Agreement between angiogram-derived and wire-derived fractional flow reserve near the clinical cut-off range

Background

Angiogram-derived fractional flow reserve (FFRangio) is a less-invasive alternative to wire-derived fractional flow reserve (FFR); however, its reliability near the clinical decision threshold remains uncertain.

Purpose

This study aimed to assess the agreement between FFRangio and wire-derived FFR in lesions with values close to the clinical cut-off.

Methods

This multicenter registry included patients with FFRangio values near the threshold for hemodynamically significant ischemia (0.70–0.89) measured during diagnostic coronary angiography. Wire-derived FFR was subsequently assessed during percutaneous coronary intervention. Agreement was analyzed using Bland–Altman analysis and a generalized linear model to identify determinants of measurement differences.

Results

Seventy patients were enrolled in this study; the mean age was 75.2 years, and 47 patients (67.1%) were male. Target lesions were located in the left anterior descending coronary artery (LAD) in 36 patients (51.4%), the left circumflex coronary artery (LCx) in 15 (21.4%), and the right coronary artery (RCA) in 19 (27.1%). Wire-derived FFR was successfully obtained in all patients at the same distance from the target vessel ostium as used for FFRangio analysis. In the overall cohort, the mean FFRangio value was significantly lower than the mean wire-derived FFR value (0.76 vs. 0.82, p<0.001). FFR values differed significantly according to target vessel; the mean FFRangio value was higher in LAD lesions than in LCx and RCA lesions (0.77 vs. 0.74 vs. 0.74, p=0.010), whereas the mean wire-derived FFR value was lower in LAD lesions (0.80 vs. 0.82 vs. 0.86, p=0.049).

In Bland–Altman analysis, the difference between FFRangio and wire-derived FFR varied across the range of mean FFR. In a generalized linear model including mean FFR, target vessel, and their interaction, mean FFR was significantly associated with the measurement difference (p = 0.0015), indicating proportional bias. Both target vessel (p < 0.0001) and the interaction between mean FFR and vessel type (p < 0.0001) were significant, demonstrating vessel-specific proportional bias. When mean FFR was fixed at 0.80 using prediction profiling, the predicted difference between FFRangio and wire-derived FFR was smallest in LAD lesions and close to zero, whereas larger negative differences were observed in RCA and LCx lesions.

Conclusions

FFRangio demonstrates vessel-specific proportional bias relative to wire-derived FFR. In lesions near the clinical decision threshold, agreement between FFRangio and wire-derived FFR is closest in LAD lesions, whereas FFRangio tends to provide lower FFR values in RCA and LCx lesions, leading to more frequent positive ischemia classification. These findings suggest that vessel type should be carefully considered when FFRangio is used for clinical decision making near the cut-off value.