Core Laboratory Finds Significantly Less Coronary Artery Disease on CTA
Wednesday, Nov. 29, 2017
Readers at a blinded central core laboratory found significantly less coronary artery disease on coronary CT angiography (CTA) than clinical site radiologists and cardiologists, suggesting an opportunity to reduce follow-up testing, according to research presented Monday at RSNA 2017 and published in the online edition of the journal Radiology.
Detection of significant coronary artery disease on CTA yields important prognostic information that may trigger downstream testing and coronary revascularization. The PROspective Multicenter Imaging Study for Evaluation of chest pain, or PROMISE, a trial of coronary CTA at 193 North American sites, found that CTA identified more coronary artery disease and led to 50 percent more coronary catheterization than functional testing strategies such as stress echocardiography and nuclear stress testing. However, the greater number of coronary catheterizations in the CTA arm did not reduce heart attacks or death.
"The higher rate of positive test results and downstream testing is a barrier to adoption of coronary CTA," said Michael T. Lu, MD, MPH, director of research in cardiovascular imaging at Massachusetts General Hospital and assistant professor of radiology at Harvard Medical School, both in Boston.
In the new study, Dr. Lu and colleagues compared coronary CTA interpretations by six readers at the PROMISE central core laboratory with those of the 582 clinical site readers for significant coronary artery disease — defined as 50 percent or more blockage in the artery — in 4,347 PROMISE subjects. Core laboratory readers were blinded to all clinical information; site readers interpreted CTA as part of the standard clinical workup for chest pain.
Results between the two groups were substantially different. The core laboratory found 41 percent less significant coronary artery disease than clinical site readers. Despite identifying less disease, the core laboratory had similar predictive power for heart attacks and cardiovascular death.
There are a number of potential explanations for why site readers called more significant disease than the core laboratory, Dr. Lu said. Chief among these are U.S. clinical practice incentives that punish false negatives while accepting a high rate of false positives. Access to the patient's medical record may also paradoxically lead to more positive findings, as current cardiovascular risk calculators overestimate actual risk.
"As a clinical radiologist who reads coronary CTA, my key takeaway is that systematic factors lead readers to call more disease," Dr. Lu added. "Our results suggest we can safely call less, which would address a major limitation of coronary CTA and reduce downstream testing for our patients."
In a commentary following Dr. Lu's presentation, Jonathon Leipsic, MD, associate professor of radiology and cardiology at the University of British Columbia in Vancouver, noted that readers in the clinic face frequent interruptions and time constraints that are largely absent from the core laboratory. U.S. medicolegal pressures also bias clinical readers toward calling more disease, he said, noting that the more conservative medical management approach is often preferred in the less litigious environments of Canada and the United Kingdom. Possible solutions to the overcalling of disease, he said, include the use of fractional flow reserve derived from CT and the employment of deep learning algorithms to help remove human bias, potentially reducing unnecessary follow-up tests.