The New England Journal of Medicine
Issue Index Table of contents for December 29, 2011 Vol. 365 No. 26 PERSPECTIVE
Building the Path to Accountable Care
2445-2447
E.S. Fisher, M.B. McClellan, and D.G. Safran
Copyright and Open Access at the Bedside
2447-2449
John C. Newman, and Robin Feldman
The Road Less Traveled
2449-2451
D.R. Fingold
The Savings Illusion — Why Clinical Quality Improvement Fails to Deliver Bottom-Line Results
e48
S.S. Rauh, E.B. Wadsworth, W.B. Weeks, and J.N. Weinstein ORIGINAL ARTICLES
Liberal or Restrictive Transfusion in High-Risk Patients after Hip Surgery
2453-2462
J.L. Carson and Others
Low-Molecular-Weight Heparin and Mortality in Acutely Ill Medical Patients
2463-2472
A.K. Kakkar and Others
Incorporation of Bevacizumab in the Primary Treatment of Ovarian Cancer
2473-2483
R.A. Burger and Others
A Phase 3 Trial of Bevacizumab in Ovarian Cancer
2484-2496
T.J. Perren and Others
SF3B1 and Other Novel Cancer Genes in Chronic Lymphocytic Leukemia
2497-2506
L. Wang and Others REVIEW ARTICLE
Mechanisms of Disease: Proprotein Convertases in Health and Disease
2507-2518
A.W. Artenstein and S.M. Opal EDITORIALS
Transfusion Thresholds in FOCUS
2532-2533
P.J. Barr and K.E.M. Bailie
Mutations in RNA Splicing Machinery in Human Cancers
2534-2535
B. Ebert and O.A. Bernard CORRESPONDENCE
Adenocarcinoma in Barrett's Esophagus
2539-2540
Change in FEV1 over Time in COPD
2540-2541
Community Transmission of Oseltamivir-Resistant A(H1N1)pdm09 Influenza
2541-2542
n engl j med 365;26
nejm.org
december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
The
NEW ENGLA ND JOURNAL
of
MEDICINE
Perspective december 29, 2011
Building the Path to Accountable Care Elliott S. Fisher, M.D., M.P.H., Mark B. McClellan, M.D., Ph.D., and Dana G. Safran, Sc.D.
T
he recent release of the regulations that will govern the early years of Medicare’s implementation of accountable care organizations (ACOs) provides an important moment to reflect on the transition to this new payment model, which offers health care providers flexible financial support for improving care in return for accepting accountability for its overall quality and cost. It’s also an appropriate moment to explore the challenges that must be overcome in order to make more rapid progress. The final regulations for the Medicare Shared Savings Program, released on October 20, 2011, represent diligent work to try to address many concerns raised in response to the draft regulations.1 The regulations allow a broader range of ACO governance structures, reduce the number of required quality measures, create more opportunities for saving
while delaying risk bearing, permit care provided by a greater variety of clinicians (including specialists) to determine which patients are assigned to a given ACO, and allow ACOs to more effectively identify and reach out to their patients. Outside of Medicare, almost 100 provider organizations are already working with private health plans toward contracts containing the core elements of the ACO model: payment tied to improving patient care across the continuum and reducing overall spending growth. Other ACOrelated initiatives are emerging at the regional or state level (e.g., Vermont) or through state Medicaid programs (e.g., New Jersey
and Texas). At least 12 states have enacted legislation to facilitate accountable care reforms. Although no systematic evaluation of emerging ACOs is yet available, early work provides important insights. There is remarkable diversity: some ACOs are starting from scratch by assembling a new network of physicians, with or without a partnering hospital, and others are well- established individual practice associations or integrated systems; they have varying levels of prior experience with performance measurement and risk bearing.2,3 Even integrated systems, however, are finding that they need substantial time and resources to develop the informational, technical, financial, and professional capabilities required to provide and reward coordinated, longitudinal, population-based care. Most private-sector initiatives are therefore emerging as negotiated part-
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2445
PERSPE C T I V E
Building the Path to Accountable Care
Challenges to the Successful Diffusion of the ACO Model and Approaches to Overcoming Them. Challenge
Approach
Providing timely and useful data
Payers provide ACOs with patient-level data to support care management. Quality measures used for accountability are also useful for care improvement.
Overcoming transition costs
Use quality-related payments to support needed ACO investments. Provide up-front funding options for provider groups that need them.
Gaining consumer support
Adopt performance measures that are more meaningful to consumers. Support consumer choice and allow consumers to share savings as well.
Learning what works; using that knowledge to inform policy and practice
Develop and test multipayer or all-payer ACOs where possible. Track and evaluate both public and private ACO implementation.
Clarifying the path forward
Create meaningful alternatives to fee for service for all providers. Measure effect on overall quality and cost in all payment reforms.
nerships between providers and a payer; both sides are investing in infrastructure, data sharing, risk management, patient engagement, and performance-improvement approaches.2,3 These observations suggest that ACO formation should be understood not through the lens of a specific settled contract, but rather as an emerging payment model that can be implemented alongside the fee-forservice payment system while supporting a transition to the goal of fully coordinated and accountable care. Implementation is still at an early stage. We have only preliminary information about the actual performance of many ACOs and no strong evidence on which features are most likely to lead to success in specific circumstances or how ACOs can best be integrated with other reforms intended to promote accountability and high-value care. On the basis of the growing set of ACO experiences and the extensive public comments regarding the Medicare ACO program, we have identified five key challenges and possible approaches to overcoming them (see table). The first challenge is providing timely data and useful performance measures. Even when providers have rich clinical data about 2446
care within their own setting, they are generally blind to the care provided elsewhere. Timely provision of consistent public and private data (including both timely claims data and increasingly sophisticated clinical data) would help ACO clinicians to improve and coordinate care. Performance measures for accountability should be based on data sources that can be used to support care improvement and should be aligned, to the extent possible, across providers and payers to minimize the burden and optimize the effect. The second challenge is overcoming transition costs. Any group that is implementing an ACO faces substantial infrastructure costs to support improved and coordinated care — a necessity that raises questions about whether ACO reforms will generate sufficient returns for providers.4 The Centers for Medicare and Medicaid Services (CMS) took steps to address this concern, including sharing first-dollar savings with ACOs that meet minimum savings thresholds, implementing a program to finance start-up costs for poorly capitalized groups (including physicianled organizations without a hospital and rural ACOs), and aligning ACO performance measures with those of Medicare’s Physician
Quality Reporting System. Other payers are linking ACOs to related reforms (medical homes or episode-based payments) to provide additional support. For example, in its global payment model, Blue Cross Blue Shield of Massachusetts uses a broad set of approaches to help practices develop the capacity to function as an ACO, including in-kind data support, shared-savings and quality payments, and in some cases, direct financial support for initial infrastructure needs. With such assistance, more than two thirds of physicians in Massachusetts are now participating in Blue Cross Blue Shield’s Alternative Quality Contract. Gaining consumer support is the third challenge. If consumers become worried that ACOs are stinting on care, the approach will be quickly rejected. The CMS and private payers should move to incorporate performance measures that have more direct meaning to consumers, such as patient-reported outcomes and additional measures of patients’ experience of care. Many new ACOs have found that making clear to patients that they can continue to choose their providers can allay consumers’ concerns; such choice is a feature of the Medicare regulations. Some ACOs now enable their members
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
PERSPECTIVE
Building the Path to Accountable Care
to share in the savings from accountable care, which can foster greater consumer engagement and support for these delivery reforms. The fourth challenge is to learn what works — and what doesn’t — and then to use that new understanding to inform policy and practice. Assuming responsibility for spending, quality, and health outcomes represents a major shift for most U.S. providers and requires changes to information systems, care processes, staffing, governance, leadership, and approaches to patient engagement. Implementing such changes is particularly challenging for ACOs that face mixed payment models ranging from fee for service to capitation, with many alternatives in between. Developing and testing multipayer ACOs will be important. Funding agencies and foundations should also support a comprehensive effort to track and evaluate both public and private ACO initiatives so that policymakers can make midcourse corrections and ACOs and payers can be better informed about successful strategies. Finally, the fifth challenge is clarifying and reinforcing the path forward. The ongoing debate about the future of health care
reform may encourage some providers to adopt a “wait and see” approach. But the current political debate is unlikely to slow what appears to be an inexorable transition in both public and private insurance programs away from traditional fee-for-service reimbursement. Opponents of the Affordable Care Act support private payers’ efforts to implement payment reforms for accountable care. Consequently, it is more important than ever for providers to consider how they can demonstrably improve care while lowering costs and how they can use emerging payment methods to help them do so. Both the CMS and private payers should also continue to emphasize their commitment to the same goal. A broad range of potentially reinforcing payment models (such as episodebased and medical-home payments) can provide additional momentum for accountable care, especially when performance measures for these other payment models incorporate accountability for overall costs and quality. The success of the ACO model in any of its specific current contractual structures is not ensured, but early results are promising.5 Traditional fee-for-service models
will almost certainly see continued cuts in payment rates — which will make physicians’ and other providers’ work even more difficult if they remain under this payment model. Additional steps to accelerate the transition toward accountable care offer the public and providers a better and more hopeful path. Disclosure forms provided by the authors are available with the full text of this article at NEJM.org. From the Dartmouth Institute for Health Policy and Clinical Practice, Lebanon, NH (E.S.F.); the Engleberg Center for Health Care Reform, Brookings Institution, Washington, DC (M.B.M.); and Blue Cross Blue Shield of Massachusetts, Boston (D.G.S.). 1. Berwick DM. Making good on ACOs’ promise — the final rule for the Medicare Shared Savings Program. N Engl J Med 2011; 365:1753-6. 2. Van Citters AD, Larson BK, Carluzzo KL, et al. Toward accountable care: four health care organizations’ efforts to improve patient care and reduce costs. New York: The Commonwealth Fund, December 2011. 3. Chernew ME, Mechanic RE, Landon BE, Safran DG. Private-payer innovation in Massachusetts: the “Alternative Quality Contract.” Health Aff (Millwood) 2011;30:51-61. 4. Haywood TT, Kosel KC. The ACO model — a three-year financial loss? N Engl J Med 2011;364(14):e27. 5. Song Z, Safran DG, Landon BE, et al. Health care spending and quality in year 1 of the Alternative Quality Contract. N Engl J Med 2011;365:909-18. Copyright © 2011 Massachusetts Medical Society.
Copyright and Open Access at the Bedside John C. Newman, M.D., Ph.D., and Robin Feldman, J.D.
F
or three decades after its publication, in 1975, the Mini– Mental State Examination (MMSE) was widely distributed in textbooks, pocket guides, and Web sites and memorized by countless residents and medical students. The simplicity and ubiquity of this 30-item screening test — covering such functions as arith-
metic, memory, language comprehension, visuospatial skills, and orientation — made it the de facto standard for cognitive screening. Yet all that time, it was under copyright protection. In 2000, its authors, Marshal Folstein, Susan Folstein, and Paul McHugh, began taking steps to enforce their rights, first transferring the copy-
right to MiniMental, a corporation the Folsteins founded, and then in 2001 granting a worldwide exclusive license to Psychological Assessment Resources (PAR) to publish, distribute, and manage all intellectual property rights.1,2 A licensed version of the MMSE can now be purchased from PAR for $1.23 per test. The
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2447
PERSPE C T I V E
Copyright and Open Access at the Bedside
MMSE form is gradually disappearing from textbooks, Web sites, and clinical tool kits.1 Clinicians’ response to this “lockdown” has been muted. A few commentators have expressed concern about continuing to use a now-proprietary tool in training2 or about implications for the developing world,1 echoing debates about patented pharmaceuticals. In our experience, many clinicians are either unaware of the MMSE’s copyright restrictions or simply ignore them, despite the risk of copyright infringement. But then in March 2011, a promising new cognitive screening tool that was to be available through “open access,” the Sweet 16 — a 16-item assessment of thinking, learning, and memory developed by Harvard’s Tamara Fong3 — was removed from the Internet at the request of PAR in an apparent copyright dispute.4 The Sweet 16 includes orientation and three-object recall items, similar to the MMSE’s, along with a digit-span item. This action, unprecedented for a bedside clinical assessment tool, has sent a chill through the academic community; clearly, clinicians and researchers can no longer live in blissful ignorance of copyright. Copyright derives from one of the few powers explicitly mentioned in the U.S. Constitution. Any new intellectual work is under copyright protection automatically from the moment it is fixed in a tangible medium of expression — a category now including blog posts, iPhone apps, and cognitive screening tools. Copyright law grants the author (or owner, for copyright can be transferred) exclusive rights to copy the work, distribute it, make works derivative of it, and perform or display it publicly. These rights last for 2448
70 years past the date of the author’s death, or up to 120 years from the time of creation if the work was done “for hire.” This duration has been retroactively extended several times, so that works published as early as 1923 may remain under copyright today (and will until at least 2019). For persons or entities other than the copyright holder to copy or distribute a work, they must have permission from the owner, usually in the form of a license. Copying or distribution without permission is copyright infringement and carries stiff civil or even criminal penalties. There is limited protection under “fair use” law for certain nonprofit uses of limited parts of a work — for example, for teaching or research — but that exception is narrower than it sounds. One need not have intended to infringe someone’s copyright to be subject to damages of up to $30,000 per work, and willful infringers pay up to $150,000 — and may, under certain circumstances, be subject to a jail term. For clinicians, the risk of infringement is real. Photocopying or downloading the MMSE probably constitutes infringement; those who publish the MMSE on a Web site or pocket card could incur more severe penalties for distribution. Even more chilling is the “takedown” of the Sweet 16, apparently under threat of legal action from PAR (although PAR has not commented publicly). Are the creators of any new cognitive test that includes orientation questions or requires a patient to recall three items subject to action by PAR? However disputable the legal niceties, few physicians or institutions would want to have to argue their case in court. The MMSE case may be a har-
binger of more to come. Many clinical tools we take for granted, such as the Katz Index of Independence in Activities of Daily Living, fall into the same “benign neglect” copyright category as the MMSE did before 2000. At any time, they might be pulled back behind a wall of active copyright enforcement by the authors or their heirs. What can researchers do to ensure that our colleagues can use the tools we develop to improve patient care? One option is to essentially place works in the public domain by declaring free and open rights for all users. The Geriatric Depression Scale, the Patient Health Questionnaire (PHQ-9) depression scale, and the Saint Louis University Mental Status (SLUMS) cognitive assessment tool are all in the public domain. That domain, however offers no mechanism for ensuring that authors are recognized or compensated and no means of guaranteeing that later improvements will be made freely available. The ability to improve a clinical tool is crucial. Even licenses granting wide permission to copy, such as those of the Montreal Cognitive Assessment and the Lawton Instrumental Activities of Daily Living (IADL) scale, while laudable, might still inhibit innovation by permitting legal challenges to improved tools perceived as derivative (as may have been the case with Sweet 16 and the MMSE). A better solution is to apply the principle of “copyleft” from the open-source technology movement to encourage innovation and access while protecting authors’ rights. Copyleft is intellectual jujitsu that uses copyright protection to guarantee the right of anyone to use, modify, copy, and distribute a work, as long as it
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Downloaded from nejm.org at Hinari Phase 2 sites on December 30, 2011. For personal use only. No other uses without permission. Copyright © 2011 Massachusetts Medical Society. All rights reserved.
PERSPECTIVE
Copyright and Open Access at the Bedside
and any derivatives remain under the same license. The author retains the right to offer the work under a different license simultaneously — for example, giving a company specific license to commercialize the work without copy left protections. Popular copyleft licenses include the Creative Commons Attribution-ShareAlike license and the GNU Free Documentation License. Google, Apple, Facebook, and Twitter all use open-source software at the heart of their products, because there is a clear economic benefit to using well-tested, well-validated, continually improved software in the core of complex products. Similarly, there is a clear clinical benefit to using well-tested, well-validated, continually improved clinical tools in complex patient care — as demonstrated by the MMSE’s use before 2000. In a sense, copyleft is how academic medicine has always been assumed to work.2 Restrictive licensing of such basic tools wastes resources, prevents standardization, and detracts from efforts to improve patient care.
We suggest that authors of widely used clinical tools provide explicit permissive licensing, ideally with a form of copyleft. Any new tool developed with public funds should be required to use a copyleft or similar license to guarantee the freedom to distribute and improve it, similar to the requirement for open-access publication of research funded by the National Institutes of Health.5 The solution can be as simple as placing a copy of the tool on the authors’ Web site, with a statement naming or linking to the license. Clinicians and researchers would be free to use, copy, and improve the tool; improvements would have to offer a similar copyleft license, perpetuating the benefits. Yet authors would maintain ownership and copyright of their tool and could profit by licensing it for a fee to commercial users or publishers who wished to include it in a noncopyleft work. The restrictions on the MMSE’s use present clinicians with difficult choices: increase practice costs and complexity, risk copyright in-
fringement, or sacrifice 30 years of practical experience and validation to adopt new cognitive assessment tools. By embracing the principles of copyleft licensing, we can avoid such setbacks and build a more open future of continually improving patient care. Disclosure forms provided by the authors are available with the full text of this article at NEJM.org. From the Division of Geriatrics, San Francisco Veterans Affairs Medical Center, and the University of California San Francisco (J.C.N.); and the Law and Bioscience Project, University of California Hastings College of the Law (R.F.) — all in San Francisco. 1. Martin R, O’Neill D. Taxing your memory. Lancet 2009;373:2009-10. 2. Powsner S, Powsner D. Cognition, copyright, and the classroom. Am J Psychiatry 2005;162:627-8. 3. Fong TG, Jones RN, Rudolph JL, et al. Development and validation of a brief cognitive assessment tool: the Sweet 16. Arch Intern Med 2011;171:432-7. 4. The Hospital Elder Life Program (HELP) (http://www.hospitalelderlifeprogram.org/ private/sweet16-disclaimer.php?pageid= 01.09.00). 5. Feldman R, Nelson C. Open source, open access, and open transfer: market approaches to research bottlenecks. Northwestern Journal of Technology and Intellectual Property. Fall 2008:14-32. Copyright © 2011 Massachusetts Medical Society.
The Road Less Traveled Diane R. Fingold, M.D.
I
worry that the primary care physician is a dying breed. Though it was once considered the noblest profession, U.S. medical students today believe the work is too hard, the hours too long, the pay too low. So they’re choosing to hit the “ROAD” — the high-paying specialties of radiology, ophthalmology, anesthesia, and dermatology. But if you think being a primary care doctor is hard, meet Mary. Not many people have it
harder than Mary. She always leaps to mind when I consider how our health care system fails our patients — and why I chose primary care. I’m Mary’s doctor, and though I care deeply about her, seeing her name on my schedule evokes mixed feelings: irritation that I’ll be an hour behind schedule the rest of the day; trepidation over the 50-50 chance she’ll need to be admitted, disrupting my busy day; and fear about intractable social problems.
Boston Irish, Mary has striking white hair and mischievous blue eyes; she’s feisty, funny, and utterly determined. Her neighborhood has seen generations of immigrants; its streets are lined with two-family homes and historic red brick churches, and there are pubs on every corner. Born and raised not far from her current apartment, she was stricken with rheumatic fever as a young girl. She remembers her youth as plagued by poor health, yet she
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2449
PERSPE C T I V E
The Road Less Traveled
graduated from high school, worked, married, and raised three children. But the early heart infection took a toll. By her mid-40s, Mary had an increasingly calcified mitral valve, an enlarging heart, and a damaged conduction system that caused atrial fibrillation and hypertension. Then came the stroke and a life forever changed. The paralysis of her right side was devastating enough, but she was also left unable to speak. Ultimately, with great difficulty, she began to utter a few simple words. Overwhelmed by her disabilities and dependent on others as never before, Mary saw her losses mount when her husband, unable to cope with her situation, left her. Soon thereafter, Mary’s creatinine level began to climb. No diabetes, no new meds, no obvious explanation for her failing kidneys. She saw a nephrologist and had innumerable blood and urine tests and finally a renal biopsy — the challenge of each experience heightened by the difficulty of communication and her loss of mobility. Eventually, she received a diagnosis of focal glomerular sclerosis. The kidney deterioration couldn’t be stopped, and she became overloaded with fluid and toxins. Mary had multiple hospital admissions for congestive heart failure and was facing the prospect of dialysis. An attempt at mitral valvuloplasty bought her a little time, but the admissions soon resumed. Then hemodialysis began. Consider the logistics of requiring a wheelchair, being unable to communicate over the phone, and needing to get to dialysis appointments three times a week. Add to the mix that Mary was unemployed, estranged from her children, took 16 medications 2450
daily, and also had to get to appointments for physical, occupational, and speech therapy — plus regular visits to the nephrologist, the cardiologist, and me. Somehow, she did it: her tenacity is astounding. Finally, Mary got a lucky break: a cadaveric kidney that was a close match. She had a rocky perioperative course but ultimately a good outcome. Mary’s joy at not having to return to dialysis was palpable; she strained to utter the words, “Thanks, thanks!” Mary’s pharmacy is one of few that still have the personal touch — Mary receives her pills in blister packs weekly by mail. There’s a real pharmacist I can speak to on the phone; he knows Mary well, knows all her meds, and is there every day, so whenever I adjust Mary’s Lasix dose, I can call him and get an extra blister pack sent to her the next day. It seems like a gift from above. Traveling by wheelchair down the sidewalk of her bustling neighborhood is part of Mary’s daily routine. Draped over the back of her wheelchair, she keeps a bag containing all her medications and a book with her doctors’ phone numbers and upcoming appointments; it makes her feel secure. But one day, a desperate stranger found it too tempting: a disabled woman, all alone, with that bag just dangling for the taking. In a moment, it was gone. Mary felt like her world was gone, too. “I understand she’s not due for her medications yet, but as I just explained to you, her medications were stolen.” I don’t know how many people I spoke to, but no one seemed to be able to fix the problem. Without her Lasix, Mary would probably be in heart failure in 48 hours, and her immunosuppressants and prednisone prevent
acute rejection of her transplanted kidney. Finding someone who understood the seriousness of the situation and could help seemed impossible. Even her pharmacist was willing to give her only a few days’ supply until Mary sorted things out with the insurance company — our government. But how could she “sort things out” when she couldn’t even speak? Somehow, miraculously, I persuaded someone to replace Mary’s pills. But for Mary, such miracles are rare. She arrives at my office for a follow-up visit, clearly distraught. She struggles through tears to choke out a few words (“darn, thanks, yes”) that don’t convey what she needs to say. She points to the phone, “You want me to call someone?” I ask. She nods emphatically. “Who? A relative? One of your specialists? Do you need a refill on one of your medications?” She lurches forward. I’m on the right track. “Which one?” I ask. She frowns. Off track. “OK, you don’t need a refill; what do you need?” Of course I know she can’t answer that. And if I’m feeling this frustrated, how does Mary feel every hour of every day? How does she go on? I finally figure out that she wants me to call her pharmacist. Without knowing why, I dial the number. “I was expecting your call,” he says. There’s a “problem” with Mary’s treatment. She’s been dropped from Medicaid and is no longer eligible for her medications. My heart sinks. Mary’s eyes are filled with fear — it’s as bad as she thinks. Again, the pharmacist offers a few days’ supply until Mary can straighten things out. I want to scream, “But she can’t speak!” I talk to social workers. Connected with every department in the vast Medicaid bureaucracy, I
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
PERSPECTIVE
The Road Less Traveled
repeat the story, getting variations on a standard response: “That shouldn’t have happened, but I’m not sure how to reverse the problem.” Luckily, the pharmacist is flexible about “a few days,” and after about a month, Mary is back on the Medicaid logs. It was a fluke, I’m told — a simple error that could send a life into a tailspin. Mary’s life is filled with such stories, but she musters awe- inspiring strength and determination for each challenge. At times, it does feel too hard to be Mary’s doctor. But it also has priceless rewards — which those medical
students on the ROAD don’t yet understand and may never get to experience. Every time Mary rolls into the emergency department, the overworked housestaff quickly tire of being unable to “get the story” from her. They just give her a squirt of Lasix or an antibiotic and head for the door. I understand their frustration. But what about Mary? I get the call and head over to the ED. As I pull back the curtain, a smile of recognition spreads over Mary’s face. She can relax now. She knows I care, that I’ll figure out her story and make sure the ED docs know all her
meds, allergies, and complications; I’ll let her specialists know she’s here. She knows that if her medicines change, I’ll contact her pharmacy to ensure she gets a new blister pack. She lies back and breathes more comfortably. And at times like this, I recognize my deep satisfaction with the road I’ve chosen to travel. Disclosure forms provided by the author are available with the full text of this article at NEJM.org. From Harvard Medical School and Massachusetts General Hospital — both in Boston. Copyright © 2011 Massachusetts Medical Society.
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2451
The
NEW ENGLA ND JOURNAL
of
MEDICINE
Perspective The Savings Illusion — Why Clinical Quality Improvement Fails to Deliver Bottom-Line Results Stephen S. Rauh, M.B.A., C.F.A., Eric B. Wadsworth, Ph.D., C.P.A., William B. Weeks, M.D., M.B.A., and James N. Weinstein, D.O.
I
t has become a core belief in U.S. health care that improving clinical quality will reduce health care costs. It seems intuitive that reducing readmissions, shortening lengths of stay, and building efficiency
into clinical processes will reduce resource utilization and thereby lower costs. Certainly, evidence suggests that there is no association between high quality and high costs.1 Yet true bottom-line savings from improved clinical quality rarely materialize, and costs continue to climb. Manufacturing and service companies around the world have demonstrated the cost benefits of improving product quality and production efficiency. So why haven’t nearly two decades of work on improving health care quality had a measurable effect on health care costs? The explanation lies in the cost
structure of the typical health care setting. Its management and organization create a rigid cost structure that is relatively insensitive to small changes in patient volume, resource use, or the severity of patients’ health conditions. This fixed-cost dilemma leaves most health care costs insensitive to changes in volume and utilization, so clinical quality improvements typically create additional capacity rather than bottom-line savings.2 An examination of the different cost layers highlights the distinction between variable costs, such as supplies and medications, where reduced use produces true savings, and
10.1056/nejmp1111662 nejm.org
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
fixed costs, such as facilities and ancillary services, where the costs persist despite reduced use. To better understand the cost structure of health care delivery, it can be useful to consider how different costs behave depending on the degree to which they are sensitive to changes in resource utilization. The four cost layers we have identified are defined in the table. Clinical improvements that reduce layer 1 costs, such as those of supplies and medications, will generally create bottom-line savings, since these are the only truly variable clinical costs in a hospital. To generate savings by reducing use of the resources that account for layer 2 costs, the need for the resource must be reduced enough to allow elimination of a payable unit. For instance, a single nursing unit might have to e48(1)
PERSPE C T I V E
The Savings Illusion
Behavior of the Various Cost Layers in the Health Care System. Cost Layer
Effects of Reduction in Use
Layer 1: truly variable costs of patient care
The item is not consumed, does not need to be replaced, and is available for later use.
Supplies, medications
Layer 2: semivariable costs of patient care
The item is not consumed, but the ability to repurpose the item is limited by time. Costs of providing the service may be reduced with sufficient reduction in volume.
Direct hourly nursing, respiratory therapists, physical therapists
Layer 3: semifixed costs of patient care
The item is not consumed, but the obligation to continue to pay for the item does not change.
Equipment, operating-room time, physician salaries, ancillary services
Layer 4: fixed costs not asso ciated with patient care
Resource consumption is not altered in the short run but may be altered in the next operating cycle.
Billing, organizational overhead, finance
discharge multiple patients before any savings in hourly nursing labor costs could be captured by allowing an hourly employee to go home early. Reducing layer 3 resources — those for equipment, operating-room time, or physicians’ salaries, for example — almost always produces additional capacity without bottom-line savings. If an intervention reduces operating-room time by 15 minutes, the costs of the equipment and salaried staff required to run the operating room do not change. Nonclinical layer 4 costs are primarily fixed in the short run, but reducing administrative labor costs by achieving administrative efficiency will produce true savings in future operating cycles. Because of these cost behaviors, quality-improvement efforts that reduce lengths of stay or readmissions or increase radiology throughput do not create substantive bottom-line savings. They generally create capacity to treat additional patients. Similarly, efforts to expand the access of disadvantaged populations to primary care under the assumption that such access will be paid for through avoiding use of highcost care sites — such as emergency departments — do not generate cost savings. The cost of e48(2)
Examples
staffing and equipping an emergency department does not change if there are small reductions in utilization. Indeed, improved access will increase health care costs if new physicians and staff are hired to serve new patients in primary care practices. Although capacity creation does not generate bottom-line savings, it does create an opportunity to admit another patient and collect additional revenue. Because health care costs are relatively fixed and do not change much at the margin, the cost of admitting a new patient is remarkably low, making volume growth a highly profitable strategy. Volume growth also can give the appearance of reducing costs, since the cost per case decreases when the high fixed costs are spread over a larger number of patients, although total costs will probably continue to rise. Growing volume and increasing revenue, rather than creating true bottom-line savings, are typically at the core of the business case for high-quality care.3 Because of the rigid cost structures, incremental reductions in resource use are unlikely to generate cost savings for either a health care setting or the health care system. The most meaningful way to achieve savings is to
focus on overall reductions in utilization rates for health care services and to eliminate the associated unnecessary capacity. In a recent article, Kaplan and Porter argue that most health care costs are not fixed.4 Postulating that personnel costs can be adjusted and space reallocated on the basis of demand and patient mix, they suggest that cost behaviors are not responsible for the inability to generate cost savings, but “management inattention” is. Although we do not dispute this logic, its practical application is dependent on both procedure volume and the time horizon required for aligning resources with demand. High-volume procedures and treatments for which resource use can be standardized across the cycle of care and for which capacity can be readily adjusted to accommodate appropriate volume appear to be best suited to the aggressive cost management advocated by Kaplan and Porter. Presumably, lower-volume treatments and procedures would have to be consolidated regionally to be more amenable to effective cost management. Until that happens, the fixed-cost dilemma will remain an obstacle. Cost layering provides management with a framework for targeting chang-
10.1056/nejmp1111662 nejm.org
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
PERSPECTIVE
The Savings Illusion
es that will generate the most immediate savings. Whereas quality improvement is producing significant benefits for patients, quality initiatives will continue to produce disappointing bottom-line savings as long as the capacity created is used to support growth in patient volume. As the U.S. health care system begins shifting its focus from volume to value, hospitals
will need to adapt their cost structures and capacity to accommodate lower per capita utilization rates as well as reductions in the per-episode intensity of care. Disclosure forms provided by the authors are available with the full text of this article at NEJM.org. From the Dartmouth Institute for Health Policy and Clinical Practice, Lebanon, NH. This article (10.1056/NEJMp1111662) was published on December 14, 2011, at NEJM.org.
10.1056/nejmp1111662 nejm.org
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
1. Yasaitis L, Fisher ES, Skinner JS, Chandra A. Hospital quality and intensity of spending: is there an association? Health Aff (Millwood) 2009;28:w566-w572. 2. Rauh SS, Wadsworth E, Weeks WB. The fixed cost dilemma: what counts when counting cost reduction efforts? Healthc Financ Manage 2010;64:60-3. 3. Gross PA, Ferguson JP, DeMauro P, et al. The business case for quality at a university teaching hospital. Jt Comm J Qual Patient Saf 2007;33:163-70. 4. Kaplan RS, Porter ME. How to solve the cost crisis in health care. Harv Bus Rev 2011; 89:46-52. Copyright © 2011 Massachusetts Medical Society.
e48(3)
new england journal of medicine The
established in 1812
december 29, 2011
vol. 365 no. 26
Liberal or Restrictive Transfusion in High-Risk Patients after Hip Surgery Jeffrey L. Carson, M.D., Michael L. Terrin, M.D., M.P.H., Helaine Noveck, M.P.H., David W. Sanders, M.D., Bernard R. Chaitman, M.D., George G. Rhoads, M.D., M.P.H., George Nemo, Ph.D., Karen Dragert, R.N., Lauren Beaupre, P.T., Ph.D., Kevin Hildebrand, M.D., William Macaulay, M.D., Courtland Lewis, M.D., Donald Richard Cook, B.M.Sc., M.D., Gwendolyn Dobbin, C.C.R.P., Khwaja J. Zakriya, M.D., Fred S. Apple, Ph.D., Rebecca A. Horney, B.A., and Jay Magaziner, Ph.D., M.S.Hyg., for the FOCUS Investigators*
A bs t r ac t Background
The hemoglobin threshold at which postoperative red-cell transfusion is warranted is controversial. We conducted a randomized trial to determine whether a higher threshold for blood transfusion would improve recovery in patients who had undergone surgery for hip fracture. Methods
We enrolled 2016 patients who were 50 years of age or older, who had either a history of or risk factors for cardiovascular disease, and whose hemoglobin level was below 10 g per deciliter after hip-fracture surgery. We randomly assigned patients to a liberal transfusion strategy (a hemoglobin threshold of 10 g per deciliter) or a restrictive transfusion strategy (symptoms of anemia or at physician discretion for a hemoglobin level of <8 g per deciliter). The primary outcome was death or an inability to walk across a room without human assistance on 60-day follow-up. Results
A median of 2 units of red cells were transfused in the liberal-strategy group and none in the restrictive-strategy group. The rates of the primary outcome were 35.2% in the liberal-strategy group and 34.7% in the restrictive-strategy group (odds ratio in the liberal-strategy group, 1.01; 95% confidence interval [CI], 0.84 to 1.22), for an absolute risk difference of 0.5 percentage points (95% CI, −3.7 to 4.7). The rates of in-hospital acute coronary syndrome or death were 4.3% and 5.2%, respectively (absolute risk difference, −0.9%; 99% CI, −3.3 to 1.6), and rates of death on 60-day follow-up were 7.6% and 6.6%, respectively (absolute risk difference, 1.0%; 99% CI, −1.9 to 4.0). The rates of other complications were similar in the two groups.
The authors’ affiliations are listed in the Appendix. Address reprint requests to Dr. Carson at the Department of Medicine, University of Medicine and Dentistry of New Jersey, Robert Wood Johnson Medical School, 125 Paterson St., New Brunswick, NJ 08903, or at
[email protected]. * Investigators in the Transfusion Trigger Trial for Functional Outcomes in Cardiovascular Patients Undergoing Surgical Hip Fracture Repair (FOCUS) are listed in the Supplementary Appendix, available at NEJM.org. This article (10.1056/NEJMoa1012452) was published on December 14, 2011, at NEJM.org. N Engl J Med 2011;365:2453-62. Copyright © 2011 Massachusetts Medical Society.
Conclusions
A liberal transfusion strategy, as compared with a restrictive strategy, did not reduce rates of death or inability to walk independently on 60-day follow-up or reduce in-hospital morbidity in elderly patients at high cardiovascular risk. (Funded by the National Heart, Lung, and Blood Institute; FOCUS ClinicalTrials.gov number, NCT00071032.) n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2453
The
n e w e ng l a n d j o u r na l
I
n the United States, more than 17 million red-cell units are collected annually, and 15 million units are transfused.1 Blood transfusions are frequently given to surgical patients and to the elderly.2,3 Yet, the indications for postoperative transfusion have not been adequately evaluated and remain controversial. Most clinical trials have been small.4 One adequately powered trial involving adults in intensive care units showed a nonsignificant decrease in 30-day mortality with a restrictive transfusion strategy, as compared with a liberal strategy (18.7% vs. 23.3%).5 However, the effect of a restrictive approach on functional recovery or risk of myocardial infarction in patients with cardiac disease has not been studied.4 We performed the Transfusion Trigger Trial for Functional Outcomes in Cardiovascular Patients Undergoing Surgical Hip Fracture Repair (FOCUS) to test the hypothesis that a higher threshold for blood transfusion (a hemoglobin level of 10 g per deciliter) would improve functional recovery and reduce morbidity and mortality, as compared with a more restrictive transfusion strategy (a hemoglobin level of <8 g per deciliter or symptoms).
Me thods Patients
From July 19, 2004, through February 28, 2009, we enrolled patients at 47 clinical sites in the United States and Canada. Telephone follow-up ended on May 4, 2009. Patients 50 years of age or older who were undergoing primary surgical repair of a hip fracture and who had clinical evidence of or risk factors for cardiovascular disease were eligible if they had a hemoglobin level of less than 10 g per deciliter within 3 days after surgery. According to the original protocol, only patients with cardiovascular disease (a history of ischemic heart disease, electrocardiographic evidence of previous myocardial infarction, a history or presence of congestive heart failure or peripheral vascular disease, or a history of stroke or transient ischemic attack) were eligible. In December 2005, eligibility criteria were expanded to enhance recruitment by including patients with any of the following cardiovascular risk factors: a history of or treatment for hypertension, diabetes mellitus, or hypercholesterolemia; a cholesterol level of 200 mg or more per deciliter or a lowdensity lipoprotein cholesterol level of 130 mg or
2454
of
m e dic i n e
more per deciliter; current tobacco use; or a creatinine level of more than 2.0 mg per deciliter.6 We excluded patients if they were unable to walk without human assistance before hip fracture, declined blood transfusions, had multiple trauma (defined as having had or planning to undergo surgery for non–hip-related traumatic injury), had a pathologic hip fracture associated with cancer, had a history of clinically recognized acute myocardial infarction within 30 days before randomization, had previously participated in the trial with a contralateral hip fracture, had symptoms associated with anemia (e.g., ischemic chest pain), or were actively bleeding at the time of potential randomization. The institutional review board or ethics committee at all 47 participating clinical sites approved the protocol (available with the full text of this article at NEJM.org). An independent data and safety monitoring board also approved the protocol and monitored the trial. Written informed consent was obtained from patients or their designated representatives. Methods were reported in detail previously.6 Treatment Assignment and Follow-up
We randomly assigned patients to the liberalstrategy group or the restrictive-strategy group using an automated telephone randomization system. Staff members at the data coordinating center prepared randomization schedules for each site using randomly ordered block sizes of two, four, six, or eight. After randomization, clinical-site staff members, clinicians, and patients were aware of study-group assignments. Patients in the liberal-strategy group received 1 unit of packed red cells and additional blood as needed to maintain a hemoglobin level of 10 g or more per deciliter. An assessment of the hemoglobin level after transfusion was required, and an additional unit of blood was transfused if the patient’s hemoglobin level was below 10 g per deciliter. Patients in the restrictive-strategy group were permitted to receive transfusions if symptoms or signs of anemia developed or at the discretion of their physicians if the hemoglobin level fell below 8 g per deciliter. Symptoms or signs that were considered indications for transfusion were chest pain that was deemed to be cardiac in origin, congestive heart failure, and unexplained tachy-
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Liber al or Restrictive Tr ansfusion in High-Risk Patients
cardia or hypotension unresponsive to fluid replacement. Blood was administered 1 unit at a time, and the presence of symptoms or signs was reassessed. Patients with clinically diagnosed dementia received transfusions when the hemoglobin level fell below 8 g per deciliter because they might not be able to report their symptoms. Hemoglobin levels were measured during hospitalization on days 1, 2, 4, and 7 after randomization. Additional hemoglobin determinations were made as clinically indicated. The assigned transfusion strategy was to be followed until discharge or up to 30 days, whichever came first. Transfusion was permitted at any time without measuring a hemoglobin level if the patient was bleeding and emergency transfusion was considered necessary by the treating physician. Nurses at the clinical coordinating center who were not involved with study implementation and were unaware of study-group assignments telephoned patients or proxies at or close to 30 days and 60 days after randomization to ascertain outcomes after hospital discharge. They spoke directly to patients who were accessible by telephone or to proxies if patients were cognitively impaired or could not talk on the telephone.
markers that were performed in hospitals for clin ical indications were also collected. Samples were analyzed at the core laboratory of the Minneapolis Medical Research Foundation of Hennepin County Medical Center for troponin I (Access 2 Immunoassay System, Beckman Coulter) with the use of a threshold of 0.06 μg per liter (1.5 times the 99th percentile [0.04 μg per liter] for healthy patients). We used the Universal Definition of Myocardial Infarction criteria7,8 to define myocardial infarction and unstable angina on the basis of review of clinical status, central interpretation of electrocardiograms at Saint Louis University, and results of core laboratory and clinical cardiac biomarkers (see the Supplementary Appendix, available at NEJM.org). Study investigators who classified cardiovascular outcomes and those who did follow-up telephone assessments were unaware of study-group assignments. Other secondary outcomes that were determined on telephone follow-up at or close to 30 days and 60 days after randomization included current residence, survival, functional measures (lowerextremity physical and instrumental activities of daily living), and fatigue. These outcomes were ascertained with the use of methods described previously.6
Primary Outcome
The primary outcome was death or an inability to walk 10 ft (or across a room) without human assistance at the 60-day follow-up. We hypothesized that an increased hemoglobin level would allow patients to participate more actively in rehabilitation and therefore increase the proportion who were walking independently 60 days after randomization.
Tertiary Study Outcomes
We evaluated in-hospital morbidity up to 30 days after randomization, including pneumonia, wound infection, thromboembolism, stroke or transient ischemic attack, and clinically recognized myocardial infarction.6 We prespecified two composite outcomes: death, myocardial infarction, or pneumonia; and death, myocardial infarction, pneumonia, thromboembolism, or stroke.
Secondary Outcomes
Secondary outcomes included a combined outcome of in-hospital myocardial infarction, unstable angina, or death for any reason; each of these outcomes was assessed individually. Electrocardiography was performed before surgery, before randomization, and on day 4 after randomization (or at the time of discharge if before day 4). Blood (plasma or serum) specimens were collected for measurement of the cardiac troponin I level before surgery, before randomization, and on days 1 and 4 after randomization or before discharge (if before day 4). Electrocardiograms and results of testing of cardiac bio-
Vital Status and Walking Confirmation
We validated the vital status of patients in the United States by searching the online Social Security Database. When discrepancies were identified between telephone reports and this database, we verified deaths using hospital records or published obituaries. We validated the vital status of Canadian patients by searching hospital medical records, vital-status records, and outpatient medical records. We validated vital status in 95.9% of patients (99.0% in the United States and 91.2% in Canada). Of 1934 vital-status confirmations, we found 7 discrepancies (0.4%) between telephone re-
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2455
The
n e w e ng l a n d j o u r na l
ports and vital-status records; in these cases, we used vital-status records. We assessed the reliability of the self-report of walking status in a subgroup of 814 patients for whom we had both self-report and proxy report and found high reliability (kappa = 0.90) between these reports.9 Adherence Definitions
We defined major protocol violations as a lack of receipt of a transfusion or hospital discharge with a hemoglobin level of less than 10 g per deciliter in the liberal-strategy group and as the receipt of transfusion with a hemoglobin level of 8 g per deciliter or more in the absence of symptoms in the restrictive-strategy group. Statistical Analysis
According to the original study design, we determined that a sample size of 2600 patients would provide a power of 90% and an experiment-wise alpha level of 0.05 allowing for interim analyses (four were performed by the data and safety monitoring board) and a level of 0.048 for the final comparison to detect an absolute between-group difference of 7 percentage points in the primary outcome (odds ratio, 0.75). In September 2007, the data and safety monitoring board approved a reduction of recruitment goal to 2000 patients. This change resulted in an absolute change of approximately 1 percentage point in the between-group difference in the primary outcome that could be excluded with a power of 90%. We used the Mantel–Haenszel method10 to conduct the primary analysis, taking into account different clinical sites. We prespecified tests for interaction of the primary outcome11,12 with sex, age, race, and cardiovascular-disease status (known cardiovascular disease vs. risk factors only) without adjustment of the alpha level. Tests for interaction and differences in outcomes are presented without adjustment for clinical site. The primary outcome analysis is presented as a Mantel–Haenszel odds ratio with 95% confidence intervals. For secondary and tertiary analyses, we used standard methods for the comparison of proportions and means without adjustment for clinical site, using an alpha level of 0.01 (with 99% confidence intervals). Analyses were performed with the use of SAS software, version 9.2.
2456
of
m e dic i n e
R e sult s Study Population
We screened 14,438 patients and randomly assigned 2016 to either the liberal-strategy group (1007 patients) or the restrictive-strategy group (1009) (see the Supplementary Appendix). There were 14 withdrawals, 2 losses to follow-up, and 1 incomplete follow-up ascertainment; follow-up for the primary analysis was obtained in 99.2% of the patients. Of the 1999 patients included in the primary analysis, we directly interviewed 1075 (53.8%) and obtained data on 923 (46.2%) by proxy; the source of information was missing for 1 patient. The mean age of the study population was 81.6 years (range, 51 to 103), and cardiovascular disease was present in 62.9%. Baseline characteristics were similar in the two study groups (Table 1). Hemoglobin Levels and Transfusion
The average hemoglobin level before transfusion was 1.3 g per deciliter higher in the liberal-strategy group than in the restrictive-strategy group (P<0.001) (Table 2). The median number of units transfused was 2.0 (interquartile range, 1 to 2) in the liberal-strategy group and 0 (interquartile range, 0 to 1) in the restrictive-strategy group; 59.0% of patients in the restrictive-strategy group did not receive a transfusion after randomization. Figure 1 shows the average daily lowest hemoglobin levels in the two groups. Violations in the transfusion protocol occurred in 9.0% of patients in the liberal-strategy group and in 5.6% of those in the restrictive-strategy group. Symptoms leading to transfusion are listed in Table 2. Outcomes
The rates of death or an inability to walk without human assistance at 60-day follow-up were similar in the liberal-strategy group and the restrictive-strategy group (35.2% vs. 34.7%, P = 0.90) (Table 3). The odds ratio for the primary outcome associated with the liberal strategy versus the restrictive strategy was 1.01 (95% confidence interval [CI], 0.84 to 1.22), for an absolute risk difference of 0.5 percentage points (95% CI, −3.7 to 4.7). There was a significant interaction according to patients’ sex (P = 0.03), with an odds ratio associ-
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Liber al or Restrictive Tr ansfusion in High-Risk Patients
Table 1. Baseline Characteristics of the Patients.* Liberal Strategy (N = 1007)
Variable Age — yr
Restrictive Strategy (N = 1009)
81.8±8.8
81.5±9.0
250 (24.8)
239 (23.7)
White
944 (93.7)
947 (93.9)
Black
40 (4.0)
42 (4.2)
Asian
14 (1.4)
13 (1.3)
Other
9 (0.9)
7 (0.7)
609 (60.5)
613 (60.8)
Male sex — no. (%) Race — no. (%)†
Residence in the United States — no. (%) Cardiovascular disease — no. (%) Any
637 (63.3)
631 (62.5)
Coronary artery disease
402 (39.9)
403 (39.9)
Congestive heart failure
184 (18.3)
167 (16.6)
Cerebrovascular disease
249 (24.7)
224 (22.2)
Peripheral vascular disease
117 (11.6)
102 (10.1)
Hypertension
824/1003 (82.2)
821/1005 (81.7)
Diabetes mellitus
252/1003 (25.1)
256/1005 (25.5)
Hypercholesterolemia
347/1002 (34.6)
360/1001 (36.0)
Tobacco use
116/1003 (11.6)
113/1004 (11.3)
83/1001 (8.3)
86/1003 (8.6)
Cardiovascular risk factors — no./total no. (%)
Creatinine >2.0 mg/dl Chronic lung disease
189/1003 (18.8)
188/1007 (18.7)
History of dementia or confusion
309/1004 (30.8)
325/1008 (32.2)
History of cancer
181/1003 (18.0)
189/1008 (18.8)
Femoral neck
432/1004 (43.0)
422/1008 (41.9)
Intertrochanteric
512/1004 (51.0)
522/1008 (51.8)
Subtrochanteric
88/1004 (8.8)
95/1008 (9.4)
Reverse oblique
13/1004 (1.3)
8/1008 (0.8)
General
543/1005 (54.0)
566/1008 (56.2)
Spinal
457/1005 (45.5)
434/1008 (43.1)
Other
5/1005 (0.5)
Type of hip fracture — no./total no. (%)
Type of anesthesia — no./total no. (%)
American Society of Anesthesiology risk score‡
3.0±0.6
8/1008 (0.8) 2.9±0.6
Residence — no./total no. (%) Home or retirement home
892/1005 (88.8)
886/1008 (87.9)
Nursing home
104/1005 (10.3)
110/1008 (10.9)
9/1005 (0.9)
12/1008 (1.2)
Other
* Plus–minus values are means ±SD. There were no significant between-group differences for any of the listed variables. † Race was self-reported. ‡ Scores range from 1 to 5, with a higher score indicating greater risk. Data in this category were missing for 38 patients in the liberal-strategy group and 39 in the restrictive-strategy group.
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2457
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. Hemoglobin Levels and Transfusions.* Variable
Liberal Strategy (N = 1007)
Restrictive Strategy (N = 1009)
11.3±1.5
11.3±1.5
0.70
9.0±0.8
9.0±0.8
0.98
9.2±0.5
7.9±0.6
<0.001
209±179
232±257
0.03
P Value
Hemoglobin level — g/dl Before surgery During eligibility screening Before transfusion Estimated blood loss during surgery — ml† Transfusions before randomization 0 units — no./total no. (%)
754/1006 (75.0)
720/1008 (71.4)
≥1 unit — no./total no. (%)
252/1006 (25.0)
288/1008 (28.6)
452
531
0 units — no./total no. (%)
33/1003 (3.3)
594/1007 (59.0)
1 unit — no./total no. (%)
420/1003 (41.9)
246/1007 (24.4)
2 units — no./total no. (%)
346/1003 (34.5)
127/1007 (12.6)
3 units — no./total no. (%)
132/1003 (13.2)
24/1007 (2.4)
≥4 units — no./total no. (%)
72/1003 (7.2)
16/1007 (1.6)
1866
652
Storage of units transfused after randomization — days‡
22.0±9.5
22.1±9.9
0.83
Leukoreduced units transfused after randomization — %§
90.2
88.6
0.25
91/1006 (9.0)
56/1007 (5.6)
0.003
Rapid bleeding
5/1006 (0.5)
14/1007 (1.4)
0.04
Chest pain
4/1006 (0.4)
9/1007 (0.9)
0.17
Total no. of units
0.07
Transfusions after randomization
Total no. of units
Major protocol violation — no./total no. (%)¶
<0.001
Transfusion because of symptoms — no./total no. (%)‖
Congestive heart failure Tachycardia or hypotension
1/1006 (0.1)
10/1007 (1.0)
0.007
43/1006 (4.3)
123/1007 (12.2)
<0.001
* Plus–minus values are means ±SD. † Data on estimated blood loss were missing for 122 patients in the liberal-strategy group and 129 in the restrictive-strategy group. ‡ Data on the length of storage of units were missing for 25 units in the liberal-strategy group and 8 in the restrictivestrategy group. § Data on leukoreduction status were missing for 19 units in the liberal-strategy group and 10 in the restrictive-strategy group. ¶ In the liberal-strategy group, there were two types of protocol violations: 30 patients (3.0%) did not receive a transfusion, and 61 patients (6.1%) were discharged with a hemoglobin level of less than 10 g per deciliter. In the restrictivestrategy group, there was only one type of violation: 56 patients (5.6%) who did not have symptoms or rapid bleeding received transfusions for a hemoglobin level of 8.0 g per deciliter or more. ‖ Patients may have had more than one symptom.
ated with the liberal strategy of 1.45 (95% CI, 1.00 to 2.10) for men versus 0.91 (95% CI, 0.74 to 1.13) for women. Interactions according to age, race, and cardiovascular-disease status were not significant (see the Supplementary Appendix). There were no significant between-group differences in the rates of death on 30-day follow-up (5.2% in the liberal-strategy group vs. 4.3% in the restrictive-strategy group), for an absolute risk dif2458
ference of 0.9 percentage points (99% CI, −1.5 to 3.4), and on 60-day follow-up (7.6% in the liberalstrategy group vs. 6.6% in the restrictive-strategy group), for an absolute risk difference of 1.0 percentage point (99% CI, −1.9 to 4.0) (Table 3). The between-group differences were also not significant in the rates of in-hospital acute myocardial infarction, unstable angina, or death (4.3% in the liberal-strategy group vs. 5.2% in the restrictive-
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Liber al or Restrictive Tr ansfusion in High-Risk Patients
Discussion We performed a randomized clinical trial involving 2016 patients undergoing surgery for hip fracture and found no evidence that maintaining the hemoglobin level above 10 g per deciliter was superior to transfusion for symptoms or maintaining a hemoglobin level of less than 8 g per deciliter with respect to the primary outcome (a composite of death or an inability to walk across the room without human assistance) and to several clinically relevant secondary outcomes, including cardiovascular event rates and other functional measures. We enrolled a high-risk group of patients with a mean age of more than 81 years for whom untreated anemia would probably be more harmful than in a healthier or younger population undergoing most surgical procedures. An ability to walk across the room at 60 days was selected as a main component of the primary outcome because such a measure is recognized to be an important functional outcome after hip fracture and is likely to be affected by factors that transfusion might influence (e.g., aerobic capacity and muscle strength). We hypothesized, in particular, that a higher hemoglobin level might facilitate more active participation in rehabilitation, leading to more successful recovery of ambulation. We achieved a clinically important difference in the use of packed red cells and a good separation in hemoglobin levels in the two transfusion groups (Fig. 1). Patients in the restrictive-strategy group received 65% fewer units of blood than those in the liberal-strategy group; more than half the patients in the restrictive-strategy group did not receive any blood transfusion. Widespread implementation of this restrictive approach to transfusion in similar patients would greatly reduce blood use. We found an interaction between the transfusion strategy and sex in the liberal-strategy group,
14
Liberal strategy
Restrictive strategy
Both strategies
13
Lowest Daily Hemoglobin (g/dl)
strategy group), for an absolute risk difference of −0.9 percentage points (99% CI, −3.3 to 1.6). The frequencies of in-hospital clinical events and serious adverse events did not differ significantly between groups (Table 4). Also similar in the two groups were the length of hospital stay, scores for lower-extremity physical activities of daily living, instrumental activities of daily living, and fatigue, as well as rates of residing at home at 30-day and 60-day follow-up (Table 3).
12 11 10 9 8 7 6 0
0
1
2
3
4
5
6
7
Days since Randomization
Figure 1. Lowest Daily Hemoglobin Levels. Shown are the lowest daily hemoglobin levels among patients in the liberalstrategy group versus those in the restrictive-strategy group. Data for the two groups are pooled on the day of randomization and are presented for days 1, 2, 4, and 7, when hemoglobin levels were required to be measured while patients remained in the hospital. The center line within each box represents the median, and the extremes the interquartile range.
suggesting a higher rate of death or an inability to walk without human assistance at 60-day follow-up in men but not in women. This difference was not anticipated and could have been due to chance. We obtained primary-outcome information (including data regarding deaths) for more than 99% of patients and validated vital status. However, we did not perform follow-up examinations, and our telephone ascertainment of functional outcomes was subject to possible miscommunication, poorly informed proxy respondents, and recording errors. Although we did not validate patients’ ability to walk, in cases in which both patients and their proxies answered the question about walking ability, we found strong agreement between the two reports. Scores for physical activities of daily living, instrumental activities of daily living, and fatigue were not validated and were not useful for analysis for 45 to 60% of patients. We revised eligibility criteria in the course of the trial to include lower-risk patients who had cardiovascular risk factors but no history of cardiovascular disease, and there was no important treatment interaction with cardiovasculardisease status. Our study had excellent statistical power for determining the primary outcome of death or inability to walk. On the basis of the 95% confi-
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2459
2460 407/995 (40.9) 52/995 (5.2)
Inability to walk independently
Death
135/994 (13.6) 402/994 (40.4)
Nursing home
Other
38.7±7.7
FACIT-Fatigue scale¶
score
38.6±7.6
3.9±0.4
7.4±3.9
413/999 (41.3)
161/999 (16.1)
425/999 (42.5)
43/1000 (4.3)
438/1000 (43.8)
481/1000 (48.1)
0.9 (−1.5 to 3.4)
−2.0 (−7.7 to 3.8)
0.84
0.10
0.72
0.17
P Value
1.23 (0.71 to 2.12)
0.92 (0.73 to 1.16)
percentage points
Absolute Risk Difference (99% CI) Restrictive Strategy (N = 1009)
41.8±7.3
3.7±0.8
5.1±4.2
242/996 (24.3)
137/996 (13.8)
617/996 (61.9)
76/998 (7.6)
275/998 (27.6)
351/998 (35.2)
score
42.3±7.4
3.7±0.9
5.1±4.3
237/1001 (23.7)
161/1001 (16.1)
603/1001 (60.2)
66/1001 (6.6)
281/1001 (28.1)
347/1001 (34.7)
no./total no. (%)
Liberal Strategy (N = 1007)
1.0 (−1.9 to 4.0)†
0.5 (−3.7 to 4.7)
percentage points
Absolute Risk Difference (95% CI)
0.26
0.94
0.85
0.34
P Value
1.17 (0.75 to 1.83)†
1.01 (0.84 to 1.22)
Odds Ratio (95% CI)
60-Day Period
n e w e ng l a n d j o u r na l
* Plus–minus values are means ±SD. Odds ratios and risk differences are for the comparison between the liberal-strategy group and the restrictive-strategy group. † Values are 99% confidence intervals. ‡ Scores on the lower-extremity physical activities of daily living (ADL) scale range from 0 to 11, with higher scores indicating greater dependency. Scores were calculated by totaling the number of dependencies with respect to 11 basic activities. Patients who reported that they had any human assistance in an activity or that they did not perform the activity for a health reason were considered to be dependent with respect to that activity. Patients who had missing data or who did not perform the activity for reasons other than those related to health were excluded from the analysis. Scores were not used in this analysis on 30-day follow-up for 535 patients in the liberal-strategy group and 502 in the restrictive-strategy group and on 60-day follow-up for 484 in the liberal-strategy group and 456 in the restrictive-strategy group. § Scores on the instrumental ADL scale range from 0 to 4, with higher scores indicating greater dependency. Scores were calculated by totaling the number of dependencies with respect to four advanced activities. Patients who reported that they needed assistance or were unable to perform a task for health reasons were considered to be dependent with respect to that activity. Patients who had missing data or did not perform the activity for reasons other than those related to health were excluded from the analysis. Scores were not used in this analysis on 30-day follow-up for 570 patients in the liberal-strategy group and 559 in the restrictive-strategy group and on 60-day follow-up for 618 in the liberal-strategy group and 598 in the restrictive-strategy group. ¶ The Functional Assessment of Chronic Illness Therapy–Fatigue (FACIT-Fatigue) scale includes 13 items with scores ranging from 0 to 4 (0, not at all; 1, a little bit; 2, somewhat; 3, quite a bit; and 4, very much), with higher scores indicating a greater energy level. Missing items were imputed as the mean of item scores within the same scale. No proxy responses were possible. Scores were missing on 30-day follow-up for 551 patients in the liberal-strategy group and 550 in the restrictive-strategy group and on 60-day follow-up for 463 in the liberalstrategy group and 484 in the restrictive-strategy group.
7.3±4.0 3.9±0.5
Lower-extremity physical ADL‡
Instrumental ADL§
Function and symptom scales
457/994 (46.0)
Home or retirement home
Residence
459/995 (46.1)
Odds Ratio (99% CI)
30-Day Period Restrictive Strategy (N = 1009)
no./total no.(%)
Liberal Strategy (N = 1007)
Death or inability to walk independently
Variable
Table 3. Outcomes at 30 Days and 60 Days.*
The of
m e dic i n e
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Liber al or Restrictive Tr ansfusion in High-Risk Patients
Table 4. Hospital Outcomes.* Liberal Strategy (N = 1007)
Variable
Restrictive Strategy (N = 1009)
Absolute Risk Difference (99% CI)
Odds Ratio (99% CI)
number/total number (percent) Myocardial infarction, unstable angina, or in-hospital death†
percentage points
43/1005 (4.3)
52/1008 (5.2)
0.82 (0.48 to 1.42)
−0.9 (−3.3 to 1.6)
23/1005 (2.3)
38/1008 (3.8)
0.60 (0.30 to 1.19)
−1.5 (−3.5 to 0.5)
Unstable angina†
2/1005 (0.2)
3/1008 (0.3)
0.67 (0.06 to 7.03)
−0.1 (−0.7 to 0.5)
In-hospital death
Myocardial infarction†
20/1005 (2.0)
14/1008 (1.4)
1.44 (0.58 to 3.56)
0.6 (−0.9 to 2.1)
Isolated troponin elevation‡
62/1005 (6.2)
59/1008 (5.9)
1.06 (0.65 to 1.71)
0.3 (−2.4 to 3.1)
Physician diagnosis of congestive heart failure
27/1005 (2.7)
35/1007 (3.5)
0.77 (0.39 to 1.50)
−0.8 (−2.8 to 1.2)
Stroke or transient ischemic attack On CT or MRI
5/1005 (0.5)
1/1007 (0.1)
5.03 (0.30 to 84.73)
0.4 (−0.2 to 1.0)
On physician diagnosis or CT or MRI
8/1005 (0.8)
3/1007 (0.3)
2.69 (0.47 to 15.42)
0.5 (−0.3 to 1.3)
60/1005 (6.0)
48/1007 (4.8)
1.27 (0.76 to 2.12)
1.2 (−1.4 to 3.8)
9/1005 (0.9)
3/1007 (0.3)
3.02 (0.54 to 16.91)
0.6 (−0.3 to 1.5)
Wound infection
14/1005 (1.4)
8/1007 (0.8)
1.76 (0.56 to 5.56)
0.6 (−0.6 to 1.8)
Deep-vein thrombosis or pulmonary embolism
12/1005 (1.2)
8/1007 (0.8)
1.51 (0.46 to 4.92)
0.4 (−0.7 to 1.5)
Death, myocardial infarction, pneumonia
89/1005 (8.9)
90/1007 (8.9)
0.99 (0.66, 1.48)
103/1005 (10.2)
94/1007 (9.3)
1.11 (0.75 to 1.63)
0.9 (−2.5 to 4.3)
Chest radiograph with new or progressive infiltrate New-onset purulent sputum
Death, myocardial infarction, pneumonia, thromboembolism, or stroke
−0.1 (−3.4 to 3.2)
Returned to operating room
15/1005 (1.5)
18/1007 (1.8)
0.83 (0.34 to 2.06)
−0.3 (−1.8 to 1.2)
Transfer to intensive care unit
30/1005 (3.0)
29/1007 (2.9)
1.04 (0.53 to 2.05)
0.1 (−1.8 to 2.0)
days
P Value
Time from randomization to discharge§ United States
3.67±3.38
3.97±3.89
0.15
Canada
12.03±9.31
12.70±9.48
0.32
* Plus–minus values are means ±SD. Odds ratios and risk differences are for the comparison between the liberal-strategy group and the restrictive-strategy group. CT denotes computed tomography, and MRI magnetic resonance imaging. † Electrocardiographic results after randomization were incomplete for 135 patients in the liberal-strategy group and 130 in the restrictivestrategy group. ‡ Blood samples obtained for troponin testing on day 4 after randomization or at the time of hospital discharge were not available for 180 patients in the liberal-strategy group and 175 in the restrictive-strategy group. § Of the 2011 patients who were evaluated (1220 in the United States and 791 in Canada), 944 patients (93.9%) in the liberal-strategy group and 934 (92.8%) in the restrictive-strategy group were discharged alive.
dence interval, the restrictive transfusion policy plausibly resulted in at most a 3.7% increase in the risk of death or inability to walk without human assistance, a composite outcome that occurred in about 35% of patients. We had less statistical power for in-hospital outcomes; our data are compatible with an absolute change in the composite outcome of in-hospital acute myocardial infarction, unstable angina, or death, ranging from an increase of 3.3 percentage points to a decrease of 1.6 percentage points for the restrictive transfusion strategy.
Our results are consistent with most of the findings of the Transfusion Requirements in Critical Care (TRICC) trial, in which outcomes did not differ significantly between a transfusion threshold of 7 g per deciliter and a threshold of 10 g per deciliter among patients in intensive care units.5,13 However, in contrast to that report, we did not find increased rates of myocardial infarction or congestive heart failure in the liberal-strategy group. Furthermore, we did not confirm findings from observational studies of markedly higher mortality in patients who received transfusion
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2461
Liber al or Restrictive Tr ansfusion in High-Risk Patients
than in patients who did not.14 A randomized clinical trial allows us to evaluate transfusion while avoiding selection bias.15 In summary, we found that a liberal transfusion strategy, as compared with a restrictive strategy, did not result in reduced rates of death or an inability to walk on 60-day follow-up or in significant reductions in rates of in-hospital complications in this population at increased cardiovascular risk. Our findings suggest that it is reasonable to withhold transfusion in patients who have undergone surgery in the absence of symptoms of anemia or a decline in the hemoglobin level below 8 g per deciliter, even in elderly patients with underlying cardiovascular disease or risk factors. Supported in part by grants from the National Heart, Lung, and Blood Institute (U01 HL073958 and U01 HL074815).
Dr. Carson reports receiving grant support to his institution from Amgen; Dr. Lewis, receiving a salary from the Orthopaedic Associates of Hartford, receiving a stipend for serving as president of the Hartford County Medical Association, and providing expert testimony representing the American Academy of Orthopaedic Surgery on the Medicare Evidence Development and Coverage Advisory Committee; Dr. Apple, serving as a scientific advisory board member for Abbott Laboratories, Alere, Beckman Coulter, Ortho Clinical Diagnostics, and Instrumentation Laboratories, receiving consulting fees from Abbott Diagnostics, Ortho Clinical Diagnostics, and Instrumentation Labora tories, receiving grant support to his institution from Abbott Diagnostics, Siemens, Ortho Clinical Diagnostics, Roche Diagnostics, BioRad, Response Biomedical, Radiometer, and BRAHMS, and receiving lecture fees and travel expenses from Abbott Diagnostics and Alere; Dr. Magazine, serving as a board member for Amgen, Novartis, and GlaxoSmithKline and receiving consulting fees from Eli Lily, Sanofi-Aventis, and Amgen, grant support to his institution from Novartis, Merck, and Eli Lilly, and lecture fees from Novartis. No other 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.
Appendix The authors’ affiliations are as follows: the Division of General Internal Medicine, Department of Medicine, University of Medicine and Dentistry of New Jersey, Robert Wood Johnson Medical School, New Brunswick (J.L.C., H.N., K.D.); the Department of Epidemiology and Public Health, University of Maryland School of Medicine (M.L.T., J.M.); and Johns Hopkins Bayview Medical Center (K.J.Z.) — both in Baltimore; the Division of Orthopaedic Surgery, University of Western Ontario, London (D.W.S.); the Department of Physical Therapy and Surgery and the Division of Orthopaedic Surgery (L.B.), University of Alberta, Edmonton; the Division of Orthopedic Surgery (K.H.) and Department of Medicine (D.R.C.), University of Calgary, Calgary, AB; and the Department of Orthopedic Surgery, QEII Health Sciences Centre, Halifax, NS (G.D.) — all in Canada; the Department of Medicine, Saint Louis University School of Medicine, St. Louis (B.R.C.); the Department of Epidemiology, University of Medicine and Dentistry of New Jersey–School of Public Health, Piscataway (G.G.R.); the Transfusion Medicine and Cellular Therapeutics Branch, Division of Blood Diseases and Resources, National Heart, Lung, and Blood Institute, Bethesda (G.N.); and the Cooperative Studies Program Coordinating Center, Veterans Affairs Medical Center, Perry Point (R.A.H.) — both in Maryland; the Department of Orthopedic Surgery, New York–Presbyterian Hospital at Columbia University, New York (W.M.); Hartford Hospital, Hartford, CT (C.L.); and Minneapolis Medical Research Foundation of Hennepin County Medical Center and University of Minnesota School of Medicine, Minneapolis (F.S.A.). References 1. Report of the Department of Health
and Human Services: the 2009 national blood collection and utilization survey report. Washington, DC: Department of Health and Human Services, Office of the Assistant Secretary for Health, 2011. 2. Anderson SA, Menis M, O’Connell K, Burwen DR. Blood use by inpatient elderly population in the United States. Transfusion 2007;47:582-92. 3. Cobain TJ, Vamvakas EC, Wells A, Titlestad K. A survey of the demographics of blood use. Transfus Med 2007;17:1-15. 4. Carless PA, Henry DA, Carson JL, Hebert PP, McClelland B, Ker K. Transfusion thresholds and other strategies for guiding allogeneic red blood cell transfusion. Cochrane Database Syst Rev 2010;10: CD002042. 5. Hébert PC, Wells G, Blajchman MA, et al. A multicenter, randomized, controlled clinical trial of transfusion requirements in critical care. N Engl J Med 1999;340:40917. [Erratum, N Engl J Med 1999;340:1056.]
2462
6. Carson JL, Terrin ML, Magaziner J, et
al. Transfusion Trigger Trial for Functional Outcomes in Cardiovascular Patients Undergoing Surgical Hip Fracture Repair (FOCUS). Transfusion 2006;46:2192-206. 7. Alpert JS, Thygesen K, Antman E, Bassand JP. Myocardial infarction re defined — a consensus document of The Joint European Society of Cardiology/ American College of Cardiology Committee for the redefinition of myocardial infarction. J Am Coll Cardiol 2000;36: 959-69. [Erratum, J Am Coll Cardiol 2001; 37:973.] 8. Thygesen K, Alpert JS, White HD. Universal definition of myocardial infarction. J Am Coll Cardiol 2007;50:2173-95. 9. Fleiss JL. Statistical methods for rates and proportions. New York: John Wiley, 1981. 10. Mantel N, Haenszel W. Statistical aspects of the analysis of data for retrospective studies of disease. J Natl Cancer Inst 1959;22:719-48.
11. Breslow NE, Day NE. The analysis of case-control studies. Vol. 1. Lyon, France: International Agency for Research on Cancer, 1980. (IARC scientific publications no. 32.) 12. Hosmer DW, Lemeshow S. Applied logistic regression. New York: John Wiley, 1989. 13. Hébert PC, Yetisir E, Martin C, et al. Is a low transfusion threshold safe in critically ill patients with cardiovascular diseases? Crit Care Med 2001;29:227-34. 14. Marik PE, Corwin HL. Efficacy of red blood cell transfusion in the critically ill: a systematic review of the literature. Crit Care Med 2008;36:2667-74. [Erratum, Crit Care Med 2008;36:3134.] 15. MacMahon S, Collins R. Reliable assessment of the effects of treatment on mortality and major morbidity, II: observational studies. Lancet 2001;357:45562. Copyright © 2011 Massachusetts Medical Society.
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
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
Low-Molecular-Weight Heparin and Mortality in Acutely Ill Medical Patients Ajay K. Kakkar, M.B., B.S., Ph.D., Claudio Cimminiello, M.D., Samuel Z. Goldhaber, M.D., Rajiv Parakh, M.D., Chen Wang, M.D., Ph.D., and Jean-François Bergmann, M.D., for the LIFENOX Investigators*
A BS T R AC T BACKGROUND
Although thromboprophylaxis reduces the incidence of venous thromboembolism in acutely ill medical patients, an associated reduction in the rate of death from any cause has not been shown. METHODS
We conducted a double-blind, placebo-controlled, randomized trial to assess the effect of subcutaneous enoxaparin (40 mg daily) as compared with placebo — both administered for 10±4 days in patients who were wearing elastic stockings with graduated compression — on the rate of death from any cause among hospitalized, acutely ill medical patients at participating sites in China, India, Korea, Malaysia, Mexico, the Philippines, and Tunisia. Inclusion criteria were an age of at least 40 years and hospitalization for acute decompensated heart failure, severe systemic infection with at least one risk factor for venous thromboembolism, or active cancer. The primary efficacy outcome was the rate of death from any cause at 30 days after randomization. The primary safety outcome was the rate of major bleeding during and up to 48 hours after the treatment period. RESULTS
A total of 8307 patients were randomly assigned to receive enoxaparin plus elastic stockings with graduated compression (4171 patients) or placebo plus elastic stockings with graduated compression (4136 patients) and were included in the intentionto-treat population. The rate of death from any cause at day 30 was 4.9% in the enoxaparin group as compared with 4.8% in the placebo group (risk ratio, 1.0; 95% confidence interval [CI], 0.8 to 1.2; P = 0.83). The rate of major bleeding was 0.4% in the enoxaparin group and 0.3% in the placebo group (risk ratio, 1.4; 95% CI, 0.7 to 3.1; P = 0.35).
From the Thrombosis Research Institute and University College London, London (A.K.K.); the Medical Department, Ospe dale Civile di Vimercate, Vimercate, Italy (C.C.); Brigham and Women’s Hospital and Harvard Medical School, Boston (S.Z.G.); Medanta–Medicity, Gurgaon, Haryana, India (R.P.); the Beijing Institute of Respiratory Medicine, Chaoyang District, and Beijing Hospital, Ministry of Health, Dongcheng District, Beijing (C.W.); and Hôpital Lariboisière, Assistance Publique–Hôpitaux de Paris, University of Paris Diderot, Paris (J.-F.B.). Address reprint requests to Dr. Kakkar at the Thrombosis Research Institute, Manresa Rd., Emmanuel Kaye Bldg., Chelsea, London SW3 6LR, United Kingdom, or at
[email protected]. * A complete list of the LIFENOX investigators is provided in the Supplementary Appendix, available at NEJM.org. N Engl J Med 2011;365:2463-72. Copyright © 2011 Massachusetts Medical Society.
CONCLUSIONS
The use of enoxaparin plus elastic stockings with graduated compression, as compared with elastic stockings with graduated compression alone, was not associated with a reduction in the rate of death from any cause among hospitalized, acutely ill medical patients. (Funded by Sanofi; LIFENOX ClinicalTrials.gov number, NCT00622648.)
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2463
The
V
n e w e ng l a n d j o u r na l
enous thromboembolism is an important complication in hospitalized patients.1-4 It is estimated that if thromboprophylaxis is not administered, objectively diagnosed deepvein thrombosis — with the potential for fatal pulmonary embolism — will develop in 10 to 20% of medical patients and in 40 to 60% of patients undergoing major orthopedic surgery.5 A retrospective review of 6833 autopsies showed that 81% of fatal cases of pulmonary embolism occurred in nonsurgical patients.6 Pharmacologic thromboprophylaxis has been proved to reduce the incidence of venous thromboembolism in both surgical patients and acutely ill medical patients.5,7-9 In surgical patients, thromboprophylaxis has been shown to reduce the incidence of fatal pulmonary embolism and the rate of death from any cause10,11; in medical patients, studies have shown that thromboprophylaxis is associated with reductions in the rate of venous thromboembolic events, including asymptomatic deep-vein thrombosis assessed as part of a composite study end point.7-9 A meta-analysis of five studies involving medical patients indicated that prophylaxis may be associated with a reduction in the rate of fatal pulmonary embolism but not in the rate of death from any cause.12 Screening for asymptomatic deep-vein thrombosis, with subsequent treatment of the condition, may favorably alter the natural history of venous thromboembolism, thereby masking potential reductions in mortality associated with thromboprophylaxis. The fact that thromboprophylaxis is used more frequently in hospitalized surgical patients than in acutely ill medical patients,13-15 even though current guidelines clearly recommend its use in both patient populations,5,16 may reflect a lack of evidence for a mortality reduction associated with pharmacologic prophylaxis in acutely ill medical patients. In this trial, we evaluated the effect of pharmacologic thromboprophylaxis on the rate of death from any cause in acutely ill medical patients. Patients were randomly assigned to receive the low-molecular-weight heparin enoxaparin or placebo, with both groups assigned to wear elastic stockings with graduated compression.
of
m e dic i n e
sites in China, India, Korea, Malaysia, Mexico, the Philippines, and Tunisia. Recruitment began in January 2008 and was completed in September 2010. The study was conducted in accordance with the principles of the Declaration of Helsinki, including all the applicable amendments set forth by the World Medical Assembly, and with the International Conference on Harmonization guidelines for Good Clinical Practice. The study was approved by the research ethics committee at each participating site. Ethical approval was also obtained from the Comité de Protection des Personnes Île-de-France XI. A steering committee (see the Supplementary Appendix, available with the full text of this article at NEJM.org) designed and led the trial and planned the analyses. Funding and study drugs were provided by the sponsor (Sanofi). The data were gathered by the sponsor and were maintained and analyzed by an independent contract research organization. The steering committee had full access to the data. The first author wrote the first draft of the manuscript, and subsequent drafts were prepared with input from the coauthors, all of whom approved the submission of the final version of the manuscript. An independent data and safety monitoring committee (see the Supplementary Appendix) performed prespecified interim analyses after approximately 10%, 25% (for safety), 50%, and 75% (for both safety and efficacy) of the patients had completed the 30-day follow-up period, with stopping rules based on the Lan–DeMets type of O’Brien–Fleming stopping boundary. The study was conducted in accordance with the research protocol, which, along with the statistical analysis plan, is available at NEJM.org. Study Population
We enrolled men and women, 40 years of age or older, who were hospitalized within 48 hours before randomization for at least one of the following conditions: acute decompensation of heart failure; active cancer (defined as histologically confirmed cancer with an initial diagnosis within the previous 6 months or with a recurrence or metastasis within the previous 6 months), unless the hospitalization was a planned hospitalization for chemotherapy; or severe systemic infection in addition to at least one of the following conditions: Me thods chronic pulmonary disease (e.g., chronic obstrucStudy Oversight tive pulmonary disease, pulmonary fibrosis, or the We conducted this international, multicenter, ran- pulmonary restrictive syndrome), obesity (a bodydomized, double-blind, parallel-group study at 193 mass index [the weight in kilograms divided by
2464
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Low-Molecular-Weight Heparin and Mortality
the square of the height in meters] ≥30), a personal history of venous thromboembolism, or an age of 60 years or older. In addition, eligible patients were required to have an anticipated duration of hospitalization of at least 6 days and an American Society of Anesthesiologists health status score of 3 or less (on a scale of 1 to 6, with higher scores indicating more severe illness) or, for patients with cancer, an Eastern Cooperative Oncology Group performance status score of 2 or less (on a scale of 0 to 5, with higher scores indicating greater severity of illness). All eligible patients were also required to provide written informed consent. The exclusion criteria are listed in the Supplementary Appendix. Study Design
Patients were randomly assigned to receive a subcutaneous injection with either enoxaparin, at a dose of 40 mg (Lovenox [United States] or Clexane [outside the United States], Sanofi), or placebo (0.9% saline) once every 24±4 hours during hospitalization, for 6 to 14 days (10±4 days). The investigators assigned the patients to a group in the sequential order of the treatment numbers available at the site. The treatment-code list of random permuted blocks was generated by an independent contract research organization and was stratified according to center. Patients who were discharged before the completion of the treatment period continued to receive the study medication at home. Knee-high elastic stockings (Ganzoni) that provided graduated pressure from 15 mm Hg (at the ankle) to 10 mm Hg (at the knee) were provided to both groups. The investigators, patients, and research personnel, as well as the members of the steering committee and of the data and safety monitoring committee, were unaware of the group assignments. The study drug was discontinued in the event of an intercurrent illness or adverse event (e.g., creatinine clearance of <30 ml per minute or thrombocytopenia with a platelet count of <50,000 per cubic millimeter), definite venous thromboembolism requiring anticoagulant treatment, or proven heparin-induced thrombocytopenia. Outcomes
The primary efficacy outcome was the rate of death from any cause between the time of randomization and day 30. Secondary efficacy outcomes were the rates of death from any cause between the time of randomization and day 14 and
between the time of randomization and day 90; the rate of cardiopulmonary death (sudden death or death due to acute myocardial infarction, heart failure, pulmonary failure, or pulmonary embolism) at days 14, 30, and 90 after randomization; and the composite of the rate of sudden death or pulmonary embolism at days 14, 30, and 90 after randomization. The main safety outcome was the rate of major hemorrhagic events during the treatment period. Other safety outcomes were clinically relevant nonmajor bleeding, minor bleeding, serious adverse events, nonserious adverse events, and adverse events of special interest, including thrombocytopenia and heparin-induced thrombocytopenia, occurring during the entire observation period (from receipt of informed consent until day 90). A major hemorrhage was defined as overt bleeding associated with one of the following: death; the need for transfusion of at least 2 units of packed red cells or whole blood; a fall in the hemoglobin level of 20 g or more per liter; the requirement for a major therapeutic intervention (e.g., surgery) to stop or control bleeding; or a bleeding site that was retroperitoneal, intracranial, or intraocular.17 Clinically relevant nonmajor bleeding was defined as a nonmajor hemorrhage leading to discontinuation of the study drug or to hospitalization. A minor hemorrhage was defined as overt bleeding that did not meet the criteria for major hemorrhage but was associated with clinical features defined in the protocol. Statistical Analysis
We estimated that with 3944 patients in each group, the study would have 90% power to show a 25% reduction with enoxaparin in the relative risk of death from any cause at 30 days, assuming a rate of death of 7% in the placebo group, at a two-sided alpha level of 0.05. Assuming that we would not be able to evaluate data from 5% of the patients, we estimated that we would have to enroll 8300 patients. The primary population for the efficacy analyses comprised all patients who underwent randomization (intention-to-treat population). The safety analyses were performed on data from all patients who received at least one dose of a study drug. For the primary efficacy analysis, the incidence of death from any cause within 30 days after randomization was compared between the two study groups with the use of a chi-square test. Time-to-event analyses were performed with
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2465
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
8392 Patients were assessed for eligibility
8323 Underwent randomization
4 Were excluded owing to erroneous randomization number
4145 Were assigned to receive placebo
4174 Were assigned to receive enoxaparin
3 Underwent randomization but did not receive study drug and had no follow-up data
9 Underwent randomization but did not receive study drug and had no follow-up data 4136 Were included in the intention-to-treat and safety populations
4171 Were included in the intention-to-treat and safety populations
Figure 1. Screening, Randomization, and Follow-up.
the use of a log-rank test. Kaplan–Meier curves infection (49.8% of whom also had chronic pulwere used to show the probability of events over monary disease). time. All analyses were performed with the SAS The median duration of hospitalization was software package, version 9.1 (SAS Institute). 9 days in both groups. The median duration of treatment was 6 days in both groups, with 93.9% of the patients receiving 6 to 14 days of treatment. R e sult s The median duration of the use of elastic stockPatients ings with graduated compression was 6 days in A total of 8392 patients were assessed for eligi- the enoxaparin group and 7 days in the placebo bility, of whom 8323 were randomly assigned to group. A total of 39 patients (0.5%) were lost to a study group (Fig. 1). A total of 16 patients (0.2%) follow-up at day 30 (20 in the enoxaparin group were subsequently excluded either because they and 19 in the placebo group), and 76 (0.9%) were had been given an erroneous randomization num- lost to follow-up at day 90 (39 in the enoxaparin ber (4 patients) or because they did not receive the group and 37 in the placebo group). study drug and had no follow-up data (12 patients). The final intention-to-treat population in- Efficacy Outcomes cluded 8307 patients, of whom 2071 were enrolled The rate of death from any cause at 30 days was in China (24.9%), 4050 in India (48.8%), 383 in 4.9% in the enoxaparin group and 4.8% in the Korea (4.6%), 292 in Malaysia (3.5%), 396 in Mex- placebo group (risk ratio, 1.0; 95% confidence inico (4.8%), 585 in the Philippines (7.0%), and 530 terval [CI], 0.8 to 1.2; P = 0.83) (Table 2 and Fig. 2). in Tunisia (6.4%). With a rate of death in the placebo group of 4.8% Table 1 shows the baseline characteristics of rather than the 7% originally anticipated, our study the patients. The mean (±SD) age was 65±12 years, had 77% power to detect a 25% reduction in the and 37.3% of the patients (3096 of the 8307 pa- rate of death from any cause and 57% power to tients in the intention-to-treat population) were detect a 20% reduction. The incidence of death women. More than half (64.4%; 5346 of the 8300 from cardiopulmonary causes (including sudden patients with data on the primary reason for hos- death and death due to acute myocardial infarcpitalization) were hospitalized for severe systemic tion, heart failure, pulmonary failure, or pulmo2466
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Low-Molecular-Weight Heparin and Mortality
Table 1. Baseline Characteristics and Primary Reason for Hospitalization in the Intention-to-Treat Population.* Variable
Placebo (N = 4136)
Enoxaparin (N = 4171)
Baseline characteristic Age — yr Female sex — no. (%) Body-mass index†
65.3±12.2
65.6±12.0
1528 (36.9)
1568 (37.6)
23.3±5.4
23.4±5.4
Renal impairment — no./total no. (%) Any Severe‡
1452/4057 (35.8)
1454/4097 (35.5)
202/4057 (5.0)
178/4097 (4.3)
1030/4136 (24.9)
1079/4171 (25.9)
21/4135 (0.5)
27/4168 (0.6)
Selected risk factors for venous thromboembolism — no./total no. (%) Age ≥75 yr Personal history of venous thromboembolism Family history of venous thromboembolism
5/4134 (0.1)
4/4166 (0.1)
Active cancer
239/4136 (5.8)
250/4170 (6.0)
Body-mass index ≥30†
429/4084 (10.5)
431/4111 (10.5)
Coagulation disorder
3/4134 (0.1)
Hospitalization within previous 3 months for acute medical illness
2/4168 (<0.1)
390/4135 (9.4)
377/4168 (9.0)
1297/4134 (31.4)
1280/4166 (30.7)
Primary reason for hospitalization — no./total no. (%) Heart failure NYHA Class I or II NYHA Class III or IV NYHA class not determined Severe systemic infection
210/4134 (5.1)
206/4166 (4.9)
1069/4134 (25.9)
1055/4166 (25.3)
18/4134 (0.4)
19/4166 (0.5)
2336/4134 (56.5)
2383/4166 (57.2)
Active cancer
170/4134 (4.1)
195/4166 (4.7)
Heart failure and severe systemic infection
262/4134 (6.3)
253/4166 (6.1)
Heart failure and active cancer Severe systemic infection and active cancer Heart failure, severe systemic infection, and active cancer None of the above
5/4134 (0.1)
7/4166 (0.2)
62/4134 (1.5)
45/4166 (1.1)
2/4134 (<0.1) 25/4136 (0.6)
3/4166 (0.1) 24/4171 (0.6)
* Plus–minus values are means ±SD. There were no significant differences between the two study groups in any of the baseline characteristics (P>0.05 for all comparisons). NYHA denotes New York Heart Association. † The body-mass index is the weight in kilograms divided by the square of the height in meters. ‡ Renal impairment was classified as severe if the creatinine clearance was 30 ml per minute or less.
nary embolism) and the composite of the rate of sudden death or pulmonary embolism at 30 days did not differ significantly between the groups (Table 2). The most common cause of death by day 30 was pulmonary failure, which occurred in 2.1% of the patients in the enoxaparin group and in 1.8% of the patients in the placebo group (Table 2). Deaths due to cancer occurred in 0.5% of the patients in the enoxaparin group and in 0.8% of the patients in the placebo group, and the rate of sudden death was 0.7% in both groups. Three patients in the enoxaparin group (0.1%) died from a hemorrhage, and one patient in each group died from
a pulmonary embolism. No autopsies were performed in either group. Multivariate logistic regression identified the following variables as factors that were independently associated with increased mortality at day 30: a diagnosis of active cancer (odds ratio, 4.0; 95% CI, 2.8 to 5.8; P<0.001), chronic pulmonary disease (odds ratio, 1.3; 95% CI, 1.1 to 1.7; P = 0.02), two or more acute illnesses (odds ratio, 1.6; 95% CI, 1.2 to 2.3; P = 0.003), and renal impairment (odds ratio, 2.0; 95% CI, 1.6 to 2.4; P<0.001). The hazard rate (i.e., a rate that is derived from a time-to-event analysis) for death from any cause at day 90 was 8.4% in the enoxaparin group
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2467
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. Efficacy Outcomes and the Characteristics and Primary Causes of Death.* Placebo (N = 4136)
Variable
Enoxaparin (N = 4171)
Risk Ratio for Enoxaparin vs. Placebo (95% CI)
P Value†
no. of patients (%) Efficacy outcomes 14-day mortality 119 (2.9)
121 (2.9)
1.0 (0.8–1.3)
0.95
Cardiopulmonary death
Death from any cause
93 (2.2)
86 (2.1)
0.9 (0.7–1.2)
0.56
Sudden death or pulmonary embolism
27 (0.7)
20 (0.5)
0.7 (0.4–1.3)
0.29
Death from any cause
199 (4.8)
205 (4.9)
1.0 (0.8–1.2)
0.83
Cardiopulmonary death
135 (3.3)
141 (3.4)
1.0 (0.8–1.3)
0.77
29 (0.7)
29 (0.7)
1.0 (0.6–1.7)
0.97
Death from any cause
355 (8.6)
348 (8.4)
1.0 (0.8–1.1)
0.71
Cardiopulmonary death
214 (5.3)
211 (5.1)
1.0 (0.8–1.2)
0.82
40 (1.0)
43 (1.1)
1.1 (0.7–1.6)
0.77
Explained
67 (1.6)
75 (1.8)
Unexplained
22 (0.5)
24 (0.6)
Explained
92 (2.2)
94 (2.3)
Unexplained
15 (0.4)
9 (0.2)
28 (0.7)
28 (0.7)
4 (0.1)
3 (0.1)
30-day mortality
Sudden death or pulmonary embolism 90-day mortality‡
Sudden death or pulmonary embolism Characteristics of deaths recorded at day 30§ Abrupt
Insidious
Primary adjudicated reason for death at day 30 Sudden death Acute myocardial infarction Stroke
2 (<0.1)
3 (0.1)
Heart failure
26 (0.6)
23 (0.6)
Cancer
33 (0.8)
22 (0.5)
Pulmonary failure
76 (1.8)
86 (2.1)
Multiorgan failure
6 (0.1)
9 (0.2)
Sepsis
8 (0.2)
10 (0.2)
Accident or trauma
0
Pulmonary embolism
1 (<0.1)
1 (<0.1)
Hemorrhage
0
3 (0.1)
Unclassified
15 (0.4)
1 (<0.1)
16 (0.4)
* Sudden death, defined as an unexpected death occurring in a short period (<1 hour after onset of symptoms) in a patient in whom there was no previous diagnosis of a fatal condition, was an adjudicated event. Abrupt death, as reported by the investigator, was a nonadjudicated event. † The P values at day 14 and day 30 were calculated with the use of a chi-square test; the P values at day 90 were calculated with the use of a log-rank test. ‡ Hazard rates (i.e., rates derived from a time-to-event analysis) and hazard ratios, rather than percentages and risk ratios, are provided for 90-day outcomes. § Information was not available on the characteristics of three deaths in each group.
(with death occurring in 348 of 4171 patients) and 8.6% in the placebo group (with death occurring in 355 of 4136 patients) (hazard ratio, 1.0; 95% CI, 0.8 to 1.1; P = 0.71) (Table 2); the 2468
hazard rates for cardiopulmonary deaths were 5.1% and 5.3% in the two groups, respectively. Up to day 90, there was clinical suspicion of venous thromboembolism in 0.5% of the pa-
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Low-Molecular-Weight Heparin and Mortality
Safety Outcomes and Adverse Events
Major bleeding events during the treatment period and up to 48 hours after the treatment period were reported in 16 patients in the enoxaparin group (0.4%) and in 11 patients in the placebo group (0.3%) (risk ratio with enoxaparin, 1.4; 95% CI, 0.7 to 3.1; P = 0.35) (Table 3). The rates of minor bleeding were higher in the enoxaparin group than in the placebo group, and the combined rates of all bleeding events were higher in the enoxaparin group (risk ratio, 1.5; 95% CI, 1.1 to 2.1). The rate of all adverse events was 37.8% in the enoxaparin group (with an event occurring in 1577 of 4171 patients) and 36.9% in the placebo group (with an event occurring in 1528 of 4136 patients). The two groups did not differ significantly with respect to the rate of either serious adverse events (5.8% [243 of 4171 patients] in the enoxaparin group and 5.3% [219 of 4136 patients] in the placebo group) or adverse events leading to death (2.9% [121 of 4171 patients] and 2.9% [119 of 4136 patients] in the two groups, respectively). The rate of adverse events leading to permanent discontinuation of the study drug was higher in the enoxaparin group than in the placebo group (3.6% [151 of 4171 patients] vs. 2.8% [116 of 4136 patients]). There were no cases of heparin-induced thrombocytopenia.
Discussion We did not detect a difference in the rate of death from any cause among patients hospitalized for an acute medical illness when a strategy of pharmacologic prophylaxis in addition to the use of elastic stockings with graduated compression was compared with the use of elastic stockings with graduated compression alone. Pharmacologic prophylaxis was not associated with increased rates of major bleeding but was associated with increased rates of total bleeding. These findings appear to be counterintuitive, given the fact that pharmacologic prophylaxis has been shown to reduce the risk of venous thromboembolism, including asymptomatic deep-vein thrombosis, by at least 45% in hospitalized, acute-
20
Placebo
18
Death from Any Cause (%)
tients in the enoxaparin group (22 of 4072 patients) and in 0.7% of the patients in the placebo group (27 of 4044 patients). The diagnosis was confirmed by objective testing in 0.2% of the patients in the enoxaparin group and in 0.1% of the patients in the placebo group.
Enoxaparin
16 14 12 10 8 6 4 2 0
1
30
60
90
Days since Randomization No. of Deaths/ No. at Risk Placebo Enoxaparin
1/4136 2/4171
199/3922 205/3950
291/3813 292/3846
355/3745 348/3785
Figure 2. Death from Any Cause. The percentage of patients in the intention-to-treat population who died from any cause up to day 90 after randomization is shown.
ly ill medical patients.7-9 It has been assumed that the natural history of deep-vein thrombosis, well established in surgical patients,18 would be the same in acutely ill medical patients, in whom it has been shown that those with asymptomatic proximal vein thrombi have a higher risk of death than those with distal thrombi.19 This assumption may be incorrect; perhaps the natural history of deep-vein thrombosis differs between medical and surgical patients. Although Halkin et al.20 reported that the rate of death among medical patients was reduced with pharmacologic prophylaxis, other studies have failed to show that result, regardless of whether the studies were evaluating the use of in-hospital prophylaxis7-9,21-24 or extended, out-of-hospital prophylaxis.25 The study by Gärdlund was promising because it showed that pharmacologic prophylaxis appeared to delay the early occurrence of fatal pulmonary embolism; however, prophylaxis did not improve the overall clinical outcome by day 60.22 Our study may have been underpowered to show a between-group difference in mortality. The rate of death from any cause was lower than expected but was similar in the two study groups at all the time points we assessed (day 14, day 30, and day 90). With an observed mortality of 4.8% rather than the 7% originally anticipated, our study had 77% power to detect a 25% reduction in the rate of death from any cause and 57% power to detect a 20% reduction. The Prophylaxis
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2469
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. Bleeding Outcomes during the Treatment Period. Placebo (N = 4136)
Outcome
Enoxaparin (N = 4171)
Risk Ratio for Enoxaparin vs. Placebo (95% CI)
P Value
no. of patients (%) Any bleeding*
60 (1.5)
91 (2.2)
1.5 (1.1–2.1)
0.01
Adjudicated major bleeding
11 (0.3)
16 (0.4)
1.4 (0.7–3.1)
0.35
Resulting in death
0
2 (<0.1)
Requiring transfusion of ≥2 units of red cells or whole blood
6 (0.1)
5 (0.1)
Resulting in fall in hemoglobin of ≥20 g/liter
8 (0.2)
12 (0.3)
Requiring surgical intervention
2 (<0.1)
1 (<0.1)
Retroperitoneal, intracranial, or intraocular
1 (<0.1)
1 (<0.1)
Other
2 (<0.1)
0
Minor bleeding Any
47 (1.1)
73 (1.8)
Clinically relevant nonmajor
14 (0.3) 4 (0.1)
Unclassified bleeding
1.5 (1.1–2.2)
0.02
18 (0.4)
1.3 (0.6–2.6)
0.49
6 (0.1)
1.5 (0.4–5.3)
0.75
* Included in this category are patients who had at least one bleeding event. Some patients may have had more than one type of bleeding event.
in Medical Patients with Enoxaparin (MEDENOX) study, which established the efficacy of the regimen of 40 mg of enoxaparin daily in acutely ill medical patients, showed a 25% reduction in mortality associated with this pharmacologic regimen.7 Although a pulmonary embolism is identified on autopsy in 9 to 21% of medical patients,26-28 in the MEDENOX study, only one death in the placebo group and two in the group receiving 40 mg of enoxaparin were attributed to pulmonary embolism by the end of the followup period. This finding suggests that the observed favorable trend in overall survival associated with effective prophylaxis may be due to a reduction not only in deaths related to venous thromboembolism but also in deaths from other causes, most likely cardiovascular events. Indeed, pharmacologic prophylaxis in surgical patients has been shown to reduce, in addition to fatal pulmonary embolism, fatal myocardial infarction and death from other causes.10,11 Multiple coexisting illnesses and numerous other potential causes of death in medical patients might make fatal pulmonary embolism a less important determinant of mortality in this group than in surgical patients, thus diminishing the ability of pharmacologic prophylaxis to improve the overall clinical outcome. In addition, the reported rate of pulmo2470
nary embolism has historically been lower in Asian populations than in Western populations,29 although contemporary studies indicate that there is little difference in the frequency of deep-vein thrombosis between Asian and Western populations.30-32 One possible explanation for our findings is that the use of elastic stockings with graduated compression alone is effective in preventing venous thromboembolism, thus reducing the frequency of fatal pulmonary embolism. Although elastic stockings with graduated compression have been shown to be effective in reducing the risk of deep-vein thrombosis in moderate-risk surgical patients and some medical patient populations,33,34 the use of stockings did not prevent the occurrence of deep-vein thrombosis in patients recuperating from severe, disabling stroke who were participants in the Clots in Legs or Stocking after Stroke trial (CLOTS; Current Controlled Trials number, ISRCTN28163533).35 Furthermore, the knee-length stockings used in our study have recently been shown to be less effective than thighlength stockings for the prevention of deep-vein thrombosis.36 The prevention of venous thromboembolism in acutely ill medical patients — for a population in which low-molecular-weight heparins have al-
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Low-Molecular-Weight Heparin and Mortality
ready been shown to be effective7,8 — was not the primary objective of our study. Therefore, we did not screen for asymptomatic deep-vein thrombosis. We observed very low rates of symptomatic venous thromboembolism in both groups. These low rates may be due either to a decreased awareness of the disease in participating countries in which less frequent diagnostic testing for suspected events was offered or to the inclusion in this study of a population that was at lower risk for venous thromboembolism, as compared with other studies. For example, the mean age of the patients in our study was approximately 9 years younger than the mean age of patients in the MEDENOX study, and the mean body-mass index was approximately 2 units lower; the proportion of obese patients was 11% in our study, as compared with 20% in the MEDENOX study. In addition, in our study, as compared with the MEDENOX study, there was a decrease by a factor of almost 10 in the proportion of patients with a history of venous thromboembolism.7 We did not collect data on mobility status, an important determinant of the risk of venous thromboembolism.5 The adherence both to the injections and to the use of elastic stockings was excellent in both groups in our study. In summary, the results of the LIFENOX trial showed that among hospitalized, acutely ill medical patients, the rate of death from any cause did not differ significantly between patients who were randomly assigned to pharmacologic prophylaxis with enoxaparin in addition to elastic stockings
with graduated compression and those who were assigned to elastic stockings with graduated compression alone. Pharmacologic thromboprophylaxis continues to have proven benefits in preventing venous thromboembolism, thus reducing the need for the treatment of symptomatic venous thromboembolism with high doses of anticoagulant agents over a prolonged period of time. Furthermore, venous thromboembolism can lead to nonfatal complications such as the post-thrombotic syndrome and chronic thromboembolic pulmonary hypertension, which are often not treated successfully.
Supported by Sanofi. Dr. Kakkar reports receiving consulting fees, grant support through his institution, and lecture fees from Bayer Healthcare, Boehringer Ingelheim, Bristol-Myers Squibb, Daiichi Sankyo, Eisai, GlaxoSmithKline, Pfizer, and Sanofi; Dr. Cimminiello, receiving consulting fees from Sanofi; Dr. Goldhaber, receiving consulting fees from Boehringer Ingelheim, Bristol-Myers Squibb, Daiichi Sankyo, Eisai, EKOS, Medscape, Merck, Portola, and Sanofi and grant support through his institution from Boehringer Ingelheim, Bristol-Myers Squibb, Eisai, EKOS, Johnson & Johnson, and Sanofi; Dr. Parakh, receiving consulting fees from Bayer, Boehringer Ingelheim, Daiichi Sankyo, GlaxoSmithKline, Novartis, Pfizer, and Sanofi and support for travel, accommodations, or meeting expenses from Bayer, Daiichi Sankyo, and Sanofi; and Dr. Bergmann, receiving consulting fees from AstraZeneca, Association des Pharmaciens de l’Industrie–Communication Globale Santé–Prioritis, Bayer, GlaxoSmithKline, Lilly, Novartis, Pfizer, and Sanofi and grant support through his institution from Agence Nationale de Recherche contre le SIDA, Bayer, GlaxoSmithKline, and Pfizer. No other 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 A. Correges (Altizem) and G. Salette (Sanofi) for performing statistical analyses, R. Spice (Excerpta Medica, funded by Sanofi) for editorial assistance, and Dr. S. Rushton Smith (Thrombosis Research Institute) for writing assistance.
References 1. Anderson FA Jr, Wheeler HB, Gold-
berg RJ, Hosmer DW, Forcier A. The prevalence of risk factors for venous thromboembolism among hospital patients. Arch Intern Med 1992;152:1660-4. 2. Moreno-Cabral R, Kistner RL, Nordyke RA. Importance of calf vein thrombophlebitis. Surgery 1976;80:735-42. 3. The Surgeon General’s call to action to prevent deep venous thrombosis and pulmonary embolism. Washington, DC: Department of Health and Human Services, 2008 (http://www.surgeongeneralgov/ topics/deepvein). 4. Oger E, Bressollette L, Nonent M, et al. High prevalence of asymptomatic deep vein thrombosis on admission in a medical unit among elderly patients. Thromb Haemost 2002;88:592-7. 5. Geerts WH, Bergqvist D, Pineo GF, et al. Prevention of venous thromboembolism: American College of Chest Physicians Evidence-Based Clinical Practice Guide-
lines (8th Edition). Chest 2008;133:Suppl: 381S-453S. 6. Alikhan R, Peters F, Wilmott R, Cohen AT. Fatal pulmonary embolism in hospitalized patients: a necropsy review. J Clin Pathol 2004;57:1254-7. 7. Samama MM, Cohen AT, Darmon JY, et al. A comparison of enoxaparin with placebo for the prevention of venous thromboembolism in acutely ill medical patients. N Engl J Med 1999;341:793-800. 8. Leizorovicz A, Cohen AT, Turpie AG, Olsson CG, Vaitkus PT, Goldhaber SZ. Randomized, placebo-controlled trial of dalteparin for the prevention of venous thromboembolism in acutely ill medical patients. Circulation 2004;110:874-9. 9. Cohen AT, Davidson BL, Gallus AS, et al. Efficacy and safety of fondaparinux for the prevention of venous thromboembolism in older acute medical patients: randomised placebo controlled trial. BMJ 2006;332:325-9.
10. Prevention of fatal postoperative pul-
monary embolism by low doses of heparin: an international multicentre trial. Lancet 1975;2:45-51. 11. Collins R, Scrimgeour A, Yusuf S, Peto R. Reduction in fatal pulmonary embolism and venous thrombosis by perioperative administration of subcutaneous heparin: overview of results of randomized trials in general, orthopedic, and urologic surgery. N Engl J Med 1988;318:1162-73. 12. Dentali F, Douketis JD, Gianni M, Lim W, Crowther MA. Meta-analysis: anticoagulant prophylaxis to prevent symptomatic venous thromboembolism in hospitalized medical patients. Ann Intern Med 2007;146:278-88. 13. Cohen AT, Tapson VF, Bergmann JF, et al. Venous thromboembolism risk and prophylaxis in the acute hospital care setting (ENDORSE study): a multinational cross-sectional study. Lancet 2008;371: 387-94. [Erratum, Lancet 2008;371:1914.]
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2471
Low-Molecular-Weight Heparin and Mortality 14. Kakkar AK, Cohen AT, Tapson VF, et
al. Venous thromboembolism risk and prophylaxis in the acute care hospital setting (ENDORSE survey): findings in surgical patients. Ann Surg 2010;251:3308. 15. Bergmann JF, Cohen AT, Tapson VF, et al. Venous thromboembolism risk and prophylaxis in hospitalised medically ill patients: the ENDORSE Global Survey. Thromb Haemost 2010;103:736-48. 16. Nicolaides AN, Fareed J, Kakkar AK, et al. Prevention and treatment of venous thromboembolism: international consensus statement (guidelines according to scientific evidence). Int Angiol 2006;25: 101-61. 17. Schulman S, Kearon C. Definition of major bleeding in clinical investigations of antihemostatic medicinal products in non-surgical patients. J Thromb Haemost 2005;3:692-4. 18. Kakkar VV, Howe CT, Flanc C, Clarke MB. Natural history of postoperative deepvein thrombosis. Lancet 1969;2:230-2. 19. Vaitkus PT, Leizorovicz A, Cohen AT, et al. Mortality rates and risk factors for asymptomatic deep vein thrombosis in medical patients. Thromb Haemost 2005; 93:76-9. 20. Halkin H, Goldberg J, Modan M, Modan B. Reduction of mortality in general medical in-patients by low-dose heparin prophylaxis. Ann Intern Med 1982; 96:561-5. 21. Dahan R, Houlbert D, Caulin C, et al. Prevention of deep vein thrombosis in elderly medical in-patients by a low molecular weight heparin: a randomized
double-blind trial. Haemostasis 1986;16: 159-64. 22. Gärdlund B. Randomised, controlled trial of low-dose heparin for prevention of fatal pulmonary embolism in patients with infectious diseases. Lancet 1996;347:135761. 23. Fraisse F, Holzapfel L, Couland JM, et al. Nadroparin in the prevention of deep vein thrombosis in acute decompensated COPD. Am J Respir Crit Care Med 2000; 161:1109-14. 24. Mahé I, Bergmann JF, dAzémar P, Vaissie JJ, Caulin C. Lack of effect of a lowmolecular-weight heparin (nadroparin) on mortality in bedridden medical in-patients: a prospective randomised double-blind study. Eur J Clin Pharmacol 2005;61:347-51. 25. Hull RD, Schellong SM, Tapson VF, et al. Extended-duration venous thromboembolism prophylaxis in acutely ill medical patients with recently reduced mobility: a randomized trial. Ann Intern Med 2010;153:8-18. 26. Karwinski B, Svendsen E. Comparison of clinical and postmortem diagnosis of pulmonary embolism. J Clin Pathol 1989;42:135-9. 27. Stein PD, Henry JW. Prevalence of acute pulmonary embolism among patients in a general hospital and at autopsy. Chest 1995;108:978-81. 28. Sandler DA, Martin JF. Autopsy proven pulmonary embolism in hospital patients: are we detecting enough deep vein thrombosis? J R Soc Med 1989;82:203-5. 29. Hwang WS. The rarity of pulmonary thromboembolism in Asians. Singapore Med J 1968;9:276-9.
30. Leizorovicz A, Turpie AG, Cohen AT, Wong L, Yoo MC, Dans A. Epidemiology of venous thromboembolism in Asian patients undergoing major orthopedic surgery without thromboprophylaxis. J Thromb Haemost 2005;3:28-34. 31. Leizorovicz A. Epidemiology of postoperative venous thromboembolism in Asian patients: results of the SMART venography study. Haematologica 2007;92:1194200. 32. Piovella F, Wang CJ, Lu H, et al. Deepvein thrombosis rates after major orthopedic surgery in Asia: an epidemiological study based on postoperative screening with centrally adjudicated bilateral venography. J Thromb Haemost 2005;3:266470. 33. Sachdeva A, Dalton M, Amaragiri SV, Lees TA. Elastic compression stockings for prevention of deep vein thrombosis. Coch rane Database Syst Rev 2010;7:CD001484. 34. Clagett GP, Reisch JS. Prevention of venous thromboembolism in general surgical patients: results of meta-analysis. Ann Surg 1988;208:227-40. 35. Dennis M, Sandercock PA, Reid J, et al. Effectiveness of thigh-length graduated compression stockings to reduce the risk of deep vein thrombosis after stroke (CLOTS trial 1): a multicentre, randomised controlled trial. Lancet 2009;373:1958-65. 36. CLOTS (Clots in Legs Or sTockings after Stroke) Trial Collaboration. Thighlength versus below-knee stockings for deep venous thrombosis prophylaxis after stroke: a randomized trial. Ann Intern Med 2010;153:553-62. Copyright © 2011 Massachusetts Medical Society.
receive immediate notification when an article is published online first
To be notified by e-mail when Journal articles are published Online First, sign up at NEJM.org.
2472
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
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
Incorporation of Bevacizumab in the Primary Treatment of Ovarian Cancer Robert A. Burger, M.D., Mark F. Brady, Ph.D., Michael A. Bookman, M.D., Gini F. Fleming, M.D., Bradley J. Monk, M.D., Helen Huang, M.S., Robert S. Mannel, M.D., Howard D. Homesley, M.D., Jeffrey Fowler, M.D., Benjamin E. Greer, M.D., Matthew Boente, M.D., Michael J. Birrer, M.D., Ph.D., and Sharon X. Liang, M.D., for the Gynecologic Oncology Group*
A bs t r ac t BACKGROUND
Vascular endothelial growth factor is a key promoter of angiogenesis and disease progression in epithelial ovarian cancer. Bevacizumab, a humanized anti–vascular endothelial growth factor monoclonal antibody, has shown single-agent activity in women with recurrent tumors. Thus, we aimed to evaluate the addition of bevacizumab to standard front-line therapy. METHODS
In our double-blind, placebo-controlled, phase 3 trial, we randomly assigned eligible patients with newly diagnosed stage III (incompletely resectable) or stage IV epithelial ovarian cancer who had undergone debulking surgery to receive one of three treatments. All three included chemotherapy consisting of intravenous paclitaxel at a dose of 175 mg per square meter of body-surface area, plus carboplatin at an area under the curve of 6, for cycles 1 through 6, plus a study treatment for cycles 2 through 22, each cycle of 3 weeks’ duration. The control treatment was chemotherapy with placebo added in cycles 2 through 22; bevacizumab-initiation treatment was chemotherapy with bevacizumab (15 mg per kilogram of body weight) added in cycles 2 through 6 and placebo added in cycles 7 through 22. Bevacizumab-throughout treatment was chemotherapy with bevacizumab added in cycles 2 through 22. The primary end point was progression-free survival. RESULTS
Overall, 1873 women were enrolled. The median progression-free survival was 10.3 months in the control group, 11.2 in the bevacizumab-initiation group, and 14.1 in the bevacizumab-throughout group. Relative to control treatment, the hazard ratio for progression or death was 0.908 (95% confidence interval [CI], 0.795 to 1.040; P = 0.16) with bevacizumab initiation and 0.717 (95% CI, 0.625 to 0.824; P<0.001) with bevacizumab throughout. At the time of analysis, 76.3% of patients were alive, with no significant differences in overall survival among the three groups. The rate of hypertension requiring medical therapy was higher in the bevacizumab-initiation group (16.5%) and the bevacizumab-throughout group (22.9%) than in the control group (7.2%). Gastrointestinal-wall disruption requiring medical intervention occurred in 1.2%, 2.8%, and 2.6% of patients in the control group, the bevacizumab-initiation group, and the bevacizumab-throughout group, respectively.
From the Fox Chase Cancer Center, Philadelphia (R.A.B.); the Gynecologic Oncology Group Statistical and Data Center, Roswell Park Cancer Institute, Buffalo, NY (M.F.B., H.H.); the Arizona Cancer Center, Tucson (M.A.B.); the University of Chicago, Chicago (G.F.F.); Creighton University School of Medicine at St. Joseph’s Hospital and Medical Center, Phoenix, AZ (B.J.M.); the University of Oklahoma Health Sciences Center, Oklahoma City (R.S.M.); Wake Forest University School of Medicine, Winston-Salem, NC (H.D.H.); James Cancer Hospital at Ohio State University, Hilliard (J.F.); Seattle Cancer Care Alliance, Seattle (B.E.G.); Minnesota Oncology and Hematology, Minneapolis (M.B.); Harvard Medical School and Massachusetts General Hospital, Boston (M.J.B.); and North Shore–Long Island Jewish Health System, New Hyde Park, NY (S.X.L.). Address reprint requests to Dr. Burger at Fox Chase Cancer Center, 333 Cottman Ave., Philadelphia, PA 19111, or at robert.a.burger@ fccc.edu. * The member institutions of the Gynecologic Oncology Group that participated in this study are listed in the Supplementary Appendix, available at NEJM.org. N Engl J Med 2011;365:2473-83. Copyright © 2011 Massachusetts Medical Society.
CONCLUSIONS
The use of bevacizumab during and up to 10 months after carboplatin and paclitaxel chemotherapy prolongs the median progression-free survival by about 4 months in patients with advanced epithelial ovarian cancer. (Funded by the National Cancer Institute and Genentech; ClinicalTrials.gov number, NCT00262847.) n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2473
The
n e w e ng l a n d j o u r na l
E
pithelial ovarian cancer and related cancers lead to 15,000 deaths in the United States annually, representing the fifth leading cause of death from cancer among women.1 The poor prognosis is usually attributed to advanced stage at diagnosis and inadequate chemotherapy. Vascular endothelial growth factor (VEGF) and angiogenesis are important promoters of ovariancancer progression.2-6 Both correlate directly with the extent of disease and inversely with progression-free survival7-9 and overall survival,8,10-13 often independently of known prognostic factors.7-10,12,13 Bevacizumab, a humanized VEGF-neutralizing monoclonal antibody, inhibits tumor angiogenesis14 and has shown single-agent activity in phase 2 epithelial ovarian cancer trials.15,16 We investigated the integration of bevacizumab into front-line ovarian cancer therapy.
Me thods Patients
Eligibility criteria (see the Supplementary Appendix, available with the full text of this article at NEJM.org) included previously untreated, incompletely resectable stage III or any stage IV epithelial ovarian, primary peritoneal, or fallopian-tube cancer histologically confirmed by the Gynecologic Oncology Group (GOG) Pathology Committee after standard abdominal surgery with maximal debulking effort within 12 weeks before study entry; a GOG performance status score (see the Supplementary Appendix) of 0 (fully active) to 2 (ambulatory and capable of self-care but unable to work; up and about more than 50% of waking hours); and no history of clinically significant vascular events or evidence of intestinal obstruction. Owing to competing trials, patients with stage III disease and no residual lesions greater than 1 cm in maximal diameter were initially excluded, but after a protocol modification they were permitted. All patients provided written informed consent before enrollment. Study Design
The study (number GOG-0218) was a double-blind, placebo-controlled phase 3 trial (see the Supplementary Appendix). The protocol is available at NEJM.org. The authors wrote the manuscript and take responsibility for the accuracy and completeness of the reported data and for the fidelity of the report to the protocol. 2474
of
m e dic i n e
Each of the three study regimens comprised 22 3-week cycles with intravenous infusions on day 1, with the first 6 cycles consisting of standard chemotherapy with carboplatin at an area under the curve of 6 and paclitaxel at a dose of 175 mg per square meter of body-surface area. Control treatment was chemotherapy with placebo added in cycles 2 through 22; bevacizumab-initiation treatment was chemotherapy with bevacizumab (15 mg per kilogram of body weight) added in cycles 2 through 6 and placebo added in cycles 7 through 22. Bevacizumab-throughout treatment was chemotherapy with bevacizumab added in cycles 2 through 22. Bevacizumab or placebo was initiated at cycle 2, rather than cycle 1, to reduce the risk of wound-healing complications. Treatment was discontinued at the onset of disease progression, unacceptable toxic effects, completion of all 22 cycles, or withdrawal — whichever came first. Disease was assessed before cycle 1 by means of computed tomography or magnetic resonance imaging of at least the abdomen and pelvis, measurement of the serum cancer antigen 125 (CA-125) level,17 and physical examination. In patients without progression, imaging was repeated after treatment cycles 3, 6, 10, 14, 18, and 22. Serum CA-125 levels were measured and physical examinations were performed at the beginning of each cycle for cycles 1 through 6 (chemotherapy) and at the beginning of alternate cycles for cycles 7 through 22 (extended therapy). After completing study treatment, disease assessments were repeated every 3 months for 2 years, then every 6 months for 3 years, and then annually. The quality of life was compared among the three groups with the use of the Trial Outcome Index of the Functional Assessment of Cancer Therapy–Ovary (FACT-O TOI) survey.18 The summary score based on the survey, with a possible total of 112 points and higher scores indicating better quality of life, encompasses aspects of quality of life in patients with advanced ovarian cancer (e.g., pain, fatigue, abdominal symptoms, and functional status). Questionnaires were completed before cycles 1, 4, 7, 13, and 22, as well as 6 months after completing the study therapy. Disease and quality-oflife evaluations were performed at these time points even if patients discontinued treatment (except if they discontinued because of disease progression, in which case disease evaluations were omitted). Safety was monitored during each cycle. Ad-
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Bevacizumab in Ovarian Cancer
ministration of myeloid growth factor was permitted only to manage febrile neutropenia or grade 4 neutropenia (absolute neutrophil count, <500 per cubic millimeter) persisting for 7 days or more or as subsequent prophylaxis. In patients with limiting peripheral neuropathy or hypersensitivity, paclitaxel was replaced with docetaxel (75 mg per square meter) (see the Supplementary Appendix). The bevacizumab (and placebo) dose was modified only in patients whose weight changed by more than 10% but could be delayed or discontinued depending on the occurrence, duration, and severity of uncontrolled hypertension (systolic blood pressure >150 mm Hg or diastolic blood pressure >90 mm Hg), proteinuria (urine proteinto-creatinine ratio ³3.5), wound or bowel-wall disruption (of any grade, during cycle 2 or later), reversible posterior leukoencephalopathy syndrome, arterial thrombosis (grade ³3 at any time or grade 2 during cycle 2 or later), and venous thrombosis, coagulopathy, intestinal obstruction, or hypersensitivity of grade 3 or greater (see the Supplementary Appendix). Statistical Analysis
The statistical analysis plan is available at NEJM .org. Patients were stratified on the basis of GOG performance-status score and cancer stage and debulking status (stage III cancer and maximal residual lesion diameter ≤1 cm vs. stage III cancer and maximal residual lesion diameter >1 cm vs. stage IV cancer) before being randomly assigned to a treatment group according to a minimization procedure.19 The primary end point was initially specified as overall survival but was changed to progression-free survival during the trial (see the Discussion section). Thereafter, treatment assignments could be revealed to the study investigators and patients if documented progression occurred. Progression-free survival and overall survival were calculated from the date of enrollment. Progression-free survival was considered to have ended at the time of cancer progression as shown on radiography, according to the Response Evaluation Criteria in Solid Tumors criteria (see the Supplementary Appendix)20; an increase in the CA-125 level according to Gynecologic Cancer InterGroup criteria21; global deterioration of health; or death from any cause. Progression defined solely on the basis of increased CA-125 level was permitted only if the patient had completed chemotherapy. If patients remained free of progres-
sion at their last follow-up visit, data on duration of progression-free survival were censored at the time of the last radiographic assessment. Differences in progression-free survival among the three groups were assessed by means of the log-rank test.22 A sample size of 1800 was estimated to provide 90% statistical power to detect a 23% reduction in the hazard for progression with either of the two bevacizumab-containing regimens versus the control regimen while limiting the overall one-sided type I error for both comparisons to 2.5%. The final analysis was planned to be conducted after at least 375 patients in the control group died or had disease progression. Relative hazard ratios were estimated with the use of a proportional-hazards model.23 The progression-free survival and overall survival analyses included all enrolled patients. All reported P values are two-sided. Differences in FACT-O TOI scores among the three groups were assessed by means of a linear mixed model with adjustment for baseline score and age. Assessment time points were treated as categorical. Hypotheses were tested at a 1.67% significance level to account for multiple comparisons. Adverse events, graded with the use of National Cancer Institute Common Terminology Criteria for adverse events (version 3),24 were reported until 30 days after the last study treatment had been administered and were summarized for patients who received at least one cycle of bevacizumab or placebo. Differences among the groups in the severity of adverse events were examined by means of Fisher’s exact test.25
R e sult s Patients
Between October 2005 and June 2009, 1873 women were enrolled from 336 institutions in the United States, Canada, South Korea, and Japan (Fig. 1). By the time eligibility was broadened in July 2007, a total of 467 patients had enrolled; 1299 had enrolled by October 2008, when the primary end point was changed to progression-free survival. A complete data sweep was initiated on January 2, 2010, and the database was locked on February 5, 2010. Factors that could influence treatment outcome were evenly distributed across treatment groups (Table 1). Over 80% of patients were non-Hispanic white, over 80% had serous adenocarcinomas,
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2475
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
1873 Patients were enrolled and underwent randomization
625 Were assigned to control therapy:
625 Were assigned to bevacizumabinitiation therapy:
623 Were assigned to bevacizumabthroughout therapy:
Cycles 1–6: Carboplatin, AUC 6 Paclitaxel, 175 mg/m2 Bevacizumab, 15 mg/kg (starting in cycle 2) every 3 wk Cycles 7–22: Placebo every 3 wk
Cycles 1–6: Carboplatin, AUC 6 Paclitaxel, 175 mg/m2 Bevacizumab, 15 mg/kg (starting in cycle 2) every 3 wk Cycles 7–22: Bevacizumab, 15 mg/kg every 3 wk
4 Did not receive study treatment 621 Were included in safety analysis
1 Did not receive study treatment 624 Were included in safety analysis
5 Did not receive study treatment 618 Were included in safety analysis
625 Were included in efficacy analysis 423 Had disease progression or death 156 Died
625 Were included in efficacy analysis 418 Had disease progression or death 150 Died
623 Were included in efficacy analysis 360 Had disease progression or death 138 Died
Cycles 1–6: Carboplatin, AUC 6 Paclitaxel, 175 mg/m2 Placebo (starting in cycle 2) every 3 wk Cycles 7–22: Placebo every 3 wk
Figure 1. Enrollment, Randomization, and Follow-up of the Study Patients. After pathological review and review of the case-report forms, 29 enrolled patients were found to have tumor characteristics that would have rendered them ineligible: 13 had a stage of cancer according to the International Federation of Gynecology and Obstetrics (FIGO) staging system that did not meet the eligibility criteria, 5 had an inadequate specimen for central review, 3 had borderline tumors (low malignant potential), 7 had histologic characteristics that made them ineligible, and 1 had an invalid primary-tumor site. Data on these patients were analyzed as if they had been eligible. AUC denotes area under the curve.
and the majority of cancers were tumor grade 3. The cohort had a relatively poor prognosis: 40% had stage III cancer with maximal residual lesion diameter greater than 1 cm and 26% had stage IV disease. Nineteen percent of patients overall (16%, 17%, and 24% in the control group, bevacizumabinitiation group, and the bevacizumab-throughout group, respectively) completed the planned treatment, and 15% overall were still receiving treatment (in the extended-therapy phase) at the time of the database lock (see the Supplementary Appendix). Sixty-six percent of the study population discontinued the study treatment prematurely; the most common reason was disease progression (affecting 48% of patients in the control group, 42% in the bevacizumab-initiation group, and 26% in the bevacizumab-throughout group). Treatment was discontinued owing to 2476
adverse events in a higher percentage of patients in the bevacizumab-initiation group (15%) and the bevacizumab-throughout group (17%) than in the control group (12%). Overall, 76% of adverse events leading to treatment discontinuation occurred during the chemotherapy phase. Efficacy
At the time of the primary analysis, 76.3% of patients were alive, with a median of 17.4 months of follow-up. The median progression-free survival was 10.3, 11.2, and 14.1 months in the control group, the bevacizumab-initiation group, and the bevacizumab-throughout group, respectively (Fig. 2A). As compared with the control group, the hazard of progression or death was lower (albeit not significantly) in the bevacizumab-initiation group (hazard ratio, 0.908; 95% confidence interval [CI], 0.795 to 1.040; P = 0.16) and signifi-
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Bevacizumab in Ovarian Cancer
Table 1. Baseline Characteristics of the Patients, According to Treatment Group.* Bevacizumab Initiation (N = 625)
Bevacizumab Throughout (N = 623)
Control (N = 625)
Median
60
60
60
Range
24–88
22–89
25–86
Characteristic Age — yr
Race or ethnic group — no. (%)† Non-Hispanic white Asian
519 (83.0)
521 (83.6)
526 (84.2)
37 (5.9)
39 (6.3)
41 (6.6) 25 (4.0)
Non-Hispanic black
28 (4.5)
27 (4.3)
Hispanic
28 (4.5)
25 (4.0)
21 (3.4)
Other or unspecified
13 (2.1)
11 (1.8)
12 (1.9)
0
315 (50.4)
305 (49.0)
311 (49.8)
1
270 (43.2)
267 (42.9)
272 (43.5)
2
40 (6.4)
51 (8.2)
42 (6.7)
III (macroscopic, ≤1 cm)
205 (32.8)
216 (34.7)
218 (34.9)
III (>1 cm)
256 (41.0)
242 (38.8)
254 (40.6)
IV
164 (26.2)
165 (26.5)
153 (24.5)
519 (83.0)
524 (84.1)
541 (86.6)
Endometrioid
14 (2.2)
24 (3.9)
21 (3.4)
Clear cell
23 (3.7)
20 (3.2)
12 (1.9)
Mucinous
5 (0.8)
8 (1.3)
6 (1.0)
64 (10.2)
47 (7.5)
45 (7.2)
3
465 (74.4)
460 (73.8)
445 (71.2)
2
86 (13.8)
97 (15.6)
102 (16.3)
GOG performance status — no. (%)‡
Stage/debulking status — no. (%)
Histologic type — no. (%)§ Serous adenocarcinoma
Other or not specified Tumor grade — no. (%)§
1
28 (4.5)
18 (2.9)
36 (5.8)
Not graded
46 (7.4)
48 (7.7)
42 (6.7)
* Percentages may not sum to 100 because of rounding. † Race or ethnic group was self-reported. ‡ A Gynecologic Oncology Group (GOG) performance status score of 0 indicates that the patient is fully active, 1 that the patient is restricted in physically strenuous activities but ambulatory, and 2 that the patient is ambulatory and capable of self-care but unable to work. § Histologic type and tumor grade were obtained from the central GOG Pathology Committee review updated in September 2010. All clear-cell tumors were classified as grade 3.
cantly lower in the bevacizumab-throughout group (hazard ratio, 0.717; 95% CI, 0.625 to 0.824; P<0.001). The maximal separation of the progression-free survival curves for the bevacizumabthroughout group and the control group occurred at 15 months, with convergence approximately 9 months later. In an analysis of progression-free survival in which data for patients with increased CA-125 levels were censored, as required by regu-
latory agencies, the median progression-free survival was 12.0 months in the control group but 18.0 months in the bevacizumab-throughout group (hazard ratio, 0.645; 95% CI, 0.551 to 0.756; P<0.001) (see the Supplementary Appendix). The estimated treatment effect on progression-free survival with bevacizumab throughout as compared with control treatment was consistent across various prognostic factors (Fig. 2C).
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2477
The
n e w e ng l a n d j o u r na l
The median overall survival was 39.3, 38.7, and 39.7 months for the control group, the bevaciz umab-initiation group, and the bevacizumabthroughout group, respectively (Fig. 3). As compared with the control group, the hazard of death was 1.036 (95% CI, 0.827 to 1.297; P = 0.76) in the bevacizumab-initiation group and 0.915 (95% CI, 0.727 to 1.152; P = 0.45) in the bevacizumabthroughout group. Results of updated analyses of progressionfree survival (Fig. 2B) and overall survival (Fig. 3B), performed on the data as of August 26, 2011, after 47% of the patients had died, were consistent with those from the original analyses. Quality of Life
Valid quality-of-life surveys were available for 93.2%, 88.3%, 85.8%, 81.1%, 75.7%, and 74.1% of patients alive before cycles 1, 4, 7, 13, and 22 and 6 months after completing the study therapy, respectively. There were no significant differences across the three treatment groups. The mean FACT-O TOI scores exceeded 65 at each time point and generally increased over the duration of the study, reaching more than 75 by 6 months after the completion of chemotherapy. During the chemotherapy phase, the mean FACT-O TOI scores were slightly lower in the bevacizumab-initiation group and the bevacizumab-throughout group than in the control group, especially before cycle 4 (with a reduction of 2.7 points [98.3% CI, 0.88 to 4.57; P<0.001] and 3.0 points [98.3% CI, 1.13 to 4.78; P<0.001], respectively). No significant differences were found in the mean FACT-O TOI scores between the control group and the bevacizumabthroughout group at any of the three time points after completion of chemotherapy.
of
m e dic i n e
Figure 2 (facing page). Primary and Subgroup Analyses of Progression-free Survival, According to Treatment Group. Panel A shows the results of primary analysis of progression-free survival for all 1873 patients randomly assigned to receive chemotherapy with carboplatin and paclitaxel (CP) plus placebo followed by placebo alone (the control group), CP plus bevacizumab (bev) followed by placebo (the bevacizumab-initiation group), or CP plus bevacizumab followed by bevacizumab (the bevacizu mab-throughout group). There was a significant, timedependent decrease in the hazard of progression in the bevacizumab-throughout group as compared with the control group (hazard ratio, 0.717; 95% confidence interval [CI], 0.625 to 0.824; P<0.001). Panel B shows the results of an updated analysis of data on progressionfree survival as of August 26, 2011. The hazard of progression remained significantly decreased with bevaciz umab-throughout versus control therapy (hazard ratio, 0.770; 95% CI, 0.681 to 0.870). Panel C shows the effect of treatment with bevacizumab (vs. control) on progression-free survival, stratified according to multiple prognostic factors. The effect was significant and consistent across all strata for bevacizumab throughout (vs. control). GOG denotes Gynecologic Oncology Group.
cally relevant bleeding or central nervous system complications, were rare. Fatal adverse events were reported in 6 of 601 patients (1.0%) in the control group, in 10 of 607 patients (1.6%) in the bevacizumab-initiation group, and in 14 of 608 patients (2.3%) in the bevacizumab-throughout group. Most adverse events were reported during the chemotherapy phase rather than the extendedtherapy phase (see the Supplementary Appendix). For example, in each of the three groups, all but one gastrointestinal perforation or fistula occurred during receipt of chemotherapy. Exceptions were hypertension, proteinuria, and pain, which were Safety more commonly reported during the extendedTable 2 shows the frequency of adverse events therapy phase than the chemotherapy phase potentially associated with bevacizumab (on the among patients in the bevacizumab-throughout basis of prior trials). Hypertension of grade 2 or group. greater was significantly (P<0.001) more common with bevacizumab than placebo but led to Discussion discontinuation of bevacizumab in only 15 of the 608 patients (2.4%) in the bevacizumab-through- This study showed a significant improvement in out group. There were no significant differences progression-free survival (i.e., an increase in meamong the three groups in the rates of other ad- dian progression-free survival by 4 months) with verse events, including gastrointestinal perfora- bevacizumab plus chemotherapy (with carboplattion or fistula, proteinuria of grade 3 or greater, in and paclitaxel) followed by extended bevacineutropenia of grade 4 or greater or febrile neu- zumab therapy, as compared with chemotherapy tropenia, venous or arterial thrombosis, and wound alone, for advanced ovarian cancer. The effect was disruption. Other adverse events, such as clini- seen consistently across prognostic subgroups. 2478
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Bevacizumab in Ovarian Cancer
B Analysis as of August 26, 2011
CP + bev Bev or or placebo placebo 1.0 Control 0.9 Bev initiation 0.8 Bev throughout 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36
No. at Risk Control 625 Bev initiation 625 Bev through- 623 out
Months since Randomization 199 219 254
Proportion with Progressionfree Survival
Proportion with Progressionfree Survival
A Primary Analysis
CP + bev Bev or or placebo placebo 1.0 Control 0.9 Bev initiation 0.8 Bev throughout 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36
Months since Randomization
No. at Risk
33 29 38
8 6 8
Control 625 Bev initiation 625 Bev through- 623 out
535 552 559
283 319 386
169 190 256
133 121 162
78 67 97
49 40 56
C Total No. of Patients
Risk Factor Cancer stage and residual lesion size III, macroscopic ≤1 cm Bev initiation vs. control Bev throughout vs. control III, >1 cm Bev initiation vs. control Bev throughout vs. control IV Bev initiation vs. control Bev throughout vs. control Histologic type Serous Bev initiation vs. control Bev throughout vs. control Nonserous Bev initiation vs. control Bev throughout vs. control Tumor grade 1 or 2 Bev initiation vs. control Bev throughout vs. control 3 Bev initiation vs. control Bev throughout vs. control GOG performance status score 0 Bev initiation vs. control Bev throughout vs. control 1 or 2 Bev initiation vs. control Bev throughout vs. control Age <60 yr Bev initiation vs. control Bev throughout vs. control 60–69 yr Bev initiation vs. control Bev throughout vs. control ≥70 yr Bev initiation vs. control Bev throughout vs. control
Hazard Ratio for Bev (95% CI)
423 434
0.780 0.618
510 496
0.981 0.763
317 318
0.923 0.698
1066 1068
0.913 0.701
184 180
0.893 0.713
232 235
1.039 0.578
847 842
0.891 0.700
626 616
0.877 0.710
624 632
0.961 0.690
616 630
0.976 0.680
414 408
0.892 0.763
220 210
0.841 0.678 0.33
0.50 0.67
1.00
Bev Better
1.50 2.00
3.00
Control Better
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2479
n e w e ng l a n d j o u r na l
The
Proportion Surviving
A Analysis at Time of Primary Analysis 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0
No. of Events (%)
Control
156 (25.0)
Bevacizumab initiation
150 (24.0)
Bevacizumab throughout 138 (22.2) 0
6
12
18
24
30
36
42
48
42
48
Months since Randomization No. at Risk Control Bevacizumab initiation Bevacizumab throughout
625 625
442 432
173 162
46 39
623
437
171
40
Proportion Surviving
B Analysis as of August 26, 2011 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0
No. of Events (%)
Control
298 (47.7)
Bevacizumab initiation
308 (49.3)
Bevacizumab throughout 269 (43.2) 0
6
12
18
24
30
36
Months since Randomization No. at Risk Control Bevacizumab initiation Bevacizumab throughout
625 625
595 598
558 557
506 486
446 440
322 304
200 191
116 108
56 54
623
587
561
519
463
321
201
114
62
Figure 3. Analyses of Overall Survival, According to Treatment Group. RETAKE: 1st AUTHOR: Burger of the overall-survival analysis Panel A shows the results performed at the 2nd time of the FIGURE: primary 3analysis of progression-free survival, when3rd 76.3% of paof 3 tients were alive. As compared with patients receivingRevised control therapy — ARTIST: chemotherapy with ts carboplatin and paclitaxel plus placebo SIZEfollowed by pla4interval col cebo — the hazard death was 1.036 4-C (95% confidence [CI], 0.827 Combo H/T TYPE: of Line 22p3 to 1.29; P = 0.76) among patients receiving carboplatin and paclitaxel plus AUTHOR,(the PLEASE NOTE: bevacizumab followed by placebo bevacizumab-initiation group) and Figure has been redrawn and type has been reset. 0.915 (95% CI, 0.727 to 1.15;Please P = 0.45) among patients receiving carboplatin check carefully. and paclitaxel plus bevacizumab followed by bevacizumab (the bevacizumab36526Panel B shows the results of an updated ISSUE: 12-29-11 throughoutJOB: group). analysis of data on overall survival as of August 26, 2011. At this time point, as compared with the control group, the hazard of death was 1.078 (95% CI, 0.919 to 1.270) in the bevacizumab-initiation group and 0.885 (95% CI, 0.750 to 1.040) in the bevacizumab-throughout group.
2480
of
m e dic i n e
This finding supports results from earlier laboratory and epidemiologic studies, indicating that VEGF promotes progression in ovarian cancer2-13 and that VEGF blockade inhibits tumor growth, metastasis, and malignant ascites formation.2-5 The 28% reduction in the risk of progression with bevacizumab throughout as compared with placebo is clinically important. Bevacizumabrelated toxic effects after chemotherapy were similar to those seen in other tumor types, with no corresponding reduction in the quality of life. Since detection methods for ovarian cancer are highly sensitive, affected women tend to be asymptomatic at the time of initial disease progression. Therefore, we could not determine whether the delay in tumor progression per se delayed physical or psychological symptoms associated with disease or subsequent therapy. No significant improvement in overall survival was shown; however, the potential to detect a difference in survival is likely to be limited by lack of control for multiple subsequent regimens, including crossover to bevacizumab or other antiVEGF agents. Study therapy was discontinued because of adverse events in 17% of patients in the bevacizu mab-throughout group versus 12% in the control group. The 5% difference between the groups may be overestimated, since more patients in the control group than in the bevacizumab-throughout group discontinued study treatment because of disease progression, after which adverse event reporting ended. Rates of gastrointestinal perforation and fistula in the two bevacizumab groups were almost twice those in the control group but were still less than 3%, consistent with rates seen in metastatic nongynecologic tumors. This is an important finding, given previous concerns about an excessive risk of gastrointestinal perforation in patients with recurrent ovarian cancer.26 As expected, hypertension of grade 2 or higher was significantly more common with bevacizumab than without it. Although the risk of hypertension appeared to be cumulative, it was controlled with the use of medical therapy, with few patients discontinuing bevacizumab. The risk of proteinuria of grade 3 or higher also appeared to be cumulative, but proteinuria developed in less than 2% of patients in the bevacizumab-throughout group. In contrast to a
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Bevacizumab in Ovarian Cancer
Table 2. Selected Adverse Events among the Study Patients, According to Treatment Group.* Bevacizumab Initiation (N = 607)
Event
Bevacizumab Throughout (N = 608)
Control (N = 601)
number of patients (percent) Gastrointestinal events (grade ≥2)†
17 (2.8)
16 (2.6)
7 (1.2)
100 (16.5)§
139 (22.9)§
43 (7.2)
4 (0.7)
10 (1.6)
4 (0.7)
Pain (grade ≥2)
252 (41.5)
286 (47.0)
250 (41.6)
Neutropenia (grade ≥4)
384 (63.3)
385 (63.3)
347 (57.7)
Febrile neutropenia
30 (4.9)
26 (4.3)
21 (3.5)
Venous thromboembolism
32 (5.3)
41 (6.7)
35 (5.8)
Arterial thromboembolism
4 (0.7)
4 (0.7)
5 (0.8)
22 (3.6)
18 (3.0)
17 (2.8)
Hypertension (grade ≥2)‡ Proteinuria (grade ≥3)
Wound disruption CNS bleeding
0
2 (0.3)
Non-CNS bleeding (grade ≥3)
8 (1.3)
13 (2.1)
Reversible posterior leukoencephalopathy syndrome
1 (0.2)
1 (0.2)
0 5 (0.8) 0
* Adverse events were those with onset between cycle 2 and 30 days after the date of the last treatment. CNS denotes central nervous system. † Gastrointestinal events of grade 2 or greater were gastrointestinal-wall disruption: perforation, fistula, necrosis, or anastomotic leak. ‡ Hypertension of grade 2 or greater consisted of recurrent or continuous hypertension for a period of more than 24 hours or symptomatic increase in blood pressure by more than 20 mm Hg (diastolic) or to over 150/100 mm Hg if the blood pressure was previously within the normal range. § P<0.05 for the comparison with the control group.
pooled analysis of phase 3 trials of nongynecologic cancers,27 we observed no significant increase in the incidence of arterial thrombotic events with bevacizumab. Slightly higher, although not significantly higher, rates of grade 4 or 5 neutropenia and febrile neutropenia were seen in the two bevacizumab groups than in the control group. The lack of a significant difference in progression-free survival between the control group and the bevacizumab-initiation group implies that bevacizumab must be continued beyond chemotherapy to delay disease progression. The rationale for combining cytotoxic and anti-VEGF therapy arose from preclinical studies showing a transient reduction in tumor microvascular permeability and interstitial pressure,28,29 with a theoretical increase in tumor perfusion, and therefore enhanced chemosensitivity.30,31 It is impossible to determine whether such a mechanism operated in this study, since a regimen of carboplatin and paclitaxel plus
placebo followed by bevacizumab was not evaluated. Therefore, we cannot rule out the possibility that bevacizumab exposure during the chemotherapy phase of the bevacizumab-throughout group contributed to the significant improvement in progression-free survival. Although bevacizumab use resulted in additional toxic effects, it was not associated with a decline in quality-of-life scores,32 and after chemotherapy completion, no significant differences in quality-of-life scores were observed across the three treatment groups. Trials of other antiangiogenic agents, with pure maintenance designs, are ongoing. For the bevacizumab-throughout group, the maximal treatment time (approximately 15 months) was selected to exceed the median expected progression-free survival for the population yet to ensure study feasibility. Although the “tail ends” of the progression-free survival curves may be relatively unreliable for patients in the bevacizumab-throughout group and the control
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2481
The
n e w e ng l a n d j o u r na l
group, with less than 3% of patients at risk for progression, a convergence of the curves was observed nonetheless. Convergence of the progression-free survival curves was also observed in an independent, positive, front-line, open-label, phase 3 ovarian cancer trial of the International Collaborative Ovarian Neoplasm (ICON) group known as ICON7 (ClinicalTrials.gov number, NCT00483782),33 in which bevacizumab use was limited to 12 months. In our bevacizumabthroughout group and the ICON7 bevacizumab group, 24% and 62% of patients, respectively, completed all study therapy without disease progression. In our bevacizumab-throughout group, an additional 19% of patients were still receiving bevacizumab at the time of the database lock. In contrast to our findings, progression-free survival curves did not converge in the placebocontrolled phase 3 Ovarian Cancer Study Comparing Efficacy and Safety of Chemotherapy and Anti-Angiogenic Therapy in Platinum-Sensitive Recurrent Disease (OCEANS, NCT00434642).34 OCEANS showed a hazard ratio for progressionfree survival of 0.484 favoring chemotherapy with bevacizumab followed by continued beva cizumab over chemotherapy with placebo followed by placebo. Unlike the front-line trials, in OCEANS, bevacizumab was continued until disease progression. Though cross-trial comparisons have clear caveats, these results suggest that the magnitude of benefit may correlate directly with treatment duration. This hypothesis is consistent with results of preclinical studies in which anti-VEGF therapy delayed tumor growth and extended survival in a variety of established tumor models and prevented regrowth of a subgroup of residual tumors after cytotoxic therapy, whereas discontinuation of anti-VEGF therapy resulted in regrowth.35 This is not unexpected, since angiogenesis is a host-related process that can be inhibited but not eradicated. The bevacizumab regimen of 15 mg per kilogram every 3 weeks in this study was based on the regimen approved in combination with carbo platin and paclitaxel for advanced non–smallcell lung cancer36 and single-agent activity shown
of
m e dic i n e
in two phase 2 trials in ovarian cancer.15,16 OCEANS used the same dose and schedule.34 ICON7 used a bevacizumab dose of 7.5 mg per kilogram,33 albeit with a smaller magnitude of benefit in a broader patient population, without obvious differences in adverse events from those in our study population. Together, these independent phase 3 trials of ovarian cancer show a benefit of bevacizumab in the dose range of 7.5 to 15 mg per kilogram. A major limitation of this study was the change of the primary end point from overall survival to progression-free survival. This change was made because maintaining the blinding of the treatment assignments after disease progression, which was required to protect the integrity of the data on overall survival, was contested by numerous investigators and patients and therefore was deemed infeasible. A primary end point of progressionfree survival is supported by the Gynecologic Cancer Intergroup,37 which noted that in trials assessing front-line therapy for advanced ovarian cancer, including those involving maintenance therapy, progression-free survival is most often the preferred primary end point, because of the confounding effect of post-recurrence or post-progression therapy on overall survival. When considering the balance of clinical benefit for progression-free survival, quality-of-life preservation, and tolerability, our study shows that bevacizumab plus carboplatin and paclitaxel, followed by bevacizumab, could be considered a front-line treatment option for patients with advanced ovarian cancer. Further investigation is needed to optimize duration and timing of treatment, assess integration into or use after other standard front-line strategies (e.g., neoadjuvant or intraperitoneal chemotherapy), examine cost-effectiveness, and, perhaps most important, identify potential tumor or host biologic factors predictive of efficacy and adverse events with the ultimate goal of decreasing morbidity and mortality from this disease. Supported by the National Cancer Institute and Genentech. Disclosure forms provided by the authors are available with the full text of this article at NEJM.org.
References 1. Siegel R, Ward E, Brawley O, Jemal A.
Cancer statistics, 2011: the impact of eliminating socioeconomic and racial disparities on premature cancer deaths. CA Cancer J Clin 2011;61:212-36.
2482
2. Xu L, Yoneda J, Herrera C, Wood J,
Killion JJ, Fidler IJ. Inhibition of malignant ascites and growth of human ovarian carcinoma by oral administration of a potent inhibitor of the vascular endothelial growth
factor receptor tyrosine kinases. Int J Oncol 2000;16:445-54. 3. Mesiano S, Ferrara N, Jaffe RB. Role of vascular endothelial growth factor in ovarian cancer: inhibition of ascites for-
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Bevacizumab in Ovarian Cancer mation by immunoneutralization. Am J Pathol 1998;153:1249-56. 4. Byrne AT, Ross L, Holash J, et al. Vascular endothelial growth factor-trap decreases tumor burden, inhibits ascites, and causes dramatic vascular remodeling in an ovarian cancer model. Clin Cancer Res 2003;9:5721-8. 5. Manenti L, Riccardi E, Marchini S, et al. Circulating plasma vascular endothelial growth factor in mice bearing human ovarian carcinoma xenograft correlates with tumor progression and response to therapy. Mol Cancer Ther 2005;4:715-25. 6. Belotti D, Calcagno C, Garofalo A, et al. Vascular endothelial growth factor stimulates organ-specific host matrix metalloproteinase-9 expression and ovarian cancer invasion. Mol Cancer Res 2008;6: 525-34. 7. Goodheart MJ, Ritchie JM, Rose SL, Fruehauf JP, De Young BR, Buller RE. The relationship of molecular markers of p53 function and angiogenesis to prognosis of stage I epithelial ovarian cancer. Clin Cancer Res 2005;11:3733-42. 8. Hollingsworth HC, Kohn EC, Steinberg SM, Rothenberg ML, Merino MJ. Tumor angiogenesis in advanced stage ovarian carcinoma. Am J Pathol 1995;147: 33-41. 9. Paley PJ, Staskus KA, Gebhard K, et al. Vascular endothelial growth factor expression in early stage ovarian carcinoma. Cancer 1997;80:98-106. 10. Gasparini G, Bonoldi E, Viale G, et al. Prognostic and predictive value of tumour angiogenesis in ovarian carcinomas. Int J Cancer 1996;69:205-11. 11. Alvarez AA, Krigman HR, Whitaker RS, Dodge RK, Rodriguez GC. The prognostic significance of angiogenesis in epithelial ovarian carcinoma. Clin Cancer Res 1999;5:587-91. 12. Shen GH, Ghazizadeh M, Kawanami O, et al. Prognostic significance of vascular endothelial growth factor expression in human ovarian carcinoma. Br J Cancer 2000;83:196-203. 13. Duncan TJ, Al-Attar A, Rolland P, et al. Vascular endothelial growth factor expression in ovarian cancer: a model for targeted use of novel therapies? Clin Cancer Res 2008;14:3030-5. 14. Presta LG, Chen H, O’Connor SJ, et al. Humanization of an anti-vascular endothelial growth factor monoclonal antibody for the therapy of solid tumors and other disorders. Cancer Res 1997;57:45939. 15. Burger RA, Sill MW, Monk BJ, Greer BE, Sorosky JI. Phase II trial of bevaciz
umab in persistent or recurrent epithelial ovarian cancer or primary peritoneal cancer: a Gynecologic Oncology Group Study. J Clin Oncol 2007;25:5165-71. 16. Cannistra SA, Matulonis UA, Penson RT, et al. Phase II study of bevacizumab in patients with platinum-resistant ovarian cancer or peritoneal serous cancer. J Clin Oncol 2007;25:5180-6. [Erratum, J Clin Oncol 2008;26:1773.] 17. Bast RC Jr, Feeney M, Lazarus H, Nadler LM, Colvin RB, Knapp RC. Reactivity of a monoclonal antibody with human ovarian carcinoma. J Clin Invest 1981;68:1331-7. 18. Basen-Engquist K, Bodurka-Bevers D, Fitzgerald MA, et al. Reliability and validity of the functional assessment of cancer therapy-ovarian. J Clin Oncol 2001;19: 1809-17. 19. Pocock SJ, Simon R. Sequential treatment assignment with balancing for prognostic factors in the controlled clinical trial. Biometrics 1975;31:103-15. 20. Therasse P, Arbuck SG, Eisenhauer EA, et al. New guidelines to evaluate the response to treatment in solid tumors: European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada. J Natl Cancer Inst 2000;92:205-16. 21. Rustin GJ, Marples M, Nelstrop AE, Mahmoudi M, Meyer T. Use of CA-125 to define progression of ovarian cancer in patients with persistently elevated levels. J Clin Oncol 2001;19:4054-7. 22. Mantel N. Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemother Rep 1966;50:163-70. 23. Cox DR. Regression models and lifetables. J R Stat Soc [B] 1972;34:187-220. 24. Common Terminology Criteria for Adverse Events (CTCAE), v3.0. Bethesda, MD: Cancer Therapy Evaluation Pro gram, 2006 (http://ctep.cancer.gov/protocol Development/electronic_applications/ctc .htm#ctc_30). 25. Mehta C, Petal N. A network algorithm for performing Fisher’s exact test in r × c contingency tables. J Am Stat Assoc 1983;78:427-34. 26. Han ES, Monk BJ. What is the risk of bowel perforation associated with beva cizumab therapy in ovarian cancer? Gynecol Oncol 2007;105:3-6. 27. Geiger-Gritsch S, Stollenwerk B, Miksad R, Guba B, Wild C, Siebert U. Safety of bevacizumab in patients with advanced cancer: a meta-analysis of randomized controlled trials. Oncologist 2010;
15:1179-91. [Erratum, Oncologist 2010; 15:1373.] 28. Gerber HP, Ferrara N. Pharmacology and pharmacodynamics of bevacizumab as monotherapy or in combination with cytotoxic therapy in preclinical studies. Cancer Res 2005;65:671-80. 29. Willett CG, Boucher Y, di Tomaso E, et al. Direct evidence that the VEGF-specific antibody bevacizumab has antivascular effects in human rectal cancer. Nat Med 2004;10:145-7. [Erratum, Nat Med 2004;10:649.] 30. Jain RK. Normalization of tumor vasculature: an emerging concept in antiangiogenic therapy. Science 2005;307:58-62. 31. Kerbel RS. Antiangiogenic therapy: a universal chemosensitization strategy for cancer? Science 2006;312:1171-5. 32. Yost KJ, Eton DT. Combining distribution- and anchor-based approaches to determine minimally important differences: the FACIT experience. Eval Health Prof 2005;28:172-91. 33. Perren TS, Pfisterer J, Ledermann J, et al. ICON7: a phase III Gynaecologic Cancer InterGroup (GCIG) trial of adding bevacizumab to standard chemotherapy in women with newly diagnosed epithelial ovarian, primary peritoneal or fallopian tube cancer. In: Program and abstracts of the 35th Annual Meeting of the European Society of Medical Oncology, Milan, October 8–12, 2010. abstract. 34. Aghajanian CAF, Rutherford T, Smith DA, et al. OCEANS: a randomized, double-blinded, placebo-controlled phase III trial of chemotherapy with or without bevacizumab (BEV) in patients with platinum-sensitive recurrent epithelial ovarian (EOC), primary peritoneal (PPC), or fallopian tube cancer (FTC). In: Program and abstracts of the 2011 American Society of Clinical Oncology Annual Meeting, Chicago, June 4–8, 2011. abstract. 35. Bagri A, Berry L, Gunter B, et al. Effects of anti-VEGF treatment duration on tumor growth, tumor regrowth, and treatment efficacy. Clin Cancer Res 2010;16: 3887-900. 36. Avastin (bevacizumab): overview. South San Francisco, CA: Genentech (http:// www.avastin.com/avastin/hcp/overview/ index.html). 37. Stuart GC, Kitchener H, Bacon M, et al. 2010 Gynecologic Cancer InterGroup (GCIG) consensus statement on clinical trials in ovarian cancer: report from the Fourth Ovarian Cancer Consensus Conference. Int J Gynecol Cancer 2011;21: 750-5. Copyright © 2011 Massachusetts Medical Society.
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2483
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
A Phase 3 Trial of Bevacizumab in Ovarian Cancer Timothy J. Perren, M.D., Ann Marie Swart, M.D., Jacobus Pfisterer, M.D., Jonathan A. Ledermann, M.D., Eric Pujade-Lauraine, M.D., Gunnar Kristensen, M.D., Mark S. Carey, M.D., Philip Beale, M.D., Andrés Cervantes, M.D., Christian Kurzeder, M.D., Andreas du Bois, M.D., Jalid Sehouli, M.D., Rainer Kimmig, M.D., Anne Stähle, M.D., Fiona Collinson, M.D., Sharadah Essapen, M.D., Charlie Gourley, M.D., Alain Lortholary, M.D., Frédéric Selle, M.D., Mansoor R. Mirza, M.D., Arto Leminen, M.D., Marie Plante, M.D., Dan Stark, M.D., Wendi Qian, Ph.D., Mahesh K.B. Parmar, Ph.D., and Amit M. Oza, M.D., for the ICON7 Investigators*
A BS T R AC T Background The authors’ affiliations are listed in the Appendix. Address reprint requests to Dr. Parmar at MRC CTU, Aviation House, 125 Kingsway, London WL2B 6NH, United Kingdom, or at
[email protected]. * Additional investigators participating in the International Collaboration on Ovarian Neoplasms (ICON7) are listed in the Supplementary Appendix, available at NEJM.org. This article (10.1056/NEJMoa1103799) was updated on December 29, 2011, at NEJM .org. N Engl J Med 2011;365:2484-96. Copyright © 2011 Massachusetts Medical Society.
Angiogenesis plays a role in the biology of ovarian cancer. We examined the effect of bevacizumab, the vascular endothelial growth factor inhibitor, on survival in women with this disease. Methods
We randomly assigned women with ovarian cancer to carboplatin (area under the curve, 5 or 6) and paclitaxel (175 mg per square meter of body-surface area), given every 3 weeks for 6 cycles, or to this regimen plus bevacizumab (7.5 mg per kilogram of body weight), given concurrently every 3 weeks for 5 or 6 cycles and continued for 12 additional cycles or until progression of disease. Outcome measures included progressionfree survival, first analyzed per protocol and then updated, and interim overall survival. Results
A total of 1528 women from 11 countries were randomly assigned to one of the two treatment regimens. Their median age was 57 years; 90% had epithelial ovarian cancer, 69% had a serous histologic type, 9% had high-risk early-stage disease, 30% were at high risk for progression, and 70% had stage IIIC or IV ovarian cancer. Progressionfree survival (restricted mean) at 36 months was 20.3 months with standard therapy, as compared with 21.8 months with standard therapy plus bevacizumab (hazard ratio for progression or death with bevacizumab added, 0.81; 95% confidence interval, 0.70 to 0.94; P = 0.004 by the log-rank test). Nonproportional hazards were detected (i.e., the treatment effect was not consistent over time on the hazard function scale) (P<0.001), with a maximum effect at 12 months, coinciding with the end of planned bevacizumab treatment and diminishing by 24 months. Bevacizumab was associated with more toxic effects (most often hypertension of grade 2 or higher) (18%, vs. 2% with chemotherapy alone). In the updated analyses, progression-free survival (restricted mean) at 42 months was 22.4 months without bevacizumab versus 24.1 months with bevacizumab (P = 0.04 by log-rank test); in patients at high risk for progression, the benefit was greater with bevacizumab than without it, with progression-free survival (restricted mean) at 42 months of 14.5 months with standard therapy alone and 18.1 months with bevacizumab added, with respective median overall survival of 28.8 and 36.6 months. Conclusions
Bevacizumab improved progression-free survival in women with ovarian cancer. The benefits with respect to both progression-free and overall survival were greater among those at high risk for disease progression. (Funded by Roche and others; ICON7 Controlled-Trials.com number, ISRCTN91273375.) 2484
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Bevacizumab in Ovarian Cancer
O
varian cancer is the fourth most common cause of cancer-related deaths in women, with an estimated 200,000 cases and 125,000 deaths occurring annually worldwide. For the past decade, the standard treatment for women with advanced ovarian cancer has been surgery and platinum-based chemotherapy. Attempts to improve this standard two-drug chemotherapy by adding a third cytotoxic drug failed to affect either progression-free survival or overall survival and resulted in an increase in toxic effects.1-4 Although intraperitoneal chemotherapy has extended overall survival by 12 to 17 months, it is an option only for women with advanced ovarian cancer who have a small amount of residual disease after surgery5-7 and is not widely used. Since the results of one trial suggested that weekly paclitaxel administration provided some benefit,8 trials are now under way to optimize carboplatin and paclitaxel scheduling. Angiogenesis contributes to solid-tumor growth and metastasis.9,10 Epithelial ovarian-cancer cell lines frequently express vascular endothelial growth factor (VEGF).11 Decreased VEGF expression is associated with reductions in tumor vascularization and angiogenesis and with prolonged survival.12 Bevacizumab (Avastin, Roche) is a monoclonal antibody that binds to all isoforms of the VEGF-receptor ligand VEGF-A, with evidence of efficacy in metastatic colorectal and lung cancers, as well as activity in renal, breast, and brain cancers.13-15 Phase 2 trials of bevacizumab in women with ovarian cancer have shown tumor responses and delayed disease progression.16-18 The Gynecologic Cancer InterGroup (GCIG) International Collaboration on Ovarian Neoplasms (ICON7) trial and the complementary Gynecologic Oncology Group study 0218 (GOG-0218) (ClinicalTrials.gov number, NCT00262847) were designed to investigate the addition of bevaciz umab to standard chemotherapy in the first-line treatment of ovarian cancer. The results of the GOG-0218 study are reported elsewhere in this issue of the Journal by Burger et al.19
tatives from GCIG and F. Hoffmann–La Roche. An independent international trial steering committee and an independent data and safety monitoring committee provided oversight. Depending on national requirements, approval by ethics committees was obtained at each clinical site, nationally, or both. Data management was conducted by independent contract research organizations and MRC CTU (see the Supplementary Appendix, available with the full text of this article at NEJM.org). Data analysis and interpretation and preparation of the manuscript were performed independently by the funders and the sponsor (MRC). All authors had full access to all study data, had responsibility for the decision to submit the manuscript for publication, and vouch for the accuracy, integrity, and completeness of the data as reported. The trial was conducted in accordance with the protocol. (The protocol, including the statistical analysis plan, is available at NEJM.org.)
Me thods
Randomization was performed centrally by an interactive telephone or Web-based system, with stratification according to GCIG group, FIGO stage and residual disease (i.e., FIGO stages I to III and ≤1 cm of residual disease, stages I to III and >1 cm of residual disease, or stage III [inoperable] or IV), and planned interval between surgery and initiation of chemotherapy (≤4 weeks or >4 weeks).
Eligibility Requirements
After surgery, women were eligible for enrollment if they had histologically confirmed, highrisk, early-stage disease (International Federation of Gynecology and Obstetrics [FIGO] stage I or IIA and clear-call or grade 3 tumors) or advanced (FIGO stage IIB to IV) epithelial ovarian cancer, primary peritoneal cancer, or fallopian-tube cancer (based on local histopathological findings). Additional eligibility criteria were an Eastern Cooperative Oncology Group (ECOG) performance status of 0 to 2 (on a scale from 0 to 4, with 0 indicating normal activity, 1 symptomatic but ambulatory self-care possible, 2 ambulatory more than 50% of the time, 3 ambulatory 50% of the time or less and nursing care required, and 4 bedridden and possibly requiring hospitalization) and adequate coagulation values and bone marrow, liver, and renal function, with no plans for further surgery before disease progression. All the patients provided written informed consent. Randomization and Treatments
Study Design and Conduct
ICON7 is led by the U.K. Medical Research Council Clinical Trials Unit (MRC CTU) and was designed by members of the Trial Management Group, who reviewed and approved the protocol. The Trial Management Group included represen-
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2485
The
n e w e ng l a n d j o u r na l
Enrollment of women with high-risk early-stage disease was restricted to 10% of the total study population. Women were randomly assigned in a 1:1 ratio to receive carboplatin (area under the curve [AUC], 5 or 6) and paclitaxel (175 mg per square meter of body-surface area), given every 3 weeks for 6 cycles (standard-chemotherapy group), or to the same regimen plus bevacizumab (7.5 mg per kilogram of body weight), given concurrently every 3 weeks for 5 or 6 cycles and continued for 12 additional cycles or until disease progression (bevacizumab group). Bevacizumab was omitted at cycle 1 to avoid delayed wound healing if chemotherapy was started within 4 weeks of surgery. Cycles of bevacizumab that were omitted for any reason were not replaced. Assessments
Assessments were performed at the same time points in the two treatment groups. Clinical assessments and cancer antigen 125 (CA-125) measurements were performed before each cycle of chemotherapy, then every 6 weeks in year 1, every 3 months during years 2 and 3, every 6 months during years 4 and 5, and then yearly. After disease progression, follow-up assessments were performed every 6 months for the first 5 years, then yearly. Computed tomography (CT) or magnetic resonance imaging (MRI) was performed at baseline, after chemotherapy cycles 3 and 6, at 9 and 12 months after randomization, every 6 months in years 2 and 3, and then as clinically indicated until disease progression. All tumor assessments were reviewed by the principal investigator, who was unaware of treatment assignments. Quality of life was assessed with the use of the European Organization for Research and Treatment of Cancer (EORTC) QLQ-C30 and QLQ-OV28 questionnaires (see the Supplementary Appendix). Outcome Measures
The principal outcomes of interest were progression-free survival and overall survival. Other outcomes included biologic progression-free interval, response to therapy, toxicity, and quality of life. Progression-free survival was calculated from the date of randomization to the date of the first indication of disease progression or death, whichever occurred first; the data for patients who were alive without disease progression were cen2486
of
m e dic i n e
sored as of the date of their last assessment. Sensitivity analyses were performed in which data from patients who were alive without disease progression were censored at the date of the last imaging assessment. Disease progression was defined according to the Response Evaluation Criteria in Solid Tumors (RECIST)20 guidelines (see the Supplementary Appendix) on the basis of radiologic, clinical, or symptomatic indicators of progression and did not include isolated asymptomatic progression on the basis of CA-125 levels. The biologic progression-free interval was calculated from the date of randomization to the date of the first CA-125–based progression21 or first RECIST–based progression, whichever occurred first (see the Supplementary Appendix). Overall survival was calculated from the date of randomization to the date of death from any cause; data for patients still alive were censored at the date the patient was last known to be alive. Best overall response was assessed in patients with measurable disease at baseline who received at least one cycle of per-protocol treatment and was defined as the best confirmed response recorded from the start of treatment until 70 days after the last dose of per-protocol treatment. Statistical Analysis
The primary analysis was carried out with the use of an unstratified log-rank test for the difference in the distribution of progression-free survival between the two groups. Other planned analyses included a log-rank test that stratified for factors used for randomization (excluding GCIG groups to limit the number of categories being tested); Cox regression analyses that adjusted for baseline covariates to assess the robustness of the result if the proportional-hazards assumption held; flexible parametric survival models22 to smooth survival curves and estimate survival differences with the use of all survival data collected; and interaction analyses to explore the difference in the relative size of treatment effects in subgroups classified according to baseline characteristics, high risk for progression (i.e., FIGO stage IV disease or FIGO stage III disease and >1.0 cm of residual disease after debulking surgery), and stratification factors. All efficacy analyses for progression-free survival included all patients who had been randomly assigned to treatment (i.e., the intention-to-treat population). Safety analyses included patients who had
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Bevacizumab in Ovarian Cancer
received at least one cycle of the assigned protocol treatment. All reported P values are two-sided, and all statistical analyses were performed at MRC CTU. The magnitude and timing of treatment effect were explored with the use of hazard functions, including a formal test of the proportional-hazards assumption. With an estimated median progression-free survival of 18 months and median overall survival of 43 months in the standard-chemotherapy group, the trial was designed to detect a 28% increase in median progression-free survival to 23 months (hazard ratio, 0.78) and a 23% increase in median overall survival to 53 months (hazard ratio, 0.81) in the bevacizumab group. At a 5% significance level and with 90% power for progression-free survival and 80% power for overall survival, a total of 684 progression-free survival events (i.e., disease progression or death) and 715 overall survival events were required. With 1520 women randomly assigned to treatment over a period of 2 years, the required progression-free survival and overall survival events were expected to occur 1 and 3 years, respectively, after entry of the last patient. After submission of the primary analysis of the protocol-defined progression-free survival, regulatory authorities requested an overall survival analysis with at least 365 deaths (50% of the required total number of deaths) and an update on progression-free survival with the use of the same data set.
R e sult s Patients
From December 2006 through February 2009, 1528 women were enrolled at 263 centers in the United Kingdom, Germany, France, Canada, Australia, New Zealand, Denmark, Finland, Norway, Sweden, and Spain. The cutoff date for the primary analysis was February 28, 2010, by which time a total of 759 progression-free survival events had been observed. The treatment groups were well balanced with respect to baseline characteristics. The median age was 57 years, and 94% of the patients had an ECOG performance status of 0 or 1; 90% had epithelial ovarian cancer; 9% had high-risk early-stage disease; 30% were at high risk for progression; 21% had FIGO stage III, IIIA, or IIIB disease; 70% had FIGO stage IIIC or IV disease; 69% had a serous histologic type; and
26% had more than 1.0 cm of residual disease after surgical debulking (Table 1S in the Supplementary Appendix). Study Treatments Received
More than 90% of the women in both groups received 6 cycles of chemotherapy (91% in the standard-therapy group and 94% in the bevacizumab group). Women in the bevacizumab group who started chemotherapy at or before 4 weeks after surgery received a median of 16 cycles (interquartile range, 11 to 17) and those who started chemotherapy more than 4 weeks after surgery received a median of 17 cycles (interquartile range, 12 to 18); 470 patients (62%) continued to receive bevacizumab through cycle 18, with 324 patients (42%) missing no infusions. In the standardtherapy group, 1 patient received one dose of bevacizumab at cycle 1 in error. A total of 75 patients (5%) (48 in the standard-therapy group and 27 in the bevacizumab group) received additional chemotherapy or bevacizumab before disease progression. After disease progression, a maximum of 44 patients (3%) (30 in the standard-therapy group and 14 in the bevacizumab group) may have received further antiangiogenic treatment; the exact numbers are not known, since many patients are enrolled in blinded studies (Fig. 1). Adverse Events
Five deaths related to treatment or to treatment and disease were reported: one in the standardtherapy group (due to central nervous system ischemia) and four in the bevacizumab group (one each from gastrointestinal perforation, intracerebral hemorrhage, recurrent bowel perforation and ovarian cancer, and neutropenic sepsis and ovarian cancer). Adverse events of grade 3 or higher were reported in 56% of the women in the standard-therapy group and in 66% of the women in the bevacizumab group. Bevacizumab treatment appeared to be associated with an increase in bleeding (mainly grade 1 mucocutaneous bleeding), hypertension of grade 2 or higher (18% with bevacizumab vs. 2% with standard therapy), thromboembolic events of grade 3 or higher (7% with bevacizumab vs. 3% with standard therapy), and gastrointestinal perforations (occurring in 10 patients in the bevacizumab group vs. 3 patients in the standard-therapy group) (Table 1). In the bevacizumab group, the percentage of patients in whom abscesses, fistulas, or
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2487
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
1528 Women were enrolled from 263 sites in 11 countries
764 Were assigned to receive standard chemotherapy
764 Were assigned to receive standard chemotherapy plus bevacizumab therapy
696 (91%) Received 6 cycles of chemotherapy 11 Did not receive chemotherapy 2 Died 8 Withdrew consent 1 Had adverse event 57 Stopped chemotherapy early 22 Had adverse events or intercurrent illness 2 Died 16 Had insufficient therapeutic response 2 Declined to participate 8 Withdrew consent 2 Had protocol violations 3 Had administrative or other reasons 2 Had unknown reason
719 (94%) Received 6 cycles of chemotherapy 8 Did not receive chemotherapy 1 Had uncontrollable hypertension 6 Withdrew consent 1 Had an adverse event 37 Stopped chemotherapy early 15 Had adverse events or intercurrent illness 1 Died 6 Had insufficient therapeutic response 2 Declined to participate 6 Withdrew consent 2 Had protocol violations 3 Had administrative or other reasons 2 Had unknown reason
470 (62%) Received 18 cycles of bevacizumab 146 Missed one or more planned doses 324 Did not miss any doses 2 Were still receiving treatment at clinical cutoff 19 Did not receive bevacizumab 1 Died 8 Withdrew consent 1 Had uncontrollable hypertension 2 Had protocol violations 2 Had adverse events 1 Had delayed wound healing 4 Had unknown reasons 273 Stopped bevacizumab early (77 during chemotherapy, 196 during maintenance phase) 112 Had adverse events or intercurrent illness 52 During chemotherapy phase 60 During maintenance phase 1 Died 104 Had insufficient therapeutic response 19 Declined to participate 11 Withdrew consent 19 Had administrative or other reasons 7 Had unknown reason
48 Received additional chemotherapy or bevacizumab before progression 1 Received a single dose of bevacizumab in error 14 Received additional carboplatin plus paclitaxel cycle (cycles 7–12) 10 Received minor modification of first-line chemotherapy 17 Received new line of chemotherapy 6 Have details of chemotherapy being queried
27 Received additional chemotherapy or bevacizumab before progression 3 Received additional bevacizumab (cycle 19) 3 Received additional carboplatin plus paclitaxel cycle (cycles 7–12) 4 Received minor modification of first-line chemotherapy 11 Received new line of chemotherapy 6 Have details of chemotherapy being queried
Patient status at clinical cutoff: 130 (17%) Died 0 (0%) Alive on treatment 604 (79%) Alive in follow-up 30 (4%) Alive, but withdrew consent
Patient status at clinical cutoff: 111 (15%) Died 2 (<1%) Alive on treatment 625 (82%) Alive in follow-up 26 (3%) Alive, but withdrew consent
Figure 1. Randomization and Outcomes.
2488
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Bevacizumab in Ovarian Cancer
gastrointestinal perforations developed was similar in the group that received bevacizumab with their first cycle of chemotherapy and in the group that did not (3% [19 of 611] and 4% [6 of 135], respectively). Best Overall Response
The rate of complete or partial remission was 48% in the standard-therapy group and 67% in the bevacizumab group — a difference of 19 percentage points (95% confidence interval [CI], 11 to 28; P<0.001) (Table 2S in the Supplementary Appendix). Progression-Free Survival
With a median follow-up of 19.4 months, disease progression or death occurred in 759 patients (392 in the standard-therapy group and 367 in the bevacizumab group). The median progressionfree survival was 17.3 months in the standardtherapy group and 19.0 months in the bevacizu mab group. A comparison of Kaplan–Meier curves for progression-free survival showed a significant difference between the two groups (estimated hazard ratio for progression or death in the bevacizumab group, 0.81; 95% CI, 0.70 to 0.94; P = 0.004) (Fig. 2A). There was clear evidence of nonproportional hazards (test of proportional hazards, P<0.001). The effect of bevacizumab changed over time (Fig. 2B) and was maximal at 12 months, with an improvement in progressionfree survival at this time of 15.1% (95% CI, 10.7 to 19.5) as compared with standard therapy; thereafter, the effect was diminished so that progression-free survival was slightly higher in the standard-therapy group at 24 months. In detecting nonproportional hazards, the log-rank test remains useful, but the conventional hazard ratio is not meaningful; a better estimate of treatment effect is provided by the restricted mean difference (the difference in areas under the whole length of the progression-free survival curves). The restricted mean values for progression-free survival in the standard-therapy and bevacizumab groups, estimated with the use of all data obtained up to 36 months after randomization, were 20.3 months and 21.8 months, respectively — a difference of 1.5 months (95% CI, 0.1 to 2.9). Sensitivity analyses censoring data obtained at the last radiologic tumor assessment led to similar conclusions (Fig. 1S, Panel a, in the Supplementary Appendix) (hazard ratio for progression or death in the bevacizumab group, 0.79; 95% CI, 0.68 to
0.91; P = 0.001 by the log-rank test [nonproportional hazards detected]; restricted mean difference, 1.2 months [progression-free survival of 17.8 months with standard therapy vs. 19.0 months with bevacizumab], estimated at 30 months after randomization). Among patients at high risk for progression (Fig. 2C), the estimated median progression-free survival was 10.5 months with standard therapy, as compared with 15.9 months with bevacizumab (hazard ratio for progression or death in the bevacizumab group, 0.68; 95% CI, 0.55 to 0.85; P<0.001). The restricted mean values at 36 months were 13.1 months with standard therapy and 16.5 months with bevacizumab. Isolated CA-125–based progression occurred in 254 women (117 in the standard-therapy group and 137 in the bevacizumab group). Disease progression as assessed on the basis of CA-125, RECIST, or both occurred in 767 women (395 in the standardtherapy group and 372 in the bevacizumab group). The hazard ratio for biologic progression-free survival with bevacizumab, as compared with standard therapy, was 0.83 (95% CI, 0.72 to 0.96; P = 0.009) (Fig. 1S, Panel b, in the Supplementary Appendix). Exploratory analyses of potential interactions are summarized in Figure 3, and in Figure 3S in the Supplementary Appendix. There is a suggestion of a greater benefit with bevacizumab in women with a worse ECOG performance score than for those with a normal performance score (P = 0.02 for interaction). Overall Survival and Quality of Life
Overall survival data are not final; the final results are due in 2013 (Fig. 2D). There were 241 deaths: 231 (96%) were disease-related (128 in the standardtherapy group and 103 in the bevacizumab group), 5 were treatment-related, and 5 were due to other causes (see the Supplementary Appendix). Both groups showed improvement in global quality of life over time; although differences between the two groups were consistently present, these were small and not considered to be clinically significant (i.e., there was less than a 10-point difference) (Fig. 3S and Table 3S in the Supplementary Appendix). Updated Analyses
The cutoff date for updated analyses of progressionfree survival and overall survival was November 30, 2010, after a median follow-up period of 28 months
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2489
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
Table 1. Summary of Adverse Events.* Standard Chemotherapy (N = 753)
Adverse Events
Bevacizumab (N = 745)
number of patients (percent) Any event None
3 (<1)
0
Grade 1 or 2
331 (44)
254 (34)
Grade 3 or 4
410 (54)
483 (65)
9 (1)
8 (1)
666 (88)
450 (60)
85 (11)
286 (38)
2 (<1)
9 (1)
Grade 5 Any bleeding None Grade 1 or 2 Grade ≥3 Bleeding other than mucocutaneous, tumor-associated, or within CNS None
712 (95)
688 (92)
39 (5)
55 (7)
Grade 1 or 2 Grade ≥3
2 (<1)
2 (<1)
698 (93)
469 (63)
55 (7)
271 (36)
Mucocutaneous bleeding None Grade 1 or 2 Grade ≥3 Tumor-associated bleeding
0
5 (1)
0
0
Bleeding within CNS None
753 (100)
743 (99)
Grade 1 or 2
0
0
Grade ≥3
0
2 (<1)
Abscess and fistula None
743 (99)
732 (98)
Grade 1 or 2
3 (<1)
7 (1)
Grade ≥3
7 (1)
6 (1)
750 (99)
735 (99)
Gastrointestinal perforation None Grade 1 or 2
0
Grade ≥3
3 (<1)
0 10 (1)
Hypertension None
706 (94)
552 (74)
Grade 1
31 (4)
57 (8)
Grade 2
14 (2)
90 (12)
Grade ≥3
2 (<1)
46 (6)
Proteinuria None
734 (97)
712 (96)
18 (2)
29 (4)
Grade 1 or 2 Grade ≥3
2490
1 (<1)
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
4 (1)
Bevacizumab in Ovarian Cancer
Table 1. (Continued.) Standard Chemotherapy (N = 753)
Adverse Events
Bevacizumab (N = 745)
number of patients (percent) Any thromboembolic event None
708 (94)
665 (89)
Grade 1 or 2
22 (3)
29 (4)
Grade ≥3
23 (3)
51 (7)
722 (96)
695 (93)
Venous thromboembolic event None Grade 1 or 2
18 (2)
18 (2)
Grade ≥3
13 (2)
32 (4)
742 (99)
718 (96)
Arterial thromboembolic event None Grade 1 or 2
1 (<1)
Grade ≥3
7 (1)
10 (1)
20 (3)
750 (99)
740 (99)
Local thrombosis None Grade 1 or 2
3 (<1)
5 (1)
Grade ≥3
0
0
Neutropenia None
534 (71)
534 (72)
Grade 1 or 2
105 (14)
88 (12)
Grade ≥3
114 (15)
123 (17)
738 (98)
724 (97)
1 (<1)
2 (<1)
Febrile neutropenia None Grade 1 or 2 Grade ≥3
14 (2)
19 (3)
Thrombocytopenia None
684 (91)
652 (88)
Grade 1 or 2
54 (7)
67 (9)
Grade ≥3
15 (2)
26 (3)
Reversible posterior leukoencephalopathy syndrome
0
0
Congestive heart failure None
750 (99)
742 (99)
Grade 1 or 2
0
1 (<1)
Grade ≥3
3 (<1)
2 (<1)
737 (98)
708 (95)
13 (2)
27 (4)
Complication of wound healing None Grade 1 or 2 Grade ≥3
3 (<1)
10 (1)
753 (100)
742 (99)
Hyperbilirubinemia None Grade 1 or 2
0
2 (<1)
Grade ≥3
0
0
* CNS denotes central nervous system.
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2491
The
n e w e ng l a n d j o u r na l
in Progression-free Survival
75
Bevacizumab
50
Standard chemotherapy
25 0
0
3
6
9
12
15
18
21
24
27
30
Months since Randomization 693
464
216
91
25
715
585
263
73
19
C Progression-free Survival in Patients at High Risk for Progression
10
0 Kaplan–Meier difference
−10
Smoothed difference 95% CI 0
3
6
9
12
15
18
21
24
27
30
Months since Randomization
D Overall Survival
100
100
75
75
Bevacizumab
Survival (%)
Alive without Progression (%)
20
−20
No. at Risk Standard chemo- 764 therapy Bevacizumab 764
m e dic i n e
B Difference between Bevacizumab and Standard Chemotherapy
100
Difference in Progression-free Survival (%)
Alive without Progression (%)
A Progression-free Survival
of
Bevacizumab 50 Standard chemotherapy
25 0
0
3
6
9
12
15
Standard chemotherapy 50 25
18
21
24
27
0
30
0
Months since Randomization
3
6
9
12
15
18
21
24
27
30
Months since Randomization
No. at Risk
No. at Risk
Standard chemo- 234 therapy Bevacizumab 231
205
98
36
14
2
213
159
56
10
1
Standard chemo- 764 therapy Bevacizumab 764
724
652
368
159
33
737
678
404
162
40
Figure 2. Progression-free Survival and Overall Survival Curves. Panel A shows differences in progression-free survival according to treatment group in the total study population. Kaplan–Meier curves (and smoothed curves) in Panel B show the difference in progression-free survival between the standard-therapy group and the bevacizumab group, along with 95% confidence intervals; the red horizontal line represents zero effect (i.e., no treatment difference). The red horizontal line indicates no difference. Panel C shows progression-free survival, according to treatment group in patients at high risk for progression. Panel D shows overall survival according to treatment group.
and a total of 378 deaths observed. Table 2 shows with standard therapy, with median durations of the results of the primary and updated analyses for 19.8 months in the bevacizumab group and 17.4 months in the standard-therapy group, and corprogression-free survival and overall survival. responding restricted means of 24.1 months and 22.4 months at 42 months (hazard ratio for proProgression-free Survival A total of 934 progression-free survival events gression or death in the bevacizumab group, (progression or death) were reported (464 in the 0.87; 95% CI, 0.77 to 0.99; P = 0.04). The test for interaction suggests that the size standard-therapy group and 470 in the bevacizu mab group). Results were very similar to those of of the effect of bevacizumab differed between the primary analysis, again showing clear evidence patients at high risk for progression and the of a nonproportional hazard (P = 0.001) (Table 2, rest of the study population (P = 0.06) (Fig. 4S, and Fig. 5S, Panel a, in the Supplementary Appen- Panels a and b, in the Supplementary Appendix). dix). A long-term improvement in progression-free Among the 465 women at high-risk for prosurvival was seen with bevacizumab, as compared gression, 386 (83%) had disease progression or 2492
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Bevacizumab in Ovarian Cancer
B Updated Data, Progression-free Survival in Patients at High Risk for Progression
100
Alive without Progression (%)
Alive without Progression (%)
A Updated Data, Progression-free Survival
Bevacizumab
75 50
Standard chemotherapy
25 0
0
3
6
9
12 15 18 21 24 27 30 33 36
100 75 Bevacizumab Standard chemotherapy
50 25 0
0
3
6
Months since Randomization
9
12 15 18 21 24 27 30 33 36
Months since Randomization
No. at Risk
No. at Risk
Standard chemo- 764 therapy 764 Bevacizumab
693
474
350
221
114
39
716
599
430
229
107
27
C Updated Data, Overall Survival
Standard chemo- 234 therapy Bevacizumab 231
205
100
63
30
13
5
213
163
94
35
13
1
D Updated Data, Overall Survival in Patients at High Risk for Progression
100
100 Bevacizumab
Survival (%)
Survival (%)
75 Standard chemotherapy
50 25 0
Bevacizumab
75
Standard chemotherapy
50 25
0
3
6
9
0
12 15 18 21 24 27 30 33 36 39
0
3
6
Months since Randomization
9
12 15 18 21 24 27 30 33 36 39
Months since Randomization
No. at Risk
No. at Risk
Standard chemo- 764 741 724 703 672 646 623 542 421 304 212 132 71 26 therapy Bevacizumab 764 753 737 717 702 680 657 592 459 329 228 129 69 19
Standard chemo- 234 226 219 208 194 175 166 137 107 67 46 25 15 therapy Bevacizumab 231 227 222 214 208 199 186 164 134 94 65 31 18
6 4
Figure 3. Updated Progression-free Survival and Overall Survival Curves. Updated data on progression-free survival are shown according to treatment group in the total study population (Panel A) and in patients at high risk for progression (Panel B). Updated data on overall survival are shown according to treatment group in the total study population (Panel C) and in patients at high risk for progression (Panel D).
died, and the benefit with respect to progressionfree survival was greater with bevacizumab than with standard therapy, with median durations of 10.5 months in the standard-therapy group and 16.0 months in the bevacizumab group and restricted means of 14.5 months and 18.1 months at 42 months in the two groups, respectively (hazard ratio for progression or death in the bevacizumab group, 0.73; 95% CI, 0.60 to 0.93; P = 0.002). The updated progression-free survival curves were similar to those obtained in the primary analyses up to 24 months of follow-up (Fig. 3B).
Overall Survival
A total of 146 new deaths (all disease-related) were reported, making the total number of deaths 378 (200 in the standard-therapy group and 178 in the bevacizumab group). In contrast to the data for progression-free survival, there was no evidence of nonproportional hazards. A comparison of the Kaplan–Meier curves resulted in a hazard ratio for death in the bevacizumab group of 0.85 (95% CI, 0.69 to 1.04; P = 0.11). The early indication of a trend toward improved overall survival with bevacizumab has not changed (Fig. 3C). The test for interaction suggests that the size
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2493
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. Results of Primary and Updated Analyses of Progression-free Survival and Overall Survival with Standard Chemotherapy Alone versus Standard Therapy plus Bevacizumab.* Variable
Primary Analysis Standard Chemotherapy
Updated Analysis
Bevacizumab
Standard Chemotherapy
Bevacizumab
Progression-free survival All patients No. of events
392
367
464
470
Median (mo)
17.3
19.0
17.4
19.8
Hazard ratio (95% CI)
0.81 (0.70–0.94)
0.87 (0.77–0.99)
P value†
0.004
0.04
P value‡
<0.001
<0.001
Restricted mean (mo) At 36 mo
20.3
21.8
20.6
22.5
At 42 mo
NA
NA
22.4
24.1
No. of progression-free survival events
158
173
196
190
Median (mo)
10.5
15.9
10.5
Patients at high risk for progression§
Hazard ratio (95% CI)
0.68 (0.55–0.85)
16.0 0.73(0.60–0.93)
P value†
<0.001
0.002
P value‡
<0.001
<0.001
Restricted mean (mo) At 36 mo
13.1
16.5
14.1
17.6
At 42 mo
NA
NA
14.5
18.1
No. of events
130
111
200
178
Median (mo)
Not yet reached
Not yet reached
Not yet reached
Not yet reached
93
95
86
92
Overall survival¶ All patients
1-year survival, per Kaplan–Meier estimates (%) Hazard ratio (95% CI) P value† Patients at high risk for progression§
No analyses performed
0.81 (0.63–1.04)
0.85 (0.69–1.04)
0.098
0.11
No analyses performed
No. of events
109
79
Median (mo)
28.8
36.6
Hazard ratio (95% CI)
0.64 (0.48–0.85)
P value†
0.002
* The cutoff dates for the initial and updated data analyses were February 28, 2010, and November 30, 2010, respectively, and the respective median follow-up periods were 19 months and 28 months. CI denotes confidence interval. † P values were obtained with the use of the unadjusted log-rank test. ‡ P values were obtained with the use of the test based on Schoenfeld residuals for the proportional-hazards assumption. § There were 465 patients in this group. ¶ There were 1528 patients in this group.
2494
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Bevacizumab in Ovarian Cancer
of the effect of bevacizumab on overall survival differs between the patients at high risk for progression and the rest of the women studied (P = 0.011) (Fig. 4S, Panels c and d, in the Supplementary Appendix). Among the women at high risk for progression, there were 188 deaths (109 in the standard-therapy group and 79 in the bevaciz umab group) (Fig. 3D, and Fig. 4S, Panel d, in the Supplementary Appendix). The median overall survival is 28.8 months in the standard-therapy group and 36.6 months in the bevacizumab group, (hazard ratio for death in the bevacizumab group, 0.64; 95% CI, 0.48 to 0.85; P = 0.002).
Discussion The use of bevacizumab (7.5 mg per kilogram) given concurrently with 5 or 6 cycles of platinumbased chemotherapy and continued for an additional 12 cycles improved progression-free survival by about 2 months and increased the response rate by 20%. These data provide clear evidence of the biologic activity of bevacizumab, with a maximum improvement in progression-free survival of approximately 15% at 12 months, which disappeared by 24 months. The restricted mean difference summarized the treatment effect for the nonproportional hazards detected and showed an improvement in the mean progression-free survival of 1.5 to 2 months with bevacizumab. The progression-free survival and overall survival benefits were much greater among the patients at high risk for progression (improvements of 3.6 months [restricted mean] and 7.8 months [median], respectively) than among patients at lower risk. Both the timing of the maximal treatment effect, which coincided with the end of planned bevacizumab treatment (and was achieved in more than 70% of the women), and the larger treatment effect observed in women with more advanced disease, who were more likely to receive bevaciz umab up to the time of disease progression, are intriguing and raise the possibility that prolonged therapy beyond 12 months, perhaps until disease progression, might further improve the outcome. Treatment with bevacizumab did not affect the delivery of chemotherapy. However, bevacizumab
did expand the range of toxic effects, including hypertension and bowel perforation. As compared with the GOG-0218 study, the ICON7 study enrolled patients with advancedstage cancer with no visible residual disease, as well as some with high-risk early-stage disease; in the ICON7 study, half the dose of bevaciz umab was used (7.5 mg per kilogram, vs. 15 mg per kilogram in the GOG-0218 study) for a shorter maintenance period (12 cycles, vs. 16 cycles in the GOG-0218 study). The dose of bevacizumab used in the ICON7 study is one of the licensed doses for metastatic colorectal cancer but is half the licensed dose for metastatic breast cancer. The prognosis for patients at high risk for progression in the ICON7 study was similar to that for patients in the GOG-0218 study, with median progression-free survival of 10.5 months in the standard-therapy group. A 3.6-month (restricted mean) improvement in progression-free survival was observed with bevacizumab, similar to that seen in the GOG-0218 study.19 For the whole ICON7 patient population, the overall magnitude of the benefit with respect to progression-free survival was relatively modest after the addition of bevacizumab to platinum-based chemotherapy and maintenance bevacizumab for the 12-cycle extension after chemotherapy was completed. Some will argue that final overall survival data are needed before the results can be fully interpreted. The apparently greater effect of bevacizumab in patients with a poor prognosis is encouraging. Final results for overall survival rates should be available in 2013. Supported by Roche and the National Institute for Health Research, through the National Cancer Research Network. Disclosure forms provided by the authors are available with the full text of this article at NEJM.org. We thank the women who participated in the trial and their families, Cancer Research U.K. (C2113/A6865) for peer review of the ICON7 trial proposal, the participating Gynecologic Cancer InterGroup groups (Arbeitsgemeinschaft Gynaekologische Onkologie Studiengruppe Ovarialkarzinom [AGO-OVAR], Australia New Zealand Gynaecological Oncology Group [ANZGOG], Grupo Español de Investigaciõn en Cancer de Ovario [GEICO], Association de Recherche sur les Cancers dont Gynécologiques– Groupe d’Investigateurs Nationaux pour l’Etude des Cancers Ova riens et du Sein [ARCAGY–GINECO], Medical Research Council– National Cancer Research Institute [MRC–NCRI], Nordic Society for Gynecologic Oncology [NSGO], and NCIC Clinical Trials Group [NCICCTG]), and the 263 clinical sites and their staff.
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2495
Bevacizumab in Ovarian Cancer APPENDIX The authors’ affiliations are as follows: From St. James’s Institute of Oncology, St. James’s University Hospital (T.J.P., F.C.), and Leeds Institute of Molecular Medicine (D.S.), Leeds; St. Luke’s Cancer Center, Royal Surrey County National Health Service (NHS) Foundation Trust, Guildford (S.E.); and Cambridge Cancer Trials Center, Cambridge University Hospitals NHS Foundation Trust, Cambridge (W.Q.) — all in the United Kingdom; the Medical Research Council Clinical Trials Unit (A.M.S., W.Q., M.K.B.P.) and University College London Biomedical Research Centre, University College London Hospitals (J.A.L.) — both in London; Städtischen Klinikum Solingen, Klinik für Gynäkologie und Geburtshilfe, Solingen (J.P.), Universitätsklinikum Ulm, Ulm (C.K.), Dr. Horst Schmidt Klinik, Wiesbaden (A.B.); Universitätsklinikum Essen, Essen (R.K.); and St. Vincentius Kliniken, Karlsruhe (A.S.) — all in Germany; Université Paris Descartes, Assistance Publique–Hôpitaux de Paris, Hôpitaux Universitaires Paris Centre, Paris (E.P.-L.); the Department of Gynecologic Oncology and Institute of Medical Informatics, Oslo University Hospital, Oslo (G.K.); the Department of Obstetrics and Gynecology, University of British Columbia, Vancouver (M.S.C.), and the Department of Obstetrics and Gynecology, Laval University, Quebec, QC (M.P.) — both in Canada; Australia New Zealand Gynaecological Oncology Group, Camperdown, NSW, Australia (P.B.); the Department of Hematology and Medical Oncology, Institute of Health Research Hospital Clinico, University of Valencia, Valencia, Spain (A.C.); Charité Campus Virchow-Klinikum, Berlin (J.S.); the University of Edinburgh Cancer Research Centre, Institute of Genetics and Molecular Medicine, Edinburgh (C.G.); Centre Catherine de Sienne, Nantes (A. Lortholary); and Service d’Oncologie Medicale, Assistance Publique–Hôpitaux de Paris, Hôpital Tenon, Paris (F.S.) — both in France; the Department of Oncology, Rigshospitalet, Copenhagen (M.R.M.); the Women’s Clinic, Helsinki University Central Hospital, Helsinki (A. Leminen); and the Princess Margaret Hospital, Toronto (A.M.O.). References 1. Bookman MA, Brady MF, McGuire WP,
et al. Evaluation of new platinum-based treatment regimens in advanced-stage ovarian cancer: a Phase III Trial of the Gynecologic Cancer Intergroup. J Clin Oncol 2009; 27:1419-25. [Erratum, J Clin Oncol 2009;27: 2305.] 2. du Bois A, Weber B, Rochon J, et al. Addition of epirubicin as a third drug to carboplatin-paclitaxel in first-line treatment of advanced ovarian cancer: a prospectively randomized gynecologic cancer intergroup trial by the Arbeitsgemeinschaft Gynaeko logische Onkologie Ovarian Cancer Study Group and the Groupe d’Investigateurs Nationaux pour l’Etude des Cancers Ovariens. J Clin Oncol 2006;24:1127-35. 3. Pfisterer J, Weber B, Reuss A, et al. Randomized phase III trial of topotecan following carboplatin and paclitaxel in first-line treatment of advanced ovarian cancer: a gynecologic cancer intergroup trial of the AGO-OVAR and GINECO. J Natl Cancer Inst 2006;98:1036-45. 4. Hoskins P, Vergote I, Cervantes A, et al. Advanced ovarian cancer: phase III randomized study of sequential cisplatin-topotecan and carboplatin-paclitaxel vs. carboplatinpaclitaxel. J Natl Cancer Inst 2010;102: 1547-56. 5. Armstrong DK, Bundy B, Wenzel L, et al. Intraperitoneal cisplatin and paclitaxel in ovarian cancer. N Engl J Med 2006; 354:34-43. 6. Alberts DS, Liu PY, Hannigan EV, et al. Intraperitoneal cisplatin plus intravenous cyclophosphamide versus intravenous cisplatin plus intravenous cyclophosphamide for stage III ovarian cancer. N Engl J Med 1996;335:1950-5. 7. Markman M, Bundy BN, Alberts DS, et al. Phase III trial of standard-dose intravenous cisplatin plus paclitaxel versus moderately high-dose carboplatin followed by intravenous paclitaxel and intraperitoneal cisplatin in small-volume stage III ovarian
2496
carcinoma: an intergroup study of the Gynecologic Oncology Group, Southwestern Oncology Group, and Eastern Cooperative Oncology Group. J Clin Oncol 2001;19:10017. 8. Katsumata N, Yasuda M, Takahashi F, et al. Dose-dense paclitaxel once a week in combination with carboplatin every 3 weeks for advanced ovarian cancer: a phase 3, open-label, randomised controlled trial. Lancet 2009;374:1331-8. 9. Folkman J. Anti-angiogenesis: new concept for therapy of solid tumors. Ann Surg 1972;175:409-16. 10. Idem. What is the evidence that tumors are angiogenesis dependent? J Natl Cancer Inst 1990;82:4-6. 11. Yoneda J, Kuniyasu H, Crispens MA, Price JE, Bucana CD, Fidler IJ. Expression of angiogenesis-related genes and progression of human ovarian carcinomas in nude mice. J Natl Cancer Inst 1998;90:447-54. 12. Huang S, Robinson JB, Deguzman A, Bucana CD, Fidler IJ. Blockade of nuclear factor-kappaB signaling inhibits angiogenesis and tumorigenicity of human ovarian cancer cells by suppressing expression of vascular endothelial growth factor and interleukin 8. Cancer Res 2000; 60:5334-9. 13. Hurwitz H, Fehrenbacher L, Novotny W, et al. Bevacizumab plus irinotecan, fluoro uracil, and leucovorin for metastatic colo rectal cancer. N Engl J Med 2004;350:233542. 14. Sandler A, Gray R, Brahmer J. Randomised phase II/III trial of paclitaxel (P) plus carboplatin (C) with or without bevacizumab (NSC #704865) in patients with advanced non-squamous non-small cell lung cancer (NSCLC): an Eastern Cooperative Oncology Group (ECOG) trial. Presented at the 41st Annual Meeting of the American Society of Clinical Oncology, Orlando, FL, May 13–17, 2005. abstract. 15. Eskens FA, Sleijfer S. The use of beva-
cizumab in colorectal, lung, breast, renal and ovarian cancer: where does it fit? Eur J Cancer 2008;44:2350-6. 16. Burger RA, Sill MW, Monk BJ, Greer BE, Sorosky JI. Phase II trial of bevaciz umab in persistent or recurrent epithelial ovarian cancer or primary peritoneal cancer: a Gynecologic Oncology Group study. J Clin Oncol 2007;25:5165-71. 17. Cannistra SA, Matulonis UA, Penson RT, et al. Phase II study of bevacizumab in patients with platinum-resistant ovarian cancer or peritoneal serous cancer. J Clin Oncol 2007;25:5180-6. [Erratum, J Clin Oncol 2008;26:1773.] 18. Garcia AA, Hirte H, Fleming G, et al. Phase II clinical trial of bevacizumab and low-dose metronomic oral cyclophosphamide in recurrent ovarian cancer: a trial of the California, Chicago, and Princess Margaret Hospital Phase II Consortia. J Clin Oncol 2008;26:76-82. 19. Burger RA, Brady MF, Bookman MA, et al. Incorporation of bevacizumab in the primary treatment of ovarian cancer. N Engl J Med 2011;365:2473-83. 20. Therasse P, Arbuck SG, Eisenhauer EA, et al. New guidelines to evaluate the response to treatment in solid tumors: European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada. J Natl Cancer Inst 2000;92:205-16. 21. Rustin GJ, Quinn M, Thigpen T, et al. Re: new guidelines to evaluate the response to treatment in solid tumors (ovarian cancer). J Natl Cancer Inst 2004;96:487-8. 22. Royston P, Parmar MKB. Flexible proportional-hazards and proportional-odds models for censored survival data, with application to prognostic modelling and estimation of treatment effects. Stat Med 2002;21:2175-97. Copyright © 2011 Massachusetts Medical Society.
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
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
SF3B1 and Other Novel Cancer Genes in Chronic Lymphocytic Leukemia Lili Wang, M.D., Ph.D., Michael S. Lawrence, Ph.D., Youzhong Wan, Ph.D., Petar Stojanov, B.A., Carrie Sougnez, B.S., Kristen Stevenson, M.S., Lillian Werner, M.S., Andrey Sivachenko, Ph.D., David S. DeLuca, Ph.D., Li Zhang, Ph.D., Wandi Zhang, M.D., Alexander R. Vartanov, B.A., Stacey M. Fernandes, B.S., Natalie R. Goldstein, B.A., Eric G. Folco, Ph.D., Kristian Cibulskis, B.S., Bethany Tesar, M.S., Quinlan L. Sievers, B.A., Erica Shefler, B.S., Stacey Gabriel, Ph.D., Nir Hacohen, Ph.D., Robin Reed, Ph.D., Matthew Meyerson, M.D., Ph.D., Todd R. Golub, M.D., Eric S. Lander, Ph.D., Donna Neuberg, Sc.D., Jennifer R. Brown, M.D., Ph.D., Gad Getz, Ph.D., and Catherine J. Wu, M.D.
A BS T R AC T Background
The somatic genetic basis of chronic lymphocytic leukemia, a common and clinically heterogeneous leukemia occurring in adults, remains poorly understood. Methods
We obtained DNA samples from leukemia cells in 91 patients with chronic lymphocytic leukemia and performed massively parallel sequencing of 88 whole exomes and whole genomes, together with sequencing of matched germline DNA, to characterize the spectrum of somatic mutations in this disease. Results
Nine genes that are mutated at significant frequencies were identified, including four with established roles in chronic lymphocytic leukemia (TP53 in 15% of patients, ATM in 9%, MYD88 in 10%, and NOTCH1 in 4%) and five with unestablished roles (SF3B1, ZMYM3, MAPK1, FBXW7, and DDX3X). SF3B1, which functions at the catalytic core of the spliceosome, was the second most frequently mutated gene (with mutations occurring in 15% of patients). SF3B1 mutations occurred primarily in tumors with deletions in chromosome 11q, which are associated with a poor prognosis in patients with chronic lymphocytic leukemia. We further discovered that tumor samples with mutations in SF3B1 had alterations in pre–messenger RNA (mRNA) splicing. Conclusions
Our study defines the landscape of somatic mutations in chronic lymphocytic leukemia and highlights pre-mRNA splicing as a critical cellular process contributing to chronic lymphocytic leukemia.
From the Cancer Vaccine Center (L.W., Y.W., L.Z., W.Z., Q.L.S., C.J.W.) and the Departments of Medical Oncology (L.W., Y.W., L.Z., W.Z., A.R.V., S.M.F., N.R.G., B.T., Q.L.S., M.M., J.R.B., C.J.W.), Biostatistics and Computational Biology (K.S., L.W., D.N.), and Pediatric Oncology (E.G.F., R.R., T.R.G.), Dana–Farber Cancer Institute; the Departments of Cell Biology (E.G.F., R.R., T.R.G.) and Medicine (J.R.B., C.J.W.), Harvard Medical School; the Division of Allergy, Immunology, and Rheumatology, Department of Medicine, Massachusetts General Hospital (N.H.); and the Department of Medicine, Brigham and Women’s Hospital (J.R.B., C.J.W.) — all in Boston; the Broad Institute, Cambridge, MA (M.S.L., P.S., C.S., A.S., D.S.D., K.C., E.S., S.G., N.H., M.M., T.R.G., E.S.L., G.G.); and the Howard Hughes Medical Institute, Chevy Chase, MD (T.R.G.). Address reprint requests to Dr. Wu at the Dana– Farber Cancer Institute, Harvard Institutes of Medicine, Rm. 416B, 77 Ave. Louis Pasteur, Boston, MA 02115, or at
[email protected]. Drs. Wang, Lawrence, and Wan and Drs. Brown, Getz, and Wu contributed equally to this article. This article (10.1056/NEJMoa1109016) was published on December 12, 2011, at NEJM .org. N Engl J Med 2011;365:2497-506. Copyright © 2011 Massachusetts Medical Society.
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2497
The
n e w e ng l a n d j o u r na l
C
hronic lymphocytic leukemia is an incurable disease characterized by extensive clinical heterogeneity despite a common diagnostic immunophenotype (surface expression of CD19+, CD20+dim, CD5+, CD23+, and sIgMdim). Whereas the course of disease is indolent in some patients, it is steadily progressive in approximately half of patients, leading to substantial morbidity and mortality.1 Our ability to predict a more aggressive disease course has improved with the use of tests for biologic markers (degree of somatic hypermutation in the variable region of the immunoglobulin heavy chain [IGHV] gene and expression of ZAP70) and the detection of cytogenetic abnormalities (deletions in chromosomes 11q, 13q, or 17p and trisomy 12).2,3 Still, even with these advances, prediction of the disease course is not highly reliable. Massively parallel sequencing technology now provides a means of systematically discovering the genetic alterations that underlie disease and identifying new therapeutic targets and clinically predictive biomarkers. To date, most studies designed to discover tumor-associated mutations have relied on sequencing the genome or exome of only a few tumors; the newly discovered mutations detected are then further studied in an expanded cohort. These efforts have led to the identification of several important diseaseassociated mutations.4-6 A more powerful approach to the process of initial discovery is to sequence a much larger set of samples.7,8 This approach increases the chances that the full range of mutated genes will be detected, allows reconstruction of the genetic pathways underlying disease pathogenesis, and reveals associations between genetic events and the clinically important features of a disease. We therefore sequenced DNA samples of leukemia cells from 91 patients with chronic lymphocytic leukemia (88 exomes and 3 genomes), representing the broad clinical spectrum of the disease.
Me thods Study Design
Samples of DNA were obtained from normal tissues and tumors in 91 patients (discovery cohort) and 101 patients (extension cohort) with chronic lymphocytic leukemia, all of whom provided written informed consent before sample collec-
2498
of
m e dic i n e
tion. DNA was extracted from blood- or marrowderived lymphocytes in tumors and from autologous epithelial cells, fibroblasts, or granulocytes in normal tissue. Genome and Exome Sequencing
Libraries were constructed and sequenced on an Illumina Genome Analyzer II with the use of 101-bp paired-end reads for whole-genome sequencing and 76-bp paired-end reads for wholeexome sequencing. Output from Illumina software was processed by the PICARD data-processing pipeline to yield BAM files containing well-calibrated, aligned reads.7,9 BAM files were processed by the Broad Institute’s Firehose Pipeline, which provides quality control and identifies somatic point mutations, insertions or deletions, and other structural chromosomal rearrangements. Pre–messenger RNA (pre-mRNA) splicing alteration in leukemic samples was identified with the use of quantitative reverse-transcriptase– polymerase-chain-reaction assays to detect spliced and unspliced forms of representative spliceosome targets BRD2 and RIOK3.10 Associations between mutation rate and clinical features were assessed with the use of the Wilcoxon rank-sum test, Fisher’s exact test, or the Kruskal–Wallis test, as appropriate. A stepwise Cox proportional-hazards model was used to identify features with a significant effect on time to initial treatment. The materials and methods used in the study are more fully described in the Supplementary Appendix, available with the full text of this article at NEJM.org.
R e sult s Somatic Mutation Rate
We sequenced DNA derived from CD19+CD5+ leukemia cells and matched germline DNA derived from autologous skin fibroblasts, epithelial cells in saliva, or blood granulocytes. Samples were obtained from patients with a broad range of clinical characteristics, including those with del(11q) and del(17p) (indicating a poor prognosis) and with either unmutated or mutated IGHV status (Fig. 1A in the Supplementary Appendix). Deep sequence coverage was obtained to provide high sensitivity in identifying mutations (Table 1). To detect point mutations and insertions or deletions, we compared sequences in each tumor
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Novel Cancer Genes in Chronic Lymphocytic Leukemia
sample with corresponding normal sequences by means of well-validated algorithms.7,8,11,12 We detected 1838 nonsynonymous and 539 synonymous mutations in protein-coding sequences, corresponding to a mean (±SD) somatic mutation rate of 0.72±0.36 per megabase (range, 0.08 to 2.70), and an average of 20 nonsynonymous mutations per patient (range, 2 to 76) (Table 1, and Table 1 in the Supplementary Appendix). This rate is similar to that previously reported in chronic lymphocytic leukemia and other hematologic cancers.5-7,13,14 We observed no significant difference in the rates of nonsynonymous mutation between tumors with mutated genes and those with unmutated genes or in tumors in different clinical stages of disease (Table 2 in the Supplementary Appendix). Prior exposure to chemotherapy (in 30 of 91 patients) was not associated with an increased rate of nonsynonymous mutation (P = 0.14) (Fig. 1B in the Supplementary Appendix).15 Identification of Genes with Significant Mutation Frequencies
To identify genes whose mutations were associated with leukemic tumorigenesis (“driver” mutations), we examined all 91 leukemic and normal pairs with the use of the MutSig algorithm for genes that were mutated at a rate significantly higher than the background rate, given their sequence composition (see the Supplementary Appendix). Nine such genes were identified (Q≤0.1 after correction for multiple-hypothesis testing): TP53, SF3B1, MYD88, ATM, FBXW7, NOTCH1, ZMYM3, DDX3X, and MAPK1 (Fig. 1). Whereas the overall ratio of nonsynonymous to synonymous mutations was 3.1:1, the mutations in these nine genes were exclusively nonsynonymous (65:0, P<5×10−6) (Table 1 in the Supplementary Appendix), a finding that further supports their functional importance. Moreover, these mutations occurred exclusively in conserved sites across species (Fig. 2 in the Supplementary Appendix). Four of the genes with significant mutation frequencies, TP53, ATM, MYD88, and NOTCH1, have been described previously in chronic lymphocytic leukemia.6,16-18 We found 15 TP53 mutations in 14 of 91 patients (15%; Q≤6.3×10−8), most of which were localized to the DNA-binding domain that is critical for its tumor-suppressor activity17 (Fig. 3A in the Supplementary Appendix). In 8 patients (9%), we detected 9 ATM
Table 1. Summary Metrics of Whole-Genome and Whole-Exome Sequencing Studies.* Variable
Value†
Whole genomes No.
3
Bases covered per genome — %
70
Genome coverage CLL samples
38x
Normal samples
33x
Whole exomes No.
88
Bases covered per exome — %
81
Exome coverage CLL samples
132x
Normal samples
146x
Nonsynonymous mutations No. of mutations per megabase
0.7±0.36
Coding mutations — no. (range)
20 (2–76)
Synonymous mutations No. of mutations per megabase
0.2±0.16
Coding mutations — no. (range)
5.8 (0–31)
* Plus–minus values are means ±SD. CLL denotes chronic lymphocytic leukemia. † In 38x, 33x, 132x, and 146x, “x” denotes the average number of reads covering each nucleotide base that was sequenced.
mutations (Q≤1.1×10−5) scattered across this large gene, including in regions where mutation has been associated with defective DNA repair in patients with chronic lymphocytic leukemia.16 MYD88, a critical adaptor molecule of the interleukin-1 receptor–toll-like receptor (TLR) signaling pathway, harbored missense mutations in 9 patients (10%) at three sites localized within 40 amino acids of the interleukin-1 receptor– TLR domain. One site was novel (P258L), whereas the other two were identical to those recently described as activating mutations of the nuclear factor κB (NF-κB)–TLR pathway in patients with diffuse large B-cell lymphoma (M232T and L265P) (Fig. 3C in the Supplementary Appendix).19 Finally, we detected a recurrent frameshift mutation (P2514fs) in the C-terminal PEST domain of NOTCH1 in 4 patients (4%) that was identical to that recently reported in other investigations of chronic lymphocytic leukemia.5,6 This mutation is associated with unmutated IGHV and a poor prog-
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2499
Genes with Significant Mutation Frequency
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
No. of Mutations (%)
No. of Base Pairs
P Value
Q Value
TP53
TP53
15 (15)
119,041
<1.0×10−11
<6.3×10−8
SF3B1
SF3B1
14 (15)
359,856
<1.0×10−11
<6.3×10−8
MYD88
MYD88
9 (10)
73,805
<1.0×10−11
<6.3×10−8
ATM
ATM
9 (9)
837,986
2.4×10−9
1.1×10−5
FBXW7
FBXW7
4 (4)
225,671
2.0×10−6
7.4×10−3
NOTCH1
NOTCH1
4 (4)
306,968
3.3×10−6
1.0×10−2
314,226
3.5×10−5
7.4×10−2
181,343
1.6×10−5
4.3×10−2
89,405
1.9×10−5
4.4×10−2
ZMYM3
ZMYM3
DDX3X
DDX3X
MAPK1
MAPK1 0
4
8
12
4 (4) 3 (3) 3 (3)
16
No. of Mutations per 91 Patients
Figure 1. Genes with Significant Mutation Frequencies in 91 Patients with Chronic Lymphocytic Leukemia. The figure shows the number of mutations per gene that were found at a significant frequency, the percentage of patients who had each mutated gene, and for each gene, the total territory in numbers of base pairs with sufficient sequencing coverage across normal and malignant samples from the 91 patients. The P values and Q values were calculated by comparing the probability of the observed constellation of mutations with the background mutation rates across the data set.
nosis,5,6 and it is predicted to cause impaired degradation of NOTCH1, leading to pathway activation. Five of the genes with significant mutation frequencies (SF3B1, FBXW7, DDX3X, MAPK1, and ZMYM3) do not have established roles in chronic lymphocytic leukemia. Strikingly, the second most frequently mutated gene in our cohort was splicing factor 3b, subunit 1 (SF3B1), with missense mutations occurring in 14 of 91 patients (15%). SF3B1 is a component of the SF3b complex, which is associated with the U2 small nuclear ribonucleoprotein (snRNP) at the catalytic center of the spliceosome.20 SF3B1, other U2 snRNP components, and defects in splicing are not typically implicated in the biology of chronic lymphocytic leukemia. Remarkably, all 14 mutations were localized within C-terminal PP2A-repeat regions 5 through 8, which are highly conserved from humans to yeasts (Fig. 2 and 3 in the Supplementary Appendix), and in 7 instances, an identical amino acid change was produced (by the K700E mutation). The clustering of heterozygous mutations within specific domains and at identical sites suggests that mutations in SF3B1, like those in MYD88 and NOTCH1, cause specific functional changes. Whereas the N-terminal domain of SF3B1 is known to interact directly with other
2500
n engl j med 365;26
spliceosome components,20 the precise role of its C-terminal domain remains unknown. Only 6 mutations have been reported in SF3B1, all in solid tumors and in the PP2A-repeat region (Table 4 in the Supplementary Appendix). The four remaining genes with significant mutation frequencies have not, to our knowledge, been reported in previous studies of chronic lymphocytic leukemia and appear to have functions that interact with the five commonly mutated genes cited above (Fig. 3 in the Supplementary Appendix). FBXW7 (four distinct mutations) is a ubiquitin ligase that is known to be a tumorsuppressor gene, with loss of expression in a wide range of cancers.21,22 Its targets include important oncoproteins such as Notch1, c-Myc, c-Jun, cyclin E1, and MCL1.21,22 Two of the four mutations in FBXW7 cause constitutive Notch signaling in T-cell acute lymphoblastic leukemia.23 DDX3X (three distinct mutations) is an RNA helicase that functions at multiple levels of RNA processing, including RNA splicing, transport, and translation initiation as well as regulation of an RNA-sensing proinflammatory pathway.24 DDX3X interacts directly with XPO1,24 which was recently reported as being mutated in 2.4% of patients with chronic lymphocytic leuke-
nejm.org
december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Novel Cancer Genes in Chronic Lymphocytic Leukemia
mia.6 MAPK1 (three distinct mutations), also known as ERK, is a kinase that is involved in core cellular processes, such as proliferation, differentiation, transcription regulation, and development, and is a key signaling component of the TLR pathway.25,26 Two of three distinct MAPK1 mutations localize to the protein kinase domain; these mutations, to our knowledge, are the first reported examples of somatic mutations within the protein kinase domain of an ERK family member in a human cancer. Finally, we identified four distinct mutations in ZMYM3, a component of multiprotein complexes containing histone deacetylase that function to silence genes by modifying chromatin structure.27 We validated the three most frequently recurring mutations — SF3B1–K700E, MYD88–L265P, and NOTCH1–P2514fs — in 101 independent paired tumor and germline DNA samples, with similar detection frequencies in the discovery and extension cohorts (P = 0.20, P = 0.58, and P = 0.38, respectively) (Table 5 in the Supplementary Appendix). The nine genes with mutations at significant frequencies appear in five core signaling pathways, in which the genes play well-established roles: DNA repair and cell-cycle control (TP53 and ATM), Notch signaling (FBXW7 and NOTCH123), inflammatory pathways (MYD88, DDX3X, and MAPK1), and RNA splicing and processing (SF3B1 and DDX3X) (Fig. 2). We also noticed that additional genes are mutated in these pathways28 (Fig. 2, and Fig. 4 and Table 6 in the Supplementary Appendix). Although the frequency of mutations in these genes does not reach statistical significance, whether they are considered alone or as a set, it might do so in a larger collection of samples. On the other hand, in our cohort, 19 of 59 genes classified as members of the Wnt signaling pathway, which has been implicated in chronic lymphocytic leukemia in gene-expression studies,29,30 were mutated. Although no individual gene had a frequency of mutation that reached significance, the genes involved in the Wnt pathway, as a set, had a high mutation frequency (P = 0.048) (Fig. 2).
with chronic lymphocytic leukemia and IGHV mutation status in samples harboring mutations in the nine genes with significant mutation frequencies. We ordered the samples in accordance with their cytogenetic features on fluorescence in situ hybridization (FISH), using an established model of hierarchical risk3; del(13q) was associated with the most favorable prognosis when present alone, trisomy 12 was associated with a less-favorable prognosis, and del(11q) and del(17p) were both associated with aggressive disease that was refractory to chemotherapy (Fig. 3, and Tables 7 and 8 in the Supplementary Appendix). The distinct prognostic implications of these cytogenetic abnormalities suggest that each abnormality may reflect a distinct pathogenesis. Our data show that different driver mutations are associated with different key abnormalities detected on FISH, providing support for this hypothesis. Most of the TP53 mutations (11 of 17) were present in samples that also harbored del(17p) (P<0.001), resulting in homozygous p53 inactivation; this finding is consistent with previous observations.17 Mutations in ATM, which lies in the minimally deleted region of chromosome 11q, were marginally associated with del(11q) — that is, in 4 of 22 del(11q) samples (P = 0.09). Strikingly, mutations in SF3B1 were associated with del(11q) — in 8 of 22 del(11q) samples (36%, P = 0.004). Of the 6 leukemic samples with mutated SF3B1 and without del(11q), 2 also harbored a heterozygous mutation in ATM. These findings strongly suggest that there is an interaction between del(11q) and SF3B1 mutation in the pathogenesis of this clinical subgroup of chronic lymphocytic leukemia. We further observed that the mutations in NOTCH1 and FBXW7 were associated with trisomy 12 (P = 0.009 and P = 0.05, respectively). As in previous studies,5,6 NOTCH1 mutations were consistently associated with unmutated IGHV. The NOTCH1 and FBXW7 mutations were present in independent samples, suggesting that they may lead to aberrant Notch signaling in patients with trisomy 12 and unmutated IGHV. All MYD88 mutations were present in samples that were heterozygous for del(13q) (P = 0.009). Association of Driver Mutations As indicated in recent reports,5,6 we found that with Distinct Clinical Groups the MYD88 mutation was always associated with To examine the association between driver muta- mutations in the IGHV region (P = 0.001), which tions and particular clinical features, we as- suggests a postgerminal-center origin. We specsessed the cytogenetic aberrations associated ulate that in chronic lymphocytic leukemia, as in
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2501
The
n e w e ng l a n d j o u r na l
diffuse large B-cell lymphoma, in which MYD88 is frequently mutated,19 the constitutive activation of the NF-κB pathway may have its greatest effect in the germinal center. SF3B1 Mutations
In the 192 leukemic samples in the discovery and extension sets, mutations in NOTCH1 were associated with unmutated IGHV status and mutations in MYD88 were associated with mutated IGHV status. The mutation SF3B1–K700E was associated with unmutated IGHV status (P = 0.048), but it was also detected in samples with mutated IGHV, which suggests that it is an independent risk factor for chronic lymphocytic leukemia (Fig. 5A in the Supplementary Appendix). Indeed, a Cox multivariable regression model designed to test for clinical factors contributing to an earlier initiation of treatment in the 91 patients in the study revealed that an SF3B1 mutation was predictive of an earlier need for treatment (hazard ratio, 2.20; P = 0.03), independently of other established predictive markers, such as IGHV mutation status, del(17p), or an ATM mutation (Fig. 4A). Consistent with the results of these analyses was the finding that the time to initial treatment for patients with the SF3B1 mutation alone — without del(11q) — was similar to that for patients with del(11q) alone or with both del(11q) and an SF3B1 mutation. All three groups had significantly shorter times to initial treatment than patients without an SF3B1 mutation or without del(11q) (Fig. 5B in the Supplementary Appendix) (P<0.001). Similarly, shorter times to initial treatment were observed among 3 patients in the extension cohort whose tumors harbored the SF3B1–K700E mutation as compared with those whose tumors did not show this mutation. Because SF3B1 encodes a splicing factor that lies at the catalytic core of the splice osome, we looked for functional evidence of alterations in splicing associated with an SF3B1 mutation. Kotake et al. previously used intron retention in the endogenous genes BRD2 and RIOK3 to test the function of the SF3b complex.10 We confirmed that E7107, which targets this complex, inhibits the splicing of BRD2 and RIOK3 in both normal cells and chronic lymphocytic leukemia cells (Fig. 6A in the Supplementary Appendix). Using this assay, we found aberrant endogenous splicing activity in tumor 2502
of
m e dic i n e
Figure 2 (facing page). Core Signaling Pathways in Chronic Lymphocytic Leukemia. The nine genes with significant mutation frequencies fall into five core signaling pathways, in which the genes play well-established roles: DNA damage repair and cellcycle control (Panel A), Notch signaling (Panel B), inflammatory pathways (Panel C), Wnt signaling (Panel D), and RNA splicing and processing (Panel E). Genes with significant mutation frequencies are shown in red, and genes with mutations that are in a signaling pathway related to chronic lymphocytic leukemia are shown in pink. A list of additional pathway-associated genes with mutations is provided in Table 6 in the Supplementary Appendix. Co-A denotes coenzyme A, TLR toll-like receptor, and TNFR tumor necrosis factor receptor.
samples from 13 patients with mutated SF3B1 as compared with 17 patients with wild-type SF3B1, and the ratio of unspliced to spliced mRNA forms of BRD2 and RIOK3 was significantly higher in patients with SF3B1 mutations (median ratio, 2.0:1 vs. 0.55:1 [P<0.001], and 4.6:1 vs. 2.1:1 [P = 0.006], respectively) (Fig. 4B). In contrast, no splicing defects were detected in samples with the del(11q) defect and wild-type SF3B1 as compared with samples with the del(11q) defect and mutated SF3B1 (Fig. 6 in the Supplementary Appendix). These studies indicate that splicing function in chronic lymphocytic leukemia is altered as a result of a mutation in SF3B1 rather than del(11q).
Discussion Massively parallel sequencing technology has dramatically accelerated the discovery of genetic alterations in cancer.5,6,14 Our analysis of samples from 91 patients with chronic lymphocytic leukemia provided the statistical power to identify the involvement of nine driver genes and to suggest the involvement of six distinct pathways in the pathogenesis of this disease. Moreover, we discovered novel associations with prognostic markers that shed light on the biology underlying this clinically heterogeneous disease. The data led us to several general conclusions. First, like other hematologic cancers,14 chronic lymphocytic leukemia has a lower rate of somatic mutation than most solid tumors.5,6 Second, the rate of nonsynonymous mutation was not strongly affected by therapy. Third, in addition to finding the expected mutations in cell-cycle and DNA-repair pathways, we found
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Novel Cancer Genes in Chronic Lymphocytic Leukemia
B
DNA Damage and Cell-Cycle Control
C
Notch Signaling
Cell membrane
DNA damage
D
Wnt Pathway DKK2
TNFR1
MYD88
Me
p ro tallo
te as
γ-Secretas
TP53
Cell-cycle control
DNA damage repair
CDC14B
SMC1A
ANAPC4
ERCC4
PTTG1
BRCA1
ESPL1
FANCA
HDAC4
MSH4
LRP5/6
Frizzled
RYK
FZD5
Delta/Jagged
Cell membrane
ATM
WNT1
Wnt
CD14
Notch
ATM
Interleukin1R
TLRs
Extracellular space
P
Inflammatory Pathways
TLR8
A
Intracellular space
Dishevelled
RIPK1
MYD88
CSNK1E
e
e
MAPK8
DDX3X
IRAK4
IC
MAPK1 (ERK)
PRICKLE1
Multiprotein destruction complex P
TRAF3
β-Catenin Proteosomal degradation
β-Catenin
PPM1A FBXW7
MAP3K14/ NIK
IC-Notch
β-Catenin
IKBα
β-Catenin
Nucleus
E2F3 CCNB3
PPM1A C-Notch Co-A CSL
Genes with significant mutation frequency
Target genes
Proinflammatory cytokines
MYC CHD8 β-Catenin CREBBP
Transcription
LEF/TCF
Other pathway-associated genes
E RNA Splicing and Processing
Protein Nucleus Pre-mRNA Exon
5'
Intron 3'
Exon
Translation initiation
XPO1 DDX3X
Splicing Spliceosome SPOP
RBM39
SF3B1 PRPF8
U2AF2
3'-End processing
mRNA Exon
Exon
mRNA export
CPSF2
genetic alterations in Notch signaling, inflam- sociations with standard prognostic markers, matory pathways, and RNA splicing and process- suggesting that particular combinations of genetic ing. Fourth, driver mutations showed striking as- alterations may act in concert to drive cancer. β-Catenin
n engl j med 365;26
nejm.org
december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2503
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
FISH Cytogenetic Features
1
91
del(13q) Trisomy 12 del(11q) del(17p)
Genes with Significant Mutation Frequency and Related Pathways
P<0.001 Cell cycle or DNA damage
RNA processing
Notch1 signaling
Inflammatory pathway
TP53 ATM Additional
P=0.009 P=0.004
SF3B1 DDX3X Additional
P=0.009
NOTCH1 FBXW7 MYD88 MAPK1 Additional
P=0.001
IGHV Mutation Status
ZMYM3
Figure 3. Associations between Gene Mutations and Clinical Characteristics. Samples from the 91 patients in the study were sorted by means of fluorescence in situ hybridization (FISH), with the use of a model of hierarchical risk established by Döhner et al.3 Samples were scored for the presence or absence of mutations in the nine genes with significant mutation frequencies (darker colors), as well as for mutations in additional pathway-associated genes (lighter colors) and in the immunoglobulin heavy-chain variable (IGHV) mutation status (blue indicates mutation, white no mutation, and hatched unknown status). (A list of additional mutated pathway-associated genes is provided in Table 6 in the Supplementary Appendix.) Associations between gene mutation status and FISH cytogenetic features or IGHV status were calculated with the use of Fisher’s exact test and corrected for multiple hypothesis testing (Q≤0.1 for all comparisons shown).
A major surprise was the finding that a core spliceosome component, SF3B1, was mutated in about 15% of the study patients. Further analysis revealed that samples with SF3B1 mutations had enhanced intron retention within two specific transcripts previously shown to be affected by compounds that disrupt SF3b spliceosome function.10,31 Studies of these compounds have suggested that rather than inducing a global change in splicing, SF3b inhibitors alter the splicing of a narrow spectrum of transcripts derived from genes involved in cancer-related processes, including cell-cycle control (p27, CCA2, STK6, and MDM2),31-33 angiogenesis, and apoptosis.34 Our results suggest that SF3B1 mutations lead to mistakes in the splicing of these and other specific transcripts that affect the pathogenesis of chronic lymphocytic leukemia. Ongoing studies 2504
n engl j med 365;26
will focus on determining how mutations in SF3B1 alter its function in the processing of critical mRNAs. Since mutations in SF3B1 are highly enriched in patients with del(11q), these mutations may be synergistic with loss of ATM, a hypothesis that is supported by the observation that two patients had point mutations in both ATM and SF3B1 but did not have del(11q). Providing further support for this hypothesis, a recent unbiased functional screen showed that core spliceosome components were required for DNA repair in mammalian cells.35 As illustrated by our findings regarding SF3B1 mutations, identification of coding mutations in chronic lymphocytic leukemia can lead to the development of mechanistic hypotheses, novel prognostic markers, and potential therapeutic
nejm.org
december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Novel Cancer Genes in Chronic Lymphocytic Leukemia
A Variable IGHV Unmutated vs. mutated Missing vs. mutated del(17p) Present vs. absent SF3B1 Mutation vs. wild type ATM Mutation vs. wild type
Hazard Ratio (95% CI)
P Value
5.07 (2.62–9.81) 1.38 (0.53–3.58)
<0.001 0.51
2.85 (1.45–5.59)
0.002
2.20 (1.07–4.54)
0.03
3.77 (1.48–9.58)
0.005
B BRD2 P<0.0003
Figure 4. Mutations in SF3B1 and Altered mRNA Splicing. In Panel A, a Cox multivariable regression model designed to test for clinical factors contributing to the need for earlier initiation of treatment showed that an SF3B1 mutation was an independent predictor of a shorter time to treatment, regardless of the status of several other independent predictive markers. Panel B shows the relative amounts of spliced and unspliced spliceosome target messenger RNA (mRNA) in BRD2 and RIOK3 in normal CD19+ B cells and chronic lymphocytic leukemia B cells with wild-type or mutated SF3B1, as measured by means of a quantitative polymerase-chain-reaction assay. The ratios of unspliced to spliced mRNA were normalized to the percentage of leukemia cells per sample, and comparisons were calculated with the use of the Wilcoxon rank-sum test. CI denotes confidence interval.
P<0.0001 8
cerning chronic lymphocytic leukemia, including which genes within chromosomal deletions and amplifications are essential, how each mutation alters cellular networks and phenotypes, which combinations of mutations are critical in the development of cancer, and how genetic events in the host may affect the importance of specific mutations and their combinations.
6
Ratio of Unspliced to Spliced mRNA
4 2 0
RIOK3 P=0.009 P=0.006 10 8 6 4 2 0
SF3B1 Normal CD19+ Wild Type B Cells (N=17) (N=6)
SF3B1 Mutation (N=13)
targets. In addition, this information provides a starting point for the systematic analyses needed to address several fundamental questions con-
Supported by the Broad Institute through a grant from the National Human Genome Research Institute at the National Institutes of Health (U54HG003067) and by grants from the National Institutes of Health (GM43375, to Drs. Folco and Reed, and K23 CA115682, to Dr. Brown), the National Cancer Institute (5R21CA115043-2), the Melton and Rosenbach Funds (to Dr. Brown), the Blavatnik Family Foundation (to Dr. Wu), the Early Career Physician-Scientist Award of the Howard Hughes Medical Institute (to Dr. Wu), and the Damon Runyon Cancer Research Foundation (CI-38-07, to Dr. Wu). Dr. Brown is a Scholar of the American Society of Hematology and a Scholar in Clinical Research of the Leukemia and Lymphoma Society. Disclosure forms provided by the authors are available with the full text of this article at NEJM.org. We thank all members of the Biological Samples Platform, Genetic Analysis Platform, and Genome Sequencing Platform at the Broad Institute, without whom this work would not have been possible; Dan-Avi Landau and Jerome Ritz for their critical reading of an earlier draft of the manuscript; James Rheinwald and Patricia Barron of the Cell Culture Core at the Harvard Skin Disease Research Center for initiating the primary human dermal fibroblast cell lines used in this study; and the Chronic Lymphocytic Leukemia Research Consortium Biorepository for providing information on some of our samples.
References 1. Zenz T, Mertens D, Küppers R, Döh-
ner H, Stilgenbauer S. From pathogenesis to treatment of chronic lymphocytic leukaemia. Nat Rev Cancer 2010;10:3750. 2. Rassenti LZ, Huynh L, Toy TL, et al. ZAP-70 compared with immunoglobulin heavy-chain gene mutation status as a
predictor of disease progression in chronic lymphocytic leukemia. N Engl J Med 2004;351:893-901. 3. Döhner H, Stilgenbauer S, Benner A, et al. Genomic aberrations and survival in chronic lymphocytic leukemia. N Engl J Med 2000;343:1910-6. 4. Tiacci E, Trifonov V, Schiavoni G, et
n engl j med 365;26
nejm.org
al. BRAF mutations in hairy-cell leukemia. N Engl J Med 2011;364:2305-15. 5. Fabbri G, Rasi S, Rossi D, et al. Analysis of the chronic lymphocytic leukemia coding genome: role of NOTCH1 mutational activation. J Exp Med 2011;208: 1389-401. 6. Puente XS, Pinyol M, Quesada V, et al.
december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2505
Novel Cancer Genes in Chronic Lymphocytic Leukemia Whole-genome sequencing identifies recurrent mutations in chronic lymphocytic leukaemia. Nature 2011;475:101-5. 7. Chapman MA, Lawrence MS, Keats JJ, et al. Initial genome sequencing and analysis of multiple myeloma. Nature 2011; 471:467-72. 8. Cancer Genome Atlas Research Network. Integrated genomic analyses of ovarian carcinoma. Nature 2011;474:609-15. 9. DePristo MA, Banks E, Poplin R, et al. A framework for variation discovery and genotyping using next-generation DNA sequencing data. Nat Genet 2011;43:491-8. 10. Kotake Y, Sagane K, Owa T, et al. Splicing factor SF3b as a target of the antitumor natural product pladienolide. Nat Chem Biol 2007;3:570-5. 11. Berger MF, Lawrence MS, Demichelis F, et al. The genomic complexity of primary human prostate cancer. Nature 2011; 470:214-20. 12. Robinson JT, Thorvaldsdóttir H, Winck ler W, et al. Integrative genomics viewer. Nat Biotechnol 2011;29:24-6. 13. Mardis ER, Ding L, Dooling DJ, et al. Recurring mutations found by sequencing an acute myeloid leukemia genome. N Engl J Med 2009;361:1058-66. 14. Ley TJ, Mardis ER, Ding L, et al. DNA sequencing of a cytogenetically normal acute myeloid leukaemia genome. Nature 2008;456:66-72. 15. Cancer Genome Atlas Research Network. Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature 2008; 455:1061-8. 16. Austen B, Powell JE, Alvi A, et al. Mutations in the ATM gene lead to impaired overall and treatment-free survival that is independent of IGVH mutation status in patients with B-CLL. Blood 2005;106: 3175-82. 17. Zenz T, Eichhorst B, Busch R, et al. TP53 mutation and survival in chronic
lymphocytic leukemia. J Clin Oncol 2010; 28:4473-9. 18. Trbusek M, Smardova J, Malcikova J, et al. Missense mutations located in structural p53 DNA-binding motifs are associated with extremely poor survival in chronic lymphocytic leukemia. J Clin Oncol 2011; 29:2703-8. 19. Ngo VN, Young RM, Schmitz R, et al. Oncogenically active MYD88 mutations in human lymphoma. Nature 2011;470:115-9. 20. Wahl MC, Will CL, Lührmann R. The spliceosome: design principles of a dynamic RNP machine. Cell 2009;136:701-18. 21. Yada M, Hatakeyama S, Kamura T, et al. Phosphorylation-dependent degradation of c-Myc is mediated by the F-box protein Fbw7. EMBO J 2004;23:2116-25. 22. Babaei-Jadidi R, Li N, Saadeddin A, et al. FBXW7 influences murine intestinal homeostasis and cancer, targeting Notch, Jun, and DEK for degradation. J Exp Med 2011;208:295-312. 23. O’Neil J, Grim J, Strack P, et al. FBW7 mutations in leukemic cells mediate NOTCH pathway activation and resistance to gamma-secretase inhibitors. J Exp Med 2007;204:1813-24. 24. Rosner A, Rinkevich B. The DDX3 subfamily of the DEAD box helicases: divergent roles as unveiled by studying different organisms and in vitro assays. Curr Med Chem 2007;14:2517-25. 25. Pepper C, Thomas A, Hoy T, Milligan D, Bentley P, Fegan C. The vitamin D3 analog EB1089 induces apoptosis via a p53-independent mechanism involving p38 MAP kinase activation and suppression of ERK activity in B-cell chronic lymphocytic leukemia cells in vitro. Blood 2003;101: 2454-60. 26. Muzio M, Apollonio B, Scielzo C, et al. Constitutive activation of distinct BCRsignaling pathways in a subset of CLL patients: a molecular signature of anergy. Blood 2008;112:188-95.
27. Lee MG, Wynder C, Cooch N, Shie khattar R. An essential role for CoREST in nucleosomal histone 3 lysine 4 demethylation. Nature 2005;437:432-5. 28. Subramanian A, Tamayo P, Mootha VK, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 2005;102:15545-50. 29. Gutierrez A Jr, Tschumper RC, Wu X, et al. LEF-1 is a prosurvival factor in chronic lymphocytic leukemia and is expressed in the preleukemic state of monoclonal B-cell lymphocytosis. Blood 2010; 116:2975-83. 30. Klein U, Tu Y, Stolovitzky GA, et al. Gene expression profiling of B cell chronic lymphocytic leukemia reveals a homogeneous phenotype related to memory B cells. J Exp Med 2001;194:1625-38. 31. Kaida D, Motoyoshi H, Tashiro E, et al. Spliceostatin A targets SF3b and inhibits both splicing and nuclear retention of pre-mRNA. Nat Chem Biol 2007;3:576-83. 32. Corrionero A, Miñana B, Valcárcel J. Reduced fidelity of branch point recognition and alternative splicing induced by the anti-tumor drug spliceostatin A. Genes Dev 2011;25:445-59. 33. Fan L, Lagisetti C, Edwards CC, Webb TR, Potter PM. Sudemycins, novel small molecule analogues of FR901464, induce alternative gene splicing. ACS Chem Biol 2011;6:582-9. 34. Massiello A, Roesser JR, Chalfant CE. SAP155 binds to ceramide-responsive RNA cis-element 1 and regulates the alternative 5′ splice site selection of Bcl-x pre-mRNA. FASEB J 2006;20:1680-2. 35. Paulsen RD, Soni DV, Wollman R, et al. A genome-wide siRNA screen reveals diverse cellular processes and pathways that mediate genome stability. Mol Cell 2009;35:228-39. Copyright © 2011 Massachusetts Medical Society.
an nejm app for iphone
The NEJM Image Challenge app brings a popular online feature to the smartphone. Optimized for viewing on the iPhone and iPod Touch, the Image Challenge app lets you test your diagnostic skills anytime, anywhere. The Image Challenge app randomly selects from 300 challenging clinical photos published in NEJM, with a new image added each week. View an image, choose your answer, get immediate feedback, and see how others answered. The Image Challenge app is available at the iTunes App Store.
2506
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
review article Mechanisms of Disease
Proprotein Convertases in Health and Disease Andrew W. Artenstein, M.D., and Steven M. Opal, M.D.
S
ecretory proteins, such as hormones, enzymes, and receptors, constitute a broad group of biochemically active molecules that are essential for cellular function. Post-translational processing of their precursor molecules, which occurs through endoproteolytic cleavage, results in the formation of biologically active secretory proteins at the intended target sites and represents an evolutionary mechanism through which complex species maintain homeostasis, nimbly responding to internal and external challenges. At least five classes of proteolytic enzymes have been identified, encompassing more than 66,000 protein sequences classified into 50 clans and 184 families in accordance with their catalytic mechanisms and evolutionary lineages.1 More than one third of proteolytic enzymes are serine proteases, named for the essential role serine plays at their active sites.2 Serine proteases are ubiquitous in nature and are grouped into 13 clans and 40 families in accordance with their structural homology, their substrate specificity, and the configuration at their active sites, the last comprising variably ordered histidine, aspartate, and serine residues — the “catalytic triad” — in nearly half the families.2 The chymotrypsin–trypsinlike serine proteases are the most abundant family in humans and are involved in myriad physiologic processes, including embryogenesis, digestion, hemostasis, immune responses, tissue repair, cell activation, and apoptosis. However, another, smaller family of serine proteases — the proprotein convertases — performs a variety of activation functions involving critical cellular pathways in humans and other organisms. The concept that physiologically important proteins may require activation by means of post-translational cleavage of an inactive precursor molecule was initially demonstrated by the discovery of proinsulin in 1967.3 A quarter of a century later, the discovery of the proprotein convertase furin — a mammalian processing enzyme analogous to the subtilisin-like kexin endoprotease in yeast — galvanized the search for other proprotein convertases and their specific substrates.4 Over the past two decades, these enzymes have been shown to play important roles in human homeostasis, as well as in a diverse array of pathophysiological states, including endocrinopathies, neoplastic diseases, infectious diseases, atherosclerosis, and neurodegenerative diseases. They therefore represent potentially fertile therapeutic targets for the treatment of an assortment of human diseases.
From the Center for Biodefense and Emerging Pathogens, Department of Medicine, Memorial Hospital of Rhode Island, Pawtucket; and the Warren Alpert Medical School of Brown University, Providence, RI. Address reprint requests to Dr. Artenstein at the Center for Biodefense and Emerging Pathogens, Memorial Hospital of Rhode Island, 111 Brewster St., Pawtucket, RI 02860, or at artenstein@ brown.edu. N Engl J Med 2011;365:2507-18. Copyright © 2011 Massachusetts Medical Society.
S t ruc t ur e a nd Bio chemis t r y The family of proprotein convertases comprises at least nine endoprotease enzymes. Seven of the nine are core members that are structurally and biochemically similar to each other and to the bacterial and yeast proteins — subtilisin and kexin, respectively — from which they were probably derived. They include proprotein convertase n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2507
The
n e w e ng l a n d j o u r na l
subtilisin/kexin type 1 (PCSK1, formerly PC1/3), PCSK2 (formerly proprotein convertase 2 [PC2]), furin, PCSK4 (formerly PC4), PCSK5 (formerly PC5/6), PCSK6 (formerly PACE4), and PCSK7 (formerly PC7). Although they vary slightly in size, they share a template of similarly constructed domains that serve discrete biochemical functions (Fig. 1).5 The two remaining known proprotein convertases, subtilisin/kexin isoenzyme 1/site-1 protease (SKI-1/S1P)6,7 and proprotein convertase subtilisin/kexin 9 (PCSK9),8 which were identified more recently than the core seven, are distinctly different from them, belonging to different subfamilies with distinct domain structures. Proprotein convertases are synthesized as inactive precursors — zymogens — that are chaperoned through the cell by their prodomains. The prodomains contribute to protein folding and activation,9 participating in the latter process as part of an intramolecular, autoproteolytic, initial cleavage step.10,11 This process is necessary for the egress of the core proprotein convertases from the endoplasmic reticulum,12 and it is followed by a pH- and calcium-dependent cleavage event in the trans-Golgi network–endosomal compartments that completes their activation.11 The proprotein convertases are generally calcium-dependent serine endoproteases that function within defined pH ranges, which correlate with their variable compartments of action.9,11 The protein-processing activity of these enzymes localizes to their highly conserved catalytic and P domains.13 Elucidation of the crystal structure of the processing domains of furin has led to the modeling of the structure of the other six core proprotein convertases, revealing a clustering of negatively charged amino acid residues within the catalytic domain.13,14 This arrangement probably accounts for their specificity for basic substrate motifs13,15; the matching of the charge distribution to discrete proprotein convertases may be a factor in the relative substrate selectivity of these molecules.14,16 Proprotein convertases are generally classified in accordance with their predilection for substrate cleavage site. The seven core enzymes activate precursor proteins through proteolysis at basic motifs generally requiring residues of arginine (R), lysine (K), or both flanking other amino acids [R/K-(X)n-R/K-R↓], with X representing an amino acid other than cysteine (C); n representing the number 0, 2, 4, or 6; and the downward-pointing 2508
of
m e dic i n e
arrow representing the cleavage site.17 However, there is some variation in preferred cleavage sites.18 SKI-1/S1P does not require basic residues for cleavage,7 and PCSK9 is unique in that it does not cleave in trans; it acts as its own substrate.19 With the exception of PCSK1 and PCSK2, which are localized to secretory granules, proprotein convertases process secreted proteins constitutively. Furin and other ubiquitously distributed proprotein convertases such as PCSK5 and PCSK7 process proproteins in a variety of cellular compartments through tightly regulated, dynamic trafficking and cycling among the trans-Golgi network, the cell surface, and endosomes (Ta ble 1).9,17,19-29 Proprotein convertases activate a broad range of distinct proteins (Table 2). Many of these enzymes have overlapping substrates and functions, and several are concurrently expressed in cells. With the exception of FURIN and PCSK6, which are closely linked on human chromosome 15, the genes encoding proprotein convertases are widely distributed throughout the human genome.16 These observations point to a system that is both redundant, conferring an evolutionary survival advantage, and differentiated, allowing for the diversity of responses required in complex organisms.30
Propro tein C on v er ta se s in He a lth a nd Dise a se Despite current knowledge of proprotein convertase structure and biochemistry, in vitro efforts to link individual proprotein convertases to specific substrates tend to be inconclusive, owing to the absence of a distinctive “cellular context” — a wide range of variable expression patterns, interacting molecules, and other contributors to the cellular microenvironment influence the function of these enzymes.16 For this reason, murine conditional gene-knockout and tissue-specific– knockout models have been used to illuminate the relative roles of proprotein convertases in homeostasis and the pathogenic consequences when their function is altered. These models are confounded to some extent by the variable expression of phenotype caused by substrate redundancy (Table 2), and they highlight the complexity of the interrelated networks of proprotein processing. Complete knockout of gene function in specific tissues is often required to generate an aberrant phenotype.16,31
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
mechanisms of disease
Prodomain
P Domain
“Intramolecular chaperone” functions: Directs compartment-specific activation Assists in folding of molecule Disassociates after second internal cleavage event May autoinhibit PCs in select circumstances
May stabilize acidic prodomain and catalytic domain Is required for catalytic activity; single mutation can inactivate PC 35–55% sequence identity to furin
D
H
Transmembrane Domain Directs PC sorting and transit control within cell compartments Directs cell-surface tethering Is present only in transmembrane PCs: furin, PCSK5, PCSK7
S
Catalytic Domain
Cysteine-rich Domain
Contains conserved aspartate (D), histidine (H), and serine (S) — the catalytic triad
Confers protein–protein interaction properties Directs cell-surface tethering Is present in furin, PCSK5, PCSK6, PCSK7
Allows negative charge clustering to complement basic residues at substrate consensus cleavage site 50–70% sequence identity to furin
Cytosolic Domain Directs PC sorting and transit control within cell compartments
Signal Peptide
Directs cell-surface tethering
Directs translocation into the endoplasmic reticulum
Is present only in transmembrane PCs: furin, PCSK5, PCSK7
Figure 1. Schematic Representation of the Structural Organization of the Seven Core Proprotein Convertases, with Functional Correlates. COLOR FIGURE The colored bar is a schematic illustration of the organization of the core proprotein convertases, divided according to protein domain. 12/09/11 Draft 8 The domains are stable units within each protein chain that have primary, secondary, and tertiary structures, each of which is associated Author Artenstein with specific functions. PC denotes proprotein convertase; PC subtilisin/kexin types 5, 6, and 7 are denoted byFig PCSK5 1 (formerly PC5/6), # PCSK6 (formerly PACE4), and PCSK7 (formerly PC7). Adapted from Molloy et al.5 Title ME
It is clear from studies of knockout mice that proprotein convertases play an important role in embryogenesis and development. Their absence or underexpression can affect multiple pathways of organogenesis (Table 3)31-42; mice that lack furin, Pcsk5, or Ski-1/S1p have an increased rate of embryonic mortality. Without adequate levels of Ski-1/S1p, embryos may not survive the earliest stages of development; defects in the formation of the primitive cardiovascular system and gut may be related to inadequate levels of the furin or Pcsk5 isoforms, which contribute to the regulation of cellular growth factors, adhesion molecules, and extracellular matrix proteins.31,38,40 Unlike nonspecific knockout models, those specifically targeted to inactivate furin in the liver do not result in adverse morphologic outcomes in transgenic mice; variable redundancy of processing has been noted for a number of hepaticderived proteins.39 The absence of Pcsk6 leads to substantial anatomical defects in axis develop-
n engl j med 365;26
DE Artist
Longo Knoper
AUTHOR PLEASE NOTE: ment and visceral orientation, but these embryos Figure has been redrawn and type has been reset Please check carefully remain largely viable.41
Issue date
12/29/11
Hormones and Endocrinopathies
Experimentally induced deficiency of Pcsk1 or Pcsk2 results in viable yet phenotypically altered mice with complex endocrinologic disturbances. Pcsk1-knockout animals have severe growth retardation associated with reduced levels of growth hormone–releasing hormone (GHRH) and growth hormone.32 Pcsk2 disruption is associated with a lesser effect on growth.37 Mice that are deficient in either Pcsk1 or Pcsk2 have variable expression of abnormalities in the differential processing of hormones in the proopiomelanocortin pathway, such as corticotropin and β-endorphin, and abnormalities related to neuroendocrine function.32,37 Both deficiencies are associated with impaired conversion of proinsulin and proglucagon to their active moieties, resulting in variable defects in glucose homeostasis.32,37
nejm.org
december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2509
The
n e w e ng l a n d j o u r na l
of
m e dic i n e
Table 1. Characteristics of Proprotein Convertases.* Proprotein Convertase
Subcellular Processing Site
Optimal pH and Ca2+ 2+
Ref. No.†
Trans Cleavage
Tissue Distribution
Function
PCSK1
Secretory granules
pH, 5.5–6.5; Ca , 2.5 mM
20
Yes
Neural, endocrine
Activates polypeptide prohormones
PCSK2
Secretory granules
pH, 5.5–6.0; Ca2+, 10 mM
21
Yes
Neural, endocrine
Activates polypeptide prohormones
Furin
TGN, cell surface through endosomes
pH, 6.0–8.5; Ca2+, 1–2 mM
22
Yes
Ubiquitous
Activates multiple mammalian and microbial precursor proteins
PCSK4
Not described
pH, 7.0; Ca2+, 2 mM
23
Yes
Germ cells
Activates proteins involved in sperm motility, reproduction
PCSK5A (isoform) PCSK5B (isoform)
TGN, secretory pH, 6.5; Ca2+, 5 mM granules TGN, cell surface pH, 6.5; Ca2+, 5 mM through endosomes (isoforms sort distinctly)25
24
Yes
Ubiquitous
Activates ECM proteins
24
Yes
Ubiquitous
Activates ECM proteins
PCSK6
TGN, cell surface
pH, 7.5–8.5; Ca2+, 2 mM
26
Yes
Ubiquitous
Activates ECM proteins
PCSK7
TGN, cell surface through endosomes
pH. 6.0–7.0; Ca2+, 1–2 mM
27
Yes
Ubiquitous
Activates multiple precursors, furin overlap
SKI-1/S1P
TGN, cell surface through endosomes
pH, 6.5; Ca2+, 2 mM
28
Yes
Ubiquitous
Processes membrane-bound transcription factors involved in lipid metabolism
PCSK9
TGN, cell surface through endosomes
pH, 8.0–11.0; not dependent on Ca2+ concentration
29
No
Liver, intestine, Regulates plasma LDL levels kidney, centhrough increased degradatral nervous tion of LDL receptor proteins system
* There are seven core proprotein convertases that are structurally and biochemically similar: proprotein convertase subtilisin/kexin type 1 (PCSK1, formerly PC1/3), PCSK2 (formerly proprotein convertase 2 [PC2]), furin, PCSK4 (formerly PC4), PCSK5 (formerly PC5/6), PCSK6 (formerly PACE4), and PCSK7 (formerly PC7). The two remaining known proprotein convertases, subtilisin/kexin isoenzyme 1/site-1 protease (SKI-1/ S1P) and proprotein convertase subtilisin/kexin 9 (PCSK9), are distinctly different from the seven core proteins and belong to different subfamilies with distinct domain structures. ECM denotes extracellular matrix, LDL low-density lipoprotein, and TGN trans-Golgi network. † References cited are the sources of the pH and Ca2+ data.
Three cases of PCSK1 deficiency in humans have been identified, each associated with distinct, compound mutations in PCSK1.33-36 All three patients, who were not related, shared clinical phenotypic features: early-onset obesity; malabsorptive enteropathy, manifested as refractory, neonatal diarrhea in two cases; hypoadrenalism; abnormal glucose homeostasis; and impaired prohormone processing. The absence of growth retardation and the presence of profound obesity in humans were not described in previous murine models. However, a recently discovered mutation in murine Pcsk1 results in an obese phenotype,43 more closely mimicking that in humans and suggesting that the phenotype may partially depend on the relative efficiency of the enzyme on different substrates. Enzyme oligomerization may also contribute to its activity levels in secretory granules.44 2510
Cancer
Proprotein convertases are implicated in the pathogenesis of neoplastic transformation, proliferation, invasiveness, and metastasis (Fig. 2A). Numerous proven and predicted proprotein convertase substrates, such as growth factors, receptors, matrix metalloproteinases (MMPs), and adhesion molecules are known to be involved in these processes through a complex network of relationships.45 Various human cancers and tumor-cell lines show high levels of proprotein convertase expression that correlate with growth and invasiveness in experimental settings.45-47 Variabilities in the expression of PCSK1 and PCSK2 have been associated with tumors of neuroendocrine origin, including pheochromocytomas,48 pituitary adenomas,49 and carcinoids,50 and they are expressed at higher levels in small-cell carcinoma of the lung than are other proprotein
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
mechanisms of disease
Table 2. Selected Secretory Proteins Regulated by Proprotein Convertase Activation of Precursors.* Proprotein Convertase
Secretory Proteins
PCSK1
Growth hormone–releasing hormone, insulin, glucagon-like peptides, corticotropin, β-lipotropin, melanocyte-stimulating hormones, met-enkephalin, β-endorphin
PCSK2
Insulin, glucagon-like peptides, corticotropin, β-lipotropin, melanocyte-stimulating hormones, met-enkephalin, β-endorphin, somatostatin
Furin
Albumin, factor IX, protein C, von Willebrand factor, neurotrophins, adhesins, α- and β-secretases, tumor necrosis factor α (through intermediaries), transforming growth factor β, IGF-1, IGF-1 receptor, integrins, platelet-derived growth factor, vascular endothelial growth factors, matrix metalloproteinases, bone morphogenetic proteins, anthrax toxin, pseudomonas exotoxin A, aerolysin toxin, Shiga toxins, Clostridium septicum α-toxin, diphtheria toxin, various viral-coat proteins (human immunodeficiency virus type 1, Ebola virus, influenza hemagglutinin, measles virus, cytomegalovirus, respiratory syncytial virus, coronavirus causing severe acute respiratory syndrome)
PCSK4
Pituitary adenylate cyclase–activating polypeptide, IGF-2
Isoforms PCSK5A and PCSK5B
In vitro overlap with furin and PCSK7, growth differentiation factor 11
PCSK6
In vitro overlap with furin/PCSK7; neural adhesion molecule L1, matrix metalloproteinases, bone morphogenetic proteins
PCSK7
Various growth factors, some substrate overlap with furin
SKI-1/S1P
Sterol regulatory-element–binding proteins involved in lipid metabolism, brain- derived neurotrophic factor, surface glycoproteins for hemorrhagic fever viruses
PCSK9
Low-density lipoprotein receptor
* IGF-1 denotes insulin-like growth factor 1.
convertases.51,52 High expression levels of other, constitutively secreted proprotein convertases — most notably and consistently, furin, but also PCSK5, PCSK6, and PCSK7 — have been associated with cancers of the head and neck53 and the breast,54 as well as non–small-cell lung cancers52 and a variety of tumor-cell lines.55 Although the precise mechanisms involved in the acquisition of a malignant phenotype remain unclear, proprotein convertase activation of MMPs and other substrates not only enhances the invasive potential of cancers56 but also contributes to the malignant transformation of benign keratinocytes in PCSK6-transfected, nontumorigenic, murine skincell lines.57 Proprotein convertases may have important roles in tumor progression and metastasis through the complex interactions among their activated substrates.45,58 Tumor-cell proliferation is probably mediated by augmented proprotein convertase activation of transforming growth factor β (TGF-β), platelet-derived growth factor, and insulin-like growth factor 1 and its receptor,55,59 although it has also been proposed that TGF-β has tumor-suppression functions.45 Tumor progression and invasiveness may be enhanced by numerous factors: hypoxia-induced up-regulation of furin-
processing activity within the tumor microenvironment60; signal-transduction–mediated changes in cell adhesion through proprotein convertase activation of integrins and related adhesion molecules61; proprotein convertase activation of vascular endothelial growth factor, leading to angiogenesis and lymphangiogenesis, both of which are pivotal to neoplastic growth62; and increased expression of MMPs through activation of furin, PCSK6, and other ubiquitously distributed proprotein convertases.56,63 The proprotein convertase–activated MMP family of enzymes, including membrane type 1 MMP,64 ADAM (a disintegrin and metalloproteinase),65 and stromelysin 3,66 also promotes metastatic spread, by contributing to basement-membrane destruction and degradation of the extracellular matrix. Infectious Diseases
Because some pathogens adapt host protein-processing mechanisms to ensure their survival, proprotein convertases represent critical elements in the pathogenesis of certain infectious diseases. Three structurally distinct forms of bacterial toxins require furin-mediated activation. For singlechain toxins of the A (toxic-subunit)–B (bindingsubunit) type, cleavage occurs within endosomes;
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2511
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. Phenotypes of Human and Murine Mutations in Proprotein Convertases. Proprotein Convertase
Knockout or Mutant Mouse Phenotype Increased prenatal and perinatal mortality, dwarfism32
Early-onset obesity, hyperphagia, impaired glucose homeostasis, hypogonadotropic hypogonadism, secondary hypoadrenalism; malabsorptive diarrhea33-36
PCSK2
Chronic hypoglycemia, growth retardation37
Not described
Furin
Impaired axial rotation, ventral closure defects, embryonic death38; unlike non specific knockout, targeted inactivation in the liver is not associated with overt abnormalities39
Not described
PCSK4
Decreased fertility
Not described
PCSK5A, PCSK5B (isoforms)
Cardiac, tracheoesophageal, and anorectal defects; renal and palatal agenesis; axial skeletal and hind-limb malformations; pulmonary hypoplasia40; embryonic death31
PCSK6
Embryogenic defects in anteroposterior and left–right axial development, complex craniofacial and central nervous system defects41
Not described
PCSK7
No abnormal phenotype known42
Not described
SKI-1/S1P
Prevention of embryonic implantation, decreased cholesterol and fatty acid synthesis with liver disruption, embryonic death31
Not described
PCSK9
Increased hepatic cholesterol uptake
examples include pseudomonas exotoxin A,67 diphtheria toxin,68 and Shiga toxins.69 After cleavage, the toxic moiety translocates to the cytosol, where it exerts its deleterious effects on the host. Anthrax toxin is a binary exotoxin that requires proprotein convertase–mediated cleavage of its binding subunit, protective antigen (Fig. 2B).70 Although furin is thought to be the primary activator of protective antigen, PCSK6 may also perform this function.71 Furin also participates in the final assembly of a third type of bacterial toxin, exemplified by the pore-forming, lytic toxin aerolysin, which is produced by Aeromonas hydrophila. Aerolysin has the capacity to autodimerize on target-cell membranes and uses proprotein convertases to complete its multimeric pore structure.72 A broad range of pathogenic viruses hijack host proprotein convertases — specifically, furin — to cleave their envelope glycoproteins and thus generate mature proteins that are capable of fusion with host cell membranes. In human immunodeficiency virus type 1, furin processes the envelope glycoprotein 160 into glycoprotein 120 and the fusogenic component glycoprotein 4173; oth2512
Clinical-Deficiency Phenotype
PCSK1
Possible association with vertebral, anorectal, cardiac, tracheoesophageal, renal, and limb malformations and with caudal regression syndromes, including sacral agenesis, spinalcord anomalies, and pulmonary hypoplasia40
Hypercholesterolemia (mutational overexpression), hypocholesterolemia (mutational underexpression)31
er proprotein convertases may also process glycoprotein 160, although less efficiently. The hemagglutinin glycoprotein on the surface of influenza A viruses mediates host receptor binding and cell entry and requires proteolytic activation (Fig. 2C). Seasonal viruses and those with low pathogenicity are generally processed by host trypsin-like proteases, whereas highly pathogenic avian influenza viruses, including the H5N1 strains responsible for a highly lethal form of disease, are preferentially processed by ubiquitous proprotein convertases, such as furin.74 The virulence of influenza viruses directly correlates with the structures of their proprotein convertase cleavage motifs.75 Other examples of viruses that use host-derived proprotein convertases include respiratory syncytial virus, which requires two distinct furin-mediated cleavage events on its inactive precursor fusion protein to elicit viral fusion, entry, and cellto-cell syncytia formation76; measles, another paramyxovirus, which requires activation of its fusion protein by furin77; cytomegalovirus78; and the highly pathogenic filoviruses Ebola79 and Mar-
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
mechanisms of disease
burg,80 both of which cause lethal hemorrhagic fevers. The Reston subtype of Ebola, which has low pathogenicity in humans, lacks a consensus furin cleavage site.79 Furin activation of the human papillomavirus, which lacks an envelope, appears to be necessary after initial binding of the virus to the cell surface.81 Lipid Disorders and Atherosclerosis
PCSK9 enhances degradation of the hepatic lowdensity lipoprotein (LDL) receptor, but it does not act by means of intrinsic catalytic activity; instead, it relies on tight binding to the LDL receptor and the probable subsequent involvement of resident hydrolases.19,82 Because receptor-mediated endocytosis is the primary route of LDL cholesterol clearance, overexpression of PCSK9 increases circulating levels of LDL cholesterol.83 Single point mutations in PCSK9 that are associated with increased proprotein convertase function may result in familial autosomal dominant hyperlipidemia, which is associated with high plasma levels of LDL cholesterol and an increased risk of early myocardial infarction and stroke.84 Loss-of-function mutations, such as those in murine knockout models, result in elevated levels of LDL receptor, which lead to increased clearance of circulating lipoproteins and reduced cholesterol levels, a pathway enhanced by statins85; human correlates have been identified in persons with hypocholesterolemia and extremely low secretion of PCSK9.86,87 Such mutations, noted in 2.6% of black subjects in one large study, were associated with an 88% reduction in the risk of coronary heart disease over a 15-year period.87 Alzheimer’s Disease
Proprotein convertases have been associated with various neurodegenerative diseases — most notably, Alzheimer’s disease — through their indirect role in the development of senile plaques. β-Amyloid, the main component of these lesions, is produced by the endoproteolysis of amyloid precursor protein through a process catalyzed by the secretase enzymes, β-site amyloid precursor protein–cleaving enzyme 1 (BACE-1) and γ-secretase.88 Furin and other proprotein convertases activate the precursor of BACE-1, a prerequisite for the stepwise development of plaques.89 Additional data suggest a role for PCSK7 90 or furin91 in the opposing, α-secretase pathway that may attenuate the pathogenesis of Alzheimer’s disease.
Propro tein C on v er ta se s a s Ther a peu t ic Ta rge t s On the basis of their important contributions to various pathologic processes, proprotein convertases are viewed as potential therapeutic targets. Although nonspecific inhibition of proprotein convertase functions could theoretically be deleterious to homeostasis, the redundancy of proprotein convertase–substrate pairings may partially mitigate this effect.39 Specific, therapeutic targeting of proprotein convertases may therefore be a viable strategy; it is supported by in vitro data and by limited in vivo data. Biodefense
Many pathogens activated by proprotein convertases are agents of bioterrorism and emerging infectious diseases. Therapy targeted to proprotein convertases may therefore represent a universal countermeasure that could be deployed in tandem with other interventions, especially in cases of bioterrorism, in which a specific pathogen may not be immediately identified. The inhibition of furin with the use of a bioengineered α1-antitrypsin variant prevents viral envelope processing and prevents bacterial exotoxins from causing cell death in vitro; the inhibition of furin with the use of polyargininebased peptides improves the survival of mice treated with pseudomonas exotoxin A.92,93 However, cross-reactivity with other host enzymes and the nonpharmacologic properties of these inhibitor molecules limit their therapeutic potential.94 High-throughput screening has identified several dicoumarol derivatives that inhibit furinmediated cell-surface and intracellular processing.95 Because dicoumarols are established agents, they may provide a platform for the future development of drugs that inhibit proprotein convertases. People with inhalational anthrax frequently die despite the use of appropriate antimicrobial agents.96 One strategy for the treatment of anthrax involves the design of specific, competitive inhibitors of furin and other proprotein convertases that make use of peptides involved in the cleavage motif of the H5N1 influenza virus.94 In an inhalational model, postexposure prophylaxis with these compounds, combined with antimicrobial agents, protected 40% of mice from death.97
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2513
The
n e w e ng l a n d j o u r na l
A strategy that targets proprotein convertases in the treatment of systemic anthrax intoxication involves a family of endogenous furin inhibitors called inter-alpha-inhibitor proteins (IαIps).98 When IαIps were administered immediately before or up to 24 hours after mice were exposed to lethal spores, 70% of the treated mice survived.99 Because IαIps are physiologic molecules that exist in high concentrations in human plasma (400 to 800 mg per liter), they are probably safe. IαIps are rapidly down-regulated in inflammatory states; very low levels are found in patients who die from severe sepsis.100 Administration of IαIps has been shown to improve survival in both experimental models and pilot clinical studies of sepsis.101 They could be used as a broad-spectrum countermeasure in repletion doses, thus restoring physiologic levels and theoretically limiting the potential for the adverse effects that might accompany nonspecific and excessive inhibition of proprotein convertases induced by other, synthetic inhibitors. Metastatic Cancer
In vitro studies have revealed the various roles played by proprotein convertases in the malignant process and have also illuminated their potential as targets for therapy.45,47 Tumor growth and invasiveness were diminished in invasive lines of squamous-cell carcinoma transfected with a furin inhibitor and inoculated into mice with severe combined immunodeficiency.102 Similar results were obtained in vitro with the use of the chemically synthesized prodomain of furin as its cognate inhibitor.103 In vivo studies also support the potential role of strategies targeting proprotein convertases for the treatment of cancer. In one study, the inactivation of Furin in association with proto-oncogene activation delayed the development of salivary-gland tumors in mice,104 and in another study, involving mice carrying a heterozygous mutation that predisposed them to intestinal polyposis, the subgroup in which duodenumspecific Pcsk5 was inactivated had a higher tumor burden than the subgroup in which the gene remained intact, suggesting that Pcsk5 plays a protective role in intestinal tumorigenesis.105 Interventions targeting proprotein convertases may represent novel approaches to the prevention of tumor metastasis. Small-molecule inhibitors of these enzymes have been shown to reduce the incidence of metastases of skin tumors in a transgenic mouse model106 and to decrease cell motil2514
of
m e dic i n e
Figure 2 (facing page). Proprotein Convertases in Neoplastic and Infectious Diseases. In the processes of tumorigenesis and metastasis (Panel A), proprotein convertases cleave a variety of precursors of distinct growth factors, their receptors, adhesion molecules, and matrix metalloproteinases (MMPs). Each of these activated substrates has been shown to affect many of the pathways involved in tumor growth and progression. In anthrax toxin formation (Panel B), 83-kD protective antigen (PA 83), once bound to its receptor on the surface of the host target cell, is proteolytically activated by proprotein convertase–mediated cleavage of a 20-kD fragment, PA 20 . The remaining 63-kD fragment, PA 63 , then heptamerizes on lipid rafts, binds lethal factor or edema factor, and is subsequently internalized in the endosome by receptor-mediated endocytosis. At this point, in the acidic conditions within the endosome, the toxin components undergo the final stages of transformation and are released into the cytoplasm, where they exert their deleterious effects on the host. In cases of highly pathogenic influenza A virus infection (Panel C), mature hemagglutinin (HA) on the virion surface mediates viral attachment to the host cell, allowing viral entry and the transcription of viral RNA, after which viral proteins are synthesized in the rough endoplasmic reticulum. HA is synthesized as HA0, the immature form of the molecule, and is subsequently cleaved by host proprotein convertases into its mature forms — HA1 and HA2 — which are linked by disulfide bridges. Mature HA, along with other viral proteins, are then assembled into virions that bud from the cell surface. These next-generation virions are capable of attachment to host cell receptors bearing residues of sialic acid, reinitiating the infectious cycle. ADAM denotes a disintegrin and metalloproteinase, ADAM-TS ADAM with thrombospondin motifs, IGF-1 insulin-like growth factor 1, IGF-1R IGF-1 receptor, MT1-MMP membrane type 1 MMP, PDGF platelet-derived growth factor; TGF-β transforming growth factor β, and VEGF vascular endothelial growth factor. Panel B adapted from Young and Collier.70
ity, thereby reducing metastatic potential, in cell lines expressing precursor membrane type 1 MMP.107 In a study reported by Scamuffa et al., molecular targeting of proprotein convertase activity in colon-carcinoma cell lines resulted in reduced processing of multiple substrates of importance in tumor metastasis, including insulin-like growth factor 1, its receptor, and adhesion molecules, and a reduced capacity for the development of metastases to the liver in vivo.108 Hyperlipidemia
The recognition of an endogenous PCSK9 inhibitor, annexin A2,109 may lead to the development
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
mechanisms of disease
A
Tumorigenesis and Metastasis Pro-IGF-1R
Pro-IGF-1
Pro-PDGF
Pro-TGFβ
Pro-VEGF-C
Proprotein convertase
Proprotein convertase
Proprotein convertase
Proprotein convertase
Proprotein convertase
IGF-1R
IGF-1
PDGF
TGF-β
VEGF-C
Regulates furin
Proprotein convertase α Integrins Change cell adhesion Increase cell motility
Pro-MMP2
Proprotein convertase
Proprotein convertase
MT1-MMP ADAM ADAM-TS Stromelysin-3
MMP2
MMP2 integrins
Antiapoptotic
MMP
Possible tumor suppression
Destruction of basement membrane and degradation of extracellular matrix
Downstream signaling through receptor tyrosine kinases
Acquisition of malignant phenotype
B
Pro-MMP
Anthrax Toxin Formation Proprotein convertase
Invasion, angiogenesis, and lymphangiogenesis
Proliferation
Heptamerization
Metastasis
Edema factor
PA20 PA63 Lethal factor
PA83 Endosome Anthrax toxin receptor
H+ Edema factor Lethal factor
C
Cell membrane Receptor-mediated endocytosis
Translocation to cytoplasm
Influenza A Infection
Host-cell membrane
Hemagglutinin
New host-cell membrane
Next-generation virions
Neuraminidase Influenza virion
Viral assembly Mature hemagglutinin
–s–s– HA2 HA1
Cleavage at basic motif
Proprotein convertase
Immature HA0 hemagglutinin
Nucleus
Endosome
Transcription Viral ribonucleoprotein
Ribosome Rough endoplasmic reticulum COLOR FIGURE
Draft 7
n engl j med 365;26
nejm.org
december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
Author Fig #
Artenstein 2
Title ME DE Artist
Longo Knoper
12/09/11
2515
The
n e w e ng l a n d j o u r na l
of novel compounds that block the activity of this proprotein convertase. Recent data in mice showed a reduction in PCSK9 expression with the use of antisense oligonucleotides or RNA interference technology.110 A single injection of a humanized monoclonal antibody that disrupts the interaction of PCSK9 and LDL receptors resulted in a rapid decline in serum levels of LDL cholesterol to a nadir value that was 80% below baseline levels, which was sustained for nearly 2 weeks in cynomolgus monkeys.111 Such strategies, perhaps combined with the use of statins, may yield additional therapeutic options for hyperlipidemias.
of
m e dic i n e
C onclusions Proprotein convertases are a group of mammalian serine endoproteases activating a variety of secretory proteins that are essential for the maintenance of homeostasis. These enzymes are also critical determinants of a wide and evolving range of pathogenic states. In this regard, proprotein convertases may represent fertile targets for future therapeutic interventions.
Disclosure forms provided by the authors are available with the full text of this article at NEJM.org. We thank Ms. Katherine Bollesen and Ms. Margo Katz for providing administrative assistance with the manuscript.
References 1. Page MJ, Di Cera E. Serine peptidases:
classification, structure and function. Cell Mol Life Sci 2008;65:1220-36. 2. Di Cera E. Serine proteases. IUBMB Life 2009;61:510-5. 3. Steiner DF, Cunningham D, Spigelman L, Aten B. Insulin biosynthesis: evidence for a precursor. Science 1967;157: 697-700. 4. Van de Ven WJM, Voorberg J, Fontijn R, et al. Furin is a subtilisin-like proprotein processing enzyme in higher eukaryotes. Mol Biol Rep 1990;14:265-75. 5. Molloy SS, Anderson ED, Jean F, Thomas G. Bi-cycling the furin pathway: from TGN localization to pathogen activation and embryogenesis. Trends Cell Biol 1999;9:28-35. 6. Sakai J, Rawson RB, Espenshade PJ, et al. Molecular identification of the sterolregulated luminal protease that cleaves SREBPs and controls lipid composition of animal cells. Mol Cell 1998;2:505-14. 7. Seidah NG, Mowla SJ, Hamelin J, et al. Mammalian subtilisin/kexin isoenzyme SKI-1: a widely expressed proprotein convertase with a unique cleavage specificity and cellular localization. Proc Natl Acad Sci U S A 1999;96:1321-6. 8. Seidah NG, Benjannet S, Wickham L, et al. The secretory proprotein convertase neural apoptosis-regulated convertase 1 (NARC-1): liver regeneration and neuronal differentiation. Proc Natl Acad Sci U S A 2003;100:928-33. 9. Thomas G. Furin at the cutting edge: from protein traffic to embryogenesis and disease. Nat Rev Mol Cell Biol 2002;3:75366. 10. Leduc R, Molloy SS, Thorne BA, Thomas G. Activation of human furin precursor processing endoprotease occurs by an intramolecular autoproteolytic cleavage. J Biol Chem 1992;267:14304-8. 11. Anderson ED, Molloy SS, Jean F, Fei H, Shimamura S, Thomas G. The ordered and compartment-specific autoproteolytic removal of the furin intramolecular chap-
2516
erone is required for enzyme activation. J Biol Chem 2002;277:12879-90. 12. Creemers JWM, Vey M, Schäfer W, et al. Endoproteolytic cleavage of its propeptide is a prerequisite for efficient transport of furin out of the endoplasmic reticulum. J Biol Chem 1995;270:2695-702. 13. Henrich S, Cameron A, Bourenkov GP, et al. The crystal structure of the proprotein processing proteinase furin explains its stringent specificity. Nat Struct Biol 2003;10:520-6. [Erratum, Nat Struct Biol 2003;10:669.] 14. Henrich S, Lindberg I, Bode W, Than ME. Proprotein convertase models based on the crystal structures of furin and kexin: explanation of their specificity. J Mol Biol 2005;345:211-27. 15. Creemers JWM, Siezen RJ, Roebroek AJM, Ayoubi TAY, Huylebroeck D, Van de Ven WJM. Modulation of furin-mediated proprotein processing activity by sitedirected mutagenesis. J Biol Chem 1993; 268:21826-34. 16. Creemers JWM, Khatib AM. Knock-out mouse models of proprotein convertases: unique functions or redundancy? Front Biosci 2008;13:4960-71. 17. Seidah NG, Chrétien M. Proprotein and prohormone convertases: a family of subtilases generating diverse bioactive polypeptides. Brain Res 1999;848:45-62. 18. Duckert P, Brunak S, Blom N. Prediction of proprotein convertase cleavage sites. Protein Eng Des Sel 2004;17:107-12. 19. Seidah NG, Mayer G, Zaid A, et al. The activation and physiological functions of the proprotein convertases. Int J Biochem Cell Biol 2008;40:1111-25. 20. Jean F, Basak A, Rondeau N, et al. Enzymic characterization of murine and human prohormone convertase-1 (mPC1 and hPC1) expressed in mammalian GH4C1 cells. Biochem J 1993;292:891-900. 21. Shennan KIJ, Taylor NA, Jermany JL, Matthews G, Docherty K. Differences in pH optima and calcium requirements for maturation of the prohormone convertas-
es PC2 and PC3 indicates different intracellular locations for these events. J Biol Chem 1995;270:1402-7. 22. Molloy SS, Bresnahan PA, Leppla SH, Klimpel KR, Thomas G. Human furin is a calcium-dependent serine endoprotease that recognizes the sequence Arg-X-X-Arg and efficiently cleaves anthrax toxin protective antigen. J Biol Chem 1992;267: 16396-402. 23. Basak A, Touré BB, Lazure C, Mbikay M, Chrétien M, Seidah NG. Enzymic characterization in vitro of recombinant proprotein convertase PC4. Biochem J 1999; 343:29-37. 24. Cain BM, Vishnuvardhan D, Wang W, et al. Production, purification, and characterization of recombinant prohormone covertase 5 from baculovirus-infected insect cells. Protein Expr Purif 2002;24:22733. 25. De Bie I, Marcinkiewicz M, Malide D, et al. The isoforms of proprotein convertase PC5 are sorted to different subcellular compartments. J Cell Biol 1996;135: 1261-75. 26. Tsuji A, Hashimoto E, Ikoma T, et al. Inactivation of proprotein convertase, PACE4, by a1-antitrypsin Portland (a1PDX), a blocker of proteolytic activation of bone morphogenetic protein during embryogenesis: evidence that PACE4 is able to form an SDS-stable acyl intermediate with a1-PDX. J Biochem 1999;126:591-603. 27. Munzer JS, Basak A, Zhong M, et al. In vitro characterization of the novel proprotein convertase PC7. J Biol Chem 1997;272:19672-81. [Erratum, J Biol Chem 1997;272:30589.] 28. Touré BB, Munzer JS, Basak A, et al. Biosynthesis and enzymatic characterization of human SKI-1/S1P and the processing of its inhibitory prosegment. J Biol Chem 2000;275:2349-58. 29. Naureckiene S, Ma L, Sreekumar K, et al. Functional characterization of Narc 1, a novel proteinase related to proteinase K. Arch Biochem Biophys 2003;420:55-67.
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
mechanisms of disease 30. Kirschner M, Gerhart J. Evolvability.
Proc Natl Acad Sci U S A 1998;95:8420-7. 31. Scamuffa N, Calvo F, Chrétien M, Seidah NG, Khatib AM. Proprotein convertases: lessons from knockouts. FASEB J 2006;20:1954-63. 32. Zhu X, Zhou A, Dey A, et al. Disruption of PC1/3 expression in mice causes dwarfism and multiple neuroendocrine peptide processing defects. Proc Natl Acad Sci U S A 2002;99:10293-8. 33. O’Rahilly S, Gray H, Humphreys PJ, et al. Impaired processing of prohormones associated with abnormalities of glucose homeostasis and adrenal function. N Engl J Med 1995;333:1386-90. 34. Jackson RS, Creemers JWM, Ohagi S, et al. Obesity and impaired prohormone processing associated with mutations in the human prohormone convertase 1 gene. Nat Genet 1997;16:303-6. 35. Jackson RS, Creemers JWM, Farooqi IS, et al. Small-intestinal dysfunction accompanies the complex endocrinopathy of human proprotein convertase 1 deficiency. J Clin Invest 2003;112:1550-60. 36. Farooqi IS, Volders K, Stanhope R, et al. Hyperphagia and early-onset obesity due to a novel homozygous missense mutation in prohormone convertase 1/3. J Clin Endocrinol Metab 2007;92:3369-73. 37. Furuta M, Yano H, Zhou A, et al. Defective prohormone processing and altered pancreatic islet morphology in mice lacking active SPC2. Proc Natl Acad Sci U S A 1997;94:6646-51. 38. Roebroek AJM, Umans L, Pauli IGL, et al. Failure of ventral closure and axial rotation in embryos lacking the proprotein convertase furin. Development 1998;125: 4863-76. 39. Roebroek AJ, Taylor NA, Louagie E, et al. Limited redundancy of the proprotein convertase furin in mouse liver. J Biol Chem 2004;279:53442-50. 40. Szumska D, Pieles G, Essalmani R, et al. VACTERL/caudal regression/Currarino syndrome-like malformations in mice with mutation in the proprotein convertase Pcsk5. Genes Dev 2008;22:1465-77. 41. Constam DB, Robertson EJ. SPC4/ PACE4 regulates a TGFβ signaling network during axis formation. Genes Dev 2000;14:1146-55. 42. Constam DB, Calfon M, Robertson EJ. SPC4, SPC6, and the novel protease SPC7 are coexpressed with bone morphogenetic proteins at distinct sites during embryogenesis. J Cell Biol 1996;134:181-91. 43. Lloyd DJ, Bohan S, Gekakis N. Obesity, hyperphagia and increased metabolic efficiency in Pc1 mutant mice. Hum Mol Genet 2006;15:1884-93. [Erratum, Hum Mol Genet 2008;17:3435.] 44. Hoshino A, Kowalska D, Jean F, Lazure C, Lindberg I. Modulation of PC1/3 activity by self-interaction and substrate binding. Endocrinology 2011;152:1402-11. 45. Bassi DE, Fu J, de Cicco RL, Klein-
Szanto AJP. Proprotein convertases: “master switches” in the regulation of tumor growth and progression. Mol Carcinog 2005;44:151-61. 46. Bassi DE, Mahloogi H, Lopez De Cicco R, Klein-Szanto A. Increased furin activity enhances the malignant phenotype of human head and neck cancer cells. Am J Pathol 2003;162:439-47. 47. Mercapide J, Lopez De Cicco R, Bassi DE, Castresana JS, Thomas G, Klein-Szanto AJ. Inhibition of furin-mediated processing results in suppression of astrocytoma cell growth and invasiveness. Clin Cancer Res 2002;8:1740-6. 48. Konoshita T, Gasc JM, Villard E, et al. Expression of PC2 and PC1/PC3 in human pheochromocytomas. Mol Cell Endocrinol 1994;99:307-14. 49. Takumi I, Steiner DF, Sanno N, Teramoto A, Osamura RY. Localization of prohormone convertases 1/3 and 2 in the human pituitary gland and pituitary adenomas: analysis by immunohistochemistry, immunoelectron microscopy, and laser scanning microscopy. Mod Pathol 1998; 11:232-8. 50. Kajiwara H, Itoh Y, Itoh J, Yasuda M, Osamura RY. Immunohistochemical expressions of prohormone convertase (PC) 1/3 and PC2 in carcinoids of various organs. Tokai J Exp Clin Med 1999;24:13-20. 51. Creemers JWM, Roebroek AJM, Van de Ven WJM. Expression in human lung tumor cells of the proprotein processing enzyme PC1/PC3: cloning and primary sequence of a 5kb cDNA. FEBS Lett 1992; 300:82-8. 52. Mbikay M, Sirois F, Yao J, Seidah NG, Chrétien M. Comparative analysis of expression of the proprotein convertases furin, PACE4, PC1 and PC2 in human lung tumours. Br J Cancer 1997;75:1509-14. 53. Bassi DE, Mahloogi H, Al-Saleem L, Lopez De Cicco R, Ridge JA, Klein-Szanto AJP. Elevated furin expression in aggressive human head and neck tumors and tumor cell lines. Mol Carcinog 2001;31: 224-32. 54. Cheng M, Watson PH, Paterson JA, Seidah N, Chrétien M, Shiu RPC. Pro-protein convertase gene expression in human breast cancer. Int J Cancer 1997;71:96671. 55. Khatib AM, Siegfried G, Prat A, et al. Inhibition of proprotein convertases is associated with loss of growth and tumorigenicity of HT-29 human colon carcinoma cells. J Biol Chem 2001;276:30686-93. 56. Hubbard FC, Goodrow TL, Liu SC, et al. Expression of PACE4 in chemically induced carcinomas is associated with spindle cell tumor conversion and increased invasive ability. Cancer Res 1997;57:5226-31. 57. Mahloogi H, Bassi DE, Klein-Szanto AJP. Malignant conversion of non-tumorigenic murine skin keratinocytes overexpressing PACE4. Carcinogenesis 2002;23: 565-72.
58. Pesu M, Watford WT, Wei L, et al.
T cell-expressed proprotein convertase furin is essential for maintenance of peripheral tolerance. Nature 2008;455:24650. 59. Dubois CM, Blanchette F, Laprise MH, Luduc R, Grondin F, Seidah NG. Evidence that furin is an authentic transforming growth factor-β1-converting enzyme. Am J Pathol 2001;158:305-16. 60. Arsenault D, Lucien F, Dubois CM. Hypoxia enhances cancer cell invasion through relocalization of the proprotein convertase furin from the trans-Golgi network to the cell surface. J Cell Physiol 2011 April 18 (Epub ahead of print). 61. Berthet V, Rigot V, Champion S, et al. Role of endoproteolytic processing in the adhesive and signaling functions of αvβ5 integrin. J Biol Chem 2000;275:33308-13. 62. Siegfried G, Basak A, Cromlish JA, et al. The secretory proprotein convertases furin, PC5, and PC7 activate VEGF-C to induce tumorigenesis. J Clin Invest 2003; 111:1723-32. 63. Bassi DE, Mahloogi H, Klein-Szanto AJP. The proprotein convertases furin and PACE4 play a significant role in tumor progression. Mol Carcinog 2000;28:63-9. 64. Yana I, Weiss SJ. Regulation of membrane type-1 matrix metalloproteinase activation by proprotein convertases. Mol Biol Cell 2000;11:2387-401. 65. Primakoff P, Myles DG. The ADAM gene family: surface proteins with adhesion and protease activity. Trends Genet 2000;16:83-7. 66. Pei D, Weiss SJ. Furin-dependent intracellular activation of the human stromelysin-3 zymogen. Nature 1995;375:244-7. 67. Gu M, Gordon VM, Fitzgerald DJP, Leppla SH. Furin regulates both the activation of Pseudomonas exotoxin A and the quantity of the toxin receptor expressed on target cells. Infect Immun 1996;64:524-7. 68. Tsuneoka M, Nakayama K, Hatsuzawa K, Komada M, Kitamura N, Mekada E. Evidence for involvement of furin in cleavage and activation of diphtheria toxin. J Biol Chem 1993;268:26461-5. 69. Lea N, Lord M, Roberts LM. Proteolytic cleavage of the A subunit is essential for maximal cytotoxicity of Escherichia coli 0157:H7 Shiga-like toxin-1. Microbiology 1999;145:999-1004. 70. Young JAT, Collier RJ. Anthrax toxin: receptor binding, internalization, pore formation, and translocation. Annu Rev Biochem 2007;76:243-65. 71. Gordon VM, Rehemtulla A, Leppla SH. A role for PACE4 in the proteolytic activation of anthrax toxin protective antigen. Infect Immun 1997;65:3370-5. 72. Abrami L, Fivaz M, Decroly E, et al. The pore-forming toxin proaerolysin is activated by furin. J Biol Chem 1998;273: 32656-61. 73. Hallenberger S, Bosch V, Angliker H, Shaw E, Klenk HD, Garten W. Inhibition
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2517
mechanisms of disease of furin-mediated cleavage activation of HIV-1 glycoprotein gp160. Nature 1992; 360:358-61. 74. Decha P, Rungrotmongkol T, Intharathep P, et al. Source of high pathogenicity of an avian influenza virus H5N1: why H5 is better cleaved by furin. Biophys J 2008; 95:128-34. 75. Hatta M, Gao P, Halfmann P, Kawaoka Y. Molecular basis for high virulence of Hong Kong H5N1 influenza A viruses. Science 2001;293:1840-2. 76. González-Reyes L, Ruiz-Argüello MB, García-Barreno B, et al. Cleavage of the human respiratory syncytial virus fusion protein at two distinct sites is required for activation of membrane fusion. Proc Natl Acad Sci U S A 2001;98:9859-64. 77. Watanabe M, Hirano A, Stenglein S, Nelson J, Thomas G, Wong TC. Engineered serine protease inhibitor prevents furin-catalyzed activation of the fusion glycoprotein and production of infectious measles virus. J Virol 1995;69:3206-10. 78. Jean F, Thomas L, Molloy SS, et al. A protein-based therapeutic for human cytomegalovirus infection. Proc Natl Acad Sci U S A 2000;97:2864-9. 79. Volchkov VE, Feldmann H, Volchkova VA, Klenk HD. Processing of the Ebola virus glycoprotein by the proprotein convertase furin. Proc Natl Acad Sci U S A 1998;95:5762-7. 80. Volchkov VE, Volchkova VA, Ströher U, et al. Proteolytic processing of Marburg virus glycoprotein. Virology 2000; 268:1-6. 81. Day PM, Schiller JT. The role of furin in papillomavirus infection. Future Microbiol 2009;4:1255-62. 82. McNutt MC, Lagace TA, Horton JD. Catalytic activity is not required for secreted PCSK9 to reduce low density lipoprotein receptors in HepG2 cells. J Biol Chem 2007;282:20799-803. 83. Maxwell KN, Breslow JL. Adenoviralmediated expression of Pcsk9 in mice results in a low-density lipoprotein receptor knockout phenotype. Proc Natl Acad Sci U S A 2004;101:7100-5. 84. Abifadel M, Varret M, Rabès JP, et al. Mutations in PCSK9 cause autosomal dominant hypercholesterolemia. Nat Genet 2003;34:154-6. 85. Rashid S, Curtis DE, Garuti R, et al. Decreased plasma cholesterol and hypersensitivity to statins in mice lacking Pcsk9. Proc Natl Acad Sci U S A 2005;102:5374-9. 86. Zhao Z, Tuakli-Wosornu Y, Lagace TA, et al. Molecular characterization of lossof-function mutations in PCSK9 and identification of a compound heterozygote. Am J Hum Genet 2006;79:514-23. 87. Cohen JC, Boerwinkle E, Mosley TH Jr, Hobbs HH. Sequence variations in
2518
PCSK9, low LDL, and protection against coronary heart disease. N Engl J Med 2006;354:1264-72. 88. Vassar R, Kovacs DM, Yan R, Wong PC. The β-secretase enzyme BACE in health and Alzheimer’s disease: regulation, cell biology, function, and therapeutic potential. J Neurosci 2009;29:1278794. 89. Bennett BD, Denis P, Haniu M, et al. A furin-like convertase mediates propeptide cleavage of BACE, the Alzheimer’s β-secretase. J Biol Chem 2000;275:377127. [Erratum, J Biol Chem 2001;276:15561.] 90. Lopez-Perez E, Seidah NG, Checler F. Proprotein convertase activity contributes to the processing of the Alzheimer’s β-amyloid precursor protein in human cells: evidence for a role of the prohormone convertase PC7 in the constitutive α-secretase pathway. J Neurochem 1999; 73:2056-62. 91. Hwang EM, Kim SK, Sohn JH, et al. Furin is an endogenous regulator of α-secretase associated APP processing. Biochem Biophys Res Commun 2006; 349:654-9. 92. Jean F, Stella K, Thomas L, et al. α1Antitrypsin Portland, a bioengineered serpin highly selective for furin: application as an antipathogenic agent. Proc Natl Acad Sci U S A 1998;95:7293-8. 93. Sarac MS, Cameron A, Lindberg I. The furin inhibitor hexa-D-arginine blocks the activation of Pseudomonas aeruginosa exotoxin A in vivo. Infect Immun 2002; 70:7136-9. 94. Shiryaev SA, Remacle AG, Ratnikov BI, et al. Targeting host cell furin proprotein convertases as a therapeutic strategy against bacterial toxins and viral pathogens. J Biol Chem 2007;282:20847-53. 95. Komiyama T, Coppola JM, Larsen MJ, et al. Inhibition of furin/proprotein convertase-catalyzed surface and intracellular processing by small molecules. J Biol Chem 2009;284:15729-38. 96. Artenstein AW. Anthrax: from antiquity to answers. J Infect Dis 2007;195:4713. 97. Remacle AG, Gawlik K, Golubkov VS, et al. Selective and potent furin inhibitors protect cells from anthrax without significant toxicity. Int J Biochem Cell Biol 2010;42:987-95. 98. Opal SM, Artenstein AW, Cristofaro PA, et al. Inter-alpha-inhibitor proteins are endogenous furin inhibitors and provide protection against experimental anthrax intoxication. Infect Immun 2005; 73:5101-5. 99. Opal SM, Lim YP, Cristofaro P, et al. Inter-α inhibitor proteins: a novel therapeutic strategy for experimental anthrax infection. Shock 2011;35:42-4.
100. Lim YP, Bendelja K, Opal SM, Siry-
aporn E, Hixson DC, Palardy JE. Correlation between mortality and the levels of inter-alpha inhibitors in the plasma of patients with severe sepsis. J Infect Dis 2003;188:919-26. 101. Yang S, Lim YP, Zhou M, et al. Administration of human inter-α-inhibitors maintains hemodynamic stability and improves survival during sepsis. Crit Care Med 2002;30:617-22. 102. Bassi DE, Lopez De Cicco R, Mahloogi H, Zucker S, Thomas G, KleinSzanto AJP. Furin inhibition results in absent or decreased invasiveness and tumorigenicity of human cancer cells. Proc Natl Acad Sci U S A 2001;98:10326-31. 103. Basak A, Chen A, Scamuffa N, Mohottalage D, Basak S, Khatib AM. Blockade of furin activity and furin-induced tumor cells malignant phenotypes by the chemically synthesized human furin pro domain. Curr Med Chem 2010;17:221421. 104. De Vos L, DeClerq J, Rosas GG, et al. MMTV-cre-mediated fur inactivation concomitant with PLAG1 proto-oncogene activation delays salivary gland tumorigenesis in mice. Int J Oncol 2008;32:1073-83. 105. Sun X, Essalmani R, Seidah NG, Prat A. The proprotein convertase PC5/6 is protective against intestinal tumorigenesis: in vivo mouse model. Mol Cancer 2009;8:73. 106. Bassi DE, Zhang J, Cenna J, Litwin S, Cukierman E, Klein-Szanto AJP. Proprotein convertase inhibition results in decreased skin cell proliferation, tumorigenesis, and metastasis. Neoplasia 2010;12: 516-26. 107. Coppola JM, Bhojani MS, Ross BD, Rehemtulla A. A small-molecule furin inhibitor inhibits cancer cell motility and invasiveness. Neoplasia 2008;10:363-70. 108. Scamuffa N, Siegfried G, Bontemps Y, et al. Selective inhibition of proprotein convertases represses the metastatic potential of human colorectal tumor cells. J Clin Invest 2008;118:352-63. 109. Mayer G, Poirier S, Seidah NG. Annexin A2 is a C-terminal PCSK9-binding protein that regulates endogenous low density lipoprotein receptor levels. J Biol Chem 2008;283:31791-801. 110. Seidah NG. What lies ahead for the proprotein convertases? Ann N Y Acad Sci 2011;1220:149-61. 111. Chan JCY, Piper DE, Cao Q, et al. A proprotein convertase subtilisin/kexin type 9 neutralizing antibody reduces serum cholesterol in mice and nonhuman primates. Proc Natl Acad Sci U S A 2009; 106:9820-5. Copyright © 2011 Massachusetts Medical Society.
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
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
Transfusion Thresholds in FOCUS Paul J. Barr, Ph.D., and Karen E.M. Bailie, M.D., Ph.D. Even though red-cell transfusion is an accepted and widely used intervention, questions regarding which patients should receive transfusions and under what circumstances continue to spark debate. There has been limited evidence from clinical trials to inform policy. Meanwhile, concern about potential risks of blood transfusion and the costs of maintaining an adequate and safe blood supply1 have heightened interest in strategies to reduce the use of red-cell transfusion. These strategies include preoperative optimization of hemoglobin levels, the use of cell salvage during and after surgery, and pharmacologic interventions to reduce blood loss. In the Transfusion Trigger Trial for Functional Outcomes in Cardiovascular Patients Undergoing Surgical Hip Fracture Repair (FOCUS),2 Carson et al. describe another strategy: the use of lower hemoglobin thresholds to initiate transfusion. A recent Cochrane review1 of 17 trials comparing variously defined transfusion triggers, including a lower hemoglobin threshold (range, 7 to 9 g per deciliter) and a higher hemoglobin threshold (range, 9 to 12 g per deciliter), concluded that although there was good evidence that lower thresholds reduced blood use, available data were inadequate to determine whether lower thresholds had adverse effects on functional status or other major outcomes. Although the review suggested that it was likely that a restrictive hemoglobin threshold for transfusion was as safe as a liberal strategy, the results were heavily weighted by the Transfusion Requirements in Critical Care (TRICC) trial.3 The TRICC trial compared a threshold of 10 g per deciliter with a threshold of 7 g per deciliter in patients in the intensive care unit (ICU). Overall, 30-day rates of death were similar in the two groups, but in the predefined subgroups of pa2532
tients under the age of 55 years and those who were less critically ill, rates of death were significantly lower in the restrictive-strategy group. Complications occurring in the ICU were also similar overall, with the notable exception of a significantly higher number of cardiac events in the restrictive-strategy group. Thus, the overall results suggesting similar outcomes with a lower versus higher transfusion threshold might not be broadly generalizable beyond the ICU setting and particularly to patients with cardiac disease. The FOCUS trial was designed to address this question among patients who had a history of or risk factors for ischemic heart disease and were undergoing surgical repair of a fractured hip.2,4 This study population is an important one in which to pose this question, since such patients are likely to be transfused because of surgical blood loss and their age, and it is plausible that they may be compromised at lower hemoglobin levels.5-7 In this study, 2016 patients over the age of 50 years (mean, 81.6) were randomly assigned to two study groups once their postoperative hemoglobin level fell below 10 g per deciliter. In the liberal-strategy group, single-unit transfusions were given to restore and maintain a hemoglobin level above 10 g per deciliter. In the restrictivestrategy group, transfusions were given when the hemoglobin level fell below 8 g per deciliter. Transfusion for symptoms of anemia was permitted in both groups. The trial had good statistical power to detect a difference in the primary outcome (death or an inability to walk 10 ft without human assistance) at 60 days after randomization, which was ascertained through telephone calls by assessors who were unaware of study-group assignments in all but 17 participants. Although the restrictive-strategy group received
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
editorials
only half the number of transfusions administered in the liberal-strategy group, rates of the primary outcome did not differ significantly between the two groups, with 35% of patients in each group unable to walk unassisted, including 142 patients (76 of whom were in the liberalstrategy group) who had died. However, the average hemoglobin level in both groups was low by World Health Organization standards8 and could have limited functional recovery regardless of the transfusion strategy.6,7 There were no significant differences between groups among predefined secondary and tertiary outcomes, including inhospital myocardial events, other coexisting illnesses, and final discharge destination, but the study was not adequately powered to assess these outcomes. Although a significant between-group separation in hemoglobin levels was obtained, the absolute difference was on average only 1 g per deciliter. This difference may not have been clinically significant in this group of elderly patients and therefore may have contributed to the lack of detectable difference in the primary outcome between the two groups. Although the interpretation that the transfusion strategy makes no difference in functional outcome in patients without symptoms of anemia is likely to be correct, the same cannot be said with confidence of the lack of adverse effects, given the study’s insufficient power to assess these events. Nevertheless, the absolute numbers of adverse events in this high-risk group was relatively low; cardiovascular events, including myocardial ischemia, heart failure, transient ischemic attack, and stroke, each occurred in no more than 6% of patients in either group. This low event rate is reassuring and supports the authors’ conclusion that a postoperative hemoglobin threshold of 8 g per deciliter in the absence of symptomatic anemia appears to be acceptable in elderly patients with or at risk of ischemic heart disease. In using lower hemoglobin thresholds to guide transfusion, the risks of undertransfusion should not be overlooked. The decision to transfuse
should be guided by an assessment of individual patients on the basis of a combination of signs, symptoms, and laboratory measures, and not by a single hemoglobin level. In the FOCUS trial, transfusion for symptoms occurred more often in the restrictive-strategy group, and protocol violations resulting in additional transfusions (a total of 56) in this group may have reduced this frequency. Despite these caveats, the FOCUS trial provides new evidence to support the view that a more restrictive transfusion threshold in the absence of symptoms of anemia may be reasonable, including in elderly patients at risk for cardiovascular events. Such a policy would reduce exposure to allogeneic blood transfusion with its attendant risks and costs. Disclosure forms provided by the author are available with the full text of this article at NEJM.org. From the Centre of Health Improvement, Clinical and Practice Research Group, School of Pharmacy, Queen’s University Belfast, Belfast (P.J.B.); and West of Scotland Blood Transfusion Centre, Glasgow (K.E.M.B.) — both in the United Kingdom. This article (10.1056/NEJMe1110087) was published on December 14, 2011, at NEJM.org. 1. Carless PA, Henry DA, Carson JL, Hebert PPC, McClelland B,
Ker K. Transfusion thresholds and other strategies for guiding allogeneic red blood cell transfusion. Cochrane Database Syst Rev 2010;10:CD002042. 2. Carson JL, Terrin ML, Noveck H, et al. Liberal or restrictive transfusion in high-risk patients after hip surgery. N Engl J Med 2011;365:2453-62. 3. Hébert PC, Wells G, Blajchman MA, et al. A multicenter, randomized, controlled clinical trial of transfusion requirements in critical care. N Engl J Med 1999;340:409-17. [Erratum, N Engl J Med 1999;340:1056.] 4. Carson JL, Terrin M, Magaziner J, et al. Transfusion Trigger Trial for Functional Outcomes in Cardiovascular Patients Undergoing Surgical Hip Fracture Repair (FOCUS). Transfusion 2006; 46:2192-206. 5. Wang JK, Klein HG. Red blood cell transfusion in the treatment and management of anaemia: the search for the elusive transfusion trigger. Vox Sang 2010;98:2-11. 6. Spahn DR. Anemia and patient blood management in hip and knee surgery: a systematic review of the literature. Anesthesiology 2010;113:482-95. 7. Beghé C, Wilson A, Ershler WB. Prevalence and outcomes of anemia in geriatrics: a systematic review of the literature. Am J Med 2004;116:Suppl 7:3S-10S. 8. Iron deficiency anaemia: assessment, prevention, and control — a guide for programme managers. Geneva: World Health Organization, 2001 (WHO/NHD/01.3). Copyright © 2011 Massachusetts Medical Society.
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2533
n e w e ng l a n d j o u r na l
The
of
m e dic i n e
Mutations in RNA Splicing Machinery in Human Cancers Benjamin Ebert, M.D., Ph.D., and Olivier A. Bernard, Ph.D. Massively parallel sequencing of cancer genomes is revealing a panoramic view of the genetic drivers of human neoplasms. In this issue of the Journal, Wang et al.1 describe an analysis of the coding sequences of samples from 91 patients with chronic lymphocytic leukemia. The disease is characterized by the accumulation of mature B lymphocytes, and its genetic basis is being rapidly elucidated.2,3
Wang et al. leveraged the large number of samples studied to identify sets of genes that are critical to the development of chronic lymphocytic leukemia. By sequencing both the malignant lymphocytes (CD19+CD5+) and matched nonmalignant control DNA from each patient, the authors pinpointed mutations that occurred somatically. By identifying genes and pathways with recurrent mutations, they highlighted the
RNA polymerase II
DNA
Nucleosome
Splicing complex Transcription
ZRSR2 U1 snRNP
U2 snRNP
PRPF40B SF1 SF3A1 ESE
GURAGU
SF3B1
YNYURAY
SR proteins U2AF65
mRNA
U2AF35 SRSF2
Y (×10 to 12)
YAG
ESE
Branch point
Skipped exon
Retained intron
Retained exon
Figure 1. Schematic Representation of RNA Splicing. After the transcription of nucleosome-free DNA into pre-messenger RNA (pre-mRNA) by RNA polymerase II, the RNA is processed and spliced into mRNA. Components of the early splicing complex are illustrated. Serine- and arginine-rich (SR) proteins bind to the exonic C OLOR FIGURE splice enhancer (ESE). The U1 small nuclear ribonucleoprotein (snRNP) binds to the 5′ splice site, and the U2 snRNP binds to the branch Draft 4 12/07/2011 point. The U2 auxiliary factor (U2AF) binds to the pyrimidine the 3′ splice site. The genes encoding SF3B1, SF1, SF3A1, SRSF2, Author stretch Bernard and Ebert–e1111584 1 cancers or in chronic lymphocytic leukemia; these proteins are shown in # U2AF35, U2AF65, ZRSR2, and PRPF40B are mutated inFigmyeloid SF3B1 and Other Novel Cancer Title red. Three splicing outcomes are illustrated (retained exon, skipped exon,Lymphocytic and retained intron), each of which could be affected by muGenes in Chronic tations that alter the splicing complex. Y denotes a pyrimidineLeukemia nucleotide, N any nucleotide, and Y (×10 to 12) a stretch of 10 to 12 DE Longo pyrimidines. ME Artist
Koopman Williams
AUTHOR PLEASE NOTE:
Figure has been redrawn and type has been reset Please check carefully DNA
Splicing complex
RNA polymerase II
2534 mRNA
n engl j med 365;26
nejm.org
december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
editorials
developmental drivers of chronic lymphocytic leukemia. Finally, the authors elegantly integrated the mutation data with known biologic information to define five key pathways of chronic lymphocytic leukemia that are affected by mutation: DNA damage and cell-cycle control, Notch signaling, inflammatory pathways, Wnt signaling, and RNA splicing. The identification of mutations in genes involved in RNA splicing was highly unexpected, but it converges remarkably with recently published studies making use of genome sequencing in myelodysplastic syndromes.4,5 SF3B1, which encodes a core member of the U2 small nuclear ribonucleoprotein (U2 snRNP) complex, was mutated in 15% of the 91 patients in the current study, primarily as a recurrent, heterozygous missense mutation, K700E. A separate study published recently in the Journal reported SF3B1 mutations in 20% of patients with myelodysplastic syndromes and 65% of patients with refractory anemia and ring sideroblasts.4 Moreover, mutations have been reported in multiple components of the spliceosome in 45 to 85% of patients with myelodysplastic syndrome.5 SF3B1 mutations also occur in 1 to 5% of samples from a wide range of tumor types, which indicates that mutations in RNA splicing factors are a widespread cause of oncogenic transformation.4 The vast majority of human genes undergo RNA splicing after transcription, which means that mutations in the RNA splicing machinery could potentially alter the maturation of messenger RNA for most genes and the subsequent production of protein (Fig. 1). In addition, RNA splicing is linked to the epigenetic regulation of gene expression. In particular, SF3B1 has been reported to interact with the polycomb repressive complex, an important regulator of hematopoiesis.6 Genes encoding members of these complexes are mutated in hematologic cancers. The finding of SF3B1 mutations in both chronic lymphocytic leukemia and myelodysplastic syndromes resonates with the recent finding of TET2 mutations in both lymphoid and myeloid cancers.7 These developments raise the provocative possibility that SF3B1 mutations might in some cases occur initially in hematopoietic stem cells, with additional mutations then being acquired in either the lymphoid or the myeloid lineages and causing chronic lymphocytic leukemia or myelodysplastic syndromes, respectively.
Consistent with this hypothesis, stem cells from patients with chronic lymphocytic leukemia have recently been reported to be abnormally lymphoidprimed, a finding that suggests that chronic lymphocytic leukemia could also derive from a stem-cell defect.8 The genetic characterization of chronic lymphocytic leukemia has the potential to refine the molecular classification and estimation of prognosis for this disease. Patterns of genetic lesions, such as the association of SF3B1 mutations with deletion in chromosome 11q and with ATM mutations, provide clues about the molecular circuitry of chronic lymphocytic leukemia cells. The identification of mutations in genes encoding the RNA splicing machinery raises the intriguing possibility that the spliceosome could be a therapeutic target for the treatment of chronic lymphocytic leukemia and myelodysplastic syndromes.9,10 Disclosure forms provided by the authors are available with the full text of this article at NEJM.org. From the Division of Hematology, Brigham and Women’s Hospital, Boston (B.E.); and INSERM, Unité 985, Institut Gustave Roussy, Villejuif, France (O.A.B.). This article (10.1056/NEJMe1111584) was published on December 12, 2011, at NEJM.org. 1. Wang L, Lawrence MS, Wan Y, et al. SF3B1 and other novel
cancer genes in chronic lymphocytic leukemia. N Engl J Med 2011;365:2497-506. 2. Fabbri G, Rasi S, Rossi D, et al. Analysis of the chronic lymphocytic leukemia coding genome: role of NOTCH1 mutational activation. J Exp Med 2011;208:1389-401. 3. Puente XS, Pinyol M, Quesada V, et al. Whole-genome sequencing identifies recurrent mutations in chronic lymphocytic leukaemia. Nature 2011;475:101-5. 4. Papaemmanuil E, Cazzola M, Boultwood J, et al. Somatic SF3B1 mutation in myelodysplasia with ring sideroblasts. N Engl J Med 2011;365:1384-95. 5. Yoshida K, Sanada M, Shiraishi Y, et al. Frequent pathway mutations of splicing machinery in myelodysplasia. Nature 2011; 478:64-9. 6. Isono K, Mizutani-Koseki Y, Komori T, Schmidt-Zachmann MS, Koseki H. Mammalian polycomb-mediated repression of Hox genes requires the essential spliceosomal protein Sf3b1. Genes Dev 2005;19:536-41. 7. Quivoron C, Couronné L, Della Valle V, et al. TET2 inactivation results in pleiotropic hematopoietic abnormalities in mouse and is a recurrent event during human lymphomagenesis. Cancer Cell 2011;20:25-38. 8. Kikushige Y, Ishikawa F, Miyamoto T, et al. Self-renewing hematopoietic stem cell is the primary target in pathogenesis of human chronic lymphocytic leukemia. Cancer Cell 2011;20: 246-59. 9. Kaida D, Motoyoshi H, Tashiro E, et al. Spliceostatin A targets SF3b and inhibits both splicing and nuclear retention of pre-mRNA. Nat Chem Biol 2007;3:576-83. 10. Yokoi A, Kotake Y, Takahashi K, et al. Biological validation that SF3b is a target of the antitumor macrolide pladienolide. FEBS J 2011;278:4870-80. Copyright © 2011 Massachusetts Medical Society.
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2535
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
Adenocarcinoma in Barrett’s Esophagus To the Editor: In their population-based cohort study, Hvid-Jensen et al. (Oct. 13 issue)1 report an annual risk of esophageal adenocarcinoma of 0.12% among patients with Barrett’s esophagus. This annual risk increased to 0.26% when highgrade dysplasia was taken into account. Since these rates are about three times lower than the annual incidence rates usually reported,2 we question the current surveillance guidelines for patients with Barrett’s esophagus, which rely on these epidemiologic data. Moreover, several studies have reported a relationship between the length of the segment of Barrett’s esophagus and adenocarcinoma risk.2-4 Weston et al. reported that the prevalence of dysplasia or adenocarcinoma and the incidence of dysplasia in patients with traditional Barrett’s esophagus are significantly higher than in patients with a segment of Barrett’s esophagus that is shorter than 3 cm.5 For instance, a prediction of progression to esophageal adenocarcinoma has been reported to be 1.53 times as high in patients with a segment of Barrett’s esophagus that is 3 cm or longer as in patients with a segment of Barrett’s esophagus that is shorter than 3 cm (odds ratio, 1.53; 95% confidence interval, 1.34 to 1.74).4 Thus, we would appreciate knowing whether, in the cohort in the study by Hvid-Jensen et al., the incidence rate of cancer among patients with a segment of Barrett’s esophagus that is long (≥3 cm) was higher, warranting endoscopic surveillance. Maximilien Barret, M.D. Stanislas Chaussade, M.D., Ph.D. Romain Coriat, M.D.
1. Hvid-Jensen F, Pedersen L, Drewes AM, Sorensen HT, Funch-
Jensen P. Incidence of adenocarcinoma among patients with Barrett’s esophagus. N Engl J Med 2011;365:1375-83. 2. Yousef F, Cardwell C, Cantwell MM, Galway K, Johnston BT, Murray L. The incidence of esophageal cancer and high-grade dysplasia in Barrett’s esophagus: a systematic review and metaanalysis. Am J Epidemiol 2008;168:237-49. 3. Gatenby PA, Caygill CP, Ramus JR, Charlett A, Fitzgerald RC, Watson A. Short segment columnar-lined oesophagus: an underestimated cancer risk? A large cohort study of the relationship between Barrett’s columnar-lined oesophagus segment length and adenocarcinoma risk. Eur J Gastroenterol Hepatol 2007;19:969-75. 4. Weston AP, Sharma P, Mathur S, et al. Risk stratification of Barrett’s esophagus: updated prospective multivariate analysis. Am J Gastroenterol 2004;99:1657-66. 5. Weston AP, Krmpotich PT, Cherian R, Dixon A, Topalosvki M. Prospective long-term endoscopic and histological follow-up of short segment Barrett’s esophagus: comparison with traditional long segment Barrett’s esophagus. Am J Gastroenterol 1997;92:407-13.
The Authors Reply: Barret et al. address an important issue. Looking at the risk among subgroups in the population with Barrett’s esophagus (thereby identifying patients with a need for special surveillance) is an important part of future research. The presence of dysplasia, as shown in our study, and the influence of the length of Barrett’s segments in estimating the risk of adenocarcinoma of the esophagus are important factors. However, we are not able to shed light on the risk related to the length of segments of Barrett’s esophagus, since a reliable classification system that includes the length of the segment this week’s letters 2539 Adenocarcinoma in Barrett’s Esophagus 2540 Change in FEV1 over Time in COPD
Université Paris Descartes Paris, France No potential conflict of interest relevant to this letter was reported.
2541 Community Transmission of Oseltamivir- Resistant A(H1N1)pdm09 Influenza
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2539
The
n e w e ng l a n d j o u r na l
of Barrett’s esophagus was not implemented in Danish practice during the first part of our study. Furthermore, the C and M classification (based on the circumference and maximum extent of a lesion) is not registered in our nationwide database, and this information would need to be extracted from every single patient’s endoscopic description. We do agree that the question raised, as well
of
m e dic i n e
as other possible subclassifications (such as biomarkers), are of importance in future studies. Frederik Hvid-Jensen, M.D. Peter Funch-Jensen, M.D., Dr.Med.Sci. Aarhus University Hospital Aarhus, Denmark
[email protected] Since publication of their article, the authors report no further potential conflict of interest.
Change in FEV1 over Time in COPD To the Editor: Vestbo and colleagues (Sept. 29 issue)1 reported on the variability in the rate of the decline in forced expiratory volume in 1 second (FEV1) in patients with chronic obstructive pulmonary disease (COPD) from the cohort in the Evaluation of COPD Longitudinally to Identify Predictive Surrogate Endpoints study (ECLIPSE) (ClinicalTrials.gov number, NCT00292552) and stress the importance of smoking cessation to attenuate this decline. Although large drug trials have provided evidence for some attenuation of lung-function decline,2,3 we do not agree with the authors’ interpretation that intensive pharmacologic treatment resulted in the rather mild decline seen in the ECLIPSE cohort. Although the mild — or even reversed — decline could be due to regression to the mean in these patients with relatively severe COPD (i.e., the mean FEV1 was 48% of predicted), we think it reflects the 64% of participants who did not smoke or did not continue to smoke. Since smoking is known to be the main determinant of a decline in lung function, and smoking cessation is by far the most effective intervention to prevent both the development and the progression of COPD,3-5 the distribution of the rate of decline within the subgroups of smokers and nonsmokers would be of interest.
To the Editor: Vestbo et al. found a mean decline in FEV1 of 33±2 ml per year in patients with COPD, which is substantially less than the 64 ml per year observed by Fletcher and Peto in 1977.1 We would like information about the effect on the decline in the ECLIPSE cohort of pulmonary rehabilitation, which is recognized as an effective treatment for COPD.2 Some studies report that the decline in FEV1 in patients with COPD who undergo such rehabilitation is significantly lower than that achieved with standard drug treatment.3,4 Gian Galeazzo Riario Sforza, M.D. Cristoforo Incorvaia, M.D. Istituti Clinici di Perfezionamento Milan, Italy
[email protected] No potential conflict of interest relevant to this letter was reported.
Wouter D. van Dijk, M.D. Tjard R. Schermer, Ph.D. Radboud University Nijmegen Medical Center Nijmegen, the Netherlands
[email protected] No potential conflict of interest relevant to this letter was reported. 1. Vestbo J, Edwards LD, Scanlon PD, et al. Changes in forced
expiratory volume in 1 second over time in COPD. N Engl J Med 2011;365:1184-92. 2. Tashkin DP, Celli B, Senn S, et al. A 4-year trial of tiotropium
2540
in chronic obstructive pulmonary disease. N Engl J Med 2008;359: 1543-54. 3. Celli BR, Thomas NE, Anderson JA, et al. Effect of pharmacotherapy on rate of decline of lung function in chronic obstructive pulmonary disease: results from the TORCH study. Am J Respir Crit Care Med 2008;178:332-8. 4. Anthonisen NR, Connett JE, Kiley JP, et al. Effects of smoking intervention and the use of an inhaled anticholinergic bronchodilator on the rate of decline of FEV1. JAMA 1994;272:1497505. 5. Kuller LH, Ockene JK, Meilahn E, Wentworth DN, Svendsen KH, Neaton JD. Cigarette smoking and mortality. Prev Med 1991; 20:638-54.
1. Fletcher C, Peto R. The natural history of chronic airflow obstruction. Br Med J 1977;1:1645-8. 2. Nici L, Donner C, Wouters E, et al. American Thoracic Society/European Respiratory Society statement on pulmonary rehabilitation. Am J Respir Crit Care Med 2006;173:1390-413. 3. Cote CG, Celli BR. Pulmonary rehabilitation and the BODE index in COPD. Eur Respir J 2005;26:630-6. 4. Stav D, Raz M, Shpirer I. Three years of pulmonary rehabilitation inhibit the decline in airflow obstruction, improves exercise endurance time, and body-mass index, in chronic obstructive pulmonary disease. BMC Pulm Med 2009;9:26-30.
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
correspondence
The Author Replies: In their attempt to understand the reason for the lower-than-expected decline in FEV1, van Dijk and Schermer think it unlikely that drug treatment is an acceptable explanation. We concur, since such statements cannot be made based on observational data — as we explained in our article. Nevertheless, since most patients were cared for in specialist settings, our findings may not be valid for patients with COPD who were followed and managed differently. We agree that smoking cessation is essential. Our analyses showed that FEV1 declined more rapidly among current smokers than among former smokers (21±4 ml per year, P<0.001); however, there is substantial overlap between these two groups in terms of the distribution of the rate of this change. We believe that the linear decline over 3 years argues against a strong effect of regression toward the mean. Had there been a strong effect, it would have most strongly affected the first measurements, thus leading to nonlinear changes over time. We maintain that a revision of the general understanding of the natural history of COPD is needed in light of our findings and of those reported from other observational studies in COPD.1,2
Riario Sforza and Incorvaia wonder whether pulmonary rehabilitation could have influenced the decline in FEV1. We are unable to comment on this issue, since we have no detailed information on such rehabilitation before the patients entered the study or during follow-up. However, even if we had such data, the biases related to studying the effects of interventions in observational cohort studies would be as relevant for pulmonary rehabilitation as for pharmacologic treatment. Jørgen Vestbo, Dr.Med.Sc. Hvidovre Hospital Hvidovre, Denmark
[email protected]
for the ECLIPSE Study Investigators Since publication of his article, the author reports no further potential conflict of interest. 1. Casanova C, de Torres JP, Aguirre-Jaíme A, et al. The pro-
gression of chronic obstructive pulmonary disease is heterogeneous: the experience of the BODE cohort. Am J Respir Crit Care Med 2011 August 11 (Epub ahead of print). 2. Nishimura M, Makita H, Nagai K, et al. Annual change in pulmonary function and clinical phenotype in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2011 October 20 (Epub ahead of print).
Community Transmission of Oseltamivir-Resistant A(H1N1)pdm09 Influenza To the Editor: Oseltamivir-resistant prepandemic seasonal influenza A (H1N1) viruses with a H275Y neuraminidase substitution spread globally in 2008,1 reducing the effectiveness of oseltamivir.2 Although oseltamivir-resistant pandemic 2009 A (H1N1) viruses, now known as A(H1N1)pdm09, have been detected in persons receiving oseltamivir treatment, they have been detected in less than 1% of untreated patients in the community, and transmission has been documented only in closed settings or settings involving close contact with infected persons.3,4 We identified sustained community transmission of oseltamivir-resistant A(H1N1)pdm09 viruses in Australia. Reverse-transcriptase-polymerase-chain-reaction-positive A(H1N1)pdm09 viruses were obtained from 182 patients in emergency departments, intensive care units, and general practitioners’ offices in the Hunter New England region of New South Wales, Australia, between May and August 2011.
Viruses were analyzed for oseltamivir resistance. A total of 29 A(H1N1)pdm09 viruses (16%) contained the H275Y neuraminidase substitution responsible for oseltamivir resistance, and all were resistant to the adamantine class of antiviral agents. Culture was attempted for all specimens and yielded 90 isolates, of which 15 contained the H275Y substitution. A fluorescence-based neuraminidase inhibition assay5 of the H275Y isolates showed that the IC50 value (the concentration of drug required to inhibit neuraminidase activity by 50%) was 513 times higher for oseltamivir and 80 times higher for peramivir than the IC50 values associated with wild-type strains. However, these isolates remained fully sensitive to zanamivir. The frequency of H275Y variants was 4 in 50 viruses in June (8%), 20 in 85 viruses in July (24%), and 4 in 45 viruses in August (9%). Hemagglutinin and neuraminidase sequence analysis (Global Initiative on Sharing All Influenza Data accession numbers, EPI334765-334790
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.
2541
correspondence
Table 1. Characteristics of Patients with Oseltamivir-Resistant Influenza in the Hunter New England Health District, According to Hospital Records and Responses from Interviews with Patients and General Practitioners. Variable
Patients (N = 29)
Male sex
12 (41)
Pregnant
3 (10)
Asthma
6 (21)
Smoker
6 (21)
Influenza vaccination in 2011
3 (10)
Treated with oseltamivir before specimen obtained
1 (3)*
no. (%)
Cough
25 (86)
Fever
22 (76)
Hospital admission
7 (24)
Intensive care unit admission
0
Death
0
* This sample was not obtained from the patient with the earliest oseltamivirresistant case of influenza from the cluster.
tients had no known epidemiologic link with the other patients with oseltamivir-resistant influenza described here. As winter approaches in the Northern Hemisphere, it remains important to ensure that A(H1N1)pdm09 strains from early in the season are analyzed rapidly for any indication that this transmissible oseltamivir-resistant variant has spread. Aeron C. Hurt, Ph.D. World Health Organization (WHO) Collaborating Centre for Reference and Research on Influenza North Melbourne, VIC, Australia
[email protected]
Kate Hardie, M.P.H. Hunter New England Population Health Newcastle, NSW, Australia
Noelene J. Wilson, B.Med.Sc. Hunter Area Pathology Service Newcastle, NSW, Australia
Yi-Mo Deng, Ph.D. WHO Collaborating Centre for Reference and Research
and 335634-335637) showed that the resistant on Influenza strains were closely related (99.9 to 100% hem- North Melbourne, VIC, Australia agglutinin nucleotide similarity and 99.6 to Maggi Osbourn, B.N., Grad.Dip.Legal Studies Hunter New England Population Health 100% neuraminidase nucleotide similarity), sug- Newcastle, NSW, Australia gesting the spread of a single variant. Nicole Gehrig, B.Sc. Most patients lived within a 50-km (31-mi) Hunter Area Pathology Service radius of Newcastle, the seventh largest city in Newcastle, NSW, Australia Australia, and three patients lived 90, 150, and Anne Kelso, Ph.D. 490 km (56, 93, and 304 mi) away. Two geneti- WHO Collaborating Centre for Reference and Research cally related strains were detected elsewhere in on Influenza North Melbourne, VIC, Australia the state, including in Sydney. Supported by a grant from the Australian Government DepartAn ethics waiver was granted and authorized ment of Health and Ageing to the Melbourne WHO Collaboratunder the New South Wales Public Health Act of ing Centre for Reference and Research on Influenza. Members of 2010 for an urgent public health investigation. the investigational team are listed in the Supplementary AppenHospital records and interviews with patients dix, available with the full text of this letter at NEJM.org. Disclosure forms provided by the authors are available at and general practitioners revealed that only 1 of NEJM.org. the 29 patients with resistant influenza had re1. Meijer A, Lackenby A, Hungnes O, et al. Oseltamivir-resisceived oseltamivir before the influenza speci- tant influenza virus A (H1N1), Europe, 2007-08 season. Emerg men collection (Table 1). The ages were similar Infect Dis 2009;15:552-60. among the 29 patients with oseltamivir-resistant 2. Saito R, Sato I, Suzuki Y, et al. Reduced effectiveness of oseltamivir in children infected with oseltamivir-resistant influinfluenza (median, 31 years; range, 4 months to enza A (H1N1) viruses with His275Tyr mutation. Pediatr Infect 62 years) and the 153 patients with sensitivity to Dis J 2010;29:898-904. oseltamivir (median, 29 years; range, 1 month 3. Moore C, Galiano M, Lackenby A, et al. Evidence of personto-person transmission of oseltamivir-resistant pandemic influto 74 years). Five patients with oseltamivir-resis- enza A(H1N1) 2009 virus in a hematology unit. J Infect Dis tant influenza were children younger than 5 years 2011;203:18-24. of age. The resistant viruses remained antigeni- 4. Le QM, Wertheim HF, Tran ND, van Doorn HR, Nguyen TH, Horby P. A community cluster of oseltamivir-resistant cases of cally similar to the vaccine strain, including 2009 H1N1 influenza. N Engl J Med 2010;362:86-7. those from 3 patients who received the 2011 5. Hurt AC, Barr IG, Hartel G, Hampson AW. Susceptibility of influenza vaccine (Table 1). Four households had human influenza viruses from Australasia and South East Asia to the neuraminidase inhibitors zanamivir and oseltamivir. An2 affected patients each, and 2 other patients tiviral Res 2004;62:37-45. shared a short car journey. The remaining pa- Correspondence Copyright © 2011 Massachusetts Medical Society. 2542
n engl j med 365;26 nejm.org december 29, 2011
The New England Journal of Medicine Copyright © 2011 Massachusetts Medical Society. All rights reserved.