TPMT and NUDT15 Genotyping Recommendations: A Joint Consensus Recommendation of the Association for Molecular Pathology, Clinical Pharmacogenetics Implementation Consortium, College of American Pathologists, Dutch Pharmacogenetics Working Group of the Royal Dutch Pharmacists Association, European Society for Pharmacogenomics and Personalized Therapy, and Pharmacogenomics Knowledgebase

Abstract

The goals of the Association for Molecular Pathology Clinical Practice Committee's Pharmacogenomics (PGx) Working Group are to define the key attributes of pharmacogenetic alleles recommended for clinical testing and a minimum set of variants that should be included in clinical PGx genotyping assays. This article provides recommendations for a minimum panel of variant alleles (Tier 1) and an extended panel of variant alleles (Tier 2) that will aid clinical laboratories when designing assays for PGx testing. The Association for Molecular Pathology PGx Working Group considered the functional impact of the variant alleles, allele frequencies in multiethnic populations, the availability of reference materials, as well as other technical considerations for PGx testing when developing these recommendations. The ultimate goal of this Working Group is to promote standardization of PGx gene/allele testing across clinical laboratories. This article focuses on clinical TPMT and NUDT15 PGx testing, which may be applied to all thiopurine S-methyltransferase (TPMT) and nudix hydrolase 15 (NUDT15)–related medications. These recommendations are not to be interpreted as prescriptive, but to provide a reference guide.

Pharmacogenomic (PGx) testing has become an important tool to help clinicians select medications and prescribe an appropriate dose for certain patients. Several studies have documented the wide variation of alleles included in clinical PGx tests.^1,2 Some tests are designed to detect many variant alleles in a PGx gene, whereas other tests interrogate only a limited number of alleles and may not include variants that are most relevant to clinical care. This can lead to a patient's genotype being reported as a ∗1 normal allele or no variants detected, even though clinically relevant variants may be present, which may cause discrepancies in interpretation, complicate advancement and implementation of PGx testing, and, ultimately, impact patient care. Until recently, there has been little effort to standardize the specific alleles that should be included in clinical PGx tests.

To address this issue, the Association for Molecular Pathology (AMP) Pharmacogenomics (PGx) Working Group has developed a series of documents that recommend a minimum set of variant alleles to include in clinical PGx assays to facilitate standardization across laboratories and ensure that the most clinically relevant variant alleles are included in clinical PGx assays. The previously reported recommendations included CYP2C19³ and CYP2C9,⁴ genes important for warfarin PGx testing,⁵ and CYP2D6.⁶ This article focuses on two genes, TPMT encoding thiopurine methyltransferase and NUDT15 encoding nudix hydrolase 15, and is intended to provide guidance to clinical laboratorians and assay manufacturers who develop, validate, and/or offer clinical TPMT and NUDT15 genotyping assays. This document should be implemented together with other relevant clinical guidelines, including those published by the Clinical Pharmacogenetics Implementation Consortium (CPIC), the Dutch Pharmacogenetics Working Group, the Canadian Pharmacogenomics Network for Drug Safety (for TPMT variants on cisplatin-induced hearing loss),⁷ and the French National Network of Pharmacogenetics, all of which focus primarily on the interpretation of PGx test results and therapeutic recommendations for specific drug–gene pairs (https://www.pharmgkb.org/guidelineAnnotations, last accessed January 20, 2022).

The AMP PGx Working Group uses a two-tier strategy for selection criteria in recommending PGx variants for clinical testing.3, 4, 5, 6 Briefly, Tier 1 recommended variant alleles are those that meet the following criteria: i) have a well-characterized effect on the function of the protein and/or gene expression, ii) have an appreciable minor allele frequency in a population/ethnicity group, iii) have publicly available reference materials, and iv) are technically feasible for clinical laboratories to interrogate using standard molecular testing methods. Tier 2 recommended variant alleles meet at least one but not all of the Tier 1 criteria; however, the Tier 2 alleles may be reclassified to Tier 1 in the future when additional information and/or reference material(s) becomes available.

Thiopurine medications (6-mercaptopurine, 6-thioguanine, and azathioprine) commonly are prescribed as chemotherapeutic agents for the treatment of acute lymphoblastic leukemia, as immunosuppressant agents for the treatment of autoimmune disorders (eg, inflammatory bowel disease and rheumatoid arthritis), and to prevent organ transplant rejection.⁸ However, they require extensive intracellular metabolism to generate active metabolites that are responsible for therapeutic efficacy. Importantly, two key detoxifying enzymes, thiopurine S-methyltransferase (TPMT) and nudix hydrolase 15 (NUDT15), act as negative modulators of both thiopurine activation and toxicity of thiopurine medications. The enzymatic activity of TPMT and NUDT15 is influenced directly by germline genetic variability in the TPMT and NUDT15 genes. Variant alleles are found in the general population at variable frequencies and contribute to interindividual differences in the pharmacologic and safety response profiles of thiopurine medications. As such, clinical PGx testing for genetic variants in TPMT and NUDT15 associated with diminished or absent enzymatic activity can identify individuals who are at increased risk of adverse drug toxicity with standard doses of thiopurine medications.

TPMT

The TPMT gene is located on chromosome 6p22.3, and was originally described as spanning 34 kb, and having 10 exons.⁹ However, the MANE (Matched Annotation from NCBI and EMBL-EBI; https://www.ncbi.nlm.nih.gov/refseq/MANE, last accessed August 17, 2022) current NM_000367.5 transcript, used as the reference in this manuscript, has 9 exons (8 coding and 1 non-coding). Functional variants in the TPMT gene were found to contribute to the trimodal distribution of TPMT enzyme activity in red blood cells among individuals of European descent, with 89% having normal to high TPMT activity, 11% having low or intermediate activity, and approximately 0.3% of individuals having very low to absent TPMT activity.¹⁰ Subsequently, three alleles of the TPMT gene were identified to account for approximately 90% of reduced activity phenotypes among individuals of European descent: TPMT∗2, ∗3A, and ∗3C.¹¹ Individuals who are heterozygous for one of these variants have approximately 50% enzyme activity and are predicted to be intermediate metabolizers, whereas individuals who are homozygous or compound heterozygous for these variants have very low to absent TPMT activity and are poor metabolizers. Importantly, both intermediate and poor TPMT metabolizers who receive a standard dose of thiopurine drugs are at increased risk of developing life-threatening myelosuppression.

NUDT15

Nudix hydrolase 15 [or nucleoside diphosphate-linked moiety X-type motif 15 (NUDT15); EC 3.6.1.9, also known as MutT homolog 2] is a member of the nudix hydrolase superfamily, which plays an important role in preventing transversion mutations in DNA. The NUDT15 enzyme preferentially hydrolyzes 6-thiodeoxyguanosine triphosphate and 6-thioguanosine triphosphate by conversion to 6-thio(deoxy)guanosine monophosphate, which prevents their incorporation into DNA and leads to reduced cytotoxic and immunosuppressant effects of thiopurine drugs.¹² The human NUDT15 gene is located on chromosome 13q14.2, spans 15 kb, and has three exons. A genetic variant, p.Arg139Cys, which is present in both NUDT15∗2 and ∗3, was first reported in 2014 to contribute to thiopurine toxicity by a genome-wide association study, and subsequently was replicated by several studies.13, 14, 15, 16 Multiple loss-of-function NUDT15 variants have been reported. The contribution of NUDT15 to thiopurine-related adverse effects such as leukopenia is as high as 22% among all ethnicities.¹⁶

Existing Clinical Guidelines and Recommendations

The US Food & Drug Administration has included information about genetic variants in TPMT (for azathioprine, mercaptopurine, and thioguanine) and NUDT15 (for mercaptopurine and thioguanine) in the associated drug labels and the Pharmacogenetic Associations for which the Data Support Therapeutic Management Recommendations table (https://www.fda.gov/medical-devices/precision-medicine/table-pharmacogenetic-associations, last accessed January 20, 2022). In addition, dosing guidelines for thiopurines (azathioprine, mercaptopurine, and thioguanine) also have been published by CPIC¹⁷ and the Dutch Pharmacogenetics Working Group (https://www.pharmgkb.org/guidelineAnnotations, last accessed January 20, 2022). Both CPIC and the Dutch Pharmacogenetics Working Group guidelines recommend specific dose reductions for individuals who have low or deficient enzyme activity (ie, TPMT or NUDT15 intermediate and poor metabolizers), predicted by TPMT and/or NUDT15 genotyping. Of note, TPMT also may be tested using an enzyme activity assay; however, there are no clinically available assays for NUDT15 enzyme activity.

Testing Platforms

Selection of a molecular platform to use for testing PGx variants is based on many factors that include, but are not limited to, the technical feasibility of interrogating the genomic region of interest, cost, laboratory workflow, and test turnaround time required. Because both TPMT and NUDT15 genes reside on genomic regions that are amenable to testing using standard molecular techniques, PGx tests offered by clinical molecular genetics laboratories may use targeted genotyping or sequencing (Sanger sequencing or next-generation sequencing) approaches. However, it is up to the discretion of the specific laboratory to select the genotyping/sequencing method for clinical PGx test development and validation. Of note, most commonly used clinical molecular platforms (eg, genotyping, short-read next-generation sequencing) are unable to phase TPMT and NUDT15 variants. As such, assigning TPMT and NUDT15 diplotypes from raw genotyping or sequencing data are typically empiric or inferred.

Materials and Methods

The AMP PGx Working Group comprises subject matter experts from the CPIC, Pharmacogenomics Knowledgebase (PharmGKB), College of American Pathologists (CAP), Dutch Pharmacogenetics Working Group, and the PGx clinical testing and research communities. TPMT and NUDT15 variant alleles were reviewed and classified into tiers on the basis of four criteria: i) functional characterization of the allele (ie, whether it is known to affect expression of the gene or function of the encoded protein); ii) an appreciable minor allele frequency (MAF) in a population and/or ethnic group [in this TPMT and NUDT15 recommendation document, the Working Group used a MAF of ≥0.1% in at least one subpopulation for Tier 1 alleles, and ≥0.01% for Tier 2 alleles based on currently available information from applicable resources (https://www.pharmgkb.org/page/tpmtRefMaterials, https://www.pharmgkb.org/page/nudt15RefMaterials and https://gnomad.broadinstitute.org, last accessed January 20, 2022)]; iii) availability of reference materials (Table 1); and iv) technical feasibility for clinical laboratories to accurately interrogate the variant.

Table 1.

Reference Materials

Allele	Coriell†	Diplotype	Allele	Coriell†	Diplotype
TPMT∗2	HG00133	(∗1/∗2)	NUDT15∗2	HG00599	(∗2/∗32)
	HG01083	(∗1/∗2)		NA19079	(∗2/∗5)
TPMT∗3A	NA12753	∗1/∗3A	NUDT15∗3	NA18992	∗1/∗3
	NA15245	∗1/∗3A		HG00524	(∗3/∗3)
TPMT∗3B	None	-		NA18564	∗1/∗3
TPMT∗3C	HG00589	∗1/∗3C	NUDT15∗4	HG01359	∗1/∗4
	NA18855	∗1/∗3C		HG01465	(∗1/∗4)
TPMT∗4	None	-	NUDT15∗5	HG00437	∗1/∗5
TPMT∗5	None	-		NA23093	∗1/∗5
TPMT∗6	NA18603	(∗1/∗6)	NUDT15∗6	HG00277	(∗1/∗6)
TPMT∗7	None	-		NA07000	(∗1/∗6)
TPMT∗8	NA19176	∗1/∗8		NA07029	(∗1/∗6)
	NA23275	∗1/∗8	NUDT15∗7	None	-
TPMT∗9	None	-	NUDT15∗8	None	-
TPMT∗10	None	-	NUDT15∗9	HG01086	(∗2/∗9)
TPMT∗11	None	-	NUDT15∗10	None	-
TPMT∗12	NA12751	(∗1/∗12)	NUDT15∗11	None	-
TPMT∗13	None	-	NUDT15∗12	None	-
TPMT∗14	None	-	NUDT15∗13	None	-
TPMT∗15	None	-	NUDT15∗14	None	-
TPMT∗16	HG00276	(∗1/∗16)	NUDT15∗16	None	-
TPMT∗16	HG00304	(∗3A/∗16)	NUDT15∗15	None	-
TPMT∗17	None	-	NUDT15∗17	None	-
TPMT∗18	None	-	NUDT15∗18	None	-
TPMT∗19	None	-	NUDT15∗19	None	-
TPMT∗20	None	-	NUDT15∗20	None	-
TPMT∗21	NA12044	(∗1/∗21)
TPMT∗22	None	-
TPMT∗23	None	-
TPMT∗24	HG02496	(∗1/∗24)
TPMT∗25	None	-
TPMT∗26	None	-
TPMT∗27	None	-
TPMT∗28	None	-
TPMT∗29	None	-
TPMT∗30	None	-
TPMT∗31	None	-
TPMT∗32	NA10855	(∗1/∗32)
TPMT∗33	NA19026	(∗8/∗33)‡
TPMT∗34	None	-
TPMT∗35	None	-
TPMT∗36	None	-
TPMT∗37	None	-
TPMT∗38	None	-
TPMT∗39	None	-
TPMT∗40	HG01474	(∗1/∗40)
TPMT∗41	None	-
TPMT∗42	None	-
TPMT∗43	None	-
TPMT∗44	None	-
TPMT∗46	HG03521	(∗1/∗46)‡

Diplotypes in parentheses are included in Pratt et al.¹⁸

^†Novel haplotype; variants identifying TPMT∗8 and ∗33 were shown to be in cis by trio analysis using 1000 Genomes data (https://www.internationalgenome.org, last accessed January 20, 2022).

^‡The ∗8 and ∗33 variants were not phased in this sample.

These criteria received equal weight during the AMP PGx Working Group deliberations. In addition, commercially available genotyping platforms (Supplemental Table S1) were reviewed for assessing the ability of laboratories to implement the Working Group recommendations; however, these data were not used as a determinant of tier assignment.

The AMP PGx Working Group used MAF and functional information from the CPIC, PharmGKB, and scientific literature. Given that the CPIC clinical dosing guideline development process includes curating allele frequencies and function assignments, the PGx Working Group incorporated CPIC's curated information. In some populations, including those of European, Latino, and Near Eastern descent, the most common variant TPMT allele is ∗3A, which is a haplotype with the ∗3B and ∗3C alleles in cis. Single-variant frequencies can be found in databases such as the Genome Aggregation Database (https://gnomad.broadinstitute.org, last accessed January 20, 2022) and the International Genome Sample Resource (or 1000 Genomes, https://www.internationalgenome.org, last accessed January 20, 2022), but these resources do not report frequencies of allele haplotypes such as TPMT∗3A. The CPIC relies on studies from peer-reviewed published literature for these allele frequencies. However, there are caveats with obtaining frequencies from published studies. For example, most studies test for a limited number of alleles, and some alleles rarely are assayed or reported in the literature. In addition, most studies default allele assignment to TPMT∗1 if none of the interrogated variants are detected. As a result, allele frequencies can be underestimated, overestimated, or simply not reported. The CPIC allele clinical function assignment is based on literature reviews of allele function by guideline (https://cpicpgx.org/resources, last accessed January 20, 2022), which may not be consistent with the CPIC-defined biochemical function of the allele.

TPMT nomenclature is maintained at Linköping University in Linköping, Sweden (https://liu.se/en/research/tpmt-nomenclature-committee, last accessed January 20, 2022). The website currently catalogs 43 variant TPMT star (∗) alleles, TPMT∗2-∗44. NUDT15 nomenclature was introduced by the Pharmacogene Variation Consortium (https://www.pharmvar.org, last accessed January 20, 202).¹⁹ The Pharmacogene Variation Consortium lists 20 variant NUDT15 star (∗) alleles, NUDT15∗2-∗20. All NUDT15 star alleles are defined by a single variant with the exception of NUDT15∗2, which is defined by two variants in cis that also can occur separately. Although the Pharmacogene Variation Consortium currently lists 20 NUDT15 alleles, the CPIC last evaluated NUDT15 alleles up to ∗9 for the 2018 thiopurine guideline update.

To determine the frequency of clinical laboratories that currently test for TPMT and/or NUDT15 variants, aggregate data from the CAP proficiency testing (PT) program was obtained from the 2021 PGX-A mailing.²⁰ The CAP PGx PT program includes both US-based and international participants.⁶ External assessment programs in Europe such as the European Molecular Genetics Quality Network and the Reference Institute for Bioanalytics also were reviewed.

Results

Tier 1 TPMT and NUDT15 Variant Alleles

TPMT and NUDT15 star (∗) alleles recommended for inclusion in Tier 1 include TPMT∗2, ∗3A, ∗3B, ∗3C, and NUDT15∗3 (Table 2). For Human Genome Variation Society nomenclature throughout, see https://www.ncbi.nlm.nih.gov/snp and http://www.ncbi.nlm.nih.gov/clinvar (last accessed January 20, 2022); allele frequency by population information is from https://www.pharmgkb.org/page/tpmtRefMaterials (last accessed January 20, 2022) unless otherwise specified.

Table 2.

Tier 1 TPMT and NUDT15 Variant Alleles

Allele	Allele functional status†	Core variant(s)‡	RefSeqGene LRG nomenclature	HGVS genomic nomenclature (GRCh38)	HGVS cDNA nomenclature	HGVS protein nomenclature	Reference material available	Multiethnic allele frequency, %
TPMT∗2	No function	rs1800462	NG_012137.3: g.16420G>C	NC_000006.12: g.18143724C>G	NM_000367.5: c.238G>C	NP_000358.1: p.Ala80Pro	Yes§	0–0.7
TPMT∗3A	No function	rs1800460, rs1142345	NG_012137.3: g.21147G>A, NG_012137.3: g.29457A>G	NC_000006.12: g.18138997C>T, NC_000006.12: g.18130687T>C	NM_000367.5: c.460G>A, NM_000367.5: c.719A>G	NP_000358.1: p.Ala154Thr, NP_000358.1: p.Tyr240Cys	Yes	0.03–4.2
TPMT∗3B	No function	rs1800460	NG_012137.3: g.21147G>A	NC_000006.12: g.18138997C>T	NM_000367.5: c.460G>A	NP_000358.1: p.Ala154Thr	Yes (∗3A)	0–0.5
TPMT∗3C	No function	rs1142345	NG_012137.3: g.29457A>G	NC_000006.12: g.18130687T>C	NM_000367.5: c.719A>G	NP_000358.1: p.Tyr240Cys	Yes	0.6–5.3
NUDT15∗3	No function	rs116855232	NG_047021.1: g.13153C>T	NC_000013.11: g.48045719C>T	NM_018283.4: c.415C>T	NP_060753.1: p.Arg139Cys	Yes§	0–6.8

HGVS, Human Genome Variation Society; LRG, Locus Reference Genomic.

^†Citations for assignment of function can be found at https://www.pharmvar.org; HGVS nomenclature https://www.ncbi.nlm.nih.gov/snp; and http://www.ncbi.nlm.nih.gov/clinvar (last accessed January 20, 2022).

^‡Core variant(s) can be found at https://www.pharmvar.org (last accessed January 20, 2022).

^§Pratt et al.¹⁸

TPMT∗2

The no-function TPMT∗2 allele is characterized by a missense variant in exon 4 (NM_000367.5:c.238G>C, p.Ala80Pro, rs18004622).21, 22, 23, 24, 25 This allele is found most commonly in those of African ancestry with an allele frequency of 0.5% and is observed in the Latino and European populations at frequencies of 0.35% and 0.2%, respectively. It is observed at a frequency of less than 0.05% in Central/South and East Asian populations.

TPMT∗3A

The no-function TPMT∗3A allele is characterized by two missense variants in cis, one in exon 6 (NM_000367.5:c.460G>A, p.Ala154Thr, rs1800460; also known as TPMT∗3B) and one in exon 9 (NM_000367.5:c.719A>G, p.Tyr240Cys, rs1142345; also known as TPMT∗3C).22, 23, 24, 25, 26 This allele is found most commonly in the Latino and European populations at allele frequencies of 4.2% and 3.4%, respectively. It is observed at a frequency of less than 1% in African ancestries and less than 0.5% in Central/South and East Asian populations. The definitive assignment of the TPMT∗3A allele may be challenging because it harbors variants that independently define TPMT∗3B and TPMT∗3C. For example, the TPMT ∗1/∗3A diplotype assumes that both variants are present on the same chromosome (in cis), which is the most likely diplotype based on population allele frequencies. However, because many testing approaches cannot determine the phase of the two variants, the rare poor metabolizer TPMT ∗3B/∗3C diplotype (with the two variants in trans) cannot be ruled out without subsequent phasing or trio studies.

TPMT∗3B

The no-function TPMT∗3B allele is characterized by a missense variant in exon 6 (NM_000367.5:c.460G>A, p.Ala154Thr, rs1800460).^22,24, 25, 26 This allele is found most commonly in the European and Latino populations at allele frequencies of 0.3% and 0.2%, respectively. It is observed at a frequency close to 0.2% in Central/South Asian populations.

TPMT∗3C

The no-function TPMT∗3C allele is characterized by a missense variant in exon 9 (NM_000367.5:c.719A>G, p.Tyr240Cys, rs1142345).^22,24, 25, 26 This allele is found most commonly in the African and East Asian populations at allele frequencies of 2.4% and 1.6%, respectively. It is observed at a frequency of 1.1% in Central/South Asian populations, and less than 1% in Latino and European populations.

NUDT15∗3

The no-function NUDT15∗3 allele is characterized by a missense variant in exon 3 (NM_018283.4:c.415C>T, p.Arg139Cys, rs116855232).^19,27 This allele is found most commonly in the Central/South and East Asian populations at allele frequencies of 6.7% and 6.1%, respectively. It is observed at a frequency of less than 1% in Latino and European populations.

Tier 2 TPMT and NUDT15 Variant Alleles

TPMT and NUDT15 star (∗) alleles recommended for inclusion in Tier 2 include TPMT∗11, ∗29, ∗42, and NUDT15 ∗2, ∗4, ∗6, ∗9, and ∗14 (Table 3).

Table 3.

Tier 2 TPMT and NUDT15 Variant Alleles

Allele	Allele functional status†	Core variant(s)‡	RefSeqGene LRG nomenclature	HGVS genomic nomenclature (GRCh38)	HGVS cDNA nomenclature	HGVS protein nomenclature	Reference material available	Multiethnic allele frequency, %
TPMT∗11	No function	rs72552738	NG_012137.3: g.20455G>A	NC_000006.12: g.18139689C>T	NM_000367.5: c.395G>A	NP_000358.1: p.Cys132Tyr	No	0–0.02
TPMT∗29	No function	rs267607275	NG_012137.3: g.11018T>C	NC_000006.12: g.18149126A>G	NM_000367.5: c.2T>C	NP_000358.1: p.Met1Thr	No	0–0.03
TPMT∗42	Likely no function	rs759836180	NG_012137.3: g.11111dup	NC_000006.12: g.18149034dup	NM_000367.5: c.95dup	NP_000358.1: p.Trp33Valfs∗26	No	0–0.11
NUDT15∗2	No function	rs746071566, rs116855232	NG_047021.1: g.5218GAGTCG[4], NG_047021.1: g.13153C>T	NC_000013.11: g.48037796_48037801dup, NC_000013.11: g.48045719C>T	NM_018283.4: c.50_55dup, NM_018283.4: c.415C>T	NP_060753.1: p.Gly17_Val18dup, NP_060753.1: p.Arg139Cys	Yes§	0–3.7
NUDT15∗4	Uncertain function	rs147390019	NG_047021.1: g.13154G>A	NC_000013.11: g.48045720G>A	NM_018283.4: c.416G>A	NP_060753.1: p.Arg139His	Yes	0–1.8
NUDT15∗6	Uncertain function	rs746071566	NG_047021.1: g.5218GAGTCG[4]	NC_000013.11: g.48037796_48037801dup	NM_018283.4: c.50_55dup	NP_060753.1: p.Gly17_Val18dup	Yes§	0–1.3
NUDT15∗9	No function	rs746071566	NG_047021.1: g.5218GAGTCG[2]	NC_000013.11: g.48037796_48037801del	NM_018283.4: c.50_55del	NP_060753.1: p.Gly17_Val18del	Yes§	0–0.2
NUDT15∗14	Likely no function	rs777311140	NG_047021.1: g.5260_5261insCGGG	NC_000013.11: g.48037826_48037827insCGGG	NM_018283.4: c.80_81insCGGG	NP_060753.1: p.Cys28Glyfs∗28	No	0–0.6

HGVS, Human Genome Variation Society; LRG, Locus Reference Genomic.

^†Citations for assignment of function can be found at https://www.pharmvar.org; HGVS nomenclature https://www.ncbi.nlm.nih.gov/snp; and http://www.ncbi.nlm.nih.gov/clinvar (last accessed January 20, 2022).

^‡Core variant(s) can be found at https://www.pharmvar.org (last accessed January 20, 2022).

^§Pratt et al.¹⁸

TPMT∗11

The no-function TPMT∗11 allele is characterized by a missense variant in exon 5 (NM_000367.5:c.395G>A, p.Cys132Tyr, rs72552738).^22,24,28 This allele has been found in the European population at a frequency of less than 0.05%.

TPMT∗29

The no-function TPMT∗29 allele is characterized by a start codon variant in exon 2 (NM_000367.5:c.2T>C, p.Met1Thr, rs267607275) that is predicted to result in loss of translation and loss of enzyme function similar to another start codon allele: TPMT∗14.²⁹ rs267607275 is triallelic and also can occur as NM_000367.5:c.2T>G (p.Met1Arg). TPMT∗29 has a frequency of less than 0.05% in the East Asian population.

TPMT∗42

The TPMT∗42 allele is characterized by an insertion of a single nucleotide in exon 2, causing a frameshift in the open reading frame (NM_000367.5:c.95dup, p.Trp33Valfs∗26, rs759836180). There is no CPIC-assigned clinical function for this allele; however, it is likely a no-function allele owing to a frameshift resulting in a premature stop codon predicted to result in a nonfunctional truncated TPMT protein or loss of protein expression via mRNA nonsense-mediated decay.³⁰ This allele is observed at a frequency of 0.11% in the Finnish European population (https://gnomad.broadinstitute.org/variant/6-18149263-C-CT, last accessed January 20, 2022).

NUDT15∗2

The no-function NUDT15∗2 allele is characterized by a missense variant in exon 3 (NM_018283.4:c.415C>T, p.Arg139Cys, rs116855232; also present in NUDT15∗3) and an in-frame insertion in exon 1 (NM_018283.4:c.50_55dup, p.Gly17_Val18dup, rs746071566; also present independently as NUDT15∗6)^19,27 (Figure 1A). NUDT15∗2 is found in the Latino and East Asian populations at allele frequencies of 3.7% and 3.5%, respectively.

Figure 1.

A and B: Graphic overview of the variable NUDT15 6-bp repeat (rs746071566). The reference sequence contains three copies of the 6-bp GAGTCG motif (top line, repeat units shown in alternating blue and purple color). A: Insertion (gain) of one repeat unit (shown in red). Although the Pharmacogene Variation Consortium (PharmVar) and the Human Genome Variation Society (HGVS) use the 3′ alignment rule positioning the inserted motif at the 3′ end, their annotations differ: PharmVar describes the insertion as c.55_56insGAGTCG (p.18_19insGV), whereas the HGVS describes the insertion as duplication c.50_55dup (p.Gly_Val18dup). In contrast, the Single Nucleotide Polymorphism Database (dbSNP) provides the number of motifs annotating the insertion as c.38GAGTCG[4] (p.13_14GV[4]). Regardless of which annotation is used, the insertion leads to a gain of two amino acids. Notably, using the 5′ alignment rule, Moriyama et al,²⁷ who first discovered this variant, described the repeat as a GGAGTC insertion (c.36insGGAGTC; p.12_13insGV, not shown), but used the 3′ rule in subsequent reports. B: Deletion (loss) of one repeat unit (shown in red, repeat units shown in alternating blue and purple color). In this instance, PharmVar and the HGVS describe the deletion as c.50delGAGTCG (p.GV17_18del) and c.50_55del (p.Gly17_Val 18del), respectively. As shown, dbSNP provides c.38GAGTCG[2] (p.13_14GV[2]), specifying the number of repeat motifs. Regardless of which annotation is used, the deletion results in the loss of two amino acids. The figure was adapted with permission from PharmVar. del, deletion; dup, duplication.

NUDT15∗4

NUDT15∗4 is an allele with uncertain function characterized by a missense variant in exon 3 (NM_018283.4:c.416G>A, p.Arg139His, rs147390019).^19,27 This allele is found most commonly in the Latino population at an allele frequency of 1.8%, and is observed at a frequency of 0.1% in the East Asian population. The defining variant in NUDT15∗4, c.416G>A, is adjacent to the defining variant in NUDT15∗3 (c.415C>T) and affects the same amino acid. Therefore, this variant was included in Tier 2 despite its uncertain function.

NUDT15∗6

NUDT15∗6 is an allele with uncertain function characterized by an in-frame insertion of a 6-bp repeat (GAGTCG) in exon 1 (c.50_55dup, p.Gly17_Val18dup, rs746071566; also present in NUDT15∗2) that results in four copies of the GAGTCG repeat^19,27 (Figure 1A). The NUDT15∗6 allele frequency is approximately 1.3% in the East Asian population, but less than 0.5% in the European, Central/South Asian, and Latino populations. This allele is included in Tier 2 despite its uncertain function because detection of the duplication is required to accurately detect and differentiate NUDT15∗2 from NUDT15∗3.

NUDT15∗9

The no-function NUDT15∗9 allele is characterized by an in-frame deletion of the 6-bp repeat (GAGTCG) in exon 1 (NM_018283.4:c.50_55del, p.Gly17_Val18del, rs746071566) that results in two GAGTCG repeat copies^19,31,32 (Figure 1B). This allele has a frequency of 0.2% in the European population, but a frequency of less than 0.05% in the Central/South Asian and Latino populations.

NUDT15∗14

NUDT15∗14 is an allele with no assigned clinical function, but is likely no-function owing to a frameshift insertion of 4 bp (CGGG) in exon 1 (NM_018283.4:c.80_81insCGGG, p.Cys28Glyfs∗28, rs777311140)¹⁹ that results in a truncated protein or mRNA nonsense-mediated decay. The NUDT15∗14 allele frequency is approximately 0.6% in the European (Finnish) population, but less than 0.05% in the non-Finnish European and African/African American populations (https://gnomad.broadinstitute.org/variant/13-48611961-C-CGCGG?dataset=gnomad_r2_1, last accessed January 20, 2022).

Discussion

In this guideline, we recommend a minimum set of alleles for clinical laboratories conducting TPMT and NUDT15 genotyping to identify patients at increased risk for thiopurine toxicity. The goals of these recommendations are to promote standardization across clinical laboratories and to ensure that they interrogate the most common multiethnic, function-altering TPMT and NUDT15 variant alleles associated with thiopurine toxicity. Pharmacogenetic variants commonly have variable allele frequencies across different populations and ancestral groups, which includes TPMT and NUDT15 alleles. Accurate and comprehensive TPMT and NUDT15 allele frequencies still are emerging, particularly for rare variants; however, it is estimated that the AMP PGx Tier 1 recommended TPMT variants account for >70% of the reported TPMT variant alleles, including uncertain function alleles across all populations. The low-frequency AMP PGx Tier 2 recommended TPMT variants account for an additional approximately 1% of the reported TPMT variant alleles in all populations. Of note, TPMT∗2, ∗3A, and ∗3C account for 90% of the reduced activity phenotypes in Europeans.^11,19,33,34 Moreover, it also is estimated that the AMP PGx Tier 1 NUDT15 variants account for >50% and the Tier 2 NUDT15 variants account for an additional >40% (Tier 1 and Tier 2 combined, approximately 90%) of the cataloged NUDT15 variant alleles in the general population (https://www.pharmvar.org/gene/NUDT15, last accessed January 20, 2022).

TPMT and NUDT15 alleles associated with loss of enzymatic activity are major contributors to thiopurine-induced myelosuppression. TPMT metabolizes mercaptopurine and thioguanine to inactive forms. Individuals with one (heterozygotes) or two (homozygotes or compound heterozygotes) no-function TPMT alleles have intermediate or poor enzyme activity, respectively, resulting in decreased thiopurine metabolism and more parent drug available for conversion to the active thioguanine (TGN) metabolites. Individuals who are heterozygous for no-function TPMT variants treated with conventional doses of thiopurines are at increased risk for myelosuppression as a result of the accumulation of TGNs, and homozygotes are at markedly increased risk. In addition, NUDT15 converts TGNs to less-toxic metabolites, and variant NUDT15 alleles that result in reduced enzyme activity also increase the risk for excessive TGN concentrations and toxicity.

Because variant TPMT and NUDT15 alleles can be associated with severe toxicity to thiopurine drugs, the Working Group recommends that no-function alleles with an observed MAF of 0.1% or greater in any human subpopulation be included in the Tier 1 minimum set of alleles, provided they meet the other qualifications for Tier 1. Variant alleles associated with no-function that are present in 0.01% or greater of any subpopulation are included in Tier 2 for laboratories that wish to offer more comprehensive genotyping. Variant alleles with >0.1% MAF, lack reference materials for validation, and/or are technically difficult to interrogate also are included in Tier 2. Very rare variant alleles and those without an effect on enzyme function are not included in the Tier 1 or Tier 2 recommendations for clinical genotype testing.

In addition to targeted TPMT genotyping, TPMT phenotyping assays directly can measure TPMT enzymatic activity, which has the advantage of detecting individuals with low TPMT activity as a result of rare variant alleles. However, this type of functional testing should not be performed in individuals with recent blood transfusions or in individuals who already have been treated with a thiopurine drug because TPMT is an enzyme found in red blood cells. In addition, NUDT15 enzymatic testing is not readily available clinically. Therefore, TPMT and NUDT15 genotype testing is a reasonable approach to identify individuals at increased risk of thiopurine toxicity and is widely available clinically.

The Tier 1 list of alleles is designed to test the most common variant alleles to maximize detection rate in most populations. As with any targeted genotyping approach, rarer alleles may exist that are not detected. Laboratories conducting genotyping assays should include a list of alleles that are detected in their laboratory reports, the detection rate (if known), and a caveat that rare alleles may not be detected. In addition, clinical laboratories should follow best practices for assay validation and adhere to the applicable regulatory requirements for their location.

More than 150 TPMT coding variants have been identified through genome and exome sequencing approaches (https://gnomad.broadinstitute.org/gene/ENSG00000137364?dataset=gnomad_r2_1, last accessed January 20, 2022).³⁵ Whether sequencing is useful in detecting additional rare variations contributing to thiopurine toxicity, beyond the known TPMT and NUDT15 alleles, has been the focus of several studies.35, 36, 37 Sequencing has detected additional alleles of uncertain function (eg, TPMT∗16) in individuals with low TPMT activity, but without a TPMT∗2, ∗3A, ∗3B, or ∗3C variant, illustrating the potential role of sequencing to increase clinical sensitivity and in defining allele function.³⁶ However, sequencing approaches also are expected to uncover few novel and rare TPMT variants of clinical significance.¹⁷ Less is known about the potential for sequencing to reveal additional rare variants contributing to low NUDT15 activity.

More than 100 NUDT15 coding variants have been identified through genome and exome sequencing approaches (https://gnomad.broadinstitute.org/gene/ENSG00000136159?dataset=gnomad_r2_1, last accessed January 20, 2022), however, most of these variants have not been cataloged by the Pharmacogene Variation Consortium because the variants have not been submitted. Many laboratories that perform NUDT15 genotyping only target the c.415C>T (rs116855232, p.Arg139Cys) variant that is present in both NUDT15∗2 and ∗3 and may not differentiate between these two alleles because the NUDT15∗2-specific variant (c.50_55dup, p.Gly17_Val18dup, rs746071566) is technically challenging to interrogate. However, this distinction is not clinically relevant because both NUDT15∗2 and ∗3 are no-function alleles. Therefore, the clinical PGx genotype may be reported as “NUDT15∗2 or ∗3” if only c.415C>T is targeted, or some laboratories may default to reporting it as NUDT15∗3. Another caveat for NUDT15 genotyping is that the NUDT15∗4 variant, c.416G>A (rs147390019, p.Arg139His), occurs adjacent to the c.415C>T variant in the NUDT15∗2 and ∗3 alleles. Depending on the chemistry of the genotyping method, this could lead to incorrect genotype calls or interfere with assay performance and allele calling. In addition, annotation of the repeat motif in NUDT15 (ie, NUDT15∗2, NUDT15∗6, and NUDT15∗9) can differ depending on how the sequences are aligned and the motif units are defined. Figure 1 illustrates that the sequence changes are the same despite differences in nomenclature. It is recommended that clinical laboratories use Human Genome Variation Society³⁸ nomenclature.

TPMT and NUDT15 Alleles Not Included in Tier 1 or Tier 2 Recommendations

The TPMT∗8, TPMT∗33, NUDT15∗5, and NUDT15∗7 alleles were not included in either Tier 1 or Tier 2 owing to their uncertain function. TPMT∗8 is characterized by a missense variant in exon 9 (NM_000367.5:c.644G>A, p.Arg215His, rs56161402).^22,24,39,40 The TPMT∗8 allele is most common in those of African ancestry, with an allele frequency ranging from 0.7% to 2.4%. This allele also has been found in the Latino population at a frequency of 0.1%, and at a less than 0.01% frequency in European and Central/South Asian populations. TPMT∗33 (NM_000367.5:c.487C>T, p.Arg163Cys, rs112339338) is common in individuals with African ancestry, with a frequency of 0.20% (https://gnomad.broadinstitute.org/variant/6-18139201-G-A?dataset=gnomad_r2_1, last accessed January 20, 2022). NUDT15∗5 is characterized by a missense variant in exon 1 (NM_018283.4:c.52G>A, p.Val18Ile, rs186364861)^19,27 and is observed in the East Asian population at an allele frequency of 1.1%. NUDT15∗7 is characterized by a missense variant in exon 1 (NM_018283.4:c.101G>C, p.Arg34Thr, rs766023281),^19,27 and its frequency is approximately 0.06% in the East Asian population.

The no-function TPMT∗4 allele is characterized by a splicing defect variant (NM_000367.5:c.626-1G>A, rs1800584) in the splice junction between intron 8 and exon 9 that results in a protein with very low enzyme activity. This allele has been found in the European population at a frequency less than 0.01%.^41,42 Although many laboratories include TPMT∗4 for clinical testing (Table 4), it was not included in Tier 2 because it is a rare allele below the MAF cut-off value of 0.01%.

Table 4.

Alleles Included in Proficiency Testing

Gene	Allele	Tier	CAP PGX		European Molecular Genetics Quality Network	Reference Institute for Bioanalytics
			n	%
TPMT	∗2	1	58	89.2	X	X
	∗3A	1	59	90.8	X	X
	∗3B	1	60	92.3	X	X
	∗3C	1	61	93.9	X	X
	∗4	No tier	36	55.4
NUDT15	rs116855232 (∗2 and ∗3)	1	33	50.8
	rs147390019 (∗4)	2	22	33.9
	rs186364861 (∗5)	2	22	33.9

The number/percentage of laboratories that report testing for specific TPMT and/or NUDT15 alleles is provided for the College of American Pathologists Pharmacogenetics (CAP PGX) proficiency testing, based on the 2021 PGX-A mailing.²⁰ The alleles included in the European Molecular Genetics Quality Network and Reference Institute for Bioanalytics surveys are indicated with an “X.”

TPMT∗41 (NM_000367.5:c.719A>C, p.Tyr240Ser, rs1142345) is also a rare no-function allele; however, it is triallelic with TPMT∗3C (NM_000367.5:c.719A>G, p.Tyr240Cys, rs1142345). The frequency of this allele is not well characterized, and thus it was not included in Tier 1 or Tier 2. It is important to note that TPMT∗41 may interfere with targeted testing for TPMT∗3C.

Proficiency Testing/External Quality Assessment

Several PT or external quality assessment programs are commercially available for TPMT genotyping; however, aside from CAP, programs that include NUDT15 are not widely available at this time. Of the 217 total participants in the CAP 2021 PGX-A mailing, 65 laboratories provided information on the TPMT and/or NUDT15 alleles that are included in testing.²⁰ Of the 65 laboratories that responded, 27 (41.5%) indicated that they offer clinical testing for both TPMT and NUDT15, 32 (49.2%) only offer TPMT, 1 (1.5%) only offers NUDT15, and 5 (7.7%) do not offer testing for either gene clinically. The number and percentage of the 65 laboratories who include each allele is listed in Table 4. Of note, TPMT∗3A includes both the c.460G>A and the c.719A>G variants. A subset of laboratories responded that among the TPMT∗3 alleles they could detect only ∗3A and ∗3C (n = 1; 1.5%), only ∗3A (n = 1; 1.5%), or only ∗3B and ∗3C (3; 4.6%). This suggests that all laboratories testing TPMT can detect both of the variants that comprise the ∗3A allele, but the differences in reporting may reflect a limitation in the format of the question in the CAP PT survey. In Europe, there are two major proficiency testing vendors, the Reference Institute for Bioanalytics and the European Molecular Genetics Quality Network. Both include TPMT but not NUDT15. In addition, the Dutch Proficiency Testing Program is available for some pharmacogenes, but not TPMT or NUDT15. The alleles that are tested in these European PT surveys are indicated in Table 4; however, the number and percentage of laboratories that test each allele and subscribe to either survey is not available.

Limitations

This article focuses only on allele recommendations for clinical laboratory genotyping assays that include TPMT and NUDT15. This professional guideline does not include mapping genotypes to phenotypes, clinical interpretation of test results, or any recommendations for medication management based on genotype because these were considered to be out of scope for this article and/or available from other resources (eg, US Food & Drug Administration labels, CPIC. and PharmGKB). Although technical challenges related to interrogating TPMT and NUDT15 have been discussed in this article, the Working Group does not recommend or endorse any particular molecular testing platforms for clinical TPMT and NUDT15 genotyping.

Conclusions

This professional guideline provides recommendations for TPMT and NUDT15 alleles to be included in clinical PGx genotyping assays. These recommendations are intended to facilitate the design and implementation of PGx testing by clinical laboratories. In addition, these recommendations are intended to promote test standardization and genotype concordance across clinical laboratories.

Disclaimers

The Association for Molecular Pathology (AMP) Clinical Practice Guidelines and Reports are developed to be of assistance to laboratory and other health care professionals by providing guidance and recommendations for particular areas of practice. The Guidelines or Report should not be considered inclusive of all proper approaches or methods, or exclusive of others. The Guidelines or Report cannot guarantee any specific outcome, nor do they establish a standard of care. The Guidelines or Report are not intended to dictate the treatment of a particular patient. Treatment decisions must be made based on the independent judgment of health care providers and each patient's individual circumstances. The AMP makes no warranty, express or implied, regarding the Guidelines or Report and specifically excludes any warranties of merchantability and fitness for a particular use or purpose. The AMP shall not be liable for direct, indirect, special, incidental, or consequential damages related to the use of the information contained herein.

The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention/Agency for Toxic Substances and Disease Registry. Use of trade names and commercial sources is for identification only and does not imply endorsement by the Centers for Disease Control and Prevention, the Public Health Service, or the US Department of Health and Human Services.