ORBITA Redux: Lancet Correspondence Takes Talk of the Sham-Controlled PCI Trial to Further Heights

Eight months after the results of the ORBITA trial shook up interventional cardiology, the Lancet has published no less than nine letters to the editor from different researchers and physicians questioning everything from statistical calculations to interpretation of the data.

The sheer volume of letters from around the world is impressive, and unsurprising given that the high-profile trial showed PCI was no better than a sham procedure for improving symptoms or exercise capacity in patients with stable, single-vessel coronary artery disease. Since its presentation at TCT 2017 in Denver, CO, there has been no shortage of opinion regarding the clinical applicability of ORBITA. 

Commenting on the letters, senior investigator Darrel Francis, MD (Imperial College London, England), said he was “amazed” by the ingenuity people put into alternative ways of representing the results to try to make the study appear positive. Positively spinning the trial might have made the results easier to swallow, but he said the investigators’ aim has been to preserve academic integrity. Francis told TCTMD that while academics will occasionally jokingly engage in banter about how statistics can be manipulated, the ORBITA trial shouldn’t be twisted to fit biases about the benefits of PCI in stable coronary artery disease.

“This makes me realize that some people habitually manipulate data to suit their preconceptions, so much so that they are surprised when other authors don’t automatically do so,” he said.

Monday-Morning Quarterbacking

In one letter, Tao Tu, MD (The Second Xiangya Hospital, Changsha, China), and colleagues suggest that the benefit of PCI on exercise time—a nonsignificant 16.6 seconds longer than in the sham-control arm—might have been statistically significant if the study was larger or longer in duration. Additionally, they suggest the long-term outcome of PCI in patients with stable angina might be affected by the status of fractional flow reserve (FFR) in target vessels.

In two separate letters, Axel Bauer, MD (Ludwig Maximilian University, Munich, Germany), and colleagues and Grant Reed, MD, and Samir Kapadia, MD (Cleveland Clinic, OH), also question the validity of the ORBITA findings. Both groups point out that PCI was performed in 29% of patients with hemodynamically irrelevant lesions (FFR > 0.80).

Similarly, Adrian Banning, MBBS (John Radcliffe Hospital, Oxford, England), and Simon Redwood, MBBS (King’s College London, England), note that approximately 25% of patients in ORBITA had class 0 or 1 angina. 

“These patients would not be expected to have symptomatic improvement with antianginal therapy or PCI and including them could have biased the results,” write Reed and Kapadia. In their letter, Banning and Redwood say they are unsure how, “without symptoms or ischemia, patients can benefit from an intervention that is not recommended by guidelines.”

In response, Francis and lead ORBITA investigator Rasha Al-Lamee, MBBS (Imperial College London), counter that several letter writers misunderstand the role of FFR testing in clinical practice. “FFR is not clinically required for severe lesions or when there is a positive ischemia test,” they write. “Instead, the role of FFR is in lesions of uncertain significance.” In fact, at least 94% of patients who participated in ORBITA had one or more positive noninvasive or invasive tests for ischemia at randomization.

Francis and Al-Lamee also point out that, in real-world practice, PCI is not just being performed in patients with class 2 or higher angina symptoms. For example, recent data from the National Cardiovascular Data Registry CathPCI registry suggests that 22% of patients undergoing elective PCI have no symptoms and no angina. Finally, they state that although 11% of patients had not had recent angina at randomization, the proportion is similar to other unblinded trials showing relief of angina symptoms, such as ACME (9%), FAME 2 (11%), and COURAGE (13%).

Overall, Francis sees the discussion about symptom improvement in patients with stable coronary artery disease as positive. In planning the design of ORBITA, Francis said some of the assumptions they made about what is known about the relationship between angina symptoms and noninvasive and invasive testing, including angiography, turned out to be not true. “This could be seen as bad news,” he said. “On the other hand, it’s also good news as it means there are some very important clinical questions that can be addressed by researchers over the next few years without millions of dollars of expense.”

Wire Dissections and T Tests

Philippe Degrell, MD (Paris Descartes University, France), and colleagues highlight the complication rate in the trial. Pressure wire-related coronary dissections occurred in 4% of patients, a rate that is unacceptably high from experienced operators treating patients with stable coronary artery disease, they say in their letter. However, Francis and Al-Lamee point out that the research FFR was measured only to test further blinded hypotheses. “We had more dissections than in clinical practice, because we measured FFR in patients with tight lesions in whom the PCI decision was already made,” they explain.

And finally, in two other letters, researchers suggest the ORBITA group should have analyzed the difference in exercise time between the two study arms using a mixed-effects model, such as ANCOVA, rather than a two-sample T test. As such, argues Stuart Pocock, PhD (London School of Hygiene and Tropical Medicine, England), the mean treatment difference between PCI and sham procedure is biased toward the null.

Francis told TCTMD that ANCOVA is a more advanced test—one that would have accounted for the small mismatch in exercise time at baseline—but they had prespecified the two-sample T test for their analysis.

“Criticism of our original plan to use the two-sample T test, like the similarly-sized ACME study, which was ragingly positive, is entirely reasonable,” said Francis. “Knowing what we know now, we agree it would have been better to specify ANCOVA. Nevertheless, the primary analysis has to be done in the way one prespecifies. Rest assured, we’ll use ANCOVA for secondary analyses.”