Enzyme:TPMT

TPMT (thiopurine S-methyltransferase) is a cytosolic phase-II conjugating enzyme whose genetic polymorphism is one of the oldest, best-characterized, and most clinically consequential stories in all of pharmacogenetics. It is encoded by the TPMT gene on chromosome 6p22.3 and catalyzes the S-methylation of thiopurine medicines, using S-adenosyl-L-methionine as the methyl donor. Its clinical relevance is unusually focused: TPMT has no meaningful role in the metabolism of any medicine outside the thiopurine class, but for that small class, the three medicines azathioprine, mercaptopurine (6-MP), and thioguanine (6-TG), it is one of the central determinants of whether a standard dose is safe or lethal.

The history-of-medicine arc begins in 1980 at the Mayo Clinic in Rochester, Minnesota, where Richard Weinshilboum and Sharon Sladek measured TPMT enzyme activity in red blood cells from 298 randomly selected blood donors. The activity values did not form a single bell curve; they fell into a clearly multimodal distribution, with roughly 89% of donors at high activity, roughly 11% at intermediate activity, and about 1 in 300 with very low or absent activity. Family studies confirmed that the trait was inherited as a monogenic codominant Mendelian polymorphism. This was one of the foundational demonstrations of a clinically important pharmacogenetic polymorphism, and it preceded the molecular-genetics era by fifteen years: the bimodal distribution was a population-genetics observation long before anyone had cloned the gene or identified a variant. By the end of the 1980s, Lennard, Van Loon, and Weinshilboum had connected low TPMT activity directly to acute azathioprine toxicity, and in the mid-1990s the molecular variants were finally identified, beginning with the work of Krynetski and colleagues on the loss-of-function point mutation that defines the TPMT\*2 allele.

Tissue distribution

TPMT is a cytosolic enzyme expressed widely across human tissues, including the liver, kidney, and the hematopoietic system. For most of the history of TPMT testing, the clinical surrogate has been red-blood-cell TPMT activity, because erythrocytes are trivially easy to sample and their TPMT activity correlates well with activity in the tissues that matter for thiopurine toxicity, especially the bone-marrow precursors whose suppression produces the clinical danger. Modern testing increasingly uses genotype rather than the enzyme-activity phenotype, but phenotype testing retains one advantage: it detects rare or novel loss-of-function alleles that a fixed-SNP genotyping panel will miss. A recent blood transfusion can confound the red-cell enzyme-activity assay, since transfused cells carry the donor's TPMT activity, which is one practical reason genotype testing has gained ground.

Function and substrate spectrum

TPMT catalyzes the S-methylation of aromatic and heterocyclic sulfhydryl compounds. In human physiology it has no known essential endogenous substrate; its clinical importance is entirely a matter of xenobiotic metabolism, and specifically of the thiopurine medicines.

The pharmacology here has a feature that is worth stating carefully, because it inverts the logic of the cytochrome-P450 prodrug-activation stories told elsewhere in this wiki. Thiopurines are metabolized along three competing routes: TPMT-mediated S-methylation, xanthine-oxidase-mediated oxidation, and the HGPRT-mediated anabolic route that produces the 6-thioguanine nucleotides (TGNs), which are the active cytotoxic and immunosuppressive metabolites. TPMT methylation is a competing inactivation-and-shunting pathway. When TPMT activity is low, less thiopurine is methylated away, more is shunted down the HGPRT route, TGN concentrations climb, and bone-marrow toxicity follows. The clinical danger of TPMT deficiency therefore comes not from failure to activate a prodrug (the CYP2D6-codeine logic) but from loss of a competing detoxification route, which diverts substrate toward the toxic activation pathway.

The clinical substrate set is small enough to list in full: there are exactly three thiopurine medicines in clinical use, and TPMT is relevant to all three and to no other medicine class. Because the list is complete rather than curated, this table is presented in full rather than as a collapsible near-complete excerpt.

Substrate	Therapeutic class	TPMT relevance	Clinical notes
Azathioprine	Immunosuppressant (IBD, transplant, autoimmune disease)	major	Prodrug; non-enzymatically converted to mercaptopurine after absorption, after which the mercaptopurine pharmacology applies. TPMT-deficiency myelosuppression risk is the same as for mercaptopurine. FDA labeling references TPMT (and NUDT15) testing.
Mercaptopurine	Antineoplastic (ALL maintenance); immunosuppressant (IBD)	major	6-MP. The canonical TPMT substrate. TPMT poor metabolizers receiving standard-dose mercaptopurine develop life-threatening myelosuppression within weeks. FDA labeling references TPMT and NUDT15 testing.
Thioguanine	Antineoplastic (AML, ALL); occasionally IBD	major	6-TG. Enters the TGN pool more directly than mercaptopurine, so the proportional contribution of TPMT differs somewhat, but TPMT-deficiency toxicity risk remains substantial. FDA labeling references TPMT testing.

TPMT also methylates downstream thionucleotide intermediates of the thiopurine pathway (for example 6-thioinosine monophosphate), but those are endogenous metabolic intermediates rather than administered medicines, and they do not change the clinical picture: the three medicines above are the entire clinically relevant substrate set.

Phenotype categories

TPMT phenotype is assigned from the two TPMT alleles, and the CPIC convention recognises three metabolizer phenotypes:

• Normal metabolizer (NM): two normal-function alleles (typically \*1/\*1). Roughly 86 to 97% of most populations.
• Intermediate metabolizer (IM): one normal-function and one no-function allele. Roughly 3 to 14% of most populations.
• Poor metabolizer (PM): two no-function alleles. Roughly 1 in 300 (about 0.3%) in European-ancestry populations, with broadly similar order-of-magnitude rarity elsewhere.

There is no ultra-rapid TPMT category in current usage; gain-of-function alleles are not established for this enzyme. Some individuals do sit at the high end of the normal-activity range, and a poorly understood high-activity phenotype has been described, but it does not carry an established star-allele basis or routine clinical action.

The clinical stakes of the phenotype are unusually stark. A TPMT poor metabolizer given a standard thiopurine dose will, with high probability, develop profound and potentially fatal myelosuppression. CPIC's dosing guidance is correspondingly dramatic: normal metabolizers receive the standard dose; intermediate metabolizers start at roughly 30 to 80% of the standard dose; poor metabolizers receive a drastically reduced dose (for azathioprine and mercaptopurine, on the order of 10% of the standard dose, and dosed three times weekly rather than daily) or an alternative non-thiopurine agent.

Major variants

More than 40 TPMT star alleles have been catalogued, but four account for the overwhelming majority of clinically encountered loss-of-function genotypes:

• \*1, the reference normal-function allele.
• \*2 (rs1800462, Ala80Pro), no function. The first TPMT loss-of-function allele characterized at the molecular level; uncommon, and largely a European-ancestry allele.
• \*3A (rs1800460 and rs1142345 in cis, Ala154Thr and Tyr240Cys together), no function. The most common loss-of-function allele in European-ancestry populations, with an allele frequency of roughly 3 to 5%.
• \*3B (rs1800460 alone, Ala154Thr), no function; rare.
• \*3C (rs1142345 alone, Tyr240Cys), no function. The most common loss-of-function allele in East Asian and African-ancestry populations, where \*3A is comparatively uncommon.

The ancestry difference between \*3A and \*3C has a direct testing-design consequence. A genotyping panel built only around the SNPs that define \*3A will detect loss-of-function alleles efficiently in European-ancestry patients but will miss the \*3C allele that carries most of the loss-of-function burden in African-ancestry and East Asian patients. An adequate TPMT panel must cover \*2, \*3A, \*3B, and \*3C at minimum. This is the same health-equity concern that applies to ancestry-skewed allele panels for CYP2C9 and UGT1A1.

Inhibitors

TPMT is not a major drug-interaction enzyme in the way the cytochromes P450 are, but one inhibitor story is genuinely clinically important: the aminosalicylates. Sulfasalazine, mesalamine (mesalazine), olsalazine, and balsalazide all inhibit TPMT. This matters because aminosalicylates and thiopurines are both used to treat inflammatory bowel disease, and a patient may be co-prescribed both. The aminosalicylate-induced reduction in TPMT activity raises TGN concentrations and increases myelosuppression risk, in effect producing a partial phenocopy of TPMT deficiency for the duration of the co-prescription. The interaction is worth anticipating with blood-count monitoring whenever an aminosalicylate is added to, or removed from, a stable thiopurine regimen.

Inducers

TPMT is essentially non-inducible by the classical xenobiotic-induction pathways (the pregnane X receptor and constitutive androstane receptor signaling that drives CYP3A4 induction). There is no clinically actionable TPMT-induction interaction. Thiopurine dosing cannot be rescued by attempting to upregulate TPMT.

Relationship to NUDT15

For most of the history of thiopurine pharmacogenetics, TPMT was understood to be the whole story. Since 2014 it has been clear that it is not. In that year, Yang and colleagues, studying a Korean inflammatory-bowel-disease cohort, identified a common missense variant in the NUDT15 gene as a determinant of thiopurine-induced leukopenia, and the following year a parallel study in children with acute lymphoblastic leukaemia confirmed NUDT15 as an inherited determinant of mercaptopurine intolerance.

NUDT15 loss-of-function produces the same thiopurine-myelosuppression phenotype as TPMT loss-of-function, by a different biochemical mechanism. The clinically decisive fact is that the two genes have nearly opposite ancestry distributions: TPMT loss-of-function is most common in European-ancestry and African-ancestry populations and uncommon in East Asian populations, while NUDT15 loss-of-function is common in East Asian, and to a lesser extent Hispanic and South Asian, populations and uncommon in European-ancestry populations. A patient of East Asian ancestry who develops severe thiopurine myelosuppression while carrying an entirely normal TPMT genotype is very likely a NUDT15 poor or intermediate metabolizer. For this reason, modern thiopurine pharmacogenetics is properly a TPMT-and-NUDT15 story, and the CPIC guideline cited throughout this page covers both genes jointly. See Enzyme:NUDT15 for the NUDT15 account in full.

Clinical implications, summary

• Pre-prescription testing. TPMT testing before thiopurine initiation is one of the most widely adopted pharmacogenetic tests in routine medicine, and the thiopurine story is the textbook example used to teach the clinical value of pre-emptive pharmacogenetics. FDA labeling for azathioprine, mercaptopurine, and thioguanine all reference TPMT status, and the modern recommendation is to test both TPMT and NUDT15 before starting a thiopurine.
• Known PM phenotype: drastically reduce the thiopurine dose (on the order of 10% of standard, dosed three times weekly) or select an alternative non-thiopurine agent. Standard dosing in a poor metabolizer risks fatal myelosuppression.
• Known IM phenotype: start at roughly 30 to 80% of standard dose, with blood-count monitoring and titration.
• The allopurinol interaction: allopurinol inhibits xanthine oxidase, one of the three competing thiopurine-metabolism routes. Removing that route shunts more thiopurine toward TGN formation, so co-prescription of allopurinol with a thiopurine requires substantial thiopurine dose reduction regardless of TPMT genotype, and is a recognised cause of inadvertent thiopurine toxicity when the interaction is overlooked.
• Aminosalicylate co-prescription: anticipate a partial TPMT-inhibition phenocopy and monitor blood counts when an aminosalicylate is started or stopped during thiopurine therapy.

Authoritative resources

Because the clinically relevant TPMT substrate set is limited to the three thiopurine medicines listed in full above, there is no large external substrate table for TPMT comparable to those that exist for the cytochromes P450 (the canonical Flockhart cytochrome P450 interaction table does not cover TPMT). The authoritative maintained resources for TPMT genotype, phenotype, allele-function assignment, and thiopurine dosing are:

• The CPIC Guideline for Thiopurine Dosing Based on TPMT and NUDT15 Genotypes, which carries the consensus allele-function table and the phenotype-specific dosing algorithm, and is updated as new alleles are characterized.
• PharmGKB, the pharmacogenomics knowledge base hosted at Stanford University; the TPMT gene page indexes the allele catalogue and the underlying primary literature, and links the gene-drug pairs to the CPIC guideline. Available at https://www.pharmgkb.org/.
• The Pharmacogene Variation Consortium (PharmVar) maintains the definitive TPMT star-allele nomenclature.

See also

• Enzyme:NUDT15, the parallel thiopurine-myelosuppression gene with a nearly complementary ancestry distribution.
• Phenotype:TPMT poor metabolizer
• Phenotype:TPMT intermediate metabolizer
• Phenotype:TPMT normal metabolizer
• Enzyme:CYP2D6, Enzyme:CYP3A4, Enzyme:CYP2C19, Enzyme:CYP2C9, Enzyme:UGT1A1 (other drug-metabolizing enzymes covered in the wiki)
• Azathioprine, Mercaptopurine, Thioguanine (the entire clinical TPMT substrate set)
• Allopurinol (the xanthine-oxidase-inhibitor interaction)
• Category:Drug-metabolizing enzymes