Meta-analysis Questions PCI vs CABG in Women With Multivessel Disease

There are new data hinting that women with multivessel disease may do better with CABG than with PCI, this time from a meta-analysis of 15 trials, although only a minority of these included gender-specific data.

While the combined data set is problematic, lacks patient-level granularity, and is underpowered for individual endpoints, it echoes a signal seen in other trials, senior author Alexandra Lansky, MD (Yale University School of Medicine, New Haven, CT), told TCTMD. A post hoc analysis of the SYNTAX trial, now more than 10 years old, suggested that mortality is higher in women following PCI than after CABG at 5 years, despite them having lower SYNTAX scores than men. In EXCEL, women had numerically higher rates of the primary composite endpoint (death, MI, and stroke) with PCI than with CABG, but the difference was not statistically significant.

“This analysis was fueled by the fact that the guidelines now promote PCI for multivessel disease when you have a low-to-intermediate SYTNAX score—both guidelines in Europe as well as the US have adopted this,” Lansky said. “But these recommendations are coming from a majority male population, whereas we are dealing with divergent results in women. What I wanted to do was gather more information from other studies to get a bigger picture.”

Results of the new meta-analysis were published February 26, 2020, as a research letter in JACC: Cardiovascular Interventions.

Ultimately, we do need some convincing evidence that our guidelines apply to females, and we need more prospective randomized controlled data to definitely address this. Alexandra Lansky

Lead authors Burcu Gul, MD, and Tayyab Shah, BA (both Yale University School of Medicine), and colleagues identified 15 trials comparing outcomes with PCI and CABG; only six reported a composite endpoint of all-cause death, MI, and stroke according to patient sex. Indeed, women made up only 1,909 patients or 25% of the entire pooled cohort. Over follow-up times ranging from 1 to 5 years, that composite endpoint was more common in women who’d undergone PCI rather than CABG (HR 1.31, 95% CI  1.05-1.63). Event rates were similar between CABG and PCI in women across various subgroups, including multivessel disease with or without left-main disease, first- or second-generation drug-eluting stents, and time of last patient enrollment (before or after 2010). The reason is likely because numbers were too small to detect a difference, Lansky said.

The authors acknowledge the many limitations of their analysis. In particular, the lack of publicly available patient-level data from these trials makes it impossible to control for demographics and disease complexity.

To TCTMD, Lansky said the principle aim of the analysis was to be provocative and to highlight the lack of support for current guideline recommendations in women. “I think point number one is only 40% of the studies that have looked at CABG versus PCI in multivessel disease or left main have actually formally published a gender subanalysis. We should be seeing gender subanalyses in all of these papers,” she said. “Number two is the fact that once again we’re seeing that just a quarter of the patients enrolled in these trials are women, so 75% are male and that’s what’s driving guidelines.”

At a minimum, she said, a patient-level pooled analysis would allow for a better understanding of a potentially different treatment effect in women. “Ultimately, we do need some convincing evidence that our guidelines apply to females, and we need more prospective randomized controlled data to definitely address this,” Lansky said.

An important consideration are those prior data suggesting that SYNTAX score is typically lower in women than men, since it is a lower SYNTAX score that is used in guidelines as a rationale for PCI as an alternative to CABG. “And yet we seem to be seeing better outcomes with bypass surgery in women,” Lansky noted.

Commenting on the paper for TCTMD, Roxana Mehran, MD (Icahn School of Medicine at Mount Sinai, New York, NY), agreed that because randomized controlled trials enroll so few women, the role of meta-analyses becomes critically important. That said, in this case, the heterogeneity between the trials makes it impossible to draw firm conclusions, she added.

Raising this issue of a possible gender-specific response to revascularization “on the heels of the PCI versus CABG conflicts and conversations [currently erupting online] will really fuel the fire,” Mehran warned. “Unfortunately, we really have to take this analysis with a grain of salt without having an all-female, randomized controlled trial. Furthermore, these hard endpoints of death, MI, and stroke are extremely important, but in these 15 trials, with only 25% women, you really are underpowered. . . . So I’m nervous to make any kind of definitive statement, although it does raise the question of what the best treatment is in women.”

What’s interesting, Mehran added, is that in TAVR:SAVR trials, TAVR’s edge over surgery has been even greater in women than men. That finding—that an open procedure is worse than a less-invasive one in the setting of aortic valve disease—would be at odds with a scenario where women did better with CABG than with PCI.

Asked what a possible mechanism might be to explain why women do better with CABG, Lansky acknowledged she had “no idea.” But, she added, that doesn’t mean it shouldn’t be studied, or that the idea is not taken seriously. She and her colleagues had written up this data as a full meta-analysis, adhering strictly to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement guidelines. After being rejected from multiple journals, she said, they agreed to have the paper accepted as a research letter, which limits publications to 800 words and no more than five references.

“I think it was because we are challenging the guidelines,” Lansky said.