The results of SYNTAX are clear. Overall, at 5 years CABG [coronary artery bybass grafting] significantly reduced major adverse cardiac and cerebrovascular events (MACCE) at 26·9% in the CABG group versus 37·3% in the PCI [percutaneous coronary intervention] group (p<0·0001), including cardiac death (5·3% vs 9·0%; p=0·003), myocardial infarction (3·8% vs 9·7%; p<0·0001), and repeat revascularisation (13·7% vs 25·9%; p<0·0001). The investigators noted no significant difference in all-cause death (11·4% vs 13·9%; p=0·10) or stroke (3·7% vs 2·4%: p=0·09). [...]
Patients with lower and intermediate severity coronary artery disease had similar survival with PCI and CABG, whereas in the group with severe coronary artery disease CABG resulted in significantly lower mortality (11·4% with CABG vs 19·2% with PCI: p=0·005), myocardial infarction (3·9% vs 10·1%: p=0·004), and repeat revascularisation (12·1% vs 30·9%: p<0·0001). CABG also seemed to have greater benefit on MACCE in patients with isolated three-vessel disease (24·2% vs 37·5%: p<0·0001) than with left main disease (31·0% vs 36·9%: p=0·12). [...]
In the 25% of patients with diabetes in the SYNTAX trial, occurrence of MACCE was also significantly higher with PCI (46·5%) versus CABG (29·0%; p=0·0002).
With results like these, I wonder if PCI is ever warranted. Taggart notes that “interventional cardiologists will argue that they could potentially achieve better results with newer generation stents.” It’s possible. It’s also true that interventional cardiologists make their living from PCI, a not insignificant conflict. Another clue is found in the exclusions from randomization into SYNTAX.
[A] further 1275 patients (around 40% of the total) were deemed ineligible for randomisation because their coronary artery disease was either thought to be too complex for PCI (1077 who underwent CABG) or too high risk for CABG (198 who underwent PCI).
Is there broad agreement that some patients are at too high risk for CABG and also benefit from PCI? If so, then that’s a clear subpopulation for which PCI makes sense. Are there others? I don’t want to be overly harsh on PCI. I don’t want to be unjustifiably lenient either.