A Best Evidence Interview With Eric J. Topol, MD
The Best Evidence Study
Dr. Topol is senior author of:
Rosenberg S, Elashoff MR, Beineke B, et al; PREDICT (Personalized Risk Evaluation and Diagnosis in the Coronary Tree) Investigators. Multicenter validation of the diagnostic accuracy of a blood-based gene expression test for assessing obstructive coronary artery disease in nondiabetic patients. Ann Intern Med. 2010;153:425-434.
About the Interviewee
Eric J. Topol, MD, is the Director of the Scripps Translational Science Institute, a National Institutes of Health-funded program of the Clinical and Translational Science Award (CTSA) Consortium. In 2009, he was named the Gary and Mary West Endowed Chair of Innovative Medicine. He is Professor of Translational Genomics at The Scripps Research Institute, the Chief Academic Officer of Scripps Health, and a senior consultant cardiologist practitioner at Scripps Clinic. Prior to coming to Scripps, he served on the faculty of Case Western as a professor in genetics, chaired the Department of Cardiovascular Medicine at Cleveland Clinic for 15 years and raised its status to rank #1 by US News and World Report for 11 consecutive years, and founded the Cleveland Clinic Lerner College of Medicine. His work in the genomics of heart attack has led to discovery of key genes (MEF2A deletion, THSB-4 variant), which led to recognition by the American Heart Association as top 10 research advances in 2002 and 2004. As a leader in clinical trials of novel therapeutics, he administered recombinant tPA to the first patient in 1984 and pioneered and led the clinical development of clopidogrel (Plavix), bivalirudin (Angiomax), and abciximab (ReoPro). He has over 1000 original peer-reviewed publications and has edited over 30 books, including the Textbook of Interventional Cardiology (5th ed), and the Textbook of Cardiovascular Medicine (3rd ed). Dr. Topol is also Editor-in-Chief of Medscape's Genomic Medicine Resource Center (http://medscape.com/resource/genomic-medicine).
Dr. Topol has been elected to the Institute of Medicine of the National Academy of Sciences, the American Association of Physicians, the American Society of Clinical Investigation, and the Johns Hopkins Society of Scholars. He has been recognized by the Institute of Scientific Information to be among the top 10 cited biomedical researchers in medicine in the past decade.
Introduction to the Interview
Conventional (catheter-based) coronary angiography, together with noninvasive methods that also expose patients to radiation and contrast agents, remains the gold standard for diagnosis of obstructive coronary artery disease (CAD). All these methods have been shown to carry some risk for misdiagnosis, so a noninvasive, accurate test for CAD would be welcomed as an alternative clinical tool. A gene expression test that involves a set of genes identified as differentially expressed between diseased and normal arterial tissue has been developed and has been available in the United States since 2009. The developers and manufacturers of the Corus™ CAD test (CardioDX®, Palo Alto, California) believe it could have clinical advantages over current noninvasive CAD diagnostic methods because it only involves taking a standard blood sample and does not require radiation, intravenous contrast agents, or physiologic or pharmacologic stressors. The test procedure measures expression of mRNA by 23 genes from the blood sample, and based on test results, increased or decreased RNA levels can be used in determining whether an individual patient has obstructive CAD, which is defined here as having at least 1 atherosclerotic plaque causing ≥ 50% luminal diameter stenosis in a major coronary artery (≥ 1.5 mm lumen diameter) as determined by coronary angiography. The test is gender-specific, with some of the genes in the test characterized as being specific to gender and others weighted differently by gender. Age can also be weighted differently by gender. A test score is derived from an algorithm that incorporates expression of the 23 genes and other patient demographics shown to be related to inflammation of the coronary arteries. The score range is from 0 to 40, with higher scores corresponding to a greater risk for obstructive CAD and a higher maximum percentage of stenosis. Corus CAD is the first and, to date, only clinically validated blood-based test for obstructive CAD. The test is intended for use in nondiabetic stable patients at clinical suspicion of CAD. The test is not intended for patients with known prior myocardial infarction or who have had a revascularization procedure; patients aged < 21 years or > 99 years; patients with diabetes; or patients who are asymptomatic and not considered at high risk for CAD.
Development of the test was carried out in 3 phases. Genes reflecting the presence of obstructive CAD were identified in over 200 nondiabetic patients. Quantitative real-time polymerase chain reaction (qRT-PCR) technology was used to further test and refine this gene set, and samples from over 600 patients were used to develop an algorithm integrating the expression levels of 23 genes and other patient characteristics. The algorithm was validated in a prospective, multicenter clinical trial, the Personalized Risk Evaluation and Diagnosis In the Coronary Tree (PREDICT) trial, of which Dr. Topol was principal investigator. PREDICT involved 526 nondiabetic patients enrolled at 39 US clinical sites between July 2007 and April 2009. All patients had been referred for elective invasive coronary angiography, and based on these results the patients were classified as cases or controls according to whether or not they had obstructive CAD. Blood samples were collected from the patients before coronary angiography and RNA purification, and RT-PCR analysis of gene expression was carried out by technicians, who were blinded as to the clinical status of the patients.
Preliminary data comparing the results of the coronary angiograms with the gene expression data were presented at the 2009 annual scientific sessions of the American Heart Association. The final results in the validation cohort were published in the Annals of Internal Medicine. The primary endpoint of the study, the area under the receiver-operating characteristic (ROC) curve, was 0.70 ± 0.02 (P < .001). Based on ROC analysis, addition of the algorithm significantly improved performance by the Diamond-Forrester method, a score comprising age, sex, and chest pain type, but showed only modest improvement over an expanded model incorporating 11 clinical factors such as medication use, blood pressure, and lipid levels, as well as age, sex, and chest pain type. In patients with an algorithm score of 14.75, corresponding to a 20% likelihood of having obstructive CAD, the sensitivity of the test was 85% (corresponding to a negative predictive value of 83%). Using reclassification of disease status, a more clinically relevant measure of comparative predictor performance than measures such as the ROC curve area according to the PREDICT investigators, the algorithm showed a net improvement over both the Diamond-Forrester method (20%) and the expanded clinical model (16%; P < .001 for both comparisons).
In an editorial published alongside the PREDICT results, Donna K. Arnett, MSPH, PhD (University of Alabama at Birmingham) urged caution in interpreting the PREDICT results because "it is unclear how the reclassification tests would perform in other populations with lower disease prevalence. Translation of the use of the test in a real clinical setting is also unclear." Dr. Arnett believes that the algorithm is "an initial proof-of-concept for the potential use of genetic risk scores in the context of cardiovascular disease," and a "first step in what may prove to be a long but hopefully rewarding journey," but that "it does not fulfill the tough evidentiary standards that modern clinicians should demand." She concluded, "Although we celebrate this innovative work, we must not exaggerate its immediate clinical value."
Dr. Topol spoke with Linda Brookes, MSc, for Medscape Cardiology, to discuss some of the implications of these results for Medscape's readers.