Predicting Adverse Events to Thiopurines in IBD: Are We a Step Closer?

Abstract

Despite the recent approval of several advanced therapies in inflammatory bowel diseases (IBDs), thiopurines still hold an essential place in the management of IBD, particularly in resource-limited settings. They are not only helpful in preventing immunogenicity and augmenting the efficacy of antitumor necrosis factor agents but are also effective as monotherapy, particularly in ulcerative colitis.¹ However, the use of thiopurines is limited by a high incidence of side effects. A wide variety of adverse events including myelosuppression, hepatotoxicity, pancreatitis, gastro-intestinal (GI) intolerance, and flu-like illness have been reported with thiopurines. Adverse events lead to dose reduction or treatment discontinuation in 15% to 40% of the patients.² Some of the adverse events, notably leukopenia, can be predicted by genetic variants: variants in the thiopurine methyltransferase (TPMT) and nudix hydroxylase (NUDT 15) genes account for approximately 30% of leukopenia in individuals of European ancestry.³ However, the relative contribution of genetic variants to side effects varies in ethnically diverse populations. For instance, variations in TPMT are associated with an increased risk of severe immune suppression with thiopurines in Europeans. More than 90% of the cases are associated with 3 variants in the TPMT gene (TPMT*3A. TPMT*3A and TPMT*3A). Since its original identification, testing for TPMT has been routinely done in clinical practice prior to patients commencing azathioprine. This is often done by measuring TPMT enzyme activity as opposed to genotype. TPMT absolute deficiency occurs in 1 in 300 people of European origin, and a further 5% have low levels. However, studies from non-European populations showed that TPMT variation only accounted for a minor proportion of side effects. Instead, NUDT15 was originally identified in East Asians to contribute much more to the risk of severe immune suppression.⁴ A subsequent large UK study identified that NUDT15 variants also contributed to the risk of severe immune suppression in Europeans, albeit to a lesser extent than TPMT.³ However, these variants do not account for all cases of thiopurine-induced toxicity. Several other genes involved in thiopurine metabolism, including inosine triphosphatase (ITPA), glutathione s-transferase (GST), and aldehyde oxidase 1 (AOX1), have also been implicated in thiopurine-induced adverse drug reactions.^5–7 However, these studies are less well replicated, and furthermore, identification of these variants have largely been performed in adult cohorts, and data from pediatric population are scant.

In this issue of Inflammatory Bowel Diseases, Coelho et al seek to address this knowledge gap by performing a large, retrospective cohort study of 487 paediatric patients, of which 396 had been exposed to thiopurines.⁸ They investigated thiopurine-induced toxicity in relation to TPMT biochemical enzyme activity and whole exome sequencing (WES) data and conducted additional subgroup analyses to assess toxicity against variation in the TPMT and NUDT15 genes. Finally, they used an in-house technology metric called “GenePy” to assess the pathogenic burden across genes implicated in thiopurine metabolism and toxicity.⁹ The GenePy score for a gene reflects an individual’s burden of pathogenic variation for that gene.

The spectrum of adverse events and findings in the study by Coelho et al is broadly consistent with other studies but adds further novel information.⁸ In their cohort, myelosuppression was observed in 11%, GI intolerance in 11%, hepatotoxicity in 4.5%, pancreatitis in 1.8% and other adverse effects in 2.8%. TPMT enzyme activity was normal in 87.4%, intermediate in 12.3%, and deficient in 0.2%; 26% of patients with intermediate activity developed toxicity to thiopurines. Routinely genotyped TPMT alleles associated with defective enzyme activity were identified in 28 patients (7%); TPMT*3A (4.5%), *3B (1%), *3C (1.5%). Interestingly, the predictive ability of these variants was low, and only 6 patients (21%) developed toxic responses. Three rare TPMT alleles (*3D, *39 and *40), not assessed on routine genotyping, were identified in 3 patients, who all developed toxic responses. The missense variant p.R139C (NUDT15*3 allele) was identified in 4 patients, but only 1 developed toxicity. One patient with an in-frame deletion variant p.G13del in NUDT15 developed myelosuppression at low doses. The GenePy score identified a significant association for toxicity in the AOX1 and DHFR genes.

Because of the low correlation between TPMT and thiopurine-induced myelosuppression, there has been growing interest in identifying other gene variants that could help explain the toxicity. The work by Coelho et al is a significant addition to the literature in this regard, and the focus on a pediatric cohort is a particular strength of the study.⁸ The association of NUDT15 variants further reinforces the importance of this variant in individuals of European ancestry. This was first identified in a landmark Korean study that described NUDT15 p. Arg139Cys as strongly associated with early onset thiopurine-induced leukopenia among individuals of East Asian ancestry.¹⁰ A subsequent large European study showed that the incidence of NUDT15 variants is relatively low among people with European ancestry, but showed a significant association with leukopenia.³ For NUDT15, the estimated number of patients needed to genotype to prevent 1 patient from developing myelosuppression is 95 patients,³ and this highlights the need for its inclusion in routine testing panels. The use of GenePy sore to identify novel variants in AOX1 and DHFR genes is a novel finding from the study by Coelho et al.⁸ A previous study reported an association of a nonsynonymous SNP in the AOX1 gene with a lack of clinical response to azathioprine; however, no association was found between AOX1 polymorphism and adverse events in that study.¹¹ The current study showed that variants AOX1 and DHFR genes were associated with thiopurine-induced toxicity in 3 patients who were TPMT and NUDT15 normal metabolizers.

It is unclear if the findings from this study are likely to lead to immediate changes in clinical practice. Despite the identification of novel pharmacogenomic variants and the relatively low cost of testing, the adoption of genetic testing in IBD clinical practice remains largely restricted to TPMT testing alone in the United Kingdom and many other healthcare settings. Positive physician attitudes and education are necessary if more recently identified variants are to become a part of routine healthcare, but currently, physician attitudes towards pharmacogenetic testing vary. In a survey of clinicians, approximately 50% did not know that genotype data are useful for prescribing decisions.¹² Indeed, only a small proportion of physicians report receiving any pharmacogenetic education.¹³ Notwithstanding this, studies among other specialists report that 50% to 75% of clinicians feel that pharmacogenetic testing is important in predicting adverse responses to treatment.¹⁴ Further, large-scale studies in ethnically diverse populations allied with physician and patient education are paramount to enable the widespread adoption of pharmacogenomics in clinical practice.

Author Contributions

M.T.S. drafted the initial version of the article. S.S. and M.P. edited the article, and both authors approved the final version of the article.

Conflicts of Interest

M.P. reports personal fees from Janssen and Takeda outside the submitted work and grants from AstraZeneca, Galapagos, Gilead, and Pfizer outside the submitted work. S.S. has received speaker fees from MSD, AbbVie, Dr Falk pharmaceuticals, Takeda, Janssen and Celltrion and received educational grants from Takeda and Janssen and is an advisory board member for AbbVie, Dr Falk pharmaceuticals, Celltrion, Janssen, Takeda and Vifor pharmaceuticals. M.T.S. declare no competing interests.

References