TPMT and NUDT15 genotyping, TPMT enzyme activity and metabolite determination for thiopurines therapy: a reference laboratory experience

ABSTRACT

Aim

To share the experience of a US national reference laboratory, offering genotyping for TPMT and NUDT15, TPMT enzyme phenotyping and detection of thiopurine metabolites.

Methods

Retrospective review of archived datasets related to thiopurines drug therapy including patients’ data that underwent TPMT and NUDT15 genotyping, and smaller data sets where genotyping was performed with TPMT enzyme levels (phenotyping) +/- therapeutic drug monitoring (TDM).

Results

Thirteen percent of patients had variants in one or both genes tested. Testing for NUDT15 revealed 3.9% additional patients requiring thiopurines dosing recommendations. A correlation between TPMT enzyme activity and TPMT polymorphisms (odds ratio OD = 71.41, p-value <0.001) and between older age and higher enzyme levels (OD = 0.98, p-value = 0.002) was identified. No correlation between sex and TPMT enzyme levels, nor between TPMT genotyping and the level of thiopurine metabolites was found.

Conclusion

Adding NUDT15 to TPMT genotyping, identified additional 3.9% patients to benefit from thiopurine dose modifications. A significant correlation between genetic variants in TPMT and TPMT enzyme levels and between age and enzyme levels was established, while no correlation was identified between sex and enzyme levels nor between TPMT variation and thiopurine metabolites. Providers rely more significantly on genotyping only approach, rather than genotyping and phenotyping.

Plain Language Summary

The current study aimed at sharing the experience of a US national laboratory, offering three kinds of testing relevant to patients receiving thiopurine drugs. The testing included detecting variants in two genes (TPMT and NUDT15) in addition to measurements of TPMT enzyme levels and detection of thiopurine metabolites in patients’ blood samples. The study analyzed archived datasets of patients that underwent such tests. 13% of examined patients had variants in either one or both genes tested (TPMT and NUDT15) and required modifications to the standard recommended dosing of thiopurines to avoid severe toxicity and side effects. This study found that adding NUDT15 variant testing to TPMT testing, identified additional patients that could benefit from thiopurine dose modifications. Additionally, the study identified a significant relationship between TPMT enzyme levels and TPMT variants (a patient having a variant that caused loss of function, would have lower levels of TPMT enzyme). Another relation between age and enzyme levels was also detected (older patients were found to have higher enzyme levels). This study could not identify effect of sex on TPMT enzyme levels, nor establish a relationship between TPMT variants and the level of thiopurine drugs metabolites. Investigating providers’ attitudes through order choices, revealed that providers rely more significantly on detecting variants in genes affecting TPMT enzyme rather than testing for both gene variants and enzyme levels which is believed to give providers a more comprehensive knowledge about patient’s response to thiopurines. Because of the toxicity of thiopurines, it is important to be able to assess whether patients, due to their genetic make-up, can be more susceptible to side effects or toxicity upon receiving thiopurines. Testing for variants in TPMT and NUDT15 and measuring TPMT enzyme levels can help providers with proper dosing of thiopurines.

1. Introduction

Thiopurines are drugs that are mainly used clinically as anticancer, immune-suppressant, or anti-inflammatory agents. They are established treatments for acute leukemia, rheumatologic, and dermatologic conditions, in addition to inflammatory bowel diseases (IBD) such as Crohn’s disease (CD) and ulcerative colitis (UC) [1–5]. They can also be used to prevent organ transplant rejection [6]. In addition to the effectiveness of thiopurines, they have the advantages of oral delivery, and low cost [7]; yet their narrow therapeutic index and potential toxicities (myelotoxicity, hepatotoxicity, and nephrotoxicity) and side effects such as gastrointestinal intolerance, particularly when starting treatment, entail modifying the dose or even treatment cessation in many cases [8,9]. Discontinuation of treatment due to toxicities has been reported to be up to 15% and 30% in IBD and rheumatoid arthritis patients, respectively, [10,11].

Thiopurine drugs, e.g., azathioprine, mercaptopurine, and thioguanine, are prodrugs that that require activation to act as antimetabolites or immunomodulatory medications. Complex metabolism of thiopurines result in 6-thioguanine nucleotides (6TGN), the main active therapeutic metabolite [12]. Three different enzymes are involved in thiopurines metabolism: thiopurine methyltransferase (TPMT), xanthine oxidase/dehydrogenase (XOD), and hypoxanthine phosphoribosyltransferase (HPRT). TPMT is responsible for the formation of the inactive metabolites, 6-methylmercaptopurine (6MMP) and 6-methylmercaptopurine ribonucleotide (6MMPR) [12,13]. Another enzyme that plays an important role in thiopurines metabolism is NUDT15 (nudix hydrolase 15). NUDT15 functions by hydrolyzing the thiopurine effector metabolites (6-TGN: TGMP, TGDP, and TGTP) which translates into their non-incorporation into DNA which in turn reduces the cytotoxic and immunosuppressant effects of thiopurine drugs [14,15] (Figure 1).

Figure 1.

Thiopurine metabolism.

Genetic polymorphisms of TPMT gene that have no function are well documented to influence TPMT enzyme activity by decreasing its level [16,17]. TPMT enzyme levels in turn are inversely correlated with the accumulation of TGNs in hematopoietic cells leading to hematopoietic toxicity [18]. Patients who have deficient levels of TPMT enzyme are at risk of developing life-threatening complications, e.g., myelosuppression [19,20]. NUDT15 genetic polymorphisms were also reported to affect NUDT15 enzymatic activity and patients’ tolerance to standard doses of thiopurine drugs. Loss of function variants result in decreased enzymatic activity leading to excessive thiopurine active metabolites and increase susceptibility to thiopurine-induced myelosuppression [15,21,22]. Various strategies have been proposed to optimize conventional thiopurine therapies. In addition to genetic polymorphisms, determining TPMT enzyme levels (phenotyping) within RBCs (U/mL peripheral red blood cell), specially prior to starting thiopurines treatment, can help titrate the proper starting and maintenance doses and avoid potential side effects that are most frequent at the beginning of treatment. TPMT genotyping and phenotyping are believed to be complementary approaches that decrease the chances of misclassifying individuals especially those with intermediate enzyme activities [23,24]. Another approach that helps with thiopurines dose optimization is therapeutic drug monitoring (TDM) which is measuring the concentrations of thiopurines metabolites after the initiation of therapy. Measuring 6-thioguanine nucleotides (6-TGN) is used to determine whether dosing is in the optimal range and assesses the risk for leukopenia and myelotoxicity. Additionally, 6-methylmercaptopurinenucleotides (6-MMPN) is measured to assess the potential risk for hepatotoxicity [25]. Whether measuring thiopurine metabolites during treatment is of significant value remains a subject of debate [26,27].

In this study, we share the experience of ARUP laboratories, a US national reference laboratory, offering pharmacogenetic testing for TPMT and NUDT15, in addition to TPMT enzyme phenotyping and detection of thiopurine metabolites. We discuss results from three cohorts. The first is for samples that only had orders for TPMT genotyping, then a smaller cohort of concurrent TPMT genotyping and TPMT enzyme-level analysis will be discussed, and lastly the smallest cohort consisted of samples for which concurrent genotyping and metabolite-level analysis was requested.

2. Subjects and methods

2.1. Patients

Archived data sets obtained at ARUP Laboratories (Salt Lake City, UT) were retrospectively reviewed from patients referred to ARUP in relation to thiopurine therapy. First, data was extracted from patients that underwent TPMT and NUDT15 genotyping after testing for NUDT15 alleles was added to TPMT genotyping at ARUP laboratories. Then data from cohorts of patients that in addition to genotyping had results for TPMT enzyme activity (phenotype) or TDM based on 6-TGN and 6-MMPN assays were also analyzed. ARUP laboratories had no access to data regarding timing of the testing, whether prior to or during thiopurine treatment, neither to data regarding genetic ancestry of patients. Data was extracted following the policies of the University of Utah Institutional Review Board (IRB) and the procedures outlined in IRB protocol #00082990. All data were analyzed anonymously.

2.2. TPMT and NUDT15 genotyping

Genomic DNA was extracted from whole blood using Chemagen M-PVA magnetic Bead Technology and Chemagic MSM I instrument (PerkinElmer). DNA purity was determined (A260/280), then samples were quantified using the Infinite reader and normalized to 50 ng/μL. TPMT and NUDT15 genotypes were determined using a custom TaqMan® OpenArray® on the QuantStudio™ 12K Flex system (Thermo Fisher Scientific). Genotyping experiments were performed according to instructions provided by the manufacturer. The TaqMan genotyping assays detected the TPMT *3C (c.719A>G), TPMT *3B (c.460 G>A), TPMT *3A (c.460 G>A and c.719A>G), TPMT *2 (c.238 G>C), TPMT *4 (c.626-1 G>A), NUDT15 *2 or * 3(c.415C>T), and NUDT15 *4 (c.416 G>A) (Table 1). It is worth noting that NUDT15 *2 and * 3 share the SNP rs116855232 and unless additional markers are tested differentiating between the two star alleles is not possible. *1 allele (normal no-risk allele) is defined by the absence of the detection of variant alleles. Interpretation of the metabolizer status was performed according to CPIC guidelines [28].

Table 1.

TPMT and NUDT15 alleles tested.

TPMT(NM_000367)	*2: rs1800462, c.238 G>C#
	*3A: rs1800460, c.460 G>A#
	rs1142345, c.719A>G#
	*3B: rs1800460, c.460 G>A#
	*3C: rs1142345, c.719A>G#
	*4: rs1800584, c.626-1 G>A
NUDT15(NM_018283)	*2 or * 3: rs116855232, c.415C>T#β
	*4: rs147390019, c.416 G>A¥

^#Tier 1 alleles.

^¥Tier 2 alleles.

βrs116855232 is a core variant for NUDT15 *2 and NUDT15 *3. Both alleles are no function alleles.

2.3. TPMT enzyme activity (phenotype)

TPMT activity was assessed in RBCs using liquid chromatography tandem mass spectrometry (LC-MS/MS) as previously described [29]. Enzyme levels were considered normal if between 24.0–44.0 U/mL, intermediate if 17.0–23.9 U/mL, low if <17.0 U/mL, and high if >44.0 U/mL based on an in-house genotype – phenotype study performed on 157 patient samples. Predicted risk of bone marrow toxicity (myelosuppression) and recommended thiopurine dose adjustments based on enzyme levels are detailed in Table 2.

Table 2.

TPMT enzyme activity.

Activity	Range	Predicted Risk of Myelosuppression	Recommended Thiopurine Dose Adjustments
Normal	24.0–44.0 U/mL	Low	None
Low	<17.0 U/mL	High	Do not use thiopurines
Intermediate	17–23.9 U/mL	Intermediate	Consider dose reduction. Manage carefully
High	>44.0 U/mL	None	Consider dose increase. Manage carefully

2.4. Thiopurine metabolites concentrations in RBCs

6-TGN and 6-MMPN concentrations were assessed in RBCs using liquid chromatography tandem mass spectrometry (LC-MS/MS) as previously described with a minor modification where the hydrolysis temperature to cleave the nucleotides was modified to 74°C for 2-h incubation [30]. The therapeutic range for 6-TGN was considered between 235–450 pmol/8×10⁸ RBCs and ≤ 5700 for 6-MMPN pmol/8×10⁸ RBCs (Table 3).

Table 3.

Thiopurine metabolites levels.

6-TGN Concentration (pmol/8×10 RBCs)	Interpretation
<235	Possible reduced response to therapy
235–450	Within therapeutic range
>450	Possible increased risk for leukopenia and myelotoxicity
6-MMPN Concentration (pmol/8×10 RBCs)	Interpretation
≤5,700	Within therapeutic range
>5,700	Possible increased risk for hepatotoxicity

TGN, thioguanine nucleotide.

MMPN, methylmercaptopurine nucleotide.

2.5. Statistics

Descriptive statistics and graphs were produced using Microsoft Excel for Microsoft 365 MSO (Version 2309, Build 16.0). The correlation between TPMT enzyme activity and age, gender, and genotype were analyzed using logistic regression. A p-value <0.05 was considered statistically significant.

3. Results

3.1. Genotyping only cohort

3.1.1. Demographics

In total, 11,348 patients underwent TPMT and NUDT15 genotyping. Age distribution of the cohort could be found in Table 4. Females were 58.4% of the patients and males were 41.6%. The largest age group represented were between 19 and 60 years comprising 55.5% and 45.2% for females and males, respectively.

Table 4.

Sex and age distribution in genotyping only cohort.

Sex	Number	%	Age (years)	No./age group
Female	6629	58.4	≤18	1493
			19–60	3678
			>60	1458
Male	4719	41.6	≤18	1731
			19–60	2135
			>60	853

3.1.2. Genotype frequency

Of the genotyped patients, about 13% had variants in either TPMT or NUDT15 (8.8% in TPMT and in 3.9% NUDT15), while only 0.4% of patients had variants in both genes. For patients who had TPMT variants, 97.4% of the patients were heterozygotes, 0.6% were compound heterozygotes and 2% were homozygotes. The most prevalent variants among TPMT carriers were * 3A (n = 710, 71%), *3C (n = 204, 20.4%), and * 2 (n = 60, 6%). Figure 2(a) displays TPMT genotypes distribution among patients who had variants (1000 patients). While for patients who had NUDT15 variants, 95.7% were heterozygotes, and 4.3% were homozygous. The most prevalent variants among NUDT15 carriers were * 2 or * 3 (n = 343, 77.4%) and * 4 (n = 81, 18.2%). Figure 2(b) displays NUDT15 genotypes distribution among patients who had variants (443 patients). Forty-eight patients were found to have variants in both TPMT and NUDT15. The most frequent genotype in those patients was TPMT (*1/*3A)/NUDT15 (*1/*2 or * 3) (Figure 2(c)).

Figure 2.

Genotypes distribution among patients who had variants in one or both genes.

3.1.3. Metabolizer status

The metabolizer status within patients who have variants in TPMT and/or NUDT15 was also determined and reported. This in turn translated into actionable events in terms of determining which patients received recommendations for change in prescription or dosage. Among patients with variants in TPMT, 97.4% were intermediate metabolizers, while 2.6% were poor metabolizers. Among patients with variants in NUDT15, 77.4% were intermediate metabolizers and 4.3% were poor metabolizers. According to the Clinical Pharmacogenetics Implementation Consortium (CPIC) guidelines [28], an individual carrying two uncertain function alleles or one normal function allele plus one uncertain function allele (i.e., NUDT15 *4) has an indeterminate metabolizer status. In our cohort 18.3% fell in that category (*1/*4 genotype, n = 81). Additionally, 48 patients had variants in both genes. Patients found to be intermediate metabolizers for both genes were the majority (66.7%, n = 32). Figure 3 displays distribution of metabolizer status among patients. In total, 11.9% of all genotyped patients (n = 1410) received a recommendation of change to standard thiopurines dosing based on CPIC Guidelines [28]. It is worth noting that at the time of preparation of this manuscript, CPIC guidelines website for thiopurines was updated to indicate that for patients who are intermediate metabolizers for both TPMT and NUDT15 (IM/IM), a substantial dose reduction to avoid toxicity is recommended (20–50% of normal dosages) [31].

Figure 3.

Distribution of metabolizer status among patients.

3.2. Genotypes and TPMT phenotype cohort

3.2.1. Demographics

In total, 755 patients underwent both genotyping and TPMT enzyme level testing (phenotyping). Females were 64.1% of the patients and males were 35.9%. Similar to the genotyping-only cohort, the largest age group represented were between 19 and 60 years comprising 43.6% and 21.6% for females and males, respectively. Ages ranged between 1 and 88 years with a mean age of 44.9 years.

3.2.2. TPMT enzyme activity distribution

As described in the methods section and in Table 2, patients were classified into four groups, those with low, normal, intermediate, or high TPMT enzyme activity. Results of such distribution based on sex and age group could be found in Table 5. Eighty percent of patients had a normal TPMT enzyme activity (24.0–44.0 U/mL), 19.7% had intermediate enzyme activity (17–23.9 U/mL), 0.26% had low enzyme activity (<17.0 U/mL), and none had high enzyme activity (>44.0 U/mL); a clear trimodal frequency distribution as described in the literature [29,32]. Logistic regressions showed a statistically significant relationship, yet of small effect size, between age and enzyme activity (OD = 0.98, p-value = 0.002) after adjusting for sex and the presence of TPMT variants. The odds ratio indicated that an increase in 1 year of age was associated with 0.98 times the odds of having low or intermediate enzyme levels; in other words, older individuals had higher TPMT enzyme activity than younger individuals. No statistically significant relationship between enzyme activity and sex was identified in any enzyme activity group.

Table 5.

TPMT enzyme level by genotype (A) and by age and sex (B)

		TPMT enzyme activity (U/mL RBC)
A	TPMT genotypes	Normal	Intermediate	Low	Total
	WT	600	96	0	696
	Heterozygous	5	52	2	59
	Homozygous mutant	0	0	0	0

B		Age	TPMT enzyme activity (U/mL RBC)
	Sex		Normal	Intermediate	Low	Total
	F	≤18	19	9	0	28
		19–60	267	61	1	329
		>60	102	25	0	127
	M	≤18	22	13	0	35
		19–60	125	37	1	163
		>60	70	3	0	73

3.2.3. Genotype frequency and metabolizer status

Of the 755 genotyped and phenotyped patients, 11.4% (n = 86) had variants in one of the two genes TPMT or NUDT15 (7.3% in TPMT and in 4.1% NUDT15), while 0.5% (n = 4) of patients had variants in both genes. The most frequent genotype for TPMT was * 1/*3A (n = 39, 5.2%), while for NUDT15 it was * 1/*2 (n = 22, 2.9%). Based on the genotypes, 82 patients were classified as intermediate metabolizers and received recommendations for thiopurines dose reductions. For the four patients who had variants in both genes and were classified as intermediate metabolizers for both TPMT and NUDT15, consideration of further reduction in thiopurines doses were recommended in the genotyping report. In this cohort, no patients were found to be poor metabolizers.

3.2.4. Genotype–phenotype correlation

A total of 755 patients had both TPMT phenotype and TPMT genotype determined. TPMT activity in patients with no TPMT variants ranged from 24.0 to 40.3 U/ml (n = 600); 96 patients had values within the intermediate enzyme levels between 19.3 and 23.9 U/ml. No patients in this group had low enzyme levels. In patients that were heterozygote for TPMT variants, intermediate TPMT activity was detected and ranged between 13.9 and 23.8 U/ml in 52 patients, while 5 patients had normal enzyme levels that ranged between 24.1 and 25.8 U/ml. The genotypes of patients carrying a TPMT variant and normal enzyme level were as follows: *1/*3A (three patients), *1/*2 (one patient) and 

* 1/*2C (one patient). Two patients had low enzyme levels (13.9 and 16.5 U/ml). Wild-type patients had 1.3-fold higher mean TPMT activity than did heterozygous variant patients (27.26 vs 20.71 U/ml, respectively). The overall concordance rate between TPMT genotype and phenotype was 86%, while it was 88.1% among heterozygotes. For the latter group, seven heterozygous patients (1%) would have been misclassified given the enzyme levels alone, since five heterozygote patients had normal TPMT activity (0.7%), and two had low TPMT activity (0.3%) (Table 5). None of the patients in this cohort carried homozygous variant alleles.

The distribution of TPMT genotypes in relation to the enzyme activity is shown in Figure 4. The most common variant genotype was * 1/*3A (n = 42, 5.6%), other genotypes were * 1/*3C (n = 15, 2%) and * 1/*2 (n = 2, 0.3%). Of heterozygous carriers, 88% had intermediate enzyme activity. In this cohort, neither TPMT *3B or * 4 alleles were detected. Logistic regression showed a statistically significant relationship between the presence of TPMT variants and low enzyme activity (OD = 71.41, p-value <0.001) after adjusting for sex and age.

Figure 4.

The distribution of TPMT genotypes (X-axis) in relation to the enzyme activity U/ml (Y-axis).

3.2.5. Order pattern distribution by clinical specialty

The distribution of combined test orders (genotyping and phenotyping) by the specialty of the ordering physician were explored. Physician specialty was available for 88.7% of the orders (n = 670) and not available for 11.3% of cases (n = 85). Orders by rheumatologists represented 39.0% of the orders, while those by gastroenterologists represented 25.1% followed by internal and family medicine at 11.5%. Oncology and hematology (adult and pediatric) ordered 3.1% of the tests. The remaining 21.3% of orders were placed by a combination of many other specialists. The specialties of ordering providers are depicted in Figure 5.

Figure 5.

Specialties od ordering providers.

3.3. Genotypes and thiopurine metabolites concentrations cohort

3.3.1. Demographics

In total, 106 patients underwent both genotyping and thiopurines metabolites concentration testing. Females were 63.2% of the patients (n = 67) and males were 36.8% (n = 39). Similar to the other two previously discussed cohorts, the largest age group represented were between 19 and 60 years comprising 62.3% of the cohort.

3.3.2. Metabolites concentrations

Thiopurine metabolites (6-TGN and 6-MMPN) were measured 125 times in 106 patients. It was measured once in 96 patients, and multiple times in 10 patients. Three patients had four measurements: one over 3 weeks period and two over 4 months period. Three patients had three measurements: one over 3 weeks period, one over 2 months, and one over 3 months. Four patients had two measurements: one over 10 days, one over 3 weeks, one over 3 months, and one over 5 months period. For 6-GT, 53 measurements were below limit of detection of the assay (20 pmol/8×10^8 RBCs); 38 measurements indicated suboptimal dosing <235 pmol 6-TGN/8×10^8 RBCs; 20 measurements were within the therapeutic range (optimal dosing) = 235–450 pmol 6-TGN/8×10^8 RBCs; and 14 measurements indicated potential for increased risk of toxicity >450 pmol 6-TGN/8×10^8. For 6-MMP, 80 measurements were below level of detection of the assay (400 pmol/8×10^8 RBCs); 31 measurements were within therapeutic levels and 14 measurements indicated potential for increased risk of toxicity >5700 pmol 6-MMPN/8×10^8 RBCs. Logistical regression did not identify statistically significant correlation between metabolites levels and genotypes results.

4. Discussion

ARUP laboratory has started offering targeted TPMT genotyping testing for patients using thiopurine therapy in the year 2015, with the NUDT15 genotyping component added later to comply with the published guidelines [28]. Additionally, ARUP laboratory offers TPMT phenotyping by evaluating TPMT enzyme activity and levels of thiopurines metabolites.

In the largest cohort of genotyping-only orders, the distribution and prevalence of both TPMT and NUDT15 genotypes were examined. Overall, about 13% of the patients had variants in either or both genes. Patients who were heterozygous for either gene were the vast majority, while compound heterozygotes or homozygotes for the mutant allele were the minority. The overall percentage of patients having TPMT variants has not differed from previously published data by this group [29], while testing for NUDT15 identified an additional 3.9% of patients in this cohort that had NUDT15 variants and 82% of them received recommendations of change to standard thiopurines dosing based on the published guidelines [28]. This result signifies the added clinical utility of testing NUDT15 variants in patients receiving thiopurines therapy. The remaining 18% were patients that had indeterminate NUDT15 metabolizer status. The most prevalent variant found in this cohort was TPMT *3A which is the most common variant detected in populations of European ancestry [29,33,34]. For NUDT15, the most prevalent variant detected was * 2 or * 3 at 3.2% in this cohort. The core variant for * 2 or * 3 (rs116855232) has high prevalence among Asians and Hispanic patients and much lower prevalence among patients of European ancestry [35,36]. As a reference laboratory, the demographics of the patients in the analyzed dataset often do not include patients’ ancestry which limits the ability to make significant conclusions on whether the observed prevalence matches published data for specific ancestry groups. Of interest, published guidelines continue to be updated, particularly with regards to recommending alleles to be tested as tier 1 (must-test alleles) or tier 2 (could test alleles) by AMP Pharmacogenomics Working Group working jointly with other collaborative professional societies [37]. This does represent a challenge for laboratories, as adding additional alleles entails new validations that require time and extensive documentation that often delay the implementation of the recommended alleles. Moreover, the technologies often used for targeted genotyping such as TaqMan® OpenArray® on the QuantStudio™ 12K Flex system used in this study, cannot resolve the issues of variants interference (e.g., NUDT15 *2 or 3 and the nearby NUDT15 *4 co-occurring in the same patient) nor the issue of phasing (determining whether identified variants fall on the same chromosome or on opposite chromosomes, i.e., in cis or in trans), and hence haplotypes are inferred rather than being precisely determined. This in turn limits the assay’s ability to accurately determine patient phenotype and often requires additional testing such as Sanger sequencing, that might resolve the interference, but not the phasing issue.

This study retrospectively examined results from a cohort of concurrent TPMT genotyping and TPMT enzyme activity assessment. Most of those patients had normal enzyme levels, followed by intermediate levels and a very small fraction had low enzyme levels. While some studies describe the distribution of TPMT enzyme level as low, intermediate, and normal-high with a subgroup for high enzyme levels [34,38], others including this study have not identified patients with high enzyme levels (>44.0 U/mL) [29,32]. There was a statistically significant relation between TPMT enzyme activity and TPMT polymorphisms (OD = 71.41, p-value <0.001). Of interest, this study identified significant correlation between older age and higher enzyme levels (OD = 0.98, p-value = 0.002). Other studies that investigated the effect of age on TPMT enzyme levels, suggest that children have higher levels of enzyme levels [38–40]. Kahlin, et al., similar to this current study, found correlation between higher age and higher enzyme levels [41], while others have failed to find any correlation between age and enzyme levels [42]. It is worth noting that studies varied with regards to whether the tested individuals were part of a healthy cohort or, as in our case, presumed unhealthy and receiving treatment. The effect of sex on TPMT enzyme levels had been investigated by many groups with conflicting findings. In this study, we could not identify significant effect of sex on TPMT enzyme levels in this cohort; this is similar to other studies [29,43]. On the other hand, multiple studies identified sex difference where males are believed to have higher enzyme levels compared to females [41,44,45]. In addition to genetic predispositions, TPMT enzyme activity levels can be influenced by many factors including concurrent diseases, red cell transfusions, patient’s genetic ancestry or therapies patients are receiving at the time of testing [34,46,47], such variables can affect and skew any conclusions made correlating age or sex to enzyme levels, particularly in a setting of a reference laboratory that lacks access to detailed clinical information. The overall genotype-phenotype correlation in this study was 86% which is lower than similar studies which reported overall concordance of at least 93% or more [32,34,48]. This lower concordance rate could be attributed to the 12.8% individuals with WT genotypes that had intermediate enzyme levels. This discordance has been attributed to, and in certain studies proven to be due to, presence of rare TPMT variants that were not commonly clinically tested by a targeted genotyping approach [34,48]. Additionally, 7 heterozygote patients had high or low enzyme levels. It had been previously reported that genotyping is rather imprecise in predicting the enzyme levels in the intermediate range [49] and this could be explained by other non-genetic factors discussed earlier, or due to non-tested TPMT promoter tandem repeats that might influence TPMT activity [23,50,51]. Another explanation of the lower concordance rate would be the sharp cutoffs established by laboratories to determine categories of enzyme levels for practical purposes, since some patients could fall near the limits between ranges. For example, around 17% of the patients in this study that were tested for enzyme levels, had results between 23–24.9 U/ml which is slightly below or above the lower limit of the normal enzyme level (≥24 U/ml). As more sequencing technologies get adopted (e.g., full-gene sequencing, whole exome and genome sequencing), the possibility of identifying rare TPMT variants becomes higher, and increases the need for a robust and fast functional testing method to enable understanding the potential role of such variants and highlights the need for an effort to standardize the classification and interpretation of novel pharmacogene variants in light of their impact on drug response phenotypes. Without such understanding, using such rare variants to modify thiopurine doses and including such variants in published guidelines becomes harder. Like many of other genetic studies, there continues to be need for larger numbers of tested individuals to have a deeper understanding of the pharmacogenetics of TPMT and NUDT15 and their additive or independent roles, as well as identify novel genes or variants that might explain the discrepancies between genotype and phenotype in clinical practice [23,32,35].

This study was limited in investigating the relationship between TPMT genotyping and the level of thiopurine metabolites by the small sample size of concurrent orders (n = 106). While a significant relationship was not detected, the reliability of such finding is limited and contradicts other studies that compared the thiopurine metabolites levels among patients with and without TPMT variants identifying a significant difference [26]. Additionally, the study was limited by using the analysis of 6-TGNs concentrations, which cleaves the phosphate groups from the mono-, di-, and triphosphate metabolites. Measuring thioguanine triphosphate (TGTP) is more indicative of measuring the nucleotide that is incorporated into DNA, thus contributing to the immunosuppressive effects. Analytical methods that quantify thioguanine nucleotides can aid in therapeutic drug monitoring for dose optimization [52–54].

As all testing in this study was performed in a reference laboratory setting, heterogenous and less characterized demographic information could contribute to the genotype-to-phenotype associations compared to published studies. Another limitation of this study was lack of information regarding whether testing was conducted prior to or during thiopurine treatment and hence conclusions with regards to tolerated doses or needed modifications based on genotypes and phenotypes were not possible.

While there is more of a consensus on the value of genotyping in helping with thiopurine therapy, the debate regarding the value of TPMT phenotyping and thiopurine metabolites assessment and the concordance between TPMT genotype and phenotype remains unresolved [23,26,55]. This seems to be corroborated in this study by analyzing common practices of providers ordering tests through ARUP laboratory with regards to concurrent ordering of TPMT genotyping and enzyme levels or ordering genotyping and metabolite levels. In this study, over 11,000 genotyping only orders were analyzed, while concurrent orders for genotyping-phenotyping were 755 orders, and concurrent genotyping-metabolite levels were only 106 orders during the same period. Rheumatologists represented the largest specialty to order combined genotyping and phenotyping, followed by gastroenterologists and internal and family medicine specialists.

5. Conclusions

This study identified significant correlation between genetic variants in TPMT and TPMT enzyme levels and between age and enzyme levels. No correlation was identified between sex and enzyme levels nor between TPMT and thiopurine metabolites, the small sample size of the latter group entails further investigation of whether this was a true finding or was merely a result of lack of statistical power. Despite the cost difference, providers continue to rely more significantly on genotyping only approach, rather than genotyping and phenotyping. The lower genotype–phenotype correlation in this study highlights the need for investigating rare TPMT variants or promoter repeats and the need for identifying treatment or environmental factors that can affect TPMT phenotyping for a meaningful approach to evaluating patients undergoing thiopurines treatment.

Funding Statement

This work has been funded by The University of Utah, Department of Pathology, University Development Funds and ARUP Research and Development funds.

Article highlights

• Adding NUDT15 genotyping to TPMT genotyping allowed for identifying additional patients that would need thiopurines dose adjustments based on published guidelines.

• There is significant correlation between TPMT polymorphisms and TPMT enzyme levels as well as between age and enzyme levels. No correlation was identified between sex and enzyme levels.

• The more common practice providers adopt is ordering TPMT and NUDT15 genotyping rather than genotyping in addition to phenotyping.

Author contributions

Authors SS, YJ, and KJD were responsible for study conception and design. Authors SS and KJD were responsible for acquisition of data. Authors SS, YJ, BSW, and KJD were responsible for data analysis and drafting and revision of the manuscript.

Disclosure statement

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Ethical conduct of research

Data used in this study was extracted following the policies of the University of Utah Institutional Review Board (IRB) and the procedures outlined in IRB protocol #00082990 and have followed the principles outlined in the Declaration of Helsinki for all experimental investigations. All data were analyzed anonymously.