Diagnostic accuracy of photon-counting CT-derived fractional flow reserve in chronic coronary syndrome patients with severe coronary calcification
M Steinboeck, H Schuchlenz, M Genger, P KullnigAbstract
Background
Coronary computed tomography angiography (CCTA) is established in clinical guidelines for diagnosing chronic coronary syndrome (CCS). New Photon-counting CT provides superior spatial resolution, markedly enhancing coronary lumen visualisation in calcified lesions. Computed fractional flow reserve (cFFR) enables a non-invasive functional evaluation of flow limitations, potentially reducing unnecessary diagnostic invasive coronary angiographies.
Purpose
This study evaluated the diagnostic accuracy of cFFR derived from photon-counting computed tomography compared to invasive fractional flow reserve (FFR) in CCS patients, with a specific focus on those with a high coronary calcification burden.
Methods
Consecutive CCS patients underwent both dual-source photon-counting CCTA and invasive coronary angiography between January 2024 and January 2025. cFFR analysis was performed using dedicated scientific software in a blinded manner. Flow-limiting stenoses were defined as invasive FFR ≤0.80 or angiographic stenosis >90%. Diagnostic performance (sensitivity, specificity, accuracy, positive predictive value (PPV), negative predictive value (NPV)) was evaluated with 95% confidence intervals (CI). Agreement was assessed via Cohen's kappa (κ), interpreted according to Landis and Koch, while McNemar (p) tests evaluated systematic differences between methods.
Results
Analysis included 145 coronary vessels from 87 patients (77% male, mean age 70, range 40-92 years). The Coronary Artery Calcium Score was available for 64 patients (108 vessels) with a mean Agatston score of 1181. Overall group (n=145 vessels): Sensitivity was 89.3% (95% CI: 82.2–93.8%), specificity 69.7% (95% CI: 52.7–82.6%), accuracy 84.8% (95% CI: 79.0–90.7%), PPV 90.9% (95% CI: 84.1–94.9%), and NPV 65.7% (95% CI: 49.2–79.2%). Agreement was moderate (κ=0.577) with no systematic bias (p=0.832). In the high-risk subgroup (Agatston >1000, n=53 vessels from 29 patients), cFFR maintained excellent diagnostic performance with sensitivity 90.4% (95% CI: 78.3–97.5%), specificity 88.9% (95% CI: 51.8–99.7%), accuracy 90.6% (95% CI: 82.7–98.4%), PPV 97.6% (95% CI: 87.1–99.9%) and NPV 66.7% (95% CI: 34.9–90.1%). Agreement was substantial (κ=0.705; p=0.375). Mean radiation dose for CCTA was 7.5 mSv.
Conclusions
Photon-counting CT–derived cFFR accurately identifies haemodynamically significant coronary stenoses, even in patients with severe coronary calcification, demonstrating high concordance with invasive FFR and a high PPV. These findings support its role as a reliable non invasive gatekeeper for invasive coronary angiography, addressing a major limitation of conventional CCTA in heavily calcified vessels.