Sports Med 2011; 41 (12): 985-987 0112-1642/11/0012-0985/$49.95/0
ACKNOWLEDGEMENT
ª 2011 Adis Data Information BV. All rights reserved.
Dear Reader As we reach the final issue of Sports Medicine for 2011, we hope that you have found the articles published throughout the year to be both interesting and informative. The editors and publishing staff have appreciated the high quality of content contributed to the journal this year and look forward to keeping you up to date with topical issues in the field of sports medicine and the exercise sciences in 2012. The high quality of Adis Journals was further recognized in the new ISI impact factors (IFs) for 2010, with the majority of our titles making strong IF gains over 2009. The most impressive gains were made by Sports Medicine (IF 5.072), with a 63% increase making it the highest ranked title in its field, and PharmacoEconomics (3.44), increasing by 34% to become the leading health outcomes research journal. Drugs in R&D (1.707), re-launched in 2010 as Adis’ first fully open-access MEDLINE-indexed journal, increased by 26%. Other stand-out performers include CNS Drugs (4.497) and BioDrugs (4.192), with increases of 16% and 20%, respectively. Another significant milestone was reached by The Patient: PatientCentered Outcomes Research, with the journal achieving MEDLINE indexing this year. The Patient was launched by Adis in 2008 as the first journal dedicated to the needs, values and role of patients in healthcare decision making. Last, but not least, we would like to say a big thank you to all the authors who have contributed articles to Sports Medicine in the last 12 months. Without their hard work and diligence we would not have been able to publish the journal. The quality of published articles reflects also the significant time and effort dedicated by the peer reviewers who ensure that we continue to publish content of the highest possible standard. In addition to the members of our Honorary Editorial Board, we would like to thank the following individuals who acted as referees for articles in Sports Medicine in 2011: Chris R. Abbiss, Australia Julie Agel, USA Helaine H.M. Alessio, USA Ajmol Ali, New Zealand David G. Allen, Australia Duarte Araujo, Portugal Neil Armstrong, UK A´rni A´rnison, Iceland Makoto Ayabe, Japan Aaron Baggish, USA Karen L. Barker, UK Thomas J. Barstow, USA Cynthia Bartok, USA Christian John Barton, Australia
Alan M. Batterham, UK Anastasia Beneka, Greece Wilma F. Bergfeld, USA Gaston Beunen, Belgium Roland M. Biedert, Switzerland Francois Billaut, Australia Walter R. Bixby, USA J. Troy Blackburn, USA Anthony Blazevich, Australia Jennifer Blitvich, Australia Richard J. Bloomer, USA Danilo Sales Bocalini, Brazil Barry P. Boden, USA Nathalie Boisseau, France
986
Andrea Bosio, Italy Larry Bowers, USA David L. Brown, USA Wendy Brown, Australia Louise M. Burke, Australia Jose Antonio Lopez Calbet, Spain Maria C. Calo, Italy Clayton L. Camic, USA Vitor Oliveira Carvalho, Brazil Carlo Castagna, Italy Derwin King-Chung Chan, UK Allen Cheadle, USA Stephen S.S. Cheung, Canada W. Lee Childers, USA Manuel J. Coelho-e-Silva, Portugal Vernon G. Coffey, Australia David Cowan, UK R.M. Daly, Australia Paul F.G. de Clercq, Belgium Julie Demartini, USA David D. Docherty, Canada Alberto Dolci, Italy Simon T. Donell, UK Brian Duscha, USA Conrad P. Earnest, USA Andrew M. Edwards, New Zealand Nir Eynon, Israel Aurelio Faria, Portugal Ioannis G. Fatouros, Greece Oliver Faude, Switzerland Antonio J. Figueiredo, Portugal Daniel Tik-Pui Fong, China Carl C. Foster, USA Richard C. Franklin, Australia Colin W. Fuller, UK Belinda J. Gabbe, Australia Andrew W. Gardner, USA Mark M. Glaister, UK Paul Stephen Glazier, UK Letha Y. Griffin, USA Narcis Gusi, Spain Markus Hu¨bscher, Germany Tobias Hu¨fner, Germany Mike Hamlin, New Zealand Rod Havriluk, USA Mark M. Haykowsky, Canada Glyn Howatson, UK Robert Hristovski, Macedonia, The Former Yugoslav Republic Stanley Hui, Hong Kong Franco M. Impellizzeri, Italy Christopher D. Ingersoll, USA W. Jelkmann, Germany Kirsten K.L. Johansen, USA Andrew M. Jones, UK
ª 2011 Adis Data Information BV. All rights reserved.
Lee W. Jones, USA Morgan H. Jones, USA Toivo Jurimae, Estonia Trine Karlsen, Norway Kristine A. Karlson, USA Georgios Karnatzikos, Italy Wolfgang Kemmler, Germany John G. Kennedy, USA Deborah Anne Kerr, Australia Marcus W. Kilpatrick, USA Arnold W. Klein, USA Robert R.R. Kraemer, USA William J. Kraemer, USA Harm H. Kuipers, the Netherlands Gregory G.L. Landry, USA Jason K.W. Lee, Singapore Luc L. Leger, Canada Romuald Lepers, France Benjamin D. Levine, USA Peter Lindholm, Sweden Anne A.B. Loucks, USA Maureen MacDonald, Canada Nicola N. Maffulli, UK Paraskevi Malliou, Greece David Mann, Australia Samuele Maria Marcora, UK Frank E. Marino, Australia Barry B.J. Maron, USA Ryouta Matsuura, Japan Neil Maxwell, UK Cian McGinley, Ireland Daniel Memmert, Germany Philipe Meurin, France Michael C. Meyers, USA Pavle Mikulic, Croatia Catherine J. Minns Lowe, UK Trine Moholdt, Norway Monique Mokha, USA Philip J. Morgan, Australia Jacob Morkeberg, Denmark Falk Muller-Riemenschneider, Germany Toby Mundel, New Zealand Claus-Martin Muth, Germany George Nassis, Greece Ceri Nicholas, UK Guillaume Nicolas, France David D.C. Nieman, USA Timothy David Noakes, South Africa Peter O’Donoghue, UK Britt Oiestad, Norway John J. Orchard, Australia Michael S. Orendurff, USA Trisha L. Parsons, Canada Pedro Passos, Portugal David Pease, Australia
Sports Med 2011; 41 (12)
Acknowledgement
987
Antonio A. Pelliccia, Italy Stephane Perrey, France Svein Arne Pettersen, Norway Shaun M. Phillips, UK Gerald E. Pierard, Belgium Yiannis Pitsalides, UK Babette Pluim, Netherlands Mike Price, UK Alberto Rainoldi, Italy Ermanno Rampinini, Italy Nicholas A. Ratamess, USA Vikki Revell, UK Michael Ristow, Germany Wayne W.D. Rosamond, USA David S. Rowlands, New Zealand Jane J.S. Rumball, Canada Alice A.S. Ryan, USA Lucy J. Salmon, Australia Amol Saxena, USA Yorck Olaf Schumacher, Germany Nick Sculthorpe, UK Eva Segura-Orti, Spain Sanjay Sharma, UK Cecilia Shing, Australia
Richard Shuttleworth, Australia J.C. Siegler, UK Malchira S. Somanna, Italy Barry Anthony Spiering, USA Arie Steinvil, Israel Emma Stevenson, UK Kay Tetzlaff, Germany Dylan Thompson, UK Mark M.D. Tillman, USA Brook Elan Tlougan, USA Jason D. Vescovi, Canada Anders Vinther, Denmark Nicola E. Walsh, UK Darren D.E.R. Warburton, Canada Stuart J. Warden, USA Jim Waterhouse, UK Eddie Weitzberg, Sweden Greg D. Wells, Canada Daphne Wezenberg, the Netherlands Laurie Wideman, USA Alun G. Williams, UK Craig C. Young, USA Kathryn R. Zalewski, USA Jerzy Zoladz, Poland
We look forward to your continued support in 2012 and to bringing you first-class content from around the globe. With best wishes from the staff of Sports Medicine and all at Adis, a Wolters Kluwer business.
ª 2011 Adis Data Information BV. All rights reserved.
Sports Med 2011; 41 (12)
Sports Med 2011; 41 (12): 989-1002 0112-1642/11/0012-0989/$49.95/0
LEADING ARTICLE
ª 2011 Adis Data Information BV. All rights reserved.
Mandatory ECG Screening of Athletes Is this Question Now Resolved? Roy J. Shephard Faculty of Physical Education & Health, University of Toronto, Toronto, ON, Canada
Abstract
European and North American cardiologists have long debated the need for mandatory ECG screening of athletes in order to prevent sudden cardiac death. European investigators have recently adduced new evidence, which they believe supports the need for such screening. They note a decrease of sudden cardiac deaths among Italian athletes following the introduction of mandatory screening in that country, clearer definitions of resting ECG abnormalities in athletes, new and more encouraging calculations of cost/ benefit ratios and direct comparisons of clinical examination alone against clinical examination plus ECG screening. Nevertheless, it seems that critical criteria for the success of any screening procedure (a substantial prevalence of the problem, coupled with an adequate test sensitivity and specificity) have yet to be satisfied. Very few athletes are liable to sudden cardiac death, only a few of those who are vulnerable will be identified by ECG screening, and even if all potential cases could be detected, restriction of their physical activity would be unlikely to have a major influence on their prognosis. At the same time, a requirement of mandatory testing would discourage engagement in physical activity, and would impose substantial direct costs on the community. Moreover, the large number of false positive test results could have important and undesirable consequences for both indirect medical costs and the overall health of competitors. ECG screening might become more effective if it could be focused on a smaller sub-group of vulnerable athletes, or if the problem of false positive tests could be addressed through an increase of test specificity. However, on the basis of current information, it would seem better to direct efforts in preventive medicine to more common causes of premature death in the young adult.
1. Introduction Mandatory pre-participation ECG screening of athletes has long been a source of controversy between European and North American cardiologists, due in part to differing cultural, social and legal environments,[1] and differing systems of healthcare delivery. Many European groups have argued strongly for mandatory testing, but
most North American investigators have maintained that such testing is ineffective and inappropriate.[2-21] One major criticism of the European position has been that ECG screening does not meet long-accepted WHO criteria for a successful screening programme.[22,23] These criteria include at least a moderate prevalence of the condition to be diagnosed, an appropriate test sensitivity and specificity and a net benefit to the
Shephard
990
patient that outweighs any negative consequences of screening. The first three of these criteria are implicit in the classical theorem of Bayes.[24,25] Recent reports have suggested that ‘important new evidence’ now favours the European position.[26-32] This brief article looks at the strength of this ‘new evidence’, weighing available information against the fundamental epidemiological ground rules of the WHO for successful screening. The text also reconsiders costs and benefits, together with the practicalities of such testing. 2. New Evidence Supporting Mandatory ECG Screening New evidence adduced to support mandatory ECG screening includes a reported reduction of sudden cardiovascular deaths (SCD) among Italian athletes after such tests became mandatory in that country,[26,33,34] improved criteria to assess the normality of resting ECG records in athletes,[35-37] new and more favourable estimates of the costs of screening versus its presumed benefits[32,38] and direct comparisons of diagnostic efficacy of clinical examination alone versus clinical examination plus ECG screening.[27] 2.1 Effects of Mandatory ECG Screening Upon Incidence of Sudden Cardiovascular Death (SCD) in Athletes
The incidence of SCD among Italian athletes (either when exercising or at rest) has decreased sharply since the introduction of mandatory annual ECG screening and submaximal exercise testing.[26,33,34] A very high annual incidence of 3.6 SCD per 100 000 athletes (either during exercise or at rest) was observed in the period immediately before ECG screening began (in 1982). After 25 years of screening, the incidence has decreased to 0.4 per 100 000 athlete-years, with most of the decrease being attributable to fewer incidents involving hypertrophic cardiomyopathy (HCM); the proportion of SCD from arrhythmias apparently remains higher than in North America. The study under discussion is limited by its pre- post-intervention design;[26] there were only ª 2011 Adis Data Information BV. All rights reserved.
2 years of data preceding the legislation, and the only control observations were made on nonathletes over the same period. The decreased incidence of SCD among the athletes could thus have arisen from factors other than their ECG screening. Potential factors that merit careful consideration include changes in climate, and thus the risk of heat stress during all-out effort, a better control of doping, a wider availability of effective cardiac resuscitation in sports facilities and even random variation in the data.[3,39] The climate is mild in much of the Veneto region, and all causes of death were ascertained by experienced pathologists, so a decrease in heat-related deaths appears improbable, defibrillators are also, as yet, not widely available in the Veneto region, so that more frequent successful cardiac resuscitations are unlikely to have been a major influence. The study also has potential for a timedependent bias,[40] in that a high death rate during the pre-screening period necessarily created a lower-risk population during the subsequent years of observation, although against this argument, the death rate remained high for several years following the introduction of mandatory screening. Conclusions were nevertheless based on only a small number of deaths (a total of 10–11 athletes per year) in one relatively small region of Italy (Veneto, accounting for some 9% of the overall Italian population); given that the legislation requiring ECG screening was nationwide, it is surprising that no information is as yet provided on any changes in the incidence of SCD in other parts of Italy. The initial incidence of SCD among the Veneto athletes (3.6 per 100 000 athlete-years) was much higher than the 0.5 per 100 000 typically reported in North America, where ECG screening is a rarity.[12,41,42] It also exceeded the incidence observed in a careful 7-year study in Denmark, a country where any formal pre-participation screening remains a rarity. A study of death certificates for Danish young people aged 12–35 years between 2000 and 2006 (5609 valid reports out of a total 5662 deaths) revealed a total of 15 deaths occurring within 1 hour of ceasing exercise; all deaths were in males, onethird being associated with soccer and one-third with running. The estimated rate of SCD for the Sports Med 2011; 41 (12)
Mandatory Screening of Athletes
Danish athletes was 1.21 per 100 000 person-years, as compared with a figure of 3.8 per 100 000 person-years for the general population (athletes and nonathletes) in the same age range.[43] The Veneto data certainly appear to show a downward trend for the incidence of SCD in Italian athletes, whereas rates for nonathletes in the same part of Italy have remained relatively constant. However, the final incidence of 0.4 deaths per 100 000 athlete-years (on which the claim of efficacy is based) was only seen during the period 2002–4 (some 20 years after implementation of ECG screening). From 1979 to 1996 (with ECG screening required for much of this time), there were 49 deaths among 33 375 athletes, an incidence of 8.5/100 000, much higher than the value of 0.71/100 000 person-years seen among nonathletes who were not screened.[18,26] In contrast to the Italian experience, Steinvil and associates[44] have recently published findings from Israel, where a National Sport law mandated ECG screening of athletes commencing in 1997. In addition to a physical examination, the required Israeli screening includes a resting ECG and Bruce protocol exercise testing performed by an accredited physician. A newspaper search for reports of exercise-related deaths among competitive athletes found 24 such incidents in the period 1985 through 2009; there was an estimated incidence of 2.5 incidents per 100 000 athleteyears in the 10 years before enactment of the legislation, and 2.7 incidents per 100 000 athleteyears in the 10 years after its enactment. The authors caution that the Italian legislation may have been introduced because of an unusual but random increase of SCD among athletes in the period immediately prior to its enactment, with regression to the normal experience for this population explaining at least a part of the decrease in SCD. We may conclude that although recent data from one part of Italy appear to show a reduction of SCD following the introduction of mandatory ECG screening, this trend is not confirmed by Israeli data for the periods before and after screening. In both countries, the number of deaths under consideration is quite small, and the data need verifying over a longer period, with inclusion of findings from other parts of the world that have ª 2011 Adis Data Information BV. All rights reserved.
991
adopted ECG screening. Moreover, any inferences of benefit based on screening that used older ECG criteria of normality need to be checked relative to revised criteria for athletes (see section 2.2).[36,37] 2.2 Revised Criteria for the Normality of Resting ECGs in Athletes
Early investigators assumed that the main cause of SCD in young athletes was HCM, that this could readily be detected by ECG screening (with criteria of normality being set by the immediate examining physician, and abnormalities being diagnosed, particularly on the basis of increased QRS complex voltages),[16,45-47] that doubtful cases could readily be resolved by 2-dimensional (2-D) echocardiography and that the prevention of sport in athletes with HCM would avert SCD.[45,48] However, most of these premises have now been challenged. ECG screening of 4450 members of Italian national teams identified neither of two athletes where genetic tests subsequently suggested the presence of HCM.[49] Moreover, a number of the original European diagnostic criteria, such as increased QRS complex voltages, T-wave flattening and/or a prolonged corrected QT (QTc) interval[50] have been observed relatively frequently in mass-screening of Japanese schoolchildren and thus they do not seem appropriate criteria to judge normality;[51,52] indeed, isolated increases of QRS complex voltage are seen in 40% of highly trained athletes, but in <2% of individuals with HCM,[53,54] Further, diagnoses of ‘abnormality’ based on QRS voltage criteria often fall into the ‘grey’ zone (a septal wall thickness of 13–16 mm) where 2-D echocardiography fails to distinguish physiological from pathological hypertrophy of the heart.[55] Finally, the health significance of diagnosing HCM is limited, since the incidence of SCD is quite low among those with this condition. Isolated increases in QRS complex voltage account for more than one-half of ECG ‘abnormalities’ in many reports,[27] but they do not give a clear indication of imminent SCD.[53,54] In children with HCM, the annual incidence of SCD is only 1–3%,[56-58] and in young adults HCM accounts for only 7% of SCD.[46] Of 33 735 competitors, Sports Med 2011; 41 (12)
Shephard
992
none of the 16 athletes with HCM died over 17 years of observation[45] (although it remains unclear how far this favourable prognosis reflects the withdrawal of these individuals from competition). ECG criteria characteristics such as sinus bradycardia, first degree atrioventricular block, incomplete right branch bundle block, early repolarization and isolated voltage characteristics of the QRS complex can all be normal consequences of athletic training[36] or of ethnic differences.[59] Revised criteria of abnormality include: T-wave inversion, ST-segmental depression, pathological Q waves, evidence of left atrial enlargement, left axis deviation or left anterior hemi-block, right axis deviation or left posterior hemi-block, right ventricular hypertrophy, ventricular pre-excitation, complete left or right branch bundle block, a long or a short QT interval, and Brugada-like early polarization.[36,37] Importantly, the last three of these items were not considered in early ECG analyses.[60] When applied to a sample of 1005 athletes, the revised criteria reduced the number of ‘abnormalities’ from 292 to 110 (although this figure still remains a relatively large 11% of all competitors).[37] Success in diagnosing true abnormalities remains problematic. A study from the Italian National Institute of Sports Medicine added an exercise ECG and 2-D echocardiography to the evaluation of a selected sample of 12 550 athletes; 81 competitors had deeply inverted T waves, but only six of these individuals were thought to have structural abnormalities. Over a follow-up averaging 8 years, one athlete from this group died, and a second had an aborted cardiac arrest. Despite the very high false-positive rate in this particular sample (between 75/81 and 79/81 of tests), the authors of this report argued that their screening had merit.[61] Given the recent changes in ECG criteria of normality, it is plain that most published estimates of test sensitivity, specificity and cost/efficacy need to be repeated. 2.3 New Estimates of the Costs and Benefits of ECG Screening
Although many clinicians dislike cost/benefit or cost/efficacy analyses, such calculations are imporª 2011 Adis Data Information BV. All rights reserved.
tant in an era when overall medical expenditures appear to be increasing at an unsustainable rate. Cost/benefit and cost/efficacy considerations have figured prominently in North American objections to mandatory ECG screening of athletes.[3] Health economists commonly argue that it is reasonable to spend somewhere between $US20 000 and $US100 000 on an intervention that will add a year of disease-free life to a patient. The efficacy of any procedure is thus strongly influenced by the age of the patient. If SCD is prevented in a 20-year-old athlete, disease-free life might well be extended by 60 years, so that an expenditure of $US1.2–6.0 million on preventing a single incident might still fall within the range of ‘reasonable’ expenditures. Early cost estimates focused mainly on the likelihood of detecting HCM. On the basis that each athlete would receive a single ECG examination, plus any additional tests that were needed to exclude false positive results, the American Heart Association estimated that in the US, the direct cost of preventing one SCD was about $US3.4 million.[12] Others have argued that the full costs of such a policy (including provision of the necessary infrastructure and indirect costs sustained by the patient and his or her relatives) might be even higher.[18] However, two recent papers have suggested much lower figures for the direct costs of mandatory ECG testing.[28,32] Factors contributing to this discrepancy include a broad search for dangerous cardiac abnormalities other than HCM (thus increasing the prevalence factor, below), an assumption that all false positive results can be clarified by further testing and a suggestion that current fee schedules might be heavily discounted if mass testing were to become mandatory. Fuller[32] thus estimated that the direct cost of ECG screening for high-school athletes would be $US44 000 per life saved. His analysis assumed a test sensitivity of 70% and a specificity of 84%.[31] He included several no longer accepted criteria of ECG abnormality in his analysis (particularly increased QRS complex voltages and T-wave flattening); such characteristics accounted for 11% of his 15.7% ‘abnormal’ tracings. He further argued that 98% of apparent abnormalities could be resolved simply by a 2-D Sports Med 2011; 41 (12)
Mandatory Screening of Athletes
echocardiogram, bringing the average cost of dealing with a positive ECG to $US365. Wheeler et al.[28] estimated that ECG screening could be added to a standard clinical examination at a cost of $US89 per patient (a total medical expense of $US199), and that this would save 2 life-years per 1000 athletes. In this analysis, the prevalence of 16 cardiac potential causes of SCD was estimated from Italian data. Some of these figures (for instance, a 0.2% prevalence of HCM) seem high, and even the overall risk of SCD (2.4 per 100 000 athletes per year over an 8-year competitive career) is greater than that found in several North American surveys. The assumed test sensitivity of 68% is in keeping with Fuller et al’s.[31] analysis, but the figure for test specificity is more problematic. Criteria of abnormality were similar to those adopted by Fuller, and some 14% of the sample was judged as having positive test results. On the basis of additional investigations (echocardiography, stress testing, cardiac MRI, Holter monitoring, and/or computed tomography), it was estimated that there would be approximately 2% true positives, individuals liable to SCD, that these individuals would otherwise die before reaching the age of 35 years and that all of the diagnosed abnormalities could be treated successfully. Costs of the various diagnostic procedures and likely treatments were based on 2004 figures from the US National Center for Health Statistics.[62] The added expense from the ECG screening was estimated at $US42 900 per life-year saved, for a total cost of $US79 000 when the initial clinical examination was also included. The Italian requirement is for annual rather than once only ECG screening, and this would immediately boost all of the above expenditures 10–20-fold. Moreover, accounting to date has not examined other possible less direct economic consequences of screening. Most physicians would probably advocate a continuation of low or moderate physical activity, which might be adequate to conserve many aspects of metabolic health. However, there could be adverse consequences from an unnecessary restriction of physical activity by either patient or physician, together with a negative impact of essential furª 2011 Adis Data Information BV. All rights reserved.
993
ther testing upon the well-being and productivity of both the next of kin and those individuals who are wrongly identified as at an increased risk of SCD.[63] Finally, although one would not wish to resolve the issue on this basis, a full accounting would also take account of the high medical costs likely during the final years of life among athletes who do not succumb to SCD. Nevertheless, the estimates of Wheeler et al. and Fuller[28,32] suggest that if one focuses simply on the direct medical costs of a single preparticipation ECG examination, then the outlay may fall within commonly accepted definitions of a ‘reasonable’ expenditure. 2.4 Direct Comparisons of Clinical/ECG Screening with Clinical Examination Alone
Plainly, some of those vulnerable to SCD have symptoms and/or signs that should be detected by clinical examination. In the series of Corrado and associates,[26] 10 of the 55 athletes who died suddenly had pre-existing symptoms. Likewise, a retrospective Swedish study of SCD found that 18% had a positive family history and 76% had previous symptoms.[64] When evaluating the usefulness of ECG screening, it is thus important to assess how much new information the ECG procedure adds to a competent clinical examination. The outcome of such a comparison necessarily depends in part on the skills deployed in making the respective evaluations and, in this respect, international comparisons are made difficult, since the long-standing Italian legislation has fostered the emergence of a cadre of experienced sports cardiologists that is not available in most other countries. Baggish et al.[27] compared the diagnostic efficacy of history and physical examination alone versus a combination of such information with the findings from a 12-lead resting ECGs. They used a rather small sample of 510 unselected US college athletes. An 8-minute clinical examination was conducted by individuals who were neither dedicated sports physicians nor cardiologists. The subsequent ECG evaluation was based on the European College of Cardiology standards of normality for the general (nonathletic) population. The gold standard of a correct diagnosis was not SCD, Sports Med 2011; 41 (12)
Shephard
994
as observed by a prospective trial, but rather a cross-sectional comparison of the two data sets with reports of suggestive abnormalities obtained from limited echocardiographical imaging. Thus, the comparison must be judged as relatively weak. The authors claim that after examining the ECG records, 11 individuals with ‘dangerous’ abnormalities were identified, as against only 5 detected by history and physical examination alone. It is hard to believe that 11 of 510 unselected college athletes were at imminent risk of SCD; many estimates put the annual risk at no higher than one incident per 100 000 athlete-years. It seems significant that only 3 of the 11 athletes with ‘dangerous’ abnormalities were asked to restrict their sport participation. Furthermore, we have no information whether this restriction of physical activity lengthened or shortened their subsequent life span. Thus, there remains a need for further studies that compare clinical evaluation alone with clinical examination plus ECG screening. 3. Epidemiological Criteria for Successful Screening What is the current epidemiological assessment of the resting ECG as a successful method
of detecting those athletes who are vulnerable to SCD? Are the incidence of SCD, test sensitivity and test specificity sufficient to meet accepted WHO criteria for the introduction of a screening procedure? 3.1 Incidence of SCD in Athletes
However sophisticated a test procedure may be, if few of the individuals who are tested suffer from the condition, the potential for saving lives will be limited. Widely varying values have been cited for the risk of SCD among athletes (table I). This reflects differences of methodology (particularly, recent changes in ECG criteria of abnormality; see section 3), success in discovering all cardiac incidents and in defining the age, sex, ethnic characteristics of the population studied and a progressive shift in test objectives from the prevention of SCD during exercise to either an avoidance of all fatal and nonfatal episodes of cardiac arrest, or simply to the detection of various electrocardiographic abnormalities that might at some stage pose cardiovascular problems.[30] SCD is quite rare among top international athletes. An 8-year follow-up of 114 Olympic competitors found no cardiac events, despite continued
Table I. Incidence of sudden cardiac arrest (SCA), sudden cardiac death (SCD), sudden death (SD), and cardiac arrest (CA) in various populations of adults and children Study
Variable
Study duration
Incidence
Sample
Pellicia[61]
SCD
8y
0
114 Olympic athletes
Borjesson et al.[29,65]
SCD
Review
1–3 per 100 000 athlete-y
Van Camp et al.[66]
SCD (exercise)
10 y
0.1 per 100 000 athlete-y (female) 0.5 per 100 000 athlete-y (male) 1.1 per 100 000 athlete-y (university)
Drezner et al.[41]
SCD (exercise)
3y
1.5 per 100 000 athlete-y
100 000 NCAA athletes 18–23 y
Drezner et al.[67]
SCA, SCD
7 mo
4.4 per 100 000 athlete-y 1.6 per 100 000 athlete-y
1710 high-school athletes, aged 14–17 y
Asif et al.[68]
SCD
6y
2.1 per 100 000 athlete-y
2 million NCAA athletes aged 17–23 y
Chugh et al.[69]
SD
3y
0.22 per 100 000 person-y
310 000 children aged 10–17 y
Atkins et al.[70]
CA
16 mo
5.4 per 100 000 person-y
Children aged 12–19 y
Maron et al.[71]
SCD
12 y
0.46 per 100 000 person-y
651 695 high-school athletes aged 16–17 y
Maron et al.[72]
SCD
26 y
0.61 per 100 000 person-y
All competitive athletes 12–35 y
Solberg et al.[73]
SCD
8y
0.9 per 100 000 person-y
Individuals 15–34 y (few involved in sport)
2.7 million high-school and university extramural athletes aged 13–24 y
NCAA = National Collegiate Athletic Association.
ª 2011 Adis Data Information BV. All rights reserved.
Sports Med 2011; 41 (12)
Mandatory Screening of Athletes
and rigorous endurance training.[74] Borjesson and Pellicia[65] argued that the risk of SCD in competitive athletes, although somewhat higher than in nonathletes,[75] was only 1–3 per 100 000 personyears. An 8-year study from Norway found a total of 23 exercise-related deaths among individuals aged 15–34 years, a rate of about 0.9 per 100 000 person-years;[73] however, the Norwegian investigation found no deaths in elite athletes, and few of the 23 affected individuals were involved in any form of competitive sport. Most (but not all) North American data also show a low incidence of SCD. Over the period 1980–2006, there were 1049 cardiovascular deaths among all US athletes aged 39 years or younger (an average of 66 per year, mostly in male competitors), with an estimated overall rate of 0.61 per 100 000 person-years.[72] A review of all 1 994 962 National Collegiate Athletic Association (NCAA) athletes found 278 sudden deaths from 2003 to 2008, 43 of these incidents being of cardiac origin (about eight per year); 26 of the fatal incidents occurring during exercise, a rate of 2.1 per 100 000 person-years, although the cause of death was not always carefully verified.[68,76] A 10-year analysis of nontraumatic deaths occurring up to 1 hour after participating in high-school or university extramural sport found that about a quarter of incidents had a noncardiovascular cause (commonly hyperthermia).[66] SCD averaged ten cases per year among a total of 2.7 million US students, with an incidence of 0.1 per 100 000 a year in female students, 0.5 per 100 000 personyears in male high-school students, and 1.1 per 100 000 person-years in male university students, with HCM the commonest pathology identified at post-mortem. Finally, a 12-year study of 651 695 Minnesota high-school athletes aged 16–17 years found 3 cardiovascular deaths, all in males, for a rate of 0.46 per 100 000 person-years,[71] Drezner and associates[77] looked at reports of SCD in connection with the provision of automated defibrillators to universities and high schools. A sampling of some 100 000 division I intercollegiate athletes across 212 institutions yielded five instances of SCD over the period 2000–3, a rate of 1.5 per 100 000 person-years; defibrillation was not successful in any of these ª 2011 Adis Data Information BV. All rights reserved.
995
five incidents. There were also 14 cases of sudden cardiac arrest among high-school student athletes (12 male, 2 female), most of them having exercised for <1 hour previously (a rate of 4.4 per 100 000 person-years); however, 9 of the 14 were successfully resuscitated, for an SCD rate of 1.57 per 100 000 person-years.[67] The incidence of SCD seems lower in children than in adults (some adult studies include competitors in the coronary-prone age range). The report of Chugh and associates[69] covers incidents in the Portland region of Oregon; over a 3-year period, there were two sudden deaths (not necessarily exercise-related) among some 311 000 individuals aged 10–17 years, for a rate of about 0.22 per 100 000 person-years. A second study from the US and Canada looked at emergency, out-of-hospital treatments of cardiac arrest in children aged 12–19 years.[70] This yielded 193 episodes over a 16-month period (a rate of 5.37 per 100 000 person-years). However, few of these incidents were exercise related; only seven individuals had ventricular fibrillation or tachycardia (a rate of 0.23 per 100 000 person-years). In summary, it seems reasonable to assume that the incidence of SCD in adult athletes is probably in the range 1–2 per 100 000 person-years. 3.2 Test Sensitivity
The sensitivity of a diagnostic procedure essentially indicates the likelihood of identifying those within a population who are suffering from a given condition. It seems easier to find abnormalities when incidents are examined in retrospect. Thus, a Swedish study suggested that among 66 cases of SCD in athletes aged <35 years, 18% had a positive family history, 76% had previous symptoms and 82% had abnormal electrocardiograms.[64] However, the success rate is much lower when tests are applied prospectively to an entire population of athletes. In the case of SCD, the clinical diagnosis of vulnerable individuals is difficult.[78,79] In one retrospective analysis of 134 US athletes, 115 had undergone standard clinical screening; however, cardiovascular disease (CVD) was suspected in only four of the 115, and a correct diagnosis had Sports Med 2011; 41 (12)
Shephard
996
Table II. Reported sensitivity and specificity of ECG screening in athletes
Table IV. Negative specificity and specificity of ECG screening in athletes
Investigator
Reported sensitivity (%)
Gold standard
Investigator
Negative specificity (%)
Gold standard
Fuller et al.[31]
60–70
Echocardiography
Fuller et al.[32]
97.7
Echocardiography
Wheeler et al.[28]
68
Interpretation of Corrado et al.[26]
Wheeler et al.[28]
95
Interpretation of Corrado et al.[26]
Pelliccia et al.[34]
51
Clinical judgement + Echo
Pelliccia et al.[34]
95.7
Clinical judgement + Echo
Lawless et al.[85]
50
Review
Corrado et al.[26]
89
Reduction in mortality
been reached in only one individual (a sensitivity of 0.9%).[80] In addition, other reports have underlined that in 60–80% of incidents, SCD is the first manifestation of CVD.[81,82] Nevertheless, a careful clinical examination should highlight potential warning symptoms and signs, including a family history of premature death, irregularity of the pulse, fainting episodes, abnormalities of blood pressure or a heart murmur. One formal evaluation put the sensitivity of a clinical examination as low as 3%,[83] although this figure has likely been increased by standardization of the clinical protocol,[84] perhaps to 6%,[32] 10%[31] or even 15%.[28] ECG screening has a higher sensitivity, although by no means all vulnerable individuals are identified (see tables II, III and IV). One problem is that some of the conditions predisposing to SCD such as the Brugada syndrome and long and short QTc interval patterns can appear only intermittently.[89] One report estimated an overall sensitivity of 60–70%, on the theoretical basis that findings would be abnormal in 95% of patients Table III. Reported false positives and specificity of ECG screening in athletes Investigator
Reported false positives (%)
Gold standard
Pelliccia et al.[34]
37
Clinical judgement + Echo
Maron et al.[86]
15
Echo + additional tests
Lawless et al.[85]
40
Review
Rowland[17]
30
Review
Pelliccia et al.[35]
6.7
Additional diagnostic tests
Hevia et al.[87]
6
Echo + additional tests
Wilson et al.[88]
1.9
ª 2011 Adis Data Information BV. All rights reserved.
with HCM, in almost all patients with prolonged QT interval or ventricular pre-excitation and, in many patients with myocarditis, coronary artery anomalies, right ventricular dysplasia, dilated cardiomyopathy and aortic stenosis.[31] Another analysis assumed a sensitivity of 68%,[28] based upon their interpretation of the Italian experience.[37] In fact, data from the Italian National Institute of Sports Science show a somewhat lower sensitivity. Among 1005 competitors (785 ostensibly healthy athletes and 220 others who had been referred because cardiac problems were suspected), the ECG was rated as ‘abnormal’ in 402, but only 27 (2.7%) were judged as having a ‘true positive’ test relative to clinical judgement and echocardiographical examination; there were also 26 false negative reports relative to this yardstick.[90] The sensitivity for this sample was thus 27 per 53, or 51%. A recent literature review also suggests a sensitivity of about 50%.[85] The ultimate measure of test sensitivity will probably come from studies such as those conducted in Italy.[26] If it were to be proven that the introduction of ECG testing reduced the incidence of SCD from 3.6 per 100 000 to 0.4 per 100 000 athlete-years, this would imply that 89% of cases had been diagnosed and treated effectively. However, the existing Italian data cannot be interpreted in this way. Some of the decrease in SCD may reflect changes in the environment, the availability and effectiveness of emergency services, a better control of doping and even random year-to-year variations. 3.3 Test Specificity
Test specificity provides a measure of the accuracy of both positive and negative diagnoses. Sports Med 2011; 41 (12)
Mandatory Screening of Athletes
In other words, the percentage of false positive responses is given by (one minus positive specificity), and the negative specificity indicates the percentage of negative tests that are correct. Values for both positive and negative specificity naturally depend on the ECG screening criteria that are used. These criteria have changed in recent years, and most estimates are based on older standards of normality.[91] Thus, a study of 658 Stanford athletes found that 32% of female and 62% male athletes had ‘abnormal’ ECGs, but the percentages of abnormalities dropped to 43% and 26%, using the European Society of Cardiology criteria, 20% and 9% using Pelliccia’s criteria of ‘distinctly abnormal’ and 12% and 7% using current Stanford University criteria. Ultimately, only 10% of the sample was thought to need further testing.[91] If the Pelliccia criteria[90] had been applied, 15% rather than 10% of athletes would have undergone further investigation. The false positive rate is an important statistic, since it determines the need for further examinations, with attendant medical costs, anxiety and time commitments for both the patient and his or her relatives. Moreover, unless follow-up procedures have 100% specificity (which is most unlikely), harm will result from unnecessary restrictions of physical activity. In some studies, as few as 3% of ‘abnormal’ ECG records have been true positive findings. The study of Pellicia and associates[90] included 220 individuals suspected of cardiac disease, yet despite this bias, 375 of 1005 ECGs (37.3%) yielded false positive results, and only 27 tracings (6.7% of all ‘abnormal’ records) were ‘true positives’. Early North American evaluations suggested a false positive rate of around 15%,[86] although one recent review indicated a figure as high as 40%,[85] and a second review quoted a figure of 30%, with fears that many of the doubtful cases would not be resolved by echocardiography.[17] Revised criteria of abnormality (particularly greater skepticism with respect to excessive left ventricular voltages) have substantially reduced the number of false positive results, but probably at the cost of some reduction in test sensitivity. One recent study of 9125 US adolescents found abnormal ECGs in only 2%.[92] Data for 32 652 ª 2011 Adis Data Information BV. All rights reserved.
997
Italian athletes (median age 17 years, but some as old as 78 years) found 11.8% with ‘abnormal’ ECGs; as judged from further tests, the probable false positive rate for this sample was 6.9%.[35] In a recent sample of 1220 Spanish athletes, 6.1% had ‘abnormal’ ECGs, but all except two of 75 were apparently false positive findings.[87] Wilson and associates[88] examined 1074 national and international junior athletes in the UK; 25 were thought to have an abnormal ECG, and in 9 of the 25 the condition identified had the potential to cause SCD; thus, they claimed a false positive rate as low as 1.9%. The specificity of a negative test is quite high. One study of 5615 US high-school athletes claimed a negative specificity of 97.8% for an abbreviated cardiac history, and 97.7% for a normal resting ECG, the latter figure being determined on the basis that 130 of 146 ‘abnormal’ ECGs (2.3% of all ECG records) had Q-wave changes suggestive of HCM, but no evidence of this was found at echocardiography; the other 16 ‘abnormalities’ were apparently counted as true positives.[31] Wheeler and associates[28] accepted an estimate of 95% specificity, based on their interpretation of Italian data. In the study of Pellicia and associates,[90] 577 per 603 negative ECGs were considered ‘true negative’ results, giving a negative sensitivity of 95.7%.[16] At first inspection, these figures seem encouraging, but they must be tempered by the observation that most patients would have remained free of SCD even if testing had not been conducted. 4. Benefits versus Costs of Screening for an Athlete 4.1 Benefits
How far do athletes wish to learn about possible contraindications to their sport participation? Both the individual and society generally accept the premise that zero risk is unlikely in competitive sports. Indeed, for activities such as mountain climbing and snow boarding, risk is an inherent and exciting part of the athletic experience.[93] Equally, in traditional sports such as North American football and ice hockey, team Sports Med 2011; 41 (12)
Shephard
998
members recognize and accept the potential dangers of major trauma and fatalities; few participants would wish to abandon competition because they cannot be assured of 100% personal safety. Nevertheless, advocates of ECG screening argue that the cardiac risk of athletic competition is different, because the athlete is unaware of the danger and benefits if screening tests reduce his or her likelihood of the remote contingency of SCD. The counter argument is that the incidence of SCD is only 1–2 cases per 100 000 competitoryear, and even if all vulnerable individuals could be identified and treated successfully, the immediate benefit to the athlete and his or her family would be much smaller than could be achieved by addressing other causes of death in the young adult, such as suicide, motor vehicle and sports injuries. Moreover, test sensitivity is such that at best the ECG will identify a half of vulnerable competitors,[85] probably fewer than 50% if revised criteria of ECG normality are used. This detection rate must further be discounted to allow for the proportion of cases that could be identified by clinical evaluation (perhaps by as much as 15%). Moreover, account must be taken of increasingly frequent successful cardiac resuscitations, as automated external defibrillators become widely available in sports facilities.[67,77] Finally, treatment does not always have a favourable influence on prognosis. Indeed, HCM deaths often do not involve vigorous physical activity.[94] Thus, the health benefit from prohibiting sports participation and adopting other forms of treatment is quite limited,[21,46,56-58,95] possibly smaller than could be achieved by other changes in personal lifestyle. 4.2 Costs
Discouragement of physical activity by mandatory ECG testing is likely to be an important cost, although detailed data on this issue are not yet available; an analysis based on the Veneto experience concluded that 791 athletes would need to be disqualified in order to prolong one life.[96] The direct financial costs of ECG screening and subsequent follow-up are large. Sometimes this expense has been borne by the state or a ª 2011 Adis Data Information BV. All rights reserved.
sports team. However, in North America, such charges would generally be met by the individual, since neither private nor government insurance plans would accept charges for testing deemed as inappropriate by WHO screening criteria. Does an athlete have the right to insist on ECG screening? This seems acceptable, provided that the competitor or the parents are prepared to bear this expense, and the fallibility of the test results has been clearly explained to the individual concerned. The possible indirect consequences of false positive diagnoses are of much greater concern than the direct costs. Currently, Italian sports physicians exclude an unacceptably large 2% of athletes from sports play as a consequence of ECG screening; however, they admit that at most 0.2% have conditions with a potential to provoke SCD.[6,38] Unwarranted avoidance of exercise necessarily has a substantial adverse effect on an athlete’s quality of life and all-cause mortality.[97] Moreover, as many as one-third of the individuals who become concerned about their hearts can develop a residual cardiac neurosis that is hard to dissipate.[98] 5. Practicalities of Mandatory ECG Screening Even if good reasons were found to support mandatory ECG screening, its introduction would impose major logistical challenges for personnel and medical infrastructure. Current estimates for the US suggest that as many as 10 million athletes might require testing, with a potential overall cost of $US2 billion per year.[12] Moreover, it would be difficult to find an adequate number of professionals to provide well qualified ECG interpretations and the secondary testing needed to exclude false positive results. In Italy, specialist training in this area now demands 4 years of intensive study[4] and, in most countries, the standard sports medicine curriculum allocates little course time to this area of instruction. The scale of potential logistical problems seems likely to increase further in the face of aging populations and efforts to involve the entire adult population in vigorous physical activity. Sports Med 2011; 41 (12)
Mandatory Screening of Athletes
6. Conclusions European investigators have recently adduced new evidence to support their claims that all athletes should undergo mandatory ECG screening. They note a decrease of SCD among Italian competitors subsequent to the introduction of mandatory screening, the development of clearer definitions of ECG abnormality for athletes, new calculations suggesting lower cost/benefit ratios for such testing, and favourable direct comparisons of clinical examination plus ECG screening versus clinical examination alone. Nevertheless, WHO criteria for the success of a screening procedure (a substantial incidence of SCD, plus adequate test sensitivity and specificity) have yet to be satisfied. Few athletes are vulnerable to SCD, only a small fraction of these individuals can be identified by ECG screening and, even if they are detected, restriction of physical activity does not seem to have a major influence on prognosis. At the same time, mandatory ECG screening discourages participation in physical activity, imposes substantial direct costs on the community and has adverse implications for both indirect medical costs and the overall health of competitors. Keys to a more successful screening programme would include a concentration of testing procedures upon a vulnerable subset of the athletic population (i.e. an increase of prevalence or incidence of the abnormality among those who are tested), and an increase in test specificity (to reduce the problem of false negative diagnoses). A family history of SCD would provide one index of vulnerability, although premature death is by no means universal in HCM. Future advances in genetic technology may also provide some clues. Genetic mutations affecting cardiac troponins T and I, the b-myosin heavy chain, a-tropomyosin, a-actin and myosin-binding protein C have all been associated with HCM.[99-105] As many as 36 possible abnormalities of the b-MHC gene have been implicated at one chromosomal locus, 14ql.[101] Unfortunately, the phenotypical penetrance of the abnormal genotype is often low, and sometimes there is an interaction between genotype and environmental factors. Thus, if myocardial hypertrophy develops at all, it may not be seen
ª 2011 Adis Data Information BV. All rights reserved.
999
until middle age. Further, about one-half of the cases of HCM have no apparent genetic basis.[101] As yet, the rarity of the relevant genetic abnormalities, the variability of their penetrance and the high costs of genetic screening are all important obstacles to a genomic triage. There have been recent efforts to increase the specificity of ECG screening, but further studies are needed to reach a consensus on how far such measures have modified test sensitivity and specificity. Thus, on the basis of current information, ECG screening seems inappropriate, and efforts in preventive medicine would be better directed to more common causes of premature death in the young adult. Acknowledgements The author has no conflicts of interest to declare that are directly relevant to the content of this review. No funding was used to assist in the preparation of this review.
References 1. Pelliccia A, Zipes DP, Maron BJ. Bethesda Conference #36 and the European Society of Cardiology Consensus Recommendations revisited: a comparison of U.S. and European criteria for eligibility and disqualification of competitive athletes with cardiovascular abnormalities. J Am Coll Cardiol 2008; 52: 1990-6 2. Bahr R. Can electrocardiographic screening prevent sudden death: No. BMJ 2010; 341: c4914 3. Chaitman BR. An electrocardiogram should not be included in routine preparticipation screening of young athletes. Circulation 2007; 116: 2610-5 4. Corrado D, Pelliccia A, Bjornstad HH, et al. Cardiovascular pre-participation screening of young competitive athletes for prevention of sudden death: proposal for a common European protocol. Eur Heart J 2005; 26: 516-24 5. De Ceuninck M, D’Hooghe M, D’Hooghe P. FIFA Sports Medical Committee: sudden cardiac death in football (based on chapter 4.5 from Football Medicine Manual). Zu¨rich, FIFA, 2005 [online]. Available from URL: eur.i1. yimg.com/eur.yimg.com/i/eu/fifa/do/commi.pdf [Accessed 2011 Sep 12] 6. Douglas PS. Saving athletes’ lives: a reason to find common ground? J Am Coll Cardiol 2008; 52: 1997-9 7. Drezner J, Khan K. Sudden cardiac death in young athletes [editorial]. BMJ 2008; 337: a309 8. International Olympic Committee Medical Commission. Sudden cardiovascular death in sport. Lausanne: International Olympic Committee, 2004 9. Ljungqvist A, Jenoure P, Engebretsen L, et al. The International Olympic Committee (IOC) Consensus statement on periodic health examination of elite athletes. Br J Sports Med 2009; 43: 631-43
Sports Med 2011; 41 (12)
1000
10. Maron BJ, Chaitman BR, Ackerman MJ, et al. Recommendations for physical activity and recreational sports participation for young patients with genetic cardiovascular diseases. Circulation 2004; 109: 2807-16 11. Maron BJ, Zipes D. 36th Bethesda Conference, introduction: eligibility recommendations for competitive athletes with cardiovascular abnormalities – general considerations. J Am Coll Cardiol 2005; 45: 81318-27 12. Maron BJ, Thompson PD, Ackerman MJ, et al. Recommendations and considerations related to preparticipation screening for cardiovascular abnormalities in competitive athletes: 2007 update: a scientific statement from the American Heart Association Council on Nutrition, Physical Activity, and Metabolism. Endorsed by the American College of Cardiology Foundation. Circulation 2007; 115: 1643-55 13. Myerburg RJ, Vetter VL. Electrocardiograms should be included in preparticipation screening of athletes. Circulation 2007; 116: 2616-26 14. Papadakis M, Whyte G, Sharma S. Preparticipation screening for cardiovascular abnormalities in young competitive athletes. BMJ 2008; 337: 806-12 15. Pelliccia A, Corrado D. Can electrocardiographic screening prevent sudden death: Yes. BMJ 2010; 341: c4923 16. Pigozzi F, Spataro A, Fagnani F, et al. Preparticipation screening for the detection of cardiovascular abnormalities that may cause sudden death in competitive athletes. Br J Sports Med 2003; 37: 4-5 17. Rowland T. Prevention of sudden cardiac death in young athletes: controversies and conundrums. In: Armstrong N, McManus AM, editors. The elite young athlete. Basel: S. Karger, 2011: 171-86 18. Shephard RJ. Preparticipation screening of young athletes: an effective investment? In: Shephard RJ, et al., editors. Year book of Sports Medicine, 2005. Philadelphia (PA): Elsevier/Mosby, 2005: xix-xxv 19. Shephard RJ. Mass ECG screening of young athletes. Br J Sports Med 2008; 42: 707-8 20. Sofi F, Capalbo A, Pucci N, et al. Cardiovascular evaluation, including resting and exercise electrocardiography, before participation in competitive sports: cross sectional study [editorial]. BMJ 2008; 337: a346 21. Viskin S. Antagonist: routine screening of all athletes prior to participation in competitive sports should be mandatory to prevent sudden cardiac death. Heart Rhythm 2007; 4: 525-8 22. Wilson JMG, Jungner G. Principles and practice of screening for disease. Geneva: WHO, 1968 23. Andermann A, Blancquaert I, Beauchamp S, et al. Revisiting Wilson and Jungner in the genomic age: a review of screening criteria over the past 40 years. Bull WHO 2008; 86: 317-9 24. Bayes T, Price R. An essay towards solving a problem in the doctrine of chance. Phil Trans Roy Soc Lond 1763; 53: 370-418 25. Barnard G. Studies in the history of probability and statistics: IX. Thomas Bayes’s essay towards solving a problem in the doctrine of chances. Biometrika 1958; 45: 296-315 26. Corrado D, Basso C, Pavei A, et al. Trends in sudden cardiovascular death in young competitive athletes after implementation of a preparticipation screening program. JAMA 2006; 296: 1593-601
ª 2011 Adis Data Information BV. All rights reserved.
Shephard
27. Baggish AL, Hutter AM, Wang F, et al. Cardiovascular screening in college athletes with and without electrocardiography: a cross-sectional study. Ann Int Med 2010; 152: 269-75 28. Wheeler MT, Heidenreich PA, Froelicher VF, et al. Costeffectiveness of preparticipation screening for prevention of sudden cardiac death in young athletes. Ann Int Med 2010; 152: 276-86 29. Borjesson M, Dellborg M. Is there evidence for mandating electrocardiogram as part of the pre-participation examination? Clin J Sports Med 2011; 21: 13-7 30. Drezner J, Corrado D. Is there evidence for recommending electrocardiogram as part of the pre-participation examination? Clin J Sports Med 2011; 21: 18-24 31. Fuller C, McNulty CM, Spring DA, et al. Prospective screening of 5,615 high school athletes for risk of sudden cardiac death. Med Sci Sports Exerc 1997; 29: 1131-8 32. Fuller CM. Cost effectiveness analysis of screening of high school athletes for risk of sudden cardiac death. Med Sci Sports Exerc 2000; 32: 887-90 33. Decree of the Italian Ministry of Health. Norme per la tutela sanitaria dell attivita` sportiva agnostica. Rules for the protection of athletic activity. Gazzetta Ufficale 1982 Mar 5; 63 34. Pellicia A, Maron BJ. Preparticipation cardiovascular evaluation of the competitive athlete: perspectives from the 30-year Italian experience. Am J Cardiol 1995; 75: 827-9 35. Pelliccia A, Culasso F, Di Paolo FM, et al. Prevalence of abnormal electrocardiograms in a large, unselected population undergoing pre-participation cardiovascular screening. Eur Heart J 2007; 28: 2006-10 36. Corrado D, Biffi A, Basso C, et al. 12-Lead ECG in the athlete: physiological versus pathological abnormalities. Br J Sports Med 2009; 43: 669-76 37. Corrado D, Pelliccia A, Heidbuchel H, et al. Recommendations for interpretation of 12-lead electrocardiogram in the athlete. Eur Heart J 2010; 31: 243-59 38. Drezner JA. Contemporary approaches to the identification of athletes at risk for sudden cardiac death. Curr Opin Cardiol 2008; 23: 494-501 39. Thompson PD, Levine BD. Protecting athletes from sudden cardiac death. JAMA 2006; 296: 1648-50 40. van Walraven C, Davis D, Forster AJ, et al. Timedependent bias was common in survival analyses published in leading clinical journals. J Clin Epidemiol 2004; 57: 672-82 41. Drezner JA, Chun JS, Harmon KG, et al.Survival trends in the United States following exercise-related sudden cardiac arrest in the youth: 2000 – 2006. Heart Rhythm 2008; 5: 794-9 42. Maron BJ. Hypertrophic cardiomyopathy and other causes of sudden cardiac death in young competitive athletes, with considerations for preparticipation screening and criteria for disqualification. Cardiol Clin 2007; 25: 399-414 43. Holst AG, Winkel BG, Theilade J, et al. Incidence and etiology of sports related sudden cardiac death in Denmark: implications for preparticipation. Heart Rhythm 2010; 7: 1365-71
Sports Med 2011; 41 (12)
Mandatory Screening of Athletes
44. Steinvil A, Chundadze T, Zeltser D, et al. Mandatory electrocardiographic screening of athletes to reduce their risk for sudden death proven fact or wishful thinking. J Am Coll Cardiol 2011; 57: 1291-6 45. Corrado D, Basso C, Schiavon M, et al. Screening for hypertrophic cardiomyopathy in young athletes. N Engl J Med 1998; 339: 364-9 46. Basso C, Thiene G, Corrado D, et al. Hypertrophic cardiomyopathy and sudden deaths in the young: pathological evidence of myocardial ischemia. Hum Pathol 2000; 31: 988-98 47. Maron BJ, Wolfson JK, Ciro´ A, et al. Relation of electrocardiographic abnormalities and patterns of left ventricular hypertrophy identified by two-dimensional echocardiography in patients with hypertrophic cardiomyopathy. Am J Cardiol 1983; 51: 189-94 48. Maron BJ. Preparticipation screening in athletes. In: Shephard RJ, A´strand P-O, editors. Endurance in Sport. 2nd ed. Oxford: Blackwell Scientific, 2000: 667-81 49. Pelliccia A, Di Paolo FM, Corrado DA. Evidence for efficacy of the Italian national pre-participation screening programme for identification of hypertrophic cardiomyopathy in competitive athletes. Eur Heart J 2006; 27: 2196-200 50. Pelliccia A, Fagard R, Bjornstad H, et al. Recommendations for competitive sports participation in athletes with cardiovascular disease: a consensus document from the Study Group of Sports Cardiology of the Working Group of Cardiac Rehabilitation and Exercise Physiology and the Working Group of Myocardial and Pericardial Diseases of the European Society of Cardiology. Eur Heart J 2005; 26: 1422-45 51. Haneda N, Mori C, Nishio T, et al. Heart diseases discovered by mass screening in the schools of Shimane Prefecture over a period of 5 years. Jap Circ J 1986; 50: 1325-9 52. Tasaki H, Hamasaki Y, Ichimaru T. Mass screening for heart disease of school children in Saga city: 7-year follow up study. Jap Circ J 1987; 51: 1415-20 53. Melacini P, Cianfrocca C, Calore C, et al. Marginal overlap between electrocardiographic abnormalities in patients with hypertrophic cardiomyopathy and trained athletes: implications for preparticipation screening [abstract]. Circulation 2007; 116: 765 54. Sathanandam S, Zimmerman F, Davis J, et al. ECG screening criteria for LVH does not correlate with diagnosis of hypertrophic cardiomyopathy [abstract]. Circulation 2009; 120: S647 55. Cheng TO. Hypertrophic cardiomyopathy. Int J Cardiol 2009; 131: 151-5 56. Colan SD, Lipshultz SE, Lowe AM, et al. Epidemiology and cause-specific outcomes of hypertrophic cardiomyopathy in children. Circulation 2007; 115: 773-81 57. Decker JA, Rossano JW, Smith EO, et al. Risk factors and mode of death in isolated hypertrophic cardiomyopathy in children. J Am Coll Cardiol 2009; 54: 250-4 58. Yetman AT, Hamilton RM, Benson LN, et al. Long-term outcome and prognostic determinants in children with hypertrophic cardiomyopathy. J Am Coll Cardiol 1998; 32: 1943-50
ª 2011 Adis Data Information BV. All rights reserved.
1001
59. Basavarajaiah S, Boraita A, Whyte G, et al. Ethnic differences in left ventricular remodeling in highly-trained athletes relevance to differentiating physiologic left ventricular hypertrophy from hypertrophic cardiomyopathy. J Am Coll Cardiol 2008; 51: 2256-62 60. Perez M, Fonda H, Le V-V, et al. Adding an electrocardiogram to the pre-participation examination in competitive athletes: a systematic review. Curr Probl Cardiol 2009; 34: 586-662 61. Pelliccia A, Di Paolo FM, Quattrini FM, et al. Outcomes in athletes with marked ECG repolarization abnormalities. N Engl J Med 2008; 358: 152-61 62. National Center for Health Statistics. Health, United States, 2007: with chartbook on trends in the health of Americans. Hyattsville (MD): National Center for Health Statistics, 2007 63. Shephard RJ. The economic benefits of enhanced fitness. Champaign (IL): Human Kinetics, 1986 64. Wisten A, Andersson S, Forsberg H, et al. Sudden cardiac death in the young in Sweden: electrocardiogram in relation to forensic diagnosis. J Int Med 2004; 255: 213-20 65. Borjesson M, Pelliccia A. Incidence and aetiology of sudden cardiac death in young athletes: an international perspective. Br J Sports Med 2009; 43: 644-8 66. Van Camp SP, Bloor CM, Mueller FO, et al. Nontraumatic sports death in high school and college athletes. Med Sci Sports Exerc 1995; 27: 641-7 67. Drezner JA, Rao AL, Heistand J, et al. Effectiveness of emergency response planning for sudden cardiac arrest in United States high schools with automated external defibrillators. Circulation 2009; 120: 518-25 68. Asif IM, Harmon KC, Drezner JA, et al. Incidence and etiology of sudden death in National Collegiate Athletic Association (NCAA) athletes [abstract]. Clin J Sports Med 2010; 20: 136 69. Chugh SS, Reiner K, Balaji S, et al. Population based analysis of sudden death in children: Oregon unexpected death study. Heart Rhythm 2009; 6: 1618-22 70. Atkins DL, Stewart SE, Sears GK, et al. Epidemiology and outcomes from out-of-hospital cardiac arrest in children: the Resuscitation Outcomes Consortium Epistry cardiac arrest. Circulation 2009; 119: 1484-91 71. Maron BJ, Gohman TE, Aeppli D. Prevalence of sudden cardiac death during competitive sports in Minnesota high school athletes. J Am Coll Cardiol 1998; 32: 1881-4 72. Maron BJ, Doerer JJ, Haas TS, et al. Sudden deaths in young competitive athletes: analysis of 1866 deaths in the United States, 1980-2006. Circulation 2009; 119: 1085-92 73. Solberg EE, Gjertsen F, Haugstad E, et al. Sudden death in sports among young adults in Norway. Eur J Cardiovasc Prev Rehabil 2010; 17: 337-41 74. Pelliccia A, Kinoshita N, Pisichio C, et al. Intense, uninterrupted endurance training in Olympic athletes. J Am Coll Cardiol 2010; 55: 1619-25 75. Corrado D, Basso C, Rizzoli G, et al. Does sports activity enhance the risk of sudden death in adolescents and young adults? J Am Coll Cardiol 2003; 42: 1959-63
Sports Med 2011; 41 (12)
1002
76. Harmon KG, Asif IM, Klossner D, et al. Incidence of sudden cardiac death in national collegiate athletic association athletes. Circulation 2011; 123: 1594-600 77. Drezner JA, Rogers KJ, Zimmer RR, et al. Use of automated external defibrillators at NCAA Division I universities. Med Sci Sports Exerc 2005; 37: 1487-92 78. Shephard RJ, Kavanagh T. Predicting the exercise catastrophe in the ‘‘post-coronary’’ patient. Can Fam Physician 1978; 24: 614-8 79. Shephard RJ. Sudden death: a significant hazard of exercise? Br J Sports Med 1974; 8: 101-10 80. Maron BJ, Shirani J, Poliac LC, et al. Sudden death in young competitive athletes: clinical, demographic and pathological profiles. JAMA 1996; 276: 199-204 81. de Noronha SV, Sharma S, Papadakis M, et al. Aetiology of sudden cardiac death in athletes in the United Kingdom: a pathological study. Heart 2009; 95: 1409-14 82. Basso C, Maron BJ, Corrado D, et al. Clinical profile of congenital coronary artery anomalies with origin from the wrong aortic sinus leading to sudden death in young competitive athletes. J Am Coll Cardiol 2000; 35: 1493-501 83. Corrado D, Thiene G, Nava A, et al. Sudden death in young competitive athletes: clinicopathologic correlations in 22 cases. Am J Med 1990; 89: 588-96 84. Glover DW, Maron BJ. Profile of preparticipation cardiovascular screening for high school athletes. JAMA 1998; 279: 1817-9 85. Lawless CE, Best TM. Electrocardiograms in athletes: interpretation and diagnostic accuracy. Med Sci Sports Exerc 2008; 40: 787-98 86. Maron BJ, Bodison SA, Wesley YE, et al. Results of screening a large group of intercollegiate competitive athletes for cardiovascular disease. J Am Coll Cardiol 1987; 10: 1214-21 87. Hevia AC, Fernandez M, Palacio JMA, et al. ECG as a part of the preparticipation screening programme: an old and still present international dilemma. Br J Sports Med 2011; 135 (6): 272-3 88. Wilson MG, Basavarajaiah S, Whyte GP, et al.Efficacy of personal symptom and family history questionnaires when screening for inherited pathologies: the role of electrocardiography. Br J Sports Med 2008; 42: 207-11 89. Restier-Miron L, Fayn J, Millat G, et al. Spatio-temporal electrocardiographic characterization of ventricular depolarization and repolarization abnormalities in long QT syndrome. J Electrocardiol 2007; 40: 369-74 90. Pelliccia A, Maron BJ, Culasso F, et al. Clinical significance of abnormal electrocardiographic patterns in trained athletes. Circulation 2000; 102: 278-84 91. Le V-V, Wheeler MT, Mandic S, et al. Addition of the electrocardiogram to the preparticipation examination of college athletes. Clin J Sports Med 2010; 20: 98-105
ª 2011 Adis Data Information BV. All rights reserved.
Shephard
92. Nora M, Zimmerman F, Ow P, et al. Preliminary findings of ECG screening in 9,125 young adults [abstract 3718]. Circulation 2007; 116: 845 93. Levine BD, Stray-Gundersen J. The medical care of competitive athletes: the role of the physician and individual assumption of risk. Med Sci Sports Exerc 1994; 26: 1190-2 94. Elliott PM, Sharma S, Varnava A, et al. Sudden death in hypertrophic cardiomyopathy: identification of high risk patients. J Am Coll Cardiol 2000; 36: 2212-8 95. Corrado D, Basso C, Schiavon M, et al. Screening for hypertrophic cardiomyopathy in young athletes. N Engl J Med 2007; 339: 364-9 96. Elston J, Stein K. Public health implications of establishing a national programme to screen young athletes in the UK. Br J Sports Med 2011; 45 (7): 576-82 97. Bouchard C, Shephard RJ, Stephens T. Physical activity, fitness and health. Champaign, (IL): Human Kinetics, 1994 98. McDonald IG, Daly J, Jelinek VM, et al. Opening Pandora’s box: the unpredictability of reassurance by a normal test result. BMJ 1996; 313: 329-32 99. Daw EW, Chen SN, Czernuszewicz G, et al. Genome-wide mapping of modifier chromosomal loci for human hypertrophic cardiomyopathy. Human Mol Genet 2007; 16: 2463-71 100. Geisterfer-Lowrance AAT, Kass S, Tanigawa G, et al. A molecular basis for familial hypertrophic cardiomyopathy: a b-myosin heavy chain gene mutation. Cell 1990; 62: 999-1006 101. Marian AJ, Roberts R. Recent advances in the molecular genetics of hypertrophic cardiomyopathy. Circulation 1995; 92: 1336-47 102. Schwarz K, Carrier L, Guicheney P, et al. Molecular basis of familial cardiomyopathies. Circulation 1995; 91: 532-40 103. Spirito P, Seidman CE, McKenna WJ, et al. The management of hypertrophic cardiomyopathy. N Engl J Med 1997; 336: 775-85 104. Thierfelder L, Watkins H, MacRae C, et al. a-Tropomyosin and cardiac troponin T mutations cause familial hypertrophic cardiomyopathy: a disease of the sarcomere. Cell 1994; 77: 701-12 105. Watkins H, Conner D, Thierfelder L, et al. Mutations in the cardiac myosin binding protein-C gene on chromosome 11 cause familial hypertrophic cardiomyopathy. Nature Genetics 1995; 11: 434-7
Correspondence: Professor Roy J. Shephard, Professor Emeritus, Faculty of Physical Education & Health, University of Toronto, PO Box 521, Brackendale, BC V90N 1H0, Canada. E-mail:
[email protected]
Sports Med 2011; 41 (12)
Sports Med 2011; 41 (12): 1003-1017 0112-1642/11/0012-1003/$49.95/0
REVIEW ARTICLE
ª 2011 Adis Data Information BV. All rights reserved.
Neural Network Modelling and Dynamical System Theory Are They Relevant to Study the Governing Dynamics of Association Football Players? Aviroop Dutt-Mazumder,1 Chris Button,1 Anthony Robins2 and Roger Bartlett1 1 School of Physical Education, University of Otago, Dunedin, New Zealand 2 Department of Computer Science, University of Otago, Dunedin, New Zealand
Contents Abstract. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Search Methodology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Literature Search Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Inclusion and Exclusion Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Performance Analysis of Team Sports: a Complex Challenge. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Principles Common to Dynamical Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Nonlinear Dynamics of Team Sports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Association Football Games as Dynamical Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Pattern Analysis by Artificial Neural Networks (ANNs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 Why do Performance Analysts Need ANN Models?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Principles of ANN Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.1 Supervised Learning (Learning With a Teacher) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.2 Unsupervised Learning (Learning Without a Teacher) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.3 Reinforcement Learning (Environmental Feedback) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.4 ANN Modelling for the Dynamical Sports Scene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.5 Kohonen Feature Map (KFM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.6 KFM Learning Algorithm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.7 Drawbacks of a KFM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Analysis of Team Games Using Network-Based Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1 Neural Network Modelling of the Spatiotemporal Characteristics of Team Players . . . . . . . . . . 5.2 Limitations of an ANN Approach. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Abstract
1003 1004 1004 1004 1004 1005 1005 1006 1007 1007 1008 1008 1009 1009 1009 1010 1010 1010 1011 1012 1013 1014 1015
Recent studies have explored the organization of player movements in team sports using a range of statistical tools. However, the factors that best explain the performance of association football teams remain elusive. Arguably, this is due to the high-dimensional behavioural outputs that illustrate the complex, evolving configurations typical of team games. According to dynamical system analysts, movement patterns in team sports exhibit nonlinear self-organizing features. Nonlinear processing tools (i.e. Artificial Neural Networks; ANNs) are becoming increasingly popular to investigate the
Dutt-Mazumder et al.
1004
coordination of participants in sports competitions. ANNs are well suited to describing high-dimensional data sets with nonlinear attributes, however, limited information concerning the processes required to apply ANNs exists. This review investigates the relative value of various ANN learning approaches used in sports performance analysis of team sports focusing on potential applications for association football. Sixty-two research sources were summarized and reviewed from electronic literature search engines such as SPORTDiscus, Google Scholar, IEEE Xplore, Scirus, ScienceDirect and Elsevier. Typical ANN learning algorithms can be adapted to perform pattern recognition and pattern classification. Particularly, dimensionality reduction by a Kohonen feature map (KFM) can compress chaotic highdimensional datasets into low-dimensional relevant information. Such information would be useful for developing effective training drills that should enhance self-organizing coordination among players. We conclude that ANN-based qualitative analysis is a promising approach to understand the dynamical attributes of association football players.
1. Introduction There have been a number of developments in the context of sports performance analysis, which were discussed recently in a stimulating review by Glazier.[1] For example, in the past, performance analysis has been much maligned for its fragmented application and lack of theoretical framework. However, dynamical systems theory (DST) has the power to potentially unify existing sub-disciplines such as sports biomechanics, notational analysis, motor control, physiology and psychology under one macroscopic platform.[1] In this review, we will continue where Glazier’s provocative position statement ended, by describing how emerging nonlinear analytical tools (i.e. neural network modelling) are being used to detect the key principles of dynamical systems conceptualized as team sports. The review begins by summarizing some of the key challenges facing performance analysts in the 21st century. With recent advances in motion tracking technology (such as global positioning system, radio-based signal detection, automatic video tracking, etc.), there are now ample opportunities to study the collective behaviours of groups of players with a high level of sensitivity and objectivity.[2] It is important that performance analysts adopt sensitive and meaningful analytical techniques suited to the high-dimensional datasets, which result from tracking the coordinate profiles ª 2011 Adis Data Information BV. All rights reserved.
of players. It is noted that many studies are beginning to identify hallmark features of DST within high-dimensional datasets. In the second half of the review we consider a number of nonlinear data analysis techniques that are potentially suited to summarize the patterns and structure underlying team sports behaviour. In particular, artificial neural network (ANN) modelling emerges as a promising candidate to best cope with the nonlinear, chaotic patterns hidden within the data. 2. Search Methodology 2.1 Literature Search Technique
A literature search was performed in February, 2011. Electronic database literature included SPORTDiscus, Google Scholar, IEEE Xplore, Scirus, ScienceDirect and Elsevier. The search terms used were individual words and/or combinations from the following list ‘team games’, ‘inter-personal coordination’, ‘dynamical system theory’, ‘complex systems’, ‘team behaviour’, ‘association football’, ‘artificial neural network’, ‘self-organizing maps’ and ‘performance analysis’. 2.2 Inclusion and Exclusion Criteria
The search was restricted to journal articles, conference proceedings and books written in Sports Med 2011; 41 (12)
Neural Network Modelling and Dynamical Systems in Association Football
English and German. The findings of each source were retrieved, as were any pertinent secondary citations for additional information that had been missed by the primary search. Although a total number of 142 sources were reviewed initially, these were later filtered to the 62 sources deemed most relevant in accordance with the scope of the journal guidelines. 3. Performance Analysis of Team Sports: a Complex Challenge Team sports (such as association football, rugby, handball, basketball, cricket, hockey and netball) constitute some of the most widely participated and watched of all sports played globally. Simple summary statistics in team sport, such as ball possession, territory occupied and number of shots created, provides only a partial explanation of the credibility of a team. Indeed, one of the main challenges facing performance analysts is how to identify common patterns in games that best explain the overall performance of teams. The complex player movements are hard to quantify and predict with most mathematical models. Loose couplings and fluctuations of player movements exist at many different levels on different timescales (e.g. individual players, sub-units of teams – i.e. defenders, dyadic groups, i.e. an attacker and a defender – and also globally between the two teams). Unsurprisingly, it is difficult even for experienced observers such as coaches to reliably detect and interpret such chaotic patterns. Fortunately, knowledge of the theoretical principles common to dynamical systems can allow performance analysts to assist in this challenging endeavour. 3.1 Principles Common to Dynamical Systems
Almost all team games share an implicit similarity in terms of the collective movements of players during the ebb and flow of invasive and defensive styles of play.[3] It is likely that common functional principles may underpin a number of different team games because they each represent dynamical systems. In general, dynamical systems are physical, chemical, biological or social systems that exhibit many independent degrees of ª 2011 Adis Data Information BV. All rights reserved.
1005
freedom or components that vary over space and time. These complex systems are typically open systems capable of interacting with the environment and are in a constant state of flux due to changes in internal and external flows.[4] However, dynamical systems are adaptive,[1] and surrounding constraints serve to form orderly[5] and stable relationships among the many degrees of freedom at different levels of the system.[6] Various sub-systems in a dynamical network provide multiple examples of ‘degeneracy’, the ability of elements that are structurally different to perform the same function or yield the same output. In the context of team sports, degeneracy can be viewed as the phenomenon where different players coordinate among themselves, which results in either attacking or defending processes. The team then interacts and adapts dynamically to attain a common purpose,[7] such as preventing the opponents from scoring. The process of coadaptation has been used to explain how sophisticated biological systems evolve and adapt their behaviour to satisfy long-term, evolutionary constraints.[8] At a microscopic level, relevant degrees of freedom (collective variables) are driven by the coordination between the synergetics that also influence the macroscopic behaviour of dynamic systems, i.e. circular causality (see Kelso,[9] p. 16). An invasive attack in a typical team game depicts circular causality. For example, an attacker tries to deceive the defender with a repertoire of dribbles whilst team-mates move to support the player in possession (thereby providing passing options) and collectively improving the chances of scoring. Whilst, there is no master equation to quantify the movements of different players, the conceptual framework of the dynamical systems predicts that similar patterns of play result from basic mechanisms and principles.[9] Dynamical systems have variable quantities of energy moving among their components at any given instance (excepting physical systems in equilibrium).[10] Internal energy within the system (e.g. phosphate utilization within the muscles) interacts with the available energy flows in the environment (e.g. gravity, light, sounds, friction, etc.) and accordingly alters the system (team) organization.[11] Sports Med 2011; 41 (12)
1006
Drawing upon Bernstein’s[12] ideas, it can be hypothesized that sports teams should organize fluently if the external environmental energy is harnessed efficiently. A dynamical system has an implicit relation to nonlinearity. According to DST, a slight perturbation in the present state (e.g. a counter-attack opportunity) can evolve and lead to unpredictable states of the game. However, in the long term, the randomness of the system is said to ascribe to a chaotic attractor once the system has reached a stable state (i.e. no more exchange of energy between the sub-systems). A number of stable states exist within dynamical systems (metastability), where the constraints imposed on the system gives rise to its present state.[13] Thus, different functional movement patterns of players can achieve a performance goal under specific environmental circumstances.[10] 3.2 Nonlinear Dynamics of Team Sports
The study of team sport games as dynamical systems can be approached in two manners. The first approach, which has received limited attention to date, involves the application of analytical tools of nonequilibrium thermodynamics.[14,15] The second approach involves the formulation of synergetic and nonlinear equations to model the dynamics of human movement.[16] Sports such as association football,[17] badminton,[18] basketball,[19] boxing,[20] rugby union,[21] squash[22] and tennis[23] have received this kind of analysis. This latter approach tries to quantify the human movement pattern by devising dynamical formulas, which equate the fluctuations in system behaviour under the influence of constant energy transactions with the environment (i.e. perturbations). However, studying dynamical systems using equations pertaining to perturbations has some practical disadvantages, since these equations are practically limited to weak nonlinearities.[24] To overcome this limitation, graphical methods such as Kohonen feature maps (KFMs) to analyse nonlinear behaviour have had an increasing impact. Complex social networks can be studied in order to identify players who most frequently interact with fellow players in a match[25] (sometimes ª 2011 Adis Data Information BV. All rights reserved.
Dutt-Mazumder et al.
called ‘hot-links’). This methodology can reveal the collective behaviour in team sports by quantifying the frequency of the internodal pairing. Team sports are composed of multiple and dynamic couplings among dyads of players, which function under similar principles of coordination dynamics.[18,26] Such couplings appear to exist whether a sport is contested by 2 versus 2, 5 versus 5, or 11 versus 11 players. For example, common attributes of dynamical systems have been demonstrated both in small-sided[27] and also conventional association football matches.[28] Bourbousson and colleagues[29] calculated intraplayer relative phases to analyse player movements in basketball matches. The ‘stretch index’ was also implemented, which measured the expansion and the contraction of the team court coverage as the game progressed. Perturbations in game stability were described as destabilizations of the phase relation from which the system either re-stabilizes some time thereafter, or otherwise remains destabilized up to some outcome of the game sequence.[30] It was demonstrated that the configuration of each team oscillates longitudinally and laterally with substantial coupling between the teams. However, quantitative analysis using statistical measures such as ‘stretch index’ provides summary, discrete interpretations of pattern analysis. The collective behaviour of a dynamical system cannot be studied by summing the behaviours of the sub-phases.[31] A robust approach is needed, which can calibrate and quantify the time evolving the dynamic sports scene. Recent studies have advocated the application of relative phase to measure inter-personal dynamics.[32] McGarry[31] investigated the attackerdefender dyads in basketball using the dyad pairs contained within an ellipse. The results revealed that strong attraction existed for small (one vs one) attacker-defender dyads, but the strength of the attraction weakens uniformly as the number of players increases.[29] In terms of system perturbations, it was found that the attacker tries to create spaces, whereas the defender tries to block spaces.[18] Encouragingly, it was noted that the phase relations exemplified the metastability principles of the dynamical system for dyadic interactions in basketball. Sports Med 2011; 41 (12)
Neural Network Modelling and Dynamical Systems in Association Football
The behaviour of a team game may be looked upon as an outcome of dyadic interactions at different scales. This outlook is based on the universality of complex systems, which testifies that complex systems will ascribe to similar descriptions on different levels of analysis and time scales.[9] In the past, performance indicators were used to quantify behaviour in sports. However, the relation between behaviour (action) and outcome (result) is not clearly substantiated.[22] If performance indicators for the player or a team are awarded to show their performance prowess, then this invariably points out the weak performance of the opponent player or a team.[31] Thus, performance indicators are influenced by the player-opponent synergetic. Instead, team behaviour may be better represented as a function of the interactions of players and teams.[29] In the next sub-section, we suggest that association football is showing encouraging signs in relation to adopting this emerging type of performance analysis. 3.3 Association Football Games as Dynamical Systems
Association football consists of an intricate web of complex athletic, physical skills that present a high degree of difficulty for performance (match) analysis. Like most team sports, football games result from a blend of tactical and emergent organization. Performance analysts can provide fascinating insights into how players interact and respond to sudden attacks (perturbations) and reorganize themselves to stop such instabilities on the pitch. For example, in the 2011 European Champions League competition, the Fe´de´ration Internationale de Football Association (FIFA) website displayed player positional profiles in the form of heat maps as an enhanced analysis feature for the general public.[33] Existing research has tended to examine only isolated phases of team games as illustrated earlier.[18,19,21] As yet, it is unclear whether the findings can be generalized to real-game play, in which numerous other variables including contextual constraints (e.g. score, the time left in the game, etc.) and organismic constraints (e.g. fatigue, injury, relative velocity of dyads) fluctuate substanª 2011 Adis Data Information BV. All rights reserved.
1007
tially. Intuitively, it seems inappropriate to study the sub-systems individually since they are interdependent and will not provide an overall measure of the complexity. Complexity is something ‘hidden’ within the time series of a movement sequence or strategy as it emerges over time.[24] A clear requirement for future research is to adopt more holistic analytical approaches to demonstrate the unifying principles of dynamical systems in football matches. For example, it would be interesting to investigate whether the trajectories of the players converge into a strange attractor on the pitch, during critical periods of the game. The considerable remaining challenge is how to make sense of the high-dimensional data sets needed to test such an idea. The application of conventional, assumption driven, linear statistical tools will not allow nonlinear investigations in association football. Primarily, this is because linear statistics are well suited to describing the discrete amounts of information flowing through a system, but are less well equipped to examine the time-evolving, integrated nature of system behaviour, which is where ANN modelling can play a significant role. 4. Pattern Analysis by Artificial Neural Networks (ANNs) Stergiou et al.[34] proposed several nonlinear tools to study specific features of complex human movements, e.g. Lyapunov exponent (LyE), correlation dimension (CoD), approximate entropy (ApEn) and ANNs. Such tools, which operate with few assumptions about the structure of the dataset, are undoubtedly useful and their application to complex datasets will become increasingly widespread in the future. However, ANNs have a global advantage over the nonlinear tools, primarily because the architecture addresses the descriptive factors of the global game process (velocity, dispersion rate, task type, team experience/learning, etc.) rather than individual features such as rate of divergence of trajectories in state space (LyE), fractal dimension in dynamical systems (CoD), or the complexity, regularity and predictability of a time series (ApEn). Also, ANNs overcome the computational burden of applying these individual measures separately.[35] Sports Med 2011; 41 (12)
1008
In its most popular form, an ANN is a data processing technique based on the computational principles of the brain, performing a task of interest. It captures the input-output behaviour of absolute or dynamic systems. The fundamental task of a typical ANN model is to establish global and complex behaviour defined by the connections between processing elements (nodes) and element parameters (learning rate, momentum, error criterion, activation function, etc). Crucially, their architecture lets them perform nonlinear processing and thereby model real, natural and physical problems such as the BienenstockCooper-Munro theory,[36] spiking neuron,[37] stock prediction,[38] weather forecasting,[39] etc. 4.1 Why do Performance Analysts Need ANN Models?
Studies suggest that descriptive parameters of players movement behaviour (see section 4), which may reveal important characteristics governing the dynamics of a team game, do not behave as simple periodic functions of space and time,[40,41] thereby limiting the application of conventional statistics. Hence, quantifying the highdimensional datasets is beyond the scope of linear statistical tools. From a statistical outlook, ANNs are considered as nonparametric data processing tools.[42] First of all, ANNs are typically nonlinear as the nodes have nonlinear activation functions. Second, an ANN can be designed to change its synaptic weights in real time.[35] Consequently, it is ideal for performing tasks, such as pattern classification and recognition, because ANNs not only select the appropriate pattern but can also provide information about the confidence of the decision made, thereby rejecting ambiguous patterns.[35] ANN models are fault tolerant and well suited to performing realistic tasks. Since all the data are stored across the network, partial damage to the network hardware will not terminate the performance instantaneously. Network-based qualitative techniques allow the extraction of specific striking features from the complex, high-dimensional data sets that are typical to team games. Finally, ANNs compute input data sets through a modelª 2011 Adis Data Information BV. All rights reserved.
Dutt-Mazumder et al.
free response mechanism, i.e. no prior assumptions are made on the statistical model.[35] It is often the case that the inputs to neural networks have a large number of elements (high dimensional). Where such inputs produce outputs of lower dimensionality they are often referred to as performing dimensionality reduction. The most common example of such a network is a KFM network. Thus, ANN models give rise to a new dimension of pattern analysis in a dynamical sports arena. Some classes of ANNs are characterized by an ‘energy’ value associated with every possible state, and therefore the dynamical behaviour of the system can be described as the movement of a point on a complex energy surface. The purpose of many learning algorithms is to create stable states as minima on these energy surfaces. Instead of random or chaotic behaviour, the state of the network evolves so as to perform gradient descent to stable energy minima.[43] Self-organization refers to the development of patterns or regularities of behaviour without the control of some central or external agent. It is often the case that these patterns are observable at some level of description (such as an apparently goal-directed behaviour) but they arise from mechanisms and some lower level of description (such as a reinforcement learning algorithm). ANNs that use unsupervised or reinforcement-based learning algorithms are often described as self-organizing. Since the behaviour of the networks arises from the interaction of very simple computational elements, their behaviour is also emergent. Hence, ANNs provide us with well understood tools for studying self-organization and emergent processes. In a subsequent section, we propose how performance analysts can use ANNs to extract from chaotic patterns of behaviour in team sport, the characteristics that determine successful performance. But first, let us consider at a fundamental level how ANNs operate. 4.2 Principles of ANN Learning
Primarily, ANNs have three phases, such as the training phase, test phase and application phase. ANNs undergo training when an empirical or Sports Med 2011; 41 (12)
Neural Network Modelling and Dynamical Systems in Association Football
random data set is presented into the network. The fundamental attribute of an ANN model is that when the network is trained, the internal parameters in the network adapt and respond to the learning rule, i.e. training phase. These prescribed learning rules are chosen according to the task to be studied, and constitute the learning algorithm of the solution. The generalization of the network is assessed in the test phase. Finally, the new datasets are fed into an ANN to evaluate the reality. In principle, an ANN model can be adapted according to the following three broad learning approaches: (i) supervised learning, (ii) unsupervised learning and (iii) reinforcement learning. 4.2.1 Supervised Learning (Learning With a Teacher)
The aim of supervised learning is to learn some input-output mapping between data sets using a feed-forward network (figure 1). For each input, the actual output is compared with the desired output. The weights are adjusted to get the actual output closer to the teacher. Earlier studies have used supervised learning to model swimming performance.[44] However, this form of learning is not ideal for studying dynamical systems as a target output is specified within the network. In a typical dynamical system, the outcome is subjected to perturbations, which keep evolving with time and hence cannot be predefined. 4.2.2 Unsupervised Learning (Learning Without a Teacher)
The goal of unsupervised learning is to learn regular or structured aspects of the input popEnvironment Input vector representing the state of the environment
Learning system Error signal
Teacher
Actual response
Desired response
Decision box
Fig. 1. Supervised learning in a typical artificial neural network model (adapted from Haykin[35]).
ª 2011 Adis Data Information BV. All rights reserved.
1009
Environment
Input vector representing the state of the environment
Learning system Fig. 2. Unsupervised learning in a typical artificial neural network model (adapted from Haykin[35]).
ulation (figure 2). Unsupervised learning is used to address temporal configurations of a dynamical system. In a feed-forward network, for each input, the ‘detector units’ (identify specific features) adjust their weights according to the neighbourhood, activation and learning functions that describe the network behaviour. Weights are optimized according to the prescribed task, which usually maximize the same measure of similarity. Examples of unsupervised learning include the competitive learning rule that operates with the ‘winner-takes-all’ strategy[34] and a KFM. Unsupervised learning is highly favoured by researchers concerned with pattern recognition/ classification. The absence of any target output vector ensures that the outcome of learning is not predetermined and the network will find its own solution. In the training phase the network is presented with several patterns, which it classifies according to the pattern category. In the test phase it categorizes new unknown patterns, based on the information it had extracted from training datasets. A similar concept was used to investigate tactical structures in handball,[45] movement variability[46] and coordination using gait data.[47] 4.2.3 Reinforcement Learning (Environmental Feedback)
The aim of reinforcement learning is to learn typical input-output mapping, using a feed-forward network (figure 3). For each input, the actual output generates some general feedback on the ‘goodness’ of the output, and the synaptic weights are adjusted to try and maximize the goodness. Sports Med 2011; 41 (12)
Dutt-Mazumder et al.
1010
Environment Input vector representing the state of the environment Actions
Learning system
Primary reinforcement
Critic Heuristic reinforcement
Fig. 3. Reinforcement learning in a typical artificial neural network model (adapted from Haykin[35]).
Desired outputs are not supplied by the teacher, but there is general feedback on how ‘good’ the outputs are, i.e., there is environmental feedback. The reinforcement rule is not favoured while analysing real-world problems, as it can be complex and time consuming. The application favours tasks that require nonlinear feedback or decision making (e.g. control of robotic arms and unmanned vehicles). 4.2.4 ANN Modelling for the Dynamical Sports Scene
A typical ANN architecture, which implements unsupervised learning, is the KFM. Dynamical controlled network (DyCoN), based on the KFM concept, supplies each node with an individual internal memory and an adaptive learning algorithm.[48] This ensures that DyCoN has no final state and can learn continuously and in separate phases. Also, recurrent networks, which are yet to be utilized in sports research, have the architecture to process temporal data sets. They can be tailored, according to the prescribed task with supervised learning[49] or reinforcement learning.[50] In short, there are many architectures in the family of an ANN, but exhaustive study of specific ANN models is beyond the scope of the article. We will now focus on the KFM, which can recognize and classify patterns based on the variables resulting in significant events in a game (i.e. a goal scoring opportunity). 4.2.5 Kohonen Feature Map (KFM)
Self-organizing maps are a class of ANN model, based on a method called competitive learning, where the output nodes compete amongst themselves to be activated on a perª 2011 Adis Data Information BV. All rights reserved.
group basis. For presented inputs, the output node that wins the competition is called a ‘winnertakes-all neuron’.[35] Each node represents process types (section of a match), and each cluster represents a class of a similar process type. Based on competitive learning, the architecture of the network can be fabricated to develop its model such as that of a KFM and the Willshaw and von der Malsburg model.[51] The term KFM comes from the capability to recognize patterns or clusters in the data without supervision/target data.[52] A KFM is an essential tool for analysing dynamical movement patterns in sports. The KFM architecture compresses surplus high-dimensional inputs to a low-dimensional structure (e.g. one- or two-dimensional [2-D]). Dimensionality reduction is performed to recognize and validate structures visually, yet preserve nonlinear topological relationships in the data sets.[53] This helpful feature retains the relevant information and discards irrelevant information in high-dimensional datasets, which is typical of dynamical systems. They consist of an ‘input layer’ and a presumed ‘competition layer’ (figure 4). The weights of the connections from the input nodes to a single node in the competition layer are interpreted as a reference vector in the input space, i.e. a self-organizing map represents a set of vectors in the input space and one vector for each node in the competition layer. 4.2.6 KFM Learning Algorithm
The aim of a KFM is to activate different nodes of the network to respond similarly to inputs.[35] Node weights are initialized to small random values or sampled evenly by the two largest prinSports Med 2011; 41 (12)
Neural Network Modelling and Dynamical Systems in Association Football
cipal component eigenvectors. An initial neighbourhood radius is defined and the subsequent distance between each input and output node is computed, according to the given equation (equation 1):[53] dj ¼
XN 1 i¼0
fxi ðtÞ wij ðtÞg
2
(Eq. 1)
where xi (t) = input to node i at time t and wij (t) = weight from input node i to output node j at time t, N = neighbouring nodes. For each input vector the ‘winning node’ is determined. There are two possible methods; maximal dot product: i* = argmax (wi, x); minimal euclidean distance: i* = argmin (wi, x). The KFM training phase consists of input vectors (e.g. coordinates of players) in a random order. The KFM training consists of weight updates of ‘winner node’ i* and its neighbouring nodes in Ni*i (t) are updated according to the learning equation (equation 2).
wij ðt þ 1Þ ¼ wij ðtÞ þ Z ðtÞ Ni i ðtÞ ½xi ðtÞ wij ðtÞ
1011
The neighbourhood function can be symmetric (e.g. rectangular and Gaussian) or anti-symmetric (equation 3):
N ðtÞ ¼ i i
1; if dM ði iÞ ðtÞ 0; otherwise
(Eq. 3)
where the rectangular neighbourhood function is based on the Manhattan distance between nodes dM (i*i) where l (t) is the neighbourhood parameter. The Gaussian neighbourhood function (equation 4):
Ni i ðtÞ ¼ exp
dE2 ði ; iÞ 2 ðtÞ
(Eq. 4)
where the neighbourhood function is based on the Euclidean distance between nodes dE (i*i). The Gaussian function is preferred over the rectangular, since it gives smoother mapping from input points to weight coordinates.[55] Regardless of the neighbourhood functional form, both shrink with time. The training can be stopped after it undergoes a number of predefined iterations (tmax) defined by t = tmax.
(Eq: 2Þ
where wij (t) = initial weight; wij (t + 1) = updated weight; Z (t) = learning rate, which decreases with time and Ni*i (t) = neighbourhood function (adapted from Lippmann[54])
4.2.7 Drawbacks of a KFM
Although a KFM is ideal for pattern recognition and classification, it has some crucial limitations. For example, real-time analysis (e.g. evaluating
Competition layer: each output node has an associated vector of N weights
Kohonen layer
Wi
X Input layer: each node represents one element in a population of N element input vectors Fig. 4. Kohonen feature map input patterns. Wi = weight from input node I; N = neighbouring node; X = input vector.
ª 2011 Adis Data Information BV. All rights reserved.
Sports Med 2011; 41 (12)
1012
tactical patterns during half-time in a typical football match) would require considerable effort. Also, the requisition for training data is quite large (20 000–30 000 epochs), depending upon the complexity and variability of the patterns. The logic behind this is that the impact of any individual training pattern is extremely small, whereas, the number of patterns that have to be learned is statistically large. On completion of training, the KFM freezes and further re-initiation of the training cannot be commenced.[53] Also, the learning mechanism in the conventional KFM is externally controlled, which implements predefined and rigid functions for controlling the learning process. In the subsequent sections, we will discuss some of the past studies in team games using network-based tools and suggest some novel techniques to analyse an association football team. 5. Analysis of Team Games Using Network-Based Techniques As mentioned in section 3.3, a typical team game can be considered as a dynamical system, which encourages the scope of applying network modelling tools to study team behaviour. DST proposes that player movements are susceptible to task dependent and environmental perturbations, which result in a new phase state.[17] Investigation of individual patterns is a time consuming and challenging task. Instead, similar patterns in the game processes can be identified and clustered in the KFM lattice,[56] which is beneficial in replacing complex patterns with simpler specifications. The clusters depict the process type, the highlighted nodes represent the occurrence of the pattern and the diameter of the neighbouring updated nodes highlight the frequency of the associated processes. This approach was used to evaluate the tactical structures in volleyball and squash games.[48] Also, it is challenging to represent the player movement patterns in terms of nonlinear, threedimensional (x, y, t) spatiotemporal mathematical models.[57] KFMs have categorically been implemented to study the dynamical structures typical to a team game.[58] Game intelligence in team games (handball and soccer) has been evaluated using a networkª 2011 Adis Data Information BV. All rights reserved.
Dutt-Mazumder et al.
based approach.[59] The distribution of clusters on the KFM lattice gives an idea of the quality of training, and also detects striking performancerelated features (correlated and nonrelated). The data matrix (figure 5a) was mapped into a structural mapping on the network (figure 5b) that helped to recognize qualitative features (figure 5c) and analyse the dynamical processes involved. This approach depicts correlated inputs in the form of clusters (figure 5d). A typical team game consists of several processes, and researchers face a daunting task to conduct conventional quantitative analysis (e.g. an ANOVA test) on these different processes. However, a KFM can cluster similar processes into definite classes and the game structure as a whole can be studied.[58] An intriguing study examined the movements of robotic players in a 2-D simulated soccer RoboCup.[53] The dynamics of a virtual team game has sharp contrasting features with the real football game. However, the generated data sets present useful information about the complexity of real games. Relative motion vectors of the players along with the relative position of the balls were aggregated over several time steps that resulted in high-dimensional data sets. These data sets were used to train the KFM. The different clusters highlighted different kinds of motion patterns. The frequency distribution of the data sets was able to demonstrate the credibility of one football team. This interesting technique could potentially allow researchers to objectively quantify the skill level of teams or different players in the game. Using the principles of supervised learning, the back propagation algorithm[59] has been applied to study the interpersonal dynamics of rugby players.[60] Although the principle of a typical dyadic system is the same, the constraints differ according to the rules of the game. For example in a Rugby game, a forward pass is not permissible, and also the unique shape of the ball makes a kick pass a challenging task. The data sets were in the form of spatial co-ordinates that were presented to the input layer. The input layer consisted of the four nodes (x and y co-ordinates of frontal and transverse camera positions) that provided the spatial configuration of the dyadic system (attacker and defender). The architecture Sports Med 2011; 41 (12)
Neural Network Modelling and Dynamical Systems in Association Football
a
1013
b Data matrix of training parameters 3
7
1
5
5
9
1
2
1
1
5
3
3
1
5
8
7
7
9
2
5
4
1
3
3
6
9
1
1
6
2
7
6
2
5
2
4
7
4
7
3
1
9
7
6
4
1
2
2
9
3
8
8
2 7
6
6
1
9
5
4
2
1
5
4
4
4
3
4
5
3
8
1
8
8
3
9
8
2
1
1
1
1
5
3
9
7
2
6
9
5
2
4
2
8
4
4
1
7
1
4
7
2
5
8
9
1
6
2
3
c
d
(0,0,1,0)
(1,0,1,0)
(0,1,0,1)
(1,1,1,0)
MD = midfield defense GD = goal defense GT = goal throw PA = position attack CA = counter attack C = penalty box crossing RWA = right wing attack LWA = left wing attack
MD
GT C PA
GD RWA CA
LWA
Fig. 5. High-dimensional dataset mapped on the Kohonen feature map lattice to illustrate game processes. (a) Data matrix; (b) structural mapping on the network; (c) qualitative features; and (d) dynamical processes. This approach depicts correlated inputs in the form of clusters (figure 5d).
was used to explore variables such as interpersonal distance with respect to the try line and vertical oscillations of the dyadic system. Although the study demonstrated that successful attacks had increased variability, it had no potential evidence of integrating the sub-phase (dyadic system) with the team (dynamical system). Isolated studies investigating the sub-phases of a team game would not make much difference if the researcher is set to study the overall dynamical configurations of a team game. This is because all the sub-phases of the game share implicit nested relations with the entire system (team) as a whole. The sub-phases are in constant interaction with each other, which results in a new state phase over time. One would argue that the boundaries of a dynamical system cannot be demarcated. Howª 2011 Adis Data Information BV. All rights reserved.
ever, in a typical team game such as association football, the dimension of the field is laid down by FIFA. Thus, investigations in a team game underpinning the principles of DST are possible, since all the changing states are confined in the dimensions of the field. 5.1 Neural Network Modelling of the Spatiotemporal Characteristics of Team Players
As discussed in section 4.2.2, nonlinear ANN architectures favour pattern recognition and classification; this should allow researchers to investigate the spatiotemporal behaviour of the interacting sub-systems (players) within the dynamical system (team). Although complete, real-time Sports Med 2011; 41 (12)
1014
match analysis is challenging, partly because highdimensional datasets slows ANN computational rate, it is also hypothesized that the synergetic interactions between players emerge and decay in different phases of the game. However, a fruitful approach, at least initially, could be to analyse games only during critical periods of the game. The critical periods of the game can be in the form of set plays (free kick, throw in, corners) and during the build up to successful goal attempts. The trajectories (spatiotemporal co-ordinates) of the players can be compiled on a data matrix using tracking devices that can be mapped on the KFM lattice, and patterns can be classified according to their behavioural characteristics. This process can be repeated several times and a library of data sets can be compiled for future pattern recognition. Pattern recognition would allow us to investigate the self-organizing behaviour of the players on the field. The study can be further extended by performing a qualitative analysis of the nodal structures on the KFM lattice (e.g. unified distance matrix, which maps high-dimensional data set on a 2-D lattice). One can add noise from tracking variables and induct them during ANN training. This would make the training phase more robust. Adding noise will not be random numbers, since all the variables would be gathered from the match itself. In an earlier study, the tactical structures of handball matches were investigated, using a KFM-based model.[45] The offensive patterns were clustered on the KFM lattice using the definite playing processes as input patterns. The tactical structures represent the evolving behaviour of the team through different time frames. It would be interesting to see how the decisionmaking process (tactical structures) of the players changes when they are trained with ‘attention broadening techniques’.[61] Such investigations would lead researchers to understand the principles of interpersonal behaviour in a team game. Based on these results, one can frame training drills that develop better interplayer bonding in a team game. Also, individual performance is a function of the whole team. Hence, investigations on team bonding and inter-relationships would be a crucial factor for the excellence of a team. ª 2011 Adis Data Information BV. All rights reserved.
Dutt-Mazumder et al.
5.2 Limitations of an ANN Approach
Although a neural network-based approach is advantageous to study typical team games, ANN models have some fundamental limitations. First of all, the parameters have to be defined before the training is initialized. This is a crucial factor in training the network. Unrealistic parameter settings give abstract clusters on the topographical maps. To avoid such ambiguous parameter settings, further heuristics can be used to extend the basic-training algorithm (e.g. every adjustable network parameter of the activation should have its individual learning-rate parameter). Relatedly, training the network model takes time; hence, for the time being, it is more likely that ANNs may be restricted for post-match analysis rather than to inform the tactical decisions of the coach during a game. In addition, the behaviour of a typical ANN is opaque; although an investigation contradicted the conception of ANNs as black boxes.[62] They produce output without explanations. Therefore, considerable effort and experience is required to study the internal structures of the network for some insight into its ‘reasoning’. Once the training has been completed in KFM, the model freezes and further re-initiation of the training cannot be commenced. Thus, a KFM behaves like a tool after training is terminated. The important analytical procedure of dimensionality reduction means that the researcher has an important role to play in interpreting the low-dimensional outputs in relation to actual high-dimensional game behaviours. Whilst network modelling appears a promising approach to identify the emerging patterns in association football, it is interesting to speculate why relatively few empirical studies have been published to date. A number of important procedures in network modelling, such as parameter setting, dimensionality reduction and interpretation of output, are skills that are shared by only a relatively small population of performance analysts around the world. Indeed, considerable effort and experience must be expended to study the internal structures of the network for some insight into its ‘reasoning’. Relatedly, training the Sports Med 2011; 41 (12)
Neural Network Modelling and Dynamical Systems in Association Football
network model takes time; hence, the practical application of an ANN analysis also remains a key challenge for the future. Nevertheless, in our opinion, ANN models represent more appropriate procedures for sport performance analysis in comparison to conventional statistical tools. 6. Conclusions In this review, we have discussed some of the common attributes of DST that may be present in team sports. Earlier studies have investigated the emerging patterns in team sports by implementing conventional, linear statistical tools. However, these studies have tended to analyse the various sub-units of the dynamical system as isolated entities. Instead, we argue that researchers should aim to evaluate game scenarios in which numerous other variables including contextual constraints (score, time, etc.) and organismic constraints (fatigue, injury, interpersonal distance and angles between the players) are present. Integration of the sub-phases of the game using a more robust and sensitive approach is required to identify to what extent the principles of DST can be identified within team sports. Nonlinear ANN models share many promising features that may be suited to processing spatiotemporal data sets, which are typical to team sports. ANN architectures, such as the KFM, can be applied to investigate various realmatch scenarios, such as offensive build ups in open play, counter-attacks or set-plays within a match. Pattern classification by a KFM can cluster similar game processes in a match that are camouflaged in the large number of complex variables. Network-based qualitative analysis allows researchers to investigate the dynamical attributes in a team game, and potentially develop modern training drills to promote self-organizing processes between the players. These advances would help us to better evaluate the traditional training procedures and investigate whether they facilitate creative performance in a team game, such as association football. Given the recent advances in performance analysis,[1] we can conceive that sports scientists in the near future will be required to provide coaches with tactical adª 2011 Adis Data Information BV. All rights reserved.
1015
vice that is based on a network modelling of game play. Acknowledgements No sources of funding were used to assist in the preparation of this review. The authors have no conflicts of interest that are directly relevant to the content of this review. The authors would like to thank Gavin Kennedy for his input into the article as part of their research discussion group.
References 1. Glazier PS. Game, set and match? Substantive issues and future directions in performance analysis. Sports Med 2010; 40 (8): 625-34 2. Barris S, Button C. A review of vision-based motion analysis in sport. Sports Med 2008; 38 (12): 1025-43 3. Wilson GE. A framework for teaching tactical game knowledge. J Physical Edu Rec Dance 2002; 73 (1): 20-6 4. Wallace SA. Dynamic pattern perspective of rhythmic movement: an introduction. In: Zelaznik HN, editor. Advances in motor learning and control. Champaign (IL): Human Kinetics, 1996 5. Clark JE. On becoming skillful: patterns and constraints. Res Q Exercise Sport 1995; 66: 173-83 6. Kugler PN. A morphological perspective on the origin and evolution of movement patterns. In: Wade MG, Whiting HTA, editors. Motor development in children: aspects of coordination and control. Dordrecht: Martinus Nijhoff, 1986: 459-525 7. Cooke NJ, Cannon-Bowers JA, Stout RJ. Measuring team knowledge. Hum Factors 2000; 42: 151-73 8. Kauffman S. The origins of order: self organization and selection in evolution. New York: Oxford University Press, 1993 9. Kelso JA. Dynamic patterns: the self-organization of brain and behavior. Cambridge (MA): MIT Press, 1995 10. Davids K, Button C, Bennett S. Dynamics of skill acquisition: a constraints-led approach. Champaign (IL): Human Kinetics, 2008 11. Kugler PN, Kelso JAS, Turvey MT. On the control and coordination of naturally developing systems. In: Kelso JAS, Clark E, editors. The development of movement control and coordination. New York: Wiley, 1982 12. Bernstein N. The coordination and regulation of movement. New York: Pergamon, 1967 13. Kelso JAS. Contrasting perspectives on order and regulation in movement. In: Long J, Baddeley A, editors. Attention and performance IX. Mahwah (NJ): Lawrence Erlbaum Associates, 1981: 437-57 14. Iberall AS, Soodak H. Physical basis for complex: some propositions relating levels of organisation. Collective Phenomena 1978; 3: 9-24 15. Prigogine I, George C, Hennin F, et al. A unified formulation of dynamics and thermodynamics. Chem Scr 1973; 4: 5-32 16. Davids K, Glazier P, Arau´jo D, et al. Movement systems as dynamical systems: the functional role of variability and its implications for sports medicine. Sports Med 2003; 3: 245-60
Sports Med 2011; 41 (12)
1016
17. Davids K, Arau´jo D, Shuttleworth R. Applications of dynamical systems theory to football. In: Cabri J, Reilly T, Arau´jo D, editors. Science and football V. London: Routledge, 2005 18. McGarry T, Anderson DI, Wallace SA, et al. Sport competition as a dynamical self-organizing system. J Sports Sci 2002; 20: 771-81 19. Bourbousson J, Seve C, McGarry T. Space-time coordination dynamics in basketball: part 2. The interaction between two teams. J Sports Sci 2010b; 28 (3): 349-58 20. Hristovski R, Davids K, Arau´jo D, et al. How boxers decide to punch a target: emergent behaviour in nonlinear dynamical movement systems. J Sports Sci Med 2006; 5 (CSSI): 60-73 21. Passos P, Arau´jo D, Davids K, et al. Information-governing dynamics of attacker-defender interactions in youth rugby union. J Sports Sci 2008; 26 (13): 1421-9 22. McGarry T, Franks IM. In search of invariant athletic behaviour in competitive sport systems: an example from championship squash match-play. J Sports Sci 1996; 14: 445-56 23. Palut Y, Zanone PG. A dynamical analysis of tennis: concepts and data. J Sports Sci 2005; 23: 1021-32 24. Beek PJ, Beek WJ. Tools for constructing dynamical models of rhythmic movement. Hum Mov Sci 1988; 7: 301-42 25. Passos P, Davids K, Arau´jo K, et al. Networks as novel tool for studying team ball sports as complex social systems. J Sci Med Sport 2011; 14 (2): 170-6 26. Passos P, Arau´jo D, Davids K, et al. Interpersonal pattern dynamics and adaptive behaviour in multiagent neurobiological systems: conceptual model and data. J Motor Behav 2009; 41 (5): 445-59 27. Frencken WGP, Lemmink KAPM. Team kinematics of small-sided soccer games: a systematic approach. In: Reill T, Korkusuz F, editors. Science and football VI: proceedings of the 6th World Congress on Science and Football. London: Routledge, 2008: 161-6 28. Grehaigne JF, Boutheir D, Bernard D. Dynamic-system analysis of opponent relationships in collective actions of soccer. J Sports Sci 1997; 15: 137-49 29. Bourbousson J, Seve C, McGarry T. Space-time coordination dynamics in basketball: part 1. Intra- and intercouplings among player dyads. J Sports Sci 2010a; 28 (3): 339-47 30. McGarry T, Khan MA, Franks IM. On the presence and absence of behavioral traits in sport: an example from championship squash match-play. J Sports Sci 1999; 17: 297-311 31. McGarry T. Applied and theoretical perspectives of performance analysis in sport: scientific issues and challenges. Int J Perform Anal Sport 2009; 9: 128-40 32. Lames M. Modelling the interaction in game sports: relative phase and moving correlations. J Sports Sci Med 2006; 5: 556-60 33. Players Heat Map. 2010 FIFA World Cup South Africa [online]. Available from URL: http://www.fifa.com/search/ index.html?q=heat+map [Accessed 2011 Mar 29] 34. Stergiou N, Buzzi UH, Kurz MJ, et al. Nonlinear tools in human movement. In: Stergiou N, editor. Innovative analysis of human movement. Champaign (IL): Human Kinetics, 2004: 66-77
ª 2011 Adis Data Information BV. All rights reserved.
Dutt-Mazumder et al.
35. Haykin S. Neural networks: a comprehensive foundation. 2nd ed. Upper Saddle River (NJ): Prentice-Hall Inc., 1999 36. Benuskov’a L, Diamond ME, Ebner FF. Dynamic synaptic modification threshold: computational model of experience dependent plasticity in adult rat barrel cortex. Proc Natl Acad Sci U S A 1994; 91: 4791-5 37. Maass W. Networks of spiking neurons: the third generation of neural network models. Neural Networks 1997; 10 (9): 1659-71 38. Kimoto T, Asakawa K, Yoda M, et al. Stock market prediction system with modular neural networks. In: Trippi RR, Turban E, editors. Neural networks in finance and investing. Chicago (IL): Probus Publishing Co., 1994: 343-57 39. Zhang G, Patuwo BE, Hu MY. Forecasting with artificial neural networks: the state of the art. Int J Forecasting 1998; 14: 35-62 40. Scho¨llhorn WI, Nigg BM, Stefanyshyn DJ, et al. Identification of individual walking pattern using time discrete and time continuous data sets. Gait & Posture 2002; 15: 180-6 41. Mateus J. In pursuit of an ecological and fractal approach to soccer coaching. In: Relly T, Cabri J, Arau´jo D, editors. Science and football V. London: Routledge, 2004: 561-73 42. Scho¨llhorn WI. Applications of artificial neural nets in clinical biomechanics. Clin Biomechanics 2004; 19: 876-98 43. Hertz J, Krough A, Palmer RG. Introduction to the theory of neural computation. Redwood City (CA): AddisonWesley, 1991 44. Silva AJ, Costa AM, Oliviera PM, et al. The use of neural network technology to model swimming performance. J Sports Sci Med 2007; 6: 117-25 45. Pfeiffer M, Perl J. Analysis of tactical structures in team handball by means of artificial neural networks. Int J Comput Sci Sport 2006; 5 (1): 4-14 46. Barton G, Lees A, Lisboa P, et al. Visualisation of gait data with Kohonen self-organizing neural maps. Gait & Posture 2006; 24: 46-53 47. Lamb PF. The use of self-organizing maps in analyzing multi-dimensional human movement coordination [PhD thesis]. Dunedin: University of Otago, 2010 48. Perl J. Modeling dynamic systems: basic aspects and application to performance analysis. Int J Comput Sci Sport 2004; 3 (2): 19-28 49. Jaeger H. Short term memory in ‘echo’ state networks. German National Research Institute for Computer Science 2002; GMD-Report No.: 152 [online]. Available from URL: http://www.faculty.iu-bremen.de/hjaeger/pubs/ STMEchoStatesTechRep.pdf [Accessed 2011 Sep 20] 50. Williams RJ, Zipser D. (University of California, Institute for Cognitive Science, San Diego). A learning algorithm for continually running fully recurrent neural networks: final report [report no. 8805]. San Diego (CA): Institute of Cognitive Science, University of California, San Diego, 1988 51. Konen W, Maurer T, von der Malsburg C. A fast dynamic link matching algorithm for invariant pattern recognition. Neural Networks 1994; 7: 1019-30 52. Kohonen T. Self-organizing maps. New York: Springer, 1997 53. Perl J, Dauscher P. Dynamic pattern recognition in sport by means of artificial neural networks. In: Begg R,
Sports Med 2011; 41 (12)
Neural Network Modelling and Dynamical Systems in Association Football
Palaniswami M, editors. Computational intelligence for movement science. Hershey (PA): Idea Group Publishing, 2006: 299-318 54. Lippmann RP. An introduction to computing with neural nets. IEEE ASSP Magazine 1987; 4 (3): 4-22 [online]. Available from URL: http://hawk.cs.csuci.edu/William. Wolfe/UCD/engineering/cse/Graduate/courses/CSC5542/ Lippmann.pdf [Accessed 2011 Sep 21] 55. Leondes CT. Algorithm and architectures. San Diego (CA): Academic Press, 1998 56. Perl J, Weber K. A neural network approach to pattern learning in sport. Int J Comput Sci Sport 2004; 3: 67-70 57. Hughes M. Notational analysis: a mathematical perspective. Int J Perform Anal Sport 2004; 4: 97-139 58. Perl J. Game analysis and control by means of continuously learning networks. Int J Perform Anal Sport 2002; 2: 21-35
ª 2011 Adis Data Information BV. All rights reserved.
1017
59. Gruen A. Fundamentals of videogrammetry: a review. Hum Movement Sci 1997; 16: 155-87 60. Passos P, Arau´jo D, Davids K, et al. Interpersonal dynamics in sport: the role of artificial neural networks and 3-D analysis. Behav Res Meth, 2006; 38 (4): 683-91 61. Memmert D. Can creativity be improved by an attentionbroadening training program? An exploratory study focusing on team sports. Creativity Res J 2007; 19: 281-92 62. Fetz EE, Cheney PD, Mewes K, et al. Control of forelimb activity by populations of corticomotoneuronal and rubromotoneuraonal cells. Progr Brain Res 1989; 80: 437-49
Correspondence: Mr Aviroop Dutt-Mazumder, School of Physical Education, University of Otago, 55 Union Street West, PO Box 56, Dunedin, New Zealand. E-mail:
[email protected]
Sports Med 2011; 41 (12)
Sports Med 2011; 41 (12): 1019-1032 0112-1642/11/0012-1019/$49.95/0
REVIEW ARTICLE
ª 2011 Adis Data Information BV. All rights reserved.
Assessing Musculo-Articular Stiffness Using Free Oscillations Theory, Measurement and Analysis Massimiliano Ditroilo,1 Mark Watsford,2 Aron Murphy2 and Giuseppe De Vito1,3 1 School of Public Health, Physiotherapy and Population Science, University College Dublin, Dublin, Ireland 2 Human Performance Centre, University of Technology, Sydney, NSW, Australia 3 Institute for Sport and Health, University College Dublin, Dublin, Ireland
Contents Abstract. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. The Free-Oscillation Technique and Musculo-Articular Stiffness. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Measurement and Analysis Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Damped Oscillation of Visco-Elastic Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Single Degree of Freedom Mass-Spring System: Advantages and Limitations . . . . . . . . . . . . . . 3.3 Administration of a Perturbation and Ensuing Oscillations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4 Calculation of Stiffness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5 The Stiffness-Load Relationship . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Validity and Reliability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Practical Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Abstract
1019 1020 1021 1022 1022 1024 1025 1026 1027 1029 1029 1030
Stiffness, the relationship between applied load and elastic deformation, is an important neuromechanical component related to muscular performance and injury risk. The free-oscillation technique is a popular method for stiffness assessment. There has been wide application of this technique assessing a variety of musculature, including the triceps surae, knee flexors, knee extensors and pectorals. The methodology involves the modelling of the system as a linear damped mass-spring system. The use of such a model has certain advantages and limitations that will be discussed within this review. Perhaps the major advantage of such a model is the specificity of the measure, whereby it is possible for the assessment conditions to simulate the type of loading witnessed during functional tasks and sporting situations. High levels of reliability and construct validity have typically been reported using such procedures. Despite these assurances of accuracy, a number of issues have also been identified. The literature reveals some concerns surrounding the use of a linear model for stiffness assessment. Further, procedural issues surrounding the administration of the perturbation, attention focus of the participant during the perturbation, signal collection, data processing and
Ditroilo et al.
1020
analysis, presentation of stiffness as a linear or torsional value, assessment load (single vs multiple vs maximal) and the stiffness-load relationship have been identified, and are all fundamentally related to the quality of the calculated output data. Finally, several important considerations for practitioners have been recommended to ensure the quality and consistency of stiffness data collection, processing and interpretation.
1. Introduction From a physical standpoint, the concept of stiffness describes the relationship between an applied load and the amount of elastic deformation that occurs in a structure. Generally speaking, given a specific load, different structures exhibit smaller or larger elongation thus being, respectively, stiffer or more compliant.[1] When referring to biological structures, stiffness properties have been studied from the microscopic level of a single muscle fibre, or even a portion of a muscle fibre, to the macroscopic level of the entire body. However, the measurement of stiffness of the muscle-tendon unit (MTU) has received the most attention for decades as part of the large body of literature on muscle mechanics.[2] Stiffness is an integral component of muscular contraction, athletic performance and injury risk.[2,3] In terms of function and performance, relatively higher levels of stiffness augment contractile properties enhancing the magnitude and rate of force output.[4-7] Further, elevated stiffness has been linked with enhanced performance in a range of athletic activities.[7-9] In addition, it has been proposed that stiffness is a modulator of muscular injury, as many soft tissue injuries occur during normal ranges of joint motion, rather than through over extension of the joint.[10] Since stiffness is a primary determinant of the shock absorption characteristics of an individual MTU or joint, a relatively stiffer system is less able to attenuate applied forces and therefore may be more susceptible to soft tissue injuries.[10,11] Specifically, above optimal stiffness may be associated with higher peak forces, loading rates and shocks that can lead to bone injuries, whereas low stiffness may be related to excessive joint motion and instability, leading primarily to soft tissue injuries.[2,12] ª 2011 Adis Data Information BV. All rights reserved.
Stiffness is related to the concept of elasticity, with the latter representing the mechanical resistance of a material to a stretching episode, whereas the former relates to a more complex structure that includes the material itself, along with its shape and size.[13] However, stiffness has been reported to be a separate entity to that of flexibility.[14-16] Specifically, stiffness refers to the amount of tension residing in a given system, whereas flexibility refers to the passive range of motion about a joint. Various methods of stiffness assessment have been presented in the literature. Early studies were carried out on the isolated MTU[17-19] and this work has represented, in most cases, the theoretical background for the implementation of a number of in-vivo methods aimed at measuring stiffness from a macro (whole body) to a micro (single MTU) level. One popular in-vivo method of assessing stiffness is the free-oscillation technique, which has been used in numerous investigations published in the last 40 years. This method is based on the frequency response of a perturbed system.[20] Specifically, a perturbation is applied to a joint and the ensuing naturally damped oscillations are measured.[21] This technique has been proven to be a valid and reliable measure of stiffness across numerous body segments, using single or multiple joint protocols for assessment under a variety of loading conditions.[4,6,22,23] However, because the freeoscillation technique relies on modelling the joint as a single-degree of freedom mass-spring system with a damping element,[21,24] there are some inherent limitations that will be addressed within this review. Namely, data collection methodology, data analysis and the interpretation of results are inconsistent across the different studies, with one of the aims of this review being to clarify the use of the most appropriate terminology to Sports Med 2011; 41 (12)
Musculo-Articular Stiffness Assessment
describe the outcomes of the free-oscillation stiffness test. The scope of this review has been delimited to a focus on the free oscillation technique of stiffness assessment. The procedures employed to assess stiffness using this technique are discussed and, ultimately, practitioners are provided with a detailed understanding of the specific methods used to calculate stiffness. A greater understanding of this area will potentially influence performance and injury screening protocols along with training methods for a wide variety of practitioners and exercise and sport scientists. The search of scientific literature relevant to this review was performed between June 2010 and January 2011, using the US National Library of Medicine (PubMed) database. The terms ‘stiffness’, ‘free-oscillation technique’, ‘free-vibration technique’, ‘muscle-tendon stiffness’, ‘damped oscillations’, ‘musculo-articular stiffness’, ‘joint stiffness’, ‘musculotendinous stiffness’, ‘viscoelasticity’ and ‘series elastic component’ were used. Relevant literature was also sourced from searches of related articles arising from the reference list of those obtained from the database searches. Articles were considered for inclusion if a free-oscillation technique was used to obtain a measure of stiffness. 2. The Free-Oscillation Technique and Musculo-Articular Stiffness In contrast to methods assessing stiffness of the isolated MTU or other in vivo methods, research has clarified that the free-oscillation technique detects a ‘global’ measure of stiffness of the body segment under consideration. This measure includes the stiffness of the MTU, along with skin, ligaments and articular capsule.[25,26] A number of other factors contribute to the overall level of stiffness, namely reflexive muscular contraction,[15,25,27,28] muscle length-tension characteristics, loading rate, level of agonist/antagonist co-contraction, anthropometric characteristics[27,29,30] and muscle fibre pennation angle.[16] Generally, most of these factors have been controlled by detecting real-time activity of agonist and antagonist muscles,[16,25] along with controlling the ª 2011 Adis Data Information BV. All rights reserved.
1021
perturbation amplitude to assess the reflex activity and co-contraction components.[27] Further, researchers have limited the joint angle at which the load is applied to determine the musclelength tension characteristics,[31] perturbation velocity in order to assess the loading rate,[29] and recorded segment or MTU lengths and moment arms to normalize stiffness to body size.[32] It has been questioned whether passive stiffness, i.e. the resistance to passive muscle lengthening, provides a significant contribution to stiffness measured with the free-oscillation technique. In an animal model, approximately half of the passive stiffness has been attributed to the joint capsule, and the other half to muscles and tendons.[33] A study by Blackburn et al.[16] indicated that passive stiffness accounted for 25% of the variance in stiffness measured with the freeoscillation technique while the musculature supported a load of about 30% of maximal voluntary contraction (MVC). However, more recent studies advocated that under active conditions, passive stiffness is negligible and that MTU stiffness would be the primary contributor towards this global measure of stiffness.[25,27-29] Despite the global nature of the free-oscillation technique, in some studies the authors used anthropometrical, ultrasonographical and electromyographical measurements, along with mathematical models in an attempt to obtain stiffness and viscosity parameters of the joint, muscle and tendon separately.[8,9,23,34] Further, such calculations have been used to obtain stress, strain and material modulus of the triceps surae[27,29] and hamstrings.[32] The measure of stiffness obtained with the freeoscillation technique, or indeed other methods utilizing a perturbation, e.g. imposed sinusoidal oscillations, has been previously named musculoarticular stiffness (MAS).[7,26,35] It is pertinent to emphasize that this kind of measurement is taken while the muscle is voluntarily contracted; namely, active MAS, which is different from what is measured in a relaxed state, i.e. passive stiffness. Previous research using the free-oscillation technique has referred to the outcome stiffness using various different names such as ‘stiffness of the series elastic component’,[36] ‘muscle tendon stiffness’ or ‘musculotendinous stiffness’,[5] ‘muscle Sports Med 2011; 41 (12)
Ditroilo et al.
1022
stiffness’,[24] ‘active stiffness’,[37] ‘effective stiffness’,[28] ‘muscle viscoelasticity’[23] or ‘structural stiffness’.[29] In revising the terminology of stiffness assessment, we will consistently refer to this type of measurement as MAS when the stiffness value has been obtained with the free-oscillation technique. While MTU stiffness remains the primary contributor towards global stiffness under active conditions,[25,27-29] the term MAS better encapsulates the comprehensive nature of this outcome measure. The free-oscillation technique was introduced in 1970 with an attempt to estimate the amount of elastic energy stored in contracted human muscles.[38] Damped vertical oscillations of the body were measured on a force platform after the subject landed from a vertical jump with contracted calf muscles.[38] A similar experiment was replicated by Aruin and Zatsiorky[39] who asked over 100 athletes from different sporting backgrounds to perform small vertical jumps landing on their toes with straight and tense legs. Following the impact, the force was recorded and the natural frequency, stiffness and damping coefficient determined. Single-leg drop jumps were used in another study,[40] with determination of MAS and damping constants of the lower limb by means of an accelerometer strapped to the hip. Further development of the assessment procedures was published by Shorten[21] who measured the MAS of the plantarflexor muscles using a modified experimental set-up. The subject was seated with the knee flexed at 90 supporting a load, and the distal metatarsals of the foot were positioned on the edge of a force platform. A perturbation was applied to the knee so that damped oscillations were initiated. MAS was then calculated from the mass and the frequency of the damped oscillations recorded via the force platform. Similar procedures have since been applied in a multitude of scenarios. Whilst it is beyond the scope of this review to provide information about the outcomes of the different studies, a brief summary is included below. Primarily, the free oscillation technique has been used to validate the MAS of the ankle flexors;[4,21,23,25,34,37,41,42] detect differences between genders;[29,32,43] relate ª 2011 Adis Data Information BV. All rights reserved.
MAS to other physiological parameters in special populations;[8,9,44,45] or study the effect of warm up,[15] stretching,[14,15,46] training,[47] supplementation[48] or hormone therapy[12] on MAS of the ankle flexors. One study also examined the influence of ankle braces on ankle MAS along the medio-lateral direction.[49,50] Other muscular groups assessed with this technique include the knee flexors, with one validation study[22] and other investigations examining the difference between genders,[16,28,32,43,51] the effect of menstrual cycle[51,52] and the relationship with injuries.[24,31,53-56] The knee extensors have been measured for MAS, with one validation study[22] and other investigations reporting gender- and age-related differences[28,57] or relating MAS to sport performance[7,35,58] or pathological conditions.[59] Some early studies examined the pectoral muscles, looking at the relationship between MAS and muscular performance[5,10,36,60] or MAS and flexibility.[10] Finally, the lower limb extensors have been assessed for MAS, with one validation study[6] and another study examining the relationship between stiffness and muscular performance.[11] These studies have provided a variety of practical applications concerning MAS; however, there are a number of mathematical considerations required to gain a complete understanding of this assessment technique. These will be outlined in section 3. 3. Measurement and Analysis Considerations 3.1 Damped Oscillation of Visco-Elastic Structures
The free-oscillation technique relies on modelling the system under examination, i.e. one or more MTUs crossing either a single- or multiplejoint system, as a single degree of freedom massspring system with a damping element. Following a mechanical perturbation whereby the mass is displaced and then released, the ensuing damped oscillations can be recorded by means of different sensors (see section 3.3). From a physical point of view this system can be considered a damped harmonic oscillator, consisting of a spring, a Sports Med 2011; 41 (12)
Musculo-Articular Stiffness Assessment
1023
mass and a viscous damping force.[61,62] When the system is acted upon by the force exerted by the spring, it starts to oscillate and its equation of motion can be written as follows (equation 1):
the solution to this linear second-order differential equation is:[61,62]
(Eq: 1Þ
where t is time, g is the damping coefficient and xm, F and o’ are constants representing the amplitude of the motion, the phase and the angular natural frequency, respectively. This is a sinusoidal, repetitive function so angular natural frequency is given by equation 8:
F ¼ kx bv
where F is force, k is the stiffness of the spring, x is the displacement, b is the viscous element and v is the velocity. From Newton’s second law (F = ma), and since acceleration ‘a’ is the second derivative of displacement, the following equation is obtained: d 2x m 2 ¼ kx bv dt
(Eq. 2)
Since v is the first derivative of displacement, the following equation is obtained: m
d 2x dx kx ¼ b dt2 dt
(Eq. 3)
If we make the substitutions: !2 ¼
k b and ¼ m 2m
(Eq. 4)
equation 3 becomes: d2x dx þ !2 x ¼ 0 þ 2 dt dt2
xðtÞ ¼ xm e t cosð!0 t þ FÞ
!0 ¼ 2f 0 ¼
(Eq. 5)
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi !2 2
(Eq: 6Þ
(Eq: 8Þ
where f’ is the frequency in hertz (Hz). Interestingly, this means that o’ is proportional to the stiffness (k) and inversely proportional to the mass (m) of the system, whereas the damping coefficient affects angular natural frequency by reducing it. For equation 8 to be considered valid o2 has to be ‡g2. When g2 ) o2, the system is said to be under-damped and oscillates with the amplitude gradually decreasing to zero (figure 1). The thin line in the figure represents the amplitude decreasing exponentially with time, i.e. the damping characteristics of the system, and its equation is: xðtÞ ¼ xm e t
Further, by setting: !0 ¼
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi !2 2
(Eq: 7Þ
(Eq: 9Þ
For more details on harmonic motion the reader is referred to Symon.[62]
1.2 1 0.8
γ1 γ
Arbitrary units
0.6
γ2
0.4 T
0.2 0 −0.2 0
1
2
3
4
5
6
7
8
9
10
−0.4 −0.6 −0.8 −1 −1.2 Time (s)
Fig. 1. Theoretical representation of damped oscillations. T = time period; c = the damping characteristics of the system; c1 = maximal amplitude at the first peak; c2 = maximal amplitude at the second peak.
ª 2011 Adis Data Information BV. All rights reserved.
Sports Med 2011; 41 (12)
Ditroilo et al.
1024
Figure 2A shows an elastic (spring) and a viscous damping force (fluid) acting on a body of mass m. This general model has been adapted to biological structures using the human body as a whole,[38,39] and across multiple[6,11] and single joints.[21,24] It is represented as depicted in figure 2B, with the m corresponding to the mass of body segments plus any external added mass; k corresponding primarily to series (SEC) and parallel (PEC) elastic components, but also to bones and soft tissues; and b representing the friction within the muscles and the articulations and between muscles and other tissues.[38]
A
k
k
x
3.2 Single Degree of Freedom Mass-Spring System: Advantages and Limitations
The approach of modelling a single- or a multiple-joint system as a single degree of freedom mass-spring system, incorporating a viscous element, is relatively straightforward. Since the MTU exhibits both elastic and viscous properties and other characteristics of visco-elastic materials, such as stress relaxation, creep, hysteresis and strain rate dependence, the model has been granted validity.[63,64] In one of the first investigations using the free-oscillation technique, the model was applied to the whole human body and was suggested to be valid because (i) the displacement of internal organs and soft tissue was negligible; (ii) despite the potential infinite number of degrees of freedom, the majority of the body segments vibrated at the same frequency; and (iii) the activity of the muscles under examination was more or less constant.[39] In a recent review on vertical, leg and joint stiffness, the authors, although recognizing the simplicity of the mass-spring model, stated that it accurately predicts mechanical parameters such as stiffness.[65] Despite general acceptance, modelling the MTU as a single degree of freedom mass-spring system is based on a number of assumptions, simplifications and postulations, which are not necessarily true for biological structures, such as that the spring is massless and unidimensional, the mass is concentrated at the end of the system, and stiffness, viscosity and mass are independent of time.[66] One of the main potential concerns ª 2011 Adis Data Information BV. All rights reserved.
m
F
F m
b
b
B
b
k k
m
Fig. 2. (A) Model of a damped harmonic oscillator resulting in displacement (x), consisting of a spring of stiffness (k), mass (m) and a viscous damping component (b). (B) Human joint (articular elements and muscle-tendon unit) represented as a damped harmonic oscillator. F = Force.
Sports Med 2011; 41 (12)
Musculo-Articular Stiffness Assessment
with the free-oscillation technique is the assumption of linearity of the system modelled as a damped harmonic oscillator (see section 3.1). Despite linear models providing an adequate description under constant conditions, it is well recognized that overall joint dynamics are not linear.[67] Nonetheless, a limited number of studies have addressed this issue for the free-oscillation technique.[25,41,68] The assumption of linearity of oscillation was directly investigated during a test of free-oscillation to assess ankle flexor MAS.[41,68] Specifically, the time period of the first cycle of oscillations (from first to second peak) was compared with the time period of the second cycle of oscillations (from second to third peak), using a fixed load (30% of MVC[41]) and multiple loads (from 1.3 kg to 16.3 kg[68]). The time period was significantly decreased from the first cycle to the second and, in turn, MAS increased. It was concluded that the free-oscillation data exhibits nonlinear characteristics. However, most studies adopting the freeoscillation technique have used a linear model to describe the damped oscillations and have based their stiffness calculation solely on the first cycle of oscillations. To this end, Spriggs and Hunter[68] raised the issue of which cycle is more representative of the actual stiffness value. The answer is not obvious and probably depends on the particular motor task to be dealt with. In another study, where evidence of nonlinearity was demonstrated, the authors speculated that if the nonlinear behaviour proved to be reliable, a change in natural frequency might be used as a diagnostic tool.[41] Notably, the nonlinear behaviour has been explored only for ankle stiffness. Further studies using the free-oscillation technique and investigating the assumption of linearity in the ankle and other joints are warranted. Nonlinearity of a vibrating system can be introduced through any of the elements (i.e. mass, damping element and spring) composing the system.[69] To take into account the nonlinear behaviour of the spring, an additional term has been added to the model as represented in equation 5 (see section 3.1).[25,68] However, while Spriggs and Hunter[68] admitted that this parameter presented an unknown contribution towards MAS, ª 2011 Adis Data Information BV. All rights reserved.
1025
Granata et al.[25] related this nonlinear element to the reflex components. For a more in-depth analysis of the nonlinear vibration the reader is referred to Rao.[69] In conclusion, the majority of published papers have adopted a linear model, which is undoubtedly more simple and practical to use. However, despite receiving limited attention in the literature, a nonlinear damped harmonic oscillator may indeed more closely represent the behaviour of a body segment as a result of a mechanical perturbation. 3.3 Administration of a Perturbation and Ensuing Oscillations
The application of a mechanical perturbation and the recording of the ensuing damped oscillations are inherent in the free-oscillation technique. Interestingly, the mechanical perturbation has been delivered in a number of different ways in the published research. The original studies by Cavagna[38] and Aruin and Zatsiorsky[39] adopted small vertical jumps or drop jumps,[40] and the impact as a consequence of the landing was used as the perturbation. The magnitude of the impact depends on the height of the jump.[40] However, the participants underwent several practice sessions in this study, so the height of the jump, and in turn the impact, was seemingly consistent across the trials. Most research investigating MAS of single or multiple joints describe the perturbation as a ‘downward gentle push’ in the order of 100– 200 N of magnitude,[4-6,10,11,36,37,42,47,70] with only one paper reporting the amount of pressure applied (»30 N[32]). This technique has been rationalized mechanically, such that a system will oscillate at its natural or resonant frequency (i.e. the frequency at which the system oscillates with maximal output and minimal energy expenditure[71]), regardless of the magnitude of perturbation.[5] Nonetheless, some research[25,29] has called this assumption into question stating that the difference in perturbation amplitude can influence the stiffness assessment, especially in terms of reflex contribution.[37,44] This notion was based on previous research,[72,73] which demonstrated that stiffness is larger at smaller displacement Sports Med 2011; 41 (12)
Ditroilo et al.
1026
amplitudes. Even so, a recent study measuring MAS of the hamstring musculature[53] failed to find any relationship between MAS and perturbation magnitude (peak acceleration). Furthermore, the loading rate, as measured by the perturbation velocity, has been identified as another factor that may affect MAS.[27,29] Currently, it appears that strict control of the perturbation magnitude is not required and the variability in MAS caused by slight changes in perturbation characteristics appears to be minimal relative to the overall stiffness value. However, further research monitoring the force application, joint displacement and velocity is warranted to gain a better insight into the effect of perturbation magnitude on MAS measurement. Some studies, in an attempt to generate a repeatable perturbation, applied the initial downward force with a device by using either a hand-held hammer[8,23,34] or by dropping an elastic medicine ball,[25,44,45] or by tapping the system with a 10 kg weight.[9] In particular, Faria et al.[44] advocated that the control of the magnitude of the perturbation applied could help in standardazing the contribution of reflex activity to stiffness. Preventing the subject from sighting the force application, along with a random administration of the perturbation, could also assist in impeding the reaction from the subject and allowing for a more accurate data collection.[25,29,44] Alternatively, some investigators successfully prevented a change in the contraction and co-contraction level by asking the subject to keep their attention on a monitor displaying the real-time level of muscular activity.[23,28] The damped oscillations of the joint under examination have been measured mostly by means of a force platform[4,5,10,21,36,37,42,47,70] or a legpress machine equipped with a load cell.[6,11] Another common sensor used to measure damped oscillations is an accelerometer that was attached to a DeLorme boot[24,28,31,56] or to an ankle splint,[16,30,32,43] or to the lever arm of a leg-extension or leg-curl machine.[7,22] In two other studies the damped oscillations have been assessed with a triaxial electrogoniometer[57] or recorded with a video capture system.[54,59] Interestingly, only a few articles have reported the amplitude and the ª 2011 Adis Data Information BV. All rights reserved.
velocity of the oscillation following the perturbation. Amplitude and velocity measured by means of an electrogoniometer when assessing MAS triceps surae were, on average, 1.34 and 18.69 per second, respectively,[29] and 3.85 and 95.91 per second, respectively.[27] Despite the same load (30% of MVC) used to assess MAS, and very similar characteristics for the participants included in the studies, the amplitude and velocity results were surprisingly different. In two separate studies, Blackburn et al.[43,53] reported an amplitude of the oscillation <5. The frequency of the damped oscillations (f’ in equation 11, section 3.4 below) is rarely reported in the literature. As a general rule, the higher the load supported during MAS assessment (see section 3.5), the lower the frequency of oscillation. Shorten[21] reported frequencies of 3–6 Hz for triceps surae musculature, similar to the 3.5–4.7 Hz documented in two other studies.[41,68] Lower frequencies are reported for knee extensors (1.5 Hz[28]) and pectoral musculature (1.5–3.5 Hz[10,36]) during MAS assessment. The signal can be filtered to remove higher frequency noise to clearly identify peak magnitudes. A low-pass filter is generally used with cut-offs at 4 Hz,[22,36] 10 Hz,[30] 12 Hz,[14] or 15 Hz.[44] 3.4 Calculation of Stiffness
When the damped oscillations have been recorded, stiffness (k) must be calculated. By recalling equations 4 and 8 from section 3.1, k can be computed as follows in equation 10: 2f 0 ¼
rffiffiffiffiffiffiffiffiffiffiffiffiffiffi k 2 m
(Eq. 10)
hence: k ¼ mð42 f 02 þ 2 Þ
(Eq: 11Þ
where f ’ is the frequency of the oscillation, m is b is the damping the mass of the system and ¼ 2m coefficient. Practitioners are advised that once the damped oscillatory signal has been recorded, equation 11 will yield the stiffness value. As outlined in section 3.3, the damped oscillations are mostly recorded as a force or acceleration Sports Med 2011; 41 (12)
Musculo-Articular Stiffness Assessment
1027
signal, whilst equation 7 refers to displacement. It is possible to calculate displacement by double integrating the acceleration signal. In doing so, the frequency of the signal does not change.[61] The damping coefficient (equation 12) can be defined from equation 9 in section 3.1: xðtÞ ¼ e t xm
(Eq. 12)
hence: ln
xðtÞ ¼ t xm
(Eq. 13)
and: ¼
ln xðtÞ xm t
(Eq. 14)
The damped oscillation signal (in most cases one single cycle of oscillation from the first to the second peak) has been considered to calculate the frequency and the damping coefficient. Therefore, as depicted in figure 1, frequency is calculated in equation 15: f0 ¼
1 T
(Eq. 15)
and the damping coefficient is equation 16: ¼ ln
1 2
(Eq. 16)
Occasionally, the damping coefficient has been neglected as it has been claimed that it contributes <5%[54] or only 0.71%[30] to the total stiffness value. Based on equation 5, and con b sidering that b ¼ 2m is given in units of -1 kg s , the units of measurement for k would be kg s-2, or Ns2m-1 s-2, which is Nm-1. Generally, only the first cycle of oscillations has been used to determine the frequency to enter into equation 11. However, Walshe et al.[6] and Walshe and Wilson[11] suggested the inclusion of an analysis of successive cycles of oscillations until the time period between cycles varied by more than 80 ms. Although it is not explicitly stated, it is plausible that they calculated one value of stiffness per cycle of oscillations considered, and then calculated the average stiffness
ª 2011 Adis Data Information BV. All rights reserved.
value per trial. The implementation of this procedure seems to acknowledge, and control for, the nonlinearity of the system under examination. Unfortunately, the authors did not provide any results or further details of this specific analysis, which may have offered an insight into the issue of nonlinearity of this model (see section 3.2). One limitation of this practice is that multiple oscillatory cycles are not always present due either to the subjects involved, the magnitude of the assessment load or the specific musculature assessed. MAS as calculated in equation 11 is a measure of linear stiffness. In a limited number of papers examining the MAS of lower limbs,[16,24,28,30,52,54] MAS has been transformed into torsional stiffness – the resistance offered by a structure to a rotational deformation – measured in Nm rad-1.[13] The relationship between linear and torsional stiffness is given by equation 17:[23] kt ¼ ki r2
(Eq: 17Þ
where kt is torsional stiffness, kl is linear stiffness and r is the moment arm length from the centre of rotation of the joint to the point of force application. A number of authors have used different definitions of moment arm length for the same musculature (triceps surae;[23,29] hamstring[24,52,54]), which may yield minor discrepancies in stiffness values calculations. More mathematical details about conversion between kt and kl can be found in Fukashiro et al.[23] 3.5 The Stiffness-Load Relationship
As originally presented by Shorten,[21] using a range of loads to measure MAS allows the determination of the relationship between MAS and force. Although the magnitude of the loads is not specifically reported, a visual inspection of the figure in the Shorten paper[21] permits the deduction that eight loads varying from approximately 20 Nm to 80 Nm were applied to the plantar flexors, yielding a curvilinear relationship that was steep at low loads and tended to plateau at higher loads. This has been explained, based on experiments with isolated MTU,[74] as differing stiffness contributions from the SEC, PEC and Sports Med 2011; 41 (12)
Ditroilo et al.
1028
contractile component (CC) towards the overall MAS. While stiffness of SEC and PEC are represented as constant, with PEC stiffness being negligible, CC stiffness is proportional to the load. SEC stiffness was deemed to be the primary contributor towards overall MAS at low loads, whilst at higher levels of muscle activation, the CC stiffness increases linearly with increasing muscle tension due to the number of cross bridges attached until it approaches SEC stiffness.[15,21,24] An example of such a curvilinear relationship is presented in figure 3, using data collected in our laboratory while assessing leg-flexor MAS. An alternative explanation for the curvilinear stiffness/torque relationship was advocated by Jennings and Seedhom.[54] It was previously demonstrated that intrinsic stiffness (arising from the purely mechanical properties of the structures involved) increased linearly with load and therefore contraction level, whereas reflex stiffness (due to a change in muscle activation as a result of a reflexive response) was maximal at low contraction levels and then decreased.[75,76] Accordingly, the total stiffness (the sum of intrinsic and reflex stiffness) fitted a second-order polynomial line.[75] This curvilinear stiffness-load relationship has been consistently reported across the research in a range of musculature. Assessment loads from 0 kg to 50 kg[34] or from 0 kg to 40 kg[8,23] were used to determine MAS of the triceps surae complex with a curvilinear stiffness/load relationship
1400
MAS (Nm−1)
1200 1000 800 600 400 200 0 0
10
20
30
40
50
60
Load (% of MVC) Fig. 3. Representation of the musculo-articular stiffness (MAS)load relationship. Values are mean – SD. MVC = maximal voluntary contraction.
ª 2011 Adis Data Information BV. All rights reserved.
presented. Alternatively, several authors have reported the use of multiple assessment loads ranging from 5% to 90% of MVC[4,9,37,47] or 50–200% of body mass[48] for this musculature. Similar relationships have been reported for pectoral,[10,36] knee flexors[22,24] and knee extensors[22] muscles with loads varying from 10% to 70%, 15% to 60% and 15% to 45% of MVC, respectively. Maximal MAS has been estimated from the stiffness-load curve. This procedure was used by several researchers in the 1990s, with stiffness values obtained from a variety of loads generally up to 70% of MVC.[5,6,36,70] The peak magnitude from a negative exponential curve fit was used to determine the maximal stiffness value. It was postulated that the determination of such a value exhibited a greater level of validity, as the evident MTU tension would more closely represent the types of loading witnessed in sporting activities, or situations where injuries were likely to occur. In contrast, two different investigations have reported the stiffness-load relationship to be linear. In the first, multiple loads (0 kg, 6 kg and 20% of MVC) were used to examine gender differences in knee extensor and flexor MAS.[28] However, the linearity reported was probably due to the light loads used, representing only the lower portion of the curve. In the second study, loads of 20–60% of MVC were used to measure hamstring MAS.[54] The authors hypothesized that the linear relationship was due to the assessment loads being concentrated in the middle range of muscle activity and cited Sinkjaer et al.,[75] suggesting that they managed to measure only intrinsic stiffness without reflex intervention. Alternatively, the use of a single assessment load is common in the literature. A single load corresponding to 10% of participants’ bodyweight,[16,30,32,43,52,53] or 45% of MVC has been used to assess hamstring MAS.[31,56] A load of 30%[15,27,29,44,45] or 70%[14,42] of MVC or 50% of bodyweight[49,50] was used for the assessment of triceps surae MAS. Furthermore, a load of 0 kg[57,59] or 50% of MVC was used for quadriceps MAS.[7,35] Clearly, a wide range of MAS assessment loads have been used in the literature. Generally speaking, MAS corresponding to a specific load describes the level of stiffness when Sports Med 2011; 41 (12)
Musculo-Articular Stiffness Assessment
the joint structures experience that specific level of tension, e.g. 30% of MVC is reflective of gait activities,[15] or 75% of MVC being representative of impact force encountered during sub-maximal running.[47] Therefore, the choice of a load must be justified by the purpose of the study and care should be taken when comparing research. Higher assessment loads are more likely to represent the level of force exerted during high-intensity activity and are therefore more relevant when investigating individuals involved in sport activity.[36] Investigations examining joint stability, gender differences or gait analysis tend to preferably employ lower loads to match the level of force experienced during physical tasks in normal daily life.[22] 4. Validity and Reliability Published papers investigating the validity and reliability of the free-oscillation technique are relatively scant. Given the in vivo nature of the assessment procedure, criterion-referenced validity is inherently difficult to quantify. The construct validity of the free-oscillation technique was assessed for the first time over 15 years ago by relating lower limb maximal MAS to some pertinent performance-based measures.[6] Significant correlations were reported with maximal isometric and concentric rate of force development, along with pre-stretch augmentation (difference between counter movement jump and squat jump). Similarly, Murphy et al.[4] presented a significant correlation between maximal MAS and maximal isometric rate of force development measured in the triceps surae. Likewise, in a more recent study, the construct-validity of kneeextensor and knee-flexor MAS was assessed by relating MAS to isometric rate of force development and electromechanical delay, with significant correlations exhibited for most of the loads employed.[22] In these studies, typically between 25% and 50% of the variance of the various performance-based measures was explained by variance in MAS. A different approach to the assessment of validity was used by Fukashiro et al.[23] while measuring MAS of the triceps surae. The authors ª 2011 Adis Data Information BV. All rights reserved.
1029
maintained that the technique proposed by Shorten[21] measured ‘apparent’ stiffness, which was the MAS of the whole ankle joint calculated from the ground reaction forces. By measuring the moment arms of the ankle joint, they claimed to be able to obtain the ‘true’ stiffness of the MTU, separating the tendon and muscle components and they compared their results with other studies examining stiffness of the isolated MTU. They concluded that the free oscillation technique is a valid method to measure viscoelastic characteristics. The inter-day reliability of the free-oscillation technique has been assessed for the lower limbs,[6] the knee extensors and flexors,[7,22,30,58] the triceps surae[4,42] and the pectoral musculature.[70] Overall, the level of absolute reliability (coefficient of variation) seems to be higher than the relative reliability (intraclass correlation coefficient [ICC]). The reported coefficient of variation is generally below 10%,[4,6,58] with scores as low as approximately 4%.[7,22] The ICCs are typically >0.85;[4,6,70] although, when assessing the knee flexors and extensors, a wide range of ICCs have been reported (i.e. 0.62–0.98).[7,16,22,58] The rationale for the low values has yet to be addressed; however, it is conceivably due to the experimental set-up and the mechanical disadvantage of the knee joint under the reported assessment conditions. Recently, different MAS assessment procedures were compared for reliability, i.e. multiple submaximal loads relative to the individual’s MVC versus one fixed load.[22] Interestingly, the former yielded a significantly lower level of reliability. It was proposed that the variability in determining MVC affected the assessment loads, thus contributing to greater inconsistency in MAS. This notion has been raised previously by McLachlan et al.[42] Accordingly, it appears that fixed loads are therefore more responsive when designing prospective research studies that aim to detect changing values over time.[22] 5. Practical Recommendations A synopsis of the literature that utilizes the free-oscillation technique has identified a number of important issues for the assessment of Sports Med 2011; 41 (12)
Ditroilo et al.
1030
MAS that should be considered prior to data collection: The subject needs to be thoroughly familiarized with the protocol and should be blinded to the administration of the perturbation, preferably looking at a real-time electromyographical signal monitor for the musculature being assessed, along with antagonist muscles. Strict control of the administration of the perturbation magnitude does not appear warranted. However, it is suggested that a standardized procedure be implemented to ensure the application of a consistent force. The selection of assessment load/s that is/are related to the population being assessed is important, i.e. relatively higher load/s for sportrelated assessments and relatively lower load/s for clinical situations and activities of daily living applications. The variability in MVC assessment affects the reliability of the MAS measurement. To enhance reliability, the use of fixed assessment loads is preferable when multiple testing sessions are performed over time. Furthermore, it has been identified that the analytical methodology for MAS calculation has a significant bearing on the resulting value. Specific issues to consider are as follows: The first cycle of oscillations is typically used to calculate MAS. However, the analysis of multiple cycles of oscillations may enhance the understanding of MAS properties. The frequency of oscillation varies from approximately 1–5 Hz, depending on the musculature examined and the load supported. Higher frequencies (>8–10 Hz) are recommended to be filtered out. The damping coefficient seems to be insignificant when considering the final stiffness value; therefore, its inclusion appears to be unnecessary. Under most assessment conditions, it appears appropriate for the value to remain as a linear rather than being transformed into a rotational value. The stiffness-load relationship is curvilinear, with an apparent plateau in MAS at higher assessment loads. ª 2011 Adis Data Information BV. All rights reserved.
6. Conclusions The assessment of MAS, as measured by the free-oscillation technique, has been widely adopted over the last 40 years to study a variety of paradigms including muscular performance, injury occurrence and gender differences. The term MAS best encapsulates the components of the outcome measure and the use of a uniform nomenclature will standardize application and interpretation of the free-oscillation technique. Satisfactory construct validity and reliability have been consistently reported for a range of musculature. The analytical procedures are based on fundamental equations of damped harmonic motion, and practitioners are advised to follow a specific set of instructions concerning data collection and analysis in order to quantify MAS. The assessment of MAS provides a comprehensive measure of the stiffness of the different components of the MTU and joint, with the inherent inclusion of reflex stiffness, and is somewhat limited by the assumption of linearity of the damped oscillatory profile. Despite the apparent limitations, the literature reveals that the measurement of MAS can be successfully used to gain a better understanding of neuromechanical properties in clinical situations, sport performance or indeed activities of daily living. Acknowledgements There are no conflicts of interest that are directly relevant to the content of this review and no funding was received to assist in the preparation of this review.
References 1. Topololeski LDT. Mechanical properties of materials. In: Oatis CA, editor. Kinesiology: the mechanics and patomechanics of human movement. 2nd ed. Baltimore (MD): Lippincott Williams & Wilkins, 2009: 21-35 2. Butler RJ, Crowell 3rd HP, Davis IM. Lower extremity stiffness: implications for performance and injury. Clin Biomech (Bristol, Avon) 2003; 18 (6): 511-7 3. Wilson JM, Flanagan EP. The role of elastic energy in activities with high force and power requirements: a brief review. J Strength Cond Res 2008; 22 (5): 1705-15 4. Murphy A, Watsford ML, Coutts AJ, et al. Reliability of a test of musculotendinous stiffness for the triceps-surae. Phys Ther Sport 2003; 4: 175-81
Sports Med 2011; 41 (12)
Musculo-Articular Stiffness Assessment
5. Wilson GJ, Murphy AJ, Pryor JF. Musculotendinous stiffness: its relationship to eccentric, isometric, and concentric performance. J Appl Physiol 1994; 76 (6): 2714-9 6. Walshe AD, Wilson GJ, Murphy AJ. The validity and reliability of a test of lower body musculotendinous stiffness. Eur J Appl Physiol Occup Physiol 1996; 73 (3-4): 332-9 7. Watsford M, Ditroilo M, Ferna´ndez-Pen˜a E, et al. Muscle stiffness and rate of torque development during sprint cycling. Med Sci Sports Exerc 2010; 42 (7): 1324-32 8. Fukashiro S, Abe T, Shibayama A, et al. Comparison of viscoelastic characteristics in triceps surae between Black and White athletes. Acta Physiol Scand 2002; 175 (3): 183-7 9. Dumke CL, Pfaffenroth CM, McBride JM, et al. Relationship between muscle strength, power and stiffness and running economy in trained male runners. Int J Sports Physiol Perform 2010; 5 (2): 249-61 10. Wilson GJ, Wood GA, Elliott BC. The relationship between stiffness of the musculature and static flexibility: an alternative explanation for the occurrence of muscular injury. Int J Sports Med 1991; 12 (4): 403-7 11. Walshe AD, Wilson GJ. The influence of musculotendinous stiffness on drop jump performance. Can J Appl Physiol 1997; 22 (2): 117-32 12. Faria A, Gabriel R, Abrantes J, et al. Ankle stiffness in postmenopausal women: influence of hormone therapy and menopause nature. Climacteric 2010; 13 (3): 265-70 13. Panjabi MM, White III AA. Structures under loads. In: Panjabi MM, White III AA, editors. Biomechanics in the musculoskeletal system. Philadelphia (PA): Churchill Livingstone, 2001: 103-29 14. Rees SS, Murphy AJ, Watsford ML, et al. Effects of proprioceptive neuromuscular facilitation stretching on stiffness and force-producing characteristics of the ankle in active women. J Strength Cond Res 2007; 21 (2): 572-7 15. McNair PJ, Stanley SN. Effect of passive stretching and jogging on the series elastic muscle stiffness and range of motion of the ankle joint. Br J Sports Med 1996; 30 (4): 313-7 16. Blackburn JT, Padua DA, Riemann BL, et al. The relationships between active extensibility, and passive and active stiffness of the knee flexors. J Electromyogr Kinesiol 2004; 14 (6): 683-91 17. Hill AV. The series elastic component of muscle. Proc R Soc Lond B Biol Sci 1950; 137 (887): 273-80 18. Jewell BR, Wilkie DR. An analysis of the mechanical components in frog’s striated muscle. J Physiol 1958; 143 (3): 515-40 19. Cavagna GA. The series elastic component of frog gastrocnemius. J Physiol 1970; 206 (2): 257-62 20. Cannon SC, Zahalak GI. The mechanical behavior of active human skeletal muscle in small oscillations. J Biomech 1982; 15 (2): 111-21 21. Shorten MR. Muscle elasticity and human performance. Med Sport Sci 1987; 25: 1-18 22. Ditroilo M, Watsford M, De Vito G. Validity and inter-day reliability of a free-oscillation test to measure knee extensor and knee flexor musculo-articular stiffness. J Electromyogr Kinesiol 2011; 21 (3): 492-8 23. Fukashiro S, Noda M, Shibayama A. In vivo determination of muscle viscoelasticity in the human leg. Acta Physiol Scand 2001; 172 (4): 241-8
ª 2011 Adis Data Information BV. All rights reserved.
1031
24. McNair PJ, Wood GA, Marshall RN. Stiffness of the hamstring muscles and its relationship to function in anterior cruciate ligament deficit individuals. Clin Biomech (Bristol, Avon) 1992; 7: 131-7 25. Granata KP, Wilson SE, Massimini AK, et al. Active stiffness of the ankle in response to inertial and elastic loads. J Electromyogr Kinesiol 2004; 14 (5): 599-609 26. Rabita G, Couturier A, Lambertz D. Influence of training background on the relationships between plantarflexor intrinsic stiffness and overall musculoskeletal stiffness during hopping. Eur J Appl Physiol 2008; 103 (2): 163-71 27. Blackburn JT, Padua DA, Guskiewicz KM. Muscle stiffness and spinal stretch reflex sensitivity in the triceps surae. J Athl Train 2008; 43 (1): 29-36 28. Granata KP, Wilson SE, Padua DA. Gender differences in active musculoskeletal stiffness. Part I: quantification in controlled measurements of knee joint dynamics. J Electromyogr Kinesiol 2002; 12 (2): 119-26 29. Blackburn JT, Padua DA, Weinhold PS, et al. Comparison of triceps surae structural stiffness and material modulus across sex. Clin Biomech (Bristol, Avon) 2006; 21 (2): 159-67 30. Blackburn JT, Riemann BL, Padua DA, et al. Sex comparison of extensibility, passive, and active stiffness of the knee flexors. Clin Biomech (Bristol, Avon) 2004; 19 (1): 36-43 31. Watsford ML, Murphy AJ, McLachlan KA, et al. A prospective study of the relationship between lower body stiffness and hamstring injury in professional Australian rules footballers. Am J Sports Med 2010; 38 (10): 2058-64 32. Blackburn JT, Bell DR, Norcross MF, et al. Sex comparison of hamstring structural and material properties. Clin Biomech (Bristol, Avon) 2009; 24 (1): 65-70 33. Johns RJ, Wright V. Relative importance of various tissues in joint stiffness. J Appl Physiol 1962; 17 (5): 824-8 34. Babic J, Lenarcic J. In vivo determination of triceps surae muscle-tendon complex viscoelastic properties. Eur J Appl Physiol 2004; 92 (4-5): 477-84 35. Ditroilo M, Watsford M, Ferna´ndez-Pen˜a E, et al. Effects of fatigue on muscle stiffness and intermittent sprinting during cycling. Med Sci Sports Exerc 2011; 43 (5): 837-45 36. Wilson GJ, Wood GA, Elliott BC. Optimal stiffness of series elastic component in a stretch-shorten cycle activity. J Appl Physiol 1991; 70 (2): 825-33 37. Hunter DG, Spriggs J. Investigation into the relationship between the passive flexibility and active stiffness of the ankle plantar-flexor muscles. Clin Biomech (Bristol, Avon) 2000; 15 (8): 600-6 38. Cavagna GA. Elastic bounce of the body. J Appl Physiol 1970; 29(3): 279-82 39. Aruin AS, Zatsiorsky VM. Biomechanical characteristics of human ankle-joint muscles. Eur J Appl Physiol Occup Physiol 1984; 52 (4): 400-6 40. Mizrahi J, Susak Z. In-vivo elastic and damping response of the human leg to impact forces. J Biomech Eng 1982; 104 (1): 63-6 41. Coveney VA, Hunter GD, Spriggs J. Is the behaviour of the leg during oscillation linear? J Biomech 2001; 34 (6): 827-30 42. McLachlan KA, Murphy AJ, Watsford ML, et al. The interday reliability of leg and ankle musculotendinous stiffness measures. J Appl Biomech 2006; 22 (4): 296-304
Sports Med 2011; 41 (12)
Ditroilo et al.
1032
43. Blackburn JT, Bell DR, Norcross MF, et al. Comparison of hamstring neuromechanical properties between healthy males and females and the influence of musculotendinous stiffness. J Electromyogr Kinesiol 2009; 19 (5): e362-9 44. Faria A, Gabriel R, Abrantes J, et al. Triceps-surae musculotendinous stiffness: relative differences between obese and non-obese postmenopausal women. Clin Biomech (Bristol, Avon) 2009; 24 (10): 866-71 45. Faria A, Gabriel R, Abrantes J, et al. The relationship of body mass index, age and triceps-surae musculotendinous stiffness with the foot arch structure of postmenopausal women. Clin Biomech (Bristol, Avon) 2010; 25 (6): 588-93 46. Hunter DG, Coveney VA, Spriggs J. Investigation into the effect of static stretching on the active stiffness and damping characteristics of the ankle joint plantar flexors. Phys Ther Sport 2001; 2: 15-22 47. Spurrs RW, Murphy AJ, Watsford ML. The effect of plyometric training on distance running performance. Eur J Appl Physiol 2003; 89 (1): 1-7 48. Watsford ML, Murphy AJ, Spinks WL, et al. Creatine supplementation and its effect on musculotendinous stiffness and performance. J Strength Cond Res 2003; 17 (1): 26-33 49. Zinder SM, Granata KP, Padua DA, et al. Validity and reliability of a new in vivo ankle stiffness measurement device. J Biomech 2007; 40 (2): 463-7 50. Zinder SM, Granata KP, Shultz SJ, et al. Ankle bracing and the neuromuscular factors influencing joint stiffness. J Athl Train 2009; 44 (4): 363-9 51. Bell DR, Blackburn JT, Norcorss MF, et al. Estrogen and muscle stiffness have a negative relationship in females. Knee Surg Sports Traumatol Arthrosc. Epub 2011 Jun 22 52. Bell DR, Myrick MP, Blackburn JT, et al. The effect of menstrual-cycle phase on hamstring extensibility and muscle stiffness. J Sport Rehabil 2009; 18 (4): 553-63 53. Blackburn JT, Norcross MF, Padua DA. Influences of hamstring stiffness and strength on anterior knee joint stability. Clin Biomech (Bristol, Avon). Epub 2010 Nov 6 54. Jennings AG, Seedhom BB. The measurement of muscle stiffness in anterior cruciate injuries: an experiment revisited. Clin Biomech (Bristol, Avon) 1998; 13 (2): 138-40 55. McNair PJ, Marshall RN. Landing characteristics in subjects with normal and anterior cruciate ligament deficient knee joints. Arch Phys Med Rehabil 1994; 75 (5): 584-9 56. Swanik CB, Lephart SM, Swanik KA, et al. Neuromuscular dynamic restraint in women with anterior cruciate ligament injuries. Clin Orthop Relat Res 2004; (425): 189-99 57. Oatis CA. The use of a mechanical model to describe the stiffness and damping characteristics of the knee joint in healthy adults. Phys Ther 1993; 73 (11): 740-9 58. Owen G, Cronin J, Gill N, et al. Knee extensor stiffness and functional performance. Phys Ther Sport 2005; 6: 38-44 59. Hamstra-Wright KL, Swanik CB, Ennis TY, et al. Joint stiffness and pain in individuals with patellofemoral syndrome. J Orthop Sports Phys Ther 2005; 35 (8): 495-501 60. Wilson GJ, Murphy AJ, Walshe AD, et al. Stretch shorten cycle performance: detrimental effects of not equating the
ª 2011 Adis Data Information BV. All rights reserved.
61.
62.
63.
64.
65.
66. 67. 68.
69.
70.
71.
72.
73.
74.
75.
76.
natural and movement frequencies. Res Q Exerc Sport 1996; 67 (4): 373-9 Halliday D, Resnick R, Krane KS. Oscillations. In: Halliday D, Resnick R, Krane KS, editors. Physics. 4th ed. New York: John Wiley & Sons, 1991: 315-42 Symon KR. Motion of a particle in one dimension. In: Symon KR, editor. Mechanics. Reading (MA): Addison Wesley, 1971: 39-50 McHugh MP, Magnusson SP, Gleim GW, et al. Viscoelastic stress relaxation in human skeletal muscle. Med Sci Sports Exerc 1992; 24 (12): 1375-82 Taylor DC, Dalton Jr JD, Seaber AV, et al. Viscoelastic properties of muscle-tendon units: the biomechanical effects of stretching. Am J Sports Med 1990; 18 (3): 300-9 Brughelli M, Cronin J. Influence of running velocity on vertical, leg and joint stiffness: modelling and recommendations for future research. Sports Med 2008; 38 (8): 647-57 Latash ML, Zatsiorsky WM. Joint stiffness: myth or reality? Hum Movement Sci 1993; 12: 653-92 Kearney RE, Hunter IW. System identification of human joint dynamics. Crit Rev Biomed Eng 1990; 18 (1): 55-87 Spriggs J, Hunter DG. Linear and non-linear analysis of lower-limb behaviour with heat and strecthing routines using a free-oscillation method. In: Herzog W, editor. Skeletal muscle mechanics. Chichester: John Wiley and Sons, 2000: 241-56 Rao SS. Nonlinear vibration. In: Rao SS, editor. Mechanical vibrations. Upper Saddle River (NJ): Pearson Prentice Hall, 2003; 901-26 Wilson GJ, Elliott BC, Wood GA. Stretch shorten cycle performance enhancement through flexibility training. Med Sci Sports Exerc 1992; 24 (1): 116-23 Bach TM, Chapman AE, Calvert TW. Mechanical resonance of the human body during voluntary oscillations about the ankle joint. J Biomech 1983; 16 (1): 85-90 Kearney RE, Hunter IW. Dynamics of human ankle stiffness: variation with displacement amplitude. J Biomech 1982; 15 (10): 753-6 Gottlieb GL, Agarwal GC. Compliance of single joints: elastic and plastic characteristics. J Neurophysiol 1988; 59 (3): 937-51 Morgan DL. Separation of active and passive components of short-range stiffness of muscle. Am J Physiol 1977; 232 (1): C45-9 Sinkjaer T, Toft E, Andreassen S, et al. Muscle stiffness in human ankle dorsiflexors: intrinsic and reflex components. J Neurophysiol 1988; 60 (3): 1110-21 Mirbagheri MM, Barbeau H, Kearney RE. Intrinsic and reflex contributions to human ankle stiffness: variation with activation level and position. Exp Brain Res 2000; 135 (4): 423-36
Correspondence: Dr Massimiliano Ditroilo, School of Public Health, Physiotherapy and Population Science, Health Science Building, Room A312, University College Dublin, Belfield, Dublin 4, Ireland. E-mail:
[email protected]
Sports Med 2011; 41 (12)
REVIEW ARTICLE
Sports Med 2011; 41 (12): 1033-1042 0112-1642/11/0012-1033/$49.95/0
ª 2011 Adis Data Information BV. All rights reserved.
Stability of Haematological Parameters and Its Relevance on the Athlete’s Biological Passport Model Giovanni Lombardi,1 Patrizia Lanteri,1 Alessandra Colombini,1 Giuseppe Lippi2 and Giuseppe Banfi1,3 1 IRCCS Istituto Ortopedico Galeazzi, Milan, Italy 2 U.O. di Diagnostica Ematochimica, Dipartimento di Patologia e Medicina di Laboratorio, Azienda Ospedaliero-Universitaria di Parma, Parma, Italy 3 School of Medicine, University of Milan, Milan, Italy
Contents Abstract. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Historical Background and Effect of Transportation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Recent Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Studies Performed on the Siemens Advia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Studies Performed on Sysmex Instruments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Abstract
1033 1034 1035 1036 1037 1037 1041
The stability of haematological parameters is crucial to guarantee accurate and reliable data for implementing and interpreting the athlete’s biological passport (ABP). In this model, the values of haemoglobin, reticulocytes and out-of-doping period (OFF)-score (Hb-60Ret) are used to monitor the possible variations of those parameters, and also to compare the thresholds developed by the statistical model for the single athlete on the basis of its personal values and the variance of parameters in the modal group. Nevertheless, a critical review of the current scientific literature dealing with the stability of the haematological parameters included in the ABP programme, and which are used for evaluating the probability of anomalies in the athlete’s profile, is currently lacking. In addition, we collected information from published studies, in order to supply a useful, practical and updated review to sports physicians and haematologists. There are some parameters that are highly stable, such as haemoglobin and erythrocytes (red blood cells [RBCs]), whereas others, (e.g. reticulocytes, mean RBC volume and haematocrit) appear less stable. Regardless of the methodology, the stability of haematological parameters is improved by sample refrigeration. The stability of all parameters is highly affected from high storage temperatures, whereas the stability of RBCs and haematocrit is affected by initial freezing followed by refrigeration. Transport and rotation
Lombardi et al.
1034
of tubes do not substantially influence any haematological parameter except for reticulocytes. In all the studies we reviewed that used Sysmex instrumentation, which is recommended for ABP measurements, stability was shown for 72 hours at 4C for haemoglobin, RBCs and mean curpuscular haemoglobin concentration (MCHC); up to 48 hours for reticulocytes; and up to 24 hours for haematocrit. In one study, Sysmex instrumentation shows stability extended up to 72 hours at 4C for all the parameters. There are significant differences among methods and instruments: Siemens Advia shows lower stability than Sysmex as regards to reticulocytes. However, the limit of 36 hours from blood collection to analysis as recommended by ABP scientists is reasonable to guarantee analytical quality, when samples are transported at 4C and are accompanied by a certified steadiness of this temperature. There are some parameters that are highly stable, such as haemoglobin and RBCs; whereas others, such as reticulocytes, mean cell volume and haematocrit are more unstable. The stability of haematological parameters might be improved independently from the analytical methodology, by refrigeration of the specimens.
1. Introduction The stability of haematological parameters is crucial in order to assure correct and valid data for implementing and interpreting the athlete’s biological passport (ABP).[1] ABP is an evaluation of haematological parameters, haemoglobin, reticulocytes and their combination in the outof-doping period (OFF)-score (Hb-60Ret). The ABP is a multi-step system, where only the very first step of evaluation (a ‘quantitative’ step to determine the level of abnormality of a blood profile) is performed by a software system based on Bayesian statistics. The next step, the ‘qualitative’ assessment, is performed by experts who evaluate the blood profile in view of possible blood manipulations/doping. If a questionable profile attributable to possible blood manipulation is found, then a disciplinary procedure may be opened. The organization of the ABP is rather complicated, because the collection of blood is not only performed during competitions, but also out of competition, often directly at the athletes’ residence or during training stages, in hotels or even on the field. The sum of pre-analytical issues during blood collection, environment, transport and the time elapsed from blood collection to analysis, is essential for evaluating the accuracy ª 2011 Adis Data Information BV. All rights reserved.
and fairness of data,[2] especially when abnormal results are produced. Nevertheless, a critical review of current scientific literature concerning the stability of the haematological parameters, which is included in the ABP programme, and which is used for assessing the probability of anomalies in the profile of an athlete, is currently lacking. As such, we have collected all the available data in order to supply a useful, practical and updated review to support both sports physicians and haematologists. The use of blood parameters, namely haematocrit, for antidoping purposes was first introduced by the Union Cycliste Internationale (UCI) in 1997. The threshold of 50% for males and 47% for females was used for eventually limiting the participation of athletes in official competitions under the aegis of the Union. Shortly afterwards, haemoglobin thresholds of 17 g/dL for men and 16 g/dL for women were introduced by the UCI, a practice that was then followed by other national and international sports federations. The specific issue of the analytical system used for testing and monitoring athletes was first raised with the introduction of the equations named the in-doping period (‘ON’)- and ‘OFFmodel’ that were adopted for detecting the current Sports Med 2011; 41 (12)
Parameter Stability and the Biological Passport
and recent abuse of blood-doping practices.[3] These equations were basically derived from parameters measured with the H*3 Hematology Analyzer (Bayer, now Siemens, Tarrytown, NY, USA). Although some parameters (i.e. haematocrit and haemoglobin) were universally available on different instrumentation; others, (e.g. the percentage of macrocytes or reticulocyte haematocrit) were instead peculiar to haemocytometers that were manufactured and commercialized by Bayer. As such, the use of only one system to calculate ON- and OFF-model scores represented a clear practical limitation for the widespread implementation of these indirect indexes to detect blood doping, even in a rather limited antidoping testing setting. The issue of standardizing methods was particularly evident for reticulocytes.[4] A third generation approach to detection of erythropoietin abuse was then proposed, using simplified equations based on the values of haematocrit and reticulocytes, two widely available parameters on all the commercially available haematological systems.[5] At present, the use of a Bayesian approach for evaluating the so-called ‘haematological passport’ is focused on biological variation of haematological values throughout an entire competitive season. The values of haemoglobin, reticulocytes and OFF-score (Hb-60Ret) are used for monitoring the possible variations of those parameters and for also comparing the thresholds developed by the statistical model for the single athlete on the basis of his personal values and the variance of parameters in the modal group.[1] Co-variables, such as ethnicity, age, gender, altitude, instrumentation type and sport discipline are also considered in the model. The accuracy of a pre-analytical phase is mandatory for obtaining a valid profile for an athlete; accordingly, appropriate rules and recommendations were developed and delivered by Sport Federations and Antidoping Agencies. 2. Historical Background and Effect of Transportation In 1981, a study was conducted by Nordic European scientists for the stability of blood, ª 2011 Adis Data Information BV. All rights reserved.
1035
plasma and serum constituents during a simulated transport.[6,7] On 13 occasions in the study, and at the beginning of a week, blood samples were drawn from three healthy subjects using 5 mL K3EDTA (Venoject; Terumo, Leuven, Belgium) evacuated blood collection tubes. Haematological analyses were performed on a Coulter Model S automatic blood counter (Beckman Coulter, Hialeah, FL, USA), that measures haemoglobin, erythrocytes (red blood cells [RBCs]), haematocrit, the mean cell volume (MCV) of RBCs, mean curpuscular haemoglobin concentration (MCHC) and white blood cells (WBCs). One aliquot of the samples was measured immediately and another was sent by regular mail to a different laboratory (via a 250 km postal route) and then returned again by mail. Samples were sent in a noninsulated package. The two-way transport required 2–4 days for pick up and delivery; the transported aliquots were always analysed on the fourth day after blood drawing and were kept at room temperature (RT) until analysis. The study was held between February and May, when the environmental temperature comprised between -5C to +18C. The statistical analysis was performed by using the truncated normal sequential test of Schneiderman and Armitage.[6] A constituent was considered stable when the change of its mean concentration was less than one standard deviation (SD) of the analytical method used at 95% of probability. Stability is the capability of a sample material to retain the initial property of a measured constituent for a period of time within specified limits when the sample is stored under defined conditions. The measure of instability is described as an absolute difference, a quotient or as a percentage deviation of results obtained from measurement at time zero and after a given period of time.[8] A parameter was considered stable in a study, when its average change was smaller than one coefficient of variation (CV) percentage of the assessed method, allowing a 5% risk of error.[9] In two other studies, a comparison between original and post-storage values was performed by a student’s t-test.[10,11] Moreover, random variations Sports Med 2011; 41 (12)
Lombardi et al.
1036
of the between-subjects and within-subjects intercepts were used in another study, and the curve parameters were tested for the interaction with the ambient temperature. The significance of the interaction between the fitted parameters and storage temperature gave models comparison by the use of a likelihood of the statistics to define the stability of a parameter.[12] Among haematological parameters, haemoglobin (original mean value: 13.6 g/dL) and RBCs (4.41 · 1012/L) were considered stable, when haematocrit (39.8%), MCV (89.8 fL), MCHC (33.9 g/dL) and WBCs (7.02 · 109/L) were considered unstable. The variations of unstable parameters showed an increase of haematocrit from 39.8% to 42.4% (2.60%, with a CV of 1.61%) and of MCV from 89.8 fL to 93.8 fL (4.45%, with a CV of 1.00%), a decrease of MCHC from 33.9 g/dL to 31.2 g/dL (7.96%, with a CV of 1.54%) and an increase of WBC from 7.02 to 7.21 · 109/L (2.71%, with a CV of 1.70%).[6] In a companion paper,[7] the stability of blood constituents was challenged against various storage conditions. The analyses were performed after 4 days of storage, the blood was drawn from 36 healthy subjects in sodium ethylenediaminetetraacetic acid and analysed on the Coulter S. The stability of a parameter was assured when the parameter was significantly unmodified after storage, i.e. the variation was not higher than one SD of CV measured on quality control material (CVQC), represented for blood parameters by commercial quality control 4C (Beckman Coulter, Hialeah, FL, USA). When the mean of the measures of a parameter fell within the interval (–1 SD of CVQC) but the SD fell outside, ‘no conclusion’ was reported. The authors considered all blood constituents as fairly stable, with the exception of platelets;
although, the higher deviation values were out of limit of the CVQC. The behaviour of RBCs, haemoglobin and haematocrit over different storage and simulated physical conditions, which can occur during the transport of specimens, are described in table I. It should be highlighted that high temperature causes immediate and remarkable variations of blood parameters; therefore, the storage should be performed at low temperatures for assuring the stability of measures. The effect of transport (4 hours by car before analysis) was also studied by Kouri et al.[13] who reported small potential biases in values of haemoglobin, RBCs and MCV as measured on a Siemens Advia (Siemens, Tarrytown, NY, USA), i.e. 0.7%, 0.6% and 0.7% original values, respectively. However, the effect was significant for reticulocyte percentage when a 20% variation (increase of the original value) was observed. The uncertainty of combined pre-analytical (sample collection, delay in pre-treatment and transport) and analytical effects on measurement could trigger a total error of reticulocyte determination up to 35%, whereas the same variation is negligible for other parameters, such as haemoglobin, MCV and RBCs.[13] 3. Recent Studies More recent studies on the stability of blood parameters have been published using modern haemocytometers that can automatically assess reticulocyte percentages and might be used routinely or also within an antidoping setting. The studies were performed in healthy individuals, physically active individuals[9,10,12] or in athletes, such as professional cyclists,[11,14] and all but one were performed for evaluating the stability of blood
Table I. Stability of haematological parameters over different physical conditions of storage and treatment[6,7] Physical conditions
Red blood cells (erythrocytes)
Haemoglobin
Haematocrit
Storage at 4C
Stable
Stable
Stable
Initial freezing at 20C for 1 h followed for 1 h at RT and finally at 4C for 4 d
Unstable
Stable
Unstable
Heating at +50C for 1 h
Unstable
Unstable
Unstable
Rotation at 40 rpm for 5 h
Stable
Stable
Stable
Shaking at a frequency of 600 cycles/min for 1 h (amplitude 2 cm)
Stable
Stable
Stable
rpm = revolutions per minute; RT = room temperature.
ª 2011 Adis Data Information BV. All rights reserved.
Sports Med 2011; 41 (12)
Parameter Stability and the Biological Passport
parameters for antidoping purposes.[10-12,14] However, different analysers have been used and this finding specifically deserves mention, because the method and type of instrumentation is one of the six covariables calculated together with the absolute values of haemoglobin and reticulocytes into the Bayesian-like statistical model of ABP.[15] A schematic summary of studies on haematological parameters stability is illustrated in table II. 3.1 Studies Performed on the Siemens Advia
The stability of different haematological parameters measured by the Siemens Advia is summarized in table III. Haemoglobin was a stable parameter; a significant decrease over time, independent from ambient temperature, was observed by Voss et al;[12] the magnitude of the effect, however, was negligible (<0.04 g/dL/day). The different studies reported acceptable stability, independently from the storage temperature, for haemoglobin, RBCs and MCHC, whilst haematocrit, reticulocytes and MCV were unstable even before 24 hours from the start of the analysis. Conversely, Kouri et al.[13] reported a better stability of reticulocytes, since a limited decrease (i.e. 12%) was described after 3 days of storage at 4C. The decrease of reticulocytes 24 hours from blood collection seems to be typical of the Siemens Advia methodology, as confirmed by a three-way comparison by Bourner et al.[16] of the haematology analysers, Siemens Advia 120, Sysmex XE 2100 and LH 750 (Beckman Coulter, Hileah, FL, USA). They tested reticulocyte stability after 24 hours at RT (CV thresholds are 11.50% for Siemens Advia 120; 9.0% for Beckman Coulter LH 750; and 5.90% for Sysmex XE 2100) and 72 hours at 4–8C (CV thresholds were 6.60% for Siemens Advia 120, 11.60% for Beckman Coulter LH 750 and 17.10% for Sysmex XE 2100) and then again at 72 hours RT (CV thresholds were 48.30% for Siemens Advia 120, 15.30% for Beckman Coulter LH 750 and 46.00% for Sysmex XE 2100), but the authors declared stability based on their findings of the CV results over time. However, they found that reticulocyte ª 2011 Adis Data Information BV. All rights reserved.
1037
counts decreased after 24 hours with the Siemens Advia 120; a finding not replicated by the other analysers. Wiegand et al.[17] described a possible reason, suggesting that the concentration of oxazine dye may be insufficient to pick up more mature reticulocytes. However, different instruments use different dyes and materials for counting reticulocytes. Coulter uses new methylene blue by impedentiometry, conductivity and light scattering, Sysmex uses polymethine by fluorescence, Siemens oxazine 750 by absorbance, and Abbott CD4K530 by fluorescence.[4] Therefore, the differences among systems performing reticulocyte counting and recognition of volume and content characteristics, remain a crucial problem in the legal use of the parameter. 3.2 Studies Performed on Sysmex Instruments
The stability of different haematological parameters measured by Sysmex instruments (series X) is summarized in table IV. A special focus on Sysmex instruments is mandatory, because this technology is recommended for measuring athletes’ ABP values. The effect size (percentage of values modification over time), when available in the original studies, is depicted in tables III and IV. After 72 hours, the effect size in Robinson et al.[10] was calculated against the original values. The values modification were +1.84% at 4C and +1.25% at RT for haemoglobin; +2.59% at 4C and 16.88% at RT for haematocrit; +1.85% at 4C and +1.88% at RT for RBCs; +2% at 4C and +16.2% at RT for MCV. Haemoglobin could be assumed as a stable parameter independently from temperature of storage, whereas MCV and haematocrit are only stable when the storage is performed at 4C. Reticulocyte stability deserves a specific discussion; instability is evident at RT, whereas the stability appears acceptable when samples are refrigerated. A peculiar behaviour of the absolute count of reticulocytes is the characteristics of the method, which shows a significant decrease at RT until 24% of basal value after 48 hours, then rises after 50 hours from drawing and reaching a value near to the original one after 72 hours.[10] This Sports Med 2011; 41 (12)
Subjects
Agea; gender
Materials; methodsb
Study design
Statistical analysis
64 healthy volunteers
51 – 23; NS
K2EDTA (BD) tubes; Advia 120, Sysmex XE 2100 and LH 750
Comparison between the original sample measured after 15–30 min from venepuncture and samples stored at RT and 4C for 4, 10, 24, 48 and 72 h
Stability of an analyte was defined in relation to the imprecision of the respective analytical method. A parameter was considered stable, when its average change was smaller than one CV percentagec of the assessed method, allowing a 5% risk of error. The imprecision within run and run to run was calculated using ANOVA
Robinson et al.[10] 2004
3 healthy volunteers
NS; 2 M, 1 F
K3EDTA (Sarstedt) tubes; Sysmex XT-2000i
Comparison between the original sample measured 15 min after venepuncture and samples stored at 4C in a transportable refrigerator for 4, 8, 24, 48 and 72 h
Two-tailed student’s t-test
Voss et al.[12] 2008
5 healthy volunteers
25–39; 4 M, 1 F
K2EDTA (BD) tubes; Siemen Advia 120
Comparison between the original sample measured 20 min after venepuncture and samples stored at 4C in a refrigerator or an RT for 6, 24, 30 and 48 h
When linear relationships between time and the respective dependent variables could be assumed, a term representing the interaction between time and temperature was initially incorporated. The term was removed when the interaction was statistically insignificant (p > 0.05). When we observed more complicated curve shapes, we tried to find an appropriate model by fitting either quadratic (linear mixed effects) or exponential models (nonlinear mixed effects)
Lippi et al.[14] 2005
25 professional cyclists
NS; M
K2EDTA (BD) tubes; Siemen Advia 120
Comparison between the original sample measured within 2 h after venepuncture and the same sample stored at 4C for 24 h
Student’s t-test or nonparametric Wilcoxon test
Robinson et al.[11] 2011
27 professional cyclists
NS; M
K2EDTA (BD) tubes; Sysmex XT-2000i
Comparison between the original sample measured immediately after arrival to the laboratory (<5 h from venepuncture, with a certified storage at 4C) and the sample stored at 4C for 24, 48 and 72 h
Linear regression using random and fixed effects to explain target variables. Subjects were defined as random effects and storage time was the fixed effect. F-tests were used to test for significant impact of fixed effects
a
Data presented as mean – SD or ranges where specified.
b
Methods and their manufacturers: BD, Rutherford, CA, USA; Sarstedt, Nu¨mbrecht-Rommeldorf, Germany; Advia (Siemens Tarrytown, NY, USA), Sysmex XE 2100 and Sysmex XT-2000i (Sysmex, Kobe, Japan), LH 750 (Beckman Coulter, Hialeah, FL, USA).
BD = Becton Dickinson; CV = coefficient of variation; F = female; M = male; NS = not specified; RT = room temperature.
Lombardi et al.
Sports Med 2011; 41 (12)
Study Imeri et al.[9] 2008
1038
ª 2011 Adis Data Information BV. All rights reserved.
Table II. Synopsis of studies performed on the stability of haematological parameters that could be used in an antidoping setting
Parameters
4 hb
6h
10 hb
24 hb
30 h
48 hb
72 hb
Stable, -0.01[9]
Stable[12] c
Stable, -0.05[9]
Stable, -0.06[9]
Stable[12]
Stable, +0.26[9]
Stable, -0.19[9]
Haemoglobin RT
Stable[12] Stable, -0.59[9]
4C
Stable[12]
Stable, -0.31[9]
Stable, -0.05[9]
Stable[12] Stable[12]
Stable, -0.05[14]
Stable, -0.39[9]
Stable, +0.38[9]
Stable[12]
Stable[15] Haematocrit Stable[12]
RT Stable[9]
4C
Stable[12]
Stable[9]
Unstable[12]
Unstable[12]
Unstable[12]
Stable[9]
Unstable[12]
Unstable[9]
[12]
Unstable[9]
[12]
Unstable
Unstable
Unstable, +0.90[14] Red Blood Cells (erythrocytes) Stable, -0.34[9]
RT
Stable[12]
Stable, -0.21[9]
Stable, -0.15[9]
Stable[12]
Stable[12] Stable, -0.52[9]
4C
Stable[12]
Stable, -0.46[9]
Stable, -0.33[9]
Stable, -1.03[9]
Unstable, -1.51[9]
Stable[12] Stable[12]
[12]
Stable, -0.19[9]
Stable, -0.04[9]
Parameter Stability and the Biological Passport
ª 2011 Adis Data Information BV. All rights reserved.
Table III. Stability of haematological parameters measured by Siemens Adviaa
[12]
Stable
Stable
Stable, -0.10[14] Reticulocytes (%)d Stable, -6.30[9]
RT
Stable[12]
Stable, -5.61[9]
Unstable, -37.59[9]
Unstable[12]
Unstable[12] Stable, 1.39[9]
4C
Stable[12]
Stable, +0.45[9]
Stable, +1.34[9]
Unstable, -63.81[9]
Unstable, -65.69[9]
Unstable[12] Stable[12]
[12]
Stable +6.27%[9]
Stable, -2.07[9]
[12]
Stable
Stable
Unstable, -4.80[14] Mean cell volume RT
Stable, +0.17[9]
Stable[12]
Unstable, +1.93[9]
Unstable, +7.78[9]
Unstable[12]
Unstable[12] 4C
Stable, -0.25[9]
Stable[12]
Unstable, +0.33[9]
Unstable, +0.84[9]
Unstable, +11.74[9]
Unstable, +12.84[9]
Unstable[12] Unstable[12]
[12]
Unstable
Unstable, +1.67[9]
Unstable, +2.52[9]
[12]
Unstable
Unstable, +1.0[14] Mean curpuscular haemoglobin RT
Stable[9]
Stable[9]
Stable[9]
Stable[9]
Unstable[9]
4C
Stable[9]
Stable[9]
Stable[9]
Stable[9]
Stable[9]
Stable, +0.5[14]
Values in the table are used in the ABP (haemoglobin and reticulocytes [%]) and are reported in addition to the parameters which are common to all the instruments manufactured by different producers. The time elapsed from basal analysis is depicted as number of h (e.g. 4 h = after 4 h from basic analysis).
b
Stable and unstable data are shown as percentages where stated.
c
The stability or instability for values described by Voss et al.[12] are extrapolated from original figures and from the text.
d
The stability or instability for values described by Imeri et al.[9] correspond to the absolute count of reticulocytes. The behaviour of absolute count and percentage values in Voss et al.[12] is identical.
ABP = athlete’s biological passport; RT = room temperature.
1039
Sports Med 2011; 41 (12)
a
Lombardi et al.
1040
Table IV. Stability of haematological parameters measured by Sysmex instrumentsa Parameters
4 hb
8h
10 hb
24 hb
48 hb
72 hb
Stable[10]
Stable, +0.03[9]
Stable, -0.21[9]
Stable, -0.31[9]
[10]
[10]
Haemoglobin Stable, -0.12[9]
RT
Stable 4C
[10]
Stable [9]
Stable, -0.31
Stable
[10]
Stable, -0.18
[9]
Stable[10]
Stable
Stable[10]
Stable, -0.27
Stable, -0.11[9]
Stable[10]
Stable[10]
Stable[10]
Stable, +0.64[11]
Stable, +0.64[11]
Stable, +0.64[11]
Unstable[10]
Unstable[10]
Unstable[10]
Stable, -0.24
[9]
Stable, +0.06[9]
[9]
Haematocrit Stable[10]
RT 4C
Stable
[9]
Stable
[10]
Stable[10] Stable
[10]
[9]
Stable
Stable
[9]
Stable
[10]
Unstable
[10]
Unstable[9] Unstable[10]
Stable, +0.21
Stable, +0.43[11]
Stable, +0.30[9]
Stable, +0.26[9]
Stable, +0.33[9]
Stable[10]
Stable[10]
Stable[10]
Stable, +0.21
[11]
Unstable
[9]
[11]
Red Blood Cells (erythrocytes) Stable, +0.14[9]
RT
Stable[10]
Stable, +0.13[9]
Stable[10] 4C
[9]
Stable, -0.03 Stable
Stable
[10]
Stable, +0.10
[9]
[10]
Stable, +0.02 Stable
[10]
Stable[10]
Stable, -5.08[9]
Stable[10] 4C
Stable, -0.15 Stable
[11]
[10]
Stable, +0.09[9] Stable[10]
[11]
Stable, +0.80[11]
Stable, -11.61[9]
Stable, -7.12[9]
Unstable, +14.47[9]
Unstable[10]
Unstable[10]
Unstable[10]
Stable, +1.50[9]
Stable, -0.30[9]
Unstable, -3.18[9]
[10]
[10]
c
Stable, -2.09[9]
RT
[9]
Stable, +0.80
Stable, +0.20 Reticulocytes (%)
[9]
Stable, +1.87[9] Stable
Stable[10]
Stable, +3.98[9]
[10]
Stable
Stable
Stable[10]
Stable, +1.00
Stable, -5.00[11]
Unstable, +8.97[9]
Unstable, +16.41[9]
Unstable, +21.00[9]
Unstable[10]
Unstable[10]
Unstable[10]
Stable, +0.44[9]
Unstable, +1.27[9]
Unstable, +2.40[9]
Stable[10]
Stable[10]
Stable[10]
Stable, -1.00
[11]
[11]
Mean cell volume Stable, +0.35[9]
RT
Stable[10]
Unstable, +2.15[9]
Stable[10] 4C
Stable, -0.04[9]
Stable[10]
Stable, +0.35[9]
Stable[10]
Stable, -0.11
[11]
[11]
Stable, -0.55
Stable, -0.22[11]
Mean curpuscular haemoglobin RT
Stable[9]
Stable[9]
Stable[9]
Stable[9]
Stable[9]
4C
Stable[9]
Stable[9]
Stable[9]
Stable[9]
Stable[9]
Stable, 0[11]
Stable, -0.33[11]
Stable, +0.33[11]
a
Values in the table are used in the ABP (haemoglobin and reticulocytes [%]) and are reported in addition to the parameters that are common to all the instruments manufactured by different producers. The time elapsed from basal analysis is depicted as number of h (e.g. 4 h = after 4 h from basic analysis).
b
Stable and unstable data are shown as percentages where stated.
c
The stability or instability for values described by Imeri et al.[9] correspond to absolute count of reticulocytes.
ABP = athlete’s biological passport; RT = room temperature.
effect, attributed to an artefact of the measurement channel of the instrument, is less evident when reticulocyte percentages are considered, as confirmed by a more recent study.[11] A particular sensitivity to the storage temperature of the specimens appears rather clear. ª 2011 Adis Data Information BV. All rights reserved.
Refrigeration of blood is thereby necessary to avoid analytical pitfalls as well as inaccuracies. Hence, the time limit from blood collection to analysis that was established at 36 hours,[1] seems reasonable, when the assurance of the experimental data and a certified (constant) Sports Med 2011; 41 (12)
Parameter Stability and the Biological Passport
refrigeration is taken into consideration. The extension up to 72 hours of this limit seems to be suitable for haemoglobin, but additional data are necessary for confirming old and new data in order to recommend that limit for reticulocytes.[11,18,19] In the study of Robinson et al.,[11] the stability of parameters is higher than that previously reported.[9,10] The authors highlight that preanalytical and analytical conditions are slightly different and particularly more variable than in the previously published studies.[9-11] This means that the accuracy of pre-analytical conditions might strongly influence the stability of blood constituents. For instance, the stability of the haematological parameters on Beckman Coulter LH 750 is very similar to that observed using Sysmex instruments for haemoglobin and RBCs,[9] confirming the older data that was collected on the S models.[20] The stability on the Coulter instrument was better for reticulocytes at 4C (until 72 hours) as well as for MCV (10 hours at RT, 72 hours at 4C).[9] In addition, Johnson et al.[21] found similar results between Sysmex and Coulter instruments in the samples of 100 patients, as well for Abbott (for haemoglobin, haematocrit, RBCs, and reticulocyte percentages) up to 24 hours at RT, whilst MCV was stable only on Coulter after the same period of time. 4. Conclusions The specific literature on haematological parameters stability suggests some conclusions, which can have peculiar reflections for the antidoping setting as follows: 1. There are some parameters that are highly stable, such as haemoglobin and RBCs, whereas others, such as reticulocytes, MCV and haematocrit, are more unstable. 2. The stability of haematological parameters might be improved independently from the analytical methodology by refrigeration of the specimens. 3. The stability of all parameters is highly affected by high-storage temperatures; the stability of RBCs and haematocrit is affected by initial freezing followed by refrigeration. ª 2011 Adis Data Information BV. All rights reserved.
1041
4. The stability of haemoglobin, haematocrit and RBCs is not significantly influenced by transport and the rotation of tubes. 5. The stability of reticulocytes is afflicted by the transport condition of the samples. 6. In all the studies for Sysmex instrumentation, which is recommended for ABP measurements, stability is shown at 4C up to 72 hours for haemoglobin, RBCs and MCHC, up to 48 hours for reticulocytes and up to 24 hours for haematocrit. 7. Sysmex instrumentation shows stability extended up to 72 hours at 4C for all the parameters in one study. 8. There are significant differences among methods and instruments: reticulocytes and MCV as measured by Siemens Advia have a lower stability when compared with Sysmex measurement. 9. The limit of 36 hours from collection of the blood to analysis, which is currently recommended by ABP scientists, seems reasonable to guarantee analytical quality when samples are transported at 4C and are accompanied by a certified steadiness of this temperature. Moreover, modifying factors within the preanalytical phase (i.e. fasting, circadian rhythm and previous exercise) can influence the stability of the parameters even if the real effect of these factors need to be specifically addressed.[22] Accurate and standardized pre-analytical activities allow valid assurance for ABP profiles, in order to discover unfair sportsmen and reduce misinterpretation of haematological parameter fluctuations over time. Studies on the stability of haematological parameters are crucial for defining time limits of the different haematological analysers within the antidoping setting. Finally, a standardization of protocols and especially statistical analysis for studying the parameters of haematological stability is advisable for reducing the variability in data interpretation, as well as for improving the comparability among the different epidemiological investigations. The method of Thiers et al.[23] is a viable approach. Specific studies on the stability of haematological parameters between 24 and 48 hours are advisable for validating the 36-hour limit with the possibility of extending it up to 48 hours. Sports Med 2011; 41 (12)
Lombardi et al.
1042
Acknowledgements The authors have no conflicts of interest that are directly relevant to the content of this review. No funds were used to assist in the preparation of this review. G. Banfi gave expert testimony for two athletes at the Court of Arbitration for Sport hearings regarding ABP, but this is not relevant to the present review.
References 1. Sottas PE, Robinson N, Saugy M. The athlete’s biological passport and indirect markers of blood doping. Handb Exp Pharmacol 2010; (195): 305-26 2. Banfi G, Dolci A. Preanalytical phase of sport biochemistry and haematology. J Sports Med Phys Fitness 2003; 43 (2): 223-30 3. Parisotto R, Gore CJ, Emslie KR, et al. A novel method utilising markers of altered erythropoiesis for the detection of recombinant human erythropoietin abuse in athletes. Haematologica 2000; 85 (6): 564-72 4. Banfi G. Reticulocytes in sports medicine. Sports Med 2008; 38 (3): 187-211 5. Sharpe K, Ashenden MJ, Schumacher YO. A third generation approach to detect erythropoietin abuse in athletes. Haematologica 2006; 91: 356-63 6. Berg B, Estborn B, Tryding N. Stability of serum and blood constituents during mail transport. Scand J Clin Lab Invest 1981; 41 (5): 425-30 7. Felding P, Petersen PH, Horder M. The stability of blood, plasma and serum constituents during simulated transport. Scand J Clin Lab Invest 1981; 41 (1): 35-40 8. WHO. Use of anticoagulants in diagnostic laboratory investigations & stability of blood, plasma and serum samples 2002; WHO/DIL/LAB/99.1 Rev. 2. Contributors: Banfi G, Bauer K, Brand W, et al. [online]. Available from URL: http://whqlibdoc.who.int/hq/2002/WHO_DIL_LAB_ 99.1_Rev.2.pdf [Accessed 2011 Sep 26] 9. Imeri F, Herklotz R, Risch L, et al. Stability of hematological analytes depends on the hematology analyser used: a stability study with Bayer Advia 120, Beckman Coulter LH 750 and Sysmex XE 2100. Clin Chim Acta 2008; 397: 68-71 10. Robinson N, Mangin P, Saugy M. Time and temperature dependant changes in red blood cell analytes used for testing recombinant erythropoietin abuse in sports. Clin Lab 2004; 50 (5-6): 317-23
ª 2011 Adis Data Information BV. All rights reserved.
11. Robinson N, Sottas PE, Pottgiesser T, et al. Stability and robustness of blood variables in an antidoping context. Int J Lab Hematol 2011; 33: 146-53 12. Voss SC, Flenker U, Majer B, et al. Stability tests for hematological parameters in antidoping analyses. Lab Hematol 2008; 14: 24-9 13. Kouri T, Siloaho M, Pohjavaara S, et al. Pre-analytical factors and measurement uncertainty. Scand J Clin Lab Invest 2005; 65: 463-75 14. Lippi G, Salvagno GL, Solero GP, et al. Stability of blood cell counts, hematologic parameters and reticulocytes indexes on the Advia A120 hematologic analyzer. J Lab Clin Med 2005; 146: 333-40 15. Sottas PE, Robinson N, Saugy M, et al. A forensic approach to the interpretation of blood doping markers. Law Prob Risk 2008; 7: 191-210 16. Bourner G, Dhaliwal J, Sumner J. Performance evaluation of the latest fully automated hematology analyzers in a large, commercial laboratory setting: a 4-way, side-by-side study. Lab Hematol 2005; 11 (4): 285-97 17. Wiegand G, Effenberger-Klein A, Weber R, et al. Potential pitfalls of comparative measurements of reticulocytes with flow cytometry and microscopy in prematures and infants. Clin Chem Lab Med 2004; 42 (10): 1150-4 18. Banfi G. Reticulocytes in sports medicine. Sports Med 2008; 38: 187-211 19. Cavill I, Kraaijenhagen R, Pradella R, et al. In vitro stability of the reticulocyte count. Clin Lab Haematol 1996; 18 Suppl. 1: 9-11 20. Hamilton PJ, Davidson RL. The interrelationship and stability of Coulter S-determined blood indices. J Clin Path 1973; 30: 54-7 21. Johnson M, Samuels C, Jozsa N, et al. Three-way evaluation of high throughput haematology analysers: Beckman Coulter LH 750, Abbott Cell-Dyn 4000, and Sysmex XE 2100. Lab Hematol 2002; 8: 230-8 22. Lippi G, Banfi G, Maffulli N. Preanalytical variability: the dark side of the moon in blood doping screening. Eur J Appl Physiol 2010; 109 (5): 1003-5 23. Thiers RE, Wu GT, Reed AH, et al. Sample stability: a suggested definition and method of determination. Clin Chem 1976; 22 (2): 176-83
Correspondence: Prof. Giuseppe Banfi, IRCCS Istituto Ortopedico Galeazzi and School of Medicine, University of Milan, Via R. Galeazzi 4-20161 Milano, Italy. E-mail:
[email protected]
Sports Med 2011; 41 (12)
Sports Med 2011; 41 (12): 1043-1069 0112-1642/11/0012-1043/$49.95/0
REVIEW ARTICLE
ª 2011 Adis Data Information BV. All rights reserved.
Antioxidant Supplementation during Exercise Training Beneficial or Detrimental? Tina-Tinkara Peternelj and Jeff S. Coombes School of Human Movement Studies, The University of Queensland, Brisbane, QLD, Australia
Contents Abstract. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1043 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1044 2. Basic Mechanisms of Oxidative Damage. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1044 2.1 Redox Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1044 2.2 The Antioxidant Defence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1045 2.3 Oxidative Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1047 2.4 Beneficial Roles of Reactive Species. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1047 3. Exercise-Induced Oxidative Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1048 3.1 Reactive Species in Skeletal Muscle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1048 3.2 Adaptation to Exercise-Induced Oxidative Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1048 3.3 Oxidative Stress and Muscle Damage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1049 4. Antioxidant Supplementation and Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1049 4.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1049 4.2 Antioxidant Supplementation and Exercise-Induced Oxidative Stress . . . . . . . . . . . . . . . . . . . . . 1049 4.3 Antioxidant Supplementation and Muscle Damage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1050 4.4 Antioxidant Supplements as Ergogenic Aids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1050 5. Antioxidant Supplementation Interferes with the Beneficial Effects of Exercise Training . . . . . . . . . . . 1051 5.1 Antioxidant Supplements Promote Exercise-Induced Oxidative Stress . . . . . . . . . . . . . . . . . . . . . 1057 5.2 Antioxidant Supplementation Hinders Cell Adaptation to Exercise-Induced Oxidative Stress . . . 1057 5.3 Reactive Oxygen Species Elimination and Physiological Processes . . . . . . . . . . . . . . . . . . . . . . . . 1058 6. Limitations of the Studies and Future Directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1059 7. Optimizing Nutrition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1060 7.1 Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1060 7.2 Current Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1060 8. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1061
Abstract
High levels of reactive oxygen species (ROS) produced in skeletal muscle during exercise have been associated with muscle damage and impaired muscle function. Supporting endogenous defence systems with additional oral doses of antioxidants has received much attention as a noninvasive strategy to prevent or reduce oxidative stress, decrease muscle damage and improve exercise performance. Over 150 articles have been published on this topic, with almost all of these being small-scale, low-quality studies. The consistent finding is that antioxidant supplementation attenuates exercise-induced
Peternelj & Coombes
1044
oxidative stress. However, any physiological implications of this have yet to be consistently demonstrated, with most studies reporting no effects on exercise-induced muscle damage and performance. Moreover, a growing body of evidence indicates detrimental effects of antioxidant supplementation on the health and performance benefits of exercise training. Indeed, although ROS are associated with harmful biological events, they are also essential to the development and optimal function of every cell. The aim of this review is to present and discuss 23 studies that have shown that antioxidant supplementation interferes with exercise training-induced adaptations. The main findings of these studies are that, in certain situations, loading the cell with high doses of antioxidants leads to a blunting of the positive effects of exercise training and interferes with important ROS-mediated physiological processes, such as vasodilation and insulin signalling. More research is needed to produce evidence-based guidelines regarding the use of antioxidant supplementation during exercise training. We recommend that an adequate intake of vitamins and minerals through a varied and balanced diet remains the best approach to maintain the optimal antioxidant status in exercising individuals.
1. Introduction Antioxidant supplementation is a common practice amongst both professional athletes and amateur sportspersons, and the market offering various nutrient supplements is immense.[1] Although these products have been touted as a means of preventing exercise-induced oxidative damage and enhancing performance, consistent evidence of their efficacy is lacking. Moreover, it is clear that reactive oxygen species (ROS) produced during exercise play important roles in various cellular processes and, therefore, suppressing their formation with high doses of antioxidants might have a deleterious impact on cell function. The studies included in the review were identified by a systematic search using the PubMed database. Search terms were ‘reactive oxygen species’, ‘oxidative stress’, ‘antioxidant’, ‘exercise’, ‘skeletal muscle’, ‘muscle damage’ and ‘performance’. Further searching was performed by using the ‘related citations’ function of PubMed and scanning of the reference lists. We located over 150 studies investigating the effects of antioxidant supplementation on exercise-induced oxidative stress, muscle damage, recovery and performance. A number of excellent reviews are already available that contain a greater discussion of these studies.[2-11] In addiª 2011 Adis Data Information BV. All rights reserved.
tion, more detail on the effects of antioxidant therapy in human disease was beyond the scope of this review and can be found elsewhere.[12-17] The aim of this review is to discuss the studies that have shown negative effects of antioxidant supplements in exercising individuals, thus demonstrating the importance of ROS in skeletal muscle function. 2. Basic Mechanisms of Oxidative Damage 2.1 Redox Reactions
Reactions of oxidation and reduction, known as redox reactions, refer to all chemical reactions in which an atom in a compound has its oxidation number changed. The oxidation number is the effective charge that the central atom in a compound would have if all the ligands, including shared electron pairs, were removed. Oxidation can be explained as the loss of electrons, or more accurately, an increase of the oxidation number. Reduction is the gain of electrons or a decrease of the oxidation number. An oxidant is a compound that can accept electrons and is therefore reduced causing another substance to be oxidized. A reductant, on the other hand, donates electrons and is oxidized causing another substance to be reduced. Oxidation and reduction, which represent Sports Med 2011; 41 (12)
Antioxidant Supplementation in Exercise Training
the basis for numerous biochemical pathways, always accompany one another in order to transfer electrons between species. In a biological environment, oxidants and reductants are often called pro-oxidants and antioxidants, respectively. A cell’s redox state describes the pro-oxidant/ antioxidant balance and plays an important role in signalling and metabolic processes.[18,19] While oxygen is obviously vital for the life of aerobic organisms, the by-products of its metabolism can be harmful to cells. During normal metabolism, oxygen is utilized in the mitochondria for energy production. In the process of oxidative phosphorylation the majority of oxygen consumed is bound to hydrogen to form water. A small percentage of oxygen is not completely reduced, which leads to the production of oxygen intermediates known as ROS.[8] When reactants are derived from nitrogen, they are called reactive nitrogen species. Reactive species can be classified into two categories: free radicals and nonradical derivatives. A radical is any chemical compound capable of independent existence possessing one or more unpaired electrons in the outer-atomic or molecular orbital. These species have an enhanced affinity to donate or obtain another electron to become more stable, which leads to the formation of new free radicals, setting up a chain reaction. The free radical group includes compounds such as the superoxide anion radical (O2-), nitric oxide radical (NO), nitric dioxide radical (NO2), hydroxyl radical (OH), alkoxyl (RO) and peroxyl (RO2) radicals. Most typical nonradical reactive species relevant to biological systems are singlet oxygen (1O2), ozone (O3), hydrogen peroxide (H2O2), peroxynitrite (ONO2-), hypochlorous acid (HOCl), organic peroxides and aldehydes. Reactive species readily react with various organic substrates and play important roles in biological environments.[20] Cells and extracellular spaces are exposed to a large variety of reactive species from both exogenous and endogenous sources. The exogenous sources include exposure to oxygen, radiation, air pollutants, xenobiotics, drugs, alcohol, heavy metals, bacteria, viruses, sunlight, food and exercise. Nonetheless, exposure to endogenous sources is much more important and extensive, ª 2011 Adis Data Information BV. All rights reserved.
1045
because it is a continuous process during the life span. Reactive species are generated by all aerobic cells as part of normal metabolism. Mitochondria have been known as the dominant source of ROS production.[18] However, it has been suggested that the actual fraction of oxygen transformed into ROS accounts for only. around 0.15% of total oxygen consumption (VO2),[21] which is considerably less than original estimate of 2–5%.[22,23] Enzymes, such as nicotinamide adenine dinucleotide phosphate oxidase (NADPH oxidase), nitric oxide synthase (NOS) and xanthine oxidase (XO), are now recognized as the main endogenous source of reactive species.[24] Furthermore, transition metals have been shown to catalyze ROS formation[25] and in order to combat bacteria and other invaders white blood cells also produce a significant amount of reactive species.[26] The most vulnerable targets of reactive species are proteins, lipids and DNA.[27] ROS can oxidize proteins and alter their structure, impair their function and affect genetic transcription.[28,29] Fragmentation or loss of certain amino acids and aggregation make proteins more susceptible to proteolytic degradation.[30] Reactive species have the ability to oxidize polyunsaturated free fatty acids and initiate lipoprotein oxidation.[31] Disruption of the lipid bilayer changes fluidity and permeability of the cell membrane and may lead to inactivity of membrane bound proteins. Free radicals cause DNA strand breaks, loss of purines and damage to deoxyribose sugar.[32] They can impair the DNA repair system and provoke mutagenesis. Oxidative damage promotes inflammation[33] and apoptosis[34] and may eventually lead to decreased cellular and physiological functioning. 2.2 The Antioxidant Defence
To counter reactive species, we are equipped with highly effective antioxidant defence systems. These include nonenzymatic, enzymatic and dietary antioxidants. Glutathione, uric acid, lipoic acid, bilirubin and coenzyme Q10 are examples of nonenzymatic antioxidants that originate from endogenous sources and are often by-products of Sports Med 2011; 41 (12)
1046
cellular metabolism. Principal enzymatic antioxidants are superoxide dismuatse (SOD), catalase, glutathione peroxidase (GPX) and glutathione reductase, while most known examples of dietary antioxidants are tocopherols (vitamin E), ascorbic acid (vitamin C) and carotenoids (b-carotene). In addition, various polyphenolic compounds have recently been promoted as nutrient antioxidants. a-Lipoic acid and pharmaceuticals N-acetylcysteine and allopurinol have also been used in supplementation studies. Vitamin E refers to a group of fat-soluble compounds that include tocopherols and tocotrienols. a-Tocopherol is the most biologically active form, and has been shown to protect the cells from lipid peroxidation[35,36] and play a role in prevention of chronic diseases associated with oxidative stress.[37,38] The oxidized form can be recycled back to the active form by other antioxidants, such as vitamin C, retinol, ubiquinol, glutathione, cysteine and a-lipoic acid.[39] It has been suggested that vitamin E has other functions apart from its antioxidative one. For instance, g-tocopherol acts as a nucleophile and is able to trap electrophilic mutagens in lipophilic compartments.[40] Vitamin C or L-ascorbic acid is an antioxidant and a co-factor in a range of essential metabolic reactions in humans (e.g. collagen synthesis).[41] This water-soluble vitamin is produced endogenously by almost all organisms, excluding humans, several other mammalian groups and some species of birds and fish. L-ascorbate, an ion form of ascorbic acid, is a strong reducing agent and its oxidized form is reduced back by enzymes and glutathione. b-Carotene belongs to a group of red, orange and yellow pigments called carotenoids.[42] Others include a-carotene, b-cryptoxanthin, lycopene, lutein and zeaxanthin. These fat-soluble substances are found in plants and play a part in photosynthesis. b-Carotene is the most active carotenoid; after consumption it converts to retinol, a readily usable form of vitamin A. In addition to its provitamin A function, b-carotene is believed to have antioxidant properties,[43] and may positively impact the immune system[44] and exhibit anticancerogenic effects.[37] ª 2011 Adis Data Information BV. All rights reserved.
Peternelj & Coombes
Coenzyme Q10, also known as ubiquinone, is a fat-soluble, vitamin-like substance, present in most eukaryotic cells, primarily in mitochondria.[45] It is a component of the electron transport chain and plays a part in the energy production of a cell. Its reduced form, ubiquinol, acts as an important antioxidant in the body. Coenzyme Q10 is synthesized endogenously, and its dietary uptake is limited. Polyphenols are a group of water-soluble, plant-derived substances, characterized by the presence of more than one phenolic group.[46] Several thousand polyphenols have been identified and they are divided into different groups according to their structure and complexity (flavonoids, lignans, stilbenes, coumarins and tannins). Flavonoids are the largest group of phenolic compounds and include anthocyanins, flavones, isoflavones, flavonols, flavanones and flavanols. Fruits and vegetables are a particularly rich source of polyphenols. For instance, red wine contains various polyphenolic compounds, such as stilbene resveratrol and flavonol quercetin, which have been well studied and have been shown to possess pharmacological properties in the treatment of chronic diseases.[47,48] The antioxidant potential of polyphenols has been well established and is exhibited through their chain-breaking and single-electron transfer abilities. However, there is compelling evidence that the protective actions of polyphenols are not simply because of their redox properties, but rather as a result of their ability to modulate cellular signalling cascades by binding to specific target proteins.[46] a-Lipoic acid is an organosulfur compound derived from octanoic acid. It is an essential co-factor of the four mitochondrial enzyme complexes, therefore, is crucially involved in aerobic metabolism. a-Lipoic acid may have potent antioxidant potential and can recycle vitamin E;[49] however, its accumulation in tissues is limited. Micronutrient functions of a-lipoic acid may act more as an effector of cellular stress response pathways.[50] N-acetylcysteine is a by-product of an endogenously synthesized antioxidant glutathione. It is a cysteine derivative and plays a role in glutathione maintenance and metabolism. N-acetylcysteine has been proposed to have Sports Med 2011; 41 (12)
Antioxidant Supplementation in Exercise Training
antioxidant effects and is used as a pharmaceutical drug (mucolytic agent) and a nutritional supplement.[51] Allopurinol, a structural isomer of hypoxanthine, is an inhibitor of XO. It is a drug primarily used to treat hyperuricaemia, as it decreases uric acid formation and purine synthesis.[52] Antioxidants are often divided into two groups: those that act either through stabilizing ROS or by removing reactive intermediates. The former, also known as preventative antioxidants, stabilize free radicals by donating electrons and become oxidized themselves, forming less active radicals. The latter, ‘scavengers’, help slow or stop the damaging chain reaction by removing free radical intermediates. In addition, transition metal sequestration and oxidative damagerepairing mechanisms support the body’s defence system. Endogenous antioxidant systems respond rapidly to an increased production of reactive species. Cells can modulate gene expression and the activity of antioxidant enzymes to cope with oxidative stress.[18,53] 2.3 Oxidative Stress
Despite the extensive defence system, an increase in ROS production or diminished antioxidants can lead to progressive cell damage and a decline in physiological function. When oxidant capacity exceeds the antioxidant capacity, homeostatic balance is disturbed and the redox state becomes more pro-oxidizing. This imbalance is called oxidative stress.[54] As we now know that individual signalling and control events occur through discrete redox pathways, rather than through global balances, the classic definition of oxidative stress has been refined and also considers oxidative stress as a disruption of redox signalling and control.[55] Therefore, oxidative stress may occur without an overall imbalance of pro-oxidants and antioxidants and can cause organ-specific and pathway-specific toxicity. Under usual lifestyle conditions we are exposed to high levels of reactive species from exogenous sources (e.g. environmental pollution)[56] and oxidative stress has been implicated in a growing list of human diseases, such as cardioª 2011 Adis Data Information BV. All rights reserved.
1047
vascular, inflammatory, metabolic and neurodegenerative diseases, as well as cancer and the ageing process.[57] A diet rich in antioxidants has been identified as a potentially noninvasive means of controlling oxidative stress.[58,59] Antioxidant supplementation has received much attention because of its capacity to support the endogenous defence by scavenging additional ROS and, therefore, by reducing oxidative damage.[60-62] However, there is little evidence for the efficacy of antioxidant supplements to treat ROS-associated diseases. This has led to considerable debate regarding the beneficial health effects of this kind of supplementation in different types of patients and with different types of antioxidants.[13,63,64] Although observational epidemiological cohort studies with large numbers of subjects and diverse populations have been largely supportive of the health-promoting effects of antioxidants,[65-68] interventional trials have been controversial, with some positive findings,[37,38,69] many null findings[70-73] and some suggesting a detrimental effect of antioxidant supplementation, particularly vitamin E, on morbidity and mortality.[74-76] 2.4 Beneficial Roles of Reactive Species
Although reactive species are associated with harmful biological events, they are essential in cellular development and optimal function.[77,78] Cells have evolved strategies to utilize reactive species as biological stimuli. They act as subcellular messengers in important molecular signalling processes and modulate enzyme and gene activation.[77] Most antioxidant enzyme genes contain regulatory sequences in their promoter and intron regions that can interact with redox sensitive transcription factors.[79] Reactive species play significant roles in cellular growth and proliferation.[77] It has been shown recently that physiological levels of ROS are required to activate DNA repair pathways for maintaining genomic stability in stem cells.[80] Furthermore, ROS are involved in the biosynthesis of other molecules,[81] the immune response of cells[26] and drug detoxification.[77] They are a requisite for vasodilation,[82] optimal muscular contraction[83] and initiation of apoptosis.[34] Sports Med 2011; 41 (12)
1048
3. Exercise-Induced Oxidative Stress 3.1 Reactive Species in Skeletal Muscle
During contraction, skeletal muscle is a major source of ROS, as well as . one of the main targets.[24] Exercise increases VO2 by up to 20 times above resting values.[84] In the mitochondria of exercising muscle cells, this translates to a 200-fold greater oxygen usage.[84] Exerciseinduced oxidative stress was first described in the late 1970s when increased levels of lipid peroxidation products were found in the expired air of exercising humans[35] and the tissues of exercised rats.[85] In 1982, Davies et al.[86] provided the first direct evidence that high-intensity exercise significantly increased ROS production in the muscles and liver of rats, and caused damage to mitochondrial membranes. It was suggested that this could, at the same time, deliver a stimulus to mitochondrial biogenesis. However, the majority of following studies focused on the damaging effects of oxidants in muscle and looked for the potential benefits of antioxidants. Over the last 30 years, an understanding of the sources and consequences of exercise-produced ROS has advanced markedly. It is now clear that reactive species play important roles in skeletal muscle function and metabolism. Redox signalling in contracting muscle is considered one of the basic elements in exercise biology.[24] 3.2 Adaptation to Exercise-Induced Oxidative Stress
Cells adapt to increased ROS production to become more resistant to the adverse effects of oxidative stress.[87] It has to be emphasized, however, that the effects of a single bout of exercise and regular exercise are quite different. Regular physical activity brings about numerous beneficial effects and the body adapts to elevated oxidant levels, whilst with acute exercise, the adaptation is only marginal. Acute adjustment involves increased vasodilation to enhance blood flow and fuel transport and a kinetic shift via the allosteric activity of enzymes, which may not be sufficient to restore oxidant-antioxidant homeostasis.[88] Long-term stimulation of endogenous ª 2011 Adis Data Information BV. All rights reserved.
Peternelj & Coombes
defence mechanisms requires the continuous presence of physiological stimuli that maintain a certain degree of pro-oxidative milieu, and effectively overload the antioxidant systems.[89] With exercise training the body adapts to exercise-induced oxidative stress and becomes more resistant to subsequent oxidative challenges. This is achieved through a number of different mechanisms, such as upregulation of redox-sensitive gene expression and antioxidant enzymes levels,[90,91] an increase in enzyme activity,[92,93] stimulation of protein turnover,[94] improvement in DNA-repair systems,[95,96] and increased mitochondrial biogenesis[97] and muscle content of heat shock proteins (HSPs).[98,99] In addition, adaptation positively affects remodelling of skeletal muscle after injury and attenuates inflammation and apoptosis.[88,100,101] Moderate levels of reactive species appear necessary for various physiological processes, whereas, an excessive ROS production causes oxidative damage. This may be described by the concept of hormesis, a dose-response relationship in which a low dose of a substance is stimulatory or beneficial and a high dose is inhibitory or toxic.[102] The adaptive response of mitochondria to increased formation of ROS is termed mitochondrial hormesis or mitohormesis.[103] The hormetic action of reactive species could represent a mechanism underlying the health and performance benefits of regular physical activity.[102] This can be seen in the role of reactive species as endogenous regulators of skeletal muscle function. Indeed, they appear obligatory for optimal contractile activity. Muscle myofilaments, such as myosin and troponin, and proteins in the sarcoplasmic reticulum are redox-sensitive, which gives ROS the ability to alter muscle contraction.[104] Based on Reid’s model for the role of redox state on muscle force production, reaction to ROS can be described by a bell-shaped curve.[104,105] At baseline, low oxidant levels appear to be suboptimal for the contraction of unfatigued muscle. The data from Reid’s studies suggest modest augmentation in ROS levels causes muscle force to increase, while antioxidants deplete oxidant levels and depress force. At higher ROS concentrations this is reversed and Sports Med 2011; 41 (12)
Antioxidant Supplementation in Exercise Training
force production decreases in a time- and dosedependent manner.[105-107] 3.3 Oxidative Stress and Muscle Damage
Despite skeletal muscle being relatively resistant to exercise-induced oxidative damage, it is clear that intense and/or prolonged muscular activity can result in harmful outcomes.[9] Repetitive eccentric contractions, if unaccustomed in particular, place skeletal muscle under considerable stress that may cause muscle damage.[108,109] Damaging exercise also induces an inflammatory response, which further increases ROS formation.[110] However, the studies often lack the information about the subjects’ redox status and therefore fail to provide evidence for the causal role of ROS in muscle damage. The majority of studies have measured indirect and nonspecific indices of muscle damage, such as muscle soreness and reduction in the muscle force production. Eccentric exercise was shown to cause structural changes of muscle fibres,[108,109,111,112] and has been associated with muscular soreness,[110,113,114] reduced range of motion[110] and loss of torque and force production.[109,111,112,115,116] This may result in muscle fatigue and development of muscular atrophy.[117-119] Extreme fatigue can lead to muscle injury and, possibly, irreversible cell alterations.[119,120] 4. Antioxidant Supplementation and Exercise 4.1 Overview
It is common practice for athletes to use antioxidant supplements with the notion that they prevent the deleterious effects of exercise-induced oxidative stress, hasten recovery of muscle function and improve performance.[1,121-125] Indeed, there is now an enormous range of vitamins, minerals and extracts marketed as antioxidant supplements. None have undergone adequate testing, and therefore lack scientific evidence regarding efficacy and long-term safety. The popularity of antioxidant supplements with athletes has led to a plethora of small research studies in this area. As expected, the ª 2011 Adis Data Information BV. All rights reserved.
1049
studies varied considerably in terms of research design, exercise protocol, population groups, supplementation regimen and analysis methods. Importantly, the studies are also of generally low quality. As commonly found in sports nutrition research, the vast majority do not adhere to all the accepted features of a high-quality trial (e.g. placebo-controlled, double-blind, randomized design with an intent-to-treat analysis). Indeed, most studies fail to provide sufficient detail regarding inclusion and exclusion criteria, justification of sample size, adverse events, data gathering and reporting, randomization, allocation and concealment methods, and an assessment of blinding success. The poor quality of the majority of studies in this field increases the possibility for bias and needs to be always considered when evaluating the findings. Supplements used in the studies include vitamin E, vitamin C, b-carotene, coenzyme Q10, alipoic acid, N-acetylcysteine, allopurinol, quercetin, resveratrol and several other polyphenolic compounds. A number of studies have used combinations of these. The range of dosages across the supplements was wide and duration of supplemention varied from acute (1–2 days) to chronic administration (from 1 week to up to 6 months). Blood, urine, breath and muscle tissue samples were collected pre-, during and postsupplementation and exercise. The most common outcome measure was a marker of oxidative stress with lipid peroxidation products predominating, followed by oxidized proteins, DNA damage markers and alterations in endogenous antioxidant systems. Direct measurement of reactive species concentration (e.g. electron spin resonance spectroscopy) was only performed in a small number of studies because of the instability of ROS, high costs and extensive workup requirements. 4.2 Antioxidant Supplementation and Exercise-Induced Oxidative Stress
The majority of studies have used measures of oxidative stress as their main outcome, and most have demonstrated that antioxidants attenuate exercise-induced increases in oxidative stress. Sports Med 2011; 41 (12)
1050
Most common antioxidants in these positive studies were vitamin E[35,36,62,126-132] and vitamin C,[60,116,133-137] followed by different combinations of antioxidants[61,138-147] and, most recently, polyphenolic compounds.[148-156] Furthermore, lower levels of oxidative stress markers have been reported after b-carotene,[157] a-lipoic acid,[158] N-acetylcysteine[159] and selenium[160] administration. However, there have been many studies showing no significant effect of antioxidant supplements on exercise-induced oxidative stress[110,161-172] and several indicating increased oxidative stress levels following antioxidant administration.[144,173-176] Although the majority of studies report that antioxidants can reduce oxidative stress levels, the physiological implications of these effects are unknown. In an attempt to determine the importance of reducing oxidative stress, investigators have studied the role of antioxidant supplementation in exercise performance and muscle damage. 4.3 Antioxidant Supplementation and Muscle Damage
Strong evidence to support the role of antioxidant supplementation in protecting against muscle damage is lacking. The majority of investigations have focused on the effects of vitamin C and E and looked at oxidative stress markers and plasma concentrations of intramuscular enzymes, e.g. creatine kinase (CK) and lactate dehydrogenase, rather than indices of muscle damage such as force loss, muscle soreness, structural changes of myoproteins and their plasma concentration.[6] As a result of the lack of direct measurement of specific indices of muscle damage, it is unclear to what extent muscle damage was induced in those studies. There are reports that antioxidant supplementation could offer some protection from exercise-induced cell damage,[127,177-181] attenuate the inflammatory response to exercise,[147,151,182-186] and reduce muscle force loss[154,156,177,187] and fatigue.[188-191] Other investigations, however, found no significant effect of antioxidants on indices of cell damage,[111,113,161,192-194] muscle soreness[114,195-199] ª 2011 Adis Data Information BV. All rights reserved.
Peternelj & Coombes
and inflammation.[111,114,127,169,194,200,201] A number of studies suggested that antioxidant supplementation may promote muscle damage and possibly hinder recovery.[165,175,197,202] These studies are the focus of this review and discussed in section 5. 4.4 Antioxidant Supplements as Ergogenic Aids
There has been a general inconsistency of outcomes when investigating the role of antioxidant supplementation in exercise performance with the majority of the studies reporting no benefits. In the early 1970s, Sharman and colleagues[203] showed that supplementation with vitamin E had no beneficial effect on endurance performance of adolescent male swimmers. Moreover, the placebo group demonstrated greater improvements of cardiorespiratory function with exercise training compared with the antioxidant group, which may be the first report of the unfavourable effect of supplementation. In the studies that followed, vitamin E proved ineffective in improving performance in swimmers,[204] professional cyclists,[132,205,206] nonresistance-trained men,[202] athletic students[167] and marathon runners.[207] Furthermore, vitamin E supplements had no additive effect beyond that of aerobic training on indices of physical performance in a group of older sedentary adults.[208] Supplementation with coenzyme Q10 did not exhibit any significant effects on exercise performance of men,[162,209,210] regardless of their age and training status. Quercetin supplements also failed to show any ergogenic effects in sedentary individuals[199,211] or cyclists.[212] Polyphenol resveratrol did not improve muscle force output and muscle fatigability in mice subjected to electrically stimulated isometric contractions.[213] In a study by Marshall et al.,[214] vitamin C was shown to slow racing greyhounds. Despite the presumption that antioxidants work synergistically and may therefore be more efficient in combating oxidative stress, combinations of vitamins E, C, coenzyme Q10 and other vitamins and minerals failed to improve the exercise performance of competitive male Sports Med 2011; 41 (12)
Antioxidant Supplementation in Exercise Training
runners,[215] cyclists,[144,216] triathletes,[217,218] soccer players,[146,219] resistance-trained men,[220] ultraendurance runners,[221] moderately trained men,[222] and trained and untrained males and females.[166] Nonetheless, there have been a number of studies showing positive, albeit, modest effects of antioxidant supplementation on physical perforassociated with immance. Coenzyme. Q10 was . proved maximal VO2 (VO2max)and aerobic and anaerobic threshold of professional crosscountry skiers that resulted in an increased exercise capacity and a faster recovery rate.[223] Similarly, supplementation with coenzyme Q10 indicated beneficial effects on performance, fatigue sensation and recovery during fatigueinducing workload trials in a group of healthy volunteers.[189] Furthermore, results from supplementation studies that involved male cyclists,[224] trained and untrained individuals[225] and sedentary men[226] supported the performance-enhancing effect of coenzyme Q10. Vitamin E supplementation was proposed to have a beneficial effect on the performance of climbers at high altitude[128] and endurance performance of mice,[227] rats[228] and sled dogs.[229] In two early studies, supplementation with vitamin C was associated with an improved exercise capacity of untrained male students[230] and athletes.[231] In a study by Aguilo et al.,[232] male athletes supplemented with a combination of vitamin E, C and b-carotene exhibited lower blood lactate levels after a maximal exercise test and . exhibited a more significant increase more in VO2max after 3 months of exercise training than the placebo group. Supplementation with different combinations of antioxidants also positively affected the exercise performance of students,[233] elderly endurance-trained athletes[234] and aged rats.[139] Medved and colleagues[235] have studied the effect of N-acetylcysteine on muscle fatigue and performance in untrained and trained men. Although N-acetylcysteine was shown to modulate blood redox status during high-intensity intermittent exercise, it did not affect time to fatigue in a group of untrained men. Similarly, N-acetylcysteine infusion during prolonged submaximal exercise had no effect on time to fatigue ª 2011 Adis Data Information BV. All rights reserved.
1051
in a group of team-sport athletes and endurancetrained cyclists. Nonetheless, the antioxidant improved regulation of plasma K+ concentration and it was suggested the ergogenic effect of N-acetylcysteine depends on an individual’s training status.[236] Finally, N-acetylcysteine infusion during prolonged submaximal exercise was reported to augment time to fatigue in a group of well trained individuals, possibly by increasing muscle cysteine and glutathione availability.[237] Recently, there have been a number of investigations showing the performance enhancing effects of polyphenols, including quercetin,[201,238-240] resveratrol,[241] and polyphenolic compounds from grape extract,[152] beetroot juice,[242-245] Rhodiola rosea plant[246] and Ecklonia cava algae.[247] Emerging evidence suggests that the antioxidant potential of phenolic compounds is unlikely to be the sole mechanism responsible for their protective action, which could also be mediated by their interaction with various key proteins in the cell-signalling cascades.[248] As mentioned above in section 4.1, many of the studies evaluating the effects of antioxidants on exercise performance have been of low quality with small subject numbers. In addition, most have had important methodological details left out of the articles (e.g. recruitment, randomization, allocation and concealment methods) leading to the assumption that they were not considered. This creates a potentially dangerous bias in regards to subject selection and the assessment of performance effects. 5. Antioxidant Supplementation Interferes with the Beneficial Effects of Exercise Training Recent studies have indicated that antioxidant supplements have a detrimental effect on the health and performance benefits of exercise training. Considering the multifunctional beneficial roles of ROS in living organisms discussed above in section 2.4, reports of unfavourable effects of antioxidant supplementation should not come as a surprise. The studies reporting negative outcomes are discussed in sections 5.1–5.3 with more details presented in table I. Sports Med 2011; 41 (12)
1052
ª 2011 Adis Data Information BV. All rights reserved.
Table I. Studies with negative outcomes using antioxidant supplementation during exercise training Subjects
Supplements (daily dose)
Duration
Study design
Findings
Malm et al.[249] (1996)
15 M
Coenzyme Q10 (120 mg)
20 d
Placebo-controlled trial: Exercise tests: anaerobic test (Wingate test, 5 min . recovery, 10 · 10 sec all-out cycling), VO2 submax and max test. Exercise training: 9 sessions (15 · 10 sec all-out cycling sprints). Samples: plasma CK activity
After exercise, CK levels › only in the supplemented group. Subjects taking supplements showed smaller traininginduced improvements in physical performance than the placebo group
Malm et al.[250] (1997)
18 M
Coenzyme Q10 (120 mg)
22 d
Placebo-controlled double-blind trial: Exercise tests: anaerobic test (30 sec all-out cycling, 5 min recovery, 10 · 10 sec all-out cycling), submax and peak cycling . . VO2 test, VO2max running test. Exercise training: 7 sessions (15 · 10 sec all-out cycling sprints). Samples: plasma lactate
There was a greater increase in anaerobic performance in the placebo group compared with the supplemented group. Moreover, supplementation was associated with reduced exercise traininginduced increase in power output and recovery rate between cycling sprints. Coenzyme Q10 had no effect on submax . . and peak cycling VO2, running VO2max and lactate levels
Childs et al.[175] (2001)
14 M
Vitamin C (12.5 mg/kg BW) + NAC (10 mg/kg BW)
1 wk (postexercise)
Double-blind placebo-controlled trial: Exercise protocol: eccentric arm exercise (3 · 10 repetitions, 80% of 1RM). Samples: serum free iron levels, plasma lipid hydroperoxides, F2-isoprostanes, myeloperoxidase and IL-6, plasma CK and LDH activities, serum SOD and GPX
Exercise › inflammatory indicators, free iron concentration and the levels of oxidative stress and muscle damage markers. The amount of iron, levels of lipid hydroperoxides and isoprostanes and LDH and CK activities were higher in the supplemented group than in the placebo group
Coombes et al.[251] (2001)
28 F rats
Vitamin E (10 000 IU/kg diet) + a-lipoic acid (1.65 g/kg diet)
8d
In situ experiment: Contractile measurements (tibialis anterior): Po, Pt and force-frequency curve, 60 min fatigue protocol. Samples: muscle MDA and lipid hydroperoxide
32 F rats
Vitamin E: 100, 200, 400 mM/DHLA; 100 mM/vitamin E; 400 mM + DHLA; 100 mM
Contracted muscles of supplemented animals had lower levels of oxidative stress than the muscles from the control group. Vitamin E and a-lipoic acid supplementation had no effect on muscle fatigue but was associated with decreased muscle force production at low stimulation frequencies (in situ). In vitro experiments indicated that vitamin E depressed force production at low stimulation frequencies
5 F racing greyhounds
Vitamin C (1 g)
Marshall et al.[214] (2002)
In vitro experiment: contractile measurements (costal diaphragm): Po, Pt and forcefrequency curve, 30 min fatigue protocol
4 wk (each treatment)
Crossover controlled trial: Treatments: no supplementation; supplementation after racing; supplementation 1 h before racing. Exercise training: 2 · 500 m races/wk. Samples: plasma TBARS and antioxidant capacity
Vitamin C showed no effect on oxidative stress and antioxidant capacity. The dogs ran slower when supplemented
Continued next page
Peternelj & Coombes
Sports Med 2011; 41 (12)
Study (y)
Subjects
Supplements (daily dose)
Duration
Study design
Findings
Avery et al.[202] (2003)
18 untrained M
Vitamin E (1200 IU)
3 wk
Randomized placebo-controlled doubleblind trial: Exercise Protocol: 3 resistance exercise sessions separated by 3 days of recovery. Measurements: muscle soreness, muscle strength and power assessment. Samples: plasma MDA and CK activity
There was no effect of supplementation on muscle soreness, performance indices and MDA levels. CK levels were greater in the supplemented group than in the placebo group
Bryant et al.[144] (2003)
7 M cyclists
Vitamin C (1 g)/vitamin C (1 g) + vitamin E (200 IU/kg)/vitamin E (400 IU/kg)
3 wk (each treatment)
Controlled crossover single-blind trial: Treatments: placebo; vitamin C; vitamin C + vitamin E; vitamin E. Exercise tests: 60 min steady state ride . (70%VO2max) and 30 min performance ride . (70%VO2max). Samples: plasma MDA and lactic acid
Supplementation had no effect on exercise performance. Vitamin E fl MDA levels, the combination of vitamins E and C had no effect, vitamin C alone › MDA levels
Khassaf et al.[98] (2003)
16 untrained M
Vitamin C (500 mg)
8 wk
Randomized controlled trial: Muscle samples (exercise protocol: 45 min . single leg cycling, 70% VO2max, vastus lateralis): HSP60 and HSP70 content. Lymphocytes (treated with H2O2 for 30 min): SOD and CAT activity, HSP60 and HSP70 content
Supplementation with vitamin C was associated with attenuated exerciseinduced increase in HSP content and SOD and CAT activity
Nieman et al.[176] (2004)
36 triathletes (26 M, 10 F)
Vitamin E (800 IU)
2 mo
Randomized placebo-controlled double-blind trial: Ironman Triathlon race – samples: plasma and urinary F2-isoprostanes, urinary 8-OHdG and 8-oxoG, plasma lipid hydroperoxides and cytokines
Post-race concentrations of isoprostanes, lipid hydroperoxides, IL-6, IL-1ra and IL-8 increased more in the vitamin E group than in the placebo group. Supplementation had no effect on race time
Gomez-Cabrera et al.[252] (2005)
20 M rats
Allopurinol (32 mg/kg)
Admin prior to exercise
Randomized controlled trial: Exercise protocol: progressive intensity treadmill test, exercise to exhaustion. Samples: plasma lactate and XO activity, muscle GSH, GSSG, carbonylated proteins, p38, ERK1 and ERK2, NF-kb DNA-binding activity and Mn-SOD, iNOS and eNOS
Allopurinol treated rats exhibited fl oxidative stress levels and fl exercisemediated increase in XO activity and induction of MAPKs. This was associated with fl DNA binding of NF-kB and blunted upregulation of Mn-SOD, eNOS and iNOS gene expression
Gomez-Cabrera et al.[253] (2006)
25 marathon runners
Allopurinol (300 mg)
2 h prior to marathon race
Randomized placebo-controlled trial: Marathon race - samples: lymphocyte NF-kb p50 activation, plasma MDA and XO activity
Allopurinol prevented XO activation and lipid peroxidation. Inhibiton of XO-derived ROS formation prevented NF-kB activation. Allopurinol had no effect on race time Continued next page
1053
Sports Med 2011; 41 (12)
Study (y)
Antioxidant Supplementation in Exercise Training
ª 2011 Adis Data Information BV. All rights reserved.
Table I. Contd
1054
ª 2011 Adis Data Information BV. All rights reserved.
Table I. Contd Study (y)
Subjects
Supplements (daily dose)
Duration
Study design
Findings
Close et al.[197] (2006)
20 M
Vitamin C (1 g)
2 h prior to and for 2 wk postexercise
Randomized placebo-controlled double-blind trial: Exercise protocol: . downhill running test (30 min, 60% VO2max). Measurements: pain assessment (visual analogue scale, pressure algometry) and muscle function (quadriceps torque assessment). Samples: serum MDA
Supplementation with vitamin C fl exercise-induced increase in MDA levels but had no effect on DOMS. Delayed recovery of muscle function was noted in the supplemented group
Fischer et al.[99] (2006)
21 M
a-Tocopherol (400 IU) + vitamin C (500 mg)
4 wk
Randomized placebo-controlled singleblind trial: Exercise protocol: 3 h, 2-legged dynamic knee extensor exercise. Samples: muscle HSP72 mRNA and protein, plasma HSP72 and F2-isoprostanes
a-Tocopherol + vitamin C treatment attenuated › in lipid peroxidation postexercise. Exercise-induced increase in HSP72 levels in skeletal muscle and circulation was abolished in a-tocopherol + g-tocopherol + vitamin C group
Dose-response relationship between adaptations of antioxidant enzymes and responses to ultraendurance exercise. Ultraendurance training upregulated endogenous antioxidant system (GPX and CAT activity). Triathletes taking supplements had elevated post-race MDA levels compared with nonsupplementers
a-Tocopherol (290 IU) + g-tocopherol (130 IU) + vitamin C (500 mg) Knez et al.[93] (2007)
Richardson et al.[254] (2007)
Vitamin C (1095 – 447 mg) + vitamin E (314 – 128 mg)
Vitamin C: 4.9 – 4.7 y; vitamin E: 5.6 – 5.2 y
Observational study: subjects recruited 4 wk before the race, controls active <3h/wk: Triathletes: training and competing for 4.7 – 2.4 y, 14.5 – 3.4 h/wk, 10 taking supplements; race: 1.9 km swim, 90.1 km cycle, 21.1 km run. Samples: plasma MDA and erythrocyte SOD, GPX and CAT activities
29 Ironman triathletes (23 M, 6 F)
Vitamin C (558 – 350 mg) + vitamin E (702 – 756 mg)
Vitamin C: 0.8 – 0.6 y; vitamin E: 1.6 – 0.8 y
Triathletes: training and competing for 6.9 – 6.4 y, 17.19 – 3.4 h/wk, 8 taking supplements; race: 3.8 km swim, 180 km cycle, 42.2 km run. Samples: plasma MDA and erythrocyte SOD, GPX and CAT activities
25 M
Dose: a-lipoic acid (300 mg) + vitamin C (500 mg) + vitamin E (200 IU)
2 h and 1.5 h prior to exercise
Randomized placebo-controlled crossover double-blind trial: Exercise protocol: forearm handgrip exercise at low-intensity workload (3, 6 and 9 kg at 0.5 Hz) for 3 min. Measurements: plasma FR, vasodilation.
8 wk
Randomized double-blind controlled trial: . Exercise test: VO2max test (bicycle ergometer). Exercise training: 40 min . cycling 3 d/wk (65%-80% VO2max)
Dose: a-lipoic acid (300 mg) + vitamin C (500 mg) + vitamin E (400 IU) Gomez-Cabrera et al.[97] (2008)
14 sedentary M
Vitamin C (1 g)
Antioxidant administration › total antioxidant capacity and fl exerciseinduced oxidative stress but fl brachial artery vasodilation during submaximal exercise.
Continued next page
Peternelj & Coombes
Sports Med 2011; 41 (12)
16 halfIronman triathletes (13 M, 3 F)
Study (y)
Supplements (daily dose)
Duration
Study design
Findings
Vitamin C: 0.24 mg/cm2 body surface area
3 wk; 6 wk
Untrained group, trained group, trained + supplemented group: RT-PCR experiment: 3 wk training. Western blotting and performance experiments: 6 wk training. Exercise . training: 5 d/wk, treadmill (75% VO2max, 25-85 min/d). Endurance test (run to . exhaustion), VO2max test (treadmill run). Samples: muscle mTFA and NRF-1 mRNA and protein, cyt c and PGC-1 protein, MnSOD and GPX mRNA
Moderate intensity exercise enhanced endogenous antioxidant defence ( › expression of Mn-SOD and GPX) and mitochondrial biogenesis (upregulation of PGC-1 - NRF-1 - mTFA - cyt c pathway) and increased endurance capacity. Vitamin C prevented these training induced adaptations
Copp et al.[255] (2009)
19 M rats
Vitamin C (76 mg/kg) + tempol (52 mg/kg)
Acute infusion (after first exercise protocol)
Exercise protocol (right spinotrapezius muscle): 1 Hz twitch contractions for 180 sec (2 sessions: pre- and postantioxidant administration);13 rats: blood flow and PO2 mv measurements; 6 rats: muscle force measurements
Antioxidant administration › serum antioxidant capacity but fl blood flow, baseline PO2 mv , muscle oxygen utilization and muscle force production
Lamprecht et al.[174] (2009)
8 trained M cyclists
Vitamin E (107 IU) + vitamin C (450 mg) + b-carotene (36 mg) + Se (100 mg)
2 wk
Randomized double-blind placebocontrolled crossover trial: Exercise test: cycle ergometer, 90 min . cycling (45% VO2max) + 30 min cycling (75% . VO2max). Samples: plasma MDA and GPX
MDA concentrations were › and GPX levels fl after antioxidant treatment (preand post-exercise)
Ristow et al.[91] (2009)
20 untrained M (<2 h of exercise/wk), 20 pretrained M (>6 h of exercise/wk)
Vitamin C (1 g) + vitamin E (400 IU)
4 wk
Controlled trial, 2 part-study – open-label study; double blind placebo-controlled study: 4 groups: untrained nonsupplemented, trained nonsupplemented, untrained supplemented, trained supplemented. Exercise training – 5 d/wk, session: 20 min biking/running, 45 min circuit training. Measurements: GIR. Samples: plasma adiponectin, muscle PGC-1a, PGC-1b, PPARg, SOD1 and SOD2, and GPX gene levels
Exercise training › insulin sensitivity, fl fasting plasma insulin levels, › gene expression of PGC-1a, PGC-1b, PPARg, SOD1 and SOD2, GPX (irrespective of training status). Supplementation with vitamins E and C was shown to prevent these health promoting effects
Teixeira et al.[165] (2009)
20 competitive kayakers (14 M, 6 F)
a-Tocopherol (272 mg) + vitamin C (400 mg) + b-carotene (30 mg) + lutein (2 mg) + Se (400 mg) + Zn (30 mg) + mg (600 mg)
4 wk
Randomized double-blind placebocontrolled trial: Exercise test: maximal flat-water kayaking trial (1000 m). Samples: plasma antioxidants, TBARS, IL-6 and CK, SOD, GR, GPX activities
Antioxidant supplementation › antioxidant capacity but had no effect on oxidative stress and inflammation markers. Supplemented athletes showed a blunted decrease in CK activity post-exercise Continued next page
1055
Sports Med 2011; 41 (12)
Subjects 36 M rats
Antioxidant Supplementation in Exercise Training
ª 2011 Adis Data Information BV. All rights reserved.
Table I. Contd
Study (y)
Subjects
Supplements (daily dose)
Duration
Study design
Findings
Wray et al.[256] (2009)
6 older, mildly hypertensive M
Dose: a-lipoic acid (300 mg), vitamin C (500 mg), vitamin E (200 IU)
Prior to and after 6 wk of training: 2 h before exercise protocol
Antioxidant administration reduced FR levels pre- and post-exercise. Exercise training reduced BP and improved vasodilation, supplementation after training negated these effects
Dose: a-lipoic acid (300 mg), vitamin C (500 mg), vitamin E (400 IU)
Prior to and after 6 wk of training: 30 min after 1
Double-blind placebo-controlled crossover trial: Exercise protocol – d 1, 2: antioxidant efficacy test; d 3–6: FMD procedure followed by knee extensor exercise, subjects crossed over, returned after 24 h. Exercise training: 3 · wk-single leg knee-extensor exercise. Measurements: plasma FR, BP and FMD
Bailey et al.[110] (2010)
38 M
Vitamin C (800 mg) + vitamin E (536 mg) + vitamin B6 (4 mg) + vitamin B9 (400 mg) + zinc sulphate monohydrate (10 mg) + vitamin B12 (2 mg)
6 wk (including. 2 d postexercise)
Randomized placebo-controlled doubleblind trial: Exercise test (d 40): 90 min intermittent high-intensity shuttle-running. Measurements: pre- and post-exercise ratings of perceived muscle soreness and assessment of muscle function (peak isometric torque of the knee flexors and extensors, range of motion at the knee joint). Samples: urine F2-isoprostanes, serum IL-6 and cortisol
Antioxidant supplementation was associated with attenuated exerciseinduced › in cortisol concentration but › post-exercise IL-6 and F2-isoprostane levels (compared with the placebo). Treatment had no effect on indices of muscle damage, muscle function measures and perception of muscle soreness
Matsumoto et al.[257] (2011)
48 M rats
a-Tocopherol (1000 IU/kg diet) + a-lipoic acid (1.6 g/kg diet)
14 wk
Controlled trial: 4 groups: untrained nonsupplemented, trained nonsupplemented, untrained supplemented, trained supplemented. Exercise training: 90 min treadmill run 4 d/wk. Samples: left ventricular and coronary artery endothelial cells (gene analysis)
IL-6 gene levels were fl by all treatments. RhoA gene expression was fl by exercise training, › by antioxidant supplementation. The combination of exercise and supplementation resulted in a blunted fl of RhoA gene levels (compared with the exercise training effect)
Peternelj & Coombes
Sports Med 2011; 41 (12)
1RM = repetition maximum; 8-OHdG = 8-hydroxy-2-deoxyguanosine; 8-oxoG = 7,8-dihydro-8-oxoguanosine; BP = blood pressure; BW = bodyweight; CAT = catalase; CK = creatine kinase; cyt c = cytochrome c; DOMS = delayed onset muscle soreness; DHLA = dihydrolipoic acid; ERK = extracellular signal-regulated protein kinases; F = female; FMD = flowmediated vasodilation; FR = free radical; GIR = glucose infusion rate; GPX = glutathione peroxidase; GR = glutathione reductase; GSH = reduced glutathione; GSSG = oxidized glutathione; H2O2 = hydrogen peroxide; HSP = heath shock protein; IL-1ra = interleukin 1 receptor antagonist; IL-6(8) = interleukin-6(8); LDH = lactate dehydrogense; M = male; MAPK = mitogen activated protein kinase; max = maximal; MDA = malondialdehyde; mRNA = messenger RNA; mTFA = mitochondrial transcription factor A; NAC = N-acetyl cysteine; NF-jB = nuclear factor kappa-light chain-enhancer of activated B cells; NOS = nitric oxide synthase; NRF-1 = nuclear respiratory factor 1; p38 = a member of MAPKs; p50 = a subunit of NF-kb complex; PGC-1 = peroxisome proliferator-activated receptor gamma coactivator 1; PPARc = peroxisome proliferator-activated receptor gamma; PO2 mv = microvascular O2 partial pressure; Po = max specific tension; Pt = twitch tension; RhoA = Ras homolog gene family member A; RT-PCR = real-time reverse transcriptase-polymerase . . chain reaction; Se = selenium; SOD = superoxide dismutase; submax = submaximal; TBARS = thiobarbituric acid reactive substances; VO2 = oxygen uptake; VO2max = maximal . VO2; XO = xanthine oxidase; Zn = zinc; › indicates increase; fl indicates decrease; - indicates ‘leads to’/outcome.
1056
ª 2011 Adis Data Information BV. All rights reserved.
Table I. Contd
Antioxidant Supplementation in Exercise Training
5.1 Antioxidant Supplements Promote Exercise-Induced Oxidative Stress
Antioxidants, especially when present in high amounts, have been shown to increase markers of exercise-induced oxidative stress. After highintensity exercise, coenzyme Q10 supplementation was associated with an increase in a marker of cell damage (CK)[249] and a decrease in exercise-training induced improvements in physical performance.[249,250] A number of important methodological details were omitted from the articles, indicating low quality. A study by Childs et al.[175] found that vitamin C and N-acetylcysteine following eccentric arm exercise increased oxidative stress and cell damage above levels induced by muscle injury alone. The effects of vitamins E and C alone and in combination were investigated in seven male cyclists.[144] Vitamin E decreased malondialdehyde, an oxidative stress marker, whereas the combination of both had no effect and vitamin C increased malondialdehyde. This indicates that the type of antioxidant (e.g. water vs lipid soluble) is likely to be an important factor. In another study, an increase in the serum CK levels following a 3-day resistance exercise was greater after the use of vitamin E supplements compared with a placebo group.[202] However, the increase was both modest and transient with no effect of supplementation on muscle soreness and exercise performance. Furthermore, variability in the baseline CK levels between groups and the large interindividual variability of the measure need to be considered. Two months of supplementation with high doses of vitamin E had no effect on the race time of Ironman Triathlon participants but was associated with increased lipid peroxidation and inflammation.[176] Knez et al.[93] demonstrated that ultraendurance training upregulated the resting activity of several antioxidant enzymes and reduced resting levels of oxidative stress, whilst supplementation with vitamins C and E had no effect on these values. Moreover, athletes taking supplements had elevated post-race malondialdehyde levels compared with nonsupplementers. It is important to recognize that this was only an observaª 2011 Adis Data Information BV. All rights reserved.
1057
tional study; although, when a randomized controlled crossover design was used, similar findings were reported with 2 weeks of supplementation with an antioxidant concentrate (vitamins E, C, b-carotene and selenium) associated with increased lipid peroxidation and decreased plasma glutathione peroxidase concentration pre- and post-exercise.[174] Finally, in a recent study by Bailey et.al.,[110] young men were supplemented with a combination of vitamins C and E for 6 weeks before and 2 days after a 90-minute intermittent shuttle run. The supplemented subjects had increased markers of oxidative stress and inflammation compared with the placebo group. However, although the overall change in isoprostane levels (baseline vs post-exercise) approached significance, the tendency for slightly higher isoprostane levels in the placebo group at baseline precluded establishment of any significant differences at the final recovery timepoint. The authors noted that a large inter-individual variability in the responses of isoprostanes and interleukin (IL)6 after supplementation could have impacted on the findings. Indeed, in all of the above mentioned studies there were no attempts to provide sample size or power calculations to assess the likelihood that the findings were real. 5.2 Antioxidant Supplementation Hinders Cell Adaptation to Exercise-Induced Oxidative Stress
Cells adapt to increased exposure to oxidation, thereby reducing the risk of tissue damage.[90,98,258] Five small studies now show that antioxidant supplements hinder the beneficial cell adaptations to exercise.[97-99,252,253] In a group of untrained males, supplementation with vitamin C resulted in the inactivation of redox-sensitive transcription factors responsible for the expression of cytoprotective proteins, including HSPs.[98] Such suppression of cell adaptation may negatively impact cell viability over the longer term. Similarly, supplementation with g-tocopherol inhibited an exerciseinduced increase of HSP levels in skeletal muscle and the circulation.[99] A research group at the University of Valencia, Valencia, Spain has published a number of important studies on this topic. In one of their Sports Med 2011; 41 (12)
1058
first studies they used allopurinol in rats and found it attenuated the exercise-induced increase of XO activity and ROS formation.[252] This was associated with a decreased activation of mitogen-activated protein kinases (MAPKs) and blunted DNA-binding of nuclear factor kappa B (NF-kB). MAPKs respond to extracellular stimuli, including oxidative stress, and regulate cell development and survival. Transcription factor NF-kB mediates gene expression of enzymes such as Mn-SOD, eNOS and iNOS. Therefore, impairing the exercise training effects on MAPKs and NF-kB would likely impact on these positive benefits. Indeed, in humans, administration of allopurinol prior to a marathon race did suppress the exercise-induced increase of antioxidant enzyme expression.[253] In another study, GomezCabrera et al.[97] showed that chronic supplementation with vitamin C impacted on exercise performance by decreasing exercise training efficiency. This was shown in both humans and rats. Analysis of animal muscles showed that the antioxidant supplementation inhibited upregulation of Mn-SOD and GPX gene expression. Moreover, attenuated mitochondrial biogenesis in the supplemented rats was indicated by reduced protein levels of cytochrome c (cyt c) and transcription factors peroxisome proliferatoractivated receptor co-activator 1 (PGC-1), nuclear respiratory factor 1 (NRF-1) and mitochondrial transcription factor A (mTFA). Cyt c, a protein in the inner membrane of mitochondria, is an essential component of the electron transport chain and serves as a marker of mitochondrial content. PGC-1 is a transcriptional coactivator of the genes involved in cellular energy metabolism. It induces messenger RNA expression of NRF-1 and mTFA and provides a link between external physiological signals and mitochondrial biogenesis. In a recent study from our laboratory,[257] the effects of 14 weeks of antioxidant supplementation (a-tocopherol and a-lipoic acid) and treadmill exercise on myocardial and vascular endothelium gene expression were investigated in rats. Both antioxidant therapy and exercise training downregulated IL-6 gene expression, while the expression of the RAS homolog gene family ª 2011 Adis Data Information BV. All rights reserved.
Peternelj & Coombes
member A (RhoA), a gene involved in cardiovascular disease progression, was upregulated by antioxidant supplementation and downregulated by exercise. The combination of supplementation and exercise resulted in a blunted downregulation of RhoA expression. These findings confirmed an unfavourable effect of antioxidants on exerciseinduced cardiovascular protection. 5.3 Reactive Oxygen Species Elimination and Physiological Processes
Given that reactive species play an important role in the regulation of muscle contractile activity, their elimination with high doses of antioxidants may result in negative effects on muscle function. We have shown that supplementation of rats with vitamin E and a-lipoic acid decreased lipid peroxidation after a fatigue protocol but had no effect on fatigue resistance.[251] Moreover, high levels of vitamin E depressed muscle force production at low stimulation frequencies. Acute supplementation of rats with vitamin C and tempol, a radical scavenger, reduced skeletal muscle blood flow, oxygen utilization and force production at rest and during electrically stimulated contractions.[255] Close and colleagues[197] found consumption of high doses of vitamin C in the days postexercise delayed the recovery of muscle function in humans. Chronic supplementation of competitive kayakers with a mixture of vitamins and minerals failed to protect from exercise-induced oxidative stress and inflammation, and hindered the recovery of muscle damage after a 1000 m race.[165] Together, these findings suggest that ROS produced post-exercise play a role in muscle regeneration. Physical activity is known to improve insulin sensitivity as the transient rise in ROS production efficiently counteracts insulin resistance.[91] In one of the most interesting studies on this topic, Ristow et al.[91] reported that supplementation with vitamins E and C inhibited the insulin sensitizing effects of exercise training, regardless of previous training status. They found that exercise-induced oxidative stress increased expression of ROS-sensitive transcriptional regulators Sports Med 2011; 41 (12)
Antioxidant Supplementation in Exercise Training
of insulin sensitivity PGC-1a, PGC-1b and peroxisome proliferator-activated receptor-g, a nuclear receptor protein involved in fatty acid storage and glucose metabolism. Exercise training also decreased fasting plasma insulin levels and caused an adaptive response promoting endogenous antioxidant defence capacity by upregulation of SOD1, SOD2 and GPX gene expression. Supplementation with antioxidants precluded these health promoting effects of exercise in both pre-trained and untrained men. Reactive species act as potent vasodilators and may be an important part of the vasodilatory response during exercise. Administration of an antioxidant cocktail (vitamins C, E and a-lipoic acid) augmented plasma antioxidant capacity and reduced circulating levels of free radicals in a group of healthy young males.[254] Importantly, brachial artery vasodilation was decreased during a submaximal handgrip exercise in the supplemented group. The direct measurement of oxidative stress is a strength of this study. Wray et al.[256] from the same research group, showed that 6 weeks of single leg knee-extensor exercise lowered blood pressure at rest and during exercise in a group of mildly hypertensive older men. Acute administration of a-lipoic acid, vitamin C and vitamin E after the training period returned blood pressure to pre-training values. Furthermore, with exercise training, vasodilation improved significantly, but the effect was blunted after consuming antioxidants. It was concluded that antioxidant administration negated the health benefits of exercise training in older individuals. Although the study only included six subjects, the authors state they had sufficient statistical power. Negative outcomes following the combination of two potentially beneficial interventions emphasize the complex nature of oxidative stress. Reactive species in skeletal muscle are generated in response to physiological and pathophysiological stimuli and are not solely by-products of aerobic metabolism. Attempts to decrease their levels, such as, for example, through antioxidant supplementation, may lead to a blunting of positive effects of exercise and even deleterious health effects. ª 2011 Adis Data Information BV. All rights reserved.
1059
6. Limitations of the Studies and Future Directions An obvious limitation of the current body of research on this topic is the lack of studies investigating antioxidants other than vitamin E, vitamin C and coenzyme Q10. Despite the vast range of antioxidant supplements commercially available, many of these compounds have not been studied based on our systematic search. Therefore, generalizing the results to all antioxidant supplements may be problematic. Furthermore, numerous methodological issues interfere with the ability to interpret the effects of antioxidant supplementation on exercise. These include differences in exercise protocols, subject population, dosage and form of supplements, duration and timing of supplementation, and the methodology used to assess oxidative stress. It should be made clear that detection of differences between treatment and control groups in measured indices does not imply cause and effect of antioxidant supplements. Most studies investigated the effect of supplementation in small groups of subjects and did not employ a crossover design that could easily lead to type I and type II errors.[99,202,249,250,259] Null findings in supplementation studies could be partially explained by insufficient dosages or treatment durations and the lack of sensitive detection techniques. Most studies lacked information on the redox state of the subjects to confirm whether their endogenous defence system was actually overwhelmed by increased ROS formation. For instance, highly trained individuals may experience an attenuated oxidative stress response, especially with long-duration, lowintensity exercise protocols. This is likely due to an enhanced endogenous antioxidant defence that is sufficient to combat an increased free radical production, thus masking any potential effect of supplementation. However, prolonged vigorous exercise can lead to a very large increase in ROS production, overwhelming antioxidant systems. In such conditions, additional doses of antioxidants may not exert any significant effect on oxidative stress levels. Furthermore, detection depends, to a large degree, on the tissue/biofluid sampled, the timing Sports Med 2011; 41 (12)
Peternelj & Coombes
1060
of sampling and the sensitivity and specificity of the chosen biomarker. For example, in some studies, oxidative stress may have occurred preceding or following the sample collection and was therefore not detected. Importantly, nearly all of the studies included in the review did not determine the actual levels of ROS but, rather, measured indirect markers of oxidative stress, such as by-products of lipid, protein and DNA damage.[93,144,174,175,197,202,259] In addition, in the majority of the studies, a single assay analysis of oxidative stress was used. Indeed, investigating only a particular oxidative stress marker does not represent universal oxidative stress status. Given the complexity of oxidative stress, a number of markers should be chosen (e.g. lipid peroxidation and protein oxidation measures). Moreover, changes in redox status within cells may be compartmentalized and regulated via specific signalling pathways. It seems highly unlikely that various potential targets in cells would show an equivalent sensitivity to specific ROS. In addition, ROS are present in low concentrations in biological systems, have short half-lives and are highly reactive. Thus, direct measurement is difficult and as reactive species cannot be targeted easily exogenous antioxidants may not scavenge the relevant ROS. Difficulty in quantifying oxidative stress and understanding the health implications of oxidative stress measures are important issues when establishing appropriate intervention strategies. Despite the increasing awareness of the importance of reactive species, screening and monitoring of oxidative stress has not yet become routinely available. Individuals are often recommended antioxidant therapy, although there is no test that advises whether to assess if they are exposed to increased levels of free radicals or have depleted antioxidant capacities. Careful reassessment of the existing evidence is warranted to better understand the conflicting data and design future studies appropriately. There is a need for more rigorous clinical trial designs with populations under high levels of oxidative stress and carefully chosen outcomes. Large randomized controlled trials with exercising individuals consuming a variety of antiª 2011 Adis Data Information BV. All rights reserved.
oxidant supplements and using hard endpoints, such as onset of disease, would need to be conducted to adequately address the question of the impact of antioxidant supplementation on exercise-induced oxidative stress. Bioavailability and pharmacokinetics of antioxidants should be examined closely to establish the dosage, timing and duration of supplementation that would significantly reduce oxidative stress levels in the study participants. In addition, nutrigenomic issues might be considered as people respond differently to particular antioxidants based on their genetic profile. Further research, supported by improved techniques to measure oxidative stress and target specific ROS, will help to clarify the potential roles of antioxidant supplements in exercise-training. 7. Optimizing Nutrition 7.1 Summary
Studies included in this review have demonstrated disparate results with regards to the effects of antioxidant supplementation on exerciseinduced oxidative stress. In summary, there is insufficient evidence to recommend antioxidant supplements for exercising individuals who consume the recommended amounts of dietary antioxidants through food. Antioxidant supplements generally do not improve physical performance. There is little proof to support their role in prevention of exercise-induced muscle damage and enhancement of recovery. Although ingesting supplemental antioxidants can decrease exerciseinduced oxidative stress, there is no evidence that this confers health benefits. Further work is warranted to illuminate the interactive effects of exercise training and antioxidant supplementation. 7.2 Current Recommendations
The outcomes of supplementation studies have important implications for nutritionists, physicians, practitioners, exercise trainers and athletes, as well as for the general population. Reports that high doses of antioxidants preclude health-promoting effects of exercise training and interfere with ROS-mediated physiological proSports Med 2011; 41 (12)
Antioxidant Supplementation in Exercise Training
cesses suggest caution in the use of antioxidant supplements. Physically active individuals need to optimize their nutrition rather than use supplements. Diets rich in antioxidants should be attained by consuming a variety of fruits, vegetables, whole grains and nuts. Whole foods, rather than capsules, contain antioxidants presented in beneficial ratios and numerous phytochemicals that may act in synergy with the former to optimize the antioxidant effect. Antioxidant supplementation may be warranted when individuals are exposed to high levels of oxidative stress and struggle to meet the dietary antioxidant requirements. Athletes, who restrict their energy intake, use severe weight loss practices and eliminate one or more food groups from their diet or consume unbalanced diets with low micronutrient density, are at risk of suboptimal antioxidant status. A qualified sports dietitian would need to provide individualized nutrition direction and advice subsequent to blood analysis and comprehensive nutritional assessment. Careful product evaluation is required prior to adopting an antioxidant regimen, which should be clinically supervised and should only represent a short-term solution while dietary changes are being implemented. 8. Conclusions The multifunctional role of reactive species in living organisms, and the beneficial and deleterious effects of antioxidant supplementation demonstrate the complexity of exercise-induced oxidative stress. Interactions of antioxidants and reactive species should be carefully considered as the redox state will dictate cell functioning. More detailed research and critical appraisal of the situations that may warrant antioxidant supplementation in exercise training are required. A balanced diet including a variety of fruits and vegetables remains the best nutritional approach to maintain optimal antioxidant status. Acknowledgements The authors wish to declare no conflicts of interest or funding that are directly relevant to the content of this review.
ª 2011 Adis Data Information BV. All rights reserved.
1061
References 1. Maughan RJ, Depiesse F, Geyer H. The use of dietary supplements by athletes. J Sports Sci 2007; 25 Suppl. 1: S103-13 2. Atalay M, Lappalainen J, Sen CK. Dietary antioxidants for the athlete. Curr Sports Med Rep 2006 Jun; 5 (4): 182-6 3. Clarkson PM, Thompson HS. Antioxidants: what role do they play in physical activity and health? Am J Clin Nutr 2000 Aug; 72 (2 Suppl.): S637-46 4. Kanter M. Free radicals, exercise and antioxidant supplementation. Proc Nutr Soc 1998 Feb; 57 (1): 9-13 5. Margaritis I, Rousseau AS. Does physical exercise modify antioxidant requirements? Nutr Res Rev 2008 Jun; 21 (1): 3-12 6. McGinley C, Shafat A, Donnelly AE. Does antioxidant vitamin supplementation protect against muscle damage? Sports Med 2009; 39 (12): 1011-32 7. Urso ML, Clarkson PM. Oxidative stress, exercise, and antioxidant supplementation. Toxicology 2003 Jul 15; 189 (1-2): 41-54 8. Williams SL, Strobel NA, Lexis LA, et al. Antioxidant requirements of endurance athletes: implications for health. Nutr Rev 2006 Mar; 64 (3): 93-108 9. Clarkson PM, Hubal MJ. Exercise-induced muscle damage in humans. Am J Phys Med Rehabil 2002 Nov; 81 (11 Suppl.): S52-69 10. Peake JM, Suzuki K, Coombes JS. The influence of antioxidant supplementation on markers of inflammation and the relationship to oxidative stress after exercise. J Nutr Biochem 2007 Jun; 18 (6): 357-71 11. Powers SK, DeRuisseau KC, Quindry J, et al. Dietary antioxidants and exercise. J Sports Sci 2004 Jan; 22 (1): 81-94 12. Farbstein D, Kozak-Blickstein A, Levy AP. Antioxidant vitamins and their use in preventing cardiovascular disease. Molecules 2010; 15 (11): 8098-110 13. Stanner SA, Hughes J, Kelly CN, et al. A review of the epidemiological evidence for the ‘antioxidant hypothesis’. Public Health Nutr 2004 May; 7 (3): 407-22 14. Willcox BJ, Curb JD, Rodriguez BL. Antioxidants in cardiovascular health and disease: key lessons from epidemiologic studies. Am J Cardiol 2008 May 22; 101 (10A): D75-86 15. Han-Yao H, Caballero B, Chang S, et al. The efficacy and safety of multivitamin and mineral supplement use to prevent cancer and chronic disease in adults: a systematic review for a National Institutes of Health State-of-theScience Conference. Ann Intern Med 2006; 145 (5): 372-85 16. Stocker R, Keaney JF. Role of oxidative modifications in atherosclerosis. Physiol Rev 2004; 84 (4): 1381-478 17. Devasagayam TP, Tilak JC, Boloor KK, et al. Free radicals and antioxidants in human health: current status and future prospects. J Assoc Physicians 2004; 52: 794-804 18. Kohen R, Nyska A. Oxidation of biological systems: oxidative stress phenomena, antioxidants, redox reactions, and methods for their quantification. Toxicol Pathol 2002; 30 (6): 620-50 19. Jacob C, Winyard PG, editor. Redox signaling and regulation in biology and medicine. Weinheim: Wiley-VCH, 2009
Sports Med 2011; 41 (12)
1062
20. Rada´k Z, editor. Free radicals in exercise and aging. Champaign (IL): Human Kinetics, 2000 21. St-Pierre J, Buckingham JA, Roebuck SJ, et al. Topology of superoxide production from different sites in the mitochondrial electron transport chain. J Biol Chem 2002 Nov 22; 277 (47): 44784-90 22. Boveris A, Chance B. The mitochondrial generation of hydrogen peroxide: general properties and effect of hyperbaric oxygen. Biochem J 1973 Jul; 134 (3): 707-16 23. Boveris A, Oshino N, Chance B. The cellular production of hydrogen peroxide. Biochem J 1972 Jul; 128 (3): 617-30 24. Powers SK, Jackson MJ. Exercise-induced oxidative stress: cellular mechanisms and impact on muscle force production. Physiol Rev 2008 Oct; 88 (4): 1243-76 25. Aruoma OI, Halliwell B, Gajewski E, et al. Copper-iondependent damage to the bases in DNA in the presence of hydrogen peroxide. Biochem J 1991 Feb 1; 273 (Pt 3): 601-4 26. Halliwell B. Phagocyte-derived reactive species: salvation or suicide? Trends Biochem Sci 2006 Sep; 31 (9): 509-15 27. Dalle-Donne I, Rossi R, Colombo R, et al. Biomarkers of oxidative damage in human disease. Clin Chem 2006 Apr 1; 52 (4): 601-23 28. Barreiro E, Hussain SNA. Protein carbonylation in skeletal muscles: impact on function. Antioxid Redox Signal 2010; 12 (3): 417-29 29. Staib JL, Tu¨mer N, Powers SK. Increased temperature and protein oxidation lead to HSP72 mRNA and protein accumulation in the in vivo exercised rat heart. Exp Physiol 2009; 94 (1): 71-80 30. Davies KJ. Protein damage and degradation by oxygen radicals. I: general aspects. J Biol Chem 1987 Jul 15; 262 (20): 9895-901 31. Halliwell B, Chirico S. Lipid peroxidation: its mechanism, measurement, and significance. Am J Clin Nutr 1993 May; 57 (5 Suppl.): S715-24; discussion S24-25 32. Dizdaroglu M, Jaruga P, Birincioglu M, et al. Free radicalinduced damage to DNA: mechanisms and measurement. Free Radic Biol Med 2002; 32 (11): 1102-15 33. Los M, Droge W, Stricker K, et al. Hydrogen peroxide as a potent activator of T lymphocyte functions. Eur J Immunol 1995; 25 (1): 159-65 34. Kannan K, Jain SK. Oxidative stress and apoptosis. Pathophysiology 2000; 7 (3): 153-63 35. Dillard CJ, Litov RE, Savin WM, et al. Effects of exercise, vitamin E, and ozone on pulmonary function and lipid peroxidation. J Appl Physiol 1978; 45 (6): 927-32 36. Jessup JV, Horne C, Yarandi H, et al. The effects of endurance exercise and vitamin E on oxidative stress in the elderly: biological research for nursing 2003; 5 (1): 47-55 37. Blot WJ, Li JY, Taylor PR, et al. Nutrition intervention trials in Linxian, China: supplementation with specific vitamin/mineral combinations, cancer incidence, and disease-specific mortality in the general population. J Natl Cancer Inst 1993; 85 (18): 1483-92 38. Stephens NG, Parsons A, Schofield PM, et al. Randomised controlled trial of vitamin E in patients with coronary disease: Cambridge Heart Antioxidant Study (CHAOS). Lancet 1996; 347 (9004): 781-6
ª 2011 Adis Data Information BV. All rights reserved.
Peternelj & Coombes
39. Niki E. Interaction of ascorbate and alpha-tocopherol. Ann N Y Acad Sci 1987; 498: 186-99 40. Brigelius-Flohe´ R, Traber MG. Vitamin E: function and metabolism. FASEB J 1999; 13 (10): 1145-55 41. Padayatty SJ, Katz A, Wang Y, et al. Vitamin C as an antioxidant: evaluation of its role in disease prevention. J Am Coll Nutr 2003; 22 (1): 18-35 42. Paiva SAR, Russell RM. beta-Carotene and other carotenoids as antioxidants. J Am Coll Nutr 1999; 18 (5): 426-33 43. Mueller L, Boehm V. Antioxidant activity of b-carotene compounds in different in vitro assays. Molecules 2011; 16 (2): 1055-69 44. Chew BP, Park JS. Carotenoid action on the immune response. J Nutr 2004 Jan 1; 134 (1): S257-61 45. Bentinger M, Tekle M, Dallner G. Coenzyme Q: biosynthesis and functions. Biochem Biophys Res Commun 2010; 396 (1): 74-9 46. Quideau S, Deffieux D, Douat-Casassus C, et al. Plant polyphenols: chemical properties, biological activities, and synthesis. In: Peter Golitz, editor. Angewandte Chemie International Edition. Weinheim: Wiley, 2011; 50 (3): 586-621 47. Miatello R, Va´zquez M, Renna N, et al. Chronic administration of resveratrol prevents biochemical cardiovascular changes in fructose-fed rats. Am J Hypertens 2005; 18 (6): 864-70 48. Knekt P, Kumpulainen J, Ja¨rvinen R, et al. Flavonoid intake and risk of chronic diseases. Am J Clin Nutr 2002; 76 (3): 560-8 49. Kagan VE, Serbinova EA, Forte T, et al. Recycling of vitamin E in human low density lipoproteins. J Lipid Res 1992; 33 (3): 385-97 50. Petersen Shay K, Moreau RF, Smith EJ, et al. Is alphalipoic acid a scavenger of reactive oxygen species in vivo? Evidence for its initiation of stress signaling pathways that promote endogenous antioxidant capacity. IUBMB Life 2000; 60 (6): 362-7 51. Kerksick C, Willoughby D. The antioxidant role of glutathione and N-acetyl-cysteine supplements and exercise-induced oxidative stress. J Int Soc Sports Nutr 2005; 2: 38-44 52. Pacher P, Nivorozhkin A, Szabo´ C. Therapeutic effects of xanthine oxidase inhibitors: renaissance half a century after the discovery of allopurinol. Pharmacol Rev 2006 Mar 1; 58 (1): 87-114 53. Sies H. Strategies of antioxidant defense. Eur J Biochem 1993; 215 (2): 213-9 54. Sies H. Oxidative stress: from basic research to clinical application. Am J Med 1991; 91 (3C): S31-8 55. Jones DP. Redefining oxidative stress. Antioxid Redox Signal 2006; 8 (9-10): 1865-79 56. Sharman JE, Cockcroft JR, Coombes JS. Cardiovascular implications of exposure to traffic air pollution during exercise. QJM 2004; 97 (10): 637-43 57. Sorg O. Oxidative stress: a theoretical model or a biological reality? C R Biol 2004; 327 (7): 649-62 58. Roberts CK, Barnard RJ. Effects of exercise and diet on chronic disease. J Appl Physiol 2005; 98 (1): 3-30
Sports Med 2011; 41 (12)
Antioxidant Supplementation in Exercise Training
59. Herrera E, Jime´nez R, Aruoma OI, et al. Aspects of antioxidant foods and supplements in health and disease. Nutr Rev 2009; 67: S140-4 60. Alessio HM, Goldfarb AH, Cao G. Exercise-induced oxidative stress before and after vitamin C supplementation. Int J Sport Nutr 1997; 7 (1): 1-9 61. Goldfarb AH, Bloomer RJ, McKenzie MJ. Combined antioxidant treatment effects on blood oxidative stress after eccentric exercise. Med Sci Sports Exerc 2005; 37 (2): 234-9 62. Meydani M, Evans WJ, Handelman G, et al. Protective effect of vitamin E on exercise-induced oxidative damage in young and older adults. Am J Physiol Regul Integr Comp Physiol 1993; 264 (5): R992-8 63. Vina J, Gomez-Cabrera MC, Borras C. Fostering antioxidant defences: up-regulation of antioxidant genes or antioxidant supplementation? Br J Nutr 2007; 98 Suppl. 1: S36-40 64. Lykkesfeldt J, Poulsen HE. Is vitamin C supplementation beneficial? Lessons learned from randomised controlled trials. Br J Nutr 2010; 103 (09): 1251-9 65. Kritchevsky SB, Shimakawa T, Tell GS, et al. Dietary antioxidants and carotid artery wall thickness. The ARIC Study. Atherosclerosis Risk in Communities Study. Circulation 1995; 92 (8): 2142-50 66. Gaziano JM, Manson JE, Branch LG, et al. A prospective study of consumption of carotenoids in fruits and vegetables and decreased cardiovascular mortality in the elderly. Ann Epidemiol 1995; 5 (4): 255-60 67. Gey KF, Brubacher GB, Stahelin HB. Plasma levels of antioxidant vitamins in relation to ischemic heart disease and cancer. Am J Clin Nutr 1987; 45 (5 Suppl.): 1368-77 68. Schunemann HJ, Grant BJ, Freudenheim JL, et al. The relation of serum levels of antioxidant vitamins C and E, retinol and carotenoids with pulmonary function in the general population. Am J Respir Crit Care Med 2001; 163 (5): 1246-55 69. Combs Jr GF, Clark LC, Turnbull BW. Reduction of cancer mortality and incidence by selenium supplementation. Med Klin (Munich) 1997; 92 Suppl. 3: 42-5 70. Lonn E, Bosch J, Yusuf S, et al. Effects of long-term vitamin E supplementation on cardiovascular events and cancer: a randomized controlled trial. JAMA 2005; 293 (11): 1338-47 71. Sesso HD, Buring JE, Christen WG, et al. Vitamins E and C in the prevention of cardiovascular disease in men: the Physicians’ Health Study II randomized controlled trial. JAMA 2008; 300 (18): 2123-33 72. Lee IM, Cook NR, Manson JE, et al. Beta-carotene supplementation and incidence of cancer and cardiovascular disease: the Women’s Health Study. J Natl Cancer Inst 1999; 91 (24): 2102-6 73. Heart Protection Study Collaborative Group. MRC/BHF Heart Protection Study of antioxidant vitamin supplementation in 20,536 high-risk individuals: a randomised placebo-controlled trial. Lancet 2002; 360 (9326): 23-33 74. Albanes D, Heinonen OP, Taylor PR, et al. Alphatocopherol and beta-carotene supplements and lung cancer incidence in the alpha-tocopherol, beta-carotene cancer prevention study: effects of base-line characteristics
ª 2011 Adis Data Information BV. All rights reserved.
1063
75.
76.
77.
78.
79. 80.
81.
82.
83.
84. 85.
86.
87.
88.
89. 90.
91.
92.
and study compliance. J Natl Cancer Inst 1996; 88 (21): 1560-70 Goodman GE, Thornquist MD, Balmes J, et al. The betacarotene and retinol efficacy trial: incidence of lung cancer and cardiovascular disease mortality during 6-year follow-up after stopping beta-carotene and retinol supplements. J Natl Cancer Inst 2004; 96 (23): 1743-50 Bjelakovic G, Nikolova D, Gluud LL, et al. Mortality in randomized trials of antioxidant supplements for primary and secondary prevention: systematic review and metaanalysis. JAMA 2007; 297 (8): 842-57 Thannickal VJ, Fanburg BL. Reactive oxygen species in cell signaling. Am J Physiol Lung Cell Mol Physiol 2000; 279 (6): L1005-28 Valko M, Leibfritz D, Moncol J, et al. Free radicals and antioxidants in normal physiological functions and human disease. Int J Biochem Cell Biol 2007; 39 (1): 44-84 Allen RG, Tresini M. Oxidative stress and gene regulation. Free Radic Biol Med 2000; 28 (3): 463-99 Li T-S, Marba´n E. Physiological levels of reactive oxygen species are required to maintain genomic stability in stem cells. Stem Cells 2010; 28 (7): 1178-85 Panasyuk A, Frati E, Ribault D, et al. Effect of reactive oxygen species on the biosynthesis and structure of newly synthesized proteoglycans. Free Radic Biol Med 1994; 16 (2): 157-67 Wagner AH, Kohler T, Ruckschloss U, et al. Improvement of nitric oxide-dependent vasodilatation by HMG-CoA reductase inhibitors through attenuation of endothelial superoxide anion formation. Arterioscler Thromb Vasc Biol 2000; 20 (1): 61-9 Jackson MJ. Free radicals generated by contracting muscle: by-products of metabolism or key regulators of muscle function? Free Radic Biol Med 2008; 44 (2): 132-41 Sen CK. Oxidants and antioxidants in exercise. J Appl Physiol 1995; 79 (3): 675-86 Brady PS, Brady LJ, Ullrey DE. Selenium, vitamin E and the response to swimming stress in the rat. J Nutr 1979; 109 (6): 1103-9 Davies KJ, Quintanilha AT, Brooks GA, et al. Free radicals and tissue damage produced by exercise. Biochem Biophys Res Commun 1982; 107 (4): 1198-205 Niess AM, Simon P. Response and adaptation of skeletal muscle to exercise: the role of reactive oxygen species. Front Biosci 2007; 12: 4826-38 Ji LL. Modulation of skeletal muscle antioxidant defense by exercise: role of redox signaling. Free Radic Biol Med 2008 Jan 15; 44 (2): 142-52 Radak Z, Chung HY, Koltai E, et al. Exercise, oxidative stress and hormesis. Ageing Res Rev 2008 Jan; 7 (1): 34-42 Gomez-Cabrera MC, Domenech E, Vina J. Moderate exercise is an antioxidant: upregulation of antioxidant genes by training. Free Radic Biol Med 2008 Jan 15; 44 (2): 126-31 Ristow M, Zarse K, Oberbach A, et al. Antioxidants prevent health-promoting effects of physical exercise in humans. Proc Natl Acad Sci U S A 2009; 106 (21): 8665-70 Chang CK, Huang HY, Tseng HF, et al. Interaction of vitamin E and exercise training on oxidative stress and
Sports Med 2011; 41 (12)
Peternelj & Coombes
1064
93.
94.
95.
96.
97.
98.
99.
100.
101.
102. 103.
104.
105.
106.
107.
108.
109.
110.
antioxidant enzyme activities in rat skeletal muscles. J Nutr Biochem 2007; 18 (1): 39-45 Knez WL, Jenkins DG, Coombes JS. Oxidative stress in half and full Ironman triathletes. Med Sci Sports Exerc 2007 Feb; 39 (2): 283-8 Pikosky MA, Gaine PC, Martin WF, et al. Aerobic exercise training increases skeletal muscle protein turnover in healthy adults at rest. J Nutr 2006; 136 (2): 379-83 Rada´k Z, Apor P, Pucsok J, et al. Marathon running alters the DNA base excision repair in human skeletal muscle. Life Sciences 2003; 72 (14): 1627-33 Okamura K, Doi T, Sakurai M, et al. Effect of endurance exercise on the tissue 8-hydroxy-deoxyguanosine content in dogs. Free Radic Res 1997; 26 (6): 523-8 Gomez-Cabrera MC, Domenech E, Romagnoli M, et al. Oral administration of vitamin C decreases muscle mitochondrial biogenesis and hampers training-induced adaptations in endurance performance. Am J Clin Nutr 2008; 87 (1): 142-9 Khassaf M, McArdle A, Esanu C, et al. Effect of vitamin C supplements on antioxidant defence and stress proteins in human lymphocytes and skeletal muscle. J Physiol 2003; 549 (Pt 2): 645-52 Fischer CP, Hiscock NJ, Basu S, et al. Vitamin E isoformspecific inhibition of the exercise-induced heat shock protein 72 expression in humans. J Appl Physiol 2006; 100 (5): 1679-87 Radak Z, Chung HY, Goto S. Systemic adaptation to oxidative challenge induced by regular exercise. Free Radic Biol Med 2008; 44 (2): 153-9 Powers SK, Duarte J, Kavazis AN, et al. Reactive oxygen species are signalling molecules for skeletal muscle adaptation. Exp Physiol 2010; 95 (1): 1-9 Mattson MP. Hormesis defined. Ageing Res Rev 2008; 7 (1): 1-7 Ristow M, Zarse K. How increased oxidative stress promotes longevity and metabolic health: the concept of mitochondrial hormesis (mitohormesis). Exp Gerontol 2010; 45 (6): 410-8 Reid MB. Invited review: redox modulation of skeletal muscle contraction: what we know and what we don’t. J Appl Physiol 2001; 90 (2): 724-31 Reid MB, Khawli FA, Moody MR. Reactive oxygen in skeletal muscle. III: contractility of unfatigued muscle. J Appl Physiol 1993; 75 (3): 1081-7 Reid MB, Moody MR. Dimethyl sulfoxide depresses skeletal muscle contractility. J Appl Physiol 1994; 76 (5): 2186-90 Reid MB. Nitric oxide, reactive oxygen species, and skeletal muscle contraction. Med Sci Sports Exerc 2001; 33 (3): 371-6 Newham DJ, McPhail G, Mills KR, et al. Ultrastructural changes after concentric and eccentric contractions of human muscle. J Neurol Sci 1983; 61 (1): 109-22 Raastad T, Owe SG, Paulsen G, et al. Changes in calpain activity, muscle structure, and function after eccentric exercise. Med Sci Sports Exerc 2010; 42 (1): 86-95 Bailey DM, Williams C, Betts JA, et al. Oxidative stress, inflammation and recovery of muscle function after da-
ª 2011 Adis Data Information BV. All rights reserved.
111.
112.
113.
114.
115.
116.
117. 118.
119.
120.
121.
122.
123.
124.
125.
126.
127.
maging exercise: effect of 6-week mixed antioxidant supplementation. Eur J Appl Physiol 2011; 111 (6): 925-36 Beaton LJ, Allan DA, Tarnopolsky MA, et al. Contraction-induced muscle damage is unaffected by vitamin E supplementation. Med Sci Sports Exerc 2002; 34 (5): 798-805 Van Der Meulen JH, McArdle A, Jackson MJ, et al. Contraction-induced injury to the extensor digitorum longus muscles of rats: the role of vitamin E. J Appl Physiol 1997; 83 (3): 817-23 Thompson D, Williams C, Kingsley M, et al. Muscle soreness and damage parameters after prolonged intermittent shuttle-running following acute vitamin C supplementation. Int J Sports Med 2001; 22 (1): 68-75 Thompson D, Bailey DM, Hill J, et al. Prolonged vitamin C supplementation and recovery from eccentric exercise. Eur J Appl Physiol 2004; 92 (1-2): 133-8 Bailey DM, Lawrenson L, Mceneny J, et al. Electron paramagnetic spectroscopic evidence of exercise-induced free radical accumulation in human skeletal muscle. Free Radic Res 2007; 41 (2): 182-90 Thompson D, Williams C, McGregor SJ, et al. Prolonged vitamin C supplementation and recovery from demanding exercise. Int J Sport Nutr Exerc Metab 2001; 11 (4): 466-81 Young IM, Thomson K. Spinning-induced rhabdomyolysis: a case report. Eur J Emerg Med 2004; 11 (6): 358-9 Millet GY, Tomazin K, Verges S, et al. Neuromuscular consequences of an extreme mountain ultra-marathon. PLoS One 2011; 6 (2): e17059 Byrd SK. Alterations in the sarcoplasmic reticulum: a possible link to exercise-induced muscle damage. Med Sci Sports Exerc 1992; 24 (5): 531-6 Thomas AC, McLean SG, Palmieri-Smith RM. Quadriceps and hamstrings fatigue alters hip and knee mechanics. J Appl Biomech 2010; 26 (2): 159-70 Sobal J, Marquart LF. Vitamin/mineral supplement use among athletes: a review of the literature. Int J Sport Nutr 1994; 4 (4): 320-34 Slater G, Tan B, Teh KC. Dietary supplementation practices of Singaporean athletes. Int J Sport Nutr Exerc Metab 2003; 13 (3): 320-32 Braun H, Koehler K, Geyer H, et al. Dietary supplement use among elite young German athletes. Int J Sport Nutr Exerc Metab 2009; 19 (1): 97-109 Froiland K, Koszewski W, Hingst J, et al. Nutritional supplement use among college athletes and their sources of information. Int J Sport Nutr Exerc Metab 2004; 14 (1): 104-20 Krumbach CJ, Ellis DR, Driskell JA. A report of vitamin and mineral supplement use among university athletes in a division I institution. Int J Sport Nutr 1999; 9 (4): 416-25 Sacheck JM, Milbury PE, Cannon JG, et al. Effect of vitamin E and eccentric exercise on selected biomarkers of oxidative stress in young and elderly men. Free Radic Biol Med 2003; 34 (12): 1575-88 Silva L, Pinho C, Silveira P, et al. Vitamin E supplementation decreases muscular and oxidative damage but not inflammatory response induced by eccentric contraction. J Physiol Sci 2010; 60 (1): 51-7
Sports Med 2011; 41 (12)
Antioxidant Supplementation in Exercise Training
128. Simon-Schnass I, Pabst H. Influence of vitamin E on physical performance. Int J Vitam Nutr Res 1988; 58 (1): 49-54 129. Satoshi S, Kiyoji T, Hiroyo K, et al. Exercise-induced lipid peroxidation and leakage of enzymes before and after vitamin E supplementation. Int J Biochem 1989; 21 (8): 835-8 130. Itoh H, Ohkuwa T, Yamazaki Y, et al. Vitamin E supplementation attenuates leakage of enzymes following 6 successive days of running training. Int J Sports Med 2000; 21 (5): 369-74 131. Reznick AZ, Witt E, Matsumoto M, et al. Vitamin E inhibits protein oxidation in skeletal muscle of resting and exercised rats. Biochem Biophys Res Comm 1992; 189 (2): 801-6 132. Rokitzki L, Logemann E, Huber G, et al. Alphatocopherol supplementation in racing cyclists during extreme endurance training. Int J Sport Nutr 1994; 4 (3): 253-64 133. Sanchez-Quesada JL, Jorba O, Payes A, et al. Ascorbic acid inhibits the increase in low-density lipoprotein (LDL) susceptibility to oxidation and the proportion of electronegative LDL induced by intense aerobic exercise. Coron Artery Dis 1998; 9 (5): 249-55 134. Tauler P, Aguilo´ A, Gimeno I, et al. Influence of vitamin C diet supplementation on endogenous antioxidant defences during exhaustive exercise. Pflu¨gers Arch Eur J Physiol 2003; 446 (6): 658-64 135. Vasankari T, Kujala U, Sarna S, et al. Effects of ascorbic acid and carbohydrate ingestion on exercise induced oxidative stress. J Sports Med Phys Fitness 1998 Dec; 38 (4): 281-5 136. Ashton T, Young IS, Peters JR, et al. Electron spin resonance spectroscopy, exercise, and oxidative stress: an ascorbic acid intervention study. J Appl Physiol 1999; 87 (6): 2032-6 137. Goldfarb AH, Patrick SW, Bryer S, et al. Vitamin C supplementation affects oxidative-stress blood markers in response to a 30-minute run at 75% VO2max. Int J Sport Nutr Exerc Metab 2005; 15 (3): 279-90 138. Rosa EF, Ribeiro RF, Pereira FMT, et al. Vitamin C and E supplementation prevents mitochondrial damage of ileum myocytes caused by intense and exhaustive exercise training. J Appl Physiol 2009; 107 (5): 1532-8 139. Ryan MJ, Dudash HJ, Docherty M, et al. Vitamin E and C supplementation reduces oxidative stress, improves antioxidant enzymes and positive muscle work in chronically loaded muscles of aged rats. Exp Gerontol 2010; 45 (11): 882-95 140. Schro¨der H, Navarro E, Mora J, et al. Effects of a-tocopherol, b-carotene and ascorbic acid on oxidative, hormonal and enzymatic exercise stress markers in habitual training activity of professional basketball players. Eur J Nutr 2001; 40 (4): 178-84 141. Tauler P, Aguilo´ A, Gimeno I, et al. Response of blood cell antioxidant enzyme defences to antioxidant diet supplementation and to intense exercise. Eur J Nutr 2006; 45 (4): 187-95 142. Kanter MM, Nolte LA, Holloszy JO. Effects of an antioxidant vitamin mixture on lipid peroxidation at rest and postexercise. J Appl Physiol 1993; 74 (2): 965-9
ª 2011 Adis Data Information BV. All rights reserved.
1065
143. Tauler P, Aguilo´ A, Fuentespina E, et al. Diet supplementation with vitamin E, vitamin C and X-carotene cocktail enhances basal neutrophil antioxidant enzymes in athletes. Pflu¨gers Arch European J Physiol 2002; 443 (5): 791-7 144. Bryant RJ, Ryder J, Martino P, et al. Effects of vitamin E and C supplementation either alone or in combination on exercise-induced lipid peroxidation in trained cyclists. J Strength Cond Res 2003 Nov; 17 (4): 792-800 145. Aguilo´ A, Tauler P, Fuentespina E, et al. Antioxidant diet supplementation influences blood iron status in endurance athletes. Int J Sport Nutr Exerc Metab 2004; 14 (2): 147-60 146. Zoppi CC, Hohl R, Silva FC, et al. Vitamin C and e supplementation effects in professional soccer players under regular training. J Int Soc Sports Nutr 2006; 3: 37-44 147. Fischer CP, Hiscock NJ, Penkowa M, et al. Supplementation with vitamins C and E inhibits the release of interleukin-6 from contracting human skeletal muscle. J Physiol 2004; 558 (Pt 2): 633-45 148. Giacomo CD, Acquaviva R, Sorrenti V, et al. Oxidative and antioxidant status in plasma of runners: effect of oral supplementation with natural antioxidants. J Med Food 2009; 12 (1): 145-50 149. Jackson JR, Ryan MJ, Hao Y, et al. Mediation of endogenous antioxidant enzymes and apoptotic signaling by resveratrol following muscle disuse in the gastrocnemius muscles of young and old rats. Am J Physiol Regul Integr Comp Physiol 2010; 299 (6): R1572-81 150. Ryan MJ, Jackson JR, Hao Y, et al. Suppression of oxidative stress by resveratrol after isometric contractions in gastrocnemius muscles of aged mice. J Gerontol Series A Biol Sci Med Sci 2010; 65 (8): 815-31 151. Chang W-H, Hu S-P, Huang Y-F, et al. Effect of purple sweet potato leaves consumption on exercise-induced oxidative stress and IL-6 and HSP72 levels. J Appl Physiol 2010; 109 (6): 1710-5 152. Lafay S, Jan C, Nardon K, et al. Grape extract improves antioxidant status and physical performance in elite male athletes. J Sports Sci Med 2009; 8 (3): 468-80 153. Pilaczynska-Szczesniak L, Skarpanska-Steinborn A, Deskur E, et al. The influence of chokeberry juice supplementation on the reduction of oxidative stress resulting from an incremental rowing ergometer exercise. Int J Sport Nutr Exerc Metab 2005; 15 (1): 48-58 154. Nakazato K, Ochi E, Waga T. Dietary apple polyphenols have preventive effects against lengthening contractioninduced muscle injuries. Mol Nutr Food Res 2010; 54 (3): 364-72 155. Morillas-Ruiz JM, Villegas Garcı´ a JA, Lo´pez FJ, et al. Effects of polyphenolic antioxidants on exercise-induced oxidative stress. Clin Nutr 2006; 25 (3): 444-53 156. Bowtell JL, Sumners DP, Dyer A, et al. Montmorency cherry juice reduces muscle damage caused by intensive strength exercise. Med Sci Sports Exerc 2011; 43 (8): 1544-51 157. Sumida S, Doi T, Sakurai M, et al. Effect of a single bout of exercise and b-carotene supplementation on the urinary excretion of 8-hydroxy-deoxyguanosine in humans. Free Radic Res 1997; 27 (6): 607-18
Sports Med 2011; 41 (12)
1066
158. Chae C-H, Shin C-H, Kim H-T. The combination of [alpha]-lipoic acid supplementation and aerobic exercise inhibits lipid peroxidation in rat skeletal muscles. Nutr Res 2008; 28 (6): 399-405 159. Sen CK, Rankinen T, Vaisanen S, et al. Oxidative stress after human exercise: effect of N-acetylcysteine supplementation. J Appl Physiol 1994; 76 (6): 2570-7 160. Akil M, Gurbuz U, Bicer M, et al. Effect of selenium supplementation on lipid peroxidation, antioxidant enzymes, and lactate levels in rats immediately after acute swimming exercise. Biol Trace Elem Res 2010; 142 (3): 651-9 161. Kaikkonen J, Kosonen L, Nyysso¨nen K, et al. Effect of combined coenzyme Q10 and d-a-tocopheryl acetate supplementation on exercise-induced lipid peroxidation and muscular damage: a placebo-controlled double-blind study in marathon runners. Free Radic Res 1998; 29 (1): 85-92 162. Laaksonen R, Fogelholm M, Himberg J, et al. Ubiquinone supplementation and exercise capacity in trained young and older men. Eur J Appl Physiol Occup Physiol 1995; 72 (1): 95-100 163. Maxwell SR, Jakeman P, Thomason H, et al. Changes in plasma antioxidant status during eccentric exercise and the effect of vitamin supplementation. Free Radic Res Commun 1993; 19 (3): 191-202 164. Viitala PE, Newhouse IJ, LaVoie N, et al. The effects of antioxidant vitamin supplementation on resistance exercise induced lipid peroxidation in trained and untrained participants [abstract]. Lipids Health Dis 2004; 3: 14 165. Teixeira VH, Valente HF, Casal SI, et al. Antioxidants do not prevent postexercise peroxidation and may delay muscle recovery. Med Sci Sports Exerc 2009; 41 (9): 1752-60 166. Bloomer RJ, Canale RE, Blankenship MM, et al. Effect of ambrotose AO(R) on resting and exercise-induced antioxidant capacity and oxidative stress in healthy adults [abstract]. Nutr J 2010; 9: 49 167. Gaeini AA, Rahnama N, Hamedinia MR. Effects of vitamin E supplementation on oxidative stress at rest and after exercise to exhaustion in athletic students. J Sports Med Phys Fitness 2006; 46 (3): 458-61 168. Cholewa J, Poprzecki S, Zajac A, et al. The influence of vitamin C on blood oxidative stress parameters in basketball players in response to maximal exercise. Sci Sports 2008; 23 (3-4): 176-82 169. Nieman DC, Henson DA, McAnulty SR, et al. Influence of vitamin C supplementation on oxidative and immune changes after an ultramarathon. J Appl Physiol 2002; 92 (5): 1970-7 170. Braun B, Clarkson PM, Freedson PS, et al. Effects of coenzyme Q10 supplementation on exercise performance, VO2max, and lipid peroxidation in trained cyclists. Int J Sport Nutr 1991; 1 (4): 353-65 171. Vasankari T, Kujala U, Vasankari T, et al. Increased serum and low-density-lipoprotein antioxidant potential after antioxidant supplementation in endurance athletes. Am J Clin Nutr 1997; 65 (4): 1052-6 172. Meijer EP, Goris AH, Senden J, et al. Antioxidant supplementation and exercise-induced oxidative stress in the 60-year-old as measured by antipyrine hydroxylates. Br J Nutr 2001; 86 (5): 569-75
ª 2011 Adis Data Information BV. All rights reserved.
Peternelj & Coombes
173. McAnulty SR, McAnulty LS, Nieman DC, et al. Effect of alpha-tocopherol supplementation on plasma homocysteine and oxidative stress in highly trained athletes before and after exhaustive exercise. J Nutr Biochem 2005; 16 (9): 530-7 174. Lamprecht M, Hofmann P, Greilberger JF, et al. Increased lipid peroxidation in trained men after 2 weeks of antioxidant supplementation. Int J Sport Nutr Exerc Metab 2009; 19 (4): 385-99 175. Childs A, Jacobs C, Kaminski T, et al. Supplementation with vitamin C and N-acetyl-cysteine increases oxidative stress in humans after an acute muscle injury induced by eccentric exercise. Free Radic Biol Med 2001; 31 (6): 745-53 176. Nieman DC, Henson DA, McAnulty SR, et al. Vitamin E and immunity after the Kona Triathlon World Championship. Med Sci Sports Exerc 2004; 36 (8): 1328-35 177. Jakeman P, Maxwell S. Effect of antioxidant vitamin supplementation on muscle function after eccentric exercise. Eur J Appl Physiol Occup Physiol 1993; 67 (5): 426-30 178. Palazzetti S, Rousseau AS, Richard MJ, et al. Antioxidant supplementation preserves antioxidant response in physical training and low antioxidant intake. Br J Nutr 2004; 91 (1): 91-100 179. Nakhostin-Roohi B, Babaei P, Rahmani-Nia F, et al. Effect of vitamin C supplementation on lipid peroxidation, muscle damage and inflammation after 30-min exercise at 75% VO2max. J Sports Med Phys Fitness 2008; 48 (2): 217-24 180. Bloomer RJ, Goldfarb AH, McKenzie MJ, et al. Effects of antioxidant therapy in women exposed to eccentric exercise. Int J Sport Nutr Exerc Metab 2004; 14 (4): 377-88 181. Bryer SC, Goldfarb AH. Effect of high dose vitamin C supplementation on muscle soreness, damage, function, and oxidative stress to eccentric exercise. Int J Sport Nutr Exerc Metab 2006; 16 (3): 270-80 182. Nieman DC, Peters EM, Henson DA, et al. Influence of vitamin C supplementation on cytokine changes following an ultramarathon. J Interferon Cytokine Res 2000; 20 (11): 1029-35 183. Phillips T, Childs AC, Dreon DM, et al. A dietary supplement attenuates IL-6 and CRP after eccentric exercise in untrained males. Med Sci Sports Exerc 2003; 35 (12): 2032-7 184. Funes L, Carrera-Quintanar L, Cerda´n-Calero M, et al. Effect of lemon verbena supplementation on muscular damage markers, proinflammatory cytokines release and neutrophils’ oxidative stress in chronic exercise. Eur J Appl Physiol 2011; 111 (4): 695-705 185. Peters EM, Anderson R, Nieman DC, et al. Vitamin C supplementation attenuates the increases in circulating cortisol, adrenaline and anti-inflammatory polypeptides following ultramarathon running. Int J Sports Med 2001; 22 (07): 537-43 186. Senturk UK, Yalcin O, Gunduz F, et al. Effect of antioxidant vitamin treatment on the time course of hematological and hemorheological alterations after an exhausting exercise episode in human subjects. J Appl Physiol 2005; 98 (4): 1272-9 187. Shafat A, Butler P, Jensen RL, et al. Effects of dietary supplementation with vitamins C and E on muscle func-
Sports Med 2011; 41 (12)
Antioxidant Supplementation in Exercise Training
188.
189.
190.
191.
192.
193.
194.
195.
196.
197.
198.
199.
200.
201.
202.
203.
tion during and after eccentric contractions in humans. Eur J Appl Physiol 2004 Oct; 93 (1-2): 196-202 Matsumoto H, Takenami E, Iwasaki-Kurashige K, et al. Effects of blackcurrant anthocyanin intake on peripheral muscle circulation during typing work in humans. Eur J Appl Physiol 2005; 94 (1): 36-45 Mizuno K, Tanaka M, Nozaki S, et al. Antifatigue effects of coenzyme Q10 during physical fatigue. Nutrition 2008; 24 (4): 293-9 Matuszczak Y, Farid M, Jones J, et al. Effects of Nacetylcysteine on glutathione oxidation and fatigue during handgrip exercise. Muscle Nerve 2005; 32 (5): 633-8 Reid MB, Stokic´ DS, Koch SM, et al. N-acetylcysteine inhibits muscle fatigue in humans. J Clin Invest 1994; 94 (6): 2468-74 Mastaloudis A, Traber MG, Carstensen K, et al. Antioxidants did not prevent muscle damage in response to an ultramarathon run. Med Sci Sports Exerc 2006; 38 (1): 72-80 Dawson B, Henry GJ, Goodman C, et al. Effect of vitamin C and E supplementation on biochemical and ultrastructural indices of muscle damage after a 21 km run. Int J Sports Med 2002; 23 (1): 10-5 Traber MG. Relationship of vitamin E metabolism and oxidation in exercising human subjects. Br J Nutr 2006; 96 Suppl. S1: S34-7 Connolly DA, Lauzon C, Agnew J, et al. The effects of vitamin C supplementation on symptoms of delayed onset muscle soreness. J Sports Med Phys Fitness 2006; 46 (3): 462-7 Rahmani-Nia F, Talebi E, Nakhostin-Roohi B, et al. Effect of two regimes of vitamin C on delayed onset of muscle soreness [special issue]. J Mov Sci Sports 2008; 5 (1): 1-5 Close GL, Ashton T, Cable T, et al. Ascorbic acid supplementation does not attenuate post-exercise muscle soreness following muscle-damaging exercise but may delay the recovery process. Br J Nutr 2006; 95 (5): 976-81 Thompson D, Williams C, Garcia-Roves P, et al. Postexercise vitamin C supplementation and recovery from demanding exercise. Eur J Appl Physiol 2003 May; 89 (34): 393-400 Ganio MS, Armstrong LE, Johnson EC, et al. Effect of quercetin supplementation on maximal oxygen uptake in men and women. J Sports Sci 2010; 28 (2): 201-8 Mastaloudis A, Morrow JD, Hopkins DW, et al. Antioxidant supplementation prevents exercise-induced lipid peroxidation, but not inflammation, in ultramarathon runners. Free Radic Biol Med 2004; 36 (10): 1329-41 Nieman DC, Henson DA, Gross SJ, et al. Quercetin reduces illness but not immune perturbations after intensive exercise. Med Sci Sports Exerc 2007; 39 (9): 1561-9 Avery NG, Kaiser JL, Sharman MJ, et al. Effects of vitamin E supplementation on recovery from repeated bouts of resistance exercise. J Strength Cond Res 2003; 17 (4): 801-9 Sharman IM, Down MG, Sen RN. The effects of vitamin E and training on physiological function and athletic performance in adolescent swimmers. Br J Nutr 1971; 26 (2): 265-76
ª 2011 Adis Data Information BV. All rights reserved.
1067
204. Lawrence J, Bower R, Riehl W, et al. Effects of alphatocopherol acetate on the swimming endurance of trained swimmers. Am J Clin Nutr 1975; 28 (3): 205-8 205. Patil SM, Chaudhuri D, Dhanakshirur GB. Role of alphatocopherol in cardiopulmonary fitness in endurance athletes, cyclists. Ind J Physiol Pharmacol 2009; 53 (4): 375-9 206. Oostenbrug GS, Mensink RP, Hardeman MR, et al. Exercise performance, red blood cell deformability, and lipid peroxidation: effects of fish oil and vitamin E. J Appl Physiol 1997; 83 (3): 746-52 207. Buchman AL, Killip D, Ou CN, et al. Short-term vitamin E supplementation before marathon running: a placebocontrolled trial. Nutrition 1999; 15 (4): 278-83 208. Nalbant O, Toktas N, Toraman NF, et al. Vitamin E and aerobic exercise: effects on physical performance in older adults. Aging Clin Exp Res 2009; 21 (2): 111-21 209. Porter DA, Costill DL, Zachwieja JJ, et al. The effect of oral coenzyme Q10 on the exercise tolerance of middleaged, untrained men. Int J Sports Med 1995; 16 (7): 421-7 210. Zhou S, Zhang Y, Davie A, et al. Muscle and plasma coenzyme Q10 concentration, aerobic power and exercise economy of healthy men in response to four weeks of supplementation. J Sports Med Phys Fitness 2005; 45 (3): 337-46 211. Cureton KJ, Tomporowski PD, Singhal A, et al. Dietary quercetin supplementation is not ergogenic in untrained men. J Appl Physiol 2009; 107 (4): 1095-104 212. Dumke CL, Nieman DC, Utter AC, et al. Quercetin’s effect on cycling efficiency and substrate utilization. Appl Physiol Nutr Metab 2009; 34 (6): 993-1000 213. Ryan MJ, Jackson JR, Hao Y, et al. Suppression of oxidative stress by resveratrol after isometric contractions in gastrocnemius muscles of aged mice. J Gerontol A Biol Sci Med Sci 2010; 65 (8): 815-31 214. Marshall RJ, Scott KC, Hill RC, et al. Supplemental vitamin C appears to slow racing greyhounds. J Nutr 2002 Jun; 132 (6 Suppl. 2): S1616-21 215. Weight LM, Myburgh KH, Noakes TD. Vitamin and mineral supplementation: effect on the running performance of trained athletes. Am J Clin Nutr 1988; 47 (2): 192-5 216. McAnulty SR, Nieman DC, Fox-Rabinovich M, et al. Effect of n-3 fatty acids and antioxidants on oxidative stress after exercise. Med Sci Sports Exerc 2010; 42 (9): 1704-11 217. Nielsen AN, Mizuno M, Ratkevicius A, et al. No effect of antioxidant supplementation in triathletes on maximal oxygen uptake, 31P-NMRS detected muscle energy metabolism and muscle fatigue. Int J Sports Med 1999; 20 (3): 154-8 218. Snider IP, Bazzarre TL, Murdoch SD, et al. Effects of coenzyme athletic performance system as an ergogenic aid on endurance performance to exhaustion. Int J Sport Nutr 1992; 2 (3): 272-86 219. Arent SM, Pellegrino JK, Williams CA, et al. Nutritional supplementation, performance, and oxidative stress in college soccer players. J Strength Cond Res 2010; 24 (4): 1117-24 220. Fry AC, Bloomer RJ, Falvo MJ, et al. Effect of a liquid multivitamin/mineral supplement on anaerobic exercise performance. Res Sports Med 2006; 14 (1): 53-64
Sports Med 2011; 41 (12)
Peternelj & Coombes
1068
221. Knechtle B, Knechtle P, Schulze I, et al. Vitamins, minerals and race performance in ultra-endurance runners: Deutschlandlauf 2006. Asia Pac J Clin Nutr 2008; 17 (2): 194-8 222. Yfanti C, Akerstrom T, Nielsen S, et al. Antioxidant supplementation does not alter endurance training adaptation. Med Sci Sports Exerc 2010; 42 (7): 1388-95 223. Ylikoski T, Piirainen J, Hanninen O, et al. The effect of coenzyme Q10 on the exercise performance of crosscountry skiers. Mol Aspects Med 1997; 18 Suppl.: S283-90 224. Bonetti A, Solito F, Carmosino G, et al. Effect of ubidecarenone oral treatment on aerobic power in middle-aged trained subjects. J Sports Med Phys Fitness 2000; 40 (1): 51-7 225. Cooke M, Iosia M, Buford T, et al. Effects of acute and 14day coenzyme Q10 supplementation on exercise performance in both trained and untrained individuals [abstract]. J Int Soc Sports Nutr 2008; 5: 8 226. Gokbel H, Gul I, Belviranl M, et al. The effects of coenzyme Q10 supplementation on performance during repeated bouts of supramaximal exercise in sedentary men. J Strength Cond Res 2010; 24 (1): 97-102 227. Novelli GP, Bracciotti G, Falsini S. Spin-trappers and vitamin E prolong endurance to muscle fatigue in mice. Free Radic Biol Med 1990; 8 (1): 9-13 228. Asha Devi S, Prathima S, Subramanyam MVV. Dietary vitamin E and physical exercise: I. Altered endurance capacity and plasma lipid profile in ageing rats. Exp Gerontol 2003; 38 (3): 285-90 229. Piercy RJ, Hinchcliff KW, Morley PS, et al. Association between vitamin E and enhanced athletic performance in sled dogs. Med Sci Sports Exerc 2001; 33 (5): 826-33 230. Hoogerwerf A, Hoitink A. The influence of vitamin C administration on the mechanical efficiency of the human organism. Eur J Appl Physiol Occup Physiol 1963; 20 (2): 164-72 231. Howald H, Segesser B, Ko¨rner WF. Ascorbic acid and athletic performance. Ann N Y Acad Sci 1975; 258 (1): 458-64 232. Aguilo A, Tauler P, Sureda A, et al. Antioxidant diet supplementation enhances aerobic performance in amateur sportsmen. J Sports Sci 2007; 25 (11): 1203-10 233. Jourkesh M, Ostojic SM, Azarbayjani MA. The effects of vitamin E and vitamin C supplementation on bioenergetics index. Res Sports Med 2007; 15 (4): 249-56 234. Louis J, Hausswirth C, Bieuzen F, et al. Vitamin and mineral supplementation effect on muscular activity and cycling efficiency in master athletes. Appl Physiol Nutr Metab 2010; 35 (3): 251-60 235. Medved I, Brown MJ, Bjorksten AR, et al. Nacetylcysteine infusion alters blood redox status but not time to fatigue during intense exercise in humans. J Appl Physiol 2003; 94 (4): 1572-82 236. Medved I, Brown MJ, Bjorksten AR, et al. Effects of intravenous N-acetylcysteine infusion on time to fatigue and potassium regulation during prolonged cycling exercise. J Appl Physiol 2004; 96 (1): 211-7 237. Medved I, Brown MJ, Bjorksten AR, et al. Nacetylcysteine enhances muscle cysteine and glutathione availability and attenuates fatigue during prolonged ex-
ª 2011 Adis Data Information BV. All rights reserved.
238.
239.
240.
241.
242.
243.
244.
245.
246.
247.
248.
249.
250.
251.
252.
ercise in endurance-trained individuals. J Appl Physiol 2004; 97 (4): 1477-85 Davis JM, Carlstedt CJ, Chen S, et al. The dietary flavonoid quercetin increases VO(2max) and endurance capacity. Int J Sport Nutr Exerc Metab 2010; 20 (1): 56-62 Davis JM, Murphy EA, Carmichael MD, et al. Quercetin increases brain and muscle mitochondrial biogenesis and exercise tolerance. Am J Physiol Regul Integr Comp Physiol 2009; 296 (4): R1071-7 MacRae HS, Mefferd KM. Dietary antioxidant supplementation combined with quercetin improves cycling time trial performance. Int J Sport Nutr Exerc Metab 2006; 16 (4): 405-19 Lagouge M, Argmann C, Gerhart-Hines Z, et al. Resveratrol improves mitochondrial function and protects against metabolic disease by activating SIRT1 and PGC1[alpha]. Cell 2006; 127 (6): 1109-22 Bailey SJ, Fulford J, Vanhatalo A, et al. Dietary nitrate supplementation enhances muscle contractile efficiency during knee-extensor exercise in humans. J Appl Physiol 2010; 109 (1): 135-48 Bailey SJ, Winyard P, Vanhatalo A, et al. Dietary nitrate supplementation reduces the O2 cost of low-intensity exercise and enhances tolerance to high-intensity exercise in humans. J Appl Physiol 2009; 107 (4): 1144-55 Vanhatalo A, Bailey SJ, Blackwell JR, et al. Acute and chronic effects of dietary nitrate supplementation on blood pressure and the physiological responses to moderate-intensity and incremental exercise. Am J Physiol Regul Integr Comp Physiol 2010; 299 (4): R1121-31 Lansley KE, Winyard PG, Fulford J, et al. Dietary nitrate supplementation reduces the O2 cost of walking and running: a placebo-controlled study. J Appl Physiol 2011; 110 (3): 591-600 Skarpanska-Stejnborn A, Pilaczynska-Szczesniak L, Basta P, et al. The influence of supplementation with Rhodiola rosea L. extract on selected redox parameters in professional rowers. Int J Sport Nutr Exerc Metab 2009; 19 (2): 186-99 Oh JK, Shin YO, Yoon JH, et al. Effect of supplementation with Ecklonia cava polyphenol on endurance performance of college students. Int J Sport Nutr Exerc Metab 2010; 20 (1): 72-9 Vauzour D, Rodriguez-Mateos A, Corona G, et al. Polyphenols and human health: prevention of disease and mechanisms of action. Nutrients 2010; 2 (11): 1106-31 Malm C, Svensson M, Sjoberg B, et al. Supplementation with ubiquinone-10 causes cellular damage during intense exercise. Acta Physiol Scand 1996; 157 (4): 511-2 Malm C, Svensson M, Ekblom B, et al. Effects of ubiquinone-10 supplementation and high intensity training on physical performance in humans. Acta Physiol Scand 1997; 161 (3): 379-84 Coombes JS, Powers SK, Rowell B, et al. Effects of vitamin E and alpha-lipoic acid on skeletal muscle contractile properties. J Appl Physiol 2001; 90 (4): 1424-30 Gomez-Cabrera MC, Borras C, Pallardo FV, et al. Decreasing xanthine oxidase-mediated oxidative stress prevents useful cellular adaptations to exercise in rats. J Physiol 2005; 567 (Pt 1): 113-20
Sports Med 2011; 41 (12)
Antioxidant Supplementation in Exercise Training
253. Gomez-Cabrera MC, Martinez A, Santangelo G, et al. Oxidative stress in marathon runners: interest of antioxidant supplementation. Br J Nutr 2006; 96 Suppl. 1: S31-3 254. Richardson RS, Donato AJ, Uberoi A, et al. Exerciseinduced brachial artery vasodilation: role of free radicals. Am J Physiol Heart Circ Physiol 2007; 292 (3): H1516-22 255. Copp SW, Ferreira LF, Herspring KF, et al. The effects of antioxidants on microvascular oxygenation and blood flow in skeletal muscle of young rats. Exp Physiol 2009; 94 (9): 961-71 256. Wray DW, Uberoi A, Lawrenson L, et al. Oral antioxidants and cardiovascular health in the exercise-trained and untrained elderly: a radically different outcome. Clin Sci (Lond) 2009; 116 (5): 433-41 257. Matsumoto A, Mason SR, Flatscher-Bader T, et al. Effects of exercise and antioxidant supplementation on endo-
ª 2011 Adis Data Information BV. All rights reserved.
1069
thelial gene expression. Int J Cardiol 2011. Epub 2011 Feb 3 258. McArdle F, Spiers S, Aldemir H, et al. Preconditioning of skeletal muscle against contraction-induced damage: the role of adaptations to oxidants in mice. J Physiol 2004; 561 (Pt 1): 233-44 259. Teixeira A, Muller L, Santos AA, et al. Beneficial effects of gradual intense exercise in tissues of rats fed with a diet deficient in vitamins and minerals: a pilot study. Nutrition 2009; 25 (5): 590-6
Correspondence: Tina-Tinkara Peternelj, School of Human Movement Studies, University of Queensland, St Lucia, Brisbane, QLD, 4072, Australia. E-mail:
[email protected]
Sports Med 2011; 41 (12)