PECTUS-obs at 5 years: OCT high-risk plaques in FFR-negative non-culprit lesions

Summary

PECTUS-obs enrolled 438 patients with recent myocardial infarction and performed OCT of all FFR-negative non-culprit lesions, which were deferred. Of the 420 with core-lab OCT analysis, 143 (34.0%) had at least one high-risk plaque (HRP). Median follow-up was 1828 days, complete in 99.0%.

The primary endpoint was MACE: cardiac death, non-fatal MI or unplanned revascularization. Periprocedural, indeterminate and stent failure-related events were excluded.

In an exploratory landmark analysis, the risk associated with HRP appeared only in the first 2 years.

Strengths

• Blinded core lab and blinded adjudication.
• 99% follow-up at 5 years.
• Biologically coherent association, in the same direction as CLIMA and COMBINE.

Methodological issues

• Fragility and power. The result rests on 57 events. Reclassifying 2 of the 27 events in the HRP arm moves Fisher p from 0.025 to 0.067: fragility index of 2. With 57 events and 34% exposure prevalence, the minimum detectable HR at 80% power is 2.19. Power for the observed HR of 1.87 is about 61%. A positive result at that power tends to overestimate the effect.
• Endpoint changed after the 2-year result. Excluding periprocedural, indeterminate and stent failure-related events was an amendment relative to the 2-year analysis (Mol, JAMA Cardiol 2023). It was approved before database closure, but the 2-year direction of effect was already published. 3 HRP patients (2.1%) and 12 non-HRP patients (4.3%) had only excluded events. Restoring them gives 30/143 (21.0%) vs 42/277 (15.2%), RR 1.38 (0.91 to 2.11).
• Events outside the imaged lesion. In Supplementary Table 2, unplanned revascularization supplies 20 of the 27 events in the HRP arm. Of the 24 revascularizations in that arm, 13 were for chronic coronary syndrome or acute heart failure, and 7 of those were not target lesion revascularizations. Event 4.1 was driven by angiographic progression alone, with no angina or ischemia recorded. Event 2.1 was a staged procedure at day 1747, after an ACS in a different segment. Event 12.1 occurred on day 1 in a non-target segment. Events 8.1, 9.1, 14.1, 14.2, 15.1, 17.2, 17.3, 19.1 and 20.1 occurred in segments unrelated to the imaged lesion. Of the 8.1 percentage-point absolute difference, about 3 points are attributable to the imaged lesion.
• Model B adjusted by univariable selection. Covariates were chosen by univariable association with the outcome in the same dataset. Lipid-lowering therapy at discharge enters with HR 0.10 (0.03 to 0.36), but 415 of 420 patients received it; 4 of the 5 untreated patients had MACE. That HR comes from 5 patients. Previous PCI also enters, and was more frequent in the no-HRP group (17.3% vs 9.8%). Adjusting for a variable imbalanced in that direction raises the HRP coefficient: from 1.89 in Model A to 2.39 in Model B. Model A, with a priori covariates, is the estimate to cite.
• Number of target lesions left out of the models. It was 1.29 vs 1.12 (p < 0.001). Imaging more lesions raises both the chance of being classified HRP and the chance of an event.
• Proportional hazards not tested. There was no Schoenfeld residual test or any other. The discussion argues that the hazard changes over follow-up, the condition under which a single HR becomes an unstable weighted average. There were 44 all-cause and 17 cardiac deaths: 27 non-cardiac deaths compete with the endpoint, and Fine-Gray was not reported.
• Composite exposure driven by one component. Plaque rupture has HR 3.20 (1.75 to 5.85). TCFA alone has HR 1.63 (0.96 to 2.77), p = 0.068. 45 of the 48 ruptured lesions met HRP criteria. The composite exposure is driven by rupture, which had already occurred at the time of imaging.