Pharmacogenomics basics
Pharmacogenomics (PGx) explains why the same drug at the same dose can be ineffective in one patient and toxic in another. Genetic variants in drug-metabolising enzymes (CYP2D6, CYP2C19, DPYD), transporters (OATP1B1/SLCO1B1), and drug targets (VKORC1, HLA alleles) directly alter pharmacokinetics and pharmacodynamics. Key resources: the CPIC guidelines ↗ and the FDA pharmacogenomic biomarker table ↗.
Metaboliser phenotype classification
For CYP enzymes, genetic variation translates into phenotype categories based on the combined activity of two alleles. The PharmGKB database ↗ and PharmVar ↗ curate allele-phenotype assignments.
Poor Metaboliser
Activity score = 0. Two non-functional alleles. Drug accumulates → toxicity for parent drug. Prodrugs (codeine, tamoxifen) are not activated → lack of efficacy. Frequency: CYP2D6 PM ~7% Caucasian, ~2% Asian.
Intermediate Metaboliser
Activity score 0.25–1.25. One reduced-function + one non-functional allele, or two reduced-function alleles. Partial metaboliser. Moderate dose adjustment may be indicated for some drugs (CPIC specific guidance).
Normal Metaboliser
Activity score 1.25–2.25. Two functional alleles (*1/*1 or similar). Standard drug dosing applies. ~65–85% of most populations depending on enzyme.
Ultrarapid Metaboliser
Activity score >2.25. Gene duplication → >2 functional copies. Drug metabolised too rapidly → subtherapeutic levels (for parent drugs). Prodrugs (codeine) produce excess active metabolite → toxicity. Black box warning: codeine in UM breastfeeding mothers.
Key gene-drug pairs with clinical guidance
| Gene | Key drugs affected | PM consequence | CPIC level | Reference |
|---|---|---|---|---|
| CYP2D6 | Codeine, tramadol, amitriptyline, tamoxifen, atomoxetine | Codeine: no analgesia. TCAs: toxicity. Tamoxifen: reduced efficacy. | A | CPIC CYP2D6 ↗ |
| CYP2C19 | Clopidogrel, PPIs (omeprazole), antidepressants (sertraline, escitalopram) | Clopidogrel: reduced active metabolite → cardiovascular events. PPIs: higher acid suppression in PM. | A | CPIC CYP2C19 ↗ |
| DPYD | 5-Fluorouracil (5-FU), capecitabine, tegafur | Severe/life-threatening fluoropyrimidine toxicity (mucositis, myelosuppression, neurotoxicity) in PM/IM | A | CPIC DPYD ↗ |
| TPMT / NUDT15 | Azathioprine, 6-mercaptopurine, thioguanine | Severe myelosuppression in PM. Dose reduction 10-fold required for TPMT PM. | A | CPIC TPMT ↗ |
| VKORC1 + CYP2C9 | Warfarin | VKORC1 variant: lower warfarin dose needed. CYP2C9 PM: reduced S-warfarin clearance → bleeding risk. | A | CPIC warfarin ↗ |
| HLA-B*15:02 | Carbamazepine, phenytoin, oxcarbazepine | Stevens-Johnson syndrome / toxic epidermal necrolysis. Prevalence ~7% in Han Chinese. | A | CPIC HLA-B ↗ |
| SLCO1B1 (OATP1B1) | Simvastatin, atorvastatin, pravastatin, rosuvastatin | *5 allele: reduced hepatic uptake → increased statin plasma levels → myopathy risk | A | CPIC SLCO1B1 ↗ |
All CPIC guidelines are freely available at cpicpgx.org ↗. Drug-label PGx information is searchable via the FDA pharmacogenomic biomarker table ↗ and PharmGKB ↗.
Clinical implementation of PGx testing
Pre-emptive testing
Test once, store results in the EHR, and use when relevant drugs are prescribed. This model is used by health systems implementing eMERGE network PGx programmes ↗ and the PREDICT programme (Vanderbilt) ↗. Cost-effective when multiple PGx variants are tested on a single array.
Reactive testing
Test when a specific drug is prescribed (e.g. DPYD before initiating 5-FU, HLA-B*15:02 before carbamazepine in at-risk populations). EMA now mandates DPYD genotyping before fluoropyrimidine initiation in the EU. FDA labelling for many drugs includes PGx biomarker information in Section 12.5.
PGx in drug labels
The FDA biomarker table ↗ lists ~350+ gene-drug pairs. Categories: Actionable (testing required or recommended) → Informative (testing may be considered) → Research (insufficient evidence for clinical action). Always check DailyMed for the most current labelling language.
For paediatric PGx considerations, see our guide on paediatric pharmacology ↗ (CYP enzyme maturation, codeine URM risk). For drug metabolism context, see the PK basics guide ↗.
Frequently asked questions
What is pharmacogenomics and how does it differ from pharmacogenetics?
Pharmacogenetics (PGx) originally referred to the study of how single genes affect drug response. Pharmacogenomics is the broader term encompassing whole-genome approaches — examining how genetic variation across all genes influences drug pharmacokinetics, pharmacodynamics, and safety. In clinical practice, the terms are often used interchangeably. The FDA and CPIC (Clinical Pharmacogenomics Implementation Consortium) use pharmacogenomics broadly to cover both.
What is a CYP2D6 poor metaboliser and why does it matter?
CYP2D6 poor metabolisers (PM) have two non-functional CYP2D6 alleles (e.g. *3, *4, *5, *6) and cannot metabolise CYP2D6 substrates normally. This affects ~7% of Caucasians. Clinical consequences: (1) codeine cannot be converted to morphine → no analgesia; (2) tricyclic antidepressants (nortriptyline, amitriptyline) accumulate → toxicity; (3) tamoxifen cannot be activated to endoxifen → reduced breast cancer efficacy. CYP2D6 also has ultrarapid metabolisers (UM) — 2 or more functional copies — where codeine may cause morphine toxicity. FDA added a black box warning for codeine in CYP2D6 UM.
What is CPIC and how are its guidelines used clinically?
CPIC (Clinical Pharmacogenomics Implementation Consortium) is an international consortium (NIH-funded) that develops freely available, peer-reviewed guidelines for translating pharmacogenomic test results into actionable prescribing decisions. CPIC guidelines are evidence-based, gene/drug-pair specific, and distinguish between strong (Level A) and weaker (Level B/C) recommendations. They are published in Clinical Pharmacology & Therapeutics and freely available at cpicpgx.org. CPIC guidelines are integrated into electronic health record (EHR) CDS tools and national formulary systems.
What is the difference between a variant allele and a star (*) allele?
Star (*) alleles are a standardised nomenclature system for pharmacogenes where *1 is the reference (wild-type, fully functional) allele, and other numbered alleles (e.g. *2, *3, *4) indicate specific variant haplotypes with known or predicted functional consequences. A single star allele may contain multiple SNPs. Activity scores (0 = no function, 0.5 = reduced function, 1 = normal function) are assigned per allele, and the sum of two allele activity scores gives the patient's phenotype (PM, IM, NM, RM, UM). This system is maintained by PharmVar (pharmacogenomics variant resource).
Which pharmacogenomic tests are FDA-approved or FDA-cleared?
FDA maintains a Table of Pharmacogenomic Biomarkers in Drug Labeling listing gene-drug pairs where genetic testing is recommended or required. Examples include: TPMT/NUDT15 for thiopurines (azathioprine, 6-mercaptopurine) — PMs at risk for severe myelosuppression; DPYD for fluoropyrimidines (5-FU, capecitabine) — PMs at risk for life-threatening toxicity; HLA-B*15:02 for carbamazepine — risk of Stevens-Johnson syndrome in Southeast Asian patients; G6PD for rasburicase and primaquine — haemolytic anaemia risk. Most of these are actionable under CPIC Level A guidance.