TOWARDS PREDICTIVE PHARMACOGENETIC PROFILING FOR AZATHIOPRINE TREATMENT: CHARACTERIZATION OF SNPS IN RELEVANT DRUG METABOLISM GENES

Azathioprine (AZA) is one of the most widely used immunosuppressant drugs for the treatment of many inflammatory diseases, in particular autoimmune diseases where it is often administered as a steroid-sparing drug. It blocks T- and B-cell proliferation through incorporation of its active metabolites, 6-thioguanine nucleotides, into DNA. AZA is mainly catabolised through the thiopurine S-methyltransferase (TPMT) and xanthine oxidase (XO) pathways. While 70% of patients respond to AZA, 30% do not respond or show intolerance to the drug. Variants of genes coding for TPMT or XO, or other gene members of either pathway, which result or not in mutated enzymes, are believed to influence individual responses to AZA. With the aim of uncovering possible genotype-phenotype correlations that would help personalize AZA treatment through response-specific SNP profiles, we have analyzed TPMT and XDH (the gene coding for XO) for SNPs in 71 Italian patients in the context of their response to AZA (intolerant patients, n = 25; unresponsive patients, n = 16; and responsive patients, n = 30). We confirm the presence of known intronic and exonic TPMT and XDH polymorphisms not correlating with particular AZA responses. We identified two new intronic polymorphisms, G420-4A in TPMT and T652-21A in XDH, each in one intolerant patient. Whether or not intolerance is related to these mutations in the patients must be further elucidated. In XDH, we detected two novel non-synonymous mutations (c.G1004A, p.Arg335His; c.C2891T, p.Thr964Ile), one new synonymous mutation (c.G1194A, p.Leu398Leu), and one novel non-synonymous polymorphism (c.C1167T, p.Ala556Val). The missense mutation, p.Arg335His, is likely to have an effect on the structural conformation of the FAD-binding domain, thereby destabilizing the protein structure. This could therefore lead to a decreased XO activity, for example through increased protein degradation or through deficient binding of FAD, hence reduced efficiency of the enzyme function, and thereby intolerance to AZA as observed. The other non-synonymous mutation, p.Thr964Ile, occurred in the molybdopterin domain; it was found in one responsive patient, suggesting that this mutation did not modify the structure of the domain sufficiently to affect the enzymatic activity. The non-synonymous polymorphism, p.Ala556Val, causes a change from Ala to Val at position 556 in the connection segment between the FAD-binding and molybdopterin domains, which is unlikely to have a significant effect on the protein. This XDH polymorphism was found in one intolerant, one unresponsive, two responsive patients, and 3/100 healthy controls; it is therefore unlikely to be of significance in pharmacogenetic profiles predictive of AZA responses. The silent mutation, p.Leu398Leu, identified in an intolerant patient should not cause functional impairment of XO. Linkage disequilibrium (LD) of TPMT SNPs and haplotype analysis thereof demonstrated a new haplotype designated TPMT*3E; it comprises the previously reported mutations of the TPMT*3A allele associated with intolerance to AZA and the intronic T140+114A SNP. TPMT*3E was detected in four of the 25 AZA-intolerant patients and was not observed in unresponsive or responsive patients. The association of TPMT*3E with AZA intolerance, and its frequency, must be ascertained in larger, ethnically different cohorts. Nevertheless, in view of the highly significant association (Psim = 0.037) between TPMT*3E and AZA intolerance in our study, this haplotype should be taken into account when considering AZA treatment. LD analysis of XDH identified four different haplotype blocks, one of which was significantly associated with intolerance in our cohort (Psim = 0.017). This block includes five SNPs, one intronic and four located in the 3' untranslated regions (3' UTRs); these were previously described as single SNPs, but were not analyzed in the context of response to AZA. It is unclear how this haplotype results in intolerance to AZA. One possibility is that the SNPs affect the regulation of protein expression through alteration of the target sites for microRNAs that interact with the 3' UTRs to regulate the expression of mRNAs. This could therefore lead to AZA intolerance in our patients, through in an increase in down-regulation of XDH mRNA and thereby XO expression. In this work, we have demonstrated new haplotypes that should be taken into consideration in pharmacogenetic profiling for AZA. In particular, SNPs in the XDH have been poorly investigated thus far in the context of response to AZA; the new XDH haplotype is of major interest in the establishment of pharmacogenetic profiles that will permit prediction of the type of response to AZA, in particular to prevent life-threatening side effects. It should be further studied in the context of its association with other response-defining haplotypes or SNPs of TPMT and other AZA metabolism pathway genes.