Variation in Liver Biochemistries in Patients with Decompensated Cirrhosis: Implications for Assessing Drug-Induced Liver Injury in Clinical Trials

Background Current approaches to detection of drug-induced liver injury (DILI) are generally based on changes from normal in biochemical liver tests. However, limited information is available regarding the proportion of patients with normal versus abnormal biochemical tests, the expected temporal pattern of biochemical abnormalities, or the applicability of current approaches to DILI detection for patients with pre-existing liver disease.Objective These analyses aimed to answer the following questions. (1) What proportion of patients have normal versus abnormal biochemical liver tests at study baseline? (2) What is the variability of these biochemical tests on repeat testing? (3) What proportion of patients meet current criteria for discontinuation of treatment? (4) Is an eDISH (evaluation of Drug-Induced Serious Hepatotoxicity) approach likely to be useful in such a study population?Methods Biochemical liver test results from 178 patients with clinically decompensated cirrhosis who participated in a 4-month clinical trial of glycerol phenylbutyrate versus placebo for reduction in recurrence rate of hepatic encephalopathy were reviewed to determine fluctuation over time and the applicability of the US FDA DILI guideline and eDISH.Results In this cohort, the biochemical tests were frequently abnormal at baseline (aspartate aminotransferase [AST] 42 %, alanine aminotransferase [ALT] 71 %, bilirubin 67 %, international normalized ratio [INR] 62 %) and exhibited substantial variation both pre-dose and following enrollment and dosing. Up to 20 % of patients met current regulatory guidance criteria for consideration of interruption of drug treatment irrespective of treatment group assignment and whether tests were normal or abnormal at baseline, and an eDISH display approach appears unlikely to be informative.Conclusions Approaches to DILI detection in patients with pre-existing liver disease will require a more nuanced approach to recognize patterns of potential liver injury.