“Vox Populi” Fractional Flow Reserve (vpFFR)—Leveraging Wisdom of the Crowd for the Assessment of Hemodynamic Severity of Intermediate Coronary Lesions

TL;DR

This study introduces a new method called vpFFR that uses averaged human predictions to assess heart artery blockages more accurately than existing methods.

Contribution

The novel vpFFR method leverages crowd-sourced human predictions to outperform 2D and 3D QCA in diagnosing coronary lesion severity.

Findings

vpFFR showed a 73% accuracy in classifying lesion severity, better than 65% and 51% for 2D and 3D QCA.

vpFFR had a correlation coefficient of 0.56 with invasive FFR, significantly higher than 2D-QCA (−0.26) and 3D-QCA (−0.01).

Abstract

Background/Objectives: Diagnostic performance of angiography-derived physiological measures has been benchmarked against two-dimensional (2D) and three-dimensional (3D) quantitative coronary angiography (QCA), which are known for their poor correlation with hemodynamic lesion severity. Relying on the statistical concept of the wisdom of the crowd, we devised a human-performance reference for FFR surrogates, called vox populi FFR (vpFFR), and examined the comparative diagnostic performance of vpFFR, as well as 2D- and 3D-QCA, using invasively measured FFR as the gold standard. Methods: Analyses were performed in a single-center, prospective registry of consecutive FFR procedures. We calculated vpFFR as a mean of five independent, blinded predictions of the invasively measured FFR. Pearson’s correlation coefficient and receiver operating characteristic (ROC) curve analyses were used for diagnostic performance comparisons. Results: In 116 patients (156 vessels), Pearson’s correlation coefficients for vpFFR, 2D-, and 3D-QCA with invasively measured FFR are 0.56, −0.26, and −0.01, respectively (p < 0.001, p = 0.001 and p = 0.918). vpFFR has a sensitivity of 56%, specificity of 84%, positive predictive value of 67%, and negative predictive value of 76%. It correctly classified hemodynamic severity of lesions in 73% of vessels compared to 65% and 51% for 2D- and 3D-QCA, respectively. vpFFR has a larger area under the ROC curve than 2D- and 3D-QCA for predicting positive FFR (0.78, 0.63, and 0.45, respectively, p < 0.001). Conclusions: vpFFR, a mean value of five predictions of invasively measured FFR, has moderate diagnostic performance, superior to 2D- and 3D-QCA using FFR as the gold standard, and can be used as a human-performance reference for existing and emerging angiography-derived physiological measures.

Figures