Opportunistic hepatic steatosis assessment in low-dose coronary artery calcium CT using liver adipose-radiomic index (LARI)

Abstract

Background: Hepatic Steatosis (HS), a key cardiometabolic risk marker, demands accurate, non-invasive diagnosis. Coronary artery calcium scans (CAC) may opportunistically detect HS. This study aims to a) develop a liver adipose-radiomic index (LARI) and b) compare LARI with clinical- and CT-based indices to assess early-grade HS using 1H MRS as the reference standard (HS + defined as ≥5.56% liver fat) c) validate LARI against liver biopsy. Methods: We identified 700 Dallas Heart Study 2 participants (NCT00344903) with CAC + 1H MRS and 159 participants with CAC + liver-biopsy from University Hospital, Cleveland. A previously developed pipeline (DeHFt) segmented liver in these CTs; texture-based radiomic features were extracted from it. A logistic regression classifier was trained on D1 (Nt = 350, 122 HS+) to obtain LARI, subsequently evaluating its performance on the independent D2 cohort (Nv = 350, 125 HS+). LARI was also validated against liver biopsy D3 (Ne = 159, 121 HS+). The diagnostic performance was evaluated via areas under the receiver operating characteristic curve (AUC), net reclassification index (NRI), and decision curve analysis (DCA). Findings: LARI improved discrimination (AUC: 0.91 (95% CI 0.87–0.94)) as compared to clinical- and CT-based indices (via human expert and DeHFt pipeline) (P < 0.05 for each). NRI ranged between 0.05 and 0.45, and DCA showed that LARI outperformed all other indices across probability thresholds. LARI also outperformed CT-based index when examined in liver-biopsy cohort, D3 (P < 0.05). Interpretation: LARI, a noninvasive CT-based index, outperforms existing methods for early HS detection, paving the way for improved cardiometabolic interventions. Funding: For a full list of funding bodies, please see the Acknowledgements.