Pharmacogenomics in Europe: DNA and Drug Response

Pharmacogenomics in Europe: How Your DNA Affects Which Drugs Work For You

TL;DR: Pharmacogenomics (PGx) studies how genetic variants in drug-metabolizing enzymes — primarily CYP2D6, CYP2C19, CYP2C9, DPYD, and TPMT — affect your response to medications. These variants can make standard drug doses ineffective or dangerously toxic. Clinical guidelines (CPIC, DPWG) already cover over 100 drug-gene interactions, and Europe leads globally in pre-emptive PGx testing, with the PREPARE study showing a 30% reduction in adverse drug reactions.

Every year, adverse drug reactions (ADRs) account for roughly 197,000 deaths across the European Union and cost healthcare systems an estimated 79 billion euros. A significant fraction of these reactions are preventable -- because they stem not from prescribing errors or allergies, but from genetic variation in how individual patients metabolize medications.

This is the domain of pharmacogenomics (PGx): the study of how your DNA influences your response to drugs. It is one of the most clinically actionable areas of genomics today, and Europe is at the forefront of integrating it into routine healthcare.

What Is Pharmacogenomics?

Pharmacogenomics (PGx): The study of how inherited genetic variation affects an individual's response to drugs, including drug efficacy, optimal dosing, and risk of adverse reactions. Clinical PGx guidelines translate genotype results into specific prescribing recommendations.

Metabolizer phenotype: A classification assigned to a patient based on their genotype for a drug-metabolizing enzyme, ranging from poor metabolizer (little or no enzyme activity) to ultra-rapid metabolizer (substantially increased activity). The phenotype determines whether standard drug doses will be effective, ineffective, or toxic.

Pharmacogenomics is the study of how your DNA influences your response to medications. Genetic variants in drug-metabolizing enzymes can make the same dose of the same drug therapeutic for one person, ineffective for another, and dangerously toxic for a third. Clinical guidelines already exist for over 100 drug-gene interactions, making PGx one of the most immediately actionable areas of genomics.

When you take a medication, your body must absorb it, distribute it to the right tissues, metabolize it (often in the liver), and eventually eliminate it. At nearly every step, enzymes encoded by your genes do the heavy lifting. Variation in those genes -- different alleles inherited from your parents -- can make these enzymes work faster, slower, or not at all.

The result: the same dose of the same drug can be therapeutic for one person, ineffective for another, and dangerously toxic for a third. Pharmacogenomics identifies these genetic differences and translates them into concrete prescribing guidance -- which drug to choose, and at what dose.

This is not theoretical medicine. Clinical guidelines already exist for over 100 drug-gene interactions, and the evidence base grows every year.

The Key Genes: A Practical Overview

Five gene families account for the majority of clinically actionable pharmacogenomic interactions. Understanding them is the foundation of PGx literacy.

CYP2D6: The Most Polymorphic Drug-Metabolizing Enzyme

CYP2D6 metabolizes approximately 25% of all commonly prescribed drugs, including codeine, tramadol, tamoxifen, and many antidepressants (venlafaxine, nortriptyline, paroxetine). It is also one of the most genetically variable enzymes in the human body, with over 130 known allelic variants and significant differences in allele frequency across populations. For a concise reference on the most clinically relevant star alleles and phenotype groupings, see our CYP2D6 gene page.

Why it matters clinically: Codeine is a prodrug -- it does nothing until CYP2D6 converts it into morphine. Poor metabolizers make very little morphine and get diminished pain relief; CPIC advises avoiding codeine in this group. Ultra-rapid metabolizers convert codeine to morphine so efficiently that standard doses can cause respiratory depression, which has proven fatal in children and breastfed infants. One of the FDA boxed warnings on codeine covers ultra-rapid metabolism and life-threatening respiratory depression in children. Codeine is contraindicated in children under 12, and in children under 18 after tonsillectomy or adenoidectomy. Elsewhere, the label says breastfeeding is not recommended during treatment.

For antidepressants, the picture differs by drug. CPIC's 2023 guideline (Bousman et al.) gives CYP2D6-based advice for some, such as paroxetine and fluvoxamine. For paroxetine, poor metabolizers may have higher drug levels and more side effects; CPIC suggests a lower starting dose and slower titration. Ultra-rapid metabolizers have lower drug levels, so benefit is less likely; CPIC suggests an alternative drug. For others, such as duloxetine, the evidence does not support a CYP2D6-based recommendation.

CYP2C19: Clopidogrel, PPIs, and SSRIs

CYP2C19 is critical for activating clopidogrel (Plavix), the antiplatelet drug prescribed to millions of patients after cardiac stenting. Clopidogrel is another prodrug: poor metabolizers cannot activate it effectively, leaving them at significantly elevated risk of stent thrombosis -- a potentially fatal event.

The FDA added a boxed warning to clopidogrel in 2010 recommending consideration of alternative therapies for CYP2C19 poor metabolizers. Yet most patients still receive clopidogrel without any genetic testing.

CYP2C19 also metabolizes proton pump inhibitors (omeprazole, lansoprazole) and several SSRIs (escitalopram, sertraline). For PPIs, CPIC (Lima et al., 2021) suggests a standard starting dose for normal and rapid metabolizers. For H. pylori infection and erosive esophagitis, it suggests considering a 50–100% higher dose. For ultra-rapid metabolizers, it suggests a 100% higher starting dose. For citalopram, escitalopram and sertraline, poor metabolizers have shown higher drug levels. CPIC suggests an alternative antidepressant or a lower dose for this group.

CYP2C9 and VKORC1: The Warfarin Story

Warfarin remains one of the most widely prescribed anticoagulants in the world, and it has one of the narrowest therapeutic windows of any drug. Too little, and patients form life-threatening clots. Too much, and they hemorrhage.

Two genes dominate warfarin pharmacogenomics. CYP2C9 metabolizes the more potent S-enantiomer of warfarin; variants like CYP2C9*2 and *3 reduce enzyme activity, causing the drug to accumulate. VKORC1 encodes the molecular target of warfarin; common variants alter sensitivity to the drug at the receptor level.

Together, CYP2C9 and VKORC1 genotypes explain roughly 40% of the variability in warfarin dose requirements -- far more than any clinical factor alone. Genotype-guided dosing algorithms have been validated in randomized controlled trials and are recommended by CPIC guidelines.

DPYD: When Chemotherapy Becomes Life-Threatening

Dihydropyrimidine dehydrogenase, encoded by the DPYD gene, is responsible for breaking down fluoropyrimidine chemotherapy agents -- 5-fluorouracil (5-FU) and its oral prodrug capecitabine. These are among the most commonly used chemotherapies worldwide, prescribed for colorectal, breast, head and neck, and gastric cancers.

Approximately 3-8% of the European population carries a partial DPYD deficiency, and roughly 0.1-0.5% are fully deficient. For these patients, standard fluoropyrimidine doses cause catastrophic toxicity: severe mucositis, myelosuppression, and in some cases, death. The European Medicines Agency (EMA) now recommends DPYD testing before prescribing fluoropyrimidines, and several European countries have made it mandatory.

This is pharmacogenomics at its most urgent. A simple genetic test costing under 200 euros can prevent a fatal drug reaction.

TPMT and NUDT15: Thiopurine Toxicity

TPMT (thiopurine S-methyltransferase) and NUDT15 metabolize thiopurine drugs -- azathioprine, mercaptopurine, and thioguanine -- used to treat autoimmune conditions, inflammatory bowel disease, and acute lymphoblastic leukemia. Patients with reduced TPMT or NUDT15 activity accumulate toxic thioguanine nucleotides, leading to severe and potentially fatal myelosuppression.

CPIC's thiopurine guideline (Relling et al., 2019) describes the benefits of pre-emptive TPMT and NUDT15 testing. It tells clinicians how to use results that are available. For intermediate metabolizers, it recommends reduced starting doses (30–80% of the normal dose, depending on the starting dose). For poor metabolizers with cancer, it recommends drastically reduced doses: about a 10-fold reduction, given three times a week. For non-malignant conditions, it suggests considering an alternative agent.

Metabolizer Phenotypes: The Classification System

Pharmacogenomics assigns each patient a metabolizer phenotype based on their genotype for a given enzyme. The standard classification includes five categories:

• Poor Metabolizer (PM): Little to no functional enzyme activity. Prodrugs are not activated; active drugs accumulate to toxic levels.
• Intermediate Metabolizer (IM): Reduced enzyme activity. Dose reductions are often needed.
• Normal Metabolizer (NM): Typical enzyme activity. Standard dosing applies. (Previously called "extensive metabolizer.")
• Rapid Metabolizer (RM): Above-average enzyme activity. Some drugs are cleared faster than expected.
• Ultra-Rapid Metabolizer (UM): Substantially increased enzyme activity, often due to gene duplication. Prodrugs may cause toxicity from excessive activation; active drugs may be ineffective.

The clinical significance of each phenotype depends entirely on the specific drug. Being a CYP2D6 ultra-rapid metabolizer is dangerous with codeine but may be clinically irrelevant for other medications.

Real Clinical Impact: What Labels and Guidelines Say

Drug labels and guidelines already spell out what is at stake for three well-studied gene–drug pairs:

Case 1: Codeine and breastfeeding. The FDA label for codeine reports at least one death in a breastfed infant exposed to high morphine levels in breast milk because the mother was a CYP2D6 ultra-rapid metabolizer. The best-known published case report (Koren et al., Lancet 2006) has carried an Expression of Concern from the journal's editors since February 2026, so it should be read with caution.

Case 2: Clopidogrel after a stent. CPIC's 2022 guideline (Lee et al.) says CYP2C19 poor metabolizers have significantly reduced formation of clopidogrel's active metabolite and a higher risk of cardiac and cerebrovascular events. For patients with acute coronary syndrome or a coronary stent procedure, it recommends avoiding clopidogrel if possible. It recommends prasugrel or ticagrelor at standard dose instead, if there is no contraindication. CYP2C19 genotype did not affect how those two drugs work.

Case 3: Capecitabine and DPD deficiency. The FDA label for capecitabine carries a boxed warning: serious adverse reactions or death may occur in patients with complete DPD deficiency. The label tells clinicians to test for DPYD variants before starting treatment, unless treatment is urgent. CPIC's DPYD guideline (Amstutz et al., 2018) recommends a 50% lower starting dose for many intermediate metabolizers, then adjusting by tolerance. For poor metabolizers, it recommends avoiding fluoropyrimidines.

CPIC Guidelines: The Evidence Standard

The Clinical Pharmacogenetics Implementation Consortium (CPIC) publishes peer-reviewed, evidence-based guidelines that translate genotype results into specific prescribing recommendations. Each guideline undergoes rigorous systematic review and assigns a level of evidence to each drug-gene interaction.

CPIC guidelines currently cover over 25 genes and nearly 100 drug-gene pairs. They are freely available at cpicpgx.org and are designed for direct clinical implementation -- they tell prescribers exactly what to do when a patient's genotype is known.

Critically, CPIC guidelines address a common misconception: they do not recommend whether to test. They assume the genotype is already available and provide guidance on how to use it. This "test-agnostic" approach means the guidelines are equally applicable whether the genotype came from a dedicated PGx panel, whole-genome sequencing, or reanalysis of existing consumer genomics data.

Europe Leading the Way

While pharmacogenomics implementation has been uneven globally, several European countries are pioneering systematic integration of PGx into clinical care.

The Netherlands: DPWG and Pre-Emptive Testing

The Dutch Pharmacogenetics Working Group (DPWG) has been publishing pharmacogenomics guidelines since 2005, and the Netherlands is arguably the world leader in clinical PGx implementation. Dutch guidelines are integrated directly into electronic prescribing systems: when a pharmacist dispenses a medication, the system automatically checks whether a relevant genotype is on file and generates an alert with dosing recommendations.

The PREPARE study, a landmark European randomized controlled trial across seven countries, demonstrated that pre-emptive PGx panel testing -- genotyping a panel of pharmacogenes before any specific drug is prescribed -- reduced adverse drug reactions by 30%. This "test once, use many times" model is the future of pharmacogenomics.

United Kingdom: NHS Pharmacogenomics Pilot

NHS England has launched pharmacogenomics pilot programs through the National Genomic Medicine Service, exploring how PGx testing can be integrated into primary care and oncology pathways. The UK Pharmacogenomics Clinical Implementation Group is developing clinical decision support tools and working to establish PGx testing as a standard component of the NHS Long Term Plan.

The UK Biobank, with genetic and health data from 500,000 participants, has also become an invaluable resource for pharmacogenomics research, enabling large-scale studies of drug-gene interactions in real-world populations.

The EU 1+ Million Genomes Initiative

The 1+ Million Genomes Initiative (1+MG), signed by 25 EU member states, aims to create cross-border access to genomic data for research and clinical care. Pharmacogenomics is one of the initiative's priority use cases, with the explicit goal of enabling PGx-guided prescribing across European healthcare systems.

This initiative addresses one of the key barriers to PGx adoption: interoperability. By establishing common standards for genomic data, 1+MG makes it possible for a PGx profile generated in one country to be used by a prescriber in another.

Getting Your PGx Profile from Existing Data

If you have already taken a consumer genomics test from 23andMe, AncestryDNA, or a similar provider, you may already have data relevant to pharmacogenomics. These services genotype hundreds of thousands of SNPs across the genome, and many of the key pharmacogenomic variants are included on their arrays.

By downloading your raw data file and taking it to a raw-data analysis service, you can extract pharmacogenomic insights from data you have already paid for. If you write code, this Python walkthrough does that for CYP2C19 and SLCO1B1 — genotypes to star alleles to CPIC phenotypes — and lists what an array cannot tell you. DeepDNA itself is an API for developers and AI agents, not a consumer service. If you are unsure how to obtain your raw data, our complete guide to using your 23andMe raw data walks through the process step by step.

Important Limitations: What SNP Arrays Cannot Detect

Transparency about limitations is essential. Consumer genotyping arrays (including those from 23andMe and Ancestry) have important blind spots for pharmacogenomics:

Structural variants and copy number variation. CYP2D6 is notorious for gene deletions, duplications, and hybrid gene arrangements. SNP arrays cannot reliably detect CYP2D6 gene copy number, which means they may miss ultra-rapid metabolizers (who carry extra gene copies) and some poor metabolizers (who carry whole-gene deletions). This is clinically significant.

Rare and novel variants. SNP arrays test for a predefined set of known variants. If you carry a rare or population-specific allele that is not on the array, it will not be detected. Your result will default to the reference allele, potentially assigning you a "normal metabolizer" phenotype when your true phenotype is different.

Star allele assignment complexity. Translating raw SNP data into star alleles (the nomenclature system used by CPIC) requires sophisticated phasing algorithms, especially for genes like CYP2D6 where the relationship between SNPs and function is complex.

For these reasons, a PGx report derived from a consumer SNP array should be considered a useful screening tool, not a definitive clinical-grade result. When a clinically significant finding is identified, confirmatory testing through a certified laboratory may be warranted, particularly before making high-stakes prescribing decisions.

What an Honest Pharmacogenomics Result Looks Like

Whatever reads your raw data — a clinical lab, a script you write yourself, or a service — the same standard applies. Genotype should be translated into metabolizer phenotypes using CPIC guidelines, and each result should carry a confidence level: where a star allele call is well supported by the available SNPs it can be reported as such; where structural variation (CYP2D6) or phasing ambiguity limits certainty, the result should say so plainly and point toward confirmatory testing.

Related areas are easy to miss when they are sold separately. Variants in the MTHFR gene, for example, affect folate metabolism and can interact with methotrexate response — an intersection of nutrigenomics and pharmacogenomics.

If you are building a tool that does this, DeepDNA is an API for developers and AI agents, not a consumer service: its knowledge endpoints return the CPIC guidance, phenotypes and sources for pairs such as CYP2C19–clopidogrel, free and without a key, and this worked example reads CYP2C19 and SLCO1B1 from a raw file in Python.

The Path Forward

Pharmacogenomics is not a promise for the future -- it is a clinical reality today. The evidence base is strong, the guidelines are mature, and the technology is accessible. What remains is implementation: getting the right test to the right patient at the right time.

Europe's pre-emptive testing programs, regulatory mandates for DPYD genotyping, and cross-border genomic data initiatives represent the most ambitious pharmacogenomics implementation efforts anywhere in the world. As these programs scale, the question will shift from "should we test?" to "why haven't we tested yet?"

Your genome does not change. A pharmacogenomics profile generated today will remain relevant for every prescription you receive for the rest of your life. Whether you are starting a new medication, managing a chronic condition, or simply want to be prepared, understanding how your genes affect your drug response is one of the most practical steps you can take with your genetic data. To understand how the testing process identifies these variants, see our explainer on how DNA testing works. If you're looking for European platforms that offer PGx analysis, our comparison of 23andMe alternatives in Europe covers the options.

Frequently Asked Questions

Can I get a pharmacogenomics report from my 23andMe data?

Yes. Consumer genotyping arrays like 23andMe cover many key pharmacogenomic SNPs. A raw-data analysis tool can extract PGx insights from data you have already paid for — though structural variants like CYP2D6 gene duplications may require confirmatory clinical testing.

Which drugs are most affected by pharmacogenomic variants?

The drugs with the strongest evidence for genetic influence include codeine and tramadol (CYP2D6), clopidogrel (CYP2C19), warfarin (CYP2C9/VKORC1), fluoropyrimidine chemotherapies like 5-FU and capecitabine (DPYD), and thiopurines like azathioprine (TPMT/NUDT15).

Is pharmacogenomic testing available through European healthcare systems?

Yes, and increasingly so. The Netherlands leads with pre-emptive PGx panel testing integrated into prescribing systems. The EMA recommends DPYD testing before fluoropyrimidine chemotherapy, and several EU countries have made it mandatory. NHS England has launched PGx pilot programs in primary care and oncology.

Does a pharmacogenomics test need to be repeated?

No. Your genome does not change. A PGx profile generated once remains relevant for every prescription you receive for the rest of your life.

Building something with pharmacogenomics? DeepDNA's knowledge endpoints return CPIC guidance for drug–gene pairs such as CYP2C19–clopidogrel and SLCO1B1–simvastatin, each with its sources — free, no key. For a worked example that reads a raw DNA file in Python, see CYP2C19 and SLCO1B1 from a raw DNA file. DeepDNA is an API for developers and AI agents, not a consumer service; see the API docs.

This article was created with AI assistance and reviewed by the DeepDNA editorial team.