1586-P: Five-Year Risk of Hepatobiliary Disease in Type 2 Diabetes: The Fremantle Diabetes Study Phase II

Nonalcoholic fatty liver disease (NAFLD) is estimated to coexist in 30-64% of people with type 2 diabetes (T2D). Several risk scores use clinical/laboratory data to screen for NAFLD. Although not developed for diabetes, some include it as a binary variable. The prognostic value of a high score in T2D is unknown. We therefore aimed to i) identify prevalent/5-year incident hepatobiliary biliary disease (HBD) in community-based people with T2D, ii) evaluate available NAFLD risk scores to predict 5-year incident HBD, and iii) if none performed acceptably, develop a novel prognostic risk score. Detailed data from the Fremantle Diabetes Study Phase II (FDS2) cohort and validated linkages were used to identify ICD coded hospitalizations/cancer registrations for/with HBD and deaths from/with HBD. Of 1,499 FDS2 T2D participants recruited in 2008-2011, 204 (13.6%) had prevalent HBD and 94 (7.3%) of the remaining 1295 had a first HBD event during 5,979 person-years’ follow-up to the event, death from non-HBD causes (124 (9.6%)), or census at 5 years, whichever came first (incidence rate 16 (95% CI 13-19)/1,000 person-years). The discrimination of available NAFLD risk scores for prediction of 5-year incident HBD was weak (AUC 0.51 to 0.61). In a novel competing risk regression equation, Aboriginality, eGFR <30 and ≥90 ml/min/1.73m², ln(AST/ALT), ln(GGT), ln(platelets), and ln(alpha-2-macroglobulin) independently predicted 5-year incident HBD. Discrimination (AUC (95% CI): 0.70 (0.64-0.76), calibration (Hosmer-Lemeshow test, P=0.07) and accuracy (Brier score 0.06 (range 0-0.96)) were acceptable. For a 10% risk cut-off, sensitivity was 38%, specificity 89%, PPV 20% and NPV 95%. This FDS2-derived prognostic risk equation for incident HBD may facilitate selection of people with T2D who would benefit from intensified management aimed at preventing the adverse outcomes of NAFLD.