400 A systematic review and metanalysis of computed tomography fractional flow reserve (CTFFR) versus invasive FFR in stable patients with suspected cardiac symptoms

Background/Aim The correlation of Computed Tomography (CT) derived fractional flow reserve (CTFFR) as an alternative to invasive FFR has been assessed previously. However, most of these published studies included CTFFR analysed by a range of variable software including Artificial Intelligence (AI) software, computational flow dynamics (CFD) by (HeartFlow®, HeartFlow Inc., Redwood City, CA, USA) and on-site CTFFR software. We undertook a systematic review to assess agreement between invasive FFR and CTFFR analysed solely by CFD HeartFlow® in stable patients with suspected stable coronary artery disease (CAD). Methods A systematic literature search was conducted in MEDLINE (Ovid), EMBASE (Ovid), PubMed, ClinicalTrials.gov, The Cochrane Library, and Web of Science from January 2010 to September 2024. The PICO algorithm was applied as follows: i)Population: Adult patients with stable symptoms suspected to be of cardiac origin. ii)Intervention: CTFFR by HeartFlow®. iii)Comparator: Invasive FFR. iv)Outcomes: Diagnostic efficacy in identifying haemodynamically significant coronary artery lesions, assessed using sensitivity, specificity, and overall diagnostic accuracy. Patient outcomes were not included in the algorithm. Functionally significant ischaemia was defined as invasive FFR threshold of ≤0.80. Editorials, narrative reviews, case reports, in-vitro and animal studies, unpublished studies, non-English publications, and studies with insufficient data were excluded. Study selection was performed by title/abstract screening followed by full-text review. Risk of bias was assessed using the QUADAS-2 tool. Meta-analysis was performed using R statistical software to calculate pooled sensitivity, specificity, positive and negative likelihood ratios, and summary receiver operating characteristic (SROC) curves with 95% confidence intervals. Results Studies reported either vessel-level data (eight studies) or patient-level data (seven studies). In total, 2,096 blood vessels from 1,078 patients were included in the analysis. The pooled sensitivity and specificity for CTFFR at the per-patient level ([figure 2][1]) were 90% (95% CI, 86%-92%) and 75% (95% CI, 65%-83%), respectively. The corresponding pooled LR+ and LR− were 3.5 (95% CI, 2.60–4.90) and 0.145 (95% CI, 0.11–0.19), respectively. The pooled sensitivity and specificity for CTFFR on the per-vessel ([figure 1][2]) or per-lesion basis were 84% (95% CI, 81%–87%) and 80% (95% CI, 73% – 85%), respectively. Corresponding pooled LR+ and LR− were 4.20 (95% CI, 3.05 to 5.83), LR− 0.203 (95% CI, 0.158 to 0.207), respectively. The area under the SROC (AUC) was 0.89 at the per-patient level and 0.87 at the per-vessel or per-lesion level. Conclusions In this systematic review which focussed only on one method of CTFFR analysis, high diagnostic performance of CTFFR was demonstrated at both patient and vessel-level analysis. ![Abstract 400 Figure 1][3] Abstract 400 Figure 1 Sensitivity and specificity per vessel ![Abstract 400 Figure 2][3] Abstract 400 Figure 2 Sensitivity and specificity per patient [1]: #F2 [2]: #F1 [3]: pending:yes