Pharmacogenomics of Thiopurine Drugs: A Bench‐To‐Bedside Success Story in Thailand

ABSTRACT

Thiopurine drugs are the cornerstone treatment for many diseases such as acute lymphoblastic leukemia (ALL), organ rejection, inflammatory bowel disease (IBD), systemic lupus erythematosus (SLE), rheumatoid arthritis (RA) and other autoimmune diseases. However, their clinical use faces limitations due to the drug‐induced adverse effects, including myelosuppression. Several genetic associations have been evaluated for their association with these adverse drug reactions. TPMT and NUDT15 polymorphisms have emerged as important clinical markers for predicting and optimizing the safety and effectiveness of thiopurine drugs. The bench‐to‐bedside approach of exploring and assessing the genetic associations of TPMT and NUDT15 variants and the new LC‐MS/MS methods for evaluating TPMT is a step forward in the advancement of precision medicine of thiopurine drugs. In Thailand, TPMT and NUDT are routinely genotyped in some hospitals to guide the prescription of thiopurine drugs for optimizing the safety or effectiveness of these drugs. However, the composite effects of these genetic variants remain unexplored at the global scale. Proper large‐scale studies with multi‐ethnic patients can provide a clear understanding of the TPMT/NUDT15 association and would pave the way towards the optimization of thiopurine drugs to achieve precision medicine.

1. Introduction to Thiopurine Drugs

Thiopurines were the breakthrough drug discovery by Gertrude Elion and George Hitchings in 1950. They were awarded the Nobel Prize in Chemistry in 1988. Three thiopurines, such as thioguanine, mercaptopurine, and azathioprine, were discovered. They remain the cornerstone treatment for many diseases, such as leukemia, organ rejection, inflammatory bowel disease (IBD) [1], systemic lupus erythematosus, and rheumatoid arthritis. Azathioprine is a prodrug; after a complex metabolic process, it is converted into 6‐mercaptopurine (6‐MP) by glutathione S transferase (GST); however, some studies show that this process is non‐enzymatic [2]. This is converted into the cytotoxic 6‐thioguanine nucleotide. This cytotoxic metabolite interferes with the S phase of the cell cycle, in which DNA synthesis occurs. This metabolite, instead of real purine bases, is a fraudulent base pair that will be incorporated into DNA and RNA and cause genetic mutations and initiate apoptosis [3]. In this pathway, 6‐MP is first converted to 6‐thioinosine monophosphate (6‐TIMP), which is then converted to thioxanthosine monophosphate (TXMP), followed by thioguanosine monophosphate (TGMP). The TGMP is then converted to thioguanosine diphosphate (TGDP), thio‐deoxyguanosine diphosphate (TdGDP), followed by the thio‐deoxyguanosine triphosphate (TdGTP). The end metabolite TdGTP enters into DNA to damage it by genetic mutations and initiate apoptosis. In another way, TGDP is converted to thioguanosine triphosphate (TGTP), which then enters into RNA to damage it and initiate apoptosis as well. In addition, this metabolite activates the Rac1 pathway in the T cell and initiates apoptosis [4]. The6‐TIMP also inhibits de novo purine synthesis [5]. Details of the mechanism of action of thiopurine drugs are shown in (Figure 1).

FIGURE 1.

Mechanism of Action of Thiopurines. 6‐MP, 6‐mercaptopurine; 6‐TG, 6‐thioguanine; 6‐TIMP, 6‐thioinosine monophosphate; ABCC4/5, ATP‐binding cassette transporters; AZA, azathioprine; GSTA1/2, GSTM1, glutathione S‐transferases; meMP, methylmercaptopurine; meMPR, methylmercaptopurine ribonucleotide; meTGMP, methylthioguanosine monophosphate; meTIMP, methylthioinosine monophosphate; MPR, mercaptopurine ribonucleotide; NUDT15, nudix hydrolase 15; PRA, phosphoribosylamine; PRPP, 5‐phospho‐D‐ribose‐1pyrophosphate; SLC28A2/3, CNT2 and CNT3 transporters; SLC29A1/2, ENT1 and ENT2 transporters; TdGDP, thio‐deoxyguanosine diphosphate; TdGTP, thio‐deoxyguanosine triphosphate; TGDP, thioguanosine diphosphate; TGMP, thioguanosine monophosphate; TGTP, thioguanosine triphosphate; TPMT, thiopurine S‐methyltransferase; TXMP, thioxanthosine monophosphate; XDH, xanthine dehydrogenase.

Interindividual variability of PK/PD is wide for thiopurine drugs; for example, azathioprine is a narrow therapeutic index drug found to be ineffective in 15% of patients, while on the other extreme, 9%–28% cause serious adverse drug reactions. Hence, for most of the thiopurine drugs, individualized dosing is required [6]. Orally, this drug is well absorbed with a 30% protein binding capacity. Azathioprine and 6‐mercaptopurine reach their peak plasma concentration in 1–2 h and are primarily excreted through urine [7]. Hepatotoxicity is the most common adverse drug reaction (20%) with thiopurine drugs, followed by leukocytopenia (4%). Other side effects include nausea, vomiting, pancreatitis, and allergic reactions [6, 8, 9].

AZA and MP are both prodrugs that are converted to their active metabolites through a complex metabolic pathway. Their bioavailability ranged from 27% to 83% [10] and 5% to 37% [11], respectively, for AZA and 6‐MP after oral administration. This 6‐mercaptopurine undergoes extensive first‐pass metabolism in the presence of xanthine oxidases (XO), which are highly expressed in the intestinal and liver cells but not in the blood cells and form thiouric acid [12]. Another enzyme, aldehyde oxidase, converts AZA directly into thiouric acid without generating 6‐MP [13].

Despite studies showing the association between the genetic variants of TPMT and NUDT15 and the adverse drug reactions (ADRs) of thiopurine drugs, especially with leukopenia, we cannot blame these defective genes as a whole for these reactions. Moreover, studies show that leukopenia developed in patients who even carried the wild type TPMT genotype [14]. Hence, many non‐genetic factors might be associated with ADR. According to some studies males have significantly higher TPMT activity than females do, whereas other studies found no such association [15, 16, 17].

Likewise, smoking habits were found to have influenced the TPMT level; smokers exhibited a significantly elevated TPMT level compared to non‐smokers. Though the mechanism behind it remained unexplained, increased levels of S‐adenosylmethionine, the methyl group donor to the sulfur of the thiopurine drugs, have been reported when human lung epithelial‐like cells are exposed to cigarette smoke [17]. On the other hand, children were found to have lower TPMT activity than adults [18]. Other comorbidities play an important role in altering the level of TPMT. For example, a study on systemic lupus erythematosus and systemic vasculitis patients taking AZA reported higher TPMT levels (29.37 and 26.24 ng/mL, respectively) compared to the healthy control (22.71 ng/mL) [19]. Similar study on Acute Lymphoblastic Leukemia (ALL) patients reported the TPMT activities to be significantly higher during therapy (27.3 nmol 6‐mMP g⁻¹ Hb h⁻¹, p = 0.0002) compared to the control group (22.8 nmol 6‐mMP g⁻¹ Hb h⁻¹) [20]. In addition to all of this, the polypharmacy‐induced drug interactions can influence the activity of TPMT. In order to increase the effectiveness of a lower dose of thiopurine, allopurinol, antigout, and a xanthine oxidase inhibitor are typically prescribed to IBD patients [10, 14]. But the same medication can interact and lead to ADRs and thiopurine intolerance.

2. Importance of TPMT , NUDT and ITPA Genetic Polymorphisms for Consideration

The metabolic pathway of thiopurines has been presented in Figure 2. The TPMT gene in humans codes for the enzyme known as thiopurine methyltransferase, also known as thiopurine S‐methyltransferase (TPMT). This enzyme metabolizes thiopurine drugs via S‐adenosyl‐L‐methionine as the S‐methyl donor. Since it methylates both mercaptopurine (MP) and thioguanine, the levels of active thioguanine nucleotide (TGN) metabolites and TPMT activity are inversely correlated. For instance, patients with two homozygous deficient alleles are found to have severe myelosuppression; this becomes moderate in patients with a heterozygous inactive allele; however, the risk is low in the wild type [21, 22]. According to the Clinical Pharmacogenetics Implementation Consortium (CPIC) the actionable TPMT alleles are TPMT*2, *3A, *3B, *3C, *4, *7 and *8. These SNPs impact certain drug therapies based on the carriage of certain diplotypes (outlined in Table 1) [24].

FIGURE 2.

Metabolic pathway of Azathioprine and 6‐mercaptopurine.

TABLE 1.

Global distribution of TPMT genetic polymorphisms.

Frequencies of TPMT alleles in biogeographical groups
TPMT allele a	Phenotypes and their implications	Thailand	Afr. American	Central/South Asian	East Asian	European	Latino	Near Eastern	SSA
*1/*1	NM (Lower TGN metabolite concentration; higher MeTIMP concentration for AZA and 6‐MP; Normal risk of ADRs)	0.95	0.92	0.981	0.979	0.95	0.94	0.96	0.92
*1/*2	PM (Extremely high TGN metabolite concentration; no MeTIMP metabolites for AZA and 6‐MP; Extremely elevated risk of ADRs)	0.0	0.005	0.0002	0.00008	0.0020	0.003	0.007	0.0
*1/*3A	IM (Moderate to high TGN metabolite concentration; low MeTIMP concentration for AZA and 6‐MP; Elevated risk of ADRs)	0.0	0.008	0.004	0.0003	0.033	0.041	0.013	0.0016
*1/*3B		0.0	0.0	0.001	0.0	0.002	0.002	0.004	0.0
*1/*3C		0.05	0.024	0.011	0.0163	0.004	0.005	0.009	0.052
*1/*4		0.0	0.0	0.0	0.0	0.00005	0.0	0.0	0.0
*1/*11		0.0	0.0	0.0	0.0	0.00024	0.0	0.0	0.0
*1/*14		0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
*1/*15		0.0	0.0	0.0	0.0	0.000045	0.0	0.0	0.0
*1/*23		0.0	0.0	0.0	0.0	0.000026	0.0	0.0	0.0
*1/*29		0.0	0.0	0.0	0.0002899	0.0	0.0	0.0	0.0

Abbreviations: 6‐MP, 6‐mercaptopurine; ADR, adverse drug reaction; Afr., African; AZA, azathioprine; CARIB, Caribbean; EU, European; IM, Intermediate metabolizer; MeTIMP, metabolites of thiopurine methyltransferase; NM, Normal metabolizer; PM, Poor metabolizer; SSA, Sub‐Saharan African; TGN, thioguanine nucleotides.

^a The data presented here were retrieved from https://cpicpgx.org/genes‐drugs [23].

Defective TGTP degradation mediated by the low enzymatic activity of NUDT15 results in more TGTP available for incorporation into DNA to generate DNA‐TG, leading to DNA strand breaks and apoptosis, and thus mitigating toxicity [25]. The most common NUDT15 variants are c.415C>T, c.416G>A, c.52G>A, and c.55_56insGAGTCG, which have been reported to be associated with mercaptopurine‐induced myelotoxicity [26, 27]. The impact of these SNPs on enzyme functionality has been outlined in Table 2. Inosine triphosphate pyrophosphatase (ITPA) is another gene said to be associated with thiopurine therapy. The ITPA enzyme catalyzes the conversion of inosine triphosphate to inosine monophosphate. A defective ITPA gene causes an accumulation of the non‐canonical nucleotides in the cells and their subsequent incorporation into nucleic acids. ITPA 94C>A and IVS2 + 21A>C are the most common SNPs said to be linked with ITPA deficiency [29].

TABLE 2.

Global distributions of NUDT15 genetic polymorphisms.

Alleles a	Af.Am/Af. carib	Central/SA	EA	Eur.	Latin	Near East	Oceanian	SSA	Enzyme functionality	Thai population
NUDT15*1	0.0	0.93	0.87	0.99	0.93	0.0	0.0	0.0	Normal	8.55 [28]
NUDT15*2	0.0	0	0.03	0	0.03	0.0	0.0	0.0	No	0.02 [28]
NUDT15*3	0.0	0.0670	0.06	0.002	0.007	0.0	0.0	0.0	No	0.07 [28]
NUDT15*9	0.0	0.0005	0.0	0.001	0.0003	0.0	0.0	0.0	No	—

Abbreviations: Afr., African; Am, American; CARIB, Caribbean; Eur, European; SSA, Sub‐Saharan African.

^a The data presented here were retrieved from https://cpicpgx.org/genes‐drugs [23].

3. Global Distribution of TPMT and NUDT15

The genetic mutations of the TPMT allele differ greatly between ethnicities. TPMT *3A is the most common defective allele found in European and Latin American populations, with frequencies of 2%–4% [30, 31]. In Europe, the highest TPMT *3A frequencies were found in Greenland (8.1%) [32], and the UK (4.5%) [33]. P. Zalizko et al. studied the frequency of TPMT genetic mutations in the Latvian population. The most frequent polymorphisms they reported were in the TPMT *1/*3A (5.3%) genotype [34]. However, in some European countries, the frequencies were found to be lower, such as Croatia (1.9%) [35]. On the other hand, in Latin America, frequencies were highest in Brazil (up to 3.9%) [36] and Argentina (3.1%) [37].

On the other hand, in Asian and African populations, TPMT*3A is very rare, and instead TPMT*3C is the predominant allele [38, 39]. These allele‐specific interethnic differences are even more striking in Sub‐Saharan Africa, where TPMT*3C is highly abundant in Ghana (7.6%) [33] and Kenya (5.4%) [40], but relatively rare in North African populations (1.3%) [41]. In Asia, frequencies of TPMT*3C range between 0.9% in Koreans, 1.3%–3% in Chinese populations, and 0.8%–2.8% across South Asia [42, 43, 44]. In Singapore, 1585 patients participated in a population‐based study. Their results showed the heterozygous TPMT alleles were present among 4.7% of Chinese (n = 30/644), 4.4% of Malays (n = 24/540), and 2.7% of Indians (n = 11/401) [45]. Another study was conducted in three different ethnicities, such as 199 British Caucasians, 99 British Southwest Asians, and 192 Chinese individuals. The prevalence of the defective allele of TPMT was found to be 10% in Caucasians, 2% in Southwest Asians, and 4.7% in Chinese [46]. Another Indian study estimated the prevalence of TPMT gene mutations in autoimmune disorders and assessed myelosuppression. Among the 176 included patients, the TPMT mutation (TPMT *1/*3C) was found to be around 1.13%, and only one patient carried TPMT *1/*3A [47]. Another South Indian study, which included 326 participants, found the estimated genotype frequency for homozygous TPMT*1/*1 to be 97.24%, for heterozygous TPMT*1/*2 and TPMT*1/*3B, 0.61% each, and for heterozygous TPMT*1/*3C, 1.53% (presented in Table 1, the data was retrieved from the website of CPIC) [23, 48].

On the other hand, in Asian and African populations, TPMT*3A is very rare, and instead TPMT*3C is the predominant allele [38, 39]. These allele‐specific interethnic differences are even more striking in Sub‐Saharan Africa, where TPMT*3C is highly abundant in Ghana (7.6%) [33] and Kenya (5.4%) [40], but relatively rare in North African populations (1.3%) [41]. In Asia, frequencies of TPMT*3C range between 0.9% in Koreans, 1.3%–3% in Chinese populations, and 0.8%–2.8% across South Asia [42, 43, 44]. In Singapore, 1585 patients participated in a population‐based study. Their results showed the heterozygous TPMT alleles were present among 4.7% of Chinese (n = 30/644), 4.4% of Malays (n = 24/540), and 2.7% of Indians (n = 11/401) [45]. Another study was conducted in three different ethnicities, such as 199 British Caucasians, 99 British Southwest Asians, and 192 Chinese individuals. The prevalence of the defective allele of TPMT was found to be 10% in Caucasians, 2% in Southwest Asians, and 4.7% in Chinese [46]. Another Indian study estimated the prevalence of TPMT gene mutations in autoimmune disorders and assessed myelosuppression. Among the 176 included patients, the TPMT mutation (TPMT *1/*3C) was found to be around 1.13%, and only one patient carried TPMT *1/*3A [47]. Another South Indian study, which included 326 participants, found the estimated genotype frequency for homozygous TPMT*1/*1 to be 97.24%, for heterozygous TPMT*1/*2 and TPMT*1/*3B, 0.61% each, and for heterozygous TPMT*1/*3C, 1.53% (presented in Table 1, the data was retrieved from the website of CPIC) [23, 48].

Recent studies have shown that nudix hydrolase 15 (NUDT15) polymorphisms have a higher mutation frequency in Asian populations compared to European and African populations NUDT15 genetic polymorphism is more common in Asian populations than in Caucasian populations. The phenotypes of more than 10 variant alleles of NUDT15 have been extensively characterized. The common variant alleles are NUDT15*2 (c.36_37insGGAGTC; c.415C > T), NUDT15*3 (c.415C > T), NUDT15*4 (c.416G > A), NUDT15*5 (c.52G > A), and NUDT15*6 (c.36_37insGGAGTC). Several lines of evidence strongly suggest that NUDT15 variants, particularly NUDT15*3, NUDT15*5, and NUDT15*6, are strongly associated with myelosuppression caused by 6‐MP or 6‐MP derivatives [49, 50, 51]. A study on 129 patients reported the prevalence of NUDT15*2 to be 14% in Chinese, 12% in Malays and 8% in Indians. Similarly, NUDT15*5 has a frequency of around 1% in Chinese and 4% in Malays. NUDT15*6 on the other hand, had a prevalence of 7%. Interestingly, NUDT15*4 was absent in the Asian population [52]. Another study on the Korean population reported the allele frequencies of 86.0% for NUDT15*1, and 4.3%, 7.6%, 1.4% and 0.7% respectively for *2, *3, *5 and *6 [53]. Among Japanese the prevalence of different genotypes has been estimated as 18.9%, 7.4%, 4.2% and 1% for NUDT15*1/*3, *3/*3, *1/*5 and *5/*5 respectively [49]. The prevalence of NUDT15*9 and *3 in the European population was reported as 2.9% and 1.2% in a study [26]. Global distributions of different NUDT15 variants have been presented in Table 1. Overall, there is a discrepancy in the allele frequencies of NUDT15 and TPMT in Asian and other world populations, indicating different clinical approaches may be needed for different populations. This again raises the question about cost‐effectiveness of such pre‐emptive genotyping prior to thiopurine therapy. But there is a paucity in such studies, hindering the mass adoption of pharmacogenomics in routine practice. Large‐scale cost‐effectiveness studies focusing on particular populations may provide better insight and guide making more informed clinical decisions.

4. Distribution of TPMT and NUDT Genetic Polymorphisms in the Thai Population

The prevalence of TPMT genetic variants was examined in 164 patients by a Thai study in 2004. by using PCR‐restriction fragment length polymorphism (PCR‐RFLP) and allele‐specific polymerase chain reaction (AS‐PCR) (RFLP). Seven of 164 (4.26%) patients had TPMT variants. Six patients had TPMT*3C/*1, and one patient had TPMT*3A/*1. This is the first report of TPMT*3A polymorphism in the Thai population. The percentage frequencies of genotypes TPMT*3C/*1 and TPMT*3A/*1 were 3.65% and 0.61%, respectively [54]. Another study in Thailand investigated the genetic variants of both TPMT and NUDT15 in 178 patients using the TaqMan SNP genotyping assay and direct sequencing methods. The frequency of TPMT*3C was 0.034. The most prevalent NUDT15 variant, NUDT15*3, has an allele frequency of 0.073, while those of NUDT15*2, NUDT15*5, and NUDT15*6 variants were 0.022, 0.011, and 0.039, respectively (presented in Tables 1 and 2; the data was retrieved from the website of CPIC) [23, 28].

A search across PubMed with different combinations of the keywords like, thiopurine, 6‐mercaptopurine, thioguanine, azathioprine, TPMT, NUDT15, ITPA, IBD, ALL, organ transplant, autoimmune diseases was performed to select the studies included in this review. We only considered the clinical studies evaluating the genetic associations of the aforementioned genes with the adverse effects of thiopurine drugs and excluded studies such as review articles, meta‐analysis, case reports, and case series.

5. Thiopurine Drugs in Inflammatory Bowel Disease (IBD): Optimizing the Safety Through Pharmacogenomics Consideration

Inflammatory bowel disease (IBD) refers to a set of chronic inflammatory disorders of the gastrointestinal tract and primarily comprises two types of remitting and relapsing diseases, namely ulcerative colitis (UC) and Crohn's disease (CD). Though thiopurines are the widely used therapy for the maintenance of IBD, serious adverse effects associated with them limit their extensive clinical application and ask for close monitoring. Myelosuppression is the key safety concern alongside other adverse effects including pancreatitis, hepatotoxicity, and the risk of lymphoma and skin cancer with long‐term use. Leukopenia induced by these drugs is a relatively common and potentially fatal adverse effect with a cumulative prevalence of 7% and a 1% mortality rate in the individuals developing myelotoxicity. Genetic variation in TPMT is well‐known for increasing 6‐thioguanine nucleotide metabolite levels contributing to the development of myelotoxicity and therefore, pre‐emptive genetic testing for TPMT has been adopted for routine clinical practice to identify patients at risk of drug‐induced leukopenia and adjusting doses [55]. Some genetic associations for the thiopurine drugs induced adverse effects have been summarized in Table 3. A meta‐analysis comparing 14 studies comprising 2206 IBD patients explored the association of TPMT polymorphisms and the adverse effects related to thiopurine therapy and revealed a strong association with overall adverse events (Odds ratio, OR = 3.36; 95% Confidence interval, 95% CI = 1.82–6.19) and bone marrow toxicities (BMT) (OR = 6.67; 95% CI = 3.88–11.47). However, they noted no association between polymorphism in TPMT and pancreatitis, hepatotoxicity, gastric intolerance, skin reactions and flu‐like symptoms [78]. However, this association falls short for the Caucasian population as drug‐induced leukopenia has been observed in many Caucasian patients with normal TPMT activity and it could explain only around 25% of the ADRs. However, this study utilized an indirect method of predicting the metabolizer status from the genotype [79]. Additionally, there is a discrepancy in the prevalence of TPMT polymorphism in individuals with European ancestry (around 10%) and the Asian population (only around 1%), which cannot explain the more frequent occurrence of leukopenia in Asians (around 15%) than in Europeans (around 4%) indicating that other genetic factors may be involved, indicating other genetic factors might be associated in the process [80, 81, 82].

TABLE 3.

Significant genetic associations for different thiopurine drug‐induced adverse effects in different diseases.

Gene	Variants	Country/population	Disease condition(s)	Adverse effects	OR/HR (95% CI)	p	Drug	References
NUDT15	c.415C>T	China	Systemic lupus erythematosus, Sjögren's syndrome, vasculitis, scleroderma, dermatomyositis, connective tissue disease, IgG4 related diseases, autoimmune hepatitis	Leukopenia	7.59 (3.16–18.21)	1.79 × 10−7	AZA	[56]
				Early leukopenia	8.85 (3.64–21.53)	1.12 × 10−7
			Rheumatoid arthritis, systemic lupus erythematosus	Myelotoxicity	5.191 (1.654–16.291)	0.005		[57]
			Autoimmune hepatitis	Leukopenia	20.41 (7.84, 53.13)	< 0.00001		[58]
			Generalized eczema, atopic dermatitis, psoriasis, bullous pemphigoid, pemphigus foliaceus, dermatomyositis, systemic lupus erythematosus	Leukopenia	9.383 (1.32–66.96).	—		[59]
			Autoimmune hepatitis	Myelosuppression	7.5 (3.08–18.3)	8.26 × 10−7		[60]
				Neutropenia	8.05 (2.96–21.9)	3.54 × 10−6
			Acute lymphoblastic leukemia	Leukopenia	3.617 (1.377–9.051)	0.009	6‐MP	[61]
				Early‐onset leukopenia	9.63 (2.764–33.514)	3.75 × 10−4
			Inflammatory bowel disease	Leukopenia	7.663 (1.893–31.023)	0.004	AZA	[62]
		Chinese, Indian, Malay, others		Leukopenia	22.9 (5.17–101.4)	3.71 × 10−5	AZA, 6‐MP	[52]
				Neutropenia	13.4 (3.30–54.2)	2.79 × 10−4
		Indian	Acute lymphoblastic leukemia	Early hematological toxicity	6.51 (1.27–65.16)	0.01	6‐MP	[50]
			Inflammatory bowel disease	Leucopenia	19.35 (11.55 32.42)	< 0.0001	AZA	[63]
				Neutropenia	21.41 (12.25–37.41)	< 0.0001
		Japanese	Acute lymphoblastic leukemia	Leucopenia	7.20 (2.49–20.80)	< 0.0001	6‐MP	[64]
				Leukopenia	2.79 (1.80–4.31)	4.41 × 10−6		[49]
			Inflammatory bowel disease	Early leukopenia	28.4 (9.78–82.3)	4.38 × 10−15	AZA, 6‐MP	[65]
		Korean	Myasthenia gravis, chronic inflammatory demyelinating polyneuropathy, neuromyelitis optica, GI discomfort, vasculitis	Leukopenia	11.844 (3.894–36.024)	1.33 × 10−5	AZA	[66]
			Inflammatory bowel disease	Early leukopenia	35.63 (22.47–56.51)	4.88 × 10−94		[67]
		Thailand	Acute lymphoblastic leukemia	Myelosuppression (2nd month)	7.44 (1.3–42.63)	0.01	6‐MP	[68]
				Myelosuppression (4th month)	14.5 (3.41–61.63)	< 0.001
				Myelosuppression (6th month)	14.44 (2.89–72.26)	< 0.001
				Neutropenia (in 3 months)	12 (3.781–38.085)	< 0.001		[51]
				Early myelotoxicity (weeks 1–8)	17.862 (4.198–75.992)	9.5 × 10−5		[69]
		UK and Other	Inflammatory bowel disease	Myelosuppression	—	1.8 × 10−4	AZA, 6‐MP	[26]
	rs746071566			Myelosuppression	38.2 (5.1–286.1)	1.3 × 10−8
	c.3637insGGAGTC	Chinese, Indian, Malay, others	Inflammatory bowel disease	Leukopenia	5.95 (1.29–27.5)	0.022	AZA, 6‐MP	[52]
	NUDT15*3	China	Systemic lupus erythematosus, Behcet's disease, rheumatoid arthritis, Sjogren's syndrome, dermatomyositis, undifferentiated connective tissue disease, scleroderma, multiple chondritis, ankylosing spondylitis, antineutrophil cytoplasmic antibody‐associated vasculitis, necrotic myopathy	Leukopenia	8.374 (3.01–23.30)	4.55 × 10−5	AZA	[70]
	NUDT15*4			Gastrointestinal effects	5.714 (1.56–20.95)	0.01
	NUDT15*5			Erythropenia	9.333 (2.96–29.47)	4.04 × 10−4
	NUDT15*6			Hypochromia	13.18 (4.15–41.87)	3.66 × 10−5
	NUDT15*7			Thrombocytopenia	20.13 (3.40–119.18)	0.001
	NUDT15*6			Hypochromia	17.77 (2.72–116.67)	0.006
ITPA	c.94C>A	Bangladesh	Acute lymphoblastic leukemia	Neutropenia	7.68 (2.21–26.61)	—	6‐MP	[71]
				Hyperbilirubinemia	4.73 (1.39–16.07)	—
				Fever	6.9 (1.99–23.91)	—
	94C>A + 138G>A	Iran		Leucopenia	9.6 (1–80.7)	0.03		[29]
				Neutropenia	10.3 (1.2–86)	0.04
				ALT hepatotoxicity	21 (4–103.3)	< 0.001
				AST hepatotoxicity	13.5 (3–61.7)	0.001
	94C>A + IVS3 + 101G>A			Leucopenia	8.2 (0.97–70)	0.05
				Neutropenia	8.8 (1.04–74.9)	0.04
				ALT hepatotoxicity	15.4 (3–75.9)	0.001
				AST hepatotoxicity	9.6 (2–44.6)	0.004
	rs41320251	USA		Neutropenia	2.98 (1.21–7.35)	0.018		[72]
TPMT	*3C	China	Systemic lupus erythematosus, Behcet's disease, rheumatoid arthritis, Sjogren's syndrome, dermatomyositis, undifferentiated connective tissue disease, scleroderma, multiple chondritis, ankylosing spondylitis, antineutrophil cytoplasmic antibody‐associated vasculitis, necrotic myopathy	Leukopenia	—	0.048	AZA	[67]
		Thailand	Kidney transplant	Myelosuppression	14.18 (3.07–65.40)	< 0.005	AZA	[73]
			Acute lymphoblastic leukemia	Leukopenia (1–8 weeks)	—	0.031	6‐MP	[74]
				Neutropenia (1–8 weeks)	—	0.014
				Thrombocytopenia (9–24 weeks)	—	0.014
	*1, *2, *3A, *3C, *9, *21, *33, *34	UK		Neutropenia	—	0.02		[75]
				Thrombocytopenia	—	< 0.001
	rs11969064	UK and Other	Inflammatory bowel disease	Myelosuppression	2.3 (1.7–3.1)	5.2 × 10−9	AZA, 6‐MP	[26]
	*1, *2, *3A	USA	Acute lymphoblastic leukemia	Hepatotoxicity	—	0.035	6‐MP	[21]
	*2, *3A, *3C	Poland	Renal transplant	Leukopenia	4.26 (1.49–12.15)	0.008	AZA	[76]
		USA	Heart transplant	Severe rejection	—	< 0.001		[77]

Abbreviations: ALT, alanine transaminase; AST, aspartate aminotransferase; AZA, azathioprine; CI, confidence interval; GI, gastrointestinal; HR, hazard ratio; MP, mercaptopurine; OR, odds ratio.

NUDT15 c.415C>T, is one such variant and was reported for a strong link with thiopurine‐induced leukopenia in a 2014 study (OR = 35.6; 95% CI 22.47–56.51; p = 4.88 × 10⁻⁹⁴) [67]. Subsequent studies on NUDT15 further identified other important variants namely, c.36_37insGGAGTC, c.52G>A and c.416G>A for predicting thiopurine‐induced myelotoxicity [83]. The associations of these variants with myelosuppression in the Asian population have been well studied and documented. For example, Xu et al. found that carriers of the heterozygous NUDT15 R139C are significantly more associated with the risk of myelosuppression with AZA when compared to the wild‐type genotype (OR = 5.4; 95% CI 2.0–14.4; p < 0.001) [84]. Banerjee et al. in the Indian population found that carriers of the NUDT15 c.415 C>T polymorphism had 21 times and 19 times elevated risk of neutropenia and leukopenia compared to the carriers of the wild‐type genotype respectively (OR = 21.41; 95% CI 12.25–37.41; p < 0.0001 and OR = 19.35; 95% CI 11.55–32.42; p < 0.0001 respectively) and concluded by saying that NUDT15 genotyping is a better predictor of AZA‐induced myelosuppression in the Indian population than TPMT [63]. Similar observations were noted by Wang et al. in the Chinese population where the NUDT15 c.415 C>T polymorphism had an OR = 7.663; 95% CI 1.893–31.023; p = 0.004 for AZA‐induced leukopenia [62]. A recent meta‐analysis incorporated these associations assessed and found that compared to the wild‐type diplotype (NUDT15*1/*1), the *1/*3 c.415C > T C/T diplotype and *3/*3 c.415C > T T/T diplotype were associated with an elevated relative risk of leukopenia (risk ratio, RR = 4.12; 95% CI 2.87–5.91 and RR = 9.38; 95% CI 5.17–17.01, respectively) [55].

These evidences strongly reinforce the importance of incorporating the pre‐emptive NUDT15 genotyping in the Asian population prior to thiopurine therapy for predicting and preventing myelotoxicities. Though the evidence associating myelotoxicity with NUDT15 in the non‐Asian population is limited and needs further exploration, the literature evidence is growing. For example, Walker et al., with an exome‐wide association study demonstrated the link between NUDT15 (rs746071566) and thiopurine‐induced myelosuppression in affected vs. unaffected patients (OR = 38.2; 95% CI 5.1–286.1; p = 1.3 × 10⁻⁸) and for early onset and late onset myelosuppression, the ORs were 74.2 (95% CI 9.6–573.5; p = 8.2 × 10⁻¹⁰) and 20.9 (95% CI 2.6–170.1; p = 4.2 × 10⁻⁴) respectively. Based on the findings, they recommended considering preemptive genotyping of NUDT15 and further detailed studies to validate the association [26].

6. Thiopurine Drugs in Acute Lymphoblastic Leukemia (ALL): Optimizing the Safety Through Pharmacogenomics Consideration

Acute Lymphoblastic Leukemia (ALL) is the leading hematological malignancy in children representing almost 25% of all pediatric malignancies [85]. The treatment phases of ALL include induction, consolidation, delayed intensification and maintenance. Combination therapy with methotrexate and 6‐MP is the backbone therapy in the maintenance phase [86]. However, because of the narrow therapeutic window, toxicities related to 6‐MP can lead to potentially fatal incidents and are considered the leading cause of discontinuation or interruption of the chemotherapy. Notably, around 15%–28% of the patients suffer from drug‐related adverse effects even with the standard dose of 6‐MP [74]. Pharmacogenomics consideration can be attributable to the reduction of toxicity and optimization of the efficacy in ALL treatment as it assists the rational selection of drug and dose individualization. Treatment response to thiopurines can be altered by certain polymorphisms in the gene involved in the biotransformation of the drugs and drug transporters [87]. Some evidence of the genetic associations for the drug‐induced adverse events in ALL patients has been presented in Table 3. Genetic polymorphisms in TPMT influence the enzyme activity and in turn the 6‐MP‐related clinical efficacy and toxicity issues. Individuals carrying homozygous variant alleles of TPMT and heterozygous variant alleles are said to have low and intermediate TPMT activities respectively which are associated with the toxicities including myelotoxicity in pediatric ALL patients induced by 6‐MP necessitating dose reduction [88]. However, these toxicities are complex and several other genetic variants, namely, triphosphate pyrophosphatase (ITPA) and nucleoside diphosphate‐linked moiety X‐type motif 15 (NUDT15), have been evaluated for the association. The polymorphisms of ITPA 94C > A and IVS2 + 21A>C have been associated with the deficiency of the ITPA activity which was associated with the accumulation of 6‐thio‐ITP, an endogenous toxic metabolite. Similarly, the variation of NUDT15 leads to poor metabolism of thiopurines resulting in drug‐induced toxicities [69, 89].

Genotyping for TPMT is the earliest adoption of pharmacogenomics in the management of ALL patients. International pharmacogenomic working bodies like CPIC recommended it as a routine clinical test for ALL patients and provided a well‐established guideline for dose adjustment based on TPMT genotype/phenotype [24]. Similarly, NUDT15 is well studied particularly among the Asian population and there is some robust evidence of the association with myelotoxicities with thiopurines in ALL patients provided by multiple clinical studies, putting an emphasis on the clinical adoption and routine practice of pharmacogenomics‐guided therapy for ensuring the safety and effectiveness [50, 51, 53, 64, 68]. CPIC also recommends the genotype‐guided therapy and dose adjustment based on the NUDT15 genotype/phenotype [24]. ITPA on the other hand was associated with the adverse effects of 6‐MP in several clinical studies [29, 49, 71, 72, 90]. These associations have been evaluated in a recent meta‐analysis revealing that the ITPA 94C>A polymorphism is associated with 6‐MP‐induced toxicities including, neutropenia (OR = 2.60; 95% CI 1.30–5.19), leukopenia (OR = 1.75; 95% CI 0.74–4.12) and hepatotoxicity (OR = 1.98; 95% CI 1.32–2.95) [91]. Despite having the growing evidence for the association between ITPA and the safety of thiopurines, no clinical guideline has been proposed by CPIC.

7. Thiopurine Drugs in Other Diseases (Autoimmune Disease; Solid Organ Transplantation): Evidence of Pharmacogenomics Association

Thiopurines, predominantly 6‐MP and AZA are immunosuppressive agents indicated in many autoimmune disorders, such as autoimmune hepatitis (AIH), rheumatoid arthritis (RA), systemic lupus erythematosus (SLE), and generalized eczematous disorders [92]. However, documented adverse effects like bone marrow toxicities (BMT), pancreatitis, gastric intolerance (GI), hepatotoxicity etc. limit their use [78]. Genetic polymorphisms in TPMT and NUDT15 have been well studied for the association behind these adverse effects. Few evidences of the genetic associations for thiopurine drug‐induced adverse effects in different autoimmune conditions have been highlighted in Table 3. A meta‐analysis evaluated the genetic associations of TPMT and the adverse effects of AZA in autoimmune disease and reported an OR = 3.12; 95% CI 1.48–6.56 for overall adverse effects, for BMT (OR = 3.76; 95% CI 1.97–7.17), for GI (OR = 6.43; 95% CI 2.04–20.25) and for hepatotoxicity (OR = 2.86; 95% CI 0.32–25.86). Subgroup analysis revealed that TPMT is linked with AZA‐induced adverse effects in several autoimmune diseases to a varying degree (for SLE, OR = 4.74; 95% CI 0.53–42.45; for RA, OR = 4.32; 95% CI 0.42–44.75; for AIH, OR = 4.11; 95% CI 1.17–14.41). They also observed that Asian people are more at risk of BMT due to TPMT genetic variants (OR = 10.86; 95% CI 1.42–83.23) compared to Caucasians (OR = 2.28; 95% CI 0.85–6.09) [78]. However, the genetic association of TPMT and AZA‐induced myelotoxicities in rheumatic diseases remained controversial. Su et al., found no significant association of TPMT variants with AZA‐induced myelotoxicities in rheumatological disease (p = 0.973), whereas Yang et al., observed TPMT genetic variations to be significantly associated with leukopenia (p = 0.007) and they also observed significant association for hypochromia (p = 0.045; OR = 10.33; 95% CI 0.61–173.66), gastrointestinal discomfort (OR = 12.08; 95% CI 0.71–204.49; p = 0.028) and alopecia (OR = 33; 95% CI 1.80–606.47; p = 2.864 × 10⁻⁴) [57, 70]. Another two studies revealed no significant association of ITPA and TPMT*3C variants and AZA‐induced myelosuppression in AIH patients [58, 60]. Among Chinese autoimmune patients Fei et al., found no association between TPMT*3C genotype and AZA‐induced leukopenia (p = 0.95) [56].

Similarly, the association of NUDT15 and thiopurine drug‐induced myelotoxicity among different autoimmune patients has been assessed in several clinical studies, such as, Yang et al., who found that the NUDT15*3 genotype is significantly linked with leukopenia induced by AZA (p = 4.475 × 10⁻⁶). Futhermore, NUDT15*3 was found to be significantly associated with other adverse effects including, gastrointestinal effects (OR = 5.714; 95% CI 1.56–20.95; p = 0.002), erythropenia (OR = 9.333; 95% CI = 2.96–29.47; p = 1.109 × 10⁻⁵), hypochromia (OR = 13.18; 95% CI 4.15–41.87; p = 1.653 × 10⁻⁷) and thrombocytopenia (OR = 20.13; 95% CI 3.40–119.18; p = 9.110 × 10⁻⁶) [70]. A similar association was reported by Miao et al., for AZA‐induced myelotoxicity and NUDT15 variants in AIH patients (leukopenia, OR = 7.5; 95% CI 3.08–18.3; p = 8.26 × 10⁻⁷ and neutropenia OR = 8.05; 95% CI 2.96–21.9; p = 3.54 × 10⁻⁶); Huang et al., in dermatologic diseases (neutropenia, OR = 9.383; 95% CI 1.32–66.96); Fan et al., in patients with AIH and related cirrhosis (leukopenia, p < 0.00001; OR = 20.41; 95% CI 7.84, 53.13); Fei et al., in autoimmune patients (leukopenia, OR = 7.59; 95% CI 3.16–18.21; p = 1.79 × 10⁻⁷) and based on these findings they suggested considering the clinical adoption of pharmacogenomics‐guided thiopurine therapy in various autoimmune diseases [56, 58, 59, 60]. Several studies however suggested that NUDT15 genotyping is a better predictor of thiopurine‐induced adverse effects in autoimmune patients; however, the ethnicity of the participants cannot be overlooked, Therefore, large‐scale studies with multi‐ethnic populations are required for further assessment [56, 57, 58, 60].

AZA for its immunosuppressing action, with calcineurin inhibitors (i.e., tacrolimus, cyclosporine) and glucocorticoids is used as the standard maintenance therapy in organ transplantation. Despite the newer immunosuppressant options, AZA continues to be used therapeutically for the prevention of organ rejection after solid organ transplant due to the relatively lower cost and as a treatment option in patients with mycophenolate mofetil intolerance (the frequency of which is as high as 35%–40% with heart transplant) [93]. The bioavailability of AZA dictates the clinical efficacy and treatment‐related adverse effects and genetic variability may explain these variations in the drug's pharmacokinetics [30]. Mostly TPMT has been associated with the AZA‐induced adverse effects in transplant recipients. A Thai study with renal transplant patients revealed individuals with TPMT*1/*3C had a significantly elevated risk of AZA‐induced myelosuppression than the wild type (OR = 14.18; 95% CI 3.07–65.40; p < 0.005) [73]. Another study on renal transplant patients evaluated the genetic association of the variants of TPMT and ITPA for AZA and reported TPMT to be significantly associated with the AZA‐induced adverse effects (leukopenia, OR = 4.26; 95% CI 1.49–12.15; p = 0.008). No significant association was found for ITPA variants [76]. Similar clinical evidence exists for the genetic association among heart transplant patients as well. Sebbag et al., observed as compared to the wild type, carriers of TPMT*3A had agranulocytosis and had a drop in neutrophils and they suggested that determining these variants potentially can deduct the risk of hematological adverse effects [94]. Another study found that the carriers of the TPMT heterozygous genotype exhibited earlier severe rejection after heart transplant (p < 0.001), higher rejection score (p = 0.02), higher discontinuation rate (p = 0.01) and leukopenia (43% vs. 40%, p = 1.0) with a similar dose of AZA when compared with the wild‐type carriers [77].

8. Clinical Implementation of TPMT and NUDT Genotyping in Thailand: Success Story From Bench to Bedside

In Thai population thiopurine drugs induced adverse effects have been studied for the genetic associations particularly the TPMT and NUDT15 variants in ALL, renal transplant patients [51, 69, 73, 74]. Carriage of TPMT*1/*3C was found to be significantly associated with AZA induced myelotoxicities after renal transplant [73]. Two other studies assessed the 6‐MP induced adverse effects in ALL patients and NUDT15 variants. Puangpetch et al., observed NUDT15 c.415C>T variants to be associated with elevated likelihood of neutropenia (OR = 17.862; 95% CI 4.198–75.992, p = 9.5 × 10⁻⁵). Similarly, Buaboonnam et al., assessed higher risk of neutropenia at 3 months (OR = 12; 95% CI 3.781–38.085; p < 0.001) and significantly lower mean dose of 6‐MP (p < 0.001) in the carriers of NUDT15 c.415C>T as compared to the wild type and recommended further study to evaluate the association and elucidate dosing guidelines for Asian population [51, 69].

Another study on pediatric ALL patients investigated the genetic association of TPMT 719A>G (*3C), ITPA 94C > A and ITPA 123G > A polymorphisms and the drug transporters (MRP4 912C>A and MRP4 2269G>A) for 6‐MP induced adverse effects including hepatotoxicity and myelotoxicity and revealed that compared to the wild typegenotype carriers, the individuals with TPMT*1/*3C genotype were at an elevated risk of leukopenia (OR = 4.10; 95% CI = 1.06–15.94; p = 0.033). Heterozygous TPMT*3C genotype was significantly linked to an elevated risk of neutropenia in the first 8 weeks (OR = 4.17; 95% CI = 1.25–13.90; p = 0.014). However, no significant association was established with ITPA 94C > A, ITPA 123G>A, MRP4 912C>A, and MRP4 2269G>A variants for 6‐MP induced adverse effects. And considering the Food and Drug Administration drug label recommendations and CPIC guidelines, the findings of this study support the genetic testing of TPMT and phenotype identification in ALL patients undergoing 6‐MP therapy to reduce myelotoxicity and other adverse effects associated [24, 74].

Such phenotype identification is based on the TPMT enzyme activity which is mostly determined by measuring the formation of 6‐methylmercaptopurine (6‐MMP) from 6‐MP using S‐adenosyl‐L‐methionine (SAM) as the methyl donor, a process mediated by TPMT. These methylated products are then detected by various methods. The radiochemical (RIA) method utilizes high‐performance liquid chromatography (HPLC) followed by fluorescence and absorbance detection. Earlier studies employed washing with saline or a tedious ficoll separation process for preparing red blood cell lysate which is time consuming and uses at least 3 mL whole blood for preparing the red blood lysate [95, 96]. However measuring TPMT with whole blood (which is more homogeneous than red blood lysate) using 6‐methylthioguanine as the substrate for HPLC exerts a more rapid and accurate evaluation [97].

Therefore a new method employing LC–MS/MS for studying TPMT enzyme activity has been developed and validated for use [98]. This method utilizes the whole blood method developed for LC–MS/MS using 6‐MP as the substrate. With the TPMT incubation optimized, a rapid chromatographic method (runtime 7 min) was performed on a C18 column followed by the detection using triple quadrupole mass spectrometry. MS/MS was tuned optimally for monitoring mass to charge ratio (for 6‐MMP, 167.2 → 151.9 and for the isotope 6 MMP‐d3, 170.5 → 152.2) as the internal standard. The calibration graph covered a range of 2.5 to 360 ng/mL with a correlation coefficient of 0.999. The accuracy was evaluated using four concentrations providing a control of quality range of 99.33%–106.33%. The coefficient of variation in intra‐assay was smaller than 4.41% and in inter‐assay smaller than 5.43%, making this LC–MS/MS method a safe, reliable and simple approach for routine adoption. Furthermore it requires a blood volume of only 100 μL, facilitating the application in pediatric patients and the patients from whom drawing a large volume of blood is not feasible. In short, it's a beneficial method and can be utilized for evaluating the reference range and for many samples. Moreover, this approach can be a great tool in the patients taking thiopurine drugs for optimization of the efficacy and reduction of the adverse effects and taking other important therapeutic decisions [98].

In Thailand, TPMT and NUDT are routinely genotyped in some hospitals, for example, at the Pharmacogenetics and Personalized Medicine (PPM) Laboratory of the Department of Pathology, Faculty of Medicine Ramathibodi Hospital, These tests are available and so far, many patients have been tested to guide the prescription of thiopurine drugs for optimizing the safety or effectiveness of these drugs. Currently TPMT*2, *3A, *3B and *3C alleles are tested for TPMT and similarly NUDT15*2, *3, *4, *5 and *6 alleles are routinely checked in Thailand. For the identification of TPMT or NUDT15 polymorphisms, the RT‐PCR method is being employed. Genotyping of TPMT and NUDT is also available in Bumrungrad International Hospital. Though there is a paucity of studies reporting the cost‐effectiveness or the benefit of this approach, it is generally acknowledged that genetic testing helps make more informed clinical decisions based on the dosing recommendations provided by the CPIC. The CPIC PGx‐based dosing guidelines for thiopurine drugs based on TPMT and NUDT15 genetic polymorphism are evidence‐based recommendations for optimizing the safety of drug therapy in clinical practice. Thai hospitals have started to follow these guidelines by undertaking TPMT/NUDT15 genotyping for optimizing the safety of thiopurine drugs since many of the clinical studies involving Thai cancer patients have found strong associations for TPMT/NUDT15 with thiopurine‐induced severe toxicities, for example, myelosuppression, leukopenia, neutropenia, and so forth [24, 28, 51, 69, 74]. It is expected that other hospitals in Thailand will adopt these testing services very soon to guide the precision medicine of thiopurine drugs.

9. Pharmacogenomics Consideration of Thiopurine Drugs in Routine Clinical Practice: Challenges and Opportunities

Pre‐emptive testing for TPMT allows the dose customization according to the TPMT status and thus reduces the risk of acute myelosuppression while maintaining effective disease control. Similarly, pre‐emptive NUDT15 genotyping, particularly in the Asian population offers therapeutic benefits as these genetic variants have a comparable risk profile as in TPMT. In ALL patients at least, no‐function NUDT15 alleles are associated with increased sensitivity to 6‐MP and therefore, theoretically, the NUDT15 genotype‐guided treatment would not disrupt the efficacy of the drug [21, 24, 83].

However, the implementation faces limitations in wide clinical application. One such potential limitation is that intermediate metabolizers of TPMT may receive lower treatment doses than they actually can tolerate, because only around 30%–60% of the patients with TPMT heterozygous genotype experience severe myelotoxicities [21, 72]. Besides, the potential for errors in genotyping cannot be overlooked. As demonstrated in preclinical models, several NUDT15 and/or TPMT variants may be excluded in the genotyping test used assigning them a wild‐type genotype which may in reality be a decreased or no function allele and as these genotypes are life‐long identification, potential error may lead to long‐term clinical complications [24]. Given the complex nature of the genetic associations and the thiopurine therapy, potential combined effects of several genetic variants cannot be ruled out. So, composite evaluations of the pharmacogenetic markers are imperative for comprehensive understanding of combined phenotypic effects. In several studies, interactions of TPMT, NUDT15 and ITPA and their combined association with toxicities related to 6‐MP have been hinted at [99, 100]. However, there is a paucity of large‐scale studies validating the association and development of polygenic risk scores.

10. Future Directions

The LC–MS/MS assessment method of TPMT, developed by Wiwattanakul et al., should be studied in different ethnic populations for further evaluation, validation and clinical implementation of the method [98]. Conducting large‐scale studies with multi‐ethnic populations taking thiopurine drugs is required for assessing the combined genetic association of TPMT, NUDT15 and ITPA to have a clear understanding of the development of drug‐induced adverse effects, for developing a polygenic risk score and more pharmacogenetics‐focused implementable and acceptable clinical guidelines.

11. Conclusions

TPMT and NUDT15 polymorphisms have emerged as important clinical markers for predicting and optimizing the safety and effectiveness of thiopurine drugs. The bench‐to‐bedside approach of exploring and assessing the genetic associations of TPMT and NUDT15 variants and the new LC–MS/MS methods for evaluating TPMT is a step forward in the advancement of precision medicine of thiopurine drugs. In Thailand, TPMT and NUDT are routinely genotyped in some hospitals to guide the prescription of thiopurine drugs for optimizing the safety or effectiveness of these drugs. However, the composite effects of these genetic variants remain unexplored at the global scale. Proper large‐scale studies with multi‐ethnic patients can provide a clear understanding of the TPMT/NUDT15 association and would pave the way towards the optimization of thiopurine drugs to achieve precision medicine.

Conflicts of Interest

The authors declare no conflicts of interest.

Biswas M., John S., Murad M. A., and Sukasem C., “Pharmacogenomics of Thiopurine Drugs: A Bench‐To‐Bedside Success Story in Thailand,” Clinical and Translational Science 19, no. 2 (2026): e70410, 10.1111/cts.70410.