Abstract Background We have previously reported strong diagnostic accuracy and validity of angiography-based vessel Fractional Flow Reserve (CAAS-vFFR) compared to conventional wire based Fractional Flow Reserve (FFR)(1). The CAAS-vFFR system requires at least two additional optimised angiographic cine images taken after intracoronary glyceryl trinitrate administration, separated by ≥ 30 degrees at a frame-rate of ≥ 15 frames/seconds with minimal overlap and foreshortening, no table movement and adequate contrast opacification. At present, there is a paucity of data regarding the validity of using baseline non-optimised diagnostic cine images in the assessment of CAAS-vFFR. Purpose To assess the concordance and validity of using baseline non-optimised diagnostic cine images in measuring angiography based CAAS-vFFR. Methods We conducted an investigator-initiated, single centre, blinded, prospective observational study performing wire based FFR and CAAS-vFFR conducted simultaneously in patients undergoing routine wire-based FFR for intermediate coronary stenoses. All eligible patients had baseline diagnostic and optimised guide-catheter based angiographic images acquired. Results 195 consecutive patients with 205 lesions were recruited over 19 months; 56 (27.3%) lesions were excluded due to inadequate baseline cine angiogram images (Picture 1). 102 (68.5%) had stable symptoms and 107 (71.8%) were male. The median age was 65.0 years (interquartile range 59.0 – 73.0) and mean body-mass-index was 29.2 kg/m2 (± 5.8). There were 117 (78.5%) lesions in the left anterior descending artery/diagonal branches, 20 (13.4%) in the right coronary artery and 12 (8.1%) in the left circumflex artery/marginal branches. There were 93 (62.4%) diffuse (≥ 20 cm), 23 (15.4%) bifurcation and 15 (10.1%) ostial lesions. The mean optimised vFFR and non-optimised vFFR were 0.80 (± 0.1) and 0.79 (± 0.1) respectively with a Pearson correlation of 0.87 (p<0.001). Overall, 52 lesions (34.9%) and 73 lesions (49.0%) had a wire-based FFR and non-optimised vFFR value of ≤ 80 respectively. The mean wire-based FFR and non-optimised vFFR were 0.82 (± 0.1) and 0.79 (± 0.1) respectively. Receiver operating characteristic (ROC) curve analysis revealed excellent diagnostic accuracy of non-optimised vFFR in predicting a wire based FFR of ≤ 80 (AUC 0.91; 95% CI; 0.87-0.96) (Picture 2). Baseline non-optimised diagnostic vFFR images produced a sensitivity of 94.2%, specificity of 75.3%, positive-predictive-value (PPV) of 67.1% and negative-predictive-value (NPV) of 96.1% compared with wire-based FFR. Conclusions CAAS-vFFR derived values from baseline non-optimised diagnostic cine images were strongly concordant with those derived from optimised images and displayed a high sensitivity and NPV compared to wire-based FFR. This reflects the potential for CAAS-vFFR to have broader clinical applications and be utilised as a retrospective screening tool for intermediate lesions.Excluded baseline angiogram lesionsROC Curve
Read full abstract