To investigate the prevalence of biopsy-proven non-alcoholic steatohepatitis (NASH) in a cohort of patients with morbid obesity and with or without type 2 diabetes (T2D) and to find non-invasive predictors of NASH severity.

We evaluated a cohort of 412 subjects (age 19-67 years, body mass index-BMI: 44.98 kg/m²), who underwent fine-needle liver biopsy during bariatric surgery. Thirty-six percent of the subjects were affected by T2D. Liver biopsies were classified according to the Kleiner's NAFLD Activity Score (NAS). NAFLD Fibrosis Score (NFS), AST/ALT ratio, AST to Platelet ratio (APRI), fibrosis-4 score (FIB4) were calculated. A neural network analysis (NNA) was run to predict NASH severity.

The prevalence of biopsy-proven NASH was 63% and 78% in subjects with obesity and without or with T2D, respectively. T2D doubled the risk of NASH [OR 2.079 (95% IC=1.31–3.29)]. The prevalence of NAFL increased with the increase of BMI, while there was an inverse correlation between BMI and NASH (r=−0.145 p=0.003). Only mild liver fibrosis was observed. HOMA-IR was positively associated with hepatocyte ballooning (r=0.208, p<0.0001) and fibrosis (r=0.159, p=0.008). The NNA highlighted a specificity of 77.3% using HDL-cholesterol, BMI, and HOMA-IR as main determinants of NASH.

Our data show a higher prevalence of NASH in patients with morbid obesity than reported in the literature and the pivotal role of T2D among the risk factors for NASH development. However, the inverse correlation observed between BMI and biopsy-proven NASH suggests that over a certain threshold adiposity can be somewhat protective against liver damage. Our model predicts NASH presence with high specificity, thus helping identifying subjects who should promptly undergo liver biopsy.