A Novel Computer-Assisted Drug-Induced Liver Injury Causality Assessment Tool (DILI-CAT)

Abstract

AbstractObjectiveWe hypothesized that a drug-characteristic DILI-phenotype could be defined and be used to develop a computer-assisted DILI causality assessment-tool (DILI-CAT).DesignA drug-specific DILI-phenotype was developed for amoxicillin-clavulanate (AMX/CLA), cefazolin, cyproterone, and polygonum multiflorum using data from published case series, and subsequently a DILI-CAT Score (DILI-CAT-S) was created for each drug. The phenotype was made up of the following three parameters: (1) latency, R-value, and (3) AST/ALT ratio (also de Ritis ratio). A point allocation system was developed with points allocated depending on the degree of deviation from the core of published data for the three phenotypic parameters.ResultsThe four drugs had a significantly different phenotype based on the three parameters utilized. For example, the median cyproterone latency was 150 days versus less than 43 days for the other three drugs (median: 26 for AMX/CLA, 20 for cefazolin, and 20 days for polygonum multiflorum; p<0·001). The R-value for the four drugs was also significantly different (median: cyproterone [12.4] and polygonum multiflorum [10.9]) from AMX/CLA (1.44) and cefazolin (1.57; p<0.001). The resulting DILI-CAT-S effectively separated cyproterone and polygonum multiflorum from AMX/CLA and cefazolin, respectively (p<0·001). Notably, because of overlap in phenotype AMX/CLA and cefazolin could not be differentiated by DILI-CAT-S.ConclusionDILI-CAT is a data-driven, diagnostic tool built to define drug-specific phenotypes. Data presented here provide proof of principle that a drug-specific, data-driven causality assessment tool can be developed for different drugs and raise the possibility that such a process could improve and standardize causality assessment methods.FundingDCR was supported by the NIH, grant P30 DK 123704