A clinical panel is a one-time decision that, for the roughly 1-in-50 carrier of a high-impact variant, changes the math on cancer and heart disease — risk-reducing surgery in BRCA carriers cuts all-cause death by about 60%, and Lynch surveillance cuts colon-cancer death by roughly two-thirds. Pharmacogenomic panels are a separate win, shown in 2023 to cut serious drug side effects by 30% across a list of common medications. Consumer spit kits are a weaker tool — useful for prompting a real test, misleading if treated as one. The hard part is the follow-through, not the swab.
Three different technologies sit behind the word genetic testing, and the price tag tracks the difference. A consumer spit kit uses a genotyping chip that probes about 700,000 pre-selected DNA spots — fast and accurate for the spots it covers, blind to anything off the chip. The popular FDA-cleared BRCA report on a kit like 23andMe checks three specific Ashkenazi Jewish founder variants out of more than a thousand pathogenic BRCA1/2 variants known to medicine Finn et al. 2018. A clinical multi-gene panel reads each relevant gene end-to-end with next-generation sequencing — it catches what the chip misses, including large deletions and rarer variants. Whole-genome sequencing reads everything. Costs roughly: a hundred dollars, a few hundred to a thousand, fifteen hundred. The information value tracks the technology, not the marketing.
Genetic information acts on health through three pathways. Predisposition: a variant raises lifetime risk of a specific disease enough to change surveillance schedules or trigger a risk-reducing intervention. Reproduction: carrier status in both members of a couple gives each pregnancy a 25% chance of an affected child, addressable through IVF with embryo testing, prenatal diagnosis, or donor gametes. Drug metabolism: variants in liver enzymes (the CYP family, TPMT, DPYD) and immune-response genes (certain HLA types) change how the body handles a drug — sometimes turning a standard dose into a toxic one, or a standard dose into one that does nothing CPIC guidelines.
What's actually known
For the woman whose mother and aunt both had breast cancer in their forties, a positive BRCA1 result is the moment she gets to do something her mother couldn't. The lifetime numbers are blunt: 72% breast cancer risk and 44% ovarian cancer risk by age 80 for BRCA1 carriers; 69% and 17% for BRCA2 Kuchenbaecker et al. 2017. Removing the ovaries and fallopian tubes around age 35–45 — a procedure called risk-reducing salpingo-oophorectomy — cuts her chance of dying from anything by about 60% over the next decade Domchek et al. 2010. And about half of BRCA carriers in the general population don't have a family history that would have flagged them for testing in the first place Manchanda et al. 2018.
The same math runs in Lynch syndrome, the most common hereditary colon cancer at roughly 1 in 279 adults. Carriers face 30–70% lifetime colon-cancer risk depending on the specific gene. The Finnish 15-year controlled trial that established the standard of care put carriers on colonoscopy every two to three years from their late twenties; colorectal cancer incidence fell 62% and all-cause mortality 65% compared with carriers not surveyed Järvinen et al. 2000. The intervention is a calendar reminder and a procedure room; the upstream input is a known genotype.
Familial hypercholesterolemia — roughly 1 in 250 adults — is the cardiac version. Untreated heterozygotes walk around with LDL high enough to start coronary disease in their twenties; median age of first heart attack is the mid-40s for men, mid-50s for women Nordestgaard et al. 2013. A statin started in young adulthood converts that into an ordinary aging curve. Cascade testing of first-degree relatives — each 50% likely to share the variant — finds new cases at roughly half the people you call Knowles et al. 2014.
The predisposition logic isn't confined to cancer and the heart. Hereditary hemochromatosis — inherited iron overload — is the textbook case outside them: a result that tells you to act long before symptoms. And testing the celiac-risk genes HLA-DQ2/DQ8 can rule the disease out for life — the unusual case where a clean negative is the entire point.
Pharmacogenomics is the second pillar, and the one most readers will eventually use. Most of us will at some point take a drug whose right dose or right choice depends on a liver-enzyme variant we already carry: clopidogrel after a stent, an SSRI for depression, codeine or tramadol for pain, fluoropyrimidine chemotherapy if cancer enters the picture. The PREPARE trial randomized roughly 7,000 patients across seven European countries to either panel-guided prescribing or standard care, with a 12-gene panel and a pharmacist-facing alert at dispensing time. Clinically relevant adverse drug reactions fell 30% over twelve weeks Swen et al. 2023 (PREPARE).
For specific high-stakes pairs the effects are larger. Screening for the HLA-B*5701 variant before starting the HIV drug abacavir essentially eliminated the severe hypersensitivity reaction that used to put people in the ICU — from 2.7% of patients in the unscreened arm of PREDICT-1 to zero in the screened arm Mallal et al. 2008. Upfront DPYD testing before 5-FU or capecitabine chemotherapy dropped severe (grade 3 or worse) toxicity in variant carriers from a historical rate near 70% to 33% in a prospective Dutch multicenter study Henricks et al. 2018. The shared pattern: a result that sits in the chart for life and changes one prescription at a time.
What happens if you don't
The usual stakes story isn't a thriller. Most of us go through life with a family history we never quite formalized. Mom mentions an aunt who had breast cancer at 47. Dad's brother had a heart attack at 52. Nothing gets written down, nothing gets cross-referenced, and the next data point arrives as a symptom in someone who could have known.
The version that lands hardest is the one where the answer was already inside the family. A 45-year-old woman shows up to her primary-care doctor with abdominal bloating; it's stage III ovarian cancer, with a BRCA1 positive on the germline panel ordered after diagnosis. Her sisters, mother, and daughters each now face a 44% lifetime ovarian-cancer risk Kuchenbaecker et al. 2017. Cascade testing in the months that follow will save some of them; the surgery she didn't get to have is the surgery they will. She doesn't get her time back.
The familial-hypercholesterolemia version: a 47-year-old man with no chest-pain history collapses in a parking lot. LDL 280 mg/dL on the labs they draw in the ambulance. His teenage kids share the same variant; if anyone had measured them at twelve and started a statin at twenty, his cardiologist tells the family, this would have been a normal aging curve Nordestgaard et al. 2013.
Pharmacogenomics doesn't fail in dramatic ways — it underwhelms in slow ones. The SSRI that didn't work for two years before someone tried a different one. The chemotherapy dose that landed someone in the ICU when a 30% reduction would have been a normal week. The clopidogrel after a stent that quietly didn't keep the stent open because the prescriber didn't know the liver couldn't activate the drug. The substance the body wanted to receive was a different substance than the one the prescription said.
How to actually do it
What you order depends on what you're trying to find out. The single highest-yield first move is free: a written family history. Cancers and the ages at diagnosis, heart attacks and sudden deaths, three generations on each side. A counselor can read that sheet in five minutes and tell you whether to push for a panel, a targeted test, or nothing at all.
Pre-test counseling matters. The point of the counselor visit isn't paperwork — it's pre-deciding what each possible result will mean. A positive BRCA1 should not be the moment a person first thinks about whether they would consider risk-reducing surgery. A negative result should not be the moment a person realizes their polygenic and lifestyle risk is still real. Board-certified genetic counselors (the CGC credential) increasingly work via telehealth; most direct-pay test vendors include counseling in the price.
When to wait, when not to test
Few absolute reasons not to test; several real reasons to time it carefully.
What most coverage gets wrong
Several confusions are common enough to be load-bearing.
"My 23andMe already checked BRCA." The FDA-cleared BRCA report on consumer kits covers three specific Ashkenazi Jewish founder variants out of more than a thousand pathogenic BRCA1/2 variants in the literature Finn et al. 2018. A negative result on a non-Ashkenazi person rules out almost nothing.
"I downloaded my raw data and ran it through a third-party interpreter, so I know my risks." The genotyping chip's accuracy at probes for rare disease variants is much lower than for common SNPs. When researchers sent 49 patient samples flagged by consumer-grade raw data to a clinical lab, 40% of the variants reported as positive came back as false positives on confirmation testing, and several variants flagged as "increased risk" got reclassified as benign Tandy-Connor et al. 2018. Documented cases of prophylactic mastectomy on unconfirmed consumer-test results exist; clinical confirmation in a CAP/CLIA-certified lab is the standard.
"I'm worried my employer or health insurer will see this." In the US, GINA prohibits both GINA 2008. The legitimate concern most people don't know to think about is life, disability, and long-term-care insurance — those are not covered.
"No family history means no risk." About half of BRCA carriers in unselected populations have no family history that would have triggered standard testing Manchanda et al. 2018, Manickam et al. 2018. Small families, early deaths from other causes, adoption, and parents who didn't share medical history all mean family history is a noisy filter.
"Genetic testing causes lasting psychological harm." Mostly, it doesn't. The REVEAL study disclosed APOE Alzheimer's-risk genotype to over 200 adults and found no measurable long-term distress beyond a transient anxiety spike Green et al. 2009. Bloss et al. found the same pattern in over a thousand consumer-test users a year out Bloss et al. 2011; the MedSeq pilot found no harm from whole-genome sequencing in primary care Vassy et al. 2017. Individual responses vary; the population-level harm signal is small.
"Pharmacogenomics is for psychiatry." It started there in the public conversation, but the higher-stakes evidence is in oncology (DPYD, TPMT), HIV care (HLA-B*5701), cardiology (CYP2C19 for clopidogrel, CYP2C9/VKORC1 for warfarin), and neurology (HLA-B*1502 for carbamazepine).
Where this goes wrong in practice
The VUS trap. "Variants of uncertain significance" — DNA changes the lab can't yet classify as benign or pathogenic — show up in 5–20% of clinical panel results. A VUS should not change clinical management; the natural history is that most get reclassified as benign over time. Patients and doctors who treat a VUS like a positive are the source of avoidable prophylactic surgery.
Results that don't show up at the right moment. A pharmacogenomic panel only helps if it surfaces at the prescribing screen. The PREPARE trial's 30% reduction in adverse drug reactions depended on a pharmacist-facing alert at dispensing time Swen et al. 2023. Without integration into the electronic health record, a CYP2C19 poor-metabolizer result sits in a chart while clopidogrel gets prescribed anyway.
The family conversation that doesn't happen. A positive cancer or cardiac result is half-useful if relatives don't get tested. Cascade testing — first-degree relatives, each with a 50% chance of carrying the same variant — is the highest-yield, lowest-cost case-finding strategy in medicine, and it only happens when the proband makes the calls. Observational series put cascade-testing rates at 30–60% in actual families.
The wrong panel. A patient with strong colorectal-cancer history who orders a BRCA-only test gets a true negative that isn't the answer to their real question. The right panel for a given family pattern is the counselor's job.
Confusing tumor with germline. A "BRCA mutation" found on biopsy of a tumor may live only in the cancer tissue and have no implications for relatives. Germline confirmation in blood or saliva is required before the family conversation begins.
Cost, logistics, and the insurance fine print
The actual numbers, since they aren't on the marketing pages:
- Consumer kits: $99 for ancestry-only (Ancestry, MyHeritage); $199–$249 for ancestry plus FDA-cleared health markers (23andMe Health+Ancestry). Sale prices common around the holidays.
- Clinical multi-gene cancer or cardiac panel: $250 to $1,000+ list price. US insurance usually covers it when NCCN family-history or personal-cancer criteria are met. Without insurance, the direct-pay labs (Color, Invitae, Ambry, GeneDx) tend to be cheaper than retail and include counseling.
- Pharmacogenomic panel: $100–$300 cash, occasionally covered by Medicare for specific drug-initiation scenarios.
- Whole-genome sequencing for healthy adults: $500 to $1,500 clinical-grade.
The piece most consumers never read about: insurance gaps. GINA covers health insurance and employment in the US GINA 2008. It does not cover life insurance, disability insurance, or long-term-care insurance. Some advisors with strong family histories recommend buying those policies before testing — the underwriting questionnaire eventually asks about known genetic findings, and a positive result on file can change the offer or the premium. UK and Canadian protections are broader; the US framework still has this hole.
Genetic counseling is what keeps the system honest. Board-certified counselors (CGC) translate the result, run the family-history math, and pre-decide what each possible answer should change. Most direct-pay test vendors include counseling in the price; the National Society of Genetic Counselors maintains a public directory, and telehealth visits are now standard.
What changes when this works
The opposite path runs through the same family. The cousin whose oncologist ordered a BRCA panel after a 38-year-old sister's breast cancer — positive, mastectomy at 41, oophorectomy at 43, watches her daughters grow into their thirties without the cancers that took the previous generation. The all-cause-mortality reduction in carriers who choose risk-reducing surgery is roughly 60% over a decade Domchek et al. 2010; the people who would have noticed her absence get to keep her instead.
The hereditary-cholesterol family that finds the variant at the index relative's autopsy, and twelve years later has six surviving relatives on a statin with normal LDL and clean cardiac CT scans. The colonoscopy room every two years for a Lynch carrier whose annual scope catches a polyp at stage 0 — the version of her week that includes a small procedure and ends with her going back to her life, instead of the version that includes a diagnosis and a year of treatment Järvinen et al. 2000.
The pharmacogenomic version is quieter. The depressed patient whose first SSRI works because the prescriber checked CYP2D6 before reaching for the prescription pad — six weeks to a working dose instead of two years of swapping. The cancer patient who keeps her hair and her appetite through chemotherapy because her DPYD status caught a metabolizer variant and the dose got cut from day one Henricks et al. 2018. The post-stent patient on ticagrelor instead of clopidogrel because the chart already knew her liver couldn't activate the prodrug, and the stent stays open.
Most of the win lands as absence — the heart attack that didn't happen, the cancer that didn't have to be treated because it was prevented, the bad drug reaction that didn't show up because the prescription was different. The PREPARE numbers are the population-scale version: 30% fewer clinically relevant adverse drug reactions across roughly 7,000 patients in twelve weeks of follow-up Swen et al. 2023. Timing varies: cancer-risk action protects you over decades; FH treatment protects you starting the year you start; pharmacogenomics pays off the next time a doctor reaches for a prescription pad. People who barely know you stop being the people at your funeral.
Adjacent territory worth knowing
Several relatives of this topic earn their own entry rather than a section here. Prenatal genetic testing during a pregnancy — non-invasive prenatal testing (NIPT), chorionic villus sampling, amniocentesis — is a different decision with different timing and stakes. Newborn screening is the state-mandated heel-prick panel in the first week of life, covering ~35 conditions where early treatment prevents disability. Somatic tumor profiling sequences cancer tissue itself to guide treatment — distinct from germline testing and managed by oncology. Polygenic risk scores beyond research settings — currently early in clinical practice, with serious ancestry-equity gaps Khera et al. 2018. Polygenic embryo screening is sold by a handful of US clinics and is ethically contested with thin outcome data. Diagnostic genetic testing for an already-symptomatic patient — a different workflow from the predictive testing this entry covered.
- — APOE e4 is among the consequential, probabilistic results clinical genetic testing can hand you.
- — Ashkenazi carrier screening is a clear example of clinical genetic testing that changes decisions.
- — BRCA is the headline example — a clinical result that, for a carrier, can cut all-cause death with preventive surgery.
- — A clinical panel can flag Lynch syndrome, which means starting colonoscopies young and repeating them far more often than usual.
- — Alpha-1 antitrypsin deficiency is an inherited cause of COPD; one test catches it, especially with early or non-smoker disease.
- — Familial hypercholesterolemia is among the high-impact variants that make a clinical gene panel worth it.
- — HFE hemochromatosis is a textbook case where a genetic test changes what you do.
- — HLA-B*57:01 is a clear instance of genetic testing actually altering what you're prescribed.
- — A clean example of a clinical panel that rewrites care: two genes that can rule celiac out for life.
- — One useful, concrete branch of genetic testing is reading the genes that decide how you handle certain drugs.
- — The alcohol-flush gene is one clear, actionable thing a genetic test can settle.
- — Inherited conditions like G6PD deficiency are the kind of high-value, act-on-it result genetic testing can give.
- — HLA-B27 is a good example of a gene test that helps only in the right clinical context.
- — Plenty of genetic tests rewrite your care; MTHFR is the one nearly every specialty society tells you to skip.
- — For a first-degree relative of someone with type 1 diabetes, a free antibody screen catches the attack years before symptoms.
Substance + claimed effects
Genetic testing covers a spectrum from direct-to-consumer (DTC) spit kits (23andMe, Ancestry, MyHeritage) to clinical testing ordered by a physician or genetic counselor: targeted single-gene tests, multi-gene cancer or cardiac panels, pharmacogenomic (PGx) panels, expanded carrier screening, and whole-exome or whole-genome sequencing. The substance is the act of obtaining and acting on germline genetic information about an asymptomatic adult. In-scope consequences: (1) identification of highly penetrant disease-risk variants (BRCA1/2, Lynch, familial hypercholesterolemia, hereditary cardiomyopathies, MODY) and the surveillance/risk-reducing surgery/medication cascades that follow; (2) reproductive carrier screening and the family-planning decisions it enables; (3) pharmacogenomic genotyping that changes medication selection or dosing; (4) psychological response to results — distress, reassurance, fatalism, or behavior change; (5) downstream burdens (cost, insurance implications under GINA 2008 and its gaps, cascade testing of relatives). Out of scope for this entry: prenatal testing of a fetus, newborn screening programs, somatic tumor profiling for cancer treatment, and ancestry-only reports without health interpretation.
Evidence by addressing question
mechanism
Three distinct test technologies, often conflated:
- Genotyping arrays (used by most DTC vendors): probe ~600,000–1,000,000 pre-selected single-nucleotide polymorphisms (SNPs). Cheap (~$0.10 per SNP equivalent), high accuracy per probed SNP, but blind to any variant not on the chip. 23andMe's FDA-cleared BRCA report checks only 3 Ashkenazi founder variants (185delAG, 5382insC, 6174delT) out of >1,000 known pathogenic BRCA1/2 variants Finn et al. 2018.
- Targeted sequencing / panels (clinical): Sanger or next-generation sequencing of specific genes end-to-end. Detects substitutions, small indels, and (with proper pipelines) large deletions/duplications. The standard for BRCA1/2, Lynch (MLH1/MSH2/MSH6/PMS2/EPCAM), LDLR/APOB/PCSK9 (FH).
- Whole-exome (WES) or whole-genome sequencing (WGS): reads coding regions or the entire genome. Increasingly used for diagnostic odysseys; emerging in screening contexts (Geisinger MyCode, UK Genomes Project).
Mechanism of clinical impact runs through three pathways: (a) actionable predispositions — a pathogenic variant raises lifetime risk of a treatable/preventable disease enough to change surveillance or trigger risk-reducing intervention; (b) reproductive risk — autosomal-recessive carrier status in both partners gives ~25% per-pregnancy risk to offspring, addressable via IVF/PGT, prenatal diagnosis, or donor gametes; (c) drug metabolism — variants in cytochrome P450 enzymes (CYP2D6, CYP2C19, CYP2C9), thiopurine S-methyltransferase (TPMT), dihydropyrimidine dehydrogenase (DPYD), and HLA loci change effective drug exposure or trigger hypersensitivity CPIC guidelines.
evidence
Highly penetrant cancer predisposition. CIMBA consortium pooled analysis of 6,036 BRCA1 and 3,820 BRCA2 carriers: cumulative breast cancer risk to age 80 is 72% (95% CI 65–79%) for BRCA1 and 69% (61–77%) for BRCA2; cumulative ovarian cancer risk is 44% (36–53%) for BRCA1 and 17% (11–25%) for BRCA2 Kuchenbaecker et al. 2017. Risk-reducing salpingo-oophorectomy in carriers is associated with hazard ratio 0.40 (0.26–0.61) for all-cause mortality and 0.21 (0.06–0.80) for breast-cancer-specific mortality Domchek et al. 2010. Risk-reducing mastectomy approaches 90%+ reduction in breast cancer incidence in carriers. USPSTF grade B recommendation: women with a personal/family history suggesting a BRCA-related cancer should receive risk assessment, and if indicated, counseling and testing USPSTF 2019.
Lynch syndrome. Lifetime colorectal cancer risk 30–70% depending on the specific MMR gene. The Finnish prospective controlled trial in 252 mutation carriers: colonoscopy every 3 years cut colorectal cancer incidence by 62% and all-cause mortality by 65% over 15 years versus no surveillance Järvinen et al. 2000. NCCN and EGAPP working group consider Lynch tier-1 (sufficient evidence to support population screening of newly diagnosed colorectal cancers and cascade testing of relatives) CDC Tier 1 Applications.
Familial hypercholesterolemia (FH). Heterozygous prevalence ~1:250; untreated heterozygotes have ~20-fold elevated coronary heart disease risk and median age of first MI in mid-40s for men, mid-50s for women Nordestgaard et al. 2013. Cascade testing of first-degree relatives identifies new cases at ~50% yield; statin therapy from early adulthood normalizes coronary risk Knowles et al. 2014. CDC tier-1 condition CDC Tier 1 Applications.
Pharmacogenomics. Multiple gene-drug pairs have actionable evidence with FDA labeling and CPIC dosing guidelines:
- HLA-B*5701 and abacavir: PREDICT-1 randomized 1,956 HIV patients to prospective screening vs standard care; screening reduced clinically suspected hypersensitivity reactions from 7.8% to 3.4% and eliminated immunologically confirmed reactions entirely (0% vs 2.7%, p<0.001) Mallal et al. 2008. Standard of care for HIV.
- DPYD and fluoropyrimidines: ~7% of Europeans carry a DPYD variant that impairs 5-FU/capecitabine clearance, causing severe (grade ≥3) toxicity in ~70% of homozygotes and ~30% of heterozygotes. Prospective Dutch multicenter trial of upfront DPYD genotyping with dose reduction in carriers: severe toxicity in variant carriers fell from a historical ~70% to 33% — a relative risk reduction of roughly 50% Henricks et al. 2018. Now standard of care across the EU.
- Pre-emptive PGx panels: The PREPARE cluster-randomized crossover trial across 7 European countries enrolled ~7,000 patients receiving any of 39 PGx-relevant drugs. A 12-gene panel with dispensing-time alerts reduced clinically relevant adverse drug reactions by 30% (OR 0.70, 95% CI 0.54–0.91) Swen et al. 2023 (PREPARE) — the first prospective evidence that broad pre-emptive PGx improves outcomes versus reactive testing.
- Other established pairs: TPMT/NUDT15 before azathioprine/6-mercaptopurine; CYP2C19 for clopidogrel (POPular Genetics trial — non-inferior with genotype-guided de-escalation); CYP2D6 for tamoxifen, codeine, tramadol; HLA-B*1502 for carbamazepine in Asian populations (FDA boxed warning); CYP2C9/VKORC1 for warfarin CPIC guidelines, Relling et al. 2020.
DTC accuracy and clinical validity. Tandy-Connor et al. analyzed 49 patient samples sent to a clinical lab after DTC genotyping flagged variants of concern: 40% of the variants reported as present in DTC raw data were false positives on Sanger confirmation, and several variants classified as "increased risk" by third-party DTC interpretation tools were re-classified as benign by the clinical lab Tandy-Connor et al. 2018. The pattern is structural: SNP arrays have low specificity for rare variants because rare-variant probes have low signal-to-noise.
Polygenic risk scores (PRS). Genome-wide PRS for coronary artery disease, atrial fibrillation, type 2 diabetes, IBD, and breast cancer can identify the top 8% of the population at risk equivalent to monogenic carriers (e.g., 3-fold elevated CAD risk for the top 8% of CAD-PRS) Khera et al. 2018. Caveats: scores derived predominantly from European-ancestry cohorts perform substantially worse in African, East Asian, and South Asian populations; clinical utility (does acting on the score improve outcomes) is still being established; no current guideline endorses population PRS screening.
Population BRCA screening. Manchanda et al. modeled population testing of all women ≥30 vs current family-history-based testing in UK/US populations: dominant in cost-effectiveness (saves money and lives), would prevent thousands of breast and ovarian cancers per million women screened, and would identify carriers missed by family-history triage (≥50% of BRCA carriers in unselected populations have no qualifying family history) Manchanda et al. 2018. Geisinger's MyCode found that 1.3% of unselected biobank participants carried a pathogenic BRCA1/2 variant, and only a minority had personal/family history that would have triggered standard testing Manickam et al. 2018.
protocol
Practical sequencing depends on the question being asked:
- Start with family history. A structured three-generation pedigree (cancers, ages at diagnosis, cardiac events, sudden deaths) is the cheapest and most informative input. NCCN and ACMG criteria for BRCA, Lynch, and FH testing are family-history driven.
- For suspected hereditary cancer: clinical multi-gene panel ordered by a genetic counselor or oncologist. Cost $250–$3,000; often covered by insurance when NCCN criteria are met. Pre-test counseling is standard of care.
- For unselected adults wanting actionable health information: a clinical multi-gene panel of CDC tier-1 conditions (BRCA1/2, Lynch genes, FH genes) plus ACMG SF v3.x secondary findings list (~80 actionable genes) is the highest-yield option. Costs $250–$600 through commercial labs (Color, Invitae, Ambry).
- For DTC: useful for ancestry and a small set of FDA-cleared health reports (3 Ashkenazi BRCA variants, APOE, hereditary hemochromatosis HFE, MUTYH, factor V Leiden). A negative DTC BRCA report does not rule out BRCA carrier status. Raw data can be exported and run through third-party interpretation services (Promethease) but these are not clinically validated.
- For pharmacogenomics: targeted PGx panel ($100–$300) covering at minimum CYP2D6, CYP2C19, CYP2C9, VKORC1, TPMT, DPYD, HLA-B*5701, HLA-B*1502, SLCO1B1. Most valuable before starting a PGx-relevant drug; PREPARE shows benefit when results are pre-emptively available PREPARE 2023.
- For carrier screening: ACOG recommends offering expanded carrier screening (100+ conditions) to all couples planning pregnancy; tested partner first, partner of carrier second. Cost $100–$400 per partner.
contraindications
Few absolute contraindications; several relative ones:
- Life/disability/long-term care insurance not purchased yet: GINA prohibits health-insurance and employment discrimination based on genetic information, but does not cover life, disability, or long-term care insurance GINA 2008. Some applicants choose to purchase these policies before testing.
- Active untreated severe depression or suicidality: receiving a high-penetrance positive result (e.g., Huntington's disease, BRCA1) without psychological support can be destabilizing; standard practice for adult-onset incurable conditions like Huntington's includes a multi-visit pre-test counseling protocol.
- Inability to act on results: testing for a condition with no actionable management (e.g., APOE for Alzheimer's risk, in current practice) without explicit informed consent for that limitation.
- Predictive testing in minors: not recommended for adult-onset conditions; the child cannot consent and the result forecloses their future autonomy to choose not to know. ACMG and ASHG guidance limit predictive testing in children to conditions where childhood/adolescent action is needed.
misconceptions
- "23andMe checks my BRCA risk." It checks 3 of >1,000 pathogenic BRCA variants — the three most common in Ashkenazi Jewish populations. A negative report rules out almost nothing in a non-Ashkenazi person Finn et al. 2018.
- "Raw data from DTC is the same as a genetic test." SNP-array raw data has high false-positive rates for rare variants; 40% of DTC-flagged variants were false positives on clinical confirmation in one analysis Tandy-Connor et al. 2018.
- "AncestryDNA tells me my health risks." Ancestry.com does not return health reports (the platform may report wellness/traits in some plans; it does not return clinical health predictions in standard subscriptions).
- "A negative test means I don't have the disease." Negative tests mean no detected pathogenic variant in the genes tested. With limited panels, that's a small ruling-out. Even with a comprehensive panel, polygenic and environmental risk remains.
- "My test will hurt my health insurance." In the US, GINA prohibits this. The legitimate concern is life/disability/LTC insurance, which GINA doesn't cover.
- "Genetic testing causes psychological harm." Multiple RCTs and prospective cohorts (REVEAL, Scripps, BabySeq) show no measurable long-term distress on group averages, even for APOE disclosure for Alzheimer's risk Green et al. 2009 (REVEAL), Bloss et al. 2011. Individual responses vary; vulnerability screening matters.
- "PGx is for psychiatry only." Oncology (DPYD, TPMT, UGT1A1), HIV (HLA-B*5701), cardiology (CYP2C19/clopidogrel, CYP2C9/VKORC1/warfarin), neurology (HLA-B*1502/carbamazepine), and rheumatology (TPMT/azathioprine) all have actionable PGx markers.
failure-modes
- Variants of uncertain significance (VUS): 5–20% of clinical panel results return at least one VUS — a variant that's neither benign nor confirmed pathogenic. VUS should not change clinical management; reclassification over time is common (most VUS resolve as benign). Patients and clinicians who treat VUS as positives drive unnecessary surgery.
- Acting on DTC results without clinical confirmation: cases of prophylactic mastectomy ordered on the basis of unconfirmed DTC BRCA reports are documented in the literature. Standard of care requires clinical confirmation in a CAP/CLIA-certified lab.
- Not telling family. A positive result in a proband makes first-degree relatives 50% likely to carry. Cascade testing is the highest-yield, lowest-cost case-finding strategy in medicine — but only happens when the proband communicates.
- Pharmacogenomics ignored in the EHR. Results obtained pre-emptively only help if they surface at prescribing time. Without clinical decision support integration, a CYP2C19 poor-metabolizer status sits in a chart while clopidogrel is started.
- Ordering wrong panel. A patient with strong colorectal history gets a BRCA-only panel; the relevant Lynch genes go untested.
- Confusing germline and somatic results. Tumor-only sequencing identifies somatic variants in the cancer; these may or may not be germline. A "BRCA mutation" found in tumor tissue requires germline confirmation before family implications follow.
- Ancestry bias in interpretation. Variant databases (ClinVar) are heavily European-ancestry-weighted; African and Asian patients have higher VUS rates and a small history of misclassified pathogenic variants that turned out to be benign-in-population (notably the HCM literature in Black patients).
practicalities
- Cost. DTC: $99 (Ancestry standard) to $199–$249 (23andMe Health+Ancestry). Clinical multi-gene cancer panel: $250 (Color/Invitae cash) to $3,000+ list price; insurance often covers when NCCN criteria met. PGx panel: $100–$300 cash, sometimes Medicare-covered. Whole-genome sequencing: $500–$1,500 clinical-grade.
- Where to get it. DTC online. Clinical via primary care referral to genetic counseling, oncology, or directly via patient-pay services (Color, Genome Medical, Invitae) that include included counseling.
- Counseling. Pre- and post-test counseling by a board-certified genetic counselor (CGC) is the standard of care for predictive testing. NSGC directory; many counselors now offer telehealth.
- Insurance. GINA covers health insurance and employment in the US. Life, disability, and long-term care insurance are NOT covered; some advisors recommend purchasing these before testing if a strong family history makes a positive result likely GINA 2008. UK has a moratorium on most insurance use; Canada's GNA 2017 covers broader protections.
- Storage and reuse. Raw genotype data can be downloaded from major DTC vendors and uploaded to free third-party interpretation (Promethease) — useful in principle, unreliable for rare variants.
stakes
Population-level: 1–2% of unselected adults carry an actionable pathogenic variant in a CDC tier-1 condition (BRCA1/2, Lynch genes, FH genes); ≥50% of those carriers have no family history that would trigger standard testing Manickam et al. 2018, Manchanda et al. 2018. The status quo is not "no problem" — it's "the problem is invisible until the cancer or the MI shows up." Untreated BRCA1 carriers face 72% lifetime breast cancer risk Kuchenbaecker et al. 2017; untreated FH heterozygotes have median first MI in their 40s Nordestgaard et al. 2013. The felt experience: a 45-year-old woman diagnosed with ovarian cancer whose mother had breast cancer at 50 — a positive germline BRCA result then is too late to prevent either cancer; cascade testing of sisters and daughters can still save them. The pharmacogenomic version: a patient on clopidogrel after a stent who is a CYP2C19 poor metabolizer has an effectively unblocked platelet system and an elevated stent-thrombosis risk that was knowable before the prescription.
payoff
For the ~1–2% with actionable variants: a different life trajectory. BRCA carriers who choose risk-reducing salpingo-oophorectomy cut all-cause mortality by ~60% Domchek et al. 2010. Lynch carriers with surveillance colonoscopy avoid roughly two-thirds of the colorectal cancers and two-thirds of the mortality they would otherwise face Järvinen et al. 2000. FH carriers identified and statin-treated from young adulthood have near-normal coronary risk. The cascade effect is the multiplier: every proband identifies an average of 3–4 relatives at 50% risk, half of whom test positive; the same intervention then prevents their disease. For the 98% without actionable variants, the payoff is calibrated reassurance (knowing the family history was sporadic, not heritable) and, with PGx, a pre-stored record that prevents the wrong drug at the wrong dose decades later. The PREPARE trial timeline: 30% fewer clinically relevant adverse drug reactions over 12 weeks of follow-up PREPARE 2023.
out-of-scope
Adjacent topics worth flagging for forward links: prenatal genetic testing (NIPT, CVS, amniocentesis — a different decision with different timing); newborn screening (state-mandated; ~35 conditions, growing); somatic tumor profiling (treatment-selection sequencing of cancer tissue); direct ancestry-only testing; polygenic embryo screening (currently sold by a handful of US clinics, ethics highly contested); nutrigenomics (largely unvalidated in current consumer form).
The credibility range
Optimist case
Germline genetic testing is one of the few interventions in modern medicine where the evidence is unambiguous for a sizeable minority and the technology is cheap and one-shot. The CDC tier-1 conditions (BRCA, Lynch, FH) collectively affect ~2 million Americans whose disease is largely preventable if identified. Family-history triage misses more than half of carriers Manchanda et al. 2018, so population-level screening is dominant in cost-effectiveness modeling for women ≥30. Pharmacogenomics has crossed from theoretical to practice-changing: PREPARE showed a 30% absolute reduction in clinically relevant adverse drug reactions with pre-emptive panel testing PREPARE 2023, validating decades of CPIC infrastructure. DTC at the low end is an inexpensive on-ramp for a fraction of users to take family history seriously and seek confirmatory clinical testing. The downside data — psychological distress, family disruption — has failed to replicate across REVEAL, Scripps, MedSeq, and BabySeq Green et al. 2009, Bloss et al. 2011, Vassy et al. 2017. The age of universal-screen-where-cost-effective is here; the friction is implementation, not science.
Skeptic case
Predictive genetic testing in unselected adults still has overdiagnosis, overtreatment, and inequity problems. VUS rates of 5–20% generate anxiety and occasional unnecessary surgery in patients and clinicians who misread them as positives. DTC SNP arrays produce false positives at rates that make their "raw data" misleading Tandy-Connor et al. 2018; the documented cases of prophylactic mastectomy on DTC reports without clinical confirmation are clinical-tragedy territory. Polygenic risk scores, the most-hyped new modality, perform substantially worse in non-European populations and lack RCT evidence that acting on them changes outcomes. The Manchanda population-screening model assumes high-quality follow-through (genetic counseling, risk-reducing surgery uptake) that may not generalize from research settings to real clinics. Insurance protections (GINA) have meaningful gaps in life/disability/LTC coverage that few patients understand at the point of consent. Pharmacogenomics' practical impact is bottlenecked on EHR integration; results in a chart that don't surface at prescribing time don't help. And cascade testing — the multiplier the optimist case rests on — happens far less than the model predicts in actual families, with proband-relative communication rates of 30–60% in observational series.
Author's call
For the high-penetrance, actionable variants (CDC tier-1: BRCA, Lynch, FH) and for established pharmacogenomic gene-drug pairs, the evidence is strong enough that genetic testing — done in a clinical lab, with counseling, and acted on — is one of the higher-leverage preventive interventions available. The hard part is delivery, not the underlying science. DTC is a much weaker tool: useful for prompting clinical investigation in some users, misleading in others, and not a substitute for a clinical panel when family history or symptoms warrant testing. The headline framing for the reader: genetic testing is a decide, not a do — the version that pays off needs a clinician in the loop and a pre-formed plan for what to do with each possible result. The version that doesn't ($99 spit kit, results filed away, no follow-up) mostly produces ancestry charts and trait reports of negligible health impact.
Stakeholder + incentive map
- DTC vendors (23andMe, Ancestry, MyHeritage): direct revenue plus database value (23andMe's deal with GSK monetized aggregated genetic data for drug discovery). Incentive: maximize volume; under-emphasize panel limitations.
- Clinical genetic labs (Invitae, Color/Natera, Ambry, GeneDx, Myriad): per-test reimbursement, increasingly competing on price and counseling inclusion. Color/Invitae historically pushed direct-pay $250 panels; Myriad historically defended high-list-price BRACAnalysis until the Supreme Court invalidated the BRCA patents (2013).
- Professional societies (ACMG, ASHG, NSGC, NCCN): conservative, evidence-driven; ACMG SF v3 secondary-findings list is the canonical "what's worth reporting incidentally" reference.
- Genetic counselors (NSGC): gatekeepers and translators; advocate for pre/post-test counseling. Incentive to preserve the role against direct-to-patient ordering models.
- Insurers: cost-containment pressure on panel breadth (tier-1 yes, broad panels resisted historically); growing coverage of NCCN-criteria-meeting tests.
- Public health (CDC OPHG, EGAPP, WHO): defines tier-1 conditions; pushes cascade testing infrastructure.
- Skeptics / counter-incentive: bioethics community (overdiagnosis, geneticization, autonomy concerns); patient advocacy groups split between "more access" and "more caution"; some primary-care physicians under-utilize for time and training reasons.
Population variability
- Ancestry. Variant databases are 80%+ European-ancestry-derived; non-European patients have 2–3× higher VUS rates and PRS that perform poorly. Specific founder variants (Ashkenazi BRCA, French Canadian Lynch, Afrikaner FH) make targeted founder-variant testing very cost-effective in those populations.
- Family history strength. The strongest single predictor of yield. A patient with multiple early-onset cancers in the family has 10–30% pretest probability of finding a pathogenic variant; an unselected patient has ~1–2%.
- Age. Predictive testing for adult-onset conditions is appropriate from adulthood (cancer panels typically 18+, or earlier if family pattern is very early-onset). PGx is age-agnostic but most valuable before chronic medication courses begin.
- Sex. BRCA implications differ (women: breast/ovarian risk + risk-reducing surgery options; men: prostate, breast, pancreatic risk + cascade implications for daughters). FH and Lynch are sex-agnostic. Carrier screening matters per reproductive partner pair.
- Reproductive plans. Carrier screening only changes decisions if children are an option; the value drops to near zero post-menopause or post-vasectomy.
- Existing diagnoses. A patient on warfarin or starting clopidogrel after a stent benefits from PGx far more than a healthy young adult; a patient with strong cancer family history benefits from BRCA/Lynch testing far more than the average reader.
Knowledge gaps
- Polygenic risk scores in non-European ancestries, and whether acting on PRS (intensified screening, statins, lifestyle counseling) actually improves outcomes vs standard-of-care risk stratification. RCTs in progress.
- Pre-emptive PGx panels in primary care (versus the specialty/hospital setting of PREPARE) and in US-payor environments. Effect-size durability over years rather than 12 weeks.
- Whole-genome sequencing as a primary-care tool: BabySeq, MedSeq pilots are positive but not powered for outcomes. What's the right secondary-findings list, what's the cost-effectiveness threshold?
- Real-world cascade testing rates: most data are research-cohort. What happens when it leaves academic centers?
- Long-term psychological data: most psychological-impact studies follow patients 6–12 months. Decade-out effects of carrying a known BRCA or Huntington's status are thinly studied.
- Polygenic embryo screening: clinically offered, ethically contested, near-zero outcomes evidence.
- Insurance landscape: GINA gaps for life/disability/LTC are mostly unaddressed legislatively in the US; how this evolves affects test uptake.
Scope held. Brief named four consequence areas — disease-risk awareness, family planning, pharmacogenomics, psychological response. All four covered. Disease risk gets the heaviest treatment (BRCA, Lynch, FH) because the actionable evidence is concentrated there. Pharmacogenomics covered through PREPARE, PREDICT-1, DPYD/Henricks. Family planning surfaced through expanded carrier screening in protocol and out-of-scope's prenatal pointer. Psychological response handled in misconceptions via REVEAL, Bloss, MedSeq.
Action choice. decide over do or test. The high-impact version requires a clinician in the loop both for panel selection and for the post-result intervention; framing this as a simple do would push readers toward unaccompanied DTC kits, which the evidence specifically argues against.
Longevity score = 4, not 5. Hard call. The effect for the 1–2% with actionable variants is large enough for 5 in isolation (BRCA RRSO HR 0.40; Lynch surveillance 65% mortality reduction). But the substance is "getting tested," and ~98% of testees get no longevity benefit because they don't carry an actionable variant. Holistic score lands at 4 — among the higher-leverage preventive tools, but not population-defining the way (say) smoking cessation would be.
Mood score = 1. Wanted to score 0 — group-average studies (REVEAL, Bloss, MedSeq) show no lasting distress. Held at 1 to acknowledge the modest but real reassurance/agency signal for those who test and act, and the family-planning relief in carrier screening contexts. Defensible as 0; chose to honor the asymmetry rather than zero it out.
Citation choices. Two refs ended up with mismatched author/year mnemonics from initial typing errors: RelinkEtAl2024 actually points to Relling et al. 2020 (CPIC 10-year retrospective); RelogioEtAl2024 actually points to Manickam et al. 2018 (Geisinger biobank). The DOIs and underlying records are correct; the refs are just badly named. Flag for cleanup if the library is ever consolidated.
Excluded with intent. Newborn screening (state-mandated, different audience), prenatal NIPT (different decision, time-sensitive, deserves its own entry), polygenic embryo screening (ethics contested, evidence thin), nutrigenomics (largely unvalidated), and diagnostic testing for symptomatic patients (workup pathway, not preventive). All surfaced in out-of-scope.
Future links. Once written, link to: brca-screening, lynch-syndrome, familial-hypercholesterolemia, pharmacogenomic-panels, expanded-carrier-screening, prenatal-genetic-testing, newborn-screening, polygenic-risk-scores.
Separate-entry candidates surfaced during writing. Cascade testing as its own systems-level intervention; CDC tier-1 conditions overview; APOE/Alzheimer's risk disclosure (REVEAL has enough material to anchor an entry on whether to test for things you can't treat).
Editorial tension noted. The Manchanda 2018 cost-effectiveness model for population BRCA screening would, if fully adopted, push action toward do for women ≥30. The article lands conservatively on decide because that's the current USPSTF position and because the model's assumptions about follow-through rates may not generalize. If population screening becomes guideline-endorsed in the US, this entry should be re-scoped.
Voice check. Pulled back twice from inline trial-design jargon (named arms, hazard ratios in body text) into science callouts and parenthetical anchors. The "felt experience leads, citations hedge" rule was the hardest one to honor in the evidence section, where the numbers are the argument; resolved by anchoring each headline number in a person ("the woman whose mother and aunt…", "a 47-year-old man in a parking lot").
Genetic Testing
Spit in a tube or get a blood draw. Two short appointments if you want it done right. No ongoing effort.
For the 1 in 50 adults carrying a high-risk cancer or heart-disease variant, knowing it means surveillance and treatment that cut all-cause death by 60% or more.
$100 for a spit kit, $250–$1,000 for a clinical panel, often covered by insurance with a family history. One-time spend.
Settled science for the high-impact variants — BRCA, Lynch, familial hypercholesterolemia, and a list of drug-metabolism genes. The fine print on consumer kits is murkier.
If you're about to start a drug like clopidogrel, an SSRI, codeine, or a chemo, a pharmacogenomic panel can flag the ones your liver handles badly before the side effects do.
For most people, a negative test is calm reassurance; for a positive, the agency of acting often offsets the initial shock. Studies find no lasting distress on average.