We compared performance of the Roussel Uclaf Causality Assessment Method (RUCAM) with multidisciplinary expert panel review in identifying a drug-induced liver injury (DILI) due to antituberculosis therapy (ATT) and/or antiretroviral therapy (ART). Cases were drawn from a prospective registry of hospitalised adults with suspected DILI due to ATT and/or ART in Cape Town, South Africa. Participants had to fulfil American Thoracic Society criteria for ATT interruption (alanine transaminase [ALT] ≥5 times upper limit of normal [ULN]/ALT ≥3 times [ULN] and symptomatic). Causality assessment by expert panel review served as reference standard. The panel ranked potentially implicated drugs as certain, probable, possible or unlikely causes guided by World Health Organization Uppsala Monitoring Centre criteria. The RUCAM was performed for each potentially implicated drug. We calculated sensitivity and specificity of the RUCAM in identifying a probable/certain drug cause for liver injury. We included 48 participants. All were people with HIV (PWH). Twenty-seven were on concomitant ART and ATT, with a median of six potentially hepatotoxic drugs per case. Sensitivity and specificity of the RUCAM in identifying a probable/certain drug cause of liver injury compared with expert panel review was 7% and 100% respectively. Implicated drugs (times ranked probable/certain by panel) were isoniazid (18/0), pyrazinamide (17/0), rifampicin (15/1), efavirenz (6/4) and lopinavir/ritonavir (1/0). PWH with liver injury received multiple potentially implicated drugs, which may increase liver injury risk and complicate causality assessment. Compared with expert panel review, the RUCAM had low sensitivity in detecting probable or certain drug causes of liver injury.
Read full abstract