Thiopurine drugs, including azathioprine, mercaptopurine, and thioguanine, are widely employed as immunosuppressants and chemotherapeutic agents. In immunosuppressive therapy, they are used to treat inflammatory bowel disease, myasthenia gravis, rheumatoid arthritis, and to prevent organ transplant rejection. As chemotherapeutics, they play a key role in managing acute leukemia and chronic myeloid leukemia. Despite their clinical utility, their use is constrained by severe, potentially life-threatening adverse drug reactions (ADRs), such as myelosuppression, alopecia, hepatotoxicity, and pancreatitis (Marinaki and Arenas-Hernandez, 2020). These toxicities are strongly linked to genetic polymorphisms in two key enzymes: thiopurine S-methyltransferase (TPMT) and nucleoside diphosphate-linked moiety X-type motif 15 (NUDT15) (Relling et al., 2019).

TPMT is primarily responsible for inactivating thiopurine drugs. Reduced or absent TPMT activity leads to elevated levels of thiopurine active metabolites, thereby increasing the risk of toxicity. Genetic polymorphisms in the TPMT gene significantly influence enzyme activity. In Western populations, the TPMT variants ∗3A and ∗3C account for 90% of cases with low enzyme activity and are associated with leukopenia (Meng et al., 2018). TPMT∗3A is composed of two single nucleotide polymorphisms (SNPs): rs1800460 (c.460G>A) and rs1142345 (c.719A>G). Unlike TPMT∗3A, which is defined by a combination of two SNPs, TPMT∗3C is composed solely of the single SNP variant rs1142345. In contrast to populations of European descent, loss-of-function TPMT alleles exhibit lower prevalence in East Asian populations. In Chinese cohorts, the c.719A>G (rs1142345) allele frequency typically ranges from ∼1 to 3%, with TPMT*3C (c.719A>G) as the predominant star allele (Mao et al., 2021; Zhou and Lauschke, 2022). By comparison, carrier prevalence of TPMT loss-of-function alleles in European populations is approximately 10% (Teml et al., 2009). Notably, the low prevalence of TPMT variants in East Asians suggests that alternative genetic factors contribute to the higher incidence of thiopurine-related ADRs in this population (Liang et al., 2016).

NUDT15 negatively regulates thiopurine activation, and loss-of-function variants lead to accumulation of cytotoxic metabolites (Man et al., 2019). The Clinical Pharmacogenetics Implementation Consortium (CPIC) included nine NUDT15 single-nucleotide polymorphisms in its 2018 dosing guidelines for thiopurines (Relling et al., 2019). The most frequent variant is NUDT15∗3 (rs116855232, c.415C>T) (Mao et al., 2021). Additionally, minor allele frequency (MAF) of NUDT15∗2 in East Asians is about 3%. It is important to note that NUDT15*2 is defined by two specific SNPs: rs116855232 (c.415C>T) and rs746071566 (c. 55_56insGAGTCG). This variant has been strongly associated with thiopurine toxicity (Yang et al., 2014; Zhang et al., 2018). It exhibits a higher prevalence in Asian populations compared to European or African populations (Yang et al., 2015; Khaeso et al., 2021). Unlike TPMT, NUDT15 variants show a distinct ethnic distribution, underscoring their critical role in predicting ADRs in East Asians.

In addition to TPMT and NUDT15, other genes involved in the thiopurine drugs metabolic pathway have been explored for their association with treatment-related adverse effects. For example, genetic variants at the ITPA locus (e.g., rs1127354 and rs7270101) are associated with reduced enzymatic activity, which may elevate toxicity risk through the accumulation of the potentially harmful metabolite thioinosine triphosphate (Moradveisi et al., 2019; Ali et al., 2023). Given their status as the most prevalent genetic variants implicated in thiopurine metabolism, TPMT c.719A>G (rs1142345) and NUDT15 c.415C>T (rs116855232) were selected for investigation in this study.

The frequencies of NUDT15 and TPMT variants exhibit significant ethnic diversity. Notably, there is a paucity of research on the NUDT15 c.415C>T polymorphism among the Chinese population, particularly the Han ethnic group. The objectives of the study were twofold: (1) to determine the allelic frequencies of these two variants in a Chinese cohort (predominantly Han ethnicity); and (2) to evaluate their comparative utility in predicting thiopurine drug toxicity, with the aim of optimizing healthcare resource allocation in China, a developing nation with a large population.

Previous studies have investigated the association between thiopurine drug-induced ADRs and genes including TPMT, ITPA, NUDT15, GST, MRP4, HGPRT, IMPDH, and XO. Among these, TPMT, NUDT15, and ITPA are the most extensively studied (Chen et al., 2021; Suzuki et al., 2023; Salazar et al., 2024). Notably, the U.S. Food and Drug Administration (FDA) recommends determining patient TPMT genotypes prior to drug administration, while the CPIC guidelines additionally advise assessing NUDT15 genotypes before initiating thiopurine therapy (Relling et al., 2019). Therefore, this study was intentionally designed to focus on NUDT15 and TPMT, given their well-established role as primary contributors to thiopurine toxicity in the study cohort, although the ITPA gene has been implicated in thiopurine-induced ADRs.

Azathioprine (AZA) acts as a prodrug of 6-mercaptopurine (6-MP) and undergoes nonenzymatic conversion to 6-MP within erythrocytes. Subsequently, 6-MP is metabolized into various derivatives by three key enzymes: xanthine oxidase (XO), TPMT, and hypoxanthine guanine phosphoribosyl transferase (HGPRT). The thiopurine metabolism pathway is depicted in Figure 1 (Matsuoka, 2020; Tanaka and Saito, 2021; Suzuki et al., 2023). Reduced TPMT activity leads to accumulation of 6-thioguanine nucleotides (6-TGN). Decreased NUDT15 activity increases thiopurine triphosphate levels. Inosine triphosphate pyrophosphatase (ITPA) converts 6-thioinosine triphosphate (6-TITP) to 6-thioinosine monophosphate (6-TIMP), and diminished ITPA activity is hypothesized to cause 6-TITP accumulation. All of these conditions could result in an increased risk of adverse drug reactions.

To date, over 40 TPMT alleles (TPMT∗2–∗41) have been identified in individuals with TPMT deficiency (Iu et al., 2017). The frequency of TPMT genetic polymorphisms varies significantly across ethnic groups, with an approximate prevalence of 3% in Asians—substantially lower than that in European populations (Fangbin et al., 2012; Matsuoka, 2020). The most common polymorphism is TPMT c.719A>G, which exhibited an overall prevalence of 2.36% in our Chinese cohort (predominantly Han ethnicity). The TPMT enzyme is central to thiopurine metabolism, and the TPMT∗3C polymorphism represents a risk factor for thiopurine intolerance (Cardoso de Carvalho et al., 2020). This variant induces protein instability and impairs TPMT enzymatic activity (Evans, 2004), leading to accumulation of thiopurine nucleoside active metabolites and subsequent cytotoxicity. Conversely, the NUDT15 enzyme dephosphorylates thiopurine triphosphate—the active metabolite incorporated into DNA—into its monophosphate form (Tanaka and Saito, 2021). NUDT15∗3 is recognized as a loss-of-function variant (Moriyama et al., 2016), causing elevated thiopurine triphosphate levels and exacerbating thiopurine-induced cytotoxic effects, including myelosuppression and alopecia. In our cohort, NUDT15 c.415C>T had an overall prevalence of 12.52% in the Chinese population (predominantly Han ethnicity).

During thiopurine treatment, the incidence of leukopenia ranges from 15% to 40% in Asian populations (Takatsu et al., 2009; Kim et al., 2010; Qiu et al., 2015), significantly higher than the approximately 3% reported in Western populations (Lewis et al., 2009; Sood et al., 2015). Severe leukopenia affects approximately 1% of Asian patients (Asada et al., 2016), underscoring the critical need for pharmacogenetic testing in these populations. As a developing East Asian nation with a population exceeding 1.4 billion, China faces regional disparities in economic development and limited public healthcare funding. From a health economics perspective, there is a pressing need to deliver cost-effective personalized medication guidance to patients within constrained financial resources.

In our cohort, the prevalence of TPMT c.719A>G was significantly lower than that of NUDT15 c.415C>T (2.36% vs. 12.52%). The positive detection rates for TPMT c.719A>G, NUDT15 c.415C>T, and both indexes testing were 4.55%, 23.47%, and 26.8%, respectively. Compared with testing NUDT15 c.415C>T alone, simultaneous testing of both polymorphisms only increased the positive rate by 3.33%. Additionally, only one patient was homozygous for TPMT c.719A>G. These findings suggest that single-locus testing for NUDT15 c.415C>T may be a clinically acceptable strategy in China, although the CPIC guideline for thiopurine dosing (Relling et al., 2019) still recommends comprehensive detection of all relevant polymorphisms.

The present study is subject to several limitations. The current study focused on NUDT15 c.415C>T and TPMT c.719A>G variants, but did not include the complete allele panel recommended by CPIC/AMP guidelines. This may lead to underreporting clinically relevant alleles and does not fully align with standardized testing protocols. It prioritized TPMT c.719A>G and NUDT15 c.415C>T due to their established clinical relevance in East Asian populations, where they represent the most frequently observed variants associated with thiopurine-induced toxicity (Relling et al., 2019; Mao et al., 2021). This may have resulted in incomplete characterization of genetic contributions to ADRs of thiopurine drugs. Due to the two-site design of our genotyping assay, inference of star alleles, which relies on haplotypic combinations of multiple variants, was not feasible in this study. Star-allele inference for TPMT and NUDT15 necessitates multi-locus haplotype analysis (e.g., TPMT*3A = c.460G>A+ c.719A>G; NUDT15*2 = c.415C>T + c.55_56insGAGTCG). Owing to our assay’s exclusive interrogation of c.719A>G and c.415C>T, we may have overestimated the prevalence of TPMT*3C (as some carriers could harbor the *3A haplotype) and NUDT15*3 (as some carriers may actually carry the *2 allele). Future investigations should incorporate all AMP/CPIC Tier-1 variants to enable accurate star-allele assignment and robust haplotype-based genotyping. Besides, the lack of metabolite data represents a limitation of the current study. Future investigations incorporating simultaneous measurement of thiopurine metabolites and genetic variants will help elucidate how allele status affects drug metabolism and, ultimately, clinical responses.

In summary, this study revealed that NUDT15 c.415C>T yielded a higher carrier rate than TPMT c.719A>G in this cohort (predominantly Chinese Han ethnicity). Compared with TPMT c.719A>G, NUDT15 c.415C>T demonstrated greater suitability for predicting thiopurine drug toxicity in Chinese patients. It is important to note that this study only interrogated two single variants. Broader genetic panels may alter the absolute carrier yields. These findings highlight the critical role of NUDT15 c.415C>T genotyping in optimizing precision therapy for thiopurine-based treatments.