BACKGROUND:
A high accuracy of CT-derived FFR (FFR-CT) to detect lesion-specific ischemia compared to the invasive gold standard has been repeatedly reported. However, the influence of CT-specific reconstruction parameters on the accuracy of FFR-CT has not been systematically evaluated. We analyzed the influence of reconstructed CT slice thickness on the diagnostic accuracy of FFR-CT in comparison to invasively measured FFR in patients with suspected CAD.

METHODS:
Patients in which dual-source coronary CT angiography had been performed due to suspected CAD and who were subsequently referred to invasive coronary angiography with invasive FFR measurement within three months of the index CT examination were prospectively included in this analysis. Patients with either left main coronary artery stenoses, bifurcation or ostial stenoses were excluded. Invasive FFR was measured using a pressure wire (CERTUS®, St. Jude Medical, Minnesota, USA or Verrata®, Volcano, San Diego, USA) with intracoronary adenosine administration. FFR-CT was determined using an on-site prototype (cFFR Version 3.2.0 and 3.5.0, Siemens Healthineers, Forchheim, Germany). For each patient, data sets were reconstruced with a slice thickness and increment of 0.5mm/0.25mm, 0.75mm/0.4mm, and 1.0mm/0.5mm, respectively. In these data sets, FFR-CT was determined for all vessels with a corresponding invasive FFR measurement, the observer being blinded to the results of invasive FFR. Stenoses with invasively measured FFR ≤ 0.80 were classified as hemodynamically significant. We evaluated the diagnostic accuracy of FFR-CT for the detection of lesion-specific ischemia depending on slice thickness (0.5mm, 0.75mm and 1.0mm).

RESULTS:
A total of 39 vessels in 22 patients (67±9 years) were included. Overall, eleven stenoses displayed hemodynamic significance in invasive FFR (28%). Compared to invasive FFR, the sensitivity, specificity and accuracy of FFR-CT to identify hemodynamically significant stenoses was 82%, 100%, and 95%  for 0.5 mm slice thickness; 91%, 100% and 97%  for 0.75 mm slice thickness, and 82%, 93%, and 90% for 1.0 mm slice thickness, respectively.  The positive predictive value was 100%, 100%, and 82% with a negative predictive value of 93%, 97%, and 93% for 0.5 mm, 0.75 mm, and 1.0 mm slice thickness, respectively. ROC-Curve analysis showed a slightly higher area under the curve for 0.75mm (AUC 0.96, p<0.0001) than for 0.5mm (AUC 0.91, p<0.0001) and 1.0mm (AUC 0.87, p<0.0001) slice thickness to detect hemodynamic lesions.

CONCLUSION:
The diagnostic accuracy of FFR-CT is influenced by reconstructed slice thickness. The loss of spatial resolution decreases accuracy when slice thickness is increased to 1.0 mm.  Higher image noise might contribute to a loss of sensitivity if very thin slices are reconstructed.