Pharmacogenomics is the study of how inherited genetic variation influences drug response, encompassing differences in drug metabolism, transport, receptor sensitivity, and immune-mediated reactions. Genetic variation underlies a substantial proportion of the interindividual variability in drug efficacy and toxicity that cannot be explained by body weight, renal function, or drug interactions alone. Understanding the vocabulary and conceptual framework of pharmacogenomics is a prerequisite for applying genetic testing results to prescribing decisions.
The most common form of genetic variation relevant to pharmacogenomics is the single nucleotide polymorphism (SNP), a variation at a single base position in the genome that occurs in at least 1% of the population. SNPs can alter protein structure and function when they fall within coding regions, alter gene expression when in regulatory regions, or have no functional consequence when synonymous. Copy number variation (CNV) -- differences in the number of copies of a gene -- is particularly important for the CYP2D6 (cytochrome P450 2D6) gene, where gene duplication results in ultrarapid metabolism. Insertions and deletions (indels) in splice sites or regulatory regions can produce null alleles, entirely eliminating enzyme activity. The aggregate effect of an individual's complement of alleles at a given gene locus defines their genotype, which in turn predicts their functional metabolizer phenotype for that enzyme or transporter.1
Metabolizer Phenotype Classification. For drug-metabolizing enzymes, four phenotypic categories are defined. Poor metabolizers (PMs) carry two non-functional alleles and have essentially absent enzyme activity; they show markedly elevated plasma concentrations of drugs that are substrates for that enzyme. Intermediate metabolizers (IMs) carry one non-functional allele and one reduced-function allele (or two reduced-function alleles), producing below-normal enzyme activity. Normal metabolizers (NMs), also called extensive metabolizers, carry two functional alleles and metabolize drugs at the expected population rate. Ultrarapid metabolizers (UMs) carry gene duplications or multiplications producing greater-than-normal enzyme activity; they metabolize substrate drugs so rapidly that standard doses produce subtherapeutic plasma concentrations. These phenotype categories are assigned by genotyping the relevant alleles and applying a gene-specific translation table developed by the Clinical Pharmacogenomics Implementation Consortium (CPIC).12
Population Frequency and Clinical Relevance. The frequency of pharmacogenomically relevant alleles varies substantially across ethnic populations, which has important implications for screening and implementation strategies. CYP2D6 poor metabolizer frequency is approximately 5-10% in European populations, 1-3% in East Asian populations, and 1-3% in African populations, while CYP2C19 (cytochrome P450 2C19) poor metabolizer frequency is approximately 2-5% in Europeans but 13-23% in East Asian populations. This population stratification means that uniform testing recommendations appropriate for one population may not translate directly to another, and that population-specific allele frequencies must be considered when interpreting prior probability of a given phenotype in the absence of genotyping data. The HLA (human leukocyte antigen) B*57:01 allele, associated with abacavir hypersensitivity, is present in approximately 5-8% of white European populations but considerably less common in Asian and African populations.2,3
The Clinical Pharmacogenomics Implementation Consortium (CPIC) publishes peer-reviewed, evidence-graded guidelines for gene-drug pairs where genotype information should change prescribing. High-evidence actionable pairs include: CYP2D6/codeine, CYP2D6/tamoxifen, CYP2C19/clopidogrel, CYP2C19/tricyclic antidepressants, HLA-B*57:01/abacavir, HLA-B*15:02/carbamazepine, TPMT/thiopurines, G6PD/rasburicase and primaquine. CPIC guidelines are freely accessible at cpicpgx.org and provide specific dosing recommendations for each phenotype category.
The cytochrome P450 enzyme family is the primary site of pharmacogenomically relevant variation affecting drug metabolism. Four isoforms -- CYP2D6 (cytochrome P450 2D6), CYP2C19 (cytochrome P450 2C19), CYP2C9 (cytochrome P450 2C9), and CYP3A5 (cytochrome P450 3A5) -- account for the vast majority of clinically important pharmacogenomic effects on drug metabolism, and each has well-characterized variant alleles with established genotype-to-phenotype translations and CPIC (Clinical Pharmacogenomics Implementation Consortium)-level evidence for clinical action.
CYP2D6. CYP2D6 is the most pharmacogenomically complex drug-metabolizing enzyme, with over 100 known variant alleles exhibiting activity spanning from zero to ultrarapid. The consequences of CYP2D6 poor metabolizer status are prototypically illustrated by codeine: codeine is a prodrug that requires CYP2D6-mediated O-demethylation to morphine for analgesic efficacy; in poor metabolizers, this conversion does not occur and codeine is essentially ineffective, while in ultrarapid metabolizers, rapid and extensive conversion to morphine produces toxic plasma concentrations -- a pharmacogenomic safety signal that has resulted in regulatory restrictions on codeine use in children and breastfeeding mothers. Tamoxifen, used in the adjuvant treatment of estrogen receptor (ER)-positive breast cancer, requires CYP2D6-mediated conversion to its active metabolite endoxifen for full therapeutic efficacy; CYP2D6 poor metabolizers and patients taking CYP2D6 inhibitors (e.g., paroxetine, fluoxetine) have substantially lower endoxifen levels and potentially reduced clinical benefit, motivating genotype-guided selection of endocrine therapy.24
CYP2C19. CYP2C19 polymorphism has its highest-profile clinical impact on clopidogrel activation and proton pump inhibitor (PPI) metabolism. Clopidogrel is a thienopyridine prodrug requiring sequential CYP2C19-mediated oxidation to its active thiol metabolite, which irreversibly inhibits the platelet adenosine diphosphate (ADP) receptor P2Y12 (platelet 12 ADP receptor) to prevent platelet aggregation. CYP2C19 poor metabolizers generate substantially less active metabolite, showing significantly reduced platelet inhibition and, in the context of coronary stenting, increased rates of major adverse cardiovascular events (MACE). The FDA (US Food and Drug Administration) added a boxed warning to clopidogrel acknowledging reduced efficacy in poor metabolizers. CPIC recommends prasugrel or ticagrelor as alternatives in CYP2C19 poor or intermediate metabolizers undergoing percutaneous coronary intervention (PCI). Conversely, CYP2C19 ultrarapid metabolizers convert clopidogrel to its active metabolite at an accelerated rate, which may increase bleeding risk at standard doses.24
CYP2C9 and CYP3A5. CYP2C9 is the principal enzyme metabolizing the S-enantiomer of warfarin, the pharmacologically active species. CYP2C9 poor metabolizers (carrying *2 and *3 variant alleles, which are most common in European populations) have markedly reduced warfarin clearance and require substantially lower maintenance doses to achieve target INR (international normalized ratio) -- as little as 20-30% of the dose required by normal metabolizers. The VKORC1 (vitamin K epoxide reductase complex subunit 1) gene, encoding warfarin's molecular target, also has clinically important polymorphisms affecting sensitivity; combined VKORC1 and CYP2C9 genotyping, along with CYP4F2 (cytochrome P450 4F2) variants, is incorporated into FDA-approved pharmacogenomically informed warfarin dosing algorithms. CYP3A5 is highly polymorphic, with the *1 allele (expressing functional enzyme) more common in individuals of African ancestry (~50%) than European ancestry (~10%); CYP3A5 expressers require higher tacrolimus doses to reach target trough concentrations after transplantation, and CPIC guidance recommends dose adjustment based on CYP3A5 genotype.3,4
Potent CYP2D6 inhibitors (paroxetine, fluoxetine, bupropion, quinidine) convert a normal metabolizer into a functional poor metabolizer -- a phenomenon called phenocopying. This has direct clinical consequences: a normal metabolizer on paroxetine who is prescribed codeine receives no analgesic benefit (cannot convert codeine to morphine); a normal metabolizer on paroxetine taking tamoxifen has reduced endoxifen production. Genotyping alone does not capture phenocopying -- drug interaction assessment must accompany genotype interpretation in every case.
Beyond CYP (cytochrome P450) enzymes, several other pharmacogenomic loci carry equal or greater clinical urgency. Thiopurine methyltransferase (TPMT) and dihydropyrimidine dehydrogenase (DPYD) deficiency represent scenarios where standard drug doses can cause life-threatening toxicity in genetically susceptible patients. HLA (human leukocyte antigen) allele associations with severe immune-mediated drug reactions have transformed management of specific drug initiations. Transporter polymorphisms increasingly influence drug exposure and tissue distribution.
TPMT and Thiopurine Toxicity. The thiopurine drugs -- azathioprine, 6-mercaptopurine (6-MP), and thioguanine -- are pro-drugs metabolized by competing pathways. Thiopurine methyltransferase (TPMT) inactivates thiopurines by S-methylation; when TPMT activity is absent or severely reduced (approximately 0.3% of the population are TPMT poor metabolizers, and 10% are intermediate metabolizers), substrate is shunted into the pathway generating cytotoxic thioguanine nucleotides (TGNs), which accumulate and cause severe myelosuppression. Standard doses of azathioprine in a TPMT poor metabolizer can produce life-threatening bone marrow failure within weeks. CPIC (Clinical Pharmacogenomics Implementation Consortium) recommends TPMT genotyping or phenotyping (by measuring red blood cell TPMT activity) before initiating thiopurine therapy, with substantial dose reduction for intermediate metabolizers and alternative drug selection for poor metabolizers. An additional complexity is introduced by the NUDT15 (nudix hydrolase 15) gene, particularly important in Asian populations: NUDT15 poor metabolizers also accumulate TGNs and are at high thiopurine toxicity risk independent of TPMT status.5
DPYD and Fluoropyrimidine Toxicity. Dihydropyrimidine dehydrogenase (DPYD) is the rate-limiting enzyme in the catabolism of fluoropyrimidines -- 5-fluorouracil (5-FU) and its oral prodrug capecitabine. Approximately 3-5% of the population carry DPYD variant alleles associated with reduced or absent enzyme activity (most commonly DPYD*2A, c.1679T>G, and c.2846A>T), and in these patients, standard fluoropyrimidine doses accumulate to toxic concentrations, causing severe or life-threatening mucositis, diarrhea, myelosuppression, and neurotoxicity. Fluoropyrimidine toxicity in DPYD-deficient patients carries mortality risk. The European Society for Medical Oncology (ESMO) and several national regulatory authorities mandate DPYD screening before fluoropyrimidine initiation, and CPIC recommends pre-emptive dose reduction of 25-50% for intermediate metabolizers and avoidance of fluoropyrimidines in poor metabolizers unless no alternative exists.3
HLA Allele Associations and Severe Cutaneous Reactions. Certain HLA alleles predispose to severe immune-mediated drug reactions by altering the presentation of drug or drug-protein hapten complexes to T lymphocytes. The associations are both strong and drug-specific. HLA (human leukocyte antigen) B*57:01 is associated with abacavir hypersensitivity syndrome (AHS), a potentially fatal multi-organ inflammatory reaction occurring in approximately 5-8% of untested patients; prospective HLA B*57:01 screening before abacavir initiation has eliminated AHS in clinical practice and is mandated internationally. HLA B*15:02, present predominantly in Southeast Asian populations, is strongly associated with carbamazepine-induced Stevens-Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN); regulatory agencies in Taiwan, Hong Kong, China, and Thailand require HLA B*15:02 screening before carbamazepine initiation. HLA (human leukocyte antigen) A*31:01, found across multiple ethnic groups, is associated with carbamazepine-induced drug reaction with eosinophilia and systemic symptoms (DRESS) and is relevant to European populations where HLA B*15:02 is rare. Allopurinol-induced SJS or TEN is associated with HLA B*58:01, particularly in Asian populations, and screening is recommended before initiation in high-risk groups.6
Glucose-6-phosphate dehydrogenase (G6PD) deficiency, an X-linked condition present in up to 400 million people worldwide (particularly those of African, Mediterranean, and South Asian ancestry), impairs erythrocyte protection against oxidative stress. Drugs that generate oxidative stress trigger acute hemolytic anemia in G6PD-deficient patients: primaquine and tafenoquine (used for malaria radical cure), rasburicase (used for tumor lysis syndrome), dapsone, nitrofurantoin, and some sulfonamides. G6PD testing is mandated before primaquine and rasburicase administration. The severity of deficiency varies by variant -- Class I variants (severe, chronic hemolysis) are rare; Class II and III variants (severe to moderate deficiency, episodic hemolysis) account for most clinically relevant cases.
The translation of pharmacogenomic knowledge into routine clinical practice has accelerated with the development of multi-gene panel testing, electronic health record (EHR) integration of genotype data, and institution-wide clinical decision support systems. Two distinct implementation models -- reactive testing at the point of prescribing and pre-emptive panel genotyping stored for future use -- reflect different balances between upfront cost, test turnaround time, and breadth of utility across a patient's lifetime medication history.
Reactive vs. Pre-Emptive Testing. In the reactive testing model, pharmacogenomic testing is ordered when a clinician is considering a specific drug for which genotype information would change the prescribing decision. This is the model currently used most widely in practice for tests such as HLA (human leukocyte antigen) B*57:01 before abacavir, HLA B*15:02 before carbamazepine in at-risk populations, and TPMT (thiopurine methyltransferase) before thiopurines. The limitation of reactive testing is that results are not available immediately and may delay initiation of urgently needed therapy; for many settings, turnaround time of 1-5 days is acceptable, but for acute-onset conditions it is not. In the pre-emptive model, a comprehensive panel of clinically relevant pharmacogenomic variants is genotyped once, typically as part of a population health or precision medicine program, and the results are stored permanently in the patient's EHR. When the patient is subsequently prescribed any drug with a relevant gene-drug interaction, the existing genotype result is automatically retrieved and used to guide dosing without any additional testing delay.16
Multi-Gene Panel Testing and EHR Integration. Commercial multi-gene pharmacogenomic panels now routinely assay 15-30 or more pharmacogenomically relevant genes simultaneously from a single buccal swab or blood sample, at costs that have declined substantially as genomic sequencing technology has matured. The analytic challenge is not the genotyping itself but the phenotype translation, clinical interpretation, and presentation of results in a form usable by the prescribing clinician at the point of care. Effective implementation requires EHR-embedded clinical decision support (CDS) that fires an alert or recommendation when a pharmacogenomically significant drug is prescribed to a patient whose stored genotype predicts an abnormal response. Large academic medical centers that have deployed pre-emptive pharmacogenomic panels with EHR-integrated CDS have demonstrated feasibility and clinical uptake, though widespread implementation in community practice settings remains limited by infrastructure, reimbursement, and clinician education barriers.16
Limitations and Evolving Evidence. Several important limitations of current pharmacogenomic testing warrant clinician awareness. Most validated gene-drug pairs involve metabolizing enzyme variants affecting drug concentration; pharmacodynamic variants affecting receptor sensitivity or downstream signaling -- which may explain a different dimension of drug response variability -- are less well characterized and less routinely tested. Genotyping panels capture known variants but may miss rare or population-specific variants not included in the panel design. As noted for CYP2D6 (cytochrome P450 2D6), phenocopying by drug interactions can override the predicted phenotype, requiring concurrent drug interaction assessment at every prescribing encounter. Additionally, epigenetic modifications, organ function, inflammation, and disease state can all shift the effective metabolizer phenotype independently of germline genetics. The evidence base for clinical utility -- defined as improved patient outcomes attributable to pharmacogenomic-guided prescribing -- is strongest for reactive testing of high-evidence gene-drug pairs (abacavir/HLA (human leukocyte antigen), clopidogrel/CYP2C19 (cytochrome P450 2C19), thiopurines/TPMT, fluoropyrimidines/DPYD (dihydropyrimidine dehydrogenase)) and is still accumulating for broader pre-emptive panel programs.6
Phenotype categories: poor metabolizer (PM), intermediate (IM), normal (NM), ultrarapid (UM). Key CYP polymorphisms: CYP2D6 (codeine, tamoxifen -- PM cannot activate; UM converts too rapidly), CYP2C19 (clopidogrel -- PM has reduced platelet inhibition; use prasugrel/ticagrelor), CYP2C9 (warfarin -- PM requires lower doses), CYP3A5 (tacrolimus -- non-expressers require lower doses). TPMT/NUDT15: test before thiopurines; PM has fatal myelosuppression risk. DPYD: test before fluoropyrimidines; PM has severe toxicity risk. HLA-B*57:01: test before abacavir. HLA-B*15:02: test before carbamazepine in SE Asian patients. G6PD: test before primaquine and rasburicase. Pre-emptive panel + EHR CDS = optimal model; phenocopying by drug interactions always requires concurrent assessment.
Relling MV, Evans WE. Pharmacogenomics in the clinic. Nature. 2015;526(7573):343-350.
doi:10.1038/nature15817Ritter JM, Flower R, Henderson G, Loke YK, MacEwan D, Rang HP. Rang & Dale's Pharmacology. 9th ed. Edinburgh: Elsevier; 2019. ISBN 9780702074486.
ISBN 9780702074486Amstutz U, Henricks LM, Offer SM, et al. Clinical Pharmacogenetics Implementation Consortium (CPIC) guideline for dihydropyrimidine dehydrogenase genotype and fluoropyrimidine dosing: 2017 update. Clin Pharmacol Ther. 2018;103(2):210-216.
doi:10.1002/cpt.911Crews KR, Monte AA, Huddart R, et al. Clinical Pharmacogenetics Implementation Consortium (CPIC) guideline for CYP2D6, OPRM1, and COMT genotype and select opioid therapy. Clin Pharmacol Ther. 2021;110(4):888-896.
doi:10.1002/cpt.2149Relling MV, Schwab M, Whirl-Carrillo M, et al. Clinical Pharmacogenetics Implementation Consortium guideline for thiopurine dosing based on TPMT and NUDT15 genotypes: 2018 update. Clin Pharmacol Ther. 2019;105(5):1095-1105.
doi:10.1002/cpt.1304Caudle KE, Keeling NJ, Klein TE, et al. Standardization can accelerate the adoption of pharmacogenomics: current status and the path forward. Pharmacogenomics. 2018;19(10):847-860.
doi:10.2217/pgs-2018-0028