Prognostic association of non-alcoholic fatty liver disease and CT-derived fractional flow reserve in suspected coronary artery disease

Abstract Background Non-alcoholic fatty liver disease (NAFLD) is associated with cardiovascular disease, whereas CT-derived fractional flow reserve (CT-FFR) provides a noninvasive functional assessment of coronary stenosis. The combined prognostic implications of NAFLD and CT-FFR in patients with suspected coronary artery disease (CAD) remain uncertain. Methods This retrospective cohort included patients who underwent coronary CT angiography (CCTA) for suspected CAD between July 2012 and December 2018 and had an eligible unenhanced CT covering the liver and spleen within 1 week of CCTA. NAFLD was defined by a mean liver-to-spleen CT attenuation ratio of ≤ 1.0 after exclusion of substantial alcohol consumption and other causes of hepatic steatosis. CT-FFR ≤ 0.80 was considered hemodynamically significant. The primary endpoint was major adverse cardiac events (MACE), comprising cardiac death, heart failure hospitalization, nonfatal myocardial infarction, and late coronary revascularization. Cox regression and Kaplan-Meier analyses were performed. Results Among 614 patients, 189 (30.8%) had NAFLD and 185 (30.1%) had CT-FFR ≤ 0.80. During a median follow-up of 62.5 months, 61 patients (9.9%) experienced MACE. In the final hierarchical model, obstructive plaque (adjusted HR, 3.01; 95% CI, 1.27–7.11; P = 0.012) and NAFLD (adjusted HR, 2.29; 95% CI, 1.34–3.91; P = 0.002) were independently associated with MACE. CT-FFR ≤ 0.80 showed a significant time-varying association with MACE ( P = 0.027 for the time-varying effect), with the association attenuating over follow-up. Sequential likelihood-ratio testing showed improved model fit after addition of CT-FFR (χ 2 = 16.44, df = 1, P < 0.001) and NAFLD (χ 2 = 9.26, df = 1, P = 0.002). The NAFLD × CT-FFR interaction was not statistically significant (P for interaction = 0.071). In the hard-endpoint sensitivity analysis, event-free survival differed according to CT-FFR status (log-rank P < 0.001), whereas no statistically significant difference was observed according to NAFLD status ( P = 0.052). Conclusion NAFLD was independently associated with MACE, whereas CT-FFR ≤ 0.80 showed a time-varying association with MACE. Nested-model likelihood-ratio tests showed improved model fit after addition of CT-FFR and NAFLD to preceding clinical and anatomic variables, but the NAFLD-by-CT-FFR interaction was not statistically significant.