms of Sudden, Unexpected Death in Epilepsy.* Risk Factor or Mechanism
Source of Data Epidemiologic Studies†
Seizure, especially GTCS Frequent seizures, recent seizure, long postictal EEG suppression
Video EEG‡
Witness§
Yes
Yes
Yes
Yes
Yes
Yes
Terminal seizure
Yes
Yes
Yes
Prone position after seizure
Yes
Yes
Yes
Postictal breathing problem
Yes
Yes
Yes
Postictal cardiac arrhythmia
No
Yes
No
Environment Lack of nighttime supervision or monitoring
Yes
No
No
Location (in bed or sleeping)
Yes
Yes
Yes
Yes
Yes
No
Antiepileptic-drug therapy Lack of use or subtherapeutic levels Polytherapy
Yes
No
No
Frequent changes in regimen¶
Yes
No
No
Rapid withdrawal
No
Yes
No
Other factors or mechanisms Early adulthood
Yes
Yes
Yes
Early onset of epilepsy‖
Yes
No
No
Long duration of epilepsy**
Yes
No
No
Yes
No
No
Neurologic status Intelligence quotient <70 Nonambulatory
Yes
No
No
Major neurologic insult
Yes
No
No
Carbamazepine or lamotrigine
Yes
No
No
* The term “No” indicates that there is a lack of evidence for the risk factor or mechanism. EEG denotes electroencephalography, and GTCS generalized tonic–clonic seizure. † The epidemiologic data were reported by Sillanpää and Shinnar,9 Tomson et al.,11,12 Walczak et al.,13 Camfield and Camfield,14 Hitiris et al.,17 Kloster and Engelskjøn,18 Nilsson et al.,19 Langan et al.,20 Lhatoo et al.,21 Nei et al.,22 Ryvlin et al.,23 Aurlien et al.,24 Opeskin et al.,25 and Nilsson et al.26 ‡ The video EEGs were summarized or reported by Tomson et al.,11 Langan et al.,27 Bateman et al.,28 Bird et al.,29 Espinosa et al.,30 So et al.,31 Tao et al.,32 and Thomas et al.33 § The witnessed cases of sudden death were reported by Annegers,3 Donner et al.,5 and Langan et al.27 ¶ The addition of treatment with an antiepileptic drug in the preceding 3 months is associated with a lower risk of sudden death.23 ‖ Early onset refers to onset before 15 years of age. ** Long duration refers to a duration of more than 15 to 30 years.
inspection. Respiratory problems predominated in 8 patients, who underwent postictal hypoventilation, apnea, cyanosis, inspiratory stridor, laryngospasm, pulmonary edema, or suffocation.11,21,32,33,38 Shutdown of brain function, with severe, diffuse EEG attenuation, was considered the primary cause of death in 2 cases; 1 patient showed “no movement” while lying prone postictally and had a pulse of 47 beats per minute; an autopsy revealed pulmonary edema.11,29 Cardiac abnormalities, such
as peaked T waves, ST-segment elevation, and asystole, usually followed collapse or, less often, coincided with respiratory problems.11,28-31 Ventricular arrhythmia came close to causing sudden death in 1 patient, who had a convulsion that lasted 4.5 minutes, at which point ventricular tachycardia progressed to fibrillation.30 In 2 patients, multiple mechanisms may have led to sudden death.21,37 The view that terminal seizure plays a role in cases of sudden, unexpected death in epilepsy is
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1803
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
Increased seizures AEDs or abrupt changes in AEDs Prolonged postictal state Acquired changes in brain function Acquired autonomic dysfunction Genetics
Simple or complex partial seizure
Cardiac arrhythmia Death Hypoventilation or hypoxia
Generalized tonic–clonic seizure
Cerebral shutdown Central apnea
Chronic epilepsy
Obstructive apnea Asphyxia Pulmonary edema
Figure 2. Mechanisms of Seizure-Induced Death. Sudden, unexpected death often follows a seizure in patients with chronic epilepsy (orange). Potential risk factors associated with chronic epilepsy (blue) and direct effects from seizures (green, pink, and purple) probably interact in various ways to cause death, which is probably a result of cerebral shutdown, hypoventilation and hypoxia, cardiac arrhythmia, or some combination thereof. AED denotes antiepileptic drug.
supported by the fact that levels of heat-shock protein 70 in hippocampal neurons are elevated in such cases.39 Seizure can cause pulmonary edema,40,41 which is the most common autopsy finding in such cases.7,17,18 It remains uncertain whether sudden death occurs without a terminal seizure. Excluding cases in which there is a terminal seizure, the rate of sudden, unexpected death among patients with epilepsy may be similar to that in the general population. Determining the precise mechanism of death is difficult, even for cases that were recorded on video EEG. In most cases, there is no monitoring of blood pressure or respiratory function, and the findings on ECG are often restricted to a single channel riddled with artifact. A respiratory disorder such as hypoventilation could go unnoticed while cardiac arrhythmia, which is actually sec1804
n engl j med 365;19
ondary, is considered causative (or vice versa). In addition to cardiorespiratory causes and cerebral shutdown, genetic, drug, metabolic, and environmental factors, as well as multifactorial processes, may contribute to sudden, unexpected death in patients with epilepsy (Fig. 2). Respiratory Factors
The concept of impaired respiration as a cause of death is supported by data from studies in animals and evidence from most witnessed and recorded instances of sudden, unexpected death in epilepsy (Table 2).27,28-33 Seizure-induced respiratory changes can be lethal and may involve pulmonary dysfunction and suppression of brainstem respiratory and arousal centers.40 In sheep, prolonged seizures cause elevated pressure in the left atrial and pulmonary arteries, pulmonary ede-
nejm.org
november 10, 2011
The New England Journal of Medicine
current concepts
Table 2. Suggested Mechanisms of Impaired Respiration in Sudden, Unexpected Death in Epilepsy (SUDEP). Mechanism
Source of Data
Respiratory problems (respiratory arrest, labored breathing, suffocation in prone position, laryngeal spasm) are recorded in most instances of near and actual SUDEP on video EEG.
Tomson et al.,11 Langan et al.,27 Bateman et al.,28 Thomas et al.33
Respiratory problems are reported in most witnessed incidents of SUDEP.
Langan et al.27
Most incidents occur during sleep and patient is usually found prone, supporting suffocation as contributing cause of death.
Kloster and Engelskjøn,18 Langan et al.27
Lack of monitoring increases risk (repositioning and stimulation can improve respiration but cannot relieve arrhythmia).
Nashef et al.,7 Langan et al.20
Hypercarbia and hypoxemia are common after seizures.
Hewertson et al.,36 Bateman et al.28
Prolonged generalized EEG suppression is a risk factor (brain stem initiates breathing and arousal; the heart can function autonomously).
Lhatoo et al.21
Infants with seizure-induced life-threatening events have hypoxemia and apnea, not arrhythmias.
Hewertson et al.36
Pulmonary edema is the most common autopsy finding.
Nashef et al.,7 Hitiris et al.,17 Kloster and Engelskjøn18
Video-EEG recordings infrequently show cardiac problem as primary cause of SUDEP.
Langan et al.,27 Bateman et al.,28 Bird et al.,29 Espinosa et al.,30 So et al.,31 Tao et al.,32 Thomas et al.33
ma, tachycardia, and death from hypoventilation.42 Serotonergic neurons that modulate breathing and arousal may be involved in sudden, unexpected death in epilepsy, as is the case with sudden infant death syndrome.40,43 Some serotonergic neurons stimulate respiratory nuclei in the brain stem, whereas others, activated by hypercapnia, contribute to the ascending arousal system.44 Postictal depression of serotonergic activity can impair respiration and reflexive repositioning if the mouth and nose are obstructed by bedding. In some mouse strains, sound-induced seizure arrests respiration — an effect that is reduced by selective serotoninreuptake inhibitors (SSRIs) and 5-HT2C –receptor agonists.45 Among patients with epilepsy, use of an SSRI is associated with reduced oxygen desaturation during partial seizure but not during tonic– clonic seizure.46 Cerebral Shutdown
Seizure and the postictal state can affect brainstem respiratory centers. Central apneas or hypop neas complicate most seizures.38 In one study, patients with epilepsy who died suddenly had longer periods of postictal generalized EEG suppression than did patients with epilepsy who did not die suddenly.21 Respiration depends on brain-stem activity; prolonged suppression of activity stops respiration. Postictal shutdown of cerebral and brain-stem function may be related to the mechanisms that stop seizures. Postictal hypercapnia and hypoxemia can occur despite increased respi-
ratory effort, possibly from ventilation–perfusion inequality, which is caused by right-to-left pulmonary shunting or neurogenic pulmonary edema.47 Sudden, unexpected death has been reported in a patient with epilepsy who had postictal pulmonary edema.36 Postictal hypercapnia can cause severe acidosis that is arrhythmogenic.48 The effects of prolonged postictal EEG suppression, apnea, pulmonary shunting and edema, suffocation in the prone position, impaired arousal to hypercapnia, laryngeal spasm, and respiratory acidosis probably combine and cascade with cardiac factors to cause many cases of sudden, unexpected death in patients with epilepsy (Table 2 and Fig. 2). Cardiac Factors
Cardiac events are considered to be likely culprits in some instances of sudden, unexpected death in patients with epilepsy.4,11,12,30,48,49 Seizure-induced arrhythmias occur in animals and humans,49,50 but in 13 case studies of near and actual sudden death in patients with epilepsy, only 1 incident was clearly due to a cardiac event.30 Hypoxemia could lower the threshold for cardiac arrhythmias during seizure, especially in patients with channelopathies affecting both brain and cardiac tissue (e.g., long-QT syndrome type 2).51 Mice lacking the Kv1.1 potassium channel have severe seizures and die prematurely, possibly because of cardiac arrhythmias.48 Interictal and ictal cardiovascular changes occur in patients with epilepsy,49,50 in-
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1805
The
n e w e ng l a n d j o u r na l
cluding prolongation of the QT interval corrected for heart rate (QTc) during the ictal and interictal periods and shortening of the QTc interval postictally.52-54 Ictal asystole occurs during video EEG in 0.1 to 0.4% of patients, but recurrence is rare after pacemaker implantation.49,55 Patients with epilepsy who die suddenly have rates of cardiac repolarization abnormalities and arrhythmias that are similar to those among other patients with epilepsy,22,53 but sudden death is associated with more severe tachycardia during nocturnal seizures.22 Intense seizures may cause greater activation of the sympathetic nervous system, possibly contributing to cardiopulmonary dysfunction56 and to prolonged suppression of brain activity,38 which can in turn cause apnea and impair arousal while the heart functions independently. Intense seizures may also trigger greater compensatory responses (e.g., elevated adenosine levels), and these responses may contribute to sudden death.57 Other Risk Factors
The severity of epilepsy partly explains some of the risk factors for sudden death, such as tonic– clonic seizures, frequent seizures, early onset and long duration of epilepsy, and polytherapy, but some factors probably contribute to the risk of sudden death directly. Tonic–clonic seizure is often documented immediately before sudden death, suggesting that seizures are responsible for many cases. The association of early onset and long duration of epilepsy with an increased risk of sudden death suggests that progressive neural changes contribute to the risk. Although polytherapy is a mark of treatment-resistant epilepsy, treatment with three or more antiepileptic drugs has been found to increase the risk of sudden death by a factor of more than 8 after adjustment for seizure frequency.19 However, in randomized, controlled trials involving patients with treatment-resistant epilepsy, the rate of sudden death was increased by a factor of more than 7 among patients who received placebo as compared with those who received an additional antiepileptic drug.23 The effects of the number of drugs taken, the frequency of dose changes, and the recent removal or addition of drugs on the risk of sudden death require further investigation. Sudden, unexpected death in epilepsy usually occurs in chronic, severe cases of epilepsy, often in patients with a history of neurologic insult. Thus, such deaths may result from the cumulative 1806
of
m e dic i n e
effects of seizures compounded by a tonic–clonic seizure (i.e., the additive effect of acute injury after chronic injury). Frequent tonic–clonic seizures can progressively damage brain structure (e.g., atrophy of the hippocampus) and function (e.g., impairment of short-term memory).1 Since greater interictal autonomic changes occur in patients with chronic epilepsy than in those with recent-onset epilepsy,35 there may also be progressive changes in the brain-stem areas that regulate cardiorespiratory function and arousal. However, sudden death can occur early in the course of epilepsy and in patients who have seizures only rarely. It is not known why sudden death occurs in some patients after only a few seizures, whereas others are spared despite a lifetime of tonic– clonic seizures. Genetic susceptibility (e.g., cardiac channelopathy), alcohol use, medication withdrawal, and fever may increase the risk of sudden death after a seizure, but these potential risk factors have not been adequately studied. Regarding genetic susceptibility, mutations in the ionchannel genes expressed in brain and cardiac tissue may underlie susceptibility to epilepsy, brainstem autonomic dysfunction, and cardiac arrhythmias.51 Abnormalities in the serotonergic system have been found in patients with epilepsy or depression and in cases of sudden death.40 Patients with epilepsy who have been treated for depression within the past 12 months have a 40% increase in the risk of death, as compared with those who have not been treated for depression.58 Carbamazepine, which has been associated with sudden death among patients with epilepsy in some studies,11,22 can suppress autonomic functions.59 Lamotrigine is also associated with an increased risk of sudden death among patients with epilepsy in some studies.24,60
Sudden, Une x pec ted, Sei zur e-Induced De ath In most cases, when sudden death occurs after a seizure, it is more precisely defined as sudden, unexpected, seizure-induced death. Use of this term can be important in helping to make patients and their families aware of the potential danger of seizure and to increase public awareness of the disorder. The phrase “sudden, unexpected death in epilepsy” communicates the message that some people with epilepsy die unexpectedly for unknown reasons, but sudden death also occurs
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
current concepts
in people who are or appear to be healthy and complicates our understanding of many medical disorders. In contrast, the use of the phrase “sudden, unexpected, seizure-induced death” underscores the fact that seizure is involved in death and thereby lays the path for education about prevention. Sudden, unexpected, seizure-induced death should be considered to be a type of seizure-induced death, along with those due to status epilepticus, drowning, motor-vehicle accidents, trauma, burns, and suicide during a postictal psychosis.
Pr e v en t ion We do not know how to prevent sudden, unexpected death in epilepsy. No prospective or controlled studies have evaluated interventions to reduce its incidence. Since a tonic–clonic seizure precedes most sudden deaths in patients with epilepsy, seizure control — and the appropriate use of medication as well as counseling on lifestyle — is the focus of prevention (Table 3). The lack of therapeutic levels of antiepileptic drugs, nonadherence to treatment regimens, and frequent changes in regimens are all risk factors for sudden death.9,11,12,19,20,25 For patients who have never been treated with antiepileptic drugs, the risk of sudden death may be more than 20 times as high as that for treated patients.20 A discussion of sudden, unexpected death in epilepsy may be worthwhile for patients with tonic–clonic seizures who are beginning an antiepileptic-drug regimen and for patients at high risk for recurrent tonic–clonic seizures who are discontinuing such a regimen. Thirty percent of patients with epilepsy have treatment-resistant epilepsy, and these patients — who are at high risk for sudden death — present the greatest challenge. Frequent changes in multidrug regimens are commonly undertaken to reduce the frequency of seizure or the side effects of medication, but the potential effects of regimen changes on the risk of sudden death are rarely considered and remain unknown. Patients who are free of seizures after surgery for epilepsy have reduced rates of sudden death,8,10,62 with mortality approaching that in the general population. In contrast, patients with postoperative seizures have very high rates of sudden death.62 Patients should be informed about the fundamentals of seizure prevention: observance of a
healthful lifestyle (e.g., avoiding sleep deprivation and excessive consumption of alcohol), adherence to their antiepileptic-drug regimen (e.g., avoiding and identifying missed doses with the use of weekly pill boxes and watch alarms), knowledge of how to make up for missed doses and of factors influencing drug levels, and avoidance of drugs that lower the threshold for seizure. Patients should also be asked whether they have symptoms of nocturnal tonic–clonic seizure. For patients who do have nocturnal tonic–clonic seizure, bedtime medication doses can be increased and seizuredetection devices (discussed below) considered. Educating persons who live with patients with epilepsy may also help to prevent sudden death. Knowledge of how to perform the appropriate first-aid responses to seizure (repositioning the patient and protecting the airway after a tonic– clonic seizure) may prevent death. A case–control study of adults showed that as compared with unmonitored controls, patients who were monitored while they slept had a reduced risk of sudden death by a factor of 2.5 if another person older than 10 years of age was in the room and by a factor of 10 if there were frequent nighttime checks or if a sound-monitoring device was used.20 In a study of 14 residents at a special-needs school who had severe epilepsy and died suddenly, all 14 died at home, and most of them were not being monitored.7 None died at school, where they were monitored during sleep by four attendants and an on-call nurse and a sound-monitoring device was used. The manufacturers of several commercial devices state that their products can detect tonic–clonic seizure, but very limited data are available on the sensitivity and specificity of these devices for the detection of tonic–clonic seizure,63 and there is no evidence that their use prevents sudden death. Patients with nocturnal tonic–clonic seizure may want to consider the use of motion-detection devices (e.g., the Emfit monitor [Emfit]) that have a configurable sensing unit that is placed under the mattress and a receiver located in another room with audiovisual alarms. Pulse oximeters and heart-rate monitors may detect seizure-induced hypoxemia and tachycardia. Nonetheless, sudden death occurs in hospitals and other medical environments, despite prompt attempts at resuscitation.11 The effectiveness of strategies intended to prevent sudden death by improving respiration or oxygenation remain unproven. Since many pa-
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1807
1808 Educate the patient about the importance of adhering to medication regimen and observing healthy lifestyle Consider dose increase; advise patient to adhere to medication regimen and explain what should be done if medications are missed or if patient has gastrointestinal illness that causes vomiting or diarrhea† Optimize and simplify the regimen and avoid frequent changes Recommend that caregivers monitor the patient (e.g., with a sound monitor) Reposition the patient to lie on side or stimulate patient; for apnea or severe hypoventilation, initiate cardiopulmonary resuscitation Recommend a cardiac monitor or, for patients with life-threatening arrhythmia, a pacemaker
Breakthrough seizure
Low levels of antiepileptic drugs
Frequent changes in antiepileptic drugs
Mental retardation, nonambulatory status
Prone position, suffocation after seizure
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
Cardiac arrhythmia, asystole
Consider use of a seizure or respiratory monitor or lattice pillows‡
of
Consider use of a seizure monitor‡
A redosing strategy is recommended for missed medications or medications lost to vomiting or severe diarrhea
Drugs that lower the seizure threshold (e.g., phenylephrine, pseudoephedrine, bupropion) should be avoided
A sound or seizure monitor may be considered, but the sensitivity and specificity of these monitors for seizure detection is uncertain, as is their role in preventing sudden death
A redosing strategy is recommended for missed medications or medications lost to vomiting or severe diarrhea
Rescue medications include clonazepam (oral, dissolving wafer), diazepam (buccal solution or rectal gel), lorazepam (oral tablet or liquid), and midazolam (nasal spray, buccal solution)*
Comment
n e w e ng l a n d j o u r na l
* Rescue medications have a rapid onset and are used to treat prolonged seizure or seizure clusters. The nasal and buccal forms of midazolam have not been approved in the United States. † In pregnancy, many drug levels decline; levels should be closely monitored and dose increased to maintain a therapeutic level. ‡ The efficacy of seizure and respiratory monitors has not been proved for the detection of seizure or the prevention of sudden, unexpected death in epilepsy. The efficacy of lattice pillows has not been proved for the prevention of suffocation.
Develop strategies to ensure adherence to medication regimen, especially evening and bedtime doses
Redistribute or increase evening or bedtime medication
Identified
Sleeping alone, especially with history of nocturnal seizure
Indications that a patient may be having nocturnal seizures without awareness of them are urinary incontinence, tongue or cheek bites, unusual headaches, muscle soreness, lethargy, or confusion on awakening; ask the patient whether any of these conditions apply
Educate patient about drug interactions (e.g., oral contraceptives lower lamotrigine levels)61 and the effects of vomiting and diarrhea on drug absorption
Unidentified
Nocturnal seizure
Fluctuating levels of antiepileptic drugs
Consider use of benzodiazepine rescue medications; urge patient to seek prompt medical attention when appropriate
Consider use of rescue medications
Seizure clusters, prolonged seizure
Preventive Strategy Improve adherence to medication regimen; recommend lifestyle modifications — regular, restorative sleep and limitation of alcohol consumption; optimize antiepileptic-drug regimen; consider epilepsy surgery, dietary therapy, or neurostimulation
Generalized tonic–clonic seizure and uncontrolled seizure
Risk Factor
Table 3. Prevention of Sudden, Unexpected Death in Epilepsy.
The
m e dic i n e
current concepts
tients are found prone, some practitioners advocate the use of antisuffocation pillows (e.g., see www.sleep-safe.co.uk) for the prevention of sudden death in epilepsy, but data on their clinical value are lacking. The potential role of oxygen administration during and after a tonic–clonic seizure — a common practice in hospitals but not in patients’ homes — deserves study. Routine ECG screening of all patients with epilepsy is of uncertain value. Patients with tonic– clonic seizure or episodic loss of consciousness who have normal or nonspecific findings on magnetic resonance imaging of the brain and EEG should undergo ECG to rule out the long-QT syndrome, a lethal disorder that mimics epilepsy 64; these patients may also benefit from Holter ECG monitoring to rule out arrhythmia. Misdiagnosis of the long-QT syndrome as epilepsy precludes the use of effective therapy, and the inappropriate administration of antiepileptic drugs can induce arrhythmias.51 One additional preventive step is to discuss sudden, unexpected death in epilepsy with patients — a step few physicians initiate because it may cause stress about an uncommon problem for which there is no proven means of prevention. However, most patients with epilepsy and their families want information about sudden death.65,66 Although national guidelines in the United Kingdom recommend that all patients with epilepsy and their families be provided with information about sudden, unexpected death in epilepsy,66 most neurologists cited in the study discuss it only with high-risk patients or when specifically asked.65,66 Guidelines and tools are needed to assist physicians and patients and their families in becoming educated about sudden, unexpected death in epilepsy, and outcome measures
are needed to assess the effectiveness of this education. Patients with risk factors that can be modified — such as nonadherence to antiepileptic-drug regimen, tonic–clonic seizure, nocturnal seizure, and treatment-resistant epilepsy — may benefit most from counseling. When patients ask whether seizures can injure their brain or kill them, a simple “no” is insufficient. Epilepsy is associated with significant risks of morbidity and death.
F u t ur e Dir ec t ions Reductions in sudden deaths among patients with epilepsy may be achieved by increasing awareness on the part of the general public and the medical community, improving the prevention and treatment of epilepsy, further developing and encouraging the use of devices that detect seizure and can alert caretakers, improving our understanding of the mechanisms of sudden, unexpected death in epilepsy, and conducting interventional trials to prevent the progression of life-threatening seizure to sudden death. If patients are aware that seizure can be deadly, they may be more motivated to adhere to antiepileptic-drug regimens and to avoid lifestyle choices that increase the likelihood of seizure. Patients with seizures that remain uncontrolled after the administration of two different drug regimens should be referred to an epilepsy center for evaluation. Preventing seizures can be lifesaving. Dr. Devinsky reports providing expert testimony in legal medical proceedings regarding possible sudden, unexpected death in epilepsy and receiving lecture fees from UCB. No other potential conflict of interest relevant to this article was reported. Disclosure forms provided by the author are available with the full text of this article at NEJM.org. I thank Drs. Daniel Friedman and Ion-Florin Talos and Brigid Staley, M.P.H., for providing invaluable input on earlier versions of the manuscript.
References 1. Hermann B, Seidenberg M, Bell B.
The neurodevelopmental impact of childhood onset temporal lobe epilepsy on brain structure and function and the risk of pro gressive cognitive effects. Prog Brain Res 2002;135:429-38. 2. Nashef L. Sudden unexpected death in epilepsy: terminology and definitions. Epilepsia 1997;38:Suppl:S6-S8. 3. Annegers JF. United States perspective on definitions and classifications. Epilepsia 1997;38:Suppl:S9-S12. 4. Dasheiff RM. Sudden unexpected death in epilepsy: a series from an epilepsy surgery program and speculation on the rela-
tionship to sudden cardiac death. J Clin Neurophysiol 1991;8:216-22. 5. Donner EJ, Smith CR, Snead OC III. Sudden unexplained death in children with epilepsy. Neurology 2001;57:430-4. 6. Ficker D, So E, Shen WK, et al. Population-based study of the incidence of sudden unexplained death in epilepsy. Neurology 1998;51:1270-4. 7. Nashef L, Fish DR, Garner S, Sander JW, Shorvon SD. Sudden death in epilepsy: a study of incidence in a young cohort with epilepsy and learning difficulty. Epilepsia 1995;36:1187-94. 8. Nilsson L, Ahlbom A, Farahmand BY,
Tomson T. Mortality in a population-based cohort of epilepsy surgery patients. Epilepsia 2003;44:575-81. 9. Sillanpää M, Shinnar S. Long-term mortality in childhood-onset epilepsy. N Engl J Med 2010;363:2522-9. 10. Sperling MR, Feldman H, Kinman J, Liporace JD, O’Connor MJ. Seizure control and mortality in epilepsy. Ann Neurol 1999;46:45-50. 11. Tomson T, Nashef L, Ryvlin P. Sudden unexpected death in epilepsy: current knowledge and future directions. Lancet Neurol 2008;7:1021-31. 12. Tomson T, Walczak T, Sillanpaa M,
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1809
The
n e w e ng l a n d j o u r na l
Sander JW. Sudden unexpected death in epilepsy: a review of incidence and risk factors. Epilepsia 2005;46:Suppl 11:54-61. 13. Walczak TS, Leppik IE, D’Amelio M, et al. Incidence and risk factors in sudden unexpected death in epilepsy: a prospective cohort study. Neurology 2001;56:519-25. 14. Camfield P, Camfield C. Special considerations for a first seizure in childhood and adolescence. Epilepsia 2008;49:Suppl 1:40-4. 15. Gillberg C, Billstedt E, Sundh V, Gillberg IC. Mortality in autism: a prospective longitudinal community-based study. J Autism Dev Disord 2010;40:352-7. 16. Le Gal F, Korff CM, Monso-Hinard C, et al. A case of SUDEP in a patient with Dravet syndrome with SCN1A mutation. Epilepsia 2010;51:1915-8. 17. Hitiris N, Suratman S, Kelly K, Stephen LJ, Sills GJ, Brodie MJ. Sudden unexpected death in epilepsy: a search for risk factors. Epilepsy Behav 2007;10:138-41. 18. Kloster R, Engelskjøn T. Sudden unexpected death in epilepsy (SUDEP): a clinical perspective and a search for risk factors. J Neurol Neurosurg Psychiatry 1999;67: 439-44. 19. Nilsson L, Farahmand BY, Persson PG, Thiblin I, Tomson T. Risk factors for sudden unexpected death in epilepsy: a casecontrol study. Lancet 1999;353:888-93. 20. Langan Y, Nashef L, Sander JW. Casecontrol study of SUDEP. Neurology 2005; 64:1131-3. 21. Lhatoo SD, Faulkner HJ, Dembny K, Trippick K, Johnson C, Bird JM. An electroclinical case-control study of sudden unexpected death in epilepsy. Ann Neurol 2010;68:787-96. 22. Nei M, Ho RT, Abou-Khalil BW, et al. EEG and ECG in sudden unexplained death in epilepsy. Epilepsia 2004;45:338-45. 23. Ryvlin P, Cucherat M, Rheims S. Risk of sudden unexpected death in epilepsy in patients given adjunctive antiepileptic treatment for refractory seizures: a metaanalysis of placebo-controlled randomised trials. Lancet Neurol 2011 September 19 (Epub ahead of print). 24. Aurlien D, Larsen J, Taubøll E, Gjerstad L. Increased incidence of sudden unexpected death in epilepsy (SUDEP) with lamotrigine in Rogaland County, Norway. Epilepsia 2010;51:Suppl 3:136. abstract. 25. Opeskin K, Burke MP, Cordner SM, Berkovic SF. Comparison of antiepileptic drug levels in sudden unexpected deaths in epilepsy with deaths from other causes. Epilepsia 1999;40:1795-8. 26. Nilsson L, Bergman U, Diwan V, Farah mand BY, Persson PG, Tomson T. Antiepileptic drug therapy and its management in sudden unexpected death in epilepsy: a casecontrol study. Epilepsia 2001;42:667-73. 27. Langan Y, Nashef L, Sander JW. Sudden unexpected death in epilepsy: a series of witnessed deaths. J Neurol Neurosurg Psychiatry 2000;68:211-3.
1810
of
m e dic i n e
28. Bateman LM, Spitz M, Seyal M. Ictal
hypoventilation contributes to cardiac arrhythmia and SUDEP: report on two deaths in video-EEG-monitored patients. Epilepsia 2010;51:916-20. 29. Bird J, Dembny K, Sandeman D, Butler S. Sudden unexplained death in epilepsy: an intracranially monitored case. Epilepsia 1997;38:Suppl 11:S52-S56. 30. Espinosa PS, Lee JW, Tedrow UB, Bromfield EB, Dworetzky BA. Sudden unexpected near death in epilepsy: malignant arrhythmia from a partial seizure. Neurology 2009;72:1702-3. 31. So EL, Sam MC, Lagerlund TL. Postictal central apnea as a cause of SUDEP: evidence from near-SUDEP incident. Epilepsia 2000;41:1494-7. 32. Tao JX, Qian S, Baldwin M, et al. SUDEP, suspected positional airway obstruction, and hypoventilation in postictal coma. Epilepsia 2010;51:2344-7. 33. Thomas P, Landre E, Suisse G, Brelooin J, Dolisi C, Chatel M. Syncope anoxoischemique par dyspne: le obstructive au cours d’une crise partielle complexe temporale droite. Epilepsies 1996;8:339-46. 34. Opeskin K, Berkovic SF. Risk factors for sudden unexpected death in epilepsy: a controlled prospective study based on coroners cases. Seizure 2003;12:456-64. 35. Sevcencu C, Struijk JJ. Autonomic alterations and cardiac changes in epilepsy. Epilepsia 2010;51:725-37. 36. Hewertson J, Poets CF, Samuels MP, Boyd SG, Neville BG, Southall DP. Epileptic seizure-induced hypoxemia in infants with apparent life-threatening events. Pediatrics 1994;94:148-56. 37. Dasheiff RM, Dickinson LJ. Sudden unexpected death of epileptic patient due to cardiac arrhythmia after seizure. Arch Neurol 1986;43:194-6. 38. Bateman LM, Li CS, Seyal M. Ictal hypoxemia in localization-related epilepsy: analysis of incidence, severity and risk factors. Brain 2008;131:3239-45. 39. Thom M, Seetah S, Sisodiya S, Koepp M, Scaravilli F. Sudden and unexpected death in epilepsy (SUDEP): evidence of acute neuronal injury using HSP-70 and c-Jun immunohistochemistry. Neuropathol Appl Neurobiol 2003;29:132-43. 40. Richerson GB, Buchanan GF. The serotonin axis: shared mechanisms in seizures, depression, and SUDEP. Epilepsia 2011; 52:Suppl 1:28-38. 41. Swallow RA, Hillier CE, Smith PE. Sudden unexplained death in epilepsy (SUDEP) following previous seizure-related pulmonary oedema: case report and review of possible preventative treatment. Seizure 2002;11:446-8. 42. Johnston SC, Siedenberg R, Min JK, Jerome EH, Laxer KD. Central apnea and acute cardiac ischemia in a sheep model of epileptic sudden death. Ann Neurol 1997;42:588-94. 43. Duncan JR, Paterson DS, Hoffman JM,
et al. Brainstem serotonergic deficiency in sudden infant death syndrome. JAMA 2010; 303:430-7. 44. Buchanan GF, Hodges MR, Richerson GB. Contribution of chemosensitive serotonergic neurons to interactions between the sleep-wake cycle and respiratory control. In: Monti JM, Pandi-Perumal SR, Jacobs BL, Nutt DJ, eds. Serotonin and sleep: molecular, functional and clinical aspects. Basel, Switzerland: Birkhäuser Verlag, 2008: 529-54. 45. Uteshev VV, Tupal S, Mhaskar Y, Faingold CL. Abnormal serotonin receptor expression in DBA/2 mice associated with susceptibility to sudden death due to respiratory arrest. Epilepsy Res 2010;88:183-8. 46. Bateman LM, Li CS, Lin TC, Seyal M. Serotonin reuptake inhibitors are associated with reduced severity of ictal hypoxe mia in medically refractory partial epilepsy. Epilepsia 2010;51:2211-4. 47. Seyal M, Bateman LM, Albertson TE, Lin TC, Li CS. Respiratory changes with seizures in localization-related epilepsy: analysis of periictal hypercapnia and airflow patterns. Epilepsia 2010;51:1359-64. 48. Glasscock E, Yoo JW, Chen TT, Klassen TL, Noebels JL. Kv1.1 potassium channel deficiency reveals brain-driven cardiac dysfunction as a candidate mechanism for sudden unexplained death in epilepsy. J Neurosci 2010;30:5167-75. 49. Schuele SU. Effects of seizures on cardiac function. J Clin Neurophysiol 2009; 26:302-8. 50. Devinsky O. Effects of seizures on autonomic and cardiovascular function. Epilepsy Curr 2004;4:43-6. 51. Johnson JN, Hofman N, Haglund CM, Cascino GD, Wilde AA, Ackerman MJ. Identification of a possible pathogenic link between congenital long QT syndrome and epilepsy. Neurology 2009;72: 224-31. 52. Brotherstone R, Blackhall B, McLellan A. Lengthening of corrected QT during epileptic seizures. Epilepsia 2010;51: 221-32. 53. Surges R, Adjei P, Kallis C, et al. Pathologic cardiac repolarization in pharmacoresistant epilepsy and its potential role in sudden unexpected death in epilepsy: a case-control study. Epilepsia 2010; 51:233-42. 54. Surges R, Scott CA, Walker MC. Enhanced QT shortening and persistent tachycardia after generalized seizures. Neurology 2010;74:421-6. 55. Rugg-Gunn FJ, Simister RJ, Squirrell M, Holdright DR, Duncan JS. Cardiac arrhythmias in focal epilepsy: a prospective long-term study. Lancet 2004;364:2212-9. 56. Dünser M, Hasibeder W. Sympathetic overstimulation during critical illness: adverse effects of adrenergic stress. J Intensive Care Med 2009;24:293-316. 57. Shen HY, Li T, Boison D. A novel mouse model for sudden unexpected death in epi-
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
current concepts lepsy (SUDEP): role of impaired adenosine clearance. Epilepsia 2010;51:465-8. 58. Ridsdale L, Charlton J, Ashworth M, Richardson MP, Gulliford MC. Epilepsy mortality and risk factors for death in epilepsy: a population-based study. Br J Gen Pract 2011;61:271-8. 59. Persson H, Ericson M, Tomson T. Carbamazepine affects autonomic cardiac control in patients with newly diagnosed epilepsy. Epilepsy Res 2003;57:69-75. 60. Hesdorffer D, Tomson T, Benn E, et al. Combined analysis of risk factors for SUDEP. Epilepsia 2011;52:1150-9.
61. Reddy DS. Clinical pharmacokinetic
interactions between antiepileptic drugs and hormonal contraceptives. Expert Rev Clin Pharmacol 2010;3:183-92. 62. Sperling M, Durhakula S, Scott L, et al. Mortality after epilepsy surgery. Epilepsia 2010;51:Suppl 2:320. abstract. 63. Carlson C, Arnedo V, Cahill M, Devinsky O. Detecting nocturnal convulsions: efficacy of the MP5 monitor. Seizure 2009; 18:225-7. 64. Pacia SV, Devinsky O, Luciano DJ, Vazquez B. The prolonged QT syndrome presenting as epilepsy: a report of two
cases and literature review. Neurology 1994;44:1408-10. 65. Gayatri NA, Morrall MC, Jain V, Kashyape P, Pysden K, Ferrie C. Parental and physician beliefs regarding the provision and content of written sudden unexpected death in epilepsy (SUDEP) information. Epilepsia 2010;51:777-82. 66. Morton B, Richardson A, Duncan S. Sudden unexpected death in epilepsy (SUDEP): don’t ask, don’t tell? J Neurol Neurosurg Psychiatry 2006;77:199-202. Copyright © 2011 Massachusetts Medical Society.
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1811
Review article Mechanisms of Disease
The Human Plasma Lipidome Oswald Quehenberger, Ph.D., and Edward A. Dennis, Ph.D.
From the Departments of Medicine and Pharmacology (O.Q.) and the Departments of Chemistry and Biochemistry and Pharmacology (E.A.D.), School of Medicine, University of California, San Diego, La Jolla. Address reprint requests to Dr. Dennis at the University of California, San Diego, School of Medicine, Department of Pharmacology, 9500 Gilman Dr., La Jolla, CA 92093-0601, or at
[email protected]. N Engl J Med 2011;365:1812-23. Copyright © 2011 Massachusetts Medical Society.
C
holesterol and triglyceride levels are now assessed routinely, and physicians regularly prescribe lipid-lowering drugs to patients found to have dyslipidemia. However, the increase in the number of patients with metabolic diseases (including type 2 diabetes and obesity, which are associated with an elevated risk of cardiovascular disease) demands more detailed lipid analyses both for diagnostic purposes and for monitoring the efficacy of prescribed therapy. Human plasma comprises nucleic acids, amino acids (mostly in the form of proteins), carbohydrates (in the form of monosaccharides and disaccharides), and lipids (Fig. 1).1-3 Much is known about the first three constituents, but among the various cellular metabolites, lipids stand out because of their structural diversity and the sheer number of discrete molecular species — in the hundreds of thousands, according to some estimates4 (Fig. 2). The lipids in plasma are solubilized and dispersed through their association with specific groups of proteins. Most free fatty acids and related structures with carboxyl functional groups associate with albumin, whereas the transport and distribution of more complex lipids are accomplished by means of plasma lipoproteins.5 The structural diversity of lipids is mirrored by the enormous variation in their physiological function. The abundance of individual lipid molecular species in plasma may be indicative of the variety of specific human diseases. In this review, we discuss the application of metabolic-profiling strategies to quantitatively measure entire lipid categories in human plasma, and we examine the potential use of the diverse lipid metabolites as diagnostic tools or therapeutic targets.
Di v er si t y of L ipids in Hum a n Pl a sm a In view of the importance of lipids to biologic and pathophysiological processes, detailed knowledge of the composition and concentration of lipid metabolites in plasma would be expected to expand our diagnostic capabilities. To this end, the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), in collaboration with the National Institute of Standards and Technology (NIST), recently issued a plasma-based standard reference material (SRM 1950 Metabolites in Human Plasma) for the purpose of metabolite analysis (www.nist.gov). Historically, wide-scale lipid profiling has been difficult to perform because the diverse physical properties of lipid metabolites require a multitude of purification systems combined with a host of complex technical procedures. The evolution of lipidomics has driven the development of new analytic platforms, specifically in the area of mass spectrometry, which have streamlined these procedures and have allowed many more lipid molecules to be analyzed in great detail. A high-resolution lipidomic 1812
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
MECHANISMS OF DISEASE
Fatty acyls
Glycerolipids
Nucleic acids Amino acids
Prenol lipids
Sterol lipids
Lipids
Glycerophospholipids Carbohydrates (sugars)
Sphingolipids
Figure 1. Relative Distribution of Biologic Molecules in Human Plasma. Amino acids and nucleic acids are shown without consideration of the contribution of proteins and DNA or RNA. The relative distribution is based on weight (grams per deciliter). Data were compiled from Lentner,1 Wishart et al.,2 and Quehenberger et al.3
method based on liquid and gas chromatography coupled with mass spectrometry and designed to systematically map the mammalian lipidome was applied to the NIDDK–NIST SRM, facilitating the first comprehensive lipid analysis of human plasma.3 The in-depth analysis performed by the LIPID MAPS consortium (www.lipidmaps .org) revealed an amazingly complex plasma lipidome. Almost 600 distinct molecular species were quantified, covering the six main mammalian lipid categories: fatty acyls, glycerolipids, glycerophospholipids, sphingolipids, sterol lipids, and prenol lipids6 (Fig. 3). Their structural diversity is particularly apparent within the sphingolipid and glycerophospholipid categories, with 204 and 160 distinctly identifiable molecular species, respectively. The somewhat lower number of defined species in the glycerolipid category is largely due to the presence of complex mixtures of isobaric entities that make it difficult to define individual molecular species. The heterogeneity of complex lipids within a category is principally determined by variations in fatty acid content and head groups. The fatty acids used in the assembly of complex lipids come from several sources, including nutrient intake, release from adipose tissue, and biosynthesis.
Figure 2. Relative Distribution of Lipids in Human Plasma. Lipidomic analysis has identified, characterized, and quantified almost 600 lipid molecular species in human plasma.3 The relative distribution in each category is given on a molar basis. The nomenclature of the lipid categories conforms to the recently developed LIPID MAPS classification system.6
Pl a sm a L ipids in the Me ta bol ic S y ndrome Obesity and Glycerolipids
Glycerolipids account for a high proportion of total lipids in plasma and comprise triacylglycerols (TAGs), diacylglycerols (DAGs), and ether-linked glycerolipids. The absolute plasma concentration of TAGs, distributed between chylomicrons and very-low-density lipoprotein (VLDL), is dependent on food intake (Fig. 4). In addition, several acquired or secondary factors (including genetic determinants, uncontrolled diabetes mellitus, obesity, and sedentary lifestyle) can cause hypertriglyceridemia, a prevalent form of dyslipidemia that is frequently associated with premature coronary artery disease.7 Public surveys have revealed that the prevalence of hypertriglyceridemia among adults 20 years of age or older in the United States is approximately 36% for men and 27% for women.8 On detailed lipidomic analyses performed independently in 14 pairs of young-adult monozygotic twins who were discordant for obesity9 and in a pool of 100 healthy adults,3 a considerable number of individual glycerolipid molecular structures were implicated, including TAGs and DAGs. In these studies, TAG 50:2, 52:2, 52:3, and 52:4 were the
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1813
The
n e w e ng l a n d j o u r na l
O
m e dic i n e
of
O
O
P
P O OH HO
HO
Isopentenyl pyrophosphate
H
O
O
H3CO H3CO
H SCoA
H
Acetyl CoA
O n
RO
O
H
H
Prenol lipids (8) OH
Sterol lipids (36)
O
O R HO
OH
Fatty acids (31)
OH
Eicosanoids (76) H OH
O O X
Sphingolipids (204)
R
R1
O O
O
R2
NH H O
R3
O H O
O
O R1
Glycerolipids (73)
O
O
R2
O H O
P
O
X
O
Glycerophospholipids (160)
Figure 3. Diversity of Human Plasma Lipids. Relationships among the major mammalian lipid categories are shown by means of representative molecules from each class as examples. The diagram starts with the 2-carbon precursor acetyl coenzyme A (CoA), which is the building block for the biosynthesis of fatty acids. (SCoA indicates the thioester bond between CoA and acetic acid.) Fatty acids, in turn, become part of complex lipids — namely, glycerolipids, glycerophospholipids, sphingolipids, and sterol lipids (as steryl esters). Some fatty acids are converted to eicosanoids. A second major biosynthetic route from acetyl CoA generates the 5-carbon isoprene precursor isopentenyl pyrophosphate, which provides the building block for the prenol and sterol lipids. (The n indicates the variable number of isoprene units.) Fatty acyl–derived substituents are shown in green, isoprene-derived atoms are shown in purple, the glycerol backbone is shown in red, and the serinederived backbone is shown in blue. Arrows indicate multistep transformations among the major lipid categories, starting with acetyl CoA. Values in parentheses indicate the number of distinct analytes within each lipid category that were quantified by means of mass spectrometry in the human plasma lipidome.3
1814
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
MECHANISMS OF DISEASE
Figure 4. Plasma Lipids in the Metabolic Syndrome. Abnormal levels of plasma lipids and lipoproteins are important risk factors for metabolic and cardiovascular diseases and are targets for therapeutic intervention. Cells need cholesterol and triacylglycerol (TAG) derived from dietary sources and the liver for membrane synthesis and energy. These lipids circulate in the blood as lipoprotein particles, including chylomicrons, very-low-density lipoprotein (VLDL), lowdensity lipoprotein (LDL), and high-density lipoprotein (HDL). In circulating chylomicrons and VLDL, TAG undergoes hydrolysis, catalyzed by lipoprotein lipase (LPL), to generate a pool of free fatty acids (FFAs) that is used as an energy source in tissues, including muscle. Excess FFAs are stored in adipocytes in the form of TAGs. Such caloric abundance leads to an unopposed expansion of adipose tissue and, ultimately, to obesity and associated metabolic complications characterized by insulin resistance and diabetes. Stored TAG in adipocytes undergoes lipolysis on demand as a result of hormone-sensitive lipase (HSL), leading to an energy-balanced level of FFAs in plasma. In insulin resistance, adipocytes exhibit a high rate of lipolysis and are highly responsive to fat-mobilizing enzymes but respond poorly to lipolysisrestraining insulin. Furthermore, insulin resistance depresses adipocyte LPL activity; however, adipocytes from obese humans use compensatory mechanisms that increase the capacity for FFA transport and uptake. In combination with increased lipolysis, this process generates abnormally high plasma levels of FFAs, allowing their increased uptake into hepatocytes in excess of metabolic requirements, which leads to storage as TAG and results in hepatic steatosis and inflammation. Some TAGs are exported as VLDL, contributing to hypertriglyceridemia. Trans-palmitoleic acid may oppose some of these effects and may stimulate insulin sensitivity in muscle and liver. In general, saturated FFAs promote cardiac disorders and systemic inflammation, whereas n−3 FFAs prevent these effects. The contribution of LDL-derived cholesterol, in both its free form (FC) and its esterified form (CE), to the development of cardiovascular disease has been well described. HDL helps remove excess FC by reverse cholesterol transport (RCT), with the formation of CE by lecithin cholesterol acyltransferase (LCAT), and subsequent uptake of the CE by the liver. High levels of HDL are correlated with low cardiovascular risk.
most abundant glycerolipid species. (TAGS are defined as the ratio of the total number of carbon atoms to the total number of double bonds in the fatty acids.) When the twin study was controlled for patient age and genetic background, it showed that several (but not all) TAG molecular species, including 56:4, correlated significantly with bodymass index and measurements of subcutaneous fat and may serve as biomarkers for the early detec-
tion of acquired obesity, especially when applied to children.9 Similarly, plasma profiling of 16 adults without diabetes with a broad range of insulin sensitivity revealed that TAG species containing saturated or monounsaturated fatty acids correlate positively with insulin resistance, whereas TAGs containing essential fatty acids correlate negatively with insulin resistance (according to the homeostasis model assessment) and waist circumference.10
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1815
The
n e w e ng l a n d j o u r na l
m e dic i n e
The study of monozygotic twins included an assessment of phospholipids and showed that obesity was associated with increases in lysoglycerophosphocholine and decreases in ether glycerophospholipids.9 In addition, a strong negative correlation was seen between docosahexaenoic acid (DHA 22:6 n−3)–containing glycerophosphocholine and levels of subcutaneous and intra abdominal fat. A separate lipidomic analysis of 19 hypertensive persons and 59 normotensive control subjects showed that hypertension was also associated with a decrease in ether glycerophospholipids — specifically, the ones containing arachidonic acid (20:4 n−6) and docosapentaenoic acid (DPA 22:5 n−3).11 In a study of patients with dilated cardiomyopathy, researchers identified specific plasma lipid abnormalities; the predominant abnormality was a decrease in TAG molecular species containing odd-chain-length fatty acids.12 No marked changes in total triglycerides were observed; consequently, these specific molecular changes would remain undetected if one were using the enzymatic triglyceride assays that are routinely applied in clinical laboratories.
is the finding that unsaturated fatty acids undergo more rapid oxidation than do their saturated analogues, suggesting that the composition of fat depots may be as important as their total quantity.17 Most obese persons have elevated plasma levels of free fatty acids, owing to increased release from expanded adipose tissue.18 Long-term increases in levels of plasma free fatty acids will ultimately inhibit the antilipolytic action of insulin, which will further increase the rate of their release into the circulation and contribute to the development of insulin resistance and type 2 diabetes (Fig. 4).19 The plasma pool of free fatty acids is also an important source of lipid for hepatocytes, and any excess of free fatty acids is likely to be converted to TAGs and stored in the liver or incorporated into lipoproteins and secreted into the circulatory system.20 Predictably, raising plasma free fatty acid levels will lead to an accumulation of TAGs in the liver, which not only triggers hepatic insulin resistance but may also cause hepatic steatosis.19,21 Thus, specific free fatty acids are responsible for several serious health problems associated with obesity.22
Saturated and Unsaturated Fatty Acids in the Regulation of Energy Metabolism
Free Fatty Acid Signaling and Insulin Sensitivity
The contribution of lipids to our energy supply is quantitatively important, especially during prolonged muscular exercise, when as much as 80% of the energy is supplied by lipid oxidation.13 However, when dietary lipid intake exceeds energy expenditure, adipose lipid storage expands in the form of TAGs and results in overweight and obesity. To achieve and maintain a favorable body weight, emphasis is commonly placed on the amount of fat in the diet, whereas lipid composition is frequently ignored. However, the fatty acid composition of fats ingested may determine lipid energy substrate utilization in humans. Since the 1960s, the intake of calories in the form of fat has steadily decreased, yet during the same period, obesity has increased in the U.S. population.14,15 The molecular structure of the fatty acids determines the kinetics of their release from adipose tissue. The mobilization of fatty acids increases with the degree to which they are unsaturated, and as predicted, arachidonic acid, eicosapentaenoic acid (EPA 20:5 n−3), and DHA, the latter two being major components of fish oil, are released at high rates relative to the more saturated fatty acids.16 Consistent with this process 1816
of
Fatty acids have broad metabolic functions, owing to their ability to serve as potent signaling molecules. As described above, long-term elevations of plasma free fatty acid levels lead to insulin resistance in muscle, desensitization of adipocytes to the lipogenic effects of insulin, and steatosis in the liver (Fig. 4).19,23 Thus, plasma free fatty acids provide a metabolic link among obesity, insulin resistance, and type 2 diabetes — all wellestablished risk factors for cardiovascular disease.24 However, not all types of fatty acids contribute equally to the systemic disruption of insulin action. Epidemiologic studies indicate that fat rich in saturated fatty acids promotes insulin resistance, whereas monounsaturated and polyunsaturated fatty acids reduce it.25 A mouse model of obesity-induced insulin resistance showed that hepatic fatty acid composition was an independent determinant of insulin resistance.26 Modification of this composition by genetic inhibition of the elongation of palmitic acid (16:0) to stearic acid (18:0) ameliorated insulin resistance, even in the presence of persistent obesity and hepatosteatosis. Inhibition of elongation leads to a relative enrichment in palmitic acid, which can then serve as a substrate
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
MECHANISMS OF DISEASE
for desaturases to form monounsaturated palmitoleic acid (16:1). Studies of mice with a deficiency of the fatty acid–binding proteins aP2 and mal1 (FABP4 and FABP5, respectively) raised the possibility that local alterations in adipose tissue might regulate systemic metabolic responses.27 Although these mice were protected from the onset of diet-induced insulin resistance and fatty liver disease, total plasma free fatty acid levels paradoxically rose significantly. Targeted lipidomic analysis showed a unique lipid profile with significant enrichment of endogenous palmitoleate in adipose tissue and plasma, which led to the hypothesis that palmitoleate levels are associated with an improved metabolic profile.28 Consistent with this idea is the finding that palmitoleate stimulates basal glucose uptake and use by cultured skeletal-muscle cells.29 As compared with the findings in animal models, the results of clinical studies are less clear. One recent study suggested a strong association between circulating levels of palmitoleate and insulin sensitivity,30 whereas another showed no correlation.31 However, the latter group of investigators found a robust association between plasma levels of transpalmitoleate and insulin sensitivity.32 The divergent findings for cis-palmitoleate and trans-palmitoleate remain unexplained, but the benefits derived from the trans isomer are supported by the clinical observation that the consumption of dairy products rich in trans-palmitoleate inversely correlates with the insulin resistance syndrome.33 Fish Oil, Eicosanoids, and Inflammation
Plasma free fatty acids have also been linked to sudden cardiac death,34 and increased levels have been observed in cancer-induced cachexia and may impair lymphocyte function.35,36 Several studies in animals and humans have shown that saturated and monounsaturated fatty acids increase the risk of ventricular arrhythmia and sudden cardiac death, whereas polyunsaturated fatty acids, and n−3 fatty acids in particular, prevent this arrhythmogenic action, with a significant reduction in the risk of death from cardiac causes.37,38 Presumably, polyunsaturated fatty acids act through their eicosanoid products, including prostaglandins, leukotrienes, and other oxidized fatty acid products (Fig. 4).39 The observed beneficial effects prompted the American Heart Association to release a statement regarding the nutritional benefits of n−3 fatty acid consumption.40 Chronic inflammation in liver and adipose
tissue is common in obesity, and mounting evidence indicates that free fatty acids directly mod ulate inflammatory responses. Several groups have characterized orphan G-protein–coupled receptors (GPR) that are specifically activated by free fatty acids.41,42 A number of these receptors are expressed at high levels on leukocytes and adipose tissue and may synchronize the inflammatory action of lipopolysaccharides. Whereas saturated fatty acids amplify the proinflammatory action of lipopolysaccharides, many polyunsaturated fatty acids have antiinflammatory effects. Consistent with these findings is the observation that n−3 fatty acids, including DHA, enhance systemic insulin sensitivity in animal models of obesity by ameliorating inflammation in adipose tissue through a receptor termed GPR120.42 These findings indicate the presence of interconnecting pathways linking fatty acid metabolism to inflammation. In addition, the n−3 fatty acids have been implicated in brain function; a decrease in DHA and increased ratios of n−6 to n−3 polyunsaturated fatty acids in the serum predict the severity of some neurologic disorders including Alzheimer’s disease and other dementia disorders,43,44 major depression,45 and bipolar affective disorder.46 Furthermore, a general population–based study showed that several lipid clusters containing mainly saturated TAGs were significantly associated with schizophrenia.47
Chol e s terol , Ox idi zed S terol s, a nd C a r diova scul a r R isk Cholesterol is the most abundant sterol in plasma and exists in both free and fatty acyl–esterified forms. In plasma, cholesterol is associated with lipoproteins, including mainly low-density lipoprotein (LDL, which accounts for 60 to 70% of plasma cholesterol) and high-density lipoprotein (HDL, which accounts for 20 to 30% of plasma cholesterol), and its level in plasma is widely used to predict a person’s risk of cardiovascular events. In fact, a high level of LDL cholesterol in the blood is considered the major risk factor for coronary artery disease and stroke, making it a prominent target for therapeutic intervention. Current public health guidelines recommend that all persons 20 years of age or older be tested every 5 years to determine their fasting lipoprotein profile.48 Typical blood tests include measurement of total cholesterol, usually performed without dis-
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1817
The
n e w e ng l a n d j o u r na l
crimination between free and esterified forms. Cholesterol exists primarily in the form of fatty acyl esters; plasma contains more than 22 distinct molecular species of cholesterol ester.3 In addition to cholesterol, human plasma contains campesterol and sitosterol, which are not synthesized in humans but are derived entirely from the diet. In normal persons, plant sterols are present in only small amounts, and cholesterol accounts for more than 99% of all circulating sterols. In sitosterolemia, an autosomal recessive disorder, the levels of plant sterols in plasma are markedly elevated.49 In patients with this disorder, plasma cholesterol levels are also elevated, and xanthomatosis and premature atherosclerosis develop.50,51 These patients have a poor response to statin treatment but have a strikingly good response to treatment with bile acid resins.52 A complete plasma sterol panel is therefore necessary to determine the differential diagnosis and select the optimal treatment. Lathosterol is also present in normal plasma and can be of diagnostic value as an indicator of whole-body cholesterol synthesis.53 The rate at which cholesterol is synthesized and the differentiation of this endogenous source from dietary contributions are of considerable clinical interest, especially in assessing the efficacy of statin therapy in patients with hypercholesterolemia. Normal plasma also contains trace amounts of 25-hydroxycholesterol, which is secreted by macrophages in response to the activation of toll-like receptors.54 Exposure of B cells to this macrophagederived oxysterol suppresses immunoglobulin A production, thereby facilitating negative regulation of the adaptive immune response by the innate immune system.
Phosphol ipids a nd Sphing ol ipids in C a ncer a nd Neurol o gic Dise a se s Glycerophospholipids constitute the main components of cell membranes and serve as precursors for signaling molecules in many cellular and physiological processes. As shown in Figures 2 and 3, phospholipids in plasma are abundant and heterogeneous, properties that are largely a function of fatty acid variations as well as structural differences of the head groups. Most phospholipids found in plasma are secreted by the liver and distributed among all lipoprotein classes.5 Their structural diversity suggests critical involvement in 1818
of
m e dic i n e
many physiological and pathophysiological processes.3 Cancer cells have aberrant glycerophosphocholine metabolism, leading to an elevation of phosphocholine, an intermediate in glycerophosphocholine biosynthesis, and other cholinecontaining phospholipids.55,56 Elevated phosphocholine levels have been reported in several types of cancers and have been evaluated as a target for anticancer therapy.57-59 Moreover, phosphocholine levels may also be used as a predictor of both outcome and the aggressiveness of carcinomas60,61 and as an early marker of the response to therapy.62-64 In addition to a direct role in cancer biology, glycerophosphocholine serves as a precursor for several lipid second messengers, including phosphatidic acid, a signaling molecule implicated in the development of drug resistance that is frequently observed in rapamycin (sirolimus)–based immunosuppressive therapy and anticancer therapy.65 Phosphatidic acid can be further metabolized to lysophosphatidic acid by phospholipases A1 and A2, which hydrolyze the fatty acid at the sn-1 position and the sn-2 position, respectively.66 Lysophosphatidic acid is also formed outside cells in plasma from lysoglycerophosphocholine by autotaxin, which has lysophospholipase D activity and cleaves the choline head group from lysoglycerophosphocholine.67 Lysophosphatidic acid is a bioactive lipid that signals through specific GPR to elicit a host of cellular responses, including proliferation, survival, and migration.68 Enhanced autotaxin expression and overproduction of lysophosphatidic acid have been noted in numerous types of cancer, including ovarian cancer.69 Ascites fluid from patients with ovarian cancer contains lysophosphatidic acid in high levels,70,71 an observation that is consistent with its role in tumor biology. Lysophosphatidic acid levels are also elevated in plasma from patients with ovarian cancer but not in plasma from healthy controls72; however, a second study could not verify these findings.73 Subsequent studies supported the potential role of plasma lysophosphatidic acid as a biomarker for ovarian cancer.74-77 To date, this issue remains controversial and is the impetus for an ongoing clinical trial (ClinicalTrials.gov number, NCT00986206). Sphingolipids are complex lipids that are particularly abundant in nervous tissue. They are also found in human plasma, which contains over 200 distinct molecular species.3 Among these, the free sphingoid base sphingosine-1-phosphate (S1P) has potent messenger functions and is centrally
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
MECHANISMS OF DISEASE
involved in the trafficking of immune cells, maintenance of vascular tone, and cell communication in the central nervous system.78 It is detected in various tissues and blood and participates in numerous fundamental biologic and pathophysiological processes through five GPR S1P subtypes (S1P1 through S1P5).79 S1P regulates lymphocyte trafficking from the thymus and lymphoid organs, which requires the engagement of the S1P1receptor subtype on lymphocytes.80 Pharmacologic down-modulation of S1P1 antagonizes the action of S1P and prevents the egress of lymphocytes from the lymph node into the lymphatic circulation. This has implications for certain autoimmune diseases characterized by autoreactive lymphocytes and has led to the development of the immunosuppressive drug fingolimod (FTY720) for the treatment of multiple sclerosis.81 Fingolimod, an S1P analogue, down-modulates four of the five S1P receptors, creating temporary S1P1-null lymphocytes and thus favors their retention in lymph nodes. Sphingolipids are also implicated in a number of neurologic diseases. Recent profiling efforts established baseline levels for more than 200 sphingolipids in unfractionated plasma and across lipoprotein fractions isolated from healthy donors.3,82,83 Plasma levels of glucosylceramide and ceramide correlate with the severity of Gaucher’s disease as well as with the response to treatment,84 and it has been proposed that the recently discovered 1-deoxy-sphingolipids mediate the neurodegeneration of hereditary sensory neuropathy type 1.85 Elevated plasma levels of many of the sphingolipids have also been correlated with chronic disease — for example, sphingomyelins with coronary heart disease86 and ceramides with type 2 diabetes87 and Alzheimer’s disease.88 Even a very minor species, such as sphinganine, has been useful as a biomarker, since increased maternal serum ratios of sphinganine to sphingosine correlate with the occurrence of neural-tube defects in offspring.89
Cl inic a l Per spec t i v e s on L ipid omic s a nd Ther a peu t ic Moni t or ing The unexpectedly large number of lipid metabolites present in human plasma establishes a broad platform for the discovery of biomarkers for diseases and efficacy of treatment. Monitoring of specific sterol, cholesterol ester, and TAG
molecular species may provide more detailed diagnostic information. Moreover, a recent metaanalysis of genomewide association studies involving more than 100,000 persons identified 95 distinct gene variants associated with lipid traits in plasma that affect blood lipid levels and thus have direct relevance to cardiovascular disease.90 Combinations of common genetic variants typically contribute to extreme lipid phenotypes, highlighting the multifactorial nature of the metabolic syndrome and cardiovascular disease and the need for comprehensive analyses of lipid metabolites in plasma. Furthermore, the number of distinct human genes and their coded proteins provides a gross underestimate of the number of lipid metabolites because an individual enzyme may catalyze the synthesis of hundreds of distinct lipid species (Fig. 5). These factors, combined with wide-ranging dietary sources of lipids from the plant, animal, and marine kingdoms, as well as the contribution of the gut microbiota,91 make it clear that our lipidome is enormous.4 The plasma lipidome discussed here is just the tip of the iceberg of what will surely be a vastly expanding library. A recent study followed quantitative changes over time of some 400 lipid molecular species in a macrophage subjected to lipopolysaccharide as a model for infection.92 This systems-level view of lipid metabolism reveals important connections between lipid signaling and several biochemical pathways that contribute to innate immune responses in inflammation. The production of eicosanoids is robustly induced by proinflammatory stimuli, and the blocking of underlying signaling events is one of the strategies used by clinicians to manage acute and chronic inflammation. Although plasma eicosanoid concentrations are generally low in the healthy state, circulating eicosanoid levels may increase in disease states as a result of spillover from inflamed tissues. Eicosanoids are also secreted from tissues as part of the catabolic and secretory pathways, and stable metabolites may accumulate in plasma. Isoprostanes derived from nonenzymatic pathways have been found in blood and urine and are currently regarded as sensitive biomarkers of oxidative stress.93 In addition, a single-nucleotide polymorphism in the gene encoding prostaglandin E synthase 2 correlates with increased blood prostaglandin E2 levels, at least in females, and has been associated with type 2 diabetes in several independent cohorts.94,95 Not only do eicosanoid levels change,
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1819
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
Genes
Eicosanoid metabolites Figure 5. Eicosanoid Metabolic Network. The number of genes and proteins in a biosynthetic pathway does not accurately reflect the enormous diversity of the lipidome. For example, in the eicosanoid biosynthetic pathways, 28 known genes and their corresponding enzymatic gene products (green dots) are responsible for the production of more than 150 bioactive lipids (yellow dots) derived from dihomo-gamma-linolenic acid (20:3), arachidonic acid (20:4), eicosapentaenoic acid (20:5), and docosahexaenoic acid (22:6),39 of which 76 have been detected in normal human plasma.3 (Fatty acids are defined by the ratio of the number of carbon atoms to the number of double bonds.) A similar discrepancy between the small numbers of genes and the correspondingly larger number of individual lipid species is observed with all other lipid categories, and overlapping enzymatic activities explain the fact that a relatively small set of enzymes generates a vast number of distinct lipid species with defined molecular structures and unique biologic functions.
but their patterns of expression may also change, depending on the type of inflammation. Therefore, a complete plasma profile of eicosanoids and their breakdown products should provide an accurate metabolic snapshot. The development of cancer involves aberrant regulation of multiple cellular processes, including cell proliferation, survival, migration, and invasion and angiogenesis. The actions of a number 1820
of phospholipids are concordant with some or all of these hallmarks of cancer, and the composition and level of phospholipids correlate with various types of cancer. Blood is an important reservoir for extracellular S1P, and specific antibodies can potently deplete S1P in blood and other tissues. Several of these antibodies have reduced tumor progression in animal models and have been formulated for clinical trials of treat-
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
MECHANISMS OF DISEASE
ment targeting cancer and age-related macular degeneration.96 S1P also plays an important role in regulating the egress of immune cells from lymphoid tissue into blood. As described above, the S1P analogue fingolimod is approved for the treatment of multiple sclerosis81,97,98 and might have a clinical application in other neurodegenerative diseases. For some neurologic disorders, the initial screening, which often consists of invasive clinical procedures, could be complemented with noninvasive analysis of lipid metabolites in blood. For example, the plasma concentrations of glucosylceramide are markedly elevated in patients with symptomatic type 1 Gaucher’s disease.84 Notably, in one study, all patients had a response to enzymereplacement therapy 99; within 6 months, plasma glycosylceramide levels were significantly reduced, and these reductions were correlated with clinical responses. Monitoring plasma glucosylceramide levels during therapy is of particular importance because several sphingolipids are effective inhibitors of thrombin generation,100 which would explain the bleeding tendency in patients with Gaucher’s disease.101
C onclusions Lipid signaling pathways are complex. In the coming years, understanding the integrated lipidomic networks and decoding the coordinately regulated pathways will constitute major goals. The modulation of intracellular and systemic lipid metabolism has enormous therapeutic potential. Recent
technological advances have provided the tools for broad lipid profiling as part of the effort to discover biomarkers, which has yielded promising results.102 A substantial portion of the human lipidome has now been identified, and methods are available to permit the accurate quantitation of individual lipid molecular species. For the most part, lipidomic studies are performed in specialized laboratories with the use of multiple analytic platforms. Unlike the genome and transcriptome, which can be readily measured with automated array technology, lipids require specific analytic procedures, owing to their diverse structural and physical properties. The next challenge will be to translate the findings of lipidomic laboratories into medical applications and to introduce these technologies into clinical laboratories. This will require the standardization of sample preparation, the streamlining of analytic procedures, and the establishment of metabolite databases for cross-reference. Despite the work ahead, the use of lipidomics as a diagnostic tool should expand enormously in the coming years as modern medicine evolves to incorporate personalized treatment plans. Disclosure forms provided by the authors are available with the full text of this article at NEJM.org. We thank our LIPID MAPS colleagues Alex H. Brown, Christopher K. Glass, Alfred H. Merrill, Robert C. Murphy, Christian R.H. Raetz, David W. Russell, Walter A. Shaw, Shankar Subramaniam, Michael S. VanNieuwenhze, Nicholas Winograd, and Joseph L. Witztum for providing valuable scientific input; Dr. M. John Chapman, Pierre and Marie Curie University, for valuable insight and encouragement to write about the plasma lipidome; Yasuyuki Kihara for helping with an earlier version of Figure 5; and Maria Farron for helping with earlier versions of the other figures.
References 1. Lentner C, ed. Geigy scientific tables.
Vol. 3. Physical chemistry, composition of blood, hematology, somatometric data. West Caldwell, NJ: Ciba-Geigy, 1981. 2. Wishart DS, Knox C, Guo AC, et al. HMDB: a knowledgebase for the human metabolome. Nucleic Acids Res 2009;37: D603-D610. 3. Quehenberger O, Armando AM, Brown AH, et al. Lipidomics reveals a remarkable diversity of lipids in human plasma. J Lipid Res 2010;51:3299-305. 4. Dennis EA. Lipidomics joins the omics evolution. Proc Natl Acad Sci U S A 2009; 106:2089-90. 5. Havel RJ, Eder HA, Bragdon JH. The distribution and chemical composition of ultracentrifugally separated lipoproteins in human serum. J Clin Invest 1955;34: 1345-53. 6. Fahy E, Subramaniam S, Murphy RC, et al. Update of the LIPID MAPS compre-
hensive classification system for lipids. J Lipid Res 2009;50:Suppl:S9-S14. 7. Hopkins PN, Heiss G, Ellison RC, et al. Coronary artery disease risk in familial combined hyperlipidemia and familial hypertriglyceridemia: a case-control comparison from the National Heart, Lung, and Blood Institute Family Heart Study. Circulation 2003;108:519-23. 8. Ervin RB. Prevalence of metabolic syndrome among adults 20 years of age and over, by sex, age, race and ethnicity, and body mass index: United States, 20032006. Natl Health Stat Rep 2009;13:1-7. 9. Pietiläinen KH, Sysi-Aho M, Rissanen A, et al. Acquired obesity is associated with changes in the serum lipidomic profile independent of genetic effects — a monozygotic twin study. PLoS ONE 2007;2(2): e218. 10. Kotronen A, Velagapudi VR, Yetukuri L, et al. Serum saturated fatty acids contain-
ing triacylglycerols are better markers of insulin resistance than total serum triacylglycerol concentrations. Diabetologia 2009; 52:684-90. 11. Graessler J, Schwudke D, Schwarz PE, Herzog R, Shevchenko A, Bornstein SR. Top-down lipidomics reveals ether lipid deficiency in blood plasma of hypertensive patients. PLoS ONE 2009;4(7):e6261. 12. Sysi-Aho M, Koikkalainen J, SeppänenLaakso T, et al. Serum lipidomics meets cardiac magnetic resonance imaging: profiling of subjects at risk of dilated cardiomyopathy. PLoS ONE 2011;6(1):e15744. 13. Pirnay F, Lacroix M, Mosora F, Luyckx A, Lefebvre P. Effect of glucose ingestion on energy substrate utilization during prolonged muscular exercise. Eur J Appl Physiol Occup Physiol 1977;36:247-54. 14. Department of Agriculture, Center for Nutrition Policy and Promotion. Is total fat consumption really decreasing? Nutrition
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1821
The
n e w e ng l a n d j o u r na l
Insights. April 1998 (http://www.cnpp.usda .gov/Publications/NutritionInsights/insight5 .pdf). 15. Ogden CL, Carroll MD, McDowell MA, Flegal KM. Obesity among adults in the United States — no statistically significant chance since 2003-2004. National Center for Health Statistics Data Brief. November 2007 (http://www.cdc.gov/nchs/ data/databriefs/db01.pdf ). 16. Conner WE, Lin DS, Colvis C. Differential mobilization of fatty acids from adipose tissue. J Lipid Res 1996;37:290-8. 17. DeLany JP, Windhauser MM, Champagne CM, Bray GA. Differential oxidation of individual dietary fatty acids in humans. Am J Clin Nutr 2000;72:905-11. 18. Reaven GM, Hollenbeck C, Jeng CY, Wu MS, Chen YD. Measurement of plasma glucose, free fatty acid, lactate, and insulin for 24 h in patients with NIDDM. Diabetes 1988;37:1020-4. 19. Boden G. Fatty acid-induced inflammation and insulin resistance in skeletal muscle and liver. Curr Diab Rep 2006;6: 177-81. 20. Donnelly KL, Smith CI, Schwarzenberg SJ, Jessurun J, Boldt MD, Parks EJ. Sources of fatty acids stored in liver and secreted via lipoproteins in patients with nonalcoholic fatty liver disease. J Clin Invest 2005;115:1343-51. 21. Mulhall BP, Ong JP, Younossi ZM. Non-alcoholic fatty liver disease: an overview. J Gastroenterol Hepatol 2002;17: 1136-43. 22. Santomauro AT, Boden G, Silva ME, et al. Overnight lowering of free fatty acids with Acipimox improves insulin resistance and glucose tolerance in obese diabetic and nondiabetic subjects. Diabetes 1999;48:1836-41. 23. Ginsberg HN, Zhang YL, HernandezOno A. Metabolic syndrome: focus on dyslipidemia. Obesity (Silver Spring) 2006;14: Suppl 1:41S-49S. 24. Charles MA, Eschwège E, Thibult N, et al. The role of non-esterified fatty acids in the deterioration of glucose tolerance in Caucasian subjects: results of the Paris Prospective Study. Diabetologia 1997;40: 1101-6. 25. Riccardi G, Giacco R, Rivellese AA. Dietary fat, insulin sensitivity and the metabolic syndrome. Clin Nutr 2004;23:447-56. 26. Matsuzaka T, Shimano H, Yahagi N, et al. Crucial role of a long-chain fatty acid elongase, Elovl6, in obesity-induced insulin resistance. Nat Med 2007;13:1193-202. 27. Maeda K, Cao H, Kono K, et al. Adipocyte/macrophage fatty acid binding proteins control integrated metabolic responses in obesity and diabetes. Cell Metab 2005;1: 107-19. 28. Cao H, Gerhold K, Mayers JR, Wiest MM, Watkins SM, Hotamisligil GS. Identification of a lipokine, a lipid hormone linking adipose tissue to systemic metabolism. Cell 2008;134:933-44. 29. Dimopoulos N, Watson M, Sakamoto K, Hundal HS. Differential effects of palmitate
1822
of
m e dic i n e
and palmitoleate on insulin action and glucose utilization in rat L6 skeletal muscle cells. Biochem J 2006;399:473-81. 30. Stefan N, Kantartzis K, Celebi N, et al. Circulating palmitoleate strongly and independently predicts insulin sensitivity in humans. Diabetes Care 2010;33:405-7. 31. Mozaffarian D, Cao H, King IB, et al. Circulating palmitoleic acid and risk of metabolic abnormalities and new-onset diabetes. Am J Clin Nutr 2010;92:1350-8. 32. Idem. Trans-palmitoleic acid, metabolic risk factors, and new-onset diabetes in U.S. adults: a cohort study. Ann Intern Med 2010;153:790-9. 33. Pereira MA, Jacobs DR Jr, Van Horn L, Slattery ML, Kartashov AI, Ludwig DS. Dairy consumption, obesity, and the insulin resistance syndrome in young adults: the CARDIA Study. JAMA 2002;287:2081-9. 34. Leaf A. Plasma nonesterified fatty acid concentration as a risk factor for sudden cardiac death: the Paris Prospective Study. Circulation 2001;104:744-5. 35. Legaspi A, Jeevanandam M, Starnes HF Jr, Brennan MF. Whole body lipid and energy metabolism in the cancer patient. Metabolism 1987;36:958-63. 36. Kleinfeld AM, Okada C. Free fatty acid release from human breast cancer tissue inhibits cytotoxic T-lymphocyte-mediated killing. J Lipid Res 2005;46:1983-90. 37. McLennan PL. Relative effects of die tary saturated, monounsaturated, and polyunsaturated fatty acids on cardiac arrhythmias in rats. Am J Clin Nutr 1993;57:207-12. 38. Burr ML, Fehily AM, Gilbert JF, et al. Effects of changes in fat, fish, and fibre intakes on death and myocardial reinfarction: Diet and Reinfarction Trial (DART). Lancet 1989;2:757-61. 39. Buczynski MW, Dumlao DS, Dennis EA. Thematic review series: proteomics — an integrated omics analysis of eicosanoid biology. J Lipid Res 2009;50:1015-38. [Erratum, J Lipid Res 2009;50:1505.] 40. Kris-Etherton PM, Harris WS, Appel LJ. Fish consumption, fish oil, omega-3 fatty acids, and cardiovascular disease. Circulation 2002;106:2747-57. [Erratum, Circulation 2003;107:512.] 41. Itoh Y, Kawamata Y, Harada M, et al. Free fatty acids regulate insulin secretion from pancreatic beta cells through GPR40. Nature 2003;422:173-6. 42. Oh DY, Talukdar S, Bae EJ, et al. GPR120 is an omega-3 fatty acid receptor mediating potent anti-inflammatory and insulinsensitizing effects. Cell 2010;142:687-98. 43. Schaefer EJ, Bongard V, Beiser AS, et al. Plasma phosphatidylcholine docosahexaenoic acid content and risk of dementia and Alzheimer disease: the Framingham Heart Study. Arch Neurol 2006;63:1545-50. 44. Barberger-Gateau P, Raffaitin C, Letenneur L, et al. Dietary patterns and risk of dementia: the Three-City cohort study. Neurology 2007;69:1921-30. 45. Maes M, Smith R, Christophe A, Cosyns P, Desnyder R, Meltzer H. Fatty acid composition in major depression: decreased
omega 3 fractions in cholesteryl esters and increased C20: 4 omega 6/C20:5 omega 3 ratio in cholesteryl esters and phospholipids. J Affect Disord 1996;38:35-46. 46. Stoll AL, Severus WE, Freeman MP, et al. Omega 3 fatty acids in bipolar disorder: a preliminary double-blind, placebocontrolled trial. Arch Gen Psychiatry 1999; 56:407-12. 47. Orešič M, Tang J, Seppänen-Laakso T, et al. Metabolome in schizophrenia and other psychotic disorders: a general population-based study. Genome Med 2011;3:19. 48. Executive summary of the Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). JAMA 2001;285:2486-97. 49. Bhattacharyya AK, Connor WE. Betasitosterolemia and xanthomatosis: a newly described lipid storage disease in two sisters. J Clin Invest 1974;53:1033-43. 50. Berge KE, Tian H, Graf GA, et al. Accumulation of dietary cholesterol in sitosterolemia caused by mutations in adjacent ABC transporters. Science 2000;290:1771-5. 51. Rader DJ, Cohen J, Hobbs HH. Monogenic hypercholesterolemia: new insights in pathogenesis and treatment. J Clin Invest 2003;111:1795-803. 52. Watts GF, Mitchell WD, Bending JJ, Reshef A, Leitersdorf E. Cerebrotendinous xanthomatosis: a family study of sterol 27-hydroxylase mutations and pharmacotherapy. QJM 1996;89:55-63. 53. Björkhem I, Miettinen T, Reihnér E, Ewerth S, Angelin B, Einarsson K. Correlation between serum levels of some cholesterol precursors and activity of HMGCoA reductase in human liver. J Lipid Res 1987;28:1137-43. 54. Bauman DR, Bitmansour AD, McDonald JG, Thompson BM, Liang G, Russell DW. 25-Hydroxycholesterol secreted by macrophages in response to Toll-like receptor activation suppresses immunoglobulin A production. Proc Natl Acad Sci U S A 2009; 106:16764-9. 55. Negendank W. Studies of human tumors by MRS: a review. NMR Biomed 1992; 5:303-24. 56. Ackerstaff E, Glunde K, Bhujwalla ZM. Choline phospholipid metabolism: a target in cancer cells? J Cell Biochem 2003;90: 525-33. 57. Nakagami K, Uchida T, Ohwada S, et al. Increased choline kinase activity and elevated phosphocholine levels in human colon cancer. Jpn J Cancer Res 1999;90:419-24. 58. Ramírez de Molina A, Gutiérrez R, Ramos MA, et al. Increased choline kinase activity in human breast carcinomas: clinical evidence for a potential novel antitumor strategy. Oncogene 2002;21:4317-22. 59. Iorio E, Ricci A, Bagnoli M, et al. Activation of phosphatidylcholine cycle enzymes in human epithelial ovarian cancer cells. Cancer Res 2010;70:2126-35. 60. Ramírez de Molina A, SarmenteroEstrada J, Belda-Iniesta C, et al. Expres-
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
MECHANISMS OF DISEASE sion of choline kinase alpha to predict outcome in patients with early-stage nonsmall-cell lung cancer: a retrospective study. Lancet Oncol 2007;8:889-97. 61. Hernando E, Sarmentero-Estrada J, Koppie T, et al. A critical role for choline kinase-alpha in the aggressiveness of bladder carcinomas. Oncogene 2009;28:2425-35. 62. Jagannathan NR, Kumar M, Seenu V, et al. Evaluation of total choline from invivo volume localized proton MR spectroscopy and its response to neoadjuvant chemotherapy in locally advanced breast cancer. Br J Cancer 2001;84:1016-22. 63. Griffiths JR, Tate AR, Howe FA, Stubbs M. Magnetic resonance spectroscopy of cancer-practicalities of multi-centre trials and early results in non-Hodgkin’s lymphoma. Eur J Cancer 2002;38:2085-93. 64. Murphy PS, Viviers L, Abson C, et al. Monitoring temozolomide treatment of low-grade glioma with proton magnetic resonance spectroscopy. Br J Cancer 2004; 90:781-6. 65. Foster DA. Phosphatidic acid signaling to mTOR: signals for the survival of human cancer cells. Biochim Biophys Acta 2009; 1791:949-55. 66. Schaloske RH, Dennis EA. The phospholipase A2 superfamily and its group numbering system. Biochim Biophys Acta 2006;1761:1246-59. 67. Tokumura A, Majima E, Kariya Y, et al. Identification of human plasma lysophospholipase D, a lysophosphatidic acidproducing enzyme, as autotaxin, a multifunctional phosphodiesterase. J Biol Chem 2002;277:39436-42. 68. Moolenaar WH, van Meeteren LA, Giepmans BN. The ins and outs of lysophosphatidic acid signaling. Bioessays 2004; 26:870-81. 69. Mills GB, Moolenaar WH. The emerging role of lysophosphatidic acid in cancer. Nat Rev Cancer 2003;3:582-91. 70. Xu Y, Gaudette DC, Boynton JD, et al. Characterization of an ovarian cancer activating factor in ascites from ovarian cancer patients. Clin Cancer Res 1995;1: 1223-32. 71. Mills GB, May C, Hill M, Campbell S, Shaw P, Marks A. Ascitic fluid from human ovarian cancer patients contains growth factors necessary for intraperitoneal growth of human ovarian adenocarcinoma cells. J Clin Invest 1990;86:851-5. 72. Xu Y, Shen Z, Wiper DW, et al. Lysophosphatidic acid as a potential biomarker for ovarian and other gynecologic cancers. JAMA 1998;280:719-23. 73. Baker DL, Morrison P, Miller B, et al. Plasma lysophosphatidic acid concentration and ovarian cancer. JAMA 2002;287:3081-2. 74. Sutphen R, Xu Y, Wilbanks GD, et al. Lysophospholipids are potential biomarkers of ovarian cancer. Cancer Epidemiol Biomarkers Prev 2004;13:1185-91.
75. Pozlep B, Meleh M, Kobal B, et al. Use
of lysophosphatidic acid in the management of benign and malignant ovarian tumors. Eur J Gynaecol Oncol 2007;28:394-9. 76. Sedláková I, Vávrová J, Tošner J, Hanousek L. Lysophosphatidic acid (LPA) — a perspective marker in ovarian cancer. Tumour Biol 2011;32:311-6. 77. Bese T, Barbaros M, Baykara E, et al. Comparison of total plasma lysophosphatidic acid and serum CA-125 as a tumor marker in the diagnosis and follow-up of patients with epithelial ovarian cancer. J Gynecol Oncol 2010;21:248-54. 78. Lee MJ, Van Brocklyn JR, Thangada S, et al. Sphingosine-1-phosphate as a ligand for the G protein-coupled receptor EDG-1. Science 1998;279:1552-5. 79. Brinkmann V. Sphingosine 1-phosphate receptors in health and disease: mechanistic insights from gene deletion studies and reverse pharmacology. Pharmacol Ther 2007;115:84-105. 80. Matloubian M, Lo CG, Cinamon G, et al. Lymphocyte egress from thymus and peripheral lymphoid organs is dependent on S1P receptor 1. Nature 2004;427:355-60. 81. Brinkmann V, Billich A, Baumruker T, et al. Fingolimod (FTY720): discovery and development of an oral drug to treat multiple sclerosis. Nat Rev Drug Discov 2010; 9:883-97. 82. Hammad SM, Pierce JS, Soodavar F, et al. Blood sphingolipidomics in healthy humans: impact of sample collection methodology. J Lipid Res 2010;51:3074-87. 83. Scherer M, Böttcher A, Schmitz G, Liebisch G. Sphingolipid profiling of human plasma and FPLC-separated lipoprotein fractions by hydrophilic interaction chromatography tandem mass spectrometry. Biochim Biophys Acta 2011;1811:68-75. 84. Groener JE, Poorthuis BJ, Kuiper S, Hollak CE, Aerts JM. Plasma glucosyl ceramide and ceramide in type 1 Gaucher disease patients: correlations with disease severity and response to therapeutic intervention. Biochim Biophys Acta 2008; 1781:72-8. 85. Penno A, Reilly MM, Houlden H, et al. Hereditary sensory neuropathy type 1 is caused by the accumulation of two neurotoxic sphingolipids. J Biol Chem 2010;285: 11178-87. 86. Yeboah J, McNamara C, Jiang XC, et al. Association of plasma sphingomyelin levels and incident coronary heart disease events in an adult population: Multi-Ethnic Study of Atherosclerosis. Arterioscler Thromb Vasc Biol 2010;30:628-33. 87. Haus JM, Kashyap SR, Kasumov T, et al. Plasma ceramides are elevated in obese subjects with type 2 diabetes and correlate with the severity of insulin resistance. Diabetes 2009;58:337-43. 88. Ewers M, Mielke MM, Hampel H. Blood-based biomarkers of microvascular
pathology in Alzheimer’s disease. Exp Gerontol 2010;45:75-9. 89. Missmer SA, Suarez L, Felkner M, et al. Exposure to fumonisins and the occurrence of neural tube defects along the TexasMexico border. Environ Health Perspect 2006;114:237-41. 90. Teslovich TM, Musunuru K, Smith AV, et al. Biological, clinical and population relevance of 95 loci for blood lipids. Nature 2010;466:707-13. 91. Ley RE, Hamady M, Lozupone C, et al. Evolution of mammals and their gut microbes. Science 2008;320:1647-51. [Erratum, Science 2008;322:1188.] 92. Dennis EA, Deems RA, Harkewicz R, et al. A mouse macrophage lipidome. J Biol Chem 2010;285:39976-85. 93. Roberts LJ, Morrow JD. Measurement of F(2)-isoprostanes as an index of oxidative stress in vivo. Free Radic Biol Med 2000;28:505-13. 94. Nitz I, Fisher E, Grallert H, et al. Association of prostaglandin E synthase 2 (PTGES2) Arg298His polymorphism with type 2 diabetes in two German study populations. J Clin Endocrinol Metab 2007;92: 3183-8. 95. Boomgaarden I, Bosy-Westphal A, Müller MJ, Doring F. Influence of a type 2 diabetes associated prostaglandin E synthase 2 polymorphism on blood prostaglandin E2 levels. Prostaglandins Leukot Essent Fatty Acids 2009;80:185-8. 96. O’Brien N, Jones ST, Williams DG, et al. Production and characterization of monoclonal anti-sphingosine-1-phosphate antibodies. J Lipid Res 2009;50:2245-57. 97. Kappos L, Radue EW, O’Connor P, et al. A placebo-controlled trial of oral fingolimod in relapsing multiple sclerosis. N Engl J Med 2010;362:387-401. 98. Cohen JA, Barkhof F, Comi G, et al. Oral fingolimod or intramuscular interferon for relapsing multiple sclerosis. N Engl J Med 2010;362:402-15. 99. de Fost M, Hollak CE, Groener JE, et al. Superior effects of high-dose enzyme replacement therapy in type 1 Gaucher disease on bone marrow involvement and chito triosidase levels: a 2-center retrospective analysis. Blood 2006;108:830-5. 100. Deguchi H, Yegneswaran S, Griffin JH. Sphingolipids as bioactive regulators of thrombin generation. J Biol Chem 2004; 279:12036-42. 101. Hollak CE, Levi M, Berends F, Aerts JM, van Oers MH. Coagulation abnormalities in type 1 Gaucher disease are due to low-grade activation and can be partly restored by enzyme supplementation therapy. Br J Haematol 1997;96:470-6. 102. Harkewicz R, Dennis EA. Applications of mass spectrometry to lipids and membranes. Annu Rev Biochem 2011;80: 301-25. Copyright © 2011 Massachusetts Medical Society.
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1823
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
images in clinical medicine
Bryant’s and Stabler’s Signs after a Difficult Delivery A
Stijn R.G. Heyman, M.D. Dirk Vervloessem, M.D. ZNA Queen Paola Children’s Hospital Antwerp, Belgium
[email protected]
B
C
A
neonate presented after birth with scrotal and right inguinal ecchymosis. He had been born at term after a difficult vaginal delivery that ultimately required both fundal pressure and vacuum extraction. The thorax was not manipulated during birth. Despite the challenging delivery, the physical examination at birth was normal. A few hours later, however, the child appeared to be in discomfort, with moaning, and was newly noted to have cyanosis. Reexamination revealed ecchymosis of the scrotum and right inguinal region (Panel A). Laboratory results showed a decrease in the hematocrit from 41.2% at birth to 32.5%, normal measures of blood coagulation, and slightly elevated liver enzymes. Ultrasonography revealed three subcapsular lacerations of the liver (in the quadrate lobe and in segments 5 through 7). This case demonstrates an example of Bryant’s sign (scrotal ecchymosis) and Stabler’s sign (inguinal ecchymosis) caused by retroperitoneal hemorrhage. In this case, the retroperitoneal bleeding was caused by a perinatal liver laceration. The infant required no specific intervention and was discharged home after 1 week, with improvement of the inguinal ecchymosis (Panel B). Follow-up after 1 month revealed no remaining ecchymosis (Panel C). Copyright © 2011 Massachusetts Medical Society.
1824
n engl j med 365;19
nejm.org
The New England Journal of Medicine
november 10, 2011
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
case records of the massachusetts general hospital Founded by Richard C. Cabot Nancy Lee Harris, m.d., Editor Eric S. Rosenberg, m.d., Associate Editor Jo-Anne O. Shepard, m.d., Associate Editor Alice M. Cort, m.d., Associate Editor Sally H. Ebeling, Assistant Editor Christine C. Peters, Assistant Editor
Case 34-2011: A 75-Year-Old Man with Memory Loss and Partial Seizures Sydney S. Cash, M.D., Ph.D., Mykol Larvie, M.D., Ph.D., and Josep Dalmau, M.D., Ph.D.
Pr e sen tat ion of C a se A 75-year-old physician was seen in an outpatient office at this hospital because of memory loss and episodes of near-syncope. The patient had been generally well, except for mild and gradual memory loss, until 7 months earlier, when episodes of diffuse tingling and a sensation of flushing began to occur intermittently, lasting about a minute. Light-headedness and difficulties with word finding developed 4 months before presentation in association with the episodes and progressed to near-syncope 4 weeks before presentation. The patient reported an intermittently irregular pulse but no other speech, vision, other sensory, or motor symptoms. The patient had borderline hypertension, lumbar disk disease, pseudogout, and decreased hearing. Four months earlier, he had slipped on ice, with no light-headedness or loss of consciousness, and ruptured his right quadriceps tendon. Medications included ibuprofen or naproxen, acetaminophen with codeine, and diazepam. He used hearing aids. He was allergic to penicillin. He lived with his wife and had retired from his medical practice 1 year earlier because of memory loss. He reported anxiety and mild depression after retirement. He drank 30 to 60 ml of alcohol daily and had stopped smoking years earlier. His father had died at 94 years of age from cerebrovascular disease, his mother had died at 76 years of age from a brain tumor, and two of his sisters had died in their 50s or 60s from strokes. Another sister and her children were healthy. On examination, the blood pressure was 130/84 mm Hg, and the pulse 64 beats per minute. The temperature, respiratory rate, and general physical examination were normal. The erythrocyte sedimentation rate, results of testing for antinuclear antibodies, and urinalysis were normal. An electrocardiogram showed nonspecific ST-segment and T-wave changes and was otherwise normal. Two days later, the blood pressure was 188/96 mm Hg. Noninvasive carotid testing revealed bilateral minimal stenosis (1 to 19%) of the internal carotid arteries. Echocardiography revealed a mildly dilated left atrium, an estimated left ventricular ejection fraction of 63%, thickened aortic leaflets, trace aortic and tricuspid insufficiency, and a thickened interatrial septum thought to be consistent with lipomatous hypertrophy. Magnetic resonance imaging (MRI) of the brain without the administration of contrast material revealed scattered T2-weighted hyperintense foci in the deep and subcortical cerebral white matter bilaterally, suggestive of mild microangiopathic changes, n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
From the Departments of Neurology (S.S.C.) and Radiology (M.L.), Massachusetts General Hospital; and the Departments of Neurology (S.S.C.) and Radiology (M.L.), Harvard Medical School — both in Boston; and the Division of Neuro-oncology, Department of Neurology, Hospital of the University of Pennsylvania, University of Pennsylvania, Philadelphia (J.D.). N Engl J Med 2011;365:1825-33. Copyright © 2011 Massachusetts Medical Society.
1825
The
n e w e ng l a n d j o u r na l
and no associated restricted diffusion. Magnetic resonance angiography of the head showed no hemodynamically significant intracranial arterial stenosis. Twelve days after his initial visit, the patient saw a neurologist. The patient, who was right-handed, reported that he had spells that lasted up to 1 minute that were preceded or followed by olfactory hallucinations of food and occasionally abnormal tastes. During the spells, he noted abnormal movements of the right arm, with tightening of the right shoulder and elbow, involuntary supination of the right hand, flexion of the right fingers, and jerking of the right hand when writing, in addition to slowed speech and word-finding difficulties, facial paresthesias (greater on the right side), and facial twitching on the right side. There was no impairment of consciousness or disturbance of speech, but he described feeling faint, light-headed, generally weak, and sometimes diaphoretic. His sleep became irregular, with awakening after 2 hours, occasionally in the midst of a spell. The symptoms were especially likely to recur when he was tense or worried. He reported short-term memory loss, fatigue, anorexia, erectile dysfunction, and night sweats. On examination, the blood pressure was 140/90 mm Hg while the patient was sitting, and the pulse 84 beats per minute with 1 premature beat. He had decreased hearing bilaterally and 1+ ankle reflexes; the remainder of the neurologic examination was normal. The hematocrit was 40.7% (reference range, 42 to 52), and the erythrocyte sedimentation rate 22 mm per hour (reference range, <15); the remainder of the complete blood count was normal, as were the serum levels of electrolytes, calcium, magnesium, phosphorus, and glucose and tests of renal function. Results of timed (24-hour) urine collections for vanillylmandelic acid, epinephrine, norepinephrine, dopamine, and metanephrine, performed 5 and 7 weeks after presentation, were normal. An ambulatory electroencephalogram (48-hour duration, with nine symptomatic episodes) showed no seizure activity. The administration of levetiracetam (500 mg twice daily) was begun, and diazepam was stopped. Episodes of speaking difficulties continued daily, and fatigue and anxiety increased. The administration of fluoxetine was begun. Seven weeks after the initial presentation, the patient returned to the neurologist for follow-up. The hematocrit was stable. The serum chloride level was 98 mmol per liter (reference range, 1826
of
m e dic i n e
100 to 108), the calcium level 11.0 mg per deciliter (2.8 mmol per liter) (reference range, 8.5 to 10.5 mg per deciliter [2.1 to 2.6 mmol per liter]), and the albumin level 5.1 g per deciliter (reference range, 3.3 to 5.0); testing for syphilis was negative. The complete blood count and measurements of serum electrolytes, glucose, total protein, globulin, and thyrotropin were normal, as were tests of renal and liver function. MRI of the brain after the administration of contrast material revealed features that were the same as those seen in the previous study, including increased T2-weighted signal intensity and mild expansion of the left amygdala and hippocampus without enhancement. There was also a mild, nonspecific change in the white matter. The dose of levetiracetam was increased to 1500 mg per day. The patient was referred to the Memory Disorders Unit. He reported increasing difficulty managing his finances. His wife reported that he had increasing memory loss, with difficulty remembering familiar telephone numbers, and increased irritability. The blood pressure was 110/60 mm Hg, and the pulse was 64 beats per minute and occasionally irregular. The results of the examination were unchanged. Neuropsychological testing revealed variability in the patient’s performances across domains and across 2 days of testing. There were declines in attention, executive functioning, and memory relative to his estimated premorbid functioning, although many of his performances were within the average range. His only impaired performances were copying a complex figure, recalling a list of words, and naming pictured objects with and without cueing. Twitching of the right side of the face, dysarthria, and posturing of the right hand were observed during testing. A paraneoplastic antibody panel was negative and included testing for acetylcholine receptor binding antibody, acetylcholine receptor ganglionic neuronal antibody, striated muscle antibody, N-type and P/Q-type calcium-channel antibodies, amphiphysin antibodies, CRMP-5 (IgG), and anti–neuronal nuclear and anti–Purkinje-cell antibodies. Two weeks later, approximately 11 weeks after the initial presentation, MRI of the brain after the administration of contrast material, magnetic resonance spectroscopy, and perfusion imaging revealed no definite spectroscopic or perfusion abnormality and no change in the previously described findings in the left temporal lobe. Computed tomography (CT) of the chest and
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
case records of the massachusetts gener al hospital
Table 1. Cerebrospinal Fluid Analysis.* Variable
Reference Range, Adults†
12 Wk after Initial Presentation
Colorless, clear
Pale pink, slightly turbid
Appearance Red cells, tube 1 (per mm
3)
White cells, tube 1 (per mm3)
0
985
0–5
8
0
92
Differential count, tube 1 (%) Lymphocytes Monocytes Protein (mg/dl)
0
8
5–55
79
Glucose (mg/dl)
50–75
69
Cytologic analysis
No malignant cells
No malignant cells
None detected
None detected
Nucleic acid testing for herpes simplex virus types 1 and 2, varicella–zoster virus, cytomegalovirus, Epstein–Barr virus, enterovirus
* To convert the values for glucose to millimoles per liter, multiply by 0.05551. † Reference values are affected by many variables, including the patient population and the laboratory methods used. The ranges used at Massachusetts General Hospital are for adults who are not pregnant and do not have medical conditions that could affect the results. They may therefore not be appropriate for all patients.
abdomen revealed multiple renal cysts, a small infrarenal abdominal aortic aneurysm measuring 3.3 cm in the greatest dimension, and no evidence of cancer in the chest, abdomen, or pelvis. The patient was referred to the neuro-oncology clinic. During the examination, two paroxysmal episodes of speech arrest were witnessed that lasted 10 to 15 seconds and were accompanied by subtle twitching on the right side of the face and transient emotional outbursts with rapid recovery. Persistent periorbital twitching occurred, more on the right side than on the left, and there was a postural tremor, slight weakness (5−) in the right arm, and slow fine movements of the fingers on the right hand. The patient had minor weakness of the dorsiflexors of the right foot and patchy loss of sensation to pin prick in the lower legs and feet bilaterally. The remainder of the examination was normal. A lumbar puncture was performed; results of analysis of the cerebrospinal fluid (CSF) are shown in Table 1. The administration of phenytoin was begun. A diagnostic test was performed.
Differ en t i a l Di agnosis Dr. Sydney S. Cash: May we review the imaging studies? Dr. Mykol Larvie: The initial imaging included MRI of the head. An axial fluid-attenuated inversion recovery (FLAIR) image (Fig. 1) shows
normal configuration of the brain and normal signal throughout the majority of the brain. In the left medial temporal lobe, there is mild expansion and abnormal T2-weighted hyperintensity that was not appreciated initially but that was identified on a follow-up examination. There was no restricted diffusion or abnormal enhancement. Dr. Cash: To summarize, this man presented with a progressive neurologic illness characterized by episodic symptoms of olfactory hallucination, dysautonomia, speech arrest or wordfinding difficulties, posturing of the right arm, and twitching of the right side of the face superimposed on a progressive decline in memory and executive function. Acquired epilepsy
This patient’s episodic symptoms are consistent with seizure activity, most likely arising from the left temporal lobe. An ambulatory electroencephalogram (EEG) (i.e., continuous EEG recording for ≤72 hours) was normal, but this result does not preclude the diagnosis of focal seizure activity, especially if the zone of irritability is relatively small and deeply placed. Approximately 50% of patients with epilepsy will have a normal routine EEG,1 although ambulatory EEG may confer a lower false negative rate.2 Moreover, our experience using foramen ovale electrodes (i.e., electrodes inserted through a foramen ovale) in patients with epilepsy provides
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1827
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
unlikely in view of the extensive screening for a primary tumor, which was negative, and the appearance of the brain on imaging, with isolated enlargement of the hippocampus and amygdala without enhancement. Primary neoplasms of the central nervous system, such as low-grade glioma, may present with a subacute course and subtle imaging changes. This patient is older than the median age of patients with highgrade gliomas, which is in the 60s; the median age is even lower for patients with low-grade gliomas.3 In the patient’s age group, primary lym phomas of the central nervous system can be manifested by a wide variety of symptoms; patients typically present with strongly enhancing periventricular lesions, and the disease may be rapidly progressive, in contrast to the findings in this case. Autoimmune diseases
Figure 1. Axial T2-Weighted Fluid-Attenuated Inversion Recovery Image through the Temporal Lobes. The left medial temporal lobe, including the amygdaloid nucleus and the anterior hippocampus, show slightly greater T2 -weighted signal than the contralateral structures, and there is mild expansion of the left amygdaloid nucleus (arrow).
convincing evidence that in some patients, seizures arising from deep mesial structures in the temporal lobe may not have a clear-cut representation of the EEG recorded at the scalp. A broad differential diagnosis of acquired epilepsy and memory loss is listed in Table 2, but the rapid course of this patient’s illness and the essentially normal evaluation, except for the findings on MRI and moderate inflammatory changes in the CSF, leave us with only two possible categories of disease — neoplastic and autoimmune. The course of his illness, along with the tests performed, effectively ruled out many common causes, including vascular, metabolic, toxic, infectious, genetic, congenital, and chronic neurodegenerative diseases (e.g., Alzheimer’s disease).
The most likely category of disease to be considered is autoimmune. In fact, this patient’s presentation is typical of a limbic encephalitis.4,5 Limbic encephalitis may be either paraneoplastic (i.e., associated with tumor) or a primary nonparaneoplastic autoimmune process.6 Although there is a broad range of temporal progression among the limbic encephalitides, paraneoplastic disease typically has a faster course than was seen in this patient; nonparaneoplastic disease may have a more chronic course, which was seen in this patient. The lack of contrast enhancement and the minimal pleocytosis and moderately elevated protein level in the CSF are typical of the autoimmune encephalitides, regardless of whether they are paraneoplastic or nonparaneoplastic. Autoimmune Limbic Encephalitis
Limbic encephalitis is an inflammatory process centered in the limbic system. It results in limbic dysfunction that is manifested by behavioral and psychiatric symptoms, cognitive and memory dysfunction, as seen in this patient, and seizures, also seen in this patient.7 The MRI scan is typically abnormal, with mesial temporal abnormalities on the FLAIR image, whereas diffusion characteristics are usually normal, as in this case. Neoplastic diseases There may be very mild enhancement (often asymNeoplastic disease is a serious consideration in metric and sometimes patchy) or, more commonly, this patient. Metastatic disease to the brain seems no enhancement, as in this case. The CSF is typi-
1828
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
case records of the massachusetts gener al hospital
cally bland but can have borderline or mildly elevated protein levels, as in this case, and minimal pleocytosis with a lymphocytic predominance, also seen in this case. The EEG is not helpful as a diagnostic tool, since patients may have interictal discharges, frank seizures, focal slowing, or a normal EEG, as this patient did. When there is a clinical suspicion of limbic encephalitis, it is essential to consider which antibodies could be involved, since this will point to a possible associated tumor.8 The paraneoplastic and nonparaneoplastic autoimmune encephalitides can be divided into those that are related to intracellular antigens (e.g., Hu and Ma) and those that are related to neuronal membrane antigens. Three prominent examples are voltage-gated potassium channels, the N-methyl-d-aspartate (NMDA) receptor, and the γ-aminobutyric acid type B (GABA B ) receptors. More recently, antibodies to the α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid (AMPA) receptor have also been found in some patients.9 Patients with antibodies directed against specific antigens may have particular presentations. For example, encephalitis associated with anti-NMDA receptors is seen almost exclusively in young women, is associated with ovarian tumors, and has a rapid and dramatic presentation.10 Because of considerable overlap among presentations, however, it is important to screen for a wide range of antibodies. A number of antibodies were tested for in this patient, but a select group were not (perhaps because the test was unavailable at the time). Three important antibodies that were not present on the initial panel were Ma2, GABA B receptor, and the voltage-gated potassium channel. The clinical presentation and testing are all consistent with the presence of one of these antibodies, despite the fact that no tumors were discovered on extensive investigations. Summary
In summary, my diagnosis is a limbic encephalitis, most likely secondary to an autoimmune process, possibly paraneoplastic. The next diagnostic step should be a more extensive serologic investigation to look for the presence of the voltagegated potassium channel, GABA B receptor, and anti-Ma antibodies. Dr. Nancy Lee Harris (Pathology): Dr. Batchelor, would you tell us the clinical thinking at the time?
Table 2. Possible Causes of Acquired Seizures and Cognitive Decline. Category
Examples
Vascular
Single or multiple strokes
Toxic
Drugs, heavy metals, ethyl alcohol
Metabolic
Renal or hepatic failure, hypoglycemia, hyperglycemia
Infectious Acute
Herpes simplex virus, varicella–zoster virus, West Nile virus, bacterial infection
Subacute
Lyme disease, syphilis, tuberculosis, prion diseases (e.g., Creutzfeldt–Jakob disease)
Neurodegenerative
Alzheimer’s disease, frontotemporal dementia, primary progressive aphasia
Genetic or congenital Dysplastic lesions, perinatal injuries Neoplastic
Primary and secondary neoplasm of the central nervous system
Autoimmune
Systemic lupus erythematosus, Sjögren’s syndrome, Hashimoto’s encephalitis, limbic encephalitis
Dr. Tracy T. Batchelor (Neuro-oncology): The leading diagnosis was limbic encephalitis; an infiltrative glioma was second on the list on the basis of the imaging characteristics. The paraneoplastic antibody panel performed at the Mayo Clinic was entirely negative. Additional testing was performed in the laboratory of Dr. Josep Dalmau at the University of Pennsylvania.
Cl inic a l Di agnosis Limbic encephalitis.
Dr . S y dne y S . C a sh’s Di agnosis Autoimmune limbic encephalitis, either paraneoplastic or nonparaneoplastic.
Pathol o gic a l Discussion Dr. Josep Dalmau: The patient’s serum showed intense reactivity with the surface of cultured, dissociated rat hippocampal neurons. Additional immunohistochemical studies showed that the reactivity was similar to that attributed to antibodies against voltage-gated potassium channels, which are very frequently associated with limbic encephalitis.5,11 However, we have recently found that the target antigen in cases of limbic encephalitis with this pattern of reactivity is a secreted neuronal protein
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1829
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
A
B
C
D
E
F
G
H
I
Figure 2. Detection of Antibodies in the Patient’s Serum. HEK293 cells (from human embryonic kidneys) transfected with LGI1 were treated with the patient’s serum (Panels A, D, and G, fluorescein). In Panels B, E, and H (rhodamine), a commercial antibody against LGI1 is used to show the transfected cells. Panels C, F, and I (fluorescein and rhodamine; cell nuclei stained with 4’,6-diamidine-2-phenylidole dihydrochloride) show the colocalization of reactivities of the patient’s antibodies with the commercial LGI1 antibody. The patient’s serum had detectable antibodies to LGI1 at presentation (Panels A, B, and C). A repeat test performed later in the course of the disorder (Panels D, E, and F) shows persistent antibodies. After substantial clinical recovery, 17 months after presentation (Panels G, H, and I), there were no detectable antibodies (immunofluorescence method).
called leucine-rich, glioma-inactivated 1 (LGI1).12 Further testing of this patient’s serum showed that it contained antibodies to LGI1 at the time of presentation (Fig. 2A, 2B, and 2C). This finding confirms that the patient had limbic encephalitis associated with antibodies to LGI1, previously attributed to voltage-gated potassium channels (this patient was included in the published study12). Patients with this disorder present in the same way this patient did, with changes in mood and behavior, seizures, and memory problems. Hyponatremia often develops. Most patients with LGI1 antibodies, including this patient, do not have an underlying tumor. LGI1 interacts with presynaptic and postsynaptic epilepsy-related proteins (disintegrin and
1830
n engl j med 365;19
metalloproteinase domain–containing protein 23 [ADAM23] and ADAM22) and most likely organizes a trans-synaptic protein complex that includes the presynaptic Kv1.1 voltage-dependent potassium channels and postsynaptic AMPAreceptor scaffolds.13 Mutations of LGI1 cause autosomal dominant partial epilepsy with auditory features,14-16 also known as autosomal dominant lateral temporal-lobe epilepsy (ADLTE).17 A transgenic mouse model expressing a truncated mutant LGI1 found in human ADLTE showed inhibition of dendritic pruning and increased spine density, resulting in increased excitatory synaptic transmission.18 LGI1-null mutant mice have a lethal epileptic phenotype characterized by myoclonic seizures.13,19 It has been suggested that
nejm.org
november 10, 2011
The New England Journal of Medicine
case records of the massachusetts gener al hospital
A
Day of Diagnostic Test Result
B
C
2-Mo Follow-up
6-Mo Follow-up
D 10-Mo Follow-up
E
13-Mo Follow-up
Figure 3. Examinations of the Brain with Serial 18F-Fluorodeoxyglucose–Positron-Emission Tomography (FDG-PET). Each panel depicts the fusion of an FDG-PET examination with T2 -weighted fluid-attenuated inversion recovery images from a contemporaneous MRI examination of the head. The first PET scan (Panel A) was obtained shortly after the diagnostic test result was received and shows hypermetabolism, represented by the light yellow-white area in the left medial temporal lobe (in the crosshairs in all five panels). The abnormality in the left temporal lobe appears worse at the follow-up examination 2 months later (Panel B) and then better at 6 months (Panel C) and 10 months (Panel D); the left and right temporal lobes are nearly symmetric at the final examination, when the patient’s symptoms had improved (Panel E). These images show the waxing and waning of the abnormalities seen on PET examination, which approximately coincided with the waxing and waning of the patient’s signs and symptoms.
antibody-mediated disruption of LGI1 causes increased excitability, resulting in seizures and other symptoms of limbic encephalopathy.12 These antibodies may also alter the function of proteins associated with LGI1, such as ADAM22 and ADAM23, resulting in a phenotype different from that caused by mutations in LGI1. Dr. Batchelor: During the initial diagnostic evaluation, total-body positron-emission tomography (PET)–CT was performed. Examination of a specimen from a fine-needle aspiration biopsy of a nodule (3.5 cm in diameter) in the left thyroid was negative for cancer. The patient was treated with phenytoin, and his seizures diminished. Additional routine laboratory testing revealed hyponatremia. After the results of testing were received from Dr. Dalmau’s laboratory, the patient was treated with a 5-day course of intravenous immune globulin and pulsed intravenous glucocorticoids for four monthly cycles. He had a moderate clinical improvement and residual short-term memory loss. The decision was then made to initiate the administration of rituximab; he received six monthly doses. During the course of rituximab treatment, he had further (although still incomplete) clinical recovery. May we see the follow-up imaging and laboratory studies? Dr. Larvie: Examination of the brain with the use of 18F-fluorodeoxyglucose (FDG)–PET, performed
n engl j med 365;19
after the diagnostic test result was received, revealed a focus of abnormal hypermetabolism in the left medial temporal lobe (Fig. 3A). A subsequent examination showed much more severe hypermetabolism in the left medial temporal lobe (Fig. 3B), most pronounced within the amygdaloid nucleus. The abnormality abated gradually, as the patient’s symptoms improved (Fig. 3C, 3D, and 3E). A primary brain neoplasm may be considered in the differential diagnosis, but such neoplasms tend to be less metabolically active than the relatively hypermetabolic normal brain. The imaging findings in this patient would be unusual for a metastatic tumor, especially one that is nonenhancing and that progresses so little. Seizure activity can give rise to hypermetabolism, but that would typically correlate with the clinical finding of seizure and would not explain the expansion in the medial temporal lobe. Inflammation, particularly an autoantibody-mediated process, more readily accounts for the imaging findings. Dr. Dalmau: The patient had lower levels of antibodies 6 months after presentation (Fig. 2D, 2E, and 2F). There were no detectable serum antibodies after substantial clinical recovery, 17 months after presentation (Fig. 2G, 2H, and 2I). These findings correlate well with the FDG-PET studies. Dr. Harris: Dr. Batchelor, would you give us further follow-up? Dr. Batchelor: The patient remained clinically stable for 2 years after the 6-month course of
nejm.org
november 10, 2011
The New England Journal of Medicine
1831
The
n e w e ng l a n d j o u r na l
rituximab. He presented to the emergency department at this hospital 3 weeks before this conference because of 3 days of worsening gait instability and escalating confusion and was admitted to the hospital. Family members revealed that the patient may have undergone a couple of months of more insidious decline in cognition. His examinations on admission and in the days after admission were notable for dramatic waxing and waning of confusion and lethargy (features consistent with a delirium) and a wide-based, unsteady gait. Results of laboratory studies showed hyponatremia. MRI of the brain was unremarkable. CSF testing revealed an elevated protein level, a low glucose level, and 350 white cells, predominantly lymphocytes. Multiple microbial studies of the CSF were negative, including Gram’s stain, culture, cryptococcal antigen, and acid-fast bacilli stain, as well as nucleic acid tests for herpes simplex virus (type 1, type 2, and type 6), varicella–zoster virus, Epstein– Barr virus, and cytomegalovirus. A repeat PET-CT scan showed no evidence of a malignant tumor. A commercial paraneoplastic antibody panel was negative, including testing for voltage-gated potassium channels. Serum and CSF tests for LGI1 antibodies were also negative. The patient was treated with acyclovir for 5 days until the serologic-test results came back, and then intravenous immune globulin and methylprednisolone were administered to treat a suspected relapse of autoimmune encephalitis. He had rapid but incomplete clinical improvement after 3 days of treatment. Dr. Harris: Are there any questions? Dr. Anne B. Young (Neurology): I am curious about why these autoimmune processes can be so asymmetric. Dr. Dalmau: The perception of asymmetry may be a result of the limitations of testing. FDG-PET often reveals areas of involvement that are not seen on MRI, and as has been found in other forms of limbic encephalitis, pathological studies often show more extensive (bilateral or multifocal) inflammatory changes than were clinically or radiologically predicted. Dr. Andrew J. Cole (Neurology): Familial temporallobe epilepsy with auditory features, due to a mutation in the gene LGI1, is associated with unilateral symptoms, despite what must be a widespread structural or biochemical abnormality. In addition, in patients with autoimmune limbic encephalitis, lesions in sites distant from the locus of the sei1832
of
m e dic i n e
zure may be revealed at autopsy, indicating that there can be a disassociation between the neuropathological abnormality and the predominant functional abnormality. Dr. John B. Herman (Psychiatry): It was interesting that when emotional issues were being discussed with the patient, noticeable exacerbation of word-finding problems and twitching of the right hand occurred, which was absolutely related to the content of the conversation. Dr. Harris: The patient is here at the conference and would like to comment. The Patient: It is extraordinarily rare for a person to be involved as the discussant of one of these exercises and as the subject of another one. The symptom that most bothers me is memory loss. I can almost date the onset of a sharp cutoff of recent memory and variable problems with distant memory. The loss is strangely selective. For instance, I am sick and tired of doing “serial sevens”; I’ve been counting down from 100 by sevens all my life and still have no trouble with it. But if you give me a telephone number, I’ll be lucky if I can remember more than three of the digits. I voluntarily gave up clinical practice because of the decline in my intellectual function. The rituximab treatment was remarkable, in that I noted a sharp improvement in my symptoms early in the first cycle. This recent episode was strikingly different. I went from being relatively normal to being in la-la land within 48 hours. I remember nothing of the first 3 or 4 days that I was in the hospital, in four-point restraints, with my friends taking care of me. That delirium has passed, and I am left once again with memory problems, probably a little worse than before, and balance problems, which are new but are responding remarkably well to physical therapy. It is fascinating to have been on both sides of this podium. I consider myself lucky that I have worked at an institution where someone had seen a case like mine. Dr. Batchelor (note added in follow-up): Treatment with rituximab was begun weekly for 8 weeks, shortly after the patient’s discharge from the hospital. His condition continued to improve and returned to his preadmission baseline. One year later, he continues to receive monthly rituximab infusions and has had no further seizure activity or episodes of confusion. Short-term memory problems remain his predominant symptom.
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
case records of the massachusetts gener al hospital
A nat omic a l Di agnosis Limbic encephalitis with antibodies to leucine-rich, glioma-inactivated 1 (LGI1). This case was discussed at the Neurology Grand Rounds. Dr. Dalmau is currently at Institució Catalana de Recerca i Estudis Avançats (ICREA), Institut d’Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS), Hospital Clinic, University of Bar-
References 1. Marsan CA, Zivin LS. Factors related to the occurrence of typical paroxysmal abnormalities in the EEG records of epileptic patients. Epilepsia 1970;11:361-81. 2. Ebersole J, Schomer D, Ives JR. Ambulatory EEG monitoring. In: Ebersole JS, Pedley TA, eds. Current practice of clinical electroencephalography. 3rd ed. New York: Lippincott Williams & Wilkins, 2003:610-38. 3. Hess KR, Broglio KR, Bondy ML. Adult glioma incidence trends in the United States, 1977-2000. Cancer 2004;101:2293-9. 4. Tüzün E, Dalmau J. Limbic encephalitis and variants: classification, diagnosis and treatment. Neurologist 2007;13:261-71. 5. Vincent A, Buckley C, Schott JM, et al. Potassium channel antibody-associated encephalopathy: a potentially immunotherapy-responsive form of limbic encephalitis. Brain 2004;127:701-12. 6. Graus F, Delattre JY, Antoine JC, et al. Recommended diagnostic criteria for paraneoplastic neurological syndromes. J Neurol Neurosurg Psychiatry 2004;75:1135-40. 7. Bien CG, Urbach H, Schramm J, et al. Limbic encephalitis as a precipitating event in adult-onset temporal lobe epilepsy. Neurology 2007;69:1236-44.
celona, Barcelona; and at the Department of Neurology, University of Pennsylvania, Philadelphia. Disclosure forms provided by the authors are available with the full text of this article at NEJM.org. We thank Drs. James Lehrich, John Growden, Emad Eskandar, Janet Sherman, and Tracy Batchelor for their assistance in the preparation of the case history; Drs. Tracy Batchelor and Thomas Byrne for helping to organize the conference; and the patient for his support of and participation in the conference.
8. Soeder BM, Gleissner U, Urbach H, et
al. Causes, presentation and outcome of lesional adult onset mediotemporal lobe epilepsy. J Neurol Neurosurg Psychiatry 2009;80:894-9. 9. Bataller L, Galiano R, Garcia-Escrig M, et al. Reversible paraneoplastic limbic encephalitis associated with antibodies to the AMPA receptor. Neurology 2010; 74:265-7. 10. Dalmau J, Rosenfeld MR. Paraneoplastic syndromes of the CNS. Lancet Neurol 2008;7:327-40. 11. Thieben MJ, Lennon VA, Boeve BF, Aksamit AJ, Keegan M, Vernino S. Potentially reversible autoimmune limbic encephalitis with neuronal potassium channel antibody. Neurology 2004;62:1177-82. 12. Lai M, Huijbers MG, Lancaster E, et al. Investigation of LGI1 as the antigen in limbic encephalitis previously attributed to potassium channels: a case series. Lancet Neurol 2010;9:776-85. 13. Fukata Y, Lovero KL, Iwanaga T, et al. Disruption of LGI1-linked synaptic complex causes abnormal synaptic transmission and epilepsy. Proc Natl Acad Sci U S A 2010;107:3799-804.
14. Gu W, Brodtkorb E, Steinlein OK.
LGI1 is mutated in familial temporal lobe epilepsy characterized by aphasic seizures. Ann Neurol 2002;52:364-7. 15. Kalachikov S, Evgrafov O, Ross B, et al. Mutations in LGI1 cause autosomal-dominant partial epilepsy with auditory features. Nat Genet 2002;30:335-41. 16. Morante-Redolat JM, Gorostidi-Pagola A, Piquer-Sirerol S, et al. Mutations in the LGI1/Epitempin gene on 10q24 cause autosomal dominant lateral temporal epilepsy. Hum Mol Genet 2002;11:1119-28. 17. Poza JJ, Sáenz A, Martínez-Gil A, et al. Autosomal dominant lateral temporal epilepsy: clinical and genetic study of a large Basque pedigree linked to chromosome 10q. Ann Neurol 1999;45:182-8. 18. Zhou YD, Lee S, Jin Z, Wright M, Smith SE, Anderson MP. Arrested maturation of excitatory synapses in autosomal dominant lateral temporal lobe epilepsy. Nat Med 2009;15:1208-14. 19. Yu YE, Wen L, Silva J, et al. Lgi1 null mutant mice exhibit myoclonic seizures and CA1 neuronal hyperexcitability. Hum Mol Genet 2010;19:1702-11. Copyright © 2011 Massachusetts Medical Society.
Lantern Slides Updated: Complete PowerPoint Slide Sets from the Clinicopathological Conferences Any reader of the Journal who uses the Case Records of the Massachusetts General Hospital as a teaching exercise or reference material is now eligible to receive a complete set of PowerPoint slides, including digital images, with identifying legends, shown at the live Clinicopathological Conference (CPC) that is the basis of the Case Record. This slide set contains all of the images from the CPC, not only those published in the Journal. Radiographic, neurologic, and cardiac studies, gross specimens, and photomicrographs, as well as unpublished text slides, tables, and diagrams, are included. Every year 40 sets are produced, averaging 50-60 slides per set. Each set is supplied on a compact disc and is mailed to coincide with the publication of the Case Record. The cost of an annual subscription is $600, or individual sets may be purchased for $50 each. Application forms for the current subscription year, which began in January, may be obtained from the Lantern Slides Service, Department of Pathology, Massachusetts General Hospital, Boston, MA 02114 (telephone 617-726-2974) or e-mail
[email protected].
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1833
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
edi t or i a l s
Deconstructing a Lethal Foodborne Epidemic Martin J. Blaser, M.D. Beginning in early May 2011, northern Germany was the principal site of a massive epidemic of bloody diarrhea and the hemolytic–uremic syndrome caused by Shiga-toxin–producing Escherichia coli. By the time the outbreak ended in early July, there were reports of more than 4000 illnesses, 800 cases of the hemolytic–uremic syndrome, and 50 deaths in Germany and in 15 other countries. As a result of remarkable efforts by public health officials, clinicians, and microbiologists who raced to control the outbreak and to treat its victims, much was learned. As reported in this issue of the Journal by Frank et al.1 and Buchholz et al.,2 as well as by other investigators elsewhere,3-5 here is what we now know. The outbreak was associated with a single clone of a strain of enterohemorrhagic E. coli classified as O104:H4. Although similar strains had been reported to cause the hemolytic–uremic syndrome before,6 the 2011 strain was novel, with an uncommon serotype, plasmid-encoded extendedspectrum beta-lactamase (ESBL), and genes from enteroaggregative E. coli. Although Shiga toxin primarily causes tissue injury, the strain may have been aided by genes from enteroaggregative E. coli that improved its efficiency in intestinal colonization, and the presence of ESBL enhanced survival when beta-lactam antibiotics (including penicillins and cephalosporins) suppressed competitors. The epidemic strain apparently arose from precursors by sequential introduction of pathogenetic elements.4 Whereas variants often arise in nature, only some can spread. This O104:H4 strain was well armed for mayhem and reminds us that evolution is a constant. The outbreak was foodborne in contaminated sprouts. The initial investigation pointed to other
uncooked salad foods, illustrating the difficulty in identifying vehicles in multisite outbreaks when exposures occurred days earlier and when answers are needed immediately.7 The chain of transmission appears to have begun in Egypt, with fecal contamination of fenugreek seeds by either humans or farm animals during storage or transportation, perhaps as long ago as 2009. The seeds then went to a European distributor and from there to farms in several countries. During sprout germination, bacteria multiplied and moved from farm to restaurants and consumers, as Buchholz et al. extensively detail in their study. The evidence for such a series of events is compelling, even though the organism was not identified at the earliest steps, since the trail often is cold in point-source outbreaks by the time investigators are able to conduct trace-back investigations.8 The modern commerce in food allows multiple opportunities for bacterial replication, and mass production and wide nets of distribution foster large, geographically diverse outbreaks.9 The outbreak began suddenly in early May, peaked within weeks, and ended in early July,1 a pattern most consistent with a single point source. With the primary transmission being foodborne and the secondary transmission through household, nosocomial, and even laboratory vectors, cases continued to occur. Unlike most outbreaks of enterohemorrhagic E. coli that cluster in young children and the elderly, most cases occurred in adults across a broad age span. Although such a transmission pattern may mostly reflect the implicated vehicle and food choices, the age distribution also suggests a lack of previous immunity to a novel pathotype. The long median in cubation period (8 days) in contrast to other
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1835
The
n e w e ng l a n d j o u r na l
outbreaks of enterohemorrhagic E. coli (3 to 4 days)10 may indicate either that relatively small inocula were consumed or that in vivo characteristics of the outbreak strains were atypical. The latter is most likely on the basis of the strain genotype, as well as incubation, age, and virulence characteristics. The predominance of the outbreak in women may only reflect food preferences, since the secondary cases were more evenly divided according to sex. The outbreak was unusual, with an atypical age distribution, a long incubation period, very high rates of the hemolytic–uremic syndrome and death, and somewhat different clinical features in children and adults. The hemolytic–uremic syndrome developed suddenly, about 5 days after the onset of diarrhea. Such a window provides an opportunity to intervene in order to minimize illness, if only we knew what to do. In this outbreak, clinicians cared for their patients diligently but without critical knowledge. Would antibiotics help or hurt? Glucocorticoids? Plasma exchange? Just as there is a system for emergency case reporting in Germany and elsewhere, we also need authorizations to conduct clinical trials in real time during public health emergencies such as this one. Patients should be randomly assigned to various clinically suitable regimens to learn what works and what does not. Infectious disease epidemics constantly arise, usually involving familiar pathogens but with combinations of known and unknown virulence factors or in new vehicles, causing novel outbreaks and clinical consequences. With our complex global food trade and multiple opportunities for microbial amplification, the next large outbreak is just around the corner. We can prepare for it by drafting generic nationwide (or
of
m e dic i n e
worldwide) protocols with preapproval by institutional review boards (e.g., for outbreaks of enterohemorrhagic E. coli) so that we can learn while trying to mitigate these tragedies. Disclosure forms provided by the author are available with the full text of this article at NEJM.org. From the Departments of Medicine and Microbiology, New York University Langone Medical Center, and the New York Harbor Department of Veterans Affairs Medical Center — all in New York. This article (10.1056/NEJMe1110896) was published on October 26, 2011, at NEJM.org. 1. Frank C, Werber D, Cramer JP, et al. Epidemic profile of
Shiga-toxin–producing Escherichia coli O104:H4 outbreak in Germany. N Engl J Med 2011;365:1771-80. 2. Buchholz U, Bernard H, Werber D, et al. German outbreak of Escherichia coli O104:H4 associated with sprouts. N Engl J Med 2011;365:1763-70. 3. Bielaszewska M, Mellmann A, Zhang W, et al. Characterisation of the Escherichia coli strain associated with an outbreak of haemolytic uraemic syndrome in Germany. Lancet Infect Dis 2011;11:671-6. 4. Mellmann A, Harmsen D, Cummings CA, et al. Prospective genomic characterization of the German enterohemorrhagic Escherichia coli O104:H4 outbreak by rapid next generation sequencing technology. PLoS ONE 2011;6(7):e22751. 5. Aldabe B, Delmas Y, Gault G, et al. Household transmission of haemolytic uraemic syndrome associated with Escherichia coli O104:H4, south-western France, June 2011. Euro Surveill 2011;16:19934. 6. Morabito S, Karch H, Mariani-Kurkdijian P, et al. Enteroaggregative, Shiga toxin-producing Escherichia coli O111:H2 associated with an outbreak of hemolytic-uremic syndrome. J Clin Microbiol 1998;36:840-2. 7. Barton Behravesh C, Mody RK, Jungk J. 2008 outbreak of Salmonella Saintpaul infections associated with raw produce. N Engl J Med 2011;364:918-27. 8. Herwaldt BL, Beach MJ. The return of cyclospora in 1997: another outbreak of cyclosporiasis in North America associated with imported raspberries. Ann Intern Med 1999;130:210-20. 9. Blaser MJ. How safe is our food? Lessons from an outbreak of salmonellosis. N Engl J Med 1996;334:1324-5. 10. Slutsker L, Ries AA, Greene KD, Wells JG, Hutwagner L, Griffin PM. Escherichia coli O157:H7 diarrhea in the United States: clinical and epidemiologic features. Ann Intern Med 1997;126:505-13. Copyright © 2011 Massachusetts Medical Society.
Transplantation for Alcoholic Hepatitis — Time to Rethink the 6-Month “Rule” Robert S. Brown, Jr., M.D., M.P.H. Liver transplantation for alcoholic liver disease has been controversial since the advent of the procedure. The perception that alcohol-related liver disease is self-inflicted, combined with concerns about recidivism to alcohol use and poor adherence to post-transplantation care, has led the public 1836
and physicians to not support transplantation in patients with alcoholism. After initially avoiding transplantations in this group of patients, the majority of programs now restrict the procedure to those meeting a requirement of a 6-month period of abstinence from alcohol. Unfortunately,
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
The
n e w e ng l a n d j o u r na l
outbreaks of enterohemorrhagic E. coli (3 to 4 days)10 may indicate either that relatively small inocula were consumed or that in vivo characteristics of the outbreak strains were atypical. The latter is most likely on the basis of the strain genotype, as well as incubation, age, and virulence characteristics. The predominance of the outbreak in women may only reflect food preferences, since the secondary cases were more evenly divided according to sex. The outbreak was unusual, with an atypical age distribution, a long incubation period, very high rates of the hemolytic–uremic syndrome and death, and somewhat different clinical features in children and adults. The hemolytic–uremic syndrome developed suddenly, about 5 days after the onset of diarrhea. Such a window provides an opportunity to intervene in order to minimize illness, if only we knew what to do. In this outbreak, clinicians cared for their patients diligently but without critical knowledge. Would antibiotics help or hurt? Glucocorticoids? Plasma exchange? Just as there is a system for emergency case reporting in Germany and elsewhere, we also need authorizations to conduct clinical trials in real time during public health emergencies such as this one. Patients should be randomly assigned to various clinically suitable regimens to learn what works and what does not. Infectious disease epidemics constantly arise, usually involving familiar pathogens but with combinations of known and unknown virulence factors or in new vehicles, causing novel outbreaks and clinical consequences. With our complex global food trade and multiple opportunities for microbial amplification, the next large outbreak is just around the corner. We can prepare for it by drafting generic nationwide (or
of
m e dic i n e
worldwide) protocols with preapproval by institutional review boards (e.g., for outbreaks of enterohemorrhagic E. coli) so that we can learn while trying to mitigate these tragedies. Disclosure forms provided by the author are available with the full text of this article at NEJM.org. From the Departments of Medicine and Microbiology, New York University Langone Medical Center, and the New York Harbor Department of Veterans Affairs Medical Center — all in New York. This article (10.1056/NEJMe1110896) was published on October 26, 2011, at NEJM.org. 1. Frank C, Werber D, Cramer JP, et al. Epidemic profile of
Shiga-toxin–producing Escherichia coli O104:H4 outbreak in Germany. N Engl J Med 2011;365:1771-80. 2. Buchholz U, Bernard H, Werber D, et al. German outbreak of Escherichia coli O104:H4 associated with sprouts. N Engl J Med 2011;365:1763-70. 3. Bielaszewska M, Mellmann A, Zhang W, et al. Characterisation of the Escherichia coli strain associated with an outbreak of haemolytic uraemic syndrome in Germany. Lancet Infect Dis 2011;11:671-6. 4. Mellmann A, Harmsen D, Cummings CA, et al. Prospective genomic characterization of the German enterohemorrhagic Escherichia coli O104:H4 outbreak by rapid next generation sequencing technology. PLoS ONE 2011;6(7):e22751. 5. Aldabe B, Delmas Y, Gault G, et al. Household transmission of haemolytic uraemic syndrome associated with Escherichia coli O104:H4, south-western France, June 2011. Euro Surveill 2011;16:19934. 6. Morabito S, Karch H, Mariani-Kurkdijian P, et al. Enteroaggregative, Shiga toxin-producing Escherichia coli O111:H2 associated with an outbreak of hemolytic-uremic syndrome. J Clin Microbiol 1998;36:840-2. 7. Barton Behravesh C, Mody RK, Jungk J. 2008 outbreak of Salmonella Saintpaul infections associated with raw produce. N Engl J Med 2011;364:918-27. 8. Herwaldt BL, Beach MJ. The return of cyclospora in 1997: another outbreak of cyclosporiasis in North America associated with imported raspberries. Ann Intern Med 1999;130:210-20. 9. Blaser MJ. How safe is our food? Lessons from an outbreak of salmonellosis. N Engl J Med 1996;334:1324-5. 10. Slutsker L, Ries AA, Greene KD, Wells JG, Hutwagner L, Griffin PM. Escherichia coli O157:H7 diarrhea in the United States: clinical and epidemiologic features. Ann Intern Med 1997;126:505-13. Copyright © 2011 Massachusetts Medical Society.
Transplantation for Alcoholic Hepatitis — Time to Rethink the 6-Month “Rule” Robert S. Brown, Jr., M.D., M.P.H. Liver transplantation for alcoholic liver disease has been controversial since the advent of the procedure. The perception that alcohol-related liver disease is self-inflicted, combined with concerns about recidivism to alcohol use and poor adherence to post-transplantation care, has led the public 1836
and physicians to not support transplantation in patients with alcoholism. After initially avoiding transplantations in this group of patients, the majority of programs now restrict the procedure to those meeting a requirement of a 6-month period of abstinence from alcohol. Unfortunately,
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
editorials
a subgroup of patients with severe, acute alcoholic hepatitis has little or no chance of surviving 6 months, and these patients usually die from their disease. In this issue of the Journal, Mathurin et al.1 report on a prospective study of liver transplantation in highly selected patients with severe alcoholic hepatitis who were unlikely to survive 6 months. The authors compared the survival in the cohort of transplant patients to that in two matched control groups of patients who were not offered transplantation. The patients were carefully selected at each center; selected patients had good social support, a contract for abstinence, no prior episodes of known alcoholic liver disease, and no evidence of severe psychological disease. This resulted in 26 patients being selected over a period of more than 4 years, representing fewer than 2% of patients admitted for an episode of severe alcoholic hepatitis. The results of transplantation were excellent, with a 6-month survival rate of 77%. The rate of recidivism to alcohol use was low, with only 3 of 26 patients resuming drinking alcohol; of these 3, none had graft dysfunction. Patients had good adherence to their medical regimens. The rate of survival was superior to that among patients who were not offered transplantation, who had 23% survival at 6 months. The results of this study are not surprising; given the poor outcomes with severe alcoholic hepatitis, one would expect that the results of transplantation would be superior. However, patients who were selected for transplantation probably had systematic differences in their demographic and socioeconomic characteristics from those who did not undergo the procedure. Given these excellent outcomes, should we be offering transplantation to patients with severe alcoholic hepatitis who have not demonstrated a period of abstinence from alcohol? It is clear that the 6-month rule is somewhat arbitrary. Although 6 months of abstinence does predict a lower rate of recidivism than shorter periods of abstinence,2-5 longer periods of abstinence (up to 36 months) predict even lower rates of recidivism.6 The rate of recidivism to alcohol use in this study was low, lower than the estimated rate of 30% or higher among patients who have met the 6-month rule. In addition, we accept recurrence rates of other liver diseases that are far higher and have worse outcomes than what we see in alcoholic liver disease. The rate of recurrence of hepatitis C is
100%, leading to graft failure at a much higher rate than with recurrent alcohol use at 5 years. When we approach these decisions, we need to balance issues of utility and justice. Certainly from a utility standpoint, we can justify liver transplantation for alcoholic liver disease, probably even with an abstinence period of shorter than 6 months in selected individuals. The outcomes for alcoholic liver disease are better than those for the natural history of the disease without liver transplantation and are as good as or superior to those for other indications for liver transplantation. The issue of justice is more complex. The argument of self-inflicted disease is not a valid one. When one looks at the causes of liver disease, much of it could be perceived as self-inflicted, such as prior drug use resulting in viral hepatitis or obesity leading to nonalcoholic fatty liver disease. Alcoholism is a disease, and it should not be used to exclude patients from transplantation. However, there are important issues regarding the selection of appropriate candidates, and this is where the justice concept becomes more difficult. Without a good objective tool to predict risk of recidivism that can be applied fairly and equally across candidates, we are likely to have bias in our selection of candidates. There may have been bias in the selection of the candidates in this study. It would have been useful to look at the overall population of patients with alcoholic hepatitis to determine whether the group that was selected mirrored the overall population of patients with equally severe alcoholic hepatitis in terms of sex, race or ethnic group, and socioeconomic background. Despite these limitations, I do think this study highlights the need to rethink our approach to transplantation for alcoholic liver disease, including applying better rules for selecting patients who are at low risk for recidivism that can be applied in a uniform and fair way. I would recommend the development and use of predictive tools and reproducible methods to objectively select patients at low risk for recidivism. Because of the scarcity of organs, we cannot afford to choose candidates in an arbitrary way that cannot be defended in the public eye. We will also need to do a better job of educating the public about the nature of alcoholism as a disease and the ability to successfully rehabilitate patients while saving lives with liver transplantation.
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1837
editorials Disclosure forms provided by the author are available with the full text of this article at NEJM.org. From the Center for Liver Disease and Transplantation, Columbia University College of Physicians and Surgeons, New York. 1. Mathurin P, Moreno C, Samuel D, et al. Early liver transplanta-
tion for severe alcoholic hepatitis. N Engl J Med 2011;365:1790-800.
2. Osorio RW, Ascher NL, Avery M, Bacchetti P, Roberts JP,
Lake JR. Predicting recidivism after orthotopic liver transplantation for alcoholic liver disease. Hepatology 1994;20:105-10. 3. Krom RA. Liver transplantation and alcohol: who should get transplants? Hepatology 1994;20:28S-32S.
1838
4. Gish RG, Lee AH, Keeffe EB, Rome H, Concepcion W, Es-
quivel CO. Liver transplantation for patients with alcoholism and end-stage liver disease. Am J Gastroenterol 1993;88:1337-42. 5. Lucey MR, Brown KA, Everson GT, et al. Minimal criteria for placement of adults on the liver transplant waiting list: a report of a national conference organized by the American Society of Transplant Physicians and the American Association for the Study of Liver Diseases. Liver Transpl Surg 1997;3:628-37. 6. Tandon P, Goodman KJ, Ma MM, et al. A shorter duration of pre-transplant abstinence predicts problem drinking after liver transplantation. Am J Gastroenterol 2009;104:1700-6. Copyright © 2011 Massachusetts Medical Society.
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
c or r e sp ondence
Early or Late Parenteral Nutrition in Critically Ill Adults To the Editor: Casaer et al. (Aug. 11 issue)1 provide evidence that early supplemental parenteral nutrition appeared to be inferior to withholding parenteral nutrition until day 8 of the intensive care unit (ICU) stay. However, we think that some limitations of the study should be addressed. Half the patients included were extubated at day 2 and discharged from the ICU at day 4, when they were already receiving two thirds of the caloric demand as oral feeding; this casts doubt on the indication for parenteral nutrition and the correct application of the European Society of Parenteral and Enteral Nutrition guidelines.2 A total of 60% of the study patients underwent cardiac surgery, for which artificial nutritional support is rarely indicated. Initiation of an oral diet after successful extubation is the standard of care. Patients with the strongest indication for nutritional support3 (body-mass index [BMI, the weight in kilograms divided by the square of the height in meters], <17) were excluded from the study. In addition, by neglecting stepwise increases in parenteral nutrition,4 hyperalimentation during stress metabolism with an energy goal up to 32 kcal per kilogram of body weight per day may have been achieved as early as day 3. Concomitant intensive insulin therapy necessarily masked meta bolic consequences such as severe hyperglycemia. In our mind, better selection of patients may have indeed shown the benefits of supplemental parenteral nutrition. Thomas W. Felbinger, M.D., Ph.D. Neuperlach Medical Center Munich, Germany
speaking fees from Abbott, Baxter, B. Braun, Fresenius Kabi, and Nestlé. No other potential conflict of interest relevant to this letter was reported. 1. Casaer MP, Mesotten D, Hermans G, et al. Early versus late
parenteral nutrition in critically ill adults. N Engl J Med 2011; 365:506-17. 2. Singer P, Berger MM, Van den Berghe G, et al. ESPEN guidelines on parenteral nutrition: intensive care. Clin Nutr 2009;28: 387-400. 3. Braunschweig CL, Levy P, Sheean PM, Wang X. Enteral compared with parenteral nutrition: a meta-analysis. Am J Clin Nutr 2001;74:534-42. 4. Krishnan JA, Parce PB, Martinez A, Diette GB, Brower RG. Caloric intake in medical ICU patients: consistency of care with guidelines and relationship to clinical outcomes. Chest 2003; 124:297-305.
To the Editor: We think that major design flaws in the trial by Casaer et al. compromise interpretation of the results. The early-initiation group first received 20% glucose solution, whereas the late-initiation group received only 5% glucose solution. The former will cause hyperglycemia and a requirement for insulin administration, both of which are associated with increased infections, longer ICU stays, and overall worse outcomes.1 this week’s letters 1839 Early or Late Parenteral Nutrition in Critically Ill Adults 1842 Horse versus Rabbit Antithymocyte Globulin in Aplastic Anemia 1844 Apixaban after Acute Coronary Syndrome 1845 Oxygen Sensing, Homeostasis, and Disease
Markus A. Weigand, M.D. Konstantin Mayer, M.D. University Hospital of Giessen and Marburg Giessen, Germany
[email protected] Dr. Felbinger reports receiving speaking fees from Abbott, Baxter, B. Braun, and Fresenius Kabi; and Dr. Mayer, receiving
1846 Case 23-2011 — Legal Considerations 1847 Noninvasive Prenatal Diagnosis of a Fetal Microdeletion Syndrome
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1839
The
n e w e ng l a n d j o u r na l
Contrary to guideline recommendations,2,3 the calorie intake was excessive and the protein intake was low; this would be expected to reduce any benefit from supplemental parenteral nutrition. A total of 60% of patients had undergone cardiac surgery. These patients, who generally have a low incidence of malnutrition and tolerate enteral nutrition, would not normally be administered early parenteral nutrition. Nevertheless, the initiation of enteral nutrition was delayed until the second day in the ICU, even though guidelines recommend that enteral nutrition commence as soon as possible after ICU admission.2-4 Last, the trial was not blinded, potentially creating a systematic bias in aspects of clinical practice such as blood sugar management, maintenance of central venous access, and patient discharge, all of which could affect the study outcomes. Michael J. O’Leary, M.D. Suzie Ferrie, R.D., C.N.S.C.
m e dic i n e
and decreased morbidity.2 Thus, meeting energy requirements increases complications, whereas meeting protein requirements may improve the outcome. Critically ill patients increase endogenous glucose production to 3 mg per kilogram per minute or 300 g of glucose per day,3 which is not reduced by feeding,3 whereas maximal glucose oxidation rates are 6 to 7 mg per kilogram per minute.4 Excess glucose appearance from full feeding plus unsuppressed endogenous production leads to hyperglycemia (an infection risk) and lipogenesis (leading to increased hepatic steatosis and the need for or increased duration of ventilatory support). Limiting carbohydrate feeding to 3 mg per kilogram per minute to reduce total energy by 25 to 30% while maintaining protein intake should be explored in future trials, particularly to determine whether the reduced inflammation seen with full feeding is retained. Bruce R. Bistrian, M.D., Ph.D.
Dr. O’Leary reports receiving honoraria for lectures from Baxter Healthcare. No other potential conflict of interest relevant to this letter was reported.
Beth Israel Deaconess Medical Center Boston, MA
[email protected] Dr. Bistrian reports receiving consulting fees from Nestlé. No other potential conflict of interest relevant to this letter was reported.
1. The NICE-SUGAR Study Investigators. Intensive versus con-
1. Singer P, Anbar R, Cohen J, et al. The Tight Calorie Control
ventional glucose control in critically ill patients. N Engl J Med 2009;360:1283-97. 2. McClave SA, Martindale RG, Vanek VW, et al. Guidelines for the provision and assessment of nutrition support therapy in the adult critically ill patient: Society of Critical Care Medicine (SCCM) and American Society for Parenteral and Enteral Nutrition (A.S.P.E.N.). JPEN J Parenter Enteral Nutr 2009;33:277-316. 3. Singer P, Berger MM, Van den Berghe G, et al. ESPEN guidelines on parenteral nutrition: intensive care. Clin Nutr 2009;28: 387-400. 4. Heyland DK, Dhaliwal R, Drover JW, Gramlich L, Dodek P. Canadian clinical practice guidelines for nutrition support in mechanically ventilated, critically ill adult patients. JPEN J Parenter Enteral Nutr 2003;27:355-73.
Study (TICACOS): a prospective, randomized, controlled pilot study of nutritional support in critically ill patients. Intensive Care Med 2011;37:601-9. 2. Arabi YM, Tamim HM, Dhar GS, et al. Permissive underfeeding and intensive insulin therapy in critically ill patients: a randomized controlled trial. Am J Clin Nutr 2011;93:569-77. 3. Tappy L, Schwarz JM, Schneiter P, et al. Effects of isoenergetic glucose-based or lipid-based parenteral nutrition on glucose metabolism, de novo lipogenesis, and respiratory gas exchanges in critically ill patients. Crit Care Med 1998;26:860-7. 4. Wolfe RR, O’Donnell TF Jr, Stone MD, Richmand DA, Burke JF. Investigation of factors determining optimal glucose infusion rate in total parenteral nutrition. Metabolism 1980;29:892900.
To the Editor: In the comparison of adequate enteral feeding with enteral plus parenteral feeding, the combined feeding group probably was mildly overfed, based on a daily measured energy expenditure of 25 kcal per kilogram in similar patients in the ICU,1 and complications increased. Comparing adequate with inadequate enteral feeding, mortality was marginally improved in the fed group, whereas complications increased significantly.1 Mild underfeeding at 60 to 75% of actual energy expenditure with the provision of adequate protein marginally improved mortality
To the Editor: The study by Casaer and colleagues shows that despite good glycemic control, the more parenteral nutrition patients receive, the greater the risk of complications and the longer the ICU length of stay. It was a pity the study did not have a third group (no parenteral nutrition), since there are no data showing that late supplemental parenteral nutrition has any benefit in patients receiving enteral nutrition. An editorial in the Journal 40 years ago questioned the safety of hyperalimentation and suggested that its use be “severely restricted.” 1 This assessment appears to have stood the test of time.2,3
Royal Prince Alfred Hospital Sydney, NSW, Australia
[email protected]
1840
of
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
correspondence
Parenteral nutrition should be reserved for patients with anatomical abnormalities of the gastrointestinal tract in whom enteral nutrition is not feasible; these were not the characteristics of the patients included in the study. It is also important to note that as compared with current practice, patients in both groups in the study by Casaer et al. were grossly underfed, receiving on average 20% of their caloric goal through the enteral route. Paul E. Marik, M.D. Eastern Virginia Medical School Norfolk, VA
[email protected] No potential conflict of interest relevant to this letter was reported. 1. Duma RJ. First of all do no harm. N Engl J Med 1971;285:
1258-9.
2. Marik PE, Pinsky MR. Death by total parenteral nutrition.
Intensive Care Med 2003;29:867-9.
3. Marik PE. Death by total parenteral nutrition: the saga con-
tinues. Crit Care Med 2011;39:1536-7.
assess these conflicting guidelines. We excluded patients with chronic, severe malnutrition (BMI, <17) because available data suggested that not providing parenteral nutrition to such patients was unethical. Second, normoglycemia was targeted and achieved in both study groups, and all patients required insulin. Hence, confounding effects from early parenteral nutrition on blood glucose were avoided.4 Evidence does not support negative effects of higher insulin doses on infections or ICU stay, and the Normoglycemia in Intensive Care Evaluation–Survival Using Glucose Algorithm Regulation trial (NCT00220987) showed no effects on such outcomes.5 Third, although intensive care providers were aware of the treatment allocation, all outcome assessors and physicians responsible for hospital discharge decisions were unaware of the allocation. Fourth, generalizability of the results was questioned with statements that the patients involved in our study would not meet community consensus for the requirement of parenteral nutrition. We disagree. The EPaNIC trial enrolled severely ill patients, including a large subgroup of 517 patients who could not be given enteral nutrients because of anatomical contraindications. Prior to our study, few clinicians would have doubted that such patients benefit from early parenteral nutrition; however, parenteral nutrition was even more deleterious in this subgroup. Thus, patient selection was not an explanation for the adverse outcome associated with early parenteral nutrition. Indeed, no subgroup, whether based on type or severity of illness, BMI, or nutritional risk benefited from early parenteral nutrition. Hence, we agree with Marik that our data also challenge the late parenteral nutrition approach and that research is needed to assess whether initiation of parenteral nutrition should be further delayed in critically ill patients. Perhaps even the widely accepted standard of early enteral nutrition during critical illness should be revisited.
The Authors Reply: The issues raised by Felbinger et al., O’Leary and Ferrie, and Bistrian indicate that the rationale and design of the Early Parenteral Nutrition Completing Enteral Nutrition in Adult Critically Ill Patients (EPaNIC; ClinicalTrials.gov number, NCT00512122) trial require clarification. First, only critically ill patients who were deemed to be at risk for malnutrition according to the nutritional risk screening score were included.1 Indeed, few patients ever achieved the nutritional target enterally while in the ICU. The late-initiation group was permitted to accumulate a pronounced nutritional deficit, whereas the early-initiation group received supplemental parenteral nutrition. The early-initiation group received progressively increased nutrients (from 400 to 800 kcal per day) followed by an individualized caloric target based on age, sex, and ideal body weight. This resulted in ±25 kcal per kilogram per day of total energy from day 4 onward — not “overfeeding” by European standards.2 The protein content of the commercially available parenteral nutrition formula was lim- Michael P. Casaer, M.D. ited; hard evidence in favor of more protein is Alexander Wilmer, M.D., Ph.D. lacking. The early-initiation group received far Greet Van den Berghe, M.D., Ph.D. more energy than suggested by the American Catholic University of Leuven guidelines, which recommend late parenteral Leuven, Belgium 3 nutrition. The EPaNIC study was designed to
[email protected]
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1841
The
n e w e ng l a n d j o u r na l
Since publication of their article, the authors report no further potential conflict of interest. 1. Kondrup J, Allison SP, Elia M, Vellas B, Plauth M. ESPEN
guidelines for nutrition screening 2002. Clin Nutr 2003;22:41521. 2. Singer P, Berger MM, Van den Berghe G, et al. ESPEN guidelines on parenteral nutrition: intensive care. Clin Nutr 2009;28: 387-400. 3. Martindale RG, McClave SA, Vanek VW, et al. Guidelines for
of
m e dic i n e
the provision and assessment of nutrition support therapy in the adult critically ill patient: Society of Critical Medicine and American Society for Parenteral and Enteral Nutrition: executive summary. Crit Care Med 2009;37:1757-61. 4. Van den Berghe G, Wouters P, Weekers F, et al. Intensive insulin therapy in critically ill patients. N Engl J Med 2001;345: 1359-67. 5. The NICE-SUGAR Study Investigators. Intensive versus conventional glucose control in critically ill patients. N Engl J Med 2009;360:1283-97.
Horse versus Rabbit Antithymocyte Globulin in Aplastic Anemia To the Editor: In their randomized trial comparing rabbit antithymocyte globulin (ATG) with horse ATG, Scheinberg et al. (Aug. 4 issue)1 report the superiority of horse ATG in terms of response and survival. We previously conducted a phase 2 study of rabbit ATG plus cyclosporine as first-line therapy in patients with severe aplastic anemia.2 The perception of our transplantation team was that rabbit ATG led to a decreased risk of infusion reactions, an impression that was largely confirmed in our study. However, Scheinberg et al. report very little infusion-related toxicity with horse ATG, which has not been our experience. Several of our patients who received horse ATG had severe chills, tachycardia, high fever, and hypotension despite aggressive premedication, hydration, and supportive care. In the article by Scheinberg et al. and its accompanying online Supplementary Appendix (available with the full text of the article at NEJM.org), several details about the treatment regimen are not clear. This is an ongoing issue with clinical-trial reporting in general, as was recently highlighted in a retrospective study3 and an associated commentary.4 Would the authors kindly provide the following information regarding the administration of horse ATG: test-dose strategy, desensitization protocol, premedication regimen (including glucocorticoid regimen), infusion duration, and management of acute reactions? Michael S. Mathisen, Pharm.D. Farhad Ravandi, M.D. University of Texas M.D. Anderson Cancer Center Houston, TX
[email protected] No potential conflict of interest relevant to this letter was reported.
1842
1. Scheinberg P, Nunez O, Weinstein B, et al. Horse versus rab-
bit antithymocyte globulin in acquired aplastic anemia. N Engl J Med 2011;365:430-8. 2. Garg R, Faderl S, Garcia-Manero G, et al. Phase II study of rabbit antithymocyte globulin, cyclosporine, and granulocyte colony-stimulating factor in patients with aplastic anemia and myelodysplastic syndrome. Leukemia 2009;23:1297-302. 3. Duff JM, Leather H, Walden EO, LaPlant KD, George TJ Jr. Adequacy of published oncology randomized controlled trials to provide therapeutic details needed for clinical application. J Natl Cancer Inst 2010;102:702-5. 4. George TJ Jr, Leather H. Protocol information provides opportunity for patient safety and quality improvement. J Clin Oncol 2011;29(18):e563.
To the Editor: Scheinberg et al. report a markedly inferior rate of hematologic response in patients with severe aplastic anemia after 6 months of treatment with rabbit ATG at a dose of 3.5 mg per kilogram of body weight per day for 5 days, as compared with horse ATG (37% vs. 68%). A previous retrospective analysis reported a similarly low response rate (35%) for rabbit ATG.1 The latter study, however, raised the issue of ATG dose, since the median applied dose was criticized as relatively low (2.5 mg per kilogram per day for 5 days).2 Since horse ATG became unavailable in Europe in 2008, we prospectively treated all nine adult patients with severe aplastic anemia who were not eligible for first-line allogeneic stem-cell transplantation at our institution with rabbit ATG (Genzyme) administered at a high dose (5 mg per kilogram per day for 5 days) (Table 1). Only three patients (33%) had a hematologic response at 6 months, as compared with 9 of 13 patients (69%) who were treated with horse ATG (at a dose of 15 mg per kilogram per day for 5 days) from 2000 through 2007. Therefore, dose escalation appears unlikely to overcome the disappointing response rate of rabbit
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
The
n e w e ng l a n d j o u r na l
Since publication of their article, the authors report no further potential conflict of interest. 1. Kondrup J, Allison SP, Elia M, Vellas B, Plauth M. ESPEN
guidelines for nutrition screening 2002. Clin Nutr 2003;22:41521. 2. Singer P, Berger MM, Van den Berghe G, et al. ESPEN guidelines on parenteral nutrition: intensive care. Clin Nutr 2009;28: 387-400. 3. Martindale RG, McClave SA, Vanek VW, et al. Guidelines for
of
m e dic i n e
the provision and assessment of nutrition support therapy in the adult critically ill patient: Society of Critical Medicine and American Society for Parenteral and Enteral Nutrition: executive summary. Crit Care Med 2009;37:1757-61. 4. Van den Berghe G, Wouters P, Weekers F, et al. Intensive insulin therapy in critically ill patients. N Engl J Med 2001;345: 1359-67. 5. The NICE-SUGAR Study Investigators. Intensive versus conventional glucose control in critically ill patients. N Engl J Med 2009;360:1283-97.
Horse versus Rabbit Antithymocyte Globulin in Aplastic Anemia To the Editor: In their randomized trial comparing rabbit antithymocyte globulin (ATG) with horse ATG, Scheinberg et al. (Aug. 4 issue)1 report the superiority of horse ATG in terms of response and survival. We previously conducted a phase 2 study of rabbit ATG plus cyclosporine as first-line therapy in patients with severe aplastic anemia.2 The perception of our transplantation team was that rabbit ATG led to a decreased risk of infusion reactions, an impression that was largely confirmed in our study. However, Scheinberg et al. report very little infusion-related toxicity with horse ATG, which has not been our experience. Several of our patients who received horse ATG had severe chills, tachycardia, high fever, and hypotension despite aggressive premedication, hydration, and supportive care. In the article by Scheinberg et al. and its accompanying online Supplementary Appendix (available with the full text of the article at NEJM.org), several details about the treatment regimen are not clear. This is an ongoing issue with clinical-trial reporting in general, as was recently highlighted in a retrospective study3 and an associated commentary.4 Would the authors kindly provide the following information regarding the administration of horse ATG: test-dose strategy, desensitization protocol, premedication regimen (including glucocorticoid regimen), infusion duration, and management of acute reactions? Michael S. Mathisen, Pharm.D. Farhad Ravandi, M.D. University of Texas M.D. Anderson Cancer Center Houston, TX
[email protected] No potential conflict of interest relevant to this letter was reported.
1842
1. Scheinberg P, Nunez O, Weinstein B, et al. Horse versus rab-
bit antithymocyte globulin in acquired aplastic anemia. N Engl J Med 2011;365:430-8. 2. Garg R, Faderl S, Garcia-Manero G, et al. Phase II study of rabbit antithymocyte globulin, cyclosporine, and granulocyte colony-stimulating factor in patients with aplastic anemia and myelodysplastic syndrome. Leukemia 2009;23:1297-302. 3. Duff JM, Leather H, Walden EO, LaPlant KD, George TJ Jr. Adequacy of published oncology randomized controlled trials to provide therapeutic details needed for clinical application. J Natl Cancer Inst 2010;102:702-5. 4. George TJ Jr, Leather H. Protocol information provides opportunity for patient safety and quality improvement. J Clin Oncol 2011;29(18):e563.
To the Editor: Scheinberg et al. report a markedly inferior rate of hematologic response in patients with severe aplastic anemia after 6 months of treatment with rabbit ATG at a dose of 3.5 mg per kilogram of body weight per day for 5 days, as compared with horse ATG (37% vs. 68%). A previous retrospective analysis reported a similarly low response rate (35%) for rabbit ATG.1 The latter study, however, raised the issue of ATG dose, since the median applied dose was criticized as relatively low (2.5 mg per kilogram per day for 5 days).2 Since horse ATG became unavailable in Europe in 2008, we prospectively treated all nine adult patients with severe aplastic anemia who were not eligible for first-line allogeneic stem-cell transplantation at our institution with rabbit ATG (Genzyme) administered at a high dose (5 mg per kilogram per day for 5 days) (Table 1). Only three patients (33%) had a hematologic response at 6 months, as compared with 9 of 13 patients (69%) who were treated with horse ATG (at a dose of 15 mg per kilogram per day for 5 days) from 2000 through 2007. Therefore, dose escalation appears unlikely to overcome the disappointing response rate of rabbit
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
correspondence
cient details with respect to ATG administration. In our study, infusion-related toxic effects were common, as has been described in reports regarding ATG administration for the past 30 years Rabbit ATG Horse ATG in diverse clinical settings.1-4 We did not intend to Variable (N = 9) (N = 13) detail familiar allergic reactions and other toxic Age (yr) effects associated with a standard regimen. Such Median 29 39 reactions can be severe but are usually managed Range 20–56 18–63 with minimal residual adverse effects. We proMale sex (no.) 6 10 vide a detailed description of the protocol and serious adverse events in the Supplementary ApVery severe aplastic anemia 5 6 (no.) pendix to our article, available at NEJM.org. Halkes and colleagues describe their similar Follow-up (yr) experience with rabbit ATG as a first-line theraMedian 1.7 3.9 py in patients with severe aplastic anemia, inRange 1.0–2.6 0.9–10.7 cluding different doses of this agent used in Response at 6 mo after practice. The ATG doses in our study were based treatment (no.) on historical-use data, not on formal dose-findAlive 9 13 ing studies. Since the mechanisms that are rePartial response 3 7 sponsible for the differences in response rates Complete response 0 2 have not been established, it is not clear whether Partial or complete 3 9 a lower (or higher) dose would alter the outresponse come. An obvious difference between the types 5 4 Allogeneic stem-cell transof ATG used in the clinic is the more profound plantation as secondlymphocytopenia after the administration of line treatment (no.) rabbit ATG. Strategies for using lower doses to Overall survival at 2 yr (%) 89 (estimated) 92 shorten the duration of this effect have been proposed, but such reduced doses may not re* The dose of rabbit antithymocyte globulin (ATG) was 5 mg per kilogram per day for 5 days for patients treated from produce the kinetics of lymphodepletion ob2008 through 2010. The dose of horse ATG was 15 mg served with horse ATG. In renal transplantation, per kilogram per day for 5 days for patients treated from for example, the use of a lower total dose of rab2000 through 2007. bit ATG (3 to 5 mg per kilogram) resulted in prolonged depletion of T cells and subsets for a ATG, as compared with horse ATG, in patients year, similar to that reported with higher total with severe aplastic anemia. doses of rabbit ATG (7 to 10 mg per kilogram).4,5 Constantijn J.M. Halkes, M.D., Ph.D. Philip Scheinberg, M.D. Hendrik Veelken, M.D., Ph.D. Neal S. Young, M.D. J.H. Frederik Falkenburg, M.D., Ph.D. National Heart, Lung, and Blood Institute Table 1. Comparison of Rabbit ATG and Horse ATG as First-Line Therapy in Patients with Severe Aplastic Anemia.*
Leiden University Medical Center Leiden, the Netherlands
[email protected] No potential conflict of interest relevant to this letter was reported.
Bethesda, MD
[email protected]
1. Atta EH, Dias DD, Marra VL, de Azevedo AM. Comparison
1. Champlin R, Ho W, Gale RP. Antithymocyte globulin treat-
between horse and rabbit antithymocyte globulin as first-line treatment for patients with severe aplastic anemia: a singlecenter retrospective study. Ann Hematol 2010;89:851-9. 2. Afable MG II, Shaik M, Sugimoto Y, et al. Efficacy of rabbit antithymocyte globulin in severe aplastic anemia. Haematologica 2011;96:1269-75.
The Authors Reply: Mathisen and Ravandi comment that in our report we provide insuffi-
Since publication of their article, the authors report no further potential conflict of interest.
ment in patients with aplastic anemia: a prospective randomized trial. N Engl J Med 1983;308:113-8. 2. Young N, Griffith P, Brittain E, et al. A multicenter trial of antithymocyte globulin in aplastic anemia and related diseases. Blood 1988;72:1861-9. 3. Gaber AO, First MR, Tesi RJ, et al. Results of the doubleblind, randomized, multicenter, phase III clinical trial of Thymoglobulin versus Atgam in the treatment of acute graft rejection episodes after renal transplantation. Transplantation 1998;66:29-37.
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1843
The
n e w e ng l a n d j o u r na l
4. Brennan DC, Flavin K, Lowell JA, et al. A randomized, dou-
ble-blinded comparison of Thymoglobulin versus Atgam for induction immunosuppressive therapy in adult renal transplant recipients. Transplantation 1999;67:1011-8. [Erratum, Transplantation 1999;67:1386.]
of
m e dic i n e
5. Pankewycz O, Leca N, Kohli R, et al. Low-dose rabbit anti-
thymocyte globulin induction therapy results in prolonged selective lymphocyte depletion irrespective of maintenance immunosuppression. Transplant Proc 2011;43:462-5.
Apixaban after Acute Coronary Syndrome To the Editor: The results of the APPRAISE-2 trial (Aug. 25 issue)1 raise doubt about whether combining a long-term anticoagulant (apixaban) with dual antiplatelet therapy in patients with coronary disease can achieve meaningful incremental efficacy with an acceptable risk of bleeding. However, 15 to 20% of these patients may have atrial fibrillation.2 Anticoagulation is the best antithrombotic therapy for the prevention of stroke in patients with atrial fibrillation, and apixaban seems a better option than warfarin by offering a lower risk of systemic embolism, less bleeding, and lower mortality.3 Combined therapy with aspirin and clopidogrel is less effective than anticoagulation in preventing stroke in patients with atrial fibrillation. Anticoagulation alone is nevertheless insufficient to prevent stent thrombosis. Even if triple therapy is associated with higher bleeding rates, it is considered superior to combined aspirin–clopidogrel therapy for the prevention of cardiac events in patients with both atrial fibrillation and the acute coronary syndrome.4,5 Such patients need triple therapy for 1 to 6 months, followed by therapy with one antiplatelet agent plus an anticoagulant there after. After 12 months, an anticoagulant alone (and not antiplatelet therapy) should be given indefinitely. Laurent Fauchier, M.D. Nicolas Clementy, M.D. Centre Hospitalier Universitaire Trousseau Tours, France No potential conflict of interest relevant to this letter was reported. 1. Alexander JH, Lopes RD, James S, et al. Apixaban with anti-
platelet therapy after acute coronary syndrome. N Engl J Med 2011;365:699-708. 2. Lau DH, Huynh LT, Chew DP, Astley CM, Soman A, Sanders P. Prognostic impact of types of atrial fibrillation in acute coronary syndromes. Am J Cardiol 2009;104:1317-23. 3. Granger CB, Alexander JH, McMurray JJ, et al. Apixaban versus warfarin in patients with atrial fibrillation. N Engl J Med 2011;365:981-2. 4. Lip GY, Huber K, Andreotti F, et al. Antithrombotic management of atrial fibrillation patients presenting with acute coro-
1844
nary syndrome and/or undergoing coronary stenting: executive summary — a Consensus Document of the European Society of Cardiology Working Group on Thrombosis, endorsed by the European Heart Rhythm Association (EHRA) and the European Association of Percutaneous Cardiovascular Interventions (EAPCI). Eur Heart J 2010;31:1311-8. 5. Faxon DP, Eikelboom JW, Berger PB, et al. Consensus Document: Antithrombotic therapy in patients with atrial fibrillation undergoing coronary stenting: a North-American perspective. Thromb Haemost 2011;106:572-84.
The Authors Reply: As noted by Fauchier and Clementy, patients with both acute coronary syndrome and atrial fibrillation represent an important population that may benefit from combination antiplatelet and anticoagulant therapy. These patients were excluded from APPRAISE-2 because of the clinical indication for anticoagulation in atrial fibrillation. Multiple observational analyses have shown a higher risk of bleeding with combination anticoagulant and antiplatelet therapy.1 Despite current expert opinion, only a minority of patients with both acute coronary syndrome and atrial fibrillation are treated with triple therapy, reflecting the uncertainty that persists in this area.2-4 It is overly simplistic to assume that a favorable risk–benefit profile can be achieved by combining established antithrombotic regimens in patients with both these conditions. The best antithrombotic regimen in these patients may include combinations of newer agents, the possible elimination of older agents, and different doses and may change over time. With new antiplatelet and anticoagulant agents entering clinical use, additional well-designed, adequately powered, randomized clinical trials that assess both bleeding and ischemic outcomes will be necessary to determine the best regimen for this important population. John H. Alexander, M.D., M.H.S. Renato D. Lopes, M.D., Ph.D. Duke Clinical Research Institute Durham, NC
[email protected]
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
correspondence 2. Lopes RD, Starr A, Pieper CF, et al. Warfarin use and out-
Lars Wallentin, M.D., Ph.D. Uppsala Clinical Research Center Uppsala, Sweden Since publication of their article, the authors report no further potential conflict of interest. 1. Stenestrand U, Lindbäck J, Wallentin L, et al. Anticoagula-
tion therapy in atrial fibrillation in combination with acute myocardial infarction influences long-term outcome. Circulation 2005;112:3225-31. [Erratum, Circulation 2006;113(14):e695.]
comes in patients with atrial fibrillation complicating acute coronary syndromes. Am J Med 2010;123:134-40. 3. Lopes RD, Elliott LE, White HD, et al. Antithrombotic therapy and outcomes of patients with atrial fibrillation following primary percutaneous coronary intervention: results from the APEX-AMI trial. Eur Heart J 2009;30:2019-28. 4. Depta JP, Cannon CP, Fonarow GC, Zhao X, Peacock WF, Bhatt DL. Patient characteristics associated with the choice of triple antithrombotic therapy in acute coronary syndromes. Am J Cardiol 2009;104:1171-8.
Oxygen Sensing, Homeostasis, and Disease To the Editor: Two points could have been added to the review article by Semenza (Aug. 11 issue)1 on a fundamental mechanism in cell, tissue, and organ metabolism. First, considering the metabolic signals modifying signal transduction, the author mentions hypoxia as the sole trigger to expression of hypoxia-inducible factor 1 (HIF-1). One study2 pointed out the possible role of nonhypoxic partial pressure of oxygen (PO2) variation as a trigger to HIF-1α. In this model, transient normobaric hyperoxia followed by a return to normoxia led to a significant increase in serum erythropoietin concentration. This increase began 8 hours after the return to normoxia, which is consistent with the time frame of gene expression. Second, in the section on co-opted adaptation to hypoxia in cancer, the author points out that intratumoral hypoxia is associated with increased risks of metastasis and death. This finding has been the trigger for research on the consequences of hyperoxia on cancer-cell survival. A recent hypothesis3 is that transient normobaric hyperoxia could be deleterious for tumor cells and, in parallel, beneficial (cytoprotective) for normal cells; this could open interesting therapeutic options in adjuvant treatments against cancer. David De Bels, M.D. Francis Corazza, M.D., Ph.D. Brugmann University Hospital Brussels, Belgium
[email protected]
2. Balestra C, Germonpré P, Poortmans JR, Marroni A. Serum
erythropoietin levels in healthy humans after a short period of normobaric and hyperbaric oxygen breathing: the “normobaric oxygen paradox.” J Appl Physiol 2006;100:512-8. 3. De Bels D, Corazza F, Germonpré P, Balestra C. The normobaric oxygen paradox: a novel way to administer oxygen as an adjuvant treatment for cancer? Med Hypotheses 2011;76:467-70.
To the Editor: The review by Semenza examines the role of HIF-1α in the context of hypoxia and ischemia. However, we think that the role of HIF-1α can be seen from another point of view. HIF-1α does not appear to be the clear hypoxic marker.1 It is stabilized not only because of hypoxia but also by metabolic imbalances, induced growth, and lower intracellular pH values. The curious association of metabolic pathways and different metabolites2 with this marker causes doubt about the function of HIF-1 as a molecular regulator to oxygen deprivation. The fact that hypoxia always decreases the intracellular pH value3 discloses another face of HIF-1 beyond the phenomenon called hypoxia. The biologic function of this protein is rather that of a sensitive controller and regulator of an adaptive metabolic response to alterations of the pH value. That this function is independent from the oxygen status of cells is reflected by a multitude of HIF-1 detections at normoxia.4 Altogether, we suggest that HIF-1 is mainly a sensor and regulator for the intracellular pH value. Matthias Kappler, Ph.D. Martin Luther University of Halle-Wittenberg Halle, Germany
[email protected]
Costantino Balestra, Ph.D. Haute Ecole Paul-Henri Spaak Brussels, Belgium
Helge Taubert, Ph.D.
No potential conflict of interest relevant to this letter was reported.
Friedrich Alexander University Erlangen, Germany
1. Semenza GL. Oxygen sensing, homeostasis, and disease.
Alexander W. Eckert, M.D., Ph.D.
N Engl J Med 2011;365:537-47. [Erratum, N Engl J Med 2011; 365:968.]
Martin Luther University of Halle-Wittenberg Halle, Germany
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
1845
The
n e w e ng l a n d j o u r na l
No potential conflict of interest relevant to this letter was reported. 1. Mayer A, Höckel M, Vaupel P. Endogenous hypoxia markers:
case not proven! Adv Exp Med Biol 2008;614:127-36.
2. Denko NC. Hypoxia, HIF1 and glucose metabolism in the
solid tumour. Nat Rev Cancer 2008;8:705-13.
3. Chiche J, Brahimi-Horn MC, Pouysségur J. Tumour hypoxia
induces a metabolic shift causing acidosis: a common feature in cancer. J Cell Mol Med 2010;14:771-94. 4. Semenza GL. Targeting HIF-1 for cancer therapy. Nat Rev Cancer 2003;3:721-32.
of
m e dic i n e
which is required for the generation of reactive oxygen species during chronic intermittent hypoxia, thus delineating a feed-forward loop linking chronic intermittent hypoxia, reactive oxygen species, and HIF-1 in the pathobiology of sleepdisordered breathing. There are compelling clinical and laboratory data associating intratumoral hypoxia with an increased risk of metastasis, treatment failure, and death.4 Several different therapeutic strategies are under investigation that target hypoxic cancer cells, including inhibitors of HIF-1 or critical downstream target gene products (such as vascular endothelial growth factor), “bioreductive” drugs (such as tirapazamine) that are converted to cytotoxins only in hypoxic cells, and anaerobic bacteriolytic therapy.4,5
The Author Replies: I am grateful for the opportunity to discuss stimuli other than continuous hypoxia that activate HIF-1. These other stimuli were not covered in my review article because of space limitations. The most medically relevant of these stimuli is chronic intermittent hypoxia due to sleep-disordered breathing, which Gregg L. Semenza, M.D., Ph.D. results in systemic hypertension.1 Exposure of Johns Hopkins University School of Medicine mice to chronic intermittent hypoxia leads to the Baltimore, MD generation of reactive oxygen species, which in-
[email protected] duce HIF-1 activity in the carotid body and cenSince publication of his article, the author reports no further tral nervous system (CNS) through mechanisms potential conflict of interest. that are distinct from those associated with con1. Peppard PE, Young T, Palta M, Skatrud J. Prospective study tinuous hypoxia.2,3 Treatment with radical scav- of the association between sleep-disordered breathing and hyengers blocks the induction of HIF-1 and the de- pertension. N Engl J Med 2000;342:1378-84. velopment of hypertension. Remarkably, mice 2. Peng YJ, Yuan G, Ramakrishnan D, et al. Heterozygous HIF1alpha deficiency impairs carotid body-mediated systemic rethat are heterozygous for a knockout allele at the sponses and reactive oxygen species generation in mice exposed locus encoding the HIF-1α subunit are protected to intermittent hypoxia. J Physiol 2006;577:705-16. against the development of hypertension and 3. Yuan G, Khan SA, Luo W, Nanduri J, Semenza GL, Prabhakar NR. Hypoxia-inducible factor 1 mediates increased expression other sequelae of chronic intermittent hypoxia, of NADPH oxidase-2 in response to intermittent hypoxia. J Cell and increased reactive oxygen species levels in Physiol 2011;226:2925-33. the CNS do not develop in these animals.2 Recent 4. Wilson WR, Hay MP. Targeting hypoxia in cancer therapy. Nat Rev Cancer 2011;11:393-410. data indicate that HIF-1 activates transcription of 5. Leschner S, Weiss S. Salmonella — allies in the fight against 3 the NOX2 gene encoding NADPH oxidase, cancer. J Mol Med (Berl) 2010;88:763-73.
Case 23-2011 — Legal Considerations To the Editor: Having received comments regarding the legal issues raised in our article about a pregnant woman with placenta accreta who declined blood products (July 28 issue),1 we wish to acknowledge that that the legal issues that were discussed are decided on a state-bystate basis and that the legal analysis presented was based on Massachusetts law. As in the case of Norwood Hospital v. Munoz that we cited, a determination was made in this case that forced blood transfusion would not be appropriate ow1846
ing to the existence of a legal caregiver (the husband) for the children. Whether a particular provider would in fact ever exercise the legal right to transfuse a patient who has no substitute caregiver for dependents against his or her will is a different question — and one for clinicians, ethicists, and others on the care team to determine. Nothing about this issue is easy, either clinically or ethically. But Massachusetts law has chosen a path that recognizes this complexity
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
correspondence
and provides for some flexibility in approach, which may be the best it can do, both ethically and legally. Joshua L. Abrams, J.D. Jeffrey L. Ecker, M.D. William H. Barth, Jr., M.D.
Since publication of their article, the authors report no further potential conflict of interest. 1. Case Records of the Massachusetts General Hospital (Case
23-2011). N Engl J Med 2011;365:359-66.
Massachusetts General Hospital Boston, MA
Noninvasive Prenatal Diagnosis of a Fetal Microdeletion Syndrome To the Editor: The definitive diagnosis of fetal aneuploidy and genomic imbalances requires invasive collection of fetal cells through amniocentesis or chorionic villus sampling. These methods are associated with fetal loss and parental anxiety. Analyses of DNA in maternal plasma have shown the potential for noninvasive diagnosis of common aneuploidies.1 A couple presented for prenatal genetic counseling at the Magee–Womens Hospital of the University of Pittsburgh Medical Center. They had previously had a child with developmental delay and dysmorphic features in whom a paternally inherited 4.2-Mb deletion on chromosome 12 between bands 12p11.22 and 12p12.1 had been diagnosed (Fig. 1A). Amniocentesis was A Family Pedigree
B Results of CGH
46, XY, del(12) (p11.22p12.1) Asperger’s syndrome Facial dysmorphia Brachydactyly Short stature
22.50 Mb
Log2
P 46, XX, del(12) (p11.22p12.1) Failure to thrive Developmental delay Dysmorphic features
performed at 21 weeks of gestation, and microarray-based comparative genomic hybridization identified the same heterozygous deletion in the male fetus (Fig. 1B). A maternal blood sample was drawn at 35 weeks of gestation, and plasma DNA was extracted without further enrichment. Real-time polymerase-chain-reaction assay revealed that the relative prevalence of fetal DNA was 5.7%. The maternal plasma DNA was then used as a substrate for Illumina HiSeq2000 DNA sequencing, generating 243,340,714 single-end reads, of which 75% mapped uniquely and perfectly to the Genome Reference Consortium human genome (build 37), GRCh37. Seven maternal plasma samples in which both the mother and fetus
46, XY, del(12) (p11.22p12.1)
12p12.2
25.00 Mb
12p12.1
27.50 Mb
12p11.23
30.00 Mb
12p11.22
12p11.21
+1 0 −1
Figure 1. Pedigree of the Affected Family and Results of Comparative Genomic Hybridization (CGH) Showing Deletion on Chromosome 12. In Panel A, the arrow points to the affected proband. Male and female family members are indicated by squares and circles, respectively. P denotes the fetus being evaluated for a possible deletion on chromosome 12. In Panel B, the CGH profile of a sample obtained on amniocentesis shows a heterozygous deletion in the short arm of chromosome 12 between bands 12p11.22 and 12p12.1. A loss in DNA copy number (deletion) was detected by oligonucleotide probes, as represented by dots showing a log2 ratio of test-versus-reference value of −1. Dots with a log2 ratio of 0 represent probes with no change in copy number.
n engl j med 365;19
nejm.org
november 10, 2011
The New England Journal of Medicine
1847
The
n e w e ng l a n d j o u r na l
were known to be diploid for chromosomes 12 and 14 were also sequenced as reference libraries. Using our previously described method,2 we determined whether the maternal sample (PL565) was diploid in each of 22 nonoverlapping 4-Mb regions on chromosomes 12 and 14 through a pairwise comparison with each of the reference libraries. We detected a 4-Mb depletion in DNA copy number on chromosome 12p in PL565 in all seven pairwise comparisons with a normal sample (adjusted P≤0.05 for all comparisons) (Fig. S1 in the Supplementary Appendix, available with the full text of this letter at NEJM .org). Tests for all other 21 regions resulted in nonsignificant adjusted P values for all seven pairwise comparisons (Table S1 in the Supplementary Appendix). In summary, we have shown proof of concept that a fetal chromosomal microdeletion can be identified by means of noninvasive analysis of DNA in maternal plasma. David Peters, Ph.D. Tianjiao Chu, Ph.D. Svetlana A. Yatsenko, M.D. Nancy Hendrix, M.D. W. Allen Hogge, M.D. Urvashi Surti, Ph.D. Kimberly Bunce, Ph.D. Mary Dunkel, M.S. Patricia Shaw, B.S. Aleksandar Rajkovic, M.D. Magee–Womens Research Institute Pittsburgh, PA
[email protected] Supported by grants from the Magee–Womens Research Institute and Foundation and from the National Institutes of Health (1R01HD068578-01, to Dr. Peters). Disclosure forms provided by the authors are available with the full text of this letter at NEJM.org. 1. Fan HC, Blumenfeld YJ, Chitkara U, Hudgins L, Quake SR.
Noninvasive diagnosis of fetal aneuploidy by shotgun sequencing DNA from maternal blood. Proc Natl Acad Sci U S A 2008; 105:16266-71. 2. Chu T, Bunce K, Hogge WA, Peters DG. Statistical model for whole genome sequencing and its application to minimally invasive diagnosis of fetal genetic disease. Bioinformatics 2009;25: 1244-50.
of
m e dic i n e
• Letters in reference to a Journal article must not exceed 175 words (excluding references) and must be received within 3 weeks after publication of the article. • Letters not related to a Journal article must not exceed 400 words. • A letter can have no more than five references and one figure or table. • A letter can be signed by no more than three authors. • Financial associations or other possible conflicts of interest must be disclosed. Disclosures will be published with the letters. (For authors of Journal articles who are responding to letters, we will only publish new relevant relationships that have developed since publication of the article.) • Include your full mailing address, telephone number, fax number, and e-mail address with your letter. • All letters must be submitted at authors.NEJM.org. Letters that do not adhere to these instructions will not be considered. We will notify you when we have made a decision about possible publication. Letters regarding a recent Journal article may be shared with the authors of that article. We are unable to provide prepublication proofs. Submission of a letter constitutes permission for the Massachusetts Medical Society, its licensees, and its assignees to use it in the Journal’s various print and electronic publications and in collections, revisions, and any other form or medium.
corrections SHANK3, the Synapse, and Autism (July 14, 2011;365:173-5). The phrase “lower amplitude and frequency of action potentials” should have been “lower amplitude and frequency of miniature excitatory postsynaptic currents (mEPSCs)” in three instances: the first sentence of the fifth paragraph (page 173), beginning “Consistent . . .”; the fourth sentence of the figure legend (page 175), beginning “Dendrites . . .”; and the third sentence of the penultimate paragraph (page 175), beginning “Given . . . .” The article is correct at NEJM.org. Specialist Physician Practices as Patient-Centered Medical Homes (April 29, 2010;362:1555-8). In the fourth paragraph, beginning “The extent . . .” (page 1556), the number of practices mentioned in the third sentence should have been 372, rather than 373. In the fifth paragraph (page 1556), the percentages of practices given in the first sentence should have been 84.6%, 10.3%, and 1.7%, rather than 81%, 12.5%, and 2.7%. Because one pulmonary practice was counted twice and the weights used to calculate values given in the table were not the final weights, the reported percentages of specialist practices serving as primary care physicians were incorrect. The article is correct at NEJM.org.
Correspondence Copyright © 2011 Massachusetts Medical Society.
notices instructions for letters to the editor
Letters to the Editor are considered for publication, subject to editing and abridgment, provided they do not contain material that has been submitted or published elsewhere. Please note the following:
1848
Notices submitted for publication should contain a mailing address and telephone number of a contact person or department. We regret that we are unable to publish all notices received. Notices also appear on the Journal’s Web site
n engl j med 365;19 nejm.org november 10, 2011
The New England Journal of Medicine
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
images in clinical medicine
Amenorrhea, Abdominal Pain, and Weight Gain A
B Left ovary
Right ovary
C
Right ovary
A
21-year-old woman was referred for secondary amenorrhea, Jeroen P.H. van Wijk, M.D. RETAKE 1st van Wijk AUTHOR tests ICM Edith W.M.T. ter Braak, M.D. abdominal pain, and weight gain. Laboratory showed elevated serum 2nd REG F FIGURE a-c estradiol (23,501 pmol per liter), suppressed luteinizing hormone (<0.1 IU per 3rd CASE University Medical Center TITLE Revised liter), and inappropriately high follicle-stimulating hormone Line (27 IU per liter). Pelvic Utrecht, the Netherlands EMail 4-C Enon ultrasonography showed bilateral enlarged polycystic (Panels A and B),SIZE which
[email protected] ARTIST: ovaries mst H/T H/T 39p6 FILL suggested a spontaneous ovarian hyperstimulation syndrome.Combo Subsequent magnetic AUTHOR, resonance imaging showed a pituitary macroadenoma withPLEASE opticNOTE: chiasm compresFigure has been redrawn and type has been reset. carefully. sion (Panel C, arrow). Transsphenoidal resection ofPlease thecheck macroadenoma was performed. Immunohistochemical staining was positive for follicle-stimulating horJOB: 36518 ISSUE: 11-10-11 mone, confirming the diagnosis of a follicle-stimulating hormone–secreting pituitary macroadenoma. After the surgery, the patient did well. Her laboratory values normalized, and she went on to conceive spontaneously and deliver a healthy daughter. The ovarian hyperstimulation syndrome is a potentially serious complication of ovulation induction with gonadotropins, which this patient had not received. As illustrated in this case, spontaneous hypersecretion of follicle-stimulating hormone by a pituitary adenoma may override normal feedback mechanisms to cause the ovarian hyperstimulation syndrome and amenorrhea. Copyright © 2011 Massachusetts Medical Society.
n engl j med 365;19
nejm.org
november 10, 2011
The New England Journal of Medicine
e39