Development and validation of MRI-PDFF cutoffs for living liver donor eligibility assessment

Hepatic steatosis (HS) criteria for living donor liver transplantation (LDLT) donor eligibility should be based on large droplet fat as per Banff consensus recommendations. We aimed to establish magnetic resonance imaging proton density fat fraction cutoffs for HS assessment in potential LDLT donors. This retrospective study included consecutive potential LDLT donors who underwent MRI and liver biopsy between 2013 and 2023 at 2 tertiary institutions, each as development (n = 3062; 2015 men; median [IQR] age of 32 [25-38] y) and external validation (n = 472; 287 men; 35 [26-44] y) data sets. Proton density fat fraction (PDFF) was measured using dedicated MRI sequences. Histologic HS, defined as a large droplet fat fraction, was used as the reference standard. Dual PDFF cutoffs aimed at 95% sensitivity or 95% specificity, for diagnosing histologic HS of ≥10%, ≥20%, ≥30%, and ≥40%, were determined in the development data set using 10-fold cross-validation. The cutoffs were then validated in the external validation data set. The equation for estimating histologic HS from PDFF was also derived using linear regression. The PDFF cutoffs for histologic HS of ≥10%, ≥20%, ≥30%, and ≥40%, targeting 95% sensitivity, were 3.7%, 5.5%, 8.0%, and 10.0%, respectively. External validation demonstrated high sensitivities ≥97.9% with specificities ranging from 60.9% to 95.1%. The PDFF cutoffs targeting 95% specificity were 6.3%, 8.0%, 9.1%, and 10.1%, respectively. External validation rendered high specificities ranging from 88.5% to 95.3%, with sensitivities ranging from 76.6% to 100%. For diagnosing histologic HS ≥30%, which is the most prevalently used threshold for LDLT donor eligibility assessment, the PDFF cutoffs achieved sensitivities and specificities of over 90%. The equation of (Histologic HS = -2.95 + 1.93 × PDFF) was derived.