Angiographic phenotypes predict FFR/iFR discordance in severe aortic stenosis: A cluster analysis

Assessing coronary artery disease in severe aortic stenosis (AS) is challenging due to altered hemodynamics, causing discordance between fractional flow reserve (FFR) and instantaneous wave-free ratio (iFR). The role of angiographic lesion characteristics in predicting this discordance remains unclear.

To identify angiographic predictors of FFR/iFR discordance and define distinct angiographic phenotypes using unsupervised machine learning.

This prospective, single-center registry evaluated 401 intermediate coronary lesions from 221 patients with severe AS (aortic valve area < 1.0 cm², mean gradient >40 mmHg) using FFR and iFR. Quantitative coronary angiography measured percent diameter stenosis (%DS), lesion length (LL), and other parameters. The primary outcome was FFR/iFR discordance. Receiver operating characteristic and decision curve analyses evaluated %DS predictive utility. K-means clustering was applied to %DS and LL to identify lesion phenotypes.

FFR/iFR discordance occurred in 30 lesions (7.5 %), exclusively as FFR-negative/iFR-positive pattern. Higher %DS was the only independent angiographic predictor of discordance (adjusted OR 1.35 per 10 % increase; 95 % CI 1.12-1.64; p = 0.002), with modest discriminative ability (AUC = 0.69). Cluster analysis identified three phenotypes: Cluster 0 (High %DS [median 75.0 %]/Intermediate LL [median 15.0 mm], n = 103); Cluster 1 (Low %DS [median 49.0 %]/Short-Intermediate LL [median 13.6 mm], n = 220); and Cluster 2 (Intermediate %DS [median 63.5 %]/Long LL [median 32.4 mm], n = 78). Discordance incidence was highest in Cluster 0 (16.5 %) versus Clusters 1 (2.7 %) and 2 (9.0 %) (p < 0.001).

In severe AS, FFR/iFR discordance manifests solely as FFR-negative/iFR-positive pattern. While %DS modestly predicts this mismatch, unsupervised clustering reveals that "High %DS / Intermediate Length" lesions primarily drive physiological discordance, identifying cases where iFR may overestimate severity due to AS-related hemodynamics.