Decision Tree for the Performance of Intraoperative Liver Biopsy During Bariatric Surgery.

BACKGROUND AND AIMS: Bariatric surgery provides a useful opportunity to perform intraoperative liver biopsy to screen for non-alcoholic steatohepatitis (NASH). There is currently no consensus on whether intraoperative liver biopsy should be systematically performed. The aim of this study was to develop and validate a decision tree to guide that choice. APPROACH AND RESULTS: This prospective study included 102 consecutive patients from the severe obesity outcome network (SOON) cohort in whom liver biopsy was systematically performed during bariatric surgery. A classification and regression tree (CART) was created to identify the nodes that best classified patients with and without NASH. External validation was performed. Seventy-one biopsies were of sufficient quality for analysis (median body mass index 43.3 [40.7; 48.0] kg/m2). NASH was diagnosed in 32.4% of cases. None of the patients with no steatosis on ultrasound had NASH. The only CART node that differentiated between a "high-risk" and a "low-risk" of NASH was alanine aminotransferase (ALT). ALT>53IU/L predicted NASH with a positive predictive value (PPV) of 68% and a negative predictive value (NPP) of 89%, a sensitivity of 77%, and a specificity of 84%. In the external cohort (n=258), PPV was 68%, NPV was 62%, sensitivity was 27%, and specificity was 90%. CONCLUSIONS: The present work supports intraoperative liver biopsy to screen for NASH in patients with ALT>53IU/L; however, patients with no steatosis on ultrasound should not undergo biopsy. The CART failed to identify an algorithm with a good sensitivity to screen for NASH in patients with ultrasonography-proven steatosis and ALT≤53IU/L.