Aims: Fractional flow reserve (FFR), the reference standard for guiding coronary revascularisation, is most commonly acquired during intravenous adenosine infusion. Results may be sensitive to system- and operator-dependent variability in how pressure data are analysed and interpreted. To quantify FFR objectively, we developed a computational protocol to process the recorded pressure signals in a consistent manner. We studied the impact on lesion (re)classification and compared this with the operator-selected FFR obtained during cardiac catheterisation.
Methods and results: The algorithm used a moving average and Fourier transformation to identify the Pd/Pa ratio at its nadir (FFRmin) and during the stable hyperaemic period (FFRstable) in <2 s with 100% repeatability, in 163 coronary stenoses (93 patients). The mean operator-selected FFR (FFRCL) was higher than FFRmin and lower than FFRstable (0.779 vs. 0.762 vs. 0.806, p=<0.01). Compared with FFRmin, FFRstable resulted in 16.5% of all lesions being reclassified, all from significant to non-significant (p<0.01). FFRCL classified lesion significance differently from both FFRstable and FFRmin (11.7% and 6.1% lesions reclassified, respectively, p<0.01).
Conclusions: Subtle differences in how pressure data are analysed and interpreted by the operator during adenosine infusion result in significant differences in the classification of physiological lesion significance. An algorithmic analysis may be helpful in standardising FFR analysis, providing an objective and repeatable result.