The genes that change how a drug works — and differ in South Asians

Clopidogrel is a 'switched-off' medicine that your body has to switch on using an enzyme called CYP2C19. Many South Asians carry a gene version that makes a weaker enzyme, so the drug never fully turns on and does less to stop dangerous clots - which is why a doctor may pick a different blood thinner instead.

South Asian signal: The loss-of-function alleles CYP2C19*2 (c.681G>A, rs4244285) and *3 (c.636G>A, rs4986893) are common in South Asians. In a British-South Asian clopidogrel cohort, ~13% were poor metabolisers (two LOF alleles) and ~57% carried at least one LOF allele (intermediate or poor) - higher than European (~2.4% PM) reference (JACC Advances 2023, PMC10550831). CYP2C19*2 allele frequency in South Asians is reported around 30-34% (Ionova et al., Clin Transl Sci 2020).

Prescriber note: In a South Asian ACS/PCI patient, a poor response to clopidogrel is more likely than European data suggest; where CYP2C19 genotype or a validated point-of-care test is available, an intermediate/poor metaboliser result supports switching to prasugrel or ticagrelor rather than escalating clopidogrel. For CYP2C19 intermediate or poor metabolisers with ACS/PCI, CPIC recommends avoiding clopidogrel and using an alternative antiplatelet (e.g. prasugrel or ticagrelor) where not contraindicated, because reduced clopidogrel activation raises the risk of major adverse cardiovascular events.

Why it matters in India: South Asians carry CYP2C19 loss-of-function alleles at roughly 2-5x the European rate, so a larger share of Indian cardiac patients get inadequate platelet inhibition from a first-line, low-cost drug.

Not a dosing tool. Genotype-guided prescribing is a clinician decision; genotyping access in India is limited.

Thiopurines are strong drugs that your body has to clear away safely using a cleanup enzyme called NUDT15. If your gene makes little or none of it, the drug builds up and can crash your blood counts - so people with this gene version need a much smaller dose or a different drug.

South Asian signal: NUDT15 c.415C>T (rs116855232, the *3 defining variant) has a minor allele frequency of ~6.7% in South Asians per gnomAD (vs ~10.5% in East Asians) - far higher than TPMT risk alleles, which are only ~1.4-2.5% in Asians. In Indian paediatric ALL patients, NUDT15 variants were found in 16.7% vs TPMT*3C in only 3.3% (PMC12722899).

Prescriber note: In South Asian patients, NUDT15 c.415C>T is the higher-yield thiopurine safety test than TPMT; a heterozygous (intermediate) or homozygous (poor metaboliser) result mandates a reduced starting dose or alternative agent with close CBC monitoring. CPIC recommends reducing the thiopurine starting dose for NUDT15 intermediate metabolisers and substantially reducing or avoiding thiopurines for poor metabolisers, to prevent life-threatening myelosuppression; in Asians, NUDT15 predicts toxicity better than TPMT.

Why it matters in India: Because NUDT15 deficiency is several times more common than TPMT deficiency in Indians, testing only TPMT (the older Western default) misses most of the South Asian patients at risk of fatal thiopurine toxicity.

Not a dosing tool. Genotype-guided prescribing is a clinician decision; genotyping access in India is limited.

TPMT is a second cleanup enzyme for the same family of drugs. Some people are born making very little of it, so the medicine piles up and harms the bone marrow. Testing it alongside NUDT15 tells the doctor how much drug is safe.

South Asian signal: TPMT no-function alleles are relatively uncommon in South/East Asians (~1.4-2.5%), notably lower than in Europeans and Africans; in Indian paediatric ALL, TPMT*3C heterozygosity was ~3.3% vs NUDT15 variants at 16.7% (PMC12722899). This is why NUDT15 carries more of the toxicity signal in this population.

Prescriber note: Test TPMT together with NUDT15 before starting thiopurines; a normal TPMT does not reassure in a South Asian patient because NUDT15 deficiency is the more common cause of toxicity here. Let the more deficient of the two genes govern dosing. CPIC recommends reduced thiopurine dosing for TPMT intermediate metabolisers and major dose reduction or avoidance for poor metabolisers; TPMT and NUDT15 are interpreted jointly, with the more severe phenotype driving the decision.

Why it matters in India: The historical Western reliance on TPMT alone underestimates thiopurine risk in Indians, where TPMT variants are rare but NUDT15 variants are common - both genes must be read together.

Not a dosing tool. Genotype-guided prescribing is a clinician decision; genotyping access in India is limited.

Warfarin is famously hard to dose - too little and clots form, too much and you bleed. Two genes largely decide how much each person needs: one controls how fast you break the drug down, the other how strongly it works. Knowing them helps a doctor get closer to the right dose sooner.

South Asian signal: CYP2C9 and VKORC1 variants together explain up to ~18% and ~30% of warfarin-dose variance respectively in Europeans, but explain proportionally less in Asian ancestry, so ancestry-specific algorithms matter. CYP2C9 and VKORC1 variant frequencies differ across Indian sub-populations; CPIC therefore provides continental-ancestry-specific dosing guidance rather than a single formula (CPIC 2017 update).

Prescriber note: Where CYP2C9/VKORC1 genotype is available, use an ancestry-appropriate validated algorithm (e.g. the CPIC-endorsed approach) for the initial dose and then titrate to INR; do not apply European-derived dose predictions unmodified to South Asian patients. CPIC recommends using a validated pharmacogenetic dosing algorithm that incorporates CYP2C9 and VKORC1 (and, for some ancestries, CYP4F2 and rs12777823) to estimate a starting warfarin dose targeting INR 2-3, rather than a fixed empirical dose, when genotype is available.

Why it matters in India: Warfarin remains widely used in India (rheumatic valve disease, mechanical valves) where DOAC alternatives may be unaffordable, so getting the initial dose right in a population with different CYP2C9/VKORC1 frequencies directly reduces early bleeding and clotting events.

Not a dosing tool. Genotype-guided prescribing is a clinician decision; genotyping access in India is limited.

Some people are born with red blood cells that can't handle certain 'stressful' medicines. Give those drugs and the red cells burst, causing sudden anaemia. A simple blood test tells the doctor whether a drug like an anti-malaria pill is safe for that person.

South Asian signal: G6PD deficiency is common across India: an all-India systematic review found an overall ~1.9% prevalence (range 0.8-6.3%), but tribal/vulnerable groups run far higher - ~7.7% overall in tribal communities and up to ~27% in some groups. The G6PD Mediterranean (563C>T) variant, a severe (Class II) variant, is the commonest deficient allele in India (~60% of deficient cases), so many Indian deficients are the high-risk severe type.

Prescriber note: Screen for G6PD status before prescribing primaquine/tafenoquine (radical cure of vivax malaria), dapsone, or rasburicase; rasburicase is contraindicated in G6PD deficiency. Because the severe Mediterranean variant predominates in India, do not assume mild African-type deficiency. CPIC advises that in G6PD-deficient individuals, drugs with haemolytic potential (e.g. rasburicase - formally contraindicated; primaquine, dapsone, and other listed oxidants) should be avoided or used only with explicit risk-benefit justification and monitoring, and that higher-risk ancestries be screened before such drugs.

Why it matters in India: India carries a large absolute burden of G6PD deficiency concentrated in malaria-endemic and tribal populations - exactly the patients who need primaquine for vivax malaria radical cure and dapsone for leprosy - and the locally common Mediterranean variant is severe, so unscreened prescribing risks serious haemolysis.

Not a dosing tool. Genotype-guided prescribing is a clinician decision; genotyping access in India is limited.

Carbamazepine is a common seizure and nerve-pain drug, but in people with a particular immune gene it can trigger a rare, severe, sometimes fatal skin reaction where the skin blisters and peels. A one-time genetic test tells the doctor whether to pick a safer drug from the start.

South Asian signal: HLA-B*15:02 is present at intermediate frequency in Indians - roughly 2-4% on average (allele frequency ~2.5%, range 0-6% across North Indian communities), higher than in Europeans (<1%) though below Han Chinese/Thai/Malay levels. Indian studies confirm HLA-B*15:02 remains a significant predictor of carbamazepine SJS/TEN (Frontiers in Genetics 2020).

Prescriber note: Test HLA-B*15:02 before starting carbamazepine (and consider it for oxcarbazepine/phenytoin) in South Asian patients; a positive result is a strong reason to choose an alternative anti-epileptic. The test is one-time and the reaction is preventable. CPIC recommends NOT using carbamazepine (and using caution with oxcarbazepine/phenytoin) in carbamazepine-naive patients who carry HLA-B*15:02, unless benefits clearly outweigh risks; patients of at-risk ancestry should be tested before starting carbamazepine.

Why it matters in India: With hundreds of millions potentially exposed to carbamazepine for epilepsy and neuralgia and an HLA-B*15:02 frequency well above European levels, unscreened use in India exposes a meaningful number of patients to preventable SJS/TEN.

Not a dosing tool. Genotype-guided prescribing is a clinician decision; genotyping access in India is limited.

5-FU and capecitabine are powerful cancer drugs your body normally breaks down quickly with the DPD enzyme. People born with a weak version can't clear the drug, so a standard dose becomes a poisoning - which is why a doctor may lower the dose or choose another plan.

South Asian signal: The classic European risk variant DPYD*2A (c.1905+1G>A, rs3918290) is markedly rarer in Indians (~0.05% allele frequency) than in Europeans (~0.3-0.5%), so a European-only variant panel will miss most at-risk Indian patients; population-specific and broader DPYD variant coverage is needed. South Asian populations may carry different or under-catalogued DPYD variants, and DPD-deficiency phenotype testing remains relevant.

Prescriber note: Where DPYD testing is used before fluoropyrimidine chemotherapy, ensure the panel is not limited to European variants (DPYD*2A is rare in Indians); consider broader genotyping and/or phenotype (DPD activity) assessment, and reduce or avoid dosing per the DPYD activity score. CPIC recommends reducing the fluoropyrimidine starting dose (guided by DPYD activity score) for DPYD intermediate metabolisers and avoiding fluoropyrimidines in poor metabolisers, because reduced DPD activity sharply raises the risk of severe or fatal toxicity.

Why it matters in India: Indians carry the well-known European DPYD risk allele far less often, which paradoxically creates false reassurance - a negative European-panel test does not rule out DPD deficiency in an Indian patient, so panel breadth matters for cancer safety here.

Not a dosing tool. Genotype-guided prescribing is a clinician decision; genotyping access in India is limited.