Risk score of drug-induced liver injury among new tuberculosis patients

Background: drug-induced liver injury (DILI) is the most serious problem complicating anti-tuberculosis therapy (ATT). The aim of the study was to develop DILI risk score for targeted prevention and monitoring of DILI in TB patients. Methods: 435 HIV-uninfected new pulmonary TB pts admitted to Moscow Research and Clinical Center for TB Control 2010 to 2015 were monitored for clinical and laboratory signs of DILI during the intensive phase of treatment. The median age was 30.0 (IQR 23-45) yrs, 220 were female, 294 received only first-line anti-tuberculosis drugs. Univariate analysis and logistic regression method were used to identify risk factors. The risk score developed by stepwise backward multiple logistic regression was derived from the fist half and validated on the second half of the study population. Results: the frequency of DILI (alanine aminotransferase (ALT) >3 x the upper limit of normal (ULN) with clinical symptoms, or asymptomatic ALT elevation>5 ULN) was 25.3% (110/435 pts). Severe hepatotoxicity (ALT>10 ULN) occured in 25.5% (28/110). Female sex (OR=4.49, 95%CI: 1.62-12.45), a history of drug allergy (OR=8.32, 95%CI:2.85-25.18), malnutrition (OR=3.10, 95%CI 1.15-8.33) and absence of smoking (OR=4.40, 95%CI 1.45-13.32) were identified as risk factors. Their regression coefficients were included to the final DILI risk score. The area under ROC curve was 0.80, 95%CI:0.69-0.91 for training sample, and 0.78, 95%CI:0.68-0.89 for tested population. The sensitivity and specificity were 80% and 75%, respectively. Conclusion: a questionnaire was developed to assess the DILI risk before ATT starting for targeted prevention and monitoring of DILI among new TB patients.