Table 5. Actual diagnostic performance of computational FFRCTA when compared with invasive FFR in detecting physiologically significant lesions that need revascularization (cut-off ≤0.80).
| Studies | Sample size [patients, vessels] | Pearson correlation coefficient | Agreement (Bias ± SD: virtual index vs. FFR) | Overall diagnostic accuracy | AUC |
|---|---|---|---|---|---|
| DISCOVER-FLOW* [2011] (126) | 103 [159] | 0.68 | 0.02±0.116 | 84% (per vessel) | 0.90 |
| DeFACTO* [2012] (127) | 252 [407] | 0.63 | 0.06 | 73% (per vessel) | 0.81 |
| HeartFlow NXT* [2014] (128) | 251 [484] | 0.82 | 0.02±0.074 | 86% (per vessel) | 0.93 |
| Kim et al. [2014] (131) | 44 | 0.60 | 0.006 | 77% | – |
| Renker et al. [2014] (132) | 53 | 0.66 | – | – | 0.92 |
| Coenen et al. [2015] (133) | 106 [189] | 0.59 | −0.04±0.13 | 74.6% | 0.83 |
| Kruk et al. [2016] (123) | 90 [96] | 0.67 | −0.01±0.095 | 74% (per vessel) | 0.83 |
| Ko et al. [2017] (124) | 42 [78] | 0.57 | −0.065±0.137 | 83.9% (per vessel) | 0.88 |
*, multicenter studies. FFR, fractional flow reserve; FFRCTA, fractional flow reserve form computed tomography angiography; AUC, area under the ROC curve; SD, standard deviation.