A randomised comparison of management strategies for drug-induced liver injury associated with tuberculous meningitis treatment.

BACKGROUND: Drug-induced liver injury (DILI) management during tuberculous meningitis treatment may result in premature rifampicin and isoniazid withdrawal and contribute to poor outcomes. METHODS: Adults with tuberculous meningitis, enrolled into the ACT HIV (N=520) and LAST ACT (N=720) corticosteroid trials (NCT03092817;NCT03100786), who developed DILI were randomised to strategy-1: continue all drugs, unless ALT ≥10x upper limit of normal [ULN], bilirubin ≥43µmol/L, or symptoms worsen; or strategy-2: stop pyrazinamide, unless ALT ≥5x ULN by day 6, bilirubin ≥43µmol/L, or symptoms worsen; or strategy-3: stop rifampicin, isoniazid, pyrazinamide, continue ethambutol, add levofloxacin and aminoglycoside. The primary outcome was the proportion of time, 60-days following DILI randomisation, during which neither rifampicin nor isoniazid were given (or the participant died). Secondary outcomes included acute liver failure, and death/new neurological events. RESULTS: Sixty-seven participants with DILI were randomised to strategy-1 (n=21), 2 (n=21), and 3 (n=25). Participants had fewer days without rifampicin and isoniazid in strategy-1 (median 7 days [1st-3rd quartile 0-31]) and 2 (9 days [0-21]) than strategy-3 (18 days [11-35]; p=0.022 and p=0.041). No participants developed liver failure. Strategy failure (requiring strategy-3 switch) occurred in 8/21 (38.1%) strategy-1 and 13/21 (61.9%) strategy-2. New neurological event or death after DILI randomisation, until 12 months from primary randomisation, occurred in 7/21 strategy-1, 9/21 strategy-2 and 14/25 strategy-3 (p=0.20 strategy 1 vs. 3; p=0.42 strategy 2 vs. 3). CONCLUSIONS: DILI management strategies that increase transaminase drug-stopping thresholds (10X ULN), or stop pyrazinamide alone, appeared safe and reduced rifampicin/isoniazid interruptions vs. standard approaches.