Pharmacogenomic Testing

Pharmacogenomic testing identifies inherited variants that alter how a person absorbs, metabolises, transports or responds to a drug, so that the choice or dose can be matched to the patient.

Variants in drug metabolising enzymes (cytochrome P450 genes CYP2D6, CYP2C19, CYP2C9, CYP3A5, CYP2B6, and the enzymes TPMT, NUDT15, DPYD, UGT1A1 and NAT2) change drug exposure; variants in transporters (SLCO1B1) change uptake; and variants in immune genes (HLA-B*57:01, HLA-B*15:02, HLA-B*58:01) predict severe hypersensitivity reactions.

Results are given as a genotype (for example CYP2C19 *1/*2) and translated to a predicted phenotype: poor, intermediate, normal, rapid or ultra-rapid metaboliser (for enzymes) or presence or absence of an at risk allele (for HLA genes).

Methods are targeted genotyping panels (PCR-based or microarray), copy number analysis (CYP2D6), and sequencing; the result is permanent and can be reused for many drugs through a patient's life.

The frequency of actionable variants is high: over 90% of people carry at least one clinically relevant variant. Poor metaboliser phenotypes occur in about 5 to 10% of Europeans for CYP2D6, 2 to 5% of Europeans and 15 to 20% of East Asians for CYP2C19. Ultra-rapid CYP2D6 occurs in 1 to 2% of Northern Europeans but 10 to 30% in parts of North Africa and East Africa.

Allele frequencies vary by ancestry:

• HLA-B*57:01 is present in 5 to 8% of Europeans and is rarer in Africans (1 to 3%).
• HLA-B*15:02 in up to 10 to 15% of Han Chinese and South-East Asian people.
• HLA-B*58:01 in 6 to 8% of Han Chinese.
• DPYD variants associated with reduced function in 3 to 8% of Europeans.
• TPMT variants in about 10% heterozygous and 0.3% homozygous.
• CYP2B6 slow metabolism (*6/*6) in 20 to 25% of Africans, with implications for efavirenz dosing.

Indications. Before starting drugs with a validated gene drug pair and serious toxicity or failure: abacavir (HLA-B*57:01), carbamazepine and phenytoin (HLA-B*15:02 in at risk ancestry; CYP2C9 for phenytoin), allopurinol (HLA-B*58:01), fluoropyrimidines such as capecitabine and fluorouracil (DPYD), thiopurines such as azathioprine and mercaptopurine (TPMT and NUDT15),

irinotecan and atazanavir (UGT1A1), clopidogrel (CYP2C19), warfarin (CYP2C9, VKORC1), simvastatin (SLCO1B1), codeine and tramadol (CYP2D6), tamoxifen (CYP2D6, with limited evidence), selective serotonin reuptake inhibitors and tricyclic antidepressants (CYP2C19, CYP2D6), tacrolimus (CYP3A5), efavirenz (CYP2B6), isoniazid (NAT2), and volatile anaesthetics or succinylcholine (RYR1, BCHE).

After an unexplained adverse reaction or non-response to a drug. Pre-emptive panel testing of patients likely to use many drugs. Not for drugs lacking validated gene drug evidence, or as a substitute for therapeutic drug monitoring.

Specimen. Peripheral blood 2 to 5 mL in EDTA, saliva, or buccal swab. The sample can be taken at any time and is not affected by current medication, transfusion (for most assays except in recent allogeneic stem cell transplant, where donor DNA may be present) or acute illness. For tumour pharmacogenomics (somatic variants), tumour tissue is used instead.

Turnaround is 1 to 7 days for targeted tests; point-of-care HLA-B*57:01 and CYP2C19 assays give results within an hour in some centres. Record the drug planned, current drugs and ancestry.

Reference values. Normal function genotype (for example CYP2C19 *1/*1, CYP2D6 *1/*1, TPMT *1/*1, DPYD wild type), and absence of at risk HLA alleles. Normal metabolisers respond to usual doses.

Interpretation.

• CYP2D6: ultra-rapid metabolisers (gene duplication) convert codeine and tramadol to active morphine and O-desmethyltramadol rapidly, with a risk of respiratory depression and death, including in breastfed infants of mothers who are ultra-rapid metabolisers.
• Poor metabolisers get no analgesia from codeine or tramadol, and have higher levels of tricyclic antidepressants, some antipsychotics and beta blockers (metoprolol).
• For tamoxifen, poor metabolism may reduce the formation of endoxifen, but evidence for changing treatment is limited.
• CYP2C19: poor metabolisers (*2/*2, *2/*3) have little activation of clopidogrel and higher rates of stent thrombosis and recurrent events, so prasugrel or ticagrelor is chosen when available.
• Poor metabolisers need lower doses of some antidepressants (citalopram, sertraline), voriconazole (high levels) and proton pump inhibitors (higher exposure). Ultra-rapid metabolisers (*17) have lower levels of citalopram and voriconazole, and may need a different agent.
• CYP2C9 and VKORC1: variants (*2, *3, and VKORC1 -1639G>A) lower the warfarin dose requirement, with genotype based dosing algorithms reducing time to stable INR. CYP2C9 poor metabolisers have higher phenytoin, celecoxib and flurbiprofen levels.
• TPMT and NUDT15: homozygous deficiency leads to life threatening myelosuppression with usual thiopurine doses, requiring dose reduction to about 10% (and 3 times weekly dosing) or an alternative; heterozygotes need a 30 to 50% dose reduction. NUDT15 variants are more frequent in East Asian people and cause leucopenia even with normal TPMT.
• DPYD: reduced function alleles (c.1905+1G>A, c.1679T>G, c.2846A>T, c.1236G>A) cause fluoropyrimidine toxicity (severe mucositis, neutropenia, diarrhoea, neurotoxicity, death). Complete deficiency means avoiding the drug; partial deficiency means a 25 to 50% dose reduction with titration.
• HLA-B*57:01: positive means abacavir is contraindicated because of a hypersensitivity reaction (fever, rash, gastrointestinal and respiratory symptoms) in about 50% of carriers, whereas a negative test lowers the risk to under 1%.
• HLA-B*15:02: strongly associated with Stevens-Johnson syndrome and toxic epidermal necrolysis from carbamazepine, oxcarbazepine and phenytoin in patients of Asian ancestry. HLA-A*31:01 is associated with carbamazepine hypersensitivity in Europeans and Japanese.
• HLA-B*58:01: associated with severe cutaneous reactions to allopurinol, so an alternative xanthine oxidase inhibitor is used in carriers.
• SLCO1B1 (*5, c.521T>C): reduced hepatic uptake of simvastatin, with an increased risk of myopathy at high doses; a lower dose or another statin is used.
• UGT1A1 *28 (7 TA repeats): reduced glucuronidation, causing Gilbert syndrome, irinotecan toxicity (neutropenia, diarrhoea) and hyperbilirubinaemia with atazanavir.
• CYP3A5: expressers (*1 allele, 70 to 80% of Africans and 15% of Europeans) need higher tacrolimus doses to reach target trough concentrations.
• CYP2B6 *6/*6: slow metabolism of efavirenz, leading to high levels and neuropsychiatric toxicity at usual doses; lower doses are used.
• NAT2 slow acetylators: higher risk of isoniazid neuropathy and hepatotoxicity (about half of Europeans and Africans, 90% in some Middle Eastern groups).
• G6PD deficiency (X-linked): haemolysis with primaquine, dapsone, rasburicase, nitrofurantoin and others; usually diagnosed by enzyme assay with genotyping when assay results are unreliable (women with intermediate activity, recent haemolysis, transfusion).
• RYR1 and CACNA1S variants: susceptibility to malignant hyperthermia with volatile anaesthetics and succinylcholine. BCHE variants: prolonged apnoea after succinylcholine or mivacurium.

Critical values and findings needing immediate action.

• HLA-B*57:01 positive: abacavir must not be given or restarted, and the allergy is recorded permanently.
• HLA-B*15:02 or HLA-B*58:01 positive: avoid carbamazepine, oxcarbazepine, phenytoin or allopurinol, respectively; choose an alternative.
• DPYD complete deficiency, or TPMT or NUDT15 homozygous deficiency: avoid or give sharply reduced doses of the fluoropyrimidine or thiopurine.
• Patient already on a drug with severe toxicity (neutropenia on thiopurine, severe mucositis on fluorouracil) and a deficiency genotype: stop the drug, give supportive care and consider rescue (uridine triacetate for fluoropyrimidine overdose within 96 hours).
• CYP2D6 ultra-rapid metaboliser needing opioid analgesia, or nursing mothers who are ultra-rapid: avoid codeine and tramadol.
• CYP2C19 poor metaboliser after percutaneous coronary intervention on clopidogrel: change to prasugrel or ticagrelor.
• RYR1 variant: avoid triggering anaesthetic agents, and alert the anaesthetic team.

Interferences and pitfalls.

• Panels test only selected alleles. Rare or population specific variants (for example African specific CYP2D6*17 and CYP2D6*29, CYP2C9*5, *6, *8 and *11, and DPYD variants) are missed by panels built on European allele frequencies, so a normal genotype may not equal a normal phenotype in under represented ancestries.
• CYP2D6 is the hardest gene to test, because of gene duplications, deletions, hybrid alleles and pseudogenes; copy number analysis is required, and results should state the method.
• Phenoconversion: drugs that inhibit enzymes (fluoxetine, paroxetine, bupropion for CYP2D6; omeprazole, fluconazole for CYP2C19) or induce them (rifampicin, carbamazepine) convert a normal metaboliser genotype into a poor or rapid phenotype. Inflammation, liver disease, kidney disease, age and pregnancy also alter enzyme activity.
• Genotype explains only part of the variation in drug response; age, weight, renal and hepatic function, interacting drugs, adherence and other genes contribute, so dose adjustment is not a substitute for clinical monitoring, drug levels and INR.
• Results are lifelong and must be stored in the health record in a form that is visible at prescribing; otherwise the value of pre-emptive testing is lost.
• Many direct-to-consumer reports include gene drug pairs without clinical validity or without actionable content; they should not alter treatment without clinical review.
• Tests for psychiatric drug selection (combinatorial panels) have limited evidence for outcome benefit; testing for CYP2C19 and CYP2D6 is useful mainly for specific drugs when side effects or non-response occur.
• Cost and availability limit use, particularly in low- and middle income settings, where gene frequencies for key variants (CYP2B6, CYP3A5, NAT2, G6PD, HLA) are important but locally validated tests are scarce.
• Incidental findings can arise (a sequencing based panel can disclose a pathogenic variant in an unrelated gene).
• Ethical and privacy questions apply to genetic data in families and employers; consent is needed.

Special groups.

• Children: codeine is avoided under 12 years regardless of genotype; TPMT and NUDT15 testing is essential before mercaptopurine in leukaemia.
• Pregnant and breastfeeding women: CYP2D6 ultra-rapid status makes codeine hazardous to the breastfed infant.
• Patients of Asian ancestry: HLA-B*15:02 and HLA-B*58:01 testing before carbamazepine and allopurinol; NUDT15 testing before thiopurines; CYP2C19 poor metaboliser status common.
• Patients of African ancestry: CYP2B6 (efavirenz), CYP3A5 (tacrolimus), G6PD (malaria treatment, primaquine and tafenoquine), NAT2 (isoniazid), and APOL1 (not a drug gene but relevant to kidney disease) are key. CYP2D6 *17 and *29 reduce activity in a pattern that panels may not capture.
• Transplant recipients: donor derived DNA after stem cell transplant can give misleading genotypes if blood is used; use buccal or skin samples.
• Cancer patients: DPYD, UGT1A1 and TPMT testing before chemotherapy, with tumour (somatic) variants used for targeted therapy.
• Regions: G6PD deficiency and HIV, tuberculosis and malaria drugs make pharmacogenomic testing most relevant to public health in sub-Saharan Africa and Asia; point-of-care assays for G6PD and HLA-B*57:01 are expanding.

Follow up tests. Therapeutic drug monitoring (warfarin INR, tacrolimus trough, voriconazole level, antiepileptic levels), full blood count and liver function after thiopurine or fluoropyrimidine starts, G6PD enzyme assay, confirmatory sequencing of rare variants. Also review interacting drugs, repeat testing by a different method when phenotype and genotype disagree, documentation of results in the record, and genetic counselling for implications to relatives.