Predictive Algorithm for Thiopurine-Induced Hepatotoxicity in Inflammatory Bowel Disease Patients.

ABSTRACT

BACKGROUND: Approximately 25% of patients with inflammatory bowel disease (IBD) discontinue azathioprine (AZA) or mercaptopurine (MP) therapy within 3 months of treatment initiation because of adverse drug reactions. Of these side-effects, about half are because of hepatotoxicity. The aim of this study was to validate and (subsequently) optimize a previously reported predictive algorithm for thiopurine-associated hepatotoxicity by increasing the number of patients with IBD benefitting from conventional thiopurine therapy. METHODS: This multicenter observational study included consecutive thiopurine-naive patients with IBD who received AZA or MP treatment. The primary outcome was hepatotoxicity within 12 weeks. The patients with and without hepatotoxicity were compared. Four determinants, namely, age, sex, body mass index (BMI), and 6-methylmercaptopurine ribonucleotide concentrations 1 week after treatment initiation (T = 1) were used to validate and optimize 2 (1 dichotomous and 1 continuous) algorithms using multivariable logistic regression analysis. RESULTS: Of 229 patients, 21 (9%) developed hepatotoxicity and 93% of the patients received MP with a median dose of 0.7 mg/kg (95% confidence interval 0.3-1.4 mg/kg). A difference in BMI was found between with and without hepatotoxicity groups (median 27.6 versus 24.2, P = 0.022). Specificities of 68% (Algorithm 1) and 77% (Algorithm 2) and sensitivities of 56% (Algorithm 1) and 50% (Algorithm 2) were obtained. CONCLUSIONS: Both algorithms demonstrated limited predictive accuracy for thiopurine-induced hepatotoxicity in the validation cohort. Relevant factors contributing to this outcome were changes in thiopurine prescription behavior over time, with more MP prescriptions at relatively lower dosages of MP.