Automated Algorithm Using Pre-Intervention Fractional Flow Reserve Pullback Curve to Predict Post-Intervention Physiological Results.

Journal Article (Journal Article)


This study sought to develop an automated algorithm using pre-percutaneous coronary intervention (PCI) fractional flow reserve (FFR) pullback recordings to predict post-PCI physiological results in the pre-PCI phase.


Both FFR and percent FFR increase measured after PCI showed incremental prognostic implications. However, there is no current method to predict post-PCI physiological results using physiological assessment in the pre-PCI phase.


An automated algorithm that analyzes instantaneous FFR gradient per unit time (dFFR(t)/dt) was developed from the derivation cohort (n = 30). Using dFFR(t)/dt, the pattern of atherosclerotic disease in each patient was classified into 3 groups (major, mixed, and minor FFR gradient groups) in both the internal validation cohort with constant pullback method (n = 234) and the external validation cohort with nonstandardized pullback methods (n = 252). All patients in the validation cohorts underwent PCI on the basis of pre-PCI FFR ≤0.80. Suboptimal post-PCI physiological results were defined as both post-PCI FFR <0.84 and percent FFR increase ≤15%. From the derivation cohort, cutoffs of dFFR(t)/dt for major and minor FFR gradient were 0.035/s and 0.015/s, respectively.


In validation cohorts, dFFR(t)/dt showed significant correlations with percent FFR increase (R = 0.801; p < 0.001) and post-PCI FFR (R = 0.099; p = 0.029). In both the internal and external validation cohorts, the major FFR gradient group showed significantly higher post-PCI FFR and percent FFR increase compared with those in the mixed or minor FFR gradient groups (all p values <0.001). The proportions of suboptimal post-PCI physiological results were significantly different among 3 groups (10.4% vs. 25.8% vs. 45.7% for the major, mixed, and minor FFR gradient groups, respectively; p < 0.001) in validation cohorts. Absence of major FFR gradient lesion (odds ratio: 2.435, 95% [CI]: 1.252 to 4.734; p = 0.009) and presence of minor FFR gradient lesion (odds ratio: 2.756, 95% confidence interval: 1.629 to 4.664; p < 0.001) were independent predictors for suboptimal post-PCI physiological results.


The automated algorithm analyzing pre-PCI pullback curve was able to predict post-PCI physiological results. The incidence of suboptimal post-PCI physiological results was significantly different according to algorithm-based classifications in the pre-PCI physiological assessment. (Automated Algorithm Detecting Physiologic Major Stenosis and Its Relationship with Post-PCI Clinical Outcomes [Algorithm-PCI]; NCT04304677).

Full Text

Duke Authors

Cited Authors

  • Lee, SH; Shin, D; Lee, JM; Lefieux, A; Molony, D; Choi, KH; Hwang, D; Lee, H-J; Jang, H-J; Kim, HK; Ha, SJ; Kwak, J-J; Park, TK; Yang, JH; Song, YB; Hahn, J-Y; Doh, J-H; Shin, E-S; Nam, C-W; Koo, B-K; Choi, S-H; Gwon, H-C

Published Date

  • November 2020

Published In

Volume / Issue

  • 13 / 22

Start / End Page

  • 2670 - 2684

PubMed ID

  • 33069650

Electronic International Standard Serial Number (EISSN)

  • 1876-7605

International Standard Serial Number (ISSN)

  • 1936-8798

Digital Object Identifier (DOI)

  • 10.1016/j.jcin.2020.06.062


  • eng