The analysis of coronary artery disease by non-invasive coronary computed tomographic analysis has been limited by low specificity. The addition of computer derived flow analysis of the CTA data has added the potential for a non-invasive test that yields both anatomic and functional data. This emerging technology aims to reduce the need for invasive cardiac procedures and associated risks.
The DeFACTO study (Determination of Fractional Flow Reserve by Anatomic Computed Tomographic Angiography) compared the first iteration of FFRct technology against invasive angiography and FFR in patients with suspected or known CAD, but did not achieve the pre-specified target accuracy.2 Another study, DISCOVER-FLOW, demonstrated an accuracy of 84.6%.6 A second iteration of the FFRct algorithm was tested against invasive angiography with FFR in 254 patients with suspected CAD in the NXT (Analysis of Coronary Blood Flow Using CT Angiography-Next Steps) trial. This study reported a diagnostic accuracy, sensitivity, specificity, positive predictive value and negative predictive value of 81%, 86%, 79%, 65% and 83% respectively.7 In this trial, 484 vessels in 254 patients and found that FFRct significantly improved the per patient specificity and PPV (32 to 84% and 40 to 65%, respectively) and the per vessel specificity and PPV (60 to 86% and 33 to 67%, respectively) compared with CTA. However, this does not translate into benefit compared to CTA alone for identifying patients whose invasive FFR will warrant intervention, and 35% of the abnormal results were false positives.8
The PROMISE study demonstrated that patients with an FFRct less than or equal to 0.80 were significantly more likely to have coronary revascularization and to meet the composite endpoint of major adverse cardiac events or revascularization than those with FFRct greater than 0.80.9 The study also showed that reserving invasive coronary angiography for patients with FFRct less than or equal to 0.80 could reduce the rate of performing invasive coronary angiography by 28%. The initial trial did include a CCTA strategy, which did not improve outcomes after two years of follow-up. The study, a retrospective, observational, cohort study, also highlighted the dependence of FFRct on the quality of the CCTA images being post-processed, with one-third of the images submitted for review of insufficient quality.10
The FFRct RIPCORD study involved 200 patients and compared the management of patients by cardiologists with and without the addition of FFRct data to the baseline CTA data. The additional FFRct data changed the management plan for 72 patients (36%) based on lesion severity assessment. With inclusion of changes in the PCI target vessel, patient management was altered in 44% of patients.11
In the PLATFORM study, 584 patients were allocated to noninvasive or invasive coronary angiography cohorts. Each cohort was divided into standard of care and FFRct guided care groups. The 90-day primary end point of invasive catheterization without obstructive coronary artery disease occurred in significantly fewer patients with FFRct guided care than usual care in both cohorts. This prospective, longitudinal comparative effectiveness study demonstrated that in an appropriately selected population for CCTA, 88% of the studies were of sufficient quality to perform FFRct on the derived images.12,13 From the initial group of patients, 116 were identified to undergo ICA. Fifty-two underwent CTA/FFRct resulting in the cancellation of the ICA in 40 (77%). At one year, they reported no major cardiovascular events in this group.14 While this is promising, there are significant limitations. The group was not allocated randomly, and patients in the FFRct arm were younger, had a lower incidence of diabetes, and lower pretest probability of obstructive CAD. The small sample size was not powered to show a difference in clinical events and major cardiac events; therefore, this data cannot be extrapolated to a larger, more diverse population.
The ADVANCE registry (Assessing Diagnostic Value of Non-Invasive FFRct in Coronary Care) is a large prospective registry that included 5,083 patients with symptoms suggestive of angina from 38 centers in Europe, North America and Japan. There was a change in management pathway in 66% of patients, after FFRct data were available compared with CCTA alone.15 At one year, the authors report that patients with FFRct ≤0.80 were less frequently referred for ICA (from 62.9% to 252.9%, P < 0.001) and reported the prevalence of major events was low and similar between the cohorts.16 This data is limited by the observational nature of the study with the inability to adjust for confounding factors such as additional studies being used to make the ultimate decision of whether or not to proceed with ICA, the risk from multiple biases including referral bias, attrition bias, and other bias, and variable experience with FFRct among the study sites.
A recent Danish observational study that included 3674 patients with suspected coronary artery disease who underwent CCTA with FFRct for stenosis between 30% and 90% had supportive findings for use of this technology in care management. Patients with FFRct greater than 0.8 had similar outcomes to patients without obstructive disease on CCTA (i.e. less than 30%). By contrast, patients with FFRct ≤0.8 had significantly more major adverse coronary events (9.6% vs. 1.4%).17
A 2017 systematic review by Cook et al. evaluated 908 vessels from 536 patients in five studies. They reported that the per-vessel diagnostic accuracy of FFRct was 81.9% (95% CI: 79.4-84.4%). They further broke down the diagnostic accuracy of FFRct by the FFRct values and reported values below 0.06, 0.06-0.07, 0.70-.80, 0.80-0.90, and above 0.9 was 86.4%, 74.7%, 46.1%, 87.3% and 97.9%, respectively. This data showed clustering of large numbers of FFRct values in the >0.9 range, where the correlation between FFRct and FFR was very high. However, when approaching the 0.8 level, the correlation becomes less robust. Of significance, the FFRct correlation showed 82% overall diagnostic accuracy through threshold was met for FFRct values lower than 0.63 or above 0.83; however, the correlation between the 0.7 to 0.8 group was notably lower at 46.1%. More stringent 95% and 98% diagnostic accuracy thresholds were met for FFRCT values lower than 0.53 or above 0.93 and lower than 0.47 or above 0.99, respectively.18
The 2019 PACIFIC trial is a prospective study comparing 208 patients with suspected stable CAD who underwent coronary CTA with FFRct calculations, SPECT, PET, and invasive FFR. 83% of vessels could be evaluated with FFRct, which demonstrated a diagnostic accuracy, sensitivity, and specificity of 87%, 90%, and 86% on a per-vessel basis and 78%, 96%, and 63% on a per-patient basis, respectively. FFRct outperformed coronary CTA and SPECT, and PET for vessel-specific ischemia, while PET performed better for per-patient and intention-to-diagnosis analysis.19
The 2019 ReASSESS trail is a single centered prospective study where 143 patients scheduled to undergo invasive coronary angiogram with stable angina also underwent FFRct and SPECT. They found FFRct had similar overall diagnostic accuracy, and FFRct had higher diagnostic sensitivity than SPECT (91% (95% CI:81% to 97%) versus 41% (95% CI; 29% to 55%; p<0.001). The specificity of FFRct compared to invasive FFR was 55% (86% SPECT) with PPV a 58% and overall accuracy of 70%. In analyzing diagnostic performance for detection of coronary artery disease on an intention to treat basis positive predictive value was 47% with accuracy of 63%.20
A 2020 meta-analysis of seventy-seven studies found that FFRct identifies clinically significant stenosis in patients with suspected CAD with a high sensitivity (mean 85%, 95% CI: 81-88%), but low to moderate specificity (mean 69%, 95% CI: 64 to 74%).21
Additional literature recommended by the ACC provided better insight into current anatomic applications and the accuracy of the determinations. The SYNTAX family of randomized controlled trials demonstrated that PCI or stenting is a viable option in patients with three-vessel disease and applying this technology to decision-making regarding CABG or PCI. Use of FFRct to determine a functional SYNTAX Score (in the SYNTAX III study) reclassified 30% of patients to a lower score.22-24 ACC also provided additional literature supporting the accuracy of CCTA in obese patients.25,26
Accuracy with FFRct has been validated in several studies compared to invasive FFR to be around 80% in patients with stable CAD. While non-invasive FFRCT shows promise as comparable or even superior performance to some existing technologies, such as stress test or SPECT, it falls behind invasive FFR in accuracy and PPV. The PPV ranges from 52-65% among the various studies.7,19,20 There is also a risk of false positive of up to 35%, so confirmation at the time of invasive coronary angiogram confirmation of the lesion is often performed prior before stent placement.8
The use of invasive FFR is endorsed by multiple societies, including the Society of Cardiac Angiography and Interventions (SCAI) and American College of Cardiology (ACC)4,27 and European Society of Cardiology.28 However, there is no societal guidance published at this time from major cardiac societies that address the role of FFRct.