Mortality Appears to Rise When SYNTAX II Score’s Advice for CABG Is Ignored: EXCEL Analysis

PARIS, France—Patients with complex CAD not treated with the optimal revascularization strategy based on the SYNTAX II score may be at an increased risk for mortality, according to an analysis from the EXCEL trial. The discordance mainly seems to apply when the score supports CABG but PCI is pursued instead. None of the differences reached statistical significance, but they do warrant further exploration, experts said at EuroPCR 2019.

The SYNTAX Score II (SSII) was derived from the SYNTAX trial to guide predictions of long-term mortality following revascularization for left main or multivessel disease. It combines anatomic lesion data and selected clinical characteristics, such as age, sex, chronic obstructive pulmonary disease, peripheral artery disease, LVEF, creatinine clearance, and left main coronary artery disease.

“Noncompliance with the [score’s] recommendation to undergo CABG (ie, treating the patient with PCI) was associated with a trend toward higher 4-year all-cause mortality,” reported Patrick Serruys, MD, PhD (Imperial College London, England), in his presentation.

According to Serruys, RCTs are ideally suited for assessing the mortality impact associated with the lack of concordance between SSII recommendations and actual treatment. The main EXCEL study included approximately 1,900 low-to-intermediate risk patients with unprotected left main CAD who were randomized to PCI or CABG. At 3 years, patients in both treatment groups had comparable rates of death, stroke, or MI.

The new analysis included 1,807 patients from EXCEL who had a SYNTAX II score for both PCI and CABG. Among patients whose SSII suggested CABG as the best option, only 78 were randomized to CABG while 81 were randomized to PCI. Similarly, among patients whose score suggested PCI as the best option, 184 were randomized to surgery and 158 to PCI. Overall, the concordance rate with SYNTAX II score recommendations was 85%.

I think a decision to go to surgery or go to PCI with potentially a big difference in mortality deserves at least 15 minutes of looking carefully. Patrick Serruys

For patients in whom there was equipoise between the two interventions, 4-year mortality was not significantly different, with a rate of 9% for randomization to PCI and 7.5% for randomization to CABG (HR 1.19; 95% CI 0.81-1.75).

However, the results were markedly different for the nonequipoise groups, Serruys noted. When the SYNTAX II score recommended CABG, 4-year mortality was 5.3% in those randomized to surgery versus 14.1% in those who had PCI instead (P = 0.07). In the group for whom the score recommended PCI, early mortality was higher with CABG (5.0% at 6 months compared with zero for those correctly allocated to PCI), but by 4 years mortality was 13.6% for those randomized to PCI and 7.8% for those who had CABG instead (P = 0.11).

Serruys pointed out that the analysis was underpowered to detect differences in mortality.

Improving SYNTAX With Machines

Although the differences in the discordant groups were trends and not significant, Serruys said they are provocative enough that the impact of not respecting SYNTAX II score recommendations should be studied further. Among the unknowns, he said, are whether operator discretion and barriers to surgery, for example, might explain why some patients did not receive the recommended revascularization strategy and also why some outcomes were poorer than others.

“This is a very important piece of information and very scary, I would say,” agreed session chairperson Olivier Varenne, MD (Hôpital Cochin, Paris, France). He took an informal poll of the room with regard to use of the SYNTAX I and II scores and found that the vast majority were routinely using them.

One issue well-known issue with the score, though, is that calculating it “is a pain in the neck,” Serruys acknowledged. “We know it takes on average seven minutes plus a standard deviation of seven minutes to complete the SYNTAX II score,” he said in an interview with TCTMD. “I think a decision to go to surgery or go to PCI with potentially a big difference in mortality deserves at least 15 minutes of looking carefully.”

Having said that, “we are in a time of machine learning, artificial intelligence, and multislice CT,” Serruys observed. The ongoing CABG REVOLUTION trial, a follow-up to SYNTAX III REVOLUTION, is testing the safety of multislice CT to plan and execute CABG surgery in patients with left main or three-vessel disease.

To TCTMD, he said his group is dedicated to transforming the score into something easier and automatic with new technology.

“[With] the diagnostic CT on top of FFR, you will have a SYNTAX score anatomy made by machine that is much more reliable and reproducible than what is done by a human being. That’s the dream, but we have to lead with dreams,” Serruys observed. “We are the first to say it’s cumbersome, it takes time, but it has a great value and it is related to the mortality of the individual . . . and I think it will survive.”