Introduction Pancreatitis is a rare, but important complication of thiopurine treatment. The aims of this project were to a) characterise the clinical features of thiopurine-induced pancreatitis and b) identify clinical useful genetic markers that might predict development of this serious adverse drug reaction.
Methods Patients were identified and recruited from 172 sites (128 UK). Inclusion criteria included a) onset of acute severe abdominal pain within three months of starting thiopurine treatment b) ≥ two-fold rise in amylase or lipase c) medical opinion implicating thiopurine therapy and drug withdrawal.
Results We recruited 303 patients. Following adjudication 48 cases classified as definite (recurrent pancreatitis on rechallenge) and 195 cases classified as probable (temporal relationship and no other cause for pancreatitis) were taken forward for analyses. 46% of patients were smokers at the time of development of pancreatitis. Patients were treated with a thiopurine for a median of 19 days (95% CI: 17 – 21) before development of pancreatitis. Most cases were mild, with only 5 cases developing single organ dysfunction. 70% of patients were hospitalised with a median length of stay of 4 days (95% CI: 3.2 – 4.8). Neither age (p = 0.08), drug dose (p = 0.11), BMI (p = 0.73) nor smoking (p = 0.59) predicted length of hospital stay or severity of pancreatitis in multivariate analysis. Using a control cohort of 4,109 Crohn’s disease and ulcerative colitis cases we conducted a genome wide association study with these 239 patients. A significant variant was identified in the Class II MHC region (Odds ratio 3.03, p = 2.63 × 10–20). Dedicated HLA and 1000 genome project imputation refined the association within the MHC (R squared > 0.8 and MAF > 0.01). This association was robust to principle component correction. TPMT genotype was not associated with pancreatitis development (p = 0.99). A second cohort of 100 cases and 500 independent disease controls treated with thiopurines but screened for pancreatitis has been generated to confirm the association.
Conclusion We describe the largest clinical characterisation of thiopurine-induced pancreatitis to date and use this cohort to undertake a pharmacogenetics genome wide association study that has identified a significant association within the Class II MHC region.
Disclosure of Interest None Declared.