Golden Helix · Clinical Genomics Guide

Pharmacogenomics (PGx) Testing
A Clinical and Laboratory Guide

How PGx works, the key genes and drug interactions, CPIC guidelines, star allele calling, metabolizer phenotypes, and how clinical labs implement pharmacogenomics programs at scale.

Introduction

A patient's genotype can predict
how a drug will behave.

Pharmacogenomics (PGx) studies how a person's genetic variants influence their response to medications: how drugs are absorbed, metabolized, transported, and how they interact with their molecular targets. The practical goal is to predict, before a drug is prescribed, whether a patient is likely to respond normally, experience toxicity, or get no therapeutic benefit, using inherited genetics rather than trial-and-error prescribing.

The biology is direct. Most drugs are processed by enzymes encoded by specific genes. Inherited single nucleotide polymorphisms, copy number changes, and structural variants alter enzyme activity. A patient with reduced activity metabolizes a drug slowly, so it accumulates and can cause toxicity. A patient with elevated activity from a gene duplication clears the drug so fast that a standard dose produces no effect. These outcomes follow from genotype, and that predictability is the foundation of PGx.

300+

Medications with PGx information in the FDA drug label

~25%

Commonly prescribed drugs metabolized by CYP2D6

150+

Defined CYP2D6 star alleles in PharmVar

Once

Germline PGx is tested once and informs prescribing for life

Metabolizer phenotypes: poor, intermediate, normal, ultra-rapid

>80%

Of fluoropyrimidine catabolism handled by the DPYD enzyme

Definition

What Is Pharmacogenomics?

The term combines pharmacology, the study of drugs, with genomics, the study of the genome. PGx testing as described here analyzes germline DNA: variants present in every cell of the body from birth, stable across a lifetime, and tested once. The result is a permanent genetic profile that informs prescribing decisions for any relevant medication throughout a patient's care.

This is distinct from somatic pharmacogenomics used in oncology, such as EGFR or BRAF tumor testing for targeted therapy selection. Those alterations are acquired in tumor cells and require separate tumor testing. Germline PGx, by contrast, sits within the broader landscape of inherited variant analysis covered in the germline analysis guide.

Clinical Rationale

Why Pharmacogenomics Matters Clinically

Adverse drug reactions are among the most prevalent and preventable causes of patient harm in healthcare. Genetic variation in drug-metabolizing enzymes and transporters accounts for a substantial proportion of these reactions, and for many of the most significant gene-drug interactions the genetic basis has been understood for decades.

The scale shows up in FDA drug labeling: more than 300 medications now include pharmacogenomic information in their prescribing label, with guidance ranging from dose adjustments to genotype-based contraindications. The Clinical Pharmacogenomics Implementation Consortium (CPIC) has published peer-reviewed, evidence-graded prescribing guidelines for dozens of gene-drug pairs, giving laboratories and prescribers actionable, regularly updated guidance.

A patient's CYP2D6 metabolizer phenotype is relevant not just for one drug but for every CYP2D6-metabolized medication they may ever be prescribed: pain medications, antidepressants, antipsychotics, tamoxifen. Tested once, the result is useful for life. Pre-emptive PGx testing, which profiles a patient before any specific drug is prescribed, is now operational in health systems across the United States and Europe, delivering just-in-time guidance at the point of prescribing through EHR integration.

The Gene Set

Key PGx Genes and Drug Interactions

Pharmacogenomics covers a broad gene set, but clinical implementation prioritizes a core group of high-evidence gene-drug pairs where the actionability is clearest and the CPIC evidence level is strongest. CYP2D6 is the most clinically consequential and the most technically demanding to genotype.

Opioids · antidepressants

CYP2D6

Metabolizes ~25% of commonly prescribed drugs: codeine, tramadol, oxycodone, tricyclics, several SSRIs and SNRIs, haloperidol, risperidone, aripiprazole, tamoxifen. The most technically complex PGx gene: 90+ star alleles, frequent copy number variation, and hybrid alleles formed with the CYP2D7 pseudogene. CPIC recommends avoiding codeine in poor and ultra-rapid metabolizers.

Antiplatelet · PPIs

CYP2C19

Activates the antiplatelet prodrug clopidogrel and metabolizes proton pump inhibitors, several SSRIs, and phenytoin. CYP2C19 poor metabolizers have inadequate clopidogrel activation and elevated stent thrombosis risk; CPIC recommends prasugrel or ticagrelor for poor and intermediate metabolizers undergoing PCI.

Warfarin · NSAIDs

CYP2C9 + VKORC1

Together determine warfarin sensitivity. CYP2C9*2 and *3 slow clearance of S-warfarin and raise bleeding risk; VKORC1 promoter variants alter warfarin sensitivity at the drug target, with CYP4F2 contributing through vitamin K metabolism. FDA-approved dosing algorithms combine genotype with clinical factors. CYP2C9 also affects celecoxib, ibuprofen, diclofenac, and sulfonylureas.

Fluoropyrimidines

DPYD

Encodes dihydropyrimidine dehydrogenase, which handles more than 80% of 5-fluorouracil and capecitabine catabolism. Poor and intermediate metabolizers face severe, potentially fatal toxicity: mucositis, myelosuppression, hand-foot syndrome, neurotoxicity. DPYD*2A, c.2846A>T, c.1679T>G, and HapB3 carry strong CPIC evidence for dose reduction or avoidance.

Thiopurines

TPMT + NUDT15

Both metabolize thiopurines (azathioprine, mercaptopurine, thioguanine) used in inflammatory bowel disease, autoimmune conditions, and pediatric acute lymphoblastic leukemia. Poor metabolizers accumulate thioguanine nucleotides and risk severe myelosuppression. NUDT15 variants are particularly prevalent in Asian populations where TPMT variants are rarer.

Statins

SLCO1B1

Encodes the OATP1B1 hepatic transporter that takes statins into hepatocytes. The c.521T>C variant (rs4149056) reduces function, raises systemic simvastatin exposure, and increases statin-induced myopathy and rhabdomyolysis risk. CPIC recommends lower simvastatin doses or alternatives such as rosuvastatin or pravastatin for reduced-function genotypes.

Nomenclature

Star Alleles & Metabolizer Phenotypes

Pharmacogenomics genes use star allele nomenclature rather than standard HGVS notation. Each star allele, such as CYP2D6 *1, CYP2D6 *4, or CYP2C19 *17, names a haplotype defined by a specific combination of variants with a characterized functional consequence. A copy carrying CYP2D6 *4 has a splice site variant (c.1846G>A) that produces a non-functional enzyme, while CYP2D6 *2xN carries a functional allele duplicated N times for elevated activity. The name encapsulates the variant combination and its predicted functional impact, giving labs and clinical systems a shared vocabulary.

The PharmVar (Pharmacogene Variation) database is the authoritative source for star allele definitions, maintained by an international consortium and updated as new alleles are characterized.

From star allele to metabolizer phenotype

Each patient carries two gene copies, one from each parent. The combination of two star alleles is a diplotype, for example CYP2D6 *1/*4. Diplotypes translate into metabolizer phenotypes, the clinical classification that informs prescribing. For CYP2D6 and CYP2C19, an activity score assigns a value to each allele: 1.0 for fully functional, 0.5 for reduced function, 0 for non-functional. The diplotype activity score, the sum of both alleles, maps to a phenotype.

Activity Score	Metabolizer Phenotype	Clinical Implication
0	Poor Metabolizer (PM)	Little or no enzyme activity: drug accumulates, or a prodrug fails to activate
>0 to <1.25	Intermediate Metabolizer (IM)	Reduced enzyme activity: dose adjustment often needed
1.25 to 2.25	Normal Metabolizer (NM)	Expected drug response at standard doses
>2.25	Ultra-Rapid Metabolizer (UM)	Elevated enzyme activity: reduced drug exposure, or prodrug toxicity risk

Not all PGx genes use activity scores. DPYD, TPMT, NUDT15, and SLCO1B1 use direct genotype-to-phenotype lookup tables based on the clinical evidence for specific alleles. Translating raw calls into this standardized vocabulary is part of the tertiary analysis stage that turns variant calls into clinical results.

Evidence Framework

CPIC Guidelines for Clinical PGx

The Clinical Pharmacogenomics Implementation Consortium (CPIC) is an international group of researchers, clinicians, and pharmacists that publishes peer-reviewed prescribing guidelines for gene-drug pairs with sufficient evidence to inform clinical decision-making. Guidelines are freely accessible and updated as evidence accumulates. CPIC organizes gene-drug pairs into tiers by the strength of the prescribing recommendation.

01
Tier A: strong evidence, specific prescribing recommendations
CPIC provides specific prescribing recommendations. Examples include CYP2D6 and codeine, CYP2C19 and clopidogrel, DPYD and fluoropyrimidines, TPMT and NUDT15 with thiopurines, and SLCO1B1 and simvastatin. These are the highest-priority pairs for a clinical PGx program.
02
Tier B: moderate evidence
CPIC provides recommendations but notes that the evidence is less definitive, or that clinical implementation is more context-dependent. Tier B pairs are useful for expanding a panel beyond the Tier A core.
03
Tier C / D: insufficient or conflicting evidence
CPIC does not currently recommend clinical implementation for these gene-drug pairs.
04
Complementary frameworks: DPWG and FDA labeling
The Dutch Pharmacogenetics Working Group (DPWG) publishes independent guidelines with different evidence thresholds, particularly relevant for European programs. FDA pharmacogenomic labeling covers over 300 medications and provides regulatory context, but it is a labeling requirement rather than a prescribing recommendation framework. The two are complementary.

In the Lab

How Clinical Labs Implement PGx Testing

Reactive vs pre-emptive testing

Clinical PGx programs run in two models. Reactive testing is ordered when a specific question arises: a patient is about to receive clopidogrel, so CYP2C19 genotyping is ordered first. It is targeted and immediately actionable, but does not provide the full profile that benefits future prescribing. Pre-emptive testing profiles a patient across a broad gene panel before any specific drug is prescribed, stores the result in the EHR, and surfaces it through clinical decision support when a relevant medication is later prescribed. Pre-emptive testing is more complex to implement but compounds in value: one test informs many prescribing decisions over a lifetime.

The laboratory PGx workflow

Building a clinical PGx program means solving the bioinformatic and interpretation infrastructure first. The workflow runs through five stages, and each stage introduces technical complexity that general-purpose tools do not handle on their own. For a closer look at how this is implemented in software, see pharmacogenomics software.

01
Variant calling from the sequencing assay
Identify SNVs, indels, and structural variants at the gene positions that define PGx star alleles. Coverage and sensitivity at each CPIC-specified position set the ceiling for everything downstream.
02
Star allele identification
Map called variants to named haplotypes. Complex loci such as CYP2D6 need specialized algorithms for combinatorial allele assignment, copy number analysis, and structural variant detection. General-purpose variant callers are not designed for this.
03
Diplotype assignment
Pair the two star alleles a patient carries (for example, CYP2D6 *1/*4), accounting for allele phasing where it affects the call.
04
Metabolizer phenotype classification
Translate the diplotype into a phenotype using activity scores (CYP2D6, CYP2C19) or direct genotype-to-phenotype lookup tables (DPYD, TPMT, NUDT15, SLCO1B1) drawn from current CPIC gene-drug pair definitions.
05
Report generation
Link phenotypes to CPIC-guided prescribing recommendations and present them as clinician-readable guidance for review by qualified laboratory personnel.

Assay selection

PGx testing can run from several assay types, each with a different trade-off between coverage, sensitivity, and operational simplicity.

01
Targeted PGx panels
Purpose-built for high-throughput clinical PGx, designed to cover the alleles in clinically actionable genes with high sensitivity. The most operationally straightforward entry point for a new PGx program.
02
SNP microarrays with PGx probe sets
Provide star allele calls for genes with well-characterized common alleles, but may miss rare alleles and are generally less sensitive for structural variants such as CYP2D6 copy number changes.
03
Whole exome sequencing
Can generate PGx results as a secondary output from sequencing already performed for rare disease or hereditary cancer. Coverage completeness for specific PGx alleles varies by capture kit, so validate per allele rather than assuming whole-gene coverage.
04
Whole genome sequencing
Provides the broadest coverage, including non-coding regulatory variants such as the VKORC1 promoter and structural variants that targeted panels and exomes may miss.

Wider Context

PGx in the Broader Genomics Program

Pharmacogenomics sits within a broader germline testing context. PGx variants are germline, present in every cell, and stable from birth. How they fit within inheritance patterns and clinical applications is covered in germline analysis.

For labs running multi-indication genomics programs, PGx is increasingly one output among several from the same sequencing assay, alongside rare disease variant interpretation, hereditary cancer risk assessment, or carrier screening. The sequencing infrastructure that supports these programs is covered in NGS analysis, and the tertiary analysis pipeline that turns raw variant calls into clinical results applies to PGx and other germline analyses alike.

Common Questions

Frequently Asked Questions

What is the difference between pharmacogenomics and pharmacogenetics?

The terms are often used interchangeably, but there is a traditional distinction. Pharmacogenetics refers to how single gene variants affect drug response: the original focus of the field, dating to 1950s observations such as G6PD deficiency causing hemolytic anemia after primaquine. Pharmacogenomics is the broader term, covering genome-scale analysis of how variation across multiple genes and the wider genomic context influences drug response. In clinical practice the terms are used synonymously, and PGx is the standard abbreviation for both.

How is PGx testing different from genetic testing for disease risk?

PGx testing analyzes how inherited variants affect drug metabolism and drug response. It does not diagnose disease or assess the risk of developing a condition. CYP2D6 poor metabolizer status does not mean a patient is sick; it means specific CYP2D6-metabolized drugs behave differently in that patient. Hereditary disease risk testing, such as BRCA1/2 testing for breast and ovarian cancer risk, assesses the probability of developing a condition from a pathogenic germline variant. The frameworks differ too: PGx uses star allele nomenclature and metabolizer phenotypes, while hereditary disease testing uses the ACMG five-tier classification. Both analyze germline DNA, but the clinical questions and outputs are distinct. See germline analysis for the wider context.

Are PGx results permanent, or do they need to be repeated over time?

PGx results based on germline DNA are permanent. They reflect constitutional variants present in every cell from birth and do not change over a lifetime. A test performed once is valid for all future prescribing decisions involving relevant gene-drug pairs, which is a central argument for pre-emptive PGx: the genotyping cost is incurred once, but the utility compounds with every relevant prescription. Somatic mutations acquired during cancer development can affect drug metabolism in tumor cells, but they are distinct from the germline PGx profile and require separate tumor testing.

What does it mean to be a poor metabolizer vs. an ultra-rapid metabolizer?

Metabolizer phenotypes describe the functional activity of a drug-metabolizing enzyme based on genotype. A poor metabolizer carries two non-functional or severely reduced-function gene copies, so the enzyme has little or no activity and drugs it normally clears accumulate, raising toxicity risk. An ultra-rapid metabolizer carries additional functional copies through gene duplication, so the enzyme is overactive and clears drugs so quickly that standard doses may be ineffective. With prodrugs such as codeine, ultra-rapid metabolism instead produces active metabolites so fast that they can reach toxic levels. Normal metabolizers respond as expected at standard doses, and intermediate metabolizers have partially reduced activity and may need dose adjustment.

Which medications are most affected by PGx variants?

The strongest clinical evidence is concentrated in several drug classes. Pain management: codeine, tramadol, and oxycodone (CYP2D6). Cardiology: clopidogrel (CYP2C19), and warfarin dosing guided by CYP2C9 and VKORC1. Psychiatry: many antidepressants and antipsychotics are CYP2D6 and CYP2C19 substrates. Oncology: fluoropyrimidines such as 5-FU and capecitabine (DPYD), and thiopurines such as azathioprine and mercaptopurine (TPMT and NUDT15). Cholesterol management: simvastatin myopathy risk (SLCO1B1). FDA drug labeling includes pharmacogenomic information for over 300 medications across these and other classes.

What is the role of CPIC guidelines in clinical PGx?

CPIC (Clinical Pharmacogenomics Implementation Consortium) guidelines are the primary clinical evidence framework for pharmacogenomics implementation. CPIC publishes peer-reviewed, freely accessible guidelines for gene-drug pairs, with specific prescribing recommendations tied to metabolizer phenotype: what dose to use, what alternative to consider, what monitoring is appropriate. Guidelines are updated as evidence accumulates and are tiered by evidence strength (Tier A through D). Tier A pairs such as CYP2D6 and codeine, CYP2C19 and clopidogrel, and DPYD and fluoropyrimidines are the highest-priority targets. CPIC guidelines are distinct from FDA pharmacogenomic labeling, which is a regulatory requirement rather than an implementation framework, though the two are complementary.

How do clinical labs validate a PGx testing program?

A CLIA-certified clinical PGx program validates its assay and interpretation pipeline before reporting patient results. Validation typically demonstrates analytical accuracy of star allele calls against reference materials such as the Genetic Testing Reference Materials Coordination Program (GeT-RM), coverage completeness for clinically relevant alleles, and reproducibility across runs and operators. For complex loci such as CYP2D6, validation must specifically address copy number detection and hybrid allele identification. The interpretation pipeline from star allele to phenotype to recommendation must also be documented and validated, and deterministic software, where the same input produces the same output every run, supports that documentation.

Keep Reading

Related Resources

Pharmacogenomics intersects germline, tertiary, and platform topics across the Golden Helix guide series.

Pharmacogenomics SoftwareHow the star-allele-to-report workflow is implemented for research and LDT development.Germline Analysis: Inherited Variant TestingWhere PGx sits among rare disease, hereditary cancer, and carrier screening.NGS Analysis: A Complete GuideThe sequencing infrastructure that feeds a multi-indication genomics program.Tertiary Analysis in NGSAnnotation, filtering, and classification steps shared across PGx and other germline analyses.VSPGx in VarSeqCPIC-aligned star allele calling and phenotyping for research workflows.Why Pharmacogenomics MattersThe clinical case for PGx and how labs put it into production.

Recommended eBook