Intermediateapplied biology
Personalized Medicine: Treatment Tailored to Your Biology
Learning objectives
- Explain why individuals respond differently to the same drug.
- Describe pharmacogenomic examples such as CYP2D6 and TPMT.
- Explain how tumor sequencing guides targeted cancer therapy.
- Interpret polygenic risk scores as probabilistic, not deterministic.
- Discuss limits including gene-environment interaction, privacy, and ancestry bias.
Why One Dose Doesn't Fit All
The standard drug dose is calibrated for an average patient who does not exist. Individuals differ in how they absorb, activate, and clear drugs because the enzymes doing that work are encoded by genes that vary between people. A dose that is therapeutic for one person can be ineffective or toxic for another. Age, weight, liver and kidney function, diet, and other medications all contribute, but genetics explains a meaningful slice of the variation. Pharmacogenomics, the study of how genes affect drug response, turns this variation from a guessing game into testable information: a genetic test can predict whether a patient will metabolize a drug too quickly, too slowly, or just right, before the first pill is taken.
Pharmacogenomics in Action
The examples are concrete. The liver enzyme CYP2D6 activates codeine into morphine; ultrarapid metabolizers convert it so fast that a normal dose can cause overdose, while poor metabolizers get no pain relief at all, which is why some guidelines advise against codeine in children. TPMT testing before thiopurine drugs for leukemia and autoimmune disease identifies patients who would suffer life-threatening bone marrow suppression at standard doses. Warfarin dosing incorporates variants in CYP2C9 and VKORC1. Certain HLA variants predict severe skin reactions to the epilepsy drug carbamazepine. In each case, a one-time genetic test replaces trial-and-error prescribing with a biologically informed starting dose. Implementation is spreading: some health systems now test pharmacogenomic panels preemptively, so the information waits in the medical record before any relevant drug is ever prescribed.
Cancer Leads the Way
Oncology is where personalized medicine is most mature, because tumors themselves can be sequenced. Testing for BRCA1 and BRCA2 mutations identifies people with high inherited risk of breast and ovarian cancer, informing screening and prevention. Within tumors, specific mutations predict drug response: HER2-positive breast cancers respond to trastuzumab, which targets the HER2 protein, and lung cancers with EGFR mutations respond to EGFR inhibitors. Companion diagnostics, tests approved alongside a drug, formalize the pairing. The logic is straightforward: sequence the tumor, find the driver mutation, and choose the drug built for it. When no targetable mutation exists, conventional therapy remains the standard, which keeps expectations honest.
Risk Prediction: Polygenic Scores
Most common diseases, heart disease, diabetes, schizophrenia, are not monogenic but influenced by thousands of variants, each contributing a tiny effect. Genome-wide association studies have catalogued these variants, and polygenic risk scores sum them into a single number estimating relative risk. A high score means higher-than-average probability, not destiny: someone in the top percentile for coronary risk might have triple the average risk yet still most likely never have a heart attack, while lifestyle dominates absolute outcomes. Scores also perform worse in non-European ancestries because the underlying studies overrepresent Europeans, a bias the field is working to correct. Used carefully, they can prioritize screening; used naively, they mislead.
Limits and Challenges
Genes are not the whole story: environment, behavior, and chance shape most health outcomes, and a genetic result never captures diet, infections, or stress. Privacy matters because DNA identifies not just the patient but their relatives; the US Genetic Information Nondiscrimination Act limits use of genetic data in health insurance and employment, though gaps remain, such as life insurance. Ancestry bias in databases means predictions are less accurate for underrepresented groups, risking a new health disparity. Cost and clinician training lag behind the science. Personalized medicine is therefore best understood as a powerful refinement of care, not a replacement for clinical judgment or public health.
The Genetic Information Nondiscrimination Act of 2008 bars health insurers and employers from using genetic information against people, though it does not cover life, disability, or long-term-care insurance.
Key points
- Genetic variation in drug-metabolizing enzymes explains much individual drug-response difference.
- Tests for CYP2D6, TPMT, and HLA variants already guide dosing and drug choice.
- Tumor sequencing matches cancers to targeted drugs via companion diagnostics.
- Polygenic risk scores estimate relative risk from thousands of small-effect variants; they are probabilistic.
- Limits include environment, privacy concerns, ancestry bias, and cost.
Common misconceptions
✓ Truth: It integrates genetics with clinical data, biomarkers, imaging, and lifestyle. A drug level, a blood marker, or a tumor's protein profile can personalize care without any DNA sequencing.
✓ Truth: It improves the odds of a good starting choice but does not eliminate trial and error. Many drug responses involve genes not yet tested, plus non-genetic factors.
✓ Truth: It is mature in oncology and some pharmacogenomics niches but preliminary or absent for most diseases. Marketing often runs ahead of clinical utility.
Frequently asked questions
Do I need my whole genome sequenced to benefit?
Usually not. Most clinical pharmacogenomic decisions use small targeted panels testing a handful of well-understood variants. Whole-genome sequencing is mainly a research tool today, generating far more uncertain findings than actionable ones.
Are polygenic risk scores destiny?
No. They shift probabilities, often modestly. A high score for heart disease might double a small baseline risk, while exercise, diet, and not smoking change absolute risk far more. Think of scores as one risk factor among many, not a forecast.
Who owns my genetic data?
It depends on the test and jurisdiction. Clinical labs operate under medical privacy laws, while direct-to-consumer companies have varied policies, and some share de-identified data with research partners. Reading the consent and privacy terms before testing is essential.
Will personalized medicine replace doctors?
No. Genetic data requires clinical interpretation in the context of symptoms, history, and examination. The realistic future is doctors augmented by genomic tools, not replaced by them.