Cardiac CT Angiography: Broken Promises
Straight Talk | Steven E. Nissen, MD, MACC, and Paul C. Cremer, MD
During the ACC 2015 Annual Scientific Session, we awoke to headlines in the daily newspaper highlighting the Prospective Multicenter Imaging Study for Evaluation of Chest Pain (PROMISE) trial, which demonstrated that coronary CT angiography (CTA) was better than stress testing with imaging for high-risk patients with an intermediate probability of obstructive coronary disease. To those of us who read the actual manuscript, this claim was both astonishing and disturbing, but a closer look shows exactly the opposite. The PROMISE Trial reported results for 10,003 patients randomly assigned to CTA compared with functional testing, primarily using either nuclear scintigraphy or echocardiography. The primary endpoint, which included a composite of major adverse cardiovascular events, was neutral, showing a hazard ratio (HR) for CTA of 1.04, 95 percent confidence intervals from 0.83 to 1.29, p = 0.75.1 However, a closer inspection of the results revealed findings that suggested significant harms arising from routine application of CT angiography, concerns that were not adequately addressed in the reporting of the results.
The PROMISE trial was designed to test the hypothesis that CTA was superior to functional testing in high-risk patients with suspected obstructive coronary disease. As noted above, statistical testing of the superiority hypothesis was unequivocally negative (p = 0.75). However, in the study design, the authors pre-specified that, if superiority was not shown, they would test a non-inferiority hypothesis, defined as ruling out a 10 percent worse outcome in major adverse cardiac events (MACE) for patients assigned to CTA. In fact, the PROMISE trial failed the non-inferiority hypothesis also, showing an up to 29 percent worse outcome with CTA. Accordingly, the PROMISE trial demonstrated neither superiority, nor non-inferiority. Describing such a result as supportive of the widespread usage of CTA defies any reasonable scientific and statistical logic.
Of great concern, CTA resulted in a high rate of unnecessary cardiac catheterization, 12.2 percent for patients assigned to CTA and 8.1 percent for patients assigned to functional testing. More disturbingly, patients assigned to CTA had approximately double the rate of revascularization, 6.2 percent vs. 3.2 percent, including coronary bypass graft surgery (72 vs. 38 patients). Since these revascularization procedures did not result in improved outcomes, by definition, the added procedures were not beneficial. The authors highlighted a 0.9% lower rate of catheterizations showing no obstructive coronary disease for CTA-assigned patients, but this result is expected when one anatomic test (CTA stenosis > 70 percent) is compared to another anatomic test (invasive angiographic stenosis > 50 percent). More importantly, this “benefit” must be viewed in the context of higher rates of catheterization and revascularization. Overall, the PROMISE study shows more harm than good with routine usage of CTA vs. functional testing.
The PROMISE trial is not the first randomized controlled trial to demonstrate poor performance for CTA in the diagnosis and management of coronary disease. The FACTOR 64 study randomized 900 diabetic patients to screening with CTA vs. standard guideline-based optimal diabetes care. After four years of follow-up, there was no difference in the rate of major adverse cardiovascular events (p = 0.38). However, similar to PROMISE, randomization to CTA resulted in an excess rate of coronary angiography (13.3 percent vs. 5.1 percent), percutaneous coronary intervention (6.0 percent vs. 1.8 percent), and coronary bypass graft surgery (2.9 percent vs. 1.3 percent). Interestingly, the absence of an outcome benefit in the CTA group was observed despite significantly higher rates of statin usage (p = 0.008) and lower LDL-C levels in the CTA group compared with the control group.
The Scottish Computed Tomography of the Heart Trial (SCOT HEART) randomized 9,849 patients with suspected obstructive coronary disease to CTA vs. standard care with one to seven years of follow-up. Clinical outcome was not the primary endpoint, but the authors pre-specified an amazing 13 secondary cardiovascular endpoints, presumably in hopes that at least one of them would show a favorable effect. Not a single one of these 13 endpoints achieved statistical significance. SCOT HEART also observed a trend toward an increase in unnecessary cardiac catheterization and coronary intervention, although these differences did not reach conventional levels of statistical significance. The authors inappropriately emphasized that one of the 13 secondary endpoints—myocardial infarction—approached significance (p = 0.053), but made no statistical correction for multiplicity. Taken in its entirety, the SCOT HEART study provides no rationale for routine usage of CTA in the diagnosis of coronary disease.
Why did these three pivotal studies fail to show a benefit for CTA in high-risk patients with suspected coronary disease? In all three studies, the events rates were incredibly low—about three percent in PROMISE, four percent in FACTOR 64, and only two percent in SCOT HEART. There is a simple reality at play. It’s difficult to improve outcomes for patients who already have a 96 percent to 98 percent likelihood of avoiding cardiovascular morbidity-mortality. These CTA trials reflect the extraordinary success in recent years of primary and secondary prevention measures. Statins, effective anti-hypertensive agents, better anti-platelet therapies, and other measures have progressively reduced the risk of MACE for both primary and secondary prevention populations. Event rates are so low that the value of any method for detecting coronary ischemia is nearly impossible to document.
A further issue arose following the publication of the Clinical Outcomes Utilizing Revascularization and Aggressive Drug Evaluation (COURAGE) Trial in 2007. There is now wide consensus that coronary intervention does not reduce the likelihood of MACE in patients with stable angina and documented obstructive coronary disease compared with optimal medical therapy. Accordingly, the detection of obstructive coronary disease is not likely to improve outcomes. In fact, the detection of coronary obstruction results in the inevitable triggering of the “oculostenotic reflex,” the irresistible desire to perform a revascularization procedure despite the absence of clinical trial data demonstrating improved outcomes. Perhaps the most important message from the three failed CTA studies is the concept that treating the systemic risk factors in patients with coronary symptoms represents the most effective and least costly approach to cardiovascular care.
- Douglas P, Hoffman U, Patel M, et al. N Engl J Med. 2015;372:1291-300.
Steven E. Nissen, MD, is Chair of the Department of Cardiovascular Medicine at the Cleveland Clinic and co-author of Heart411: The Only Guide to Heart Health You’ll Ever Need.
Paul C. Cremer, MD, is a cardiovascular fellow (imaging) in residency at Massachusetts General Hospital.
< Back to Listings