Portal de Periódicos da CAPES

Separating the Contenders From the Pretenders

2010; Lippincott Williams & Wilkins; Volume: 121; Issue: 22 Linguagem: Inglês

10.1161/circulationaha.110.953463

1524-4539

James A. de Lemos, Anand Rohatgi,

Advanced Causal Inference Techniques

HomeCirculationVol. 121, No. 22Separating the Contenders From the Pretenders Free AccessEditorialPDF/EPUBAboutView PDFView EPUBSections ToolsAdd to favoritesDownload citationsTrack citationsPermissions ShareShare onFacebookTwitterLinked InMendeleyReddit Jump toFree AccessEditorialPDF/EPUBSeparating the Contenders From the PretendersCompetitive High-Throughput Biomarker Screening in Large Population-Based Studies James A. de Lemos, MD and Anand Rohatgi, MD James A. de LemosJames A. de Lemos From the Cardiology Division and Department of Medicine, University of Texas Southwestern Medical Center, Dallas, Tex. and Anand RohatgiAnand Rohatgi From the Cardiology Division and Department of Medicine, University of Texas Southwestern Medical Center, Dallas, Tex. Originally published24 May 2010https://doi.org/10.1161/CIRCULATIONAHA.110.953463Circulation. 2010;121:2381–2383Other version(s) of this articleYou are viewing the most recent version of this article. Previous versions: May 24, 2010: Previous Version 1 Despite great enthusiasm for biomarkers as tools to enhance risk prediction and to lead the way in a transformation towards personalized cardiovascular medicine, progress in the biomarker field has been painstakingly slow, particularly in the area of population screening. Some individual biomarkers such as C-reactive protein (CRP) have demonstrated consistent associations with incident cardiovascular events across multiple studies, but the magnitude of these associations is modest,1 and only small improvements in discrimination and reclassification are seen.2,3 One attractive solution to the limitations of individual biomarkers is to combine nonredundant biomarkers into panels to enhance risk assessment. However, results of studies testing multiple biomarkers for risk prediction in primary prevention populations have not provided a clear picture, with some studies showing qualified promise4–6 and others suggesting limited value.2,7,8Article see p 2388In this issue of Circulation, Blankenberg et al9 report findings from an impressive large-scale effort to screen multiple biomarkers for risk assessment in the general population. The investigators evaluated 30 biomarkers reflecting different components of the pathophysiology of atherosclerotic heart disease in 2 cohorts totaling almost 10 000 individuals. Unfortunately, most of the novel biomarkers tested either did not associate with incident cardiovascular events at all or showed promising results in 1 cohort that did not replicate. Although the novel biomarkers largely failed, consistent, albeit modest, associations with incident events were seen across the 2 cohorts for CRP, N-terminal pro B-type natriuretic peptide (NT-proBNP), and cardiac troponin (cTn) I, 3 well-established cardiovascular biomarkers. A weighted biomarker score was developed in men from one of the cohorts (FINRISK 97) and then validated in a second cohort of men (PRIME Belfast). Adding the biomarker score to a standard risk factor model improved multiple metrics of risk prediction, including measures of association, discrimination, and most important from a clinical standpoint, correct reclassification of individuals to higher or lower risk categories. Because soft end points of unstable angina and coronary revascularization were included in the composite end point, the categories used for the net reclassification index analyses do not correlate directly with typical risk categories based on 10-year rates of hard coronary events. With this caveat, the biomarker score did improve correct reclassification across the entire PRIME Belfast cohort as well as among those in the intermediate risk category. Findings were similar when data-derived cut points were used for each biomarker.9This study has a number of strengths: It represents one of the largest biomarker projects yet reported in cardiovascular medicine; the investigators developed and validated their biomarker panels in separate data sets with sufficient numbers of cardiac end points; and they used comprehensive and state-of-the-art metrics for evaluating the contributions of the biomarkers to standard risk prediction variables. As would be expected given the "real-world" observational design, the huge numbers of data points, and the complexity of the statistical analyses, there are several potential problems as well. A substantial number of individuals were missing data for at least 1 biomarker, and the authors used multiple imputation techniques to estimate the missing values. This strategy assumes that the biomarkers are "predictable" on the basis of other data collected from the individual. However, this argument has an inherent flaw in that the biomarkers were selected because they offered additional information not contained in the other variables. Reassuringly, the authors do note that results were similar in sensitivity analyses using only individuals with complete biomarker data.A more important issue to consider is that the reclassification analyses were performed only in the PRIME Belfast cohort, which may provide an overly optimistic impression about the potential clinical value of the biomarkers. The baseline risk factor model did not perform as well in PRIME men compared with FINRISK men and women, almost certainly because PRIME enrolled a much narrower age range of individuals, mitigating the dominant effect of age in the standard risk factor models. Clearly, the PRIME Belfast cohort offered an easier "victory" for the biomarker panel compared with the FINRISK men or women. Indeed, it is notable that in prior large population studies evaluating multiple biomarkers, favorable conclusions have been reported from studies enrolling individuals with a restricted age range,4,5 whereas those enrolling a broader age range have typically shown less favorable results.7,8 In a recent quantitative review of 79 articles evaluating the incremental performance of a new predictor over the Framingham Risk Score alone, a strong inverse association was observed between the performance of the baseline Framingham Risk Score model and the increment in the C statistic reported with the new predictor.10We believe, however, that a reasonable case can be made for the authors' approach of evaluating biomarkers in sex-specific cohorts with restricted age ranges. First, this approach helps deal with the dominant effects of age and, to a lesser effect, sex, which marginalize other variables in standard risk factor models. Second, this strategy provides clinically relevant information for the age and sex groups tested. Of course, the results cannot be generalized to other age ranges (or race and sex groups). It would have been instructive to also perform the reclassification analyses in the more heterogeneous FINRISK cohort because this would have likely highlighted the complicated influences of age and sex on the interpretation of results of these studies.What are the implications for the winners and losers from this biomarker competition? First, although the international debate on the utility of biomarkers for population screening has long focused on CRP, it is now time to broaden consideration to include other biomarkers that appear to provide predictive information at least as robust as CRP in this setting. For example, although the incremental value of adding NT-proBNP to CRP was not directly assessed in the present study, as an individual biomarker, NT-proBNP displayed performance characteristics similar to CRP in terms of discrimination and reclassification. In previous population-based studies that directly compared NT-proBNP or BNP with CRP for risk estimation, the natriuretic peptides typically show greater prognostic value.4,7,8,11,12 At this point, however, the diagnostic and therapeutic implications of elevated natriuretic peptide levels in the general population have not been clearly defined.These findings also require that we reconsider long-held notions about where and when cTn measurement may be informative. It is increasingly clear that cTnT and cTnI may be released in response to chronic as well as acute cardiac injury and may be elevated in asymptomatic individuals at increased cardiovascular risk.4,5,13 In distinction to the acute care setting, it appears that elevated troponin levels in ambulatory subjects reflect cardiac structural and functional abnormalities more than ischemic risk13 and, like NT-proBNP, are particularly informative for death and heart failure events.5,14 In this study, cTnI was measured with a contemporary sensitive assay but not an emerging higher-sensitivity assay. Higher-sensitivity assays may allow even greater risk discrimination among apparently healthy individuals.Although the study represents a qualified victory for multiple biomarker panels that include CRP, NT-proBNP, and cTnI, does it represent the end of the road for the other biomarkers that were tested and failed? More important, does the failure here and in prior studies of the more novel biomarkers suggest that biomarker discovery in this area is likely to be futile? We believe such a conclusion would be premature. First, characteristics of the cohort studied, including size and demographic makeup, can significantly influence biomarker performance. In the present study, for example, differences between subjects enrolled in the 2 cohorts may have contributed to the failure of some potentially informative biomarkers to replicate. Second, biomarker selection may also vary depending on how the standard risk factors are modeled, with subtle differences having potentially important implications.10 Third, in our experience with similar data sets, selection and prioritization of biomarkers for inclusion in the multiple biomarker panels are sensitive to several statistical considerations, including degree of collinearity between biomarkers, analysis of biomarkers as continuous versus categorical covariates, and choice of selection and optimization criteria (ie, based on strength of association, indices of model fit, C statistic, or integrated discrimination index). Finally, statistical power is an important issue to consider, particularly when evaluating large numbers of covariates. In the present study, soft end points such as unstable angina and revascularization were included in the composite, and actual numbers of hard coronary end points were modest relative to the numbers of covariates evaluated. A careful review of the data from the present study suggests that each of these factors may have contributed to the exclusion of some biomarkers from further consideration. Although the present study is one of few to report multiple performance characteristics on a host of novel markers, it is clear that much work is needed to define optimal methods for biomarker screening, selection, optimization, and validation.Progress forward requires movement in several directions. For the established biomarkers, further clinical validation of panels containing CRP, NT-proBNP (or BNP), and a sensitive troponin assay is required in different age, race, and sex groups, given the known influences of these factors on levels of these biomarkers. We also need carefully designed observational and interventional studies to help us understand the full implications of reclassification based on these biomarkers. In particular, it is critical to establish the safety of deferral of preventive therapies for individuals who are reclassified to a lower risk category by biomarkers. With regard to the more novel biomarkers, careful thought is needed with regard to the appropriate target populations for discovery and validation, as well as the criteria used to sort out the contenders from the pretenders.The opinions expressed in this article are not necessarily those of the editors or of the American Heart Association.DisclosuresDr de Lemos has received grant support from Biosite Inc and Roche Diagnostics and consulting income from Tethys Biomedical and Johnson and Johnson. Dr Rohatgi reports no conflicts.FootnotesCorrespondence to James A. de Lemos, MD, UT Southwestern Medical Center, 5909 Harry Hines Blvd, HA 9.133, Dallas, TX 75390-9047. E-mail [email protected] References 1 Kaptoge S, Di Angelantonio E, Lowe G, Pepys MB, Thompson SG, Collins R, Danesh J. C-reactive protein concentration and risk of coronary heart disease, stroke, and mortality: an individual participant meta-analysis. Lancet. 2010; 375: 132–140.CrossrefMedlineGoogle Scholar2 Folsom AR, Chambless LE, Ballantyne CM, Coresh J, Heiss G, Wu KK, Boerwinkle E, Mosley TH Jr, Sorlie P, Diao G, Sharrett AR. An assessment of incremental coronary risk prediction using c-reactive protein and other novel risk markers: the Atherosclerosis Risk in Communities Study. Arch Intern Med. 2006; 166: 1368–1373.CrossrefMedlineGoogle Scholar3 Shah T, Casas JP, Cooper JA, Tzoulaki I, Sofat R, McCormack V, Smeeth L, Deanfield JE, Lowe GD, Rumley A, Fowkes FG, Humphries SE, Hingorani AD. Critical appraisal of CRP measurement for the prediction of coronary heart disease events: new data and systematic review of 31 prospective cohorts. Int J Epidemiol. 2009; 38: 217–231.CrossrefMedlineGoogle Scholar4 Zethelius B, Berglund L, Sundstrom J, Ingelsson E, Basu S, Larsson A, Venge P, Arnlov J. Use of multiple biomarkers to improve the prediction of death from cardiovascular causes. N Engl J Med. 2008; 358: 2107–2116.CrossrefMedlineGoogle Scholar5 Daniels LB, Laughlin GA, Clopton P, Maisel AS, Barrett-Connor E. Minimally elevated cardiac troponin t and elevated n-terminal pro-b-type natriuretic peptide predict mortality in older adults: results from the Rancho Bernardo study. J Am Coll Cardiol. 2008; 52: 450–459.CrossrefMedlineGoogle Scholar6 Nordestgaard BG, Adourian AS, Freiberg JJ, Guo Y, Muntendam P, Falk E. Risk factors for near-term myocardial infarction in apparently healthy men and women. Clin Chem. 2010; 56: 559–567.CrossrefMedlineGoogle Scholar7 Wang TJ, Gona P, Larson MG, Tofler GH, Levy D, Newton-Cheh C, Jacques PF, Rifai N, Selhub J, Robins SJ, Benjamin EJ, D'Agostino RB, Vasan RS. Multiple biomarkers for the prediction of first major cardiovascular events and death. N Engl J Med. 2006; 355: 2631–2639.CrossrefMedlineGoogle Scholar8 Melander O, Newton-Cheh C, Almgren P, Hedblad B, Berglund G, Engstrom G, Persson M, Smith JG, Magnusson M, Christensson A, Struck J, Morgenthaler NG, Bergmann A, Pencina MJ, Wang TJ. Novel and conventional biomarkers for prediction of incident cardiovascular events in the community. JAMA. 2009; 302: 49–57.CrossrefMedlineGoogle Scholar9 Blankenberg S, Zeller T, Saarela O, Havulinna AS, Kee F, Tunstall-Pedoe H, Kuulasmaa K, Yarnell J, Schnabel RB, Wild PS, Münzel TF, Lackner KJ, Tiret L, Evans A, Salomaa V, for the MORGAM Project. Contribution of 30 biomarkers to 10-year cardiovascular risk estimation in 2 population cohorts: The MONICA, Risk, Genetics, Archiving, and Monograph (MORGAM) Biomarker Project. Circulation. 2010; 121: 2388–2397.LinkGoogle Scholar10 Tzoulaki I, Liberopoulos G, Ioannidis JP. Assessment of claims of improved prediction beyond the Framingham risk score. JAMA. 2009; 302: 2345–2352.CrossrefMedlineGoogle Scholar11 Olsen MH, Hansen TW, Christensen MK, Gustafsson F, Rasmussen S, Wachtell K, Ibsen H, Torp-Pedersen C, Hildebrandt PR. N-terminal pro-brain natriuretic peptide, but not high sensitivity c-reactive protein, improves cardiovascular risk prediction in the general population. Eur Heart J. 2007; 28: 1374–1381.CrossrefMedlineGoogle Scholar12 Kistorp C, Raymond I, Pedersen F, Gustafsson F, Faber J, Hildebrandt P. N-terminal pro-brain natriuretic peptide, C-reactive protein, and urinary albumin levels as predictors of mortality and cardiovascular events in older adults. JAMA. 2005; 293: 1609–1616.CrossrefMedlineGoogle Scholar13 Wallace TW, Abdullah SM, Drazner MH, Das SR, Khera A, McGuire DK, Wians F, Sabatine MS, Morrow DA, de Lemos JA. Prevalence and determinants of troponin t elevation in the general population. Circulation. 2006; 113: 1958–1965.LinkGoogle Scholar14 Omland T, de Lemos JA, Sabatine MS, Christophi CA, Rice MM, Jablonski KA, Tjora S, Domanski MJ, Gersh BJ, Rouleau JL, Pfeffer MA, Braunwald E. A sensitive cardiac troponin T assay in stable coronary artery disease. N Engl J Med. 2009; 361: 2538–2547.CrossrefMedlineGoogle Scholar Previous Back to top Next FiguresReferencesRelatedDetailsCited By Metelskaya V, Gavrilova N, Zhatkina M, Yarovaya E and Drapkina O (2022) A Novel Integrated Biomarker for Evaluation of Risk and Severity of Coronary Atherosclerosis, and Its Validation, Journal of Personalized Medicine, 10.3390/jpm12020206, 12:2, (206) Metelskaya V, Zhatkina M, Gavrilova N, Yarovaya E, Bogdanova N, Litinskaya O, Pokrovskaya M and Drapkina O (2020) Validation of combined markers of the presence and severity of coronary atherosclerosis, Profilakticheskaya meditsina, 10.17116/profmed20202306265, 23:6, (65), . Metelskaya V, Zhatkina M, Gavrilova N, Yarovaya E, Bogdanova N, Litinskaya O, Pokrovskaya M and Drapkina O (2020) Validation of combined markers of the presence and severity of coronary atherosclerosis, Profilakticheskaya meditsina, 10.17116/profmed20202306194, 23:6, (94), . Hughes M, Saarela O, Blankenberg S, Zeller T, Havulinna A, Kuulasmaa K, Yarnell J, Schnabel R, Tiret L, Salomaa V, Evans A and Kee F (2011) A multiple biomarker risk score for guiding clinical decisions using a decision curve approach, European Journal of Preventive Cardiology, 10.1177/1741826711417341, 19:4, (874-884), Online publication date: 1-Aug-2012. Wang T (2011) Assessing the Role of Circulating, Genetic, and Imaging Biomarkers in Cardiovascular Risk Prediction, Circulation, 123:5, (551-565), Online publication date: 8-Feb-2011. Enriquez J and De Lemos J (2010) Should We Focus on Novel Risk Markers and Screening Tests to Better Predict and Prevent Cardiovascular Disease?, Preventive Cardiology, 10.1111/j.1751-7141.2010.00080.x, 13:4, (152-159), Online publication date: 1-Sep-2010. June 8, 2010Vol 121, Issue 22 Advertisement Article InformationMetrics https://doi.org/10.1161/CIRCULATIONAHA.110.953463PMID: 20497975 Originally publishedMay 24, 2010 KeywordstroponinEditorialsbiomarkersbrain natriuretic peptideC-reactive proteinPDF download Advertisement SubjectsEpidemiologyPrimary Prevention