Peripheral microvascular function and in-stent late lumen loss after percutaneous coronary intervention: a retrospective cohort study with an exploratory analysis by implanted stent diameter

Introduction Peripheral microvascular endothelial function measured by the reactive hyperemia index (RHI) with EndoPAT predicts cardiovascular outcomes, and post-procedural RHI has been associated with binary in-stent restenosis. Whether pre-procedural RHI relates to in-stent late lumen loss (LLL) measured by quantitative coronary angiography (QCA) after percutaneous coronary intervention (PCI), and whether any association depends on device caliber, has not been established. Methods We retrospectively analyzed 132 patients who underwent EndoPAT before PCI (index PCI 2009-2016). The primary endpoint was in-stent LLL at follow-up angiography ( n = 91; median follow-up 9.4 months). Because in-stent LLL was highly skewed, the primary analysis used Spearman correlation and a robust (HC3) multivariable regression adjusting for age, diabetes, hypertension, number of diseased vessels, and pre-PCI minimal lumen diameter. Pre-specified sensitivity analyses included multiple imputation (MICE, m = 20) for missing follow-up and lesion-level generalized estimating equations (GEE). Subgroup analyses by implanted stent diameter (≤2.75 versus >2.75 mm) and diabetes were exploratory. Results RHI showed only a borderline inverse correlation with in-stent LLL ( ρ = −0.19, p = 0.07) and was not independently associated with it (adjusted β = −0.11, 95% CI −0.39 to +0.18, p = 0.45); multiple imputation ( β = −0.08) and GEE ( β = −0.07) were concordant, and no correlation survived false-discovery-rate correction. Stratification by native reference vessel diameter showed no association, whereas stratification by implanted stent diameter revealed an inverse RHI–LLL correlation confined to small-diameter stents (≤2.75 mm, n = 33; ρ = −0.375, p = 0.032) but not larger stents ( ρ = −0.055). However, the stent-diameter × RHI interaction was not significant ( p = 0.48), so this caliber-dependent signal is hypothesis-generating and tracks the achieved implanted caliber rather than the native reference vessel. Inflammation explained neither result. High-sensitivity C-reactive protein (hs-CRP) was unrelated to RHI ( ρ = −0.02) and to in-stent LLL ( ρ = +0.15), and adding it to the primary model left the RHI estimate unchanged. Small-caliber recipients had similar hs-CRP but more diabetes (57.6% versus 27.6%, p = 0.009) and worse renal function ( p = 0.008), indicating a higher-risk phenotype rather than a pure caliber stratum. Discussion Pre-procedural EndoPAT RHI was not independently associated with in-stent LLL. The exploratory signal confined to small-caliber stents is hypothesis-generating and requires prospective, device-caliber-stratified validation pairing pre- with post-procedural RHI.