Irrational Exuberance or Evidence-Based Medicine: What Role for CT Angiography in Stable Chest Pain?

In 2016, the National Institute for Health and Care Excellence (NICE) in the United Kingdom broke new ground when it recommended coronary CT angiography as the test of first choice for all patients without established coronary artery disease who present with chest pain. Elsewhere in Europe and in the United States, CT angiography does not occupy such a lofty position in the guidelines for the management of patients with stable ischemic heart disease.

According to one physician, that’s the way it should stay.

In an opinion piece published online December 26, 2018, in JAMA Cardiology, Sanjay Kaul, MD (Cedars-Sinai Medical Center, Los Angeles, CA), writes that guideline committees should hold the line on CT angiography, arguing that the strength of evidence is insufficient to support a stronger recommendation for the noninvasive imaging test in the setting of stable ischemic chest pain.

The timing of Kaul’s plea is not accidental: new US guidelines are expected later this year, likely released to coincide with the American Heart Association (AHA) 2019 Scientific Sessions in Philadelphia, PA.

Speaking with TCTMD, Kaul said CT angiography will provide “additive” information, but a class I indication should be based on whether there is a meaningful advantage with the noninvasive imaging test. He simply doesn’t see it in the available data. 

“I don’t think the quality or quantity of evidence is sufficient enough to upgrade the recommendation to a class I indication,” said Kaul. “At the same time, I don’t want to be viewed as being overly critical of CT angiography . . . . All I’m saying is that I’m not convinced by the data that CT angiography trumps stress testing or the standard of care.” 

The 2012 guidelines from the American College of Cardiology/AHA consider CT angiography a “reasonable” option for patients with a low-to-intermediate pretest probability of stable ischemic disease who are unable to exercise (class IIa, level of evidence B). For those patients who can exercise, CT angiography might also be a reasonable test (class IIb, level of evidence B). The European guidelines are in line with the cautious American recommendations, stating that CT angiography can be an alternative to stress imaging in selected patients.

Trying to Knock SCOT-HEART Down

Coronary CT angiography has been shown to be a sensitive and specific test for the detection of coronary artery disease, but recent studies have sought to determine whether its use could translate into a reduction in cardiovascular events. One such trial, the one Kaul believes lacks robust data, is SCOT-HEART.

In August, the 5-year results from the landmark trial were published in the New England Journal of Medicine and presented at the European Society of Cardiology (ESC) Congress 2018. As reported by TCTMD at the time, impressions of the study couldn’t have been more mixed, and physicians immediately took to social media to hash it out. For example, one physician called it one of the most impactful clinical trials in all of cardiovascular medicine while another said he simply didn’t believe the results.

For his part, Kaul said he was “intrigued by the buzz” SCOT-HEART generated at the ESC meeting but felt there were some misconceptions that needed to be addressed. David Newby, MD, PhD (University of Edinburgh, Scotland), who led the SCOT-HEART study, said it’s always healthy to have a balanced discussion about clinical trial results, particularly trials that have the potential to alter clinical practice, but added “it’s a bit frustrating that people keep trying to knock the findings down.”

I don’t think the quality or quantity of evidence is sufficient enough to upgrade the recommendation to a class I indication. Sanjay Kaul

“All of the evidence points to CT being superior to stress testing in terms of [reducing] clinical outcomes,” Newby told TCTMD. “No trial has ever shown the opposite. Meta-analyses, observational data, and all of the trials are consistent in that regard. Secondly, from first principles, it’s obvious a test that actually shows whether you have disease or not is going to be more powerful than a test that doesn’t.” 

Newby argued that stress testing will never prove superior to coronary CT angiography for reducing clinical events because ischemia is a surrogate endpoint designed to identify advanced lesions caused by severe luminal narrowing. In SCOT-HEART, as well as in PROMISE—a similar trial comparing CT angiography with functional stress testing, which didn’t show an advantage of CT—most MIs occurred in individuals without obstructive coronary artery disease, which stress testing is unable to pick up.

“Over half of the people have a heart attack with only a small narrowing that wouldn’t give them symptoms beforehand,” said Newby. “They have a little plaque that ruptures and causes the heart attack. Those plaques will never be seen on the stress test, but they are seen on the CT. That’s what the trials—SCOT-HEART, PROMISE—have shown. Knowing you have heart disease predicts people who have heart attacks. Stress imaging does do that, but if you have a normal [stress test], it doesn’t mean you have no risk.” 

Debate Continues

In SCOT-HEART, which included 4,080 patients with stable chest pain across 12 hospitals in Scotland, the 5-year follow-up data showed that use of CT angiography instead of standard care to guide patient management reduced the risk of coronary heart disease death or nonfatal MI by 41%. The difference was largely driven by the 40% reduction in the risk of nonfatal MI. Rates of invasive angiography and revascularization were similar between the two study arms.    

In his dissection of the trial, Kaul takes issue with the design of SCOT-HEART and questions why the study showed such a robust reduction in clinical outcomes, specifically MI, when PROMISE did not. For Kaul, PROMISE is a more credible trial based on its design and it was stopped because of futility. In SCOT-HEART, investigators anticipated 485 primary outcome events, but only 129 were observed. Despite a lower number of events accrued, the observed 41% reduction was nearly twice the expected reduction in risk.

“There’s obviously something contributing to this misestimate,” said Kaul. “I don’t buy the argument from the investigators saying better diagnostic performance led to better secondary preventive therapies. This is essentially a primary prevention patient population and statins, or aspirin for that matter, are not likely to yield such a reduction in outcomes.”

It’s a bit frustrating that people keep trying to knock the findings down. David Newby  

In SCOT-HEART, preventive medical therapy was initiated for 402 patients in the coronary CT angiography arm compared with 305 in the standard care arm, a difference of 97 patients. Treating 97 patients with aggressive medical therapy would not be expected to translate into such a large reduction in clinical outcomes—there were 33 fewer events in the CT angiography arm—“even if we assume that preventive therapies were specifically targeting highly unstable but nonobstructive plaques that are more likely to be picked up with coronary CT angiography,” according to Kaul.

Additionally, Kaul points to a recent meta-analysis showing a modest 0.4% absolute reduction in MI with CT angiography, which translates into a number needed to treat of 250. Countering that, Newby said the SCOT-HEART trial showed that just 63 people need to be referred for CT angiography to prevent one MI.

“The reason for [the difference] is because of time,” said Newby. “A lot of the trials that have been reported to date looked at outcomes over a year. As we know, if you give people statins, aspirin, and other things, you have to wait 4, 5, or 6 years before you see a difference in outcomes. The same was true in SCOT-HEART—we had to wait 5 years. All of the other trials that were done followed people for a year.”

While follow-up in PROMISE was a median 25 months, Newby noted that the trial lacked sufficient funding to maintain complete follow-up beyond the 1-year mark. As for the magnitude of the effect size observed with respect to their primary clinical endpoint, Newby said the confidence intervals range from 16% to 59%, which is consistent with other trials and meta-analyses.

Speaking with TCTMD, Ron Blankstein, MD (Brigham and Women’s Hospital, Boston, MA), a preventive cardiologist who uses multiple imaging modalities and who was not involved in the trial, noted SCOT-HEART was not designed to identify the mechanism of benefit.

“My thought is that it’s multifactorial,” said Blankstein, speculating on possible ways CT angiography reduced long-term outcomes. The initiation of preventive therapies, improved adherence, and changes to lifestyle, along with revascularization, could all be associated with the reduction in clinical events. Additionally, patients in the stress testing arm might have de-escalated therapy based on the results.

“While people might say we can’t have a 41% risk reduction based on statins or aspirin alone, it’s important to note two facts,” said Blankstein. “Normally when we do trials to assess the magnitude of benefit, we’re only isolating the effect of one therapy. For example, statins. Here, there is an additive effect of multiple mechanisms, which is one reason why the magnitude might be higher.”

Additionally, the use of preventive therapies in the SCOT-HEART trial would be directly targeted to patients identified as those with highest risk, such as patients with coronary plaque on CT angiography. “When you do targeted therapy, the magnitude of effect will be greater.” That said, Blankstein added that the absolute change in preventive medical therapy between the CT angiography and stress testing groups was not that large.

“I do think there’s more at play here,” he said.

Venkatesh Murthy, MD (University of Michigan, Ann Arbor), who also weighed in on the discussion for TCTMD, called the effect size in SCOT-HEART “implausibly large.” He noted that in the era of high-sensitivity troponin testing, type II myocardial infarctions can be seen in patients without no epicardial disease.

“Knowing a patient does not have obstructive epicardial coronary artery disease may very well have prompted doctors to check biomarkers less often,” he said in explaining the between-group differences. “This will be extremely difficult to tease out after the fact.”

In clinical practice, Murthy noted that CT angiography interpretation can differ between core labs and clinical centers, and that the oculostenotic reflex could lead to increased rates of downstream PCI, which may not be helpful in stable ischemic patients.

Explaining Away the Statistical Significance

To TCTMD, Kaul stressed that clinical endpoints were not formally adjudicated in SCOT-HEART but rather based on diagnostic codes, which raises the potential of misclassification and ascertainment errors. Like Murthy, he said knowledge of normal coronary anatomy on CT angiography could also potentially bias physicians not to pursue diagnostic testing, which would amplify the difference in MIs between treatment arms. Kaul takes issue with the study’s prespecified statistical plan, noting that had they adjusted for multiple comparisons, the results wouldn’t have been statistically significant.

The bottom line, according to Kaul, is that, at best, coronary CT angiography is an alternative to standard care, including conventional stress testing, in patients with stable ischemic chest pain. “Any claim to the contrary is perhaps a case of enthusiasm exceeding the evidence,” he writes.

There is an additive effect of multiple mechanisms, which is one reason why the magnitude might be higher. Ron Blankstein 

Countering Kaul’s statistical arguments, Newby said adjustments for multiple comparisons are analytically inappropriate in this setting because subsequent treatments and outcomes resulting from the initial diagnosis using CT angiography are interdependent. Corrections for multiplicity should only be done when additional comparisons (downstream clinical outcomes) are independent of the first comparison (initial diagnosis), he said. 

Ultimately, such criticisms are disappointing because they appear to go to “extreme lengths to explain away the statistical significance of the trial,” said Newby.

To TCTMD, Blankstein said he doesn’t believe that foreknowledge of the coronary anatomy biased the trial results. In fact, he said, if physicians identify coronary plaque or stenosis on CT angiography, they would be more likely to diagnose patients with MI if they present with chest pain or biomarker elevations, which would end up favoring the stress testing study arm.

In terms of the big picture, Blankstein believes the evidence is robust enough to support a class I indication for CT angiography. Randomized trials, meta-analyses, and observational data all point to the same direction showing that use of coronary CT angiography is associated with a subsequent reduction in MI, he said.

“I do think the current data is certainly sufficient to support a class I recommendation for the use of coronary CT angiography for evaluating patients with stable chest pain who do not have known coronary artery disease,” said Blankstein. “In fact, there is now stronger data to support the use of CT angiography in this context than any other modality.” Ultimately, however, not everyone will be a good candidate, he added, and “the ability to get a high-quality CT angiography may not be available everywhere, so patient selection remains very important.”

In the UK, use of coronary CT angiography is growing. Unpublished data looking at the impact of the NICE guidelines shows that the noninvasive test is about to overtake nuclear SPECT testing in diagnosing patients with stable chest pain, according to Newby.

“The UK has switched,” he said. “CT angiography really has taken off.”