NUDT15 Genotype

If a doctor is about to start you on azathioprine, mercaptopurine, or thioguanine, this one genetic result changes the safe starting dose. In people who carry two broken copies of this gene, standard dosing causes severe bone marrow suppression in the large majority of cases.

The gene is NUDT15 (nudix hydrolase 15). It is a drug-response gene. In the absence of thiopurines, known variants are not linked to disease or health traits.

What the Gene Actually Does

Azathioprine and mercaptopurine are called thiopurines. They work by getting converted inside your cells into molecules that resemble the building blocks of DNA. Those look-alike molecules get slotted into the DNA of fast-dividing cells, which stops those cells from multiplying. That is the point in leukemia and in autoimmune disease.

NUDT15 is the brake. The enzyme it makes chops the most damaging of those look-alike molecules down into a harmless form before they can be built into DNA. Lose the brake and the damaging form piles up, gets loaded into DNA at high levels, and your bone marrow stops producing blood cells.

So the enzyme is a cleanup crew, not a target. In someone who never takes a thiopurine, a broken copy causes no known illness.

Reading Your Result

You have two copies of the gene, one from each parent. The lab reports them as a pair, like 1/1 or 1/3, and then translates that pair into a metabolizer category. The *1 allele is the normal, fully working version.

• *Normal metabolizer (1/1):* both copies work. Standard thiopurine dosing is usually used, but blood counts still need monitoring.
• *Intermediate metabolizer (one working copy, e.g. 1/3, 1/2, 1/4, 1/6):* enzyme activity falls below normal, with the size of the drop depending on which variant you carry. Standard dosing raises the risk of low blood counts.
• *Poor metabolizer (two broken copies, e.g. 3/3, 2/3, 3/6):* little to no enzyme activity is left. Standard dosing is severely toxic.

The variants are not all equally broken. The 3 allele (c.415C>T, also written R139C) wipes out activity almost entirely, and it does so mainly by destabilizing the protein. In lab studies, the 4, 5, and 6 alleles left only a small fraction of activity intact. That gradation is why two people labeled intermediate can tolerate quite different doses.

How Much Risk a Variant Carries

The strongest signal here is in poor metabolizers. In people with two broken copies, early leukopenia, a dangerous drop in white blood cells, is expected on standard doses. Pooled estimates put the rate near the top of the range, with wide uncertainty because so few people have been dosed at full strength before anyone knew their genotype. This is not a probabilistic nudge. It is an expected event.

For carriers of a single variant copy the picture is still serious but less absolute. Across studies in inflammatory bowel disease and leukemia, carrying a variant raises the odds of thiopurine-induced low white counts several-fold to more than twenty-fold compared to people with two normal copies. A pooled analysis of sixteen studies put the diagnostic odds ratio for the common *3 variant at 8.44. One Chinese autoimmune cohort found odds of leukopenia around 21.7.

What this means for you: an intermediate result is not a reason to abandon the drug. It is a reason to start low, and to watch blood counts closely. In a Chinese autoimmune program, screening before treatment and changing the plan for carriers cut leukopenia rates substantially compared with unscreened historical care.

Hair Loss and Gut Intolerance

Marrow suppression gets the attention, but it is not the only consequence. In a large Japanese inflammatory bowel disease cohort, the *3 variant was strongly tied to complete, severe hair loss, with substantially higher odds than in people without it. Digestive intolerance was also more common, though the effect there was modest.

Hair follicle cells divide fast, same as marrow cells. Both are casualties of the same unbraked chemistry.

Why This Is Different From TPMT

TPMT (thiopurine S-methyltransferase) is the older, better-known gene in thiopurine safety. A normal TPMT result should not be treated as clearance to dose. The two genes act at different points in the same drug pathway and fail independently.

In East Asian and South Asian populations, TPMT variants are rare and NUDT15 accounts for much of the thiopurine blood toxicity. In Hispanic and Latino populations, NUDT15 variants are also more common than in Europeans, and some are missed by basic panels. Even in European cohorts, where NUDT15 variants are uncommon, they explain an additional share of severe early blood toxicity beyond what TPMT catches. Taken together the two genes explain a substantial minority of severe thiopurine blood toxicity in Europeans, which also tells you the rest has causes neither test sees.

There is a related trap in drug monitoring. Labs often track a thiopurine metabolite in red blood cells called 6-TGN to judge whether a dose is too high. In people with normal NUDT15, higher 6-TGN does track with low white counts. In variant carriers it does not, because this enzyme acts downstream of where those metabolites are counted. A reassuring 6-TGN level does not rule out NUDT15-driven toxicity.

Ancestry Changes the Yield

This is where the arithmetic gets stark. In South Asian populations, one UK inflammatory bowel disease analysis estimated that genotyping roughly a couple of dozen people would prevent one case of severe marrow suppression. In Europeans, the same UK analysis estimated that many hundreds would need testing to prevent one severe case.

Risk variants are common in East Asian cohorts, often in the range of one in five to one in three across reported cohorts. They also show up in South Asian, Central Asian, and Amerindian-linked ancestry groups. They are rare in African ancestry cohorts, but rare is not the same as impossible. Ancestry shifts the odds; it does not make the test useless in any group, because a European carrier with two broken copies faces the same catastrophic toxicity as anyone else.

What Sequencing Catches

Some basic panels test one spot: the 3 variant at codon 139. That misses several alleles that matter. The 4 allele is enriched in Hispanic and Latino populations. The 6 allele is a small insertion that many genotyping platforms cannot see at all. A published case describes a child with the 1/*6 genotype who developed severe marrow suppression that a codon-139-only assay would have called normal.

Sequencing also resolves a subtler problem. If a person carries both the *3 change and a second variant, those two changes can sit on the same copy of the gene or on opposite copies. Same copy means one working copy left and an intermediate result. Opposite copies means no working copies and a poor metabolizer result. Standard short-read methods sometimes cannot tell these apart, and the clinical difference between them is the difference between a reduced dose and no thiopurine at all.

One caveat in the other direction. A pediatric institution comparing sequencing against genotyping in 320 people found the simpler method gave accurate phenotype calls and the same clinical recommendations in their population. Sequencing earns its keep in ancestrally diverse or admixed settings and when a result looks inconsistent with what the person is experiencing, not in every case.

When a Result Can Mislead

A negative result only rules out the variants the assay looked for. A single-variant screen can miss 4, 6, *9, and rare alleles. Sequencing lowers that risk, but it can still find a variant whose drug effect is not yet known.

The sample should reflect your inherited DNA. Active leukemia with many circulating blast cells, a recent blood transfusion, or a past donor stem cell or bone marrow transplant can confuse blood-based genotyping. In those settings, confirmation from another tissue or another method may be needed.

Direct-to-consumer raw DNA is not the same as a clinical pharmacogenetic result. It may omit relevant alleles, use older allele translation, or fail to tell whether two changes are on the same gene copy or opposite copies.

A One-Time Result You Keep for Life

Your genotype does not change. Test once, get the result into your medical record, and never repeat it. There is no trend to track here, and no reason to retest unless the lab call itself is in doubt, for instance a weak or ambiguous signal that a different method should confirm.

What does need ongoing tracking is your blood. A complete blood count, especially the absolute neutrophil count, still needs regular checking through thiopurine treatment. This is the single most common misreading of a normal result. A 1/1 genotype removes one known cause of marrow suppression, not the possibility of it. Much of the severe blood toxicity in European patients has causes outside both NUDT15 and TPMT, and other genes including AOX1 and DHFR have been implicated in people negative for both.

Timing is the part worth getting right. This result is useful before the first dose and much less useful after. Get it during the conversation about starting the drug, not after your counts have already dropped.

What an Unexpected Result Should Make You Do

If you come back an intermediate metabolizer, the next steps are concrete. Order TPMT genotyping alongside this if you have not already, since the two genes fail independently and carrying defects in both compounds the danger. Bring the result to whoever is prescribing, a gastroenterologist for inflammatory bowel disease, a rheumatologist for autoimmune disease, an oncologist for leukemia, and expect a reduced starting dose with tighter blood count monitoring in the first weeks.

If you come back a poor metabolizer, the conversation changes. Standard thiopurine dosing is not an option. For Crohn's disease, colitis, and other non-cancer uses, guidelines usually favor a different drug. In leukemia protocols, thiopurines may still be used, but at a sharply reduced dose and with close blood count monitoring. That decision needs a specialist, because the alternatives differ by whether you are being treated for leukemia, Crohn's disease, or an autoimmune condition.

If the result looks odd, a weak signal, an unexpected variant of unclear meaning, or a genotype that does not match how you actually responded to the drug, confirm it by a different method. Ambiguous signals on high-throughput platforms are a known problem, and direct sequencing often settles them.

Genetic counseling is not mandatory for a straightforward drug-response result, but it helps when the call is ambiguous, when family members need to understand their own chance of carrying it, or when the result is part of a broader genome or exome report.

One thing this test will not do: tell you anything about your health if you never take one of these drugs. It predicts drug response, not disease. Screening people with no plan to start a thiopurine mostly produces results nobody acts on. The exception is worth naming, though. If your genotype is banked in your record now, it is there the day someone writes the prescription, which is the one moment it matters.