Multicentre European Study of Drug-Induced Liver Injury

All study participants provided written informed consent. Studies complied with the Declaration of Helsinki, Good Clinical Practice (Directive 2001/20/EC) and general data protection regulations (EU) 2016/679. Suspected DILI cases were consecutively recruited at centers across Europe between April 2016 and July 2021: UK, Spain, Germany, Iceland, Portugal; approved by: Yorkshire and the Humber - Leeds East Research Ethics Committee (Ref15./YH/0294); Biomedical Investigation Ethics Committee of Andalucia (Ref: AND-HEP-2015-01); Ethical Commission of Ludwig Maximilian University of Munich (Project 85-16); Bioethics Committee Iceland (Ref: 15–104-V1); Ethics Comission of Centro Académico Médico de Lisboa (Ref: 126/15) and National Data Protection Comission Portugal (Authorization 479/2016), respectively. Standard clinical investigations including imaging were performed and cases were followed until liver profile normalization where possible.
Causality assessment was performed^{48 (link)} and the cases were reviewed by an expert panel of at least three experienced clinical hepatologists or clinical pharmacologists from three different European academic centres⁴⁷ . The panel adjudicated episodes as DILI or alternative diseases termed ‘acute non-DILI. They excluded patients where the investigations were inconclusive or when disagreement of probable diagnosis occurred within the panel. Both DILI and non-DILI groups met the same biochemical criteria as defined previously^{48 (link)}, having serum ALT ≥ 5x ULN or ALT ≥ 3x ULN plus TBL ≥ 2x ULN or ALP ≥ 2x ULN with accompanying elevations of gamma-glutamyl transferase. DILI was diagnosed based on the presence of a compatible temporal sequence between drug intake and detection of liver injury as well as test results to exclude alternative conditions, which included, but were not limited to, viral serology, viral load, imaging tests, presence of autoantibodies, immunoglobulin G values and biopsy findings (when available). Table 1 and Supplementary Table 1 show clinical and demographic data, the primary causative drug implicated in DILI and etiology of acute non-DILI controls. Cases of acetaminophen overdose were not excluded and were eligible as an acute control, although none were recruited during the period of this study. The ‘onset’ samples for both DILI cases (DO) and acute non-DILI controls (NDO) were collected at the time of enrollment into the study. A second sample termed ‘follow-up’ was collected from both DILI (DF) and acute non-DILI control (NDF) group patients (median follow-up of 35 days after the enrollment). This was to explore the trajectory of biomarkers in relation to the course of the liver injury. Patients who had liver enzymes below ULN at the follow-up visit were considered to have ‘complete recovery’. Those patients where liver enzymes decreased at follow-up but remained above ULN were termed ‘partial recovery’ (Supplementary Table 6). Where there was no follow-up visit, levels recorded in clinical notes were used to define recovery type.
HV and patients with biopsy-proven chronic liver disease (NAFLD) were enrolled as additional control groups. These controls were sex and age matched to the DILI cases and were recruited in parallel in Nottingham for research approved by East Midlands Nottingham 2 research ethics committee (Ref: GM0102010). HV had no diagnosis of liver disease and for those included in the discovery cohort, absence of fatty liver was confirmed using controlled attenuation parameter of transient elastography or transabdominal ultrasound examination. Subsequent cohorts were not screened for fatty liver but had no diagnosis of liver disease. Serum samples obtained were stored at −80 °C until analysis. Both sexes were included in the study design. Sex data was collected based on self-reporting. Proportions in sub-groups are reported, but sex-based analyses could not be performed due to the small cohort size.