Less MI, CV Death With Noninvasive Testing for Chest Pain: Observational Data

Noninvasive testing is associated with lower risks of cardiovascular mortality and myocardial infarction in patients presenting with chest pain when compared with individuals not sent for testing, according to a large observational study.

In an analysis of nearly 1.5 million patients from the Canadian province of Ontario, diagnostic testing with one of four noninvasive modalities was linked with a nearly 25% reduction in the risk of cardiovascular mortality, acute MI, or unstable angina. Although the study was not designed to assess the mechanisms behind the downstream clinical benefit of noninvasive testing, investigators say the reduction in clinical outcomes does not appear to be driven by increased use of coronary angiography or revascularization, but rather by greater uptake of guideline-recommended medical therapy and increased surveillance by physicians.

“Over the past 5 to 10 years, there has been a push to look at outcomes related to imaging and testing, and the pushback has been: what’s the mechanism?” lead investigator Idan Roifman, MD (Sunnybrook Health Sciences Center, Toronto, Canada), told TCTMD. “In our study, we looked at a couple of possibilities, one of which was downstream angiography and PCI, but we didn’t actually find a difference between the two groups. We don’t think it’s driven by that, but there was more use of evidence-based medication in the testing group versus the no-testing group.”

The results, which were published July 1, 2020, in the Journal of the American Heart Association,  mirror the 5-year clinical outcomes from SCOT-HEART, a study showing that a chest-pain workup strategy using coronary CT angiography (CCTA) reduced the risk of coronary heart disease death or nonfatal MI by 41% compared with standard care alone. In SCOT-HEART, like in the Ontario analysis, there was no difference in the use of invasive angiography or coronary revascularization between the two strategies.

James de Lemos, MD (UT Southwestern Medical Center, Dallas, TX), who wasn’t involved in the new study, expressed some skepticism of the magnitude of benefit seen with noninvasive testing, noting that comparative effectiveness studies using observational data have limitations. “The researchers do multivariate adjustment and use appropriate methods for the data that they have, but it can’t account for unmeasured factors that lead doctors to test or not to test,” said de Lemos. “That leaves these types of studies at risk for the influence of unmeasured confounding [variables] and selection bias.”

One clue, according to de Lemos, is that the “massive” reduction in cardiovascular mortality (HR 0.68; 95% CI 0.65-0.72) is larger than the reduction in the composite endpoint, as well as the reduction in nonfatal outcomes. “It’s a 32% reduction in cardiovascular mortality and this is for performing a test,” said de Lemos, who also expressed doubts—as others have previously—about the scale of benefit seen in SCOT-HEART with CCTA. “There are very, very few tests, if any, that lead to measurable differences in patient outcomes. If there is a benefit, it’s usually very small, because it’s only the proportion of individuals that have a certain test result that is actionable with the therapy that makes the difference.”

To TCTMD, Roifman acknowledged the study’s limitations, but noted that the results were consistent across different sensitivity analyses and in their propensity score-matched analysis. He added that the group also lacked detailed data on the patient’s chest pain characteristics and they were unable to account for the pretest probability of obstructive disease.

Observational Analysis From Ontario

While SCOT-HEART and PROMISE both looked at a range of different modalities for testing, Roifman said they wanted to evaluate noninvasive testing against no testing to determine if there was a downstream difference in MACE rates, something that is hard to do in a randomized trial since it would be difficult to gain ethical approval for not providing testing to patients with chest pain.

“Cardiac noninvasive diagnostic testing is a pretty big issue, pretty much across the world. In the US, there are about 4 million tests done each year, so it’s a huge budgetary issue for Medicare and private insurers alike,” said Roifman, a scientist at Ontario’s ICES research organization.

The investigators included 1,491,642 patients 20 years and older who were evaluated for chest pain using diagnostic codes from the provincial healthcare system. After they were included in the cohort, these individuals were followed to assess if they received one of four noninvasive diagnostic tests. Of the 317,056 patients who underwent diagnostic testing, 58.7% received a graded exercise stress test, 26.9% had a myocardial perfusion imaging scan, 14.2% underwent stress echocardiography, and 0.3% were sent for CCTA. Patients were followed for a median of 4.1 years.

After a 90-day landmark period, the unadjusted incidence of invasive angiography and revascularization with PCI or CABG surgery was numerically similar between those undergoing noninvasive testing and those not tested. After adjusting for clinically relevant covariates, noninvasive testing reduced the composite endpoint of cardiovascular death, acute MI, and unstable angina by 23% compared with those who didn’t undergo testing (HR 0.77; 95% CI 0.75-0.79). Noninvasive diagnostic testing was also associated with a 17% reduction in MI (HR 0.83; 95% CI 0.79-0.86) and a 13% reduction in unstable angina (HR 0.87; 95% CI 0.82-0.93) in addition to the reduction cardiovascular mortality.  

Medication use after noninvasive testing, including the use of ACE inhibitors/ARBs, statins, or beta-blockers, was significantly better compared with no testing. While medication use partly explains the benefit of noninvasive testing, other factors, including greater surveillance of patients once the testing cascade of starts, could explain the results, said Roifman. He noted that the present study was submitted and accepted for publication well before the COVID-19 pandemic, but said it suggests there may be long-term implications resulting from the loss of elective imaging. 

“During the height of COVID-19, essentially all of the hospitals in Toronto shut down elective care,” he said. “You couldn’t get a stress test or a stress echo for a long time. I understand why they did it, but at the same time studies such as this show there may be a cost. You’re preventing people from getting testing that might actually reduce their risk of death or MI down the road.”

Less Testing an Indirect Result of ISCHEMIA

To TCTMD, de Lemos said noninvasive testing, including CCTA in SCOT-HEART, are excellent diagnostic tools when used in appropriately selected patients. Traditionally, such testing is reserved for patients with an intermediate pretest probability of coronary artery disease. Given the results of the ISCHEMIA trial, which showed there was no benefit of revascularization with PCI or CABG surgery over medical therapy in patients with stable, moderate-to-severe coronary artery disease, he expects there to be less use of such diagnostic tests.

“Given that routine revascularization doesn’t improve outcomes it makes less sense to do so much testing,” he said. “I think we will see an indirect effect of ISCHEMIA being reduced unnecessary testing in this population. Patients with compelling symptoms or high-risk features certainly do merit noninvasive testing but doctors rarely only test the intermediate-risk people. Testing is done much more broadly than that.”

In an editorial, Neel Butala, MD (Massachusetts General Hospital, Boston, MA), points out that SCOT-HEART included patients with a 50% chance of having coronary heart disease, whereas the new observational study includes lower-risk “all comers” with chest pain. In addition, the vast majority of patients in SCOT-HEART randomized to usual care underwent exercise stress testing. “The study by Roifman and colleagues demonstrates that the benefit of such an evaluation may also exist upstream of the clinical decision examined in SCOT-HEART (ie, which testing strategy to choose), with the benefit of further testing potentially extending to the clinical decision of whether to test at all,” he says.

Butala also acknowledges the study’s limitations, noting that while the multivariate-adjusted model accounted for “an array of covariates, it is likely that some subtleties that are not captured, particularly for a syndrome such as chest pain, for which the clinical history is integral to clinical management.”