Introduction 15% of patients with inflammatory bowel disease (IBD) prescribed azathioprine (AZA)/mercaptopurine (MP) demonstrate a skewed drug metabolism with low thioguanine nucleotides (TGN) and unexpectedly high methylmercaptopurine (MeMP) levels. This predicts a lack of treatment efficacy and excess adverse events; however studies on the effect of thiopurine hypermethylation (TH) in adult IBD patients are lacking. Importantly, where identified TH can be circumvented with the use of low dose AZA/MP and allopurinol. The aim of this study is to characterise the impact of TH in IBD patients and determine if thiopurine metabolite profiles at week 4 of treatment can predict its occurrence to allow early combination treatment.

Methods 273 patients with IBD, who were anti-TNF-α therapy naïve and matched for red blood cell thiopurine-S-methyltransferase activity, were retrospectively identified. Of these 181 patients demonstrated average MeMP:TGN ratios <11 (at least 2 profiles between weeks 12 – 52) and were used as a control group that was compared to 92 patients with average MeMP:TGN ratios ≥11. Clinical outcomes were recorded for the first 12 months of therapy. A failure of AZA/MP as monotherapy was determined by 3 gastroenterologists with expertise in IBD. Thiopurine metabolite profiles were measured in 139 patients at week 4 of therapy and were compared to average metabolite profiles between 12 – 52 weeks of treatment.

Results The average MeMP:TGN ratios in those with and without TH were 21.0 and 2.52 respectively (p = <0.0001). The normalised dose of thiopurine was higher in patients with TH (1.91 vs. 2.09 mg/ kg; p = < 0.0001). Patients with TH were more likely to fail AZA/MP monotherapy as first line treatment during the first 12 months of therapy, in comparison with patients with normal methylation profiles (p = 0.0088, Gehan-Breslow-Wilcoxon Test). The difference in the number of treatment failures at 12 months was 15.7%. There was no difference in the occurrence of gastrointestinal intolerance, myelotoxicity or pancreatitis between groups, however there was an excess of hepatotoxicity with TH (p = 0.0006; OR = 8.058; 95% CI: 2.188 – 29.670; Fisher’s exact). TH was demonstrated in 83.7% of cases by 12 weeks of therapy. A MeMP : TGN level > 6.17 at week 4 accurately predicted TH (ratio ≥11) after 12 weeks of therapy (AUC = 0.839; p = < 0.0001; sensitivity = 75.4%; specificity = 88.4%).

Conclusion TH is associated with an excess of thiopurine treatment failures as monotherapy immunosuppression during the first 12 months of treatment. We also confirm its association with hepatotoxicity. A MeMP:TGN >6.17 may be useful in early identification of patients likely to benefit from combination treatment.

Disclosure of Interest None Declared.