DNA Test: What Can It Really Predict About Your Disease Risk?

Can a DNA test really tell you which diseases lie ahead? For a handful of rare mutations, such as BRCA1 or those behind familial hypercholesterolemia, the answer is yes, with useful precision. For common diseases (heart attack, diabetes, Alzheimer’s), the result remains a modest probability. In France, this kind of analysis is done on medical prescription, in an accredited laboratory, and that framework is what gives it its value.

Definition: what is a health DNA test?

A health genetic test reads all or part of your genetic sequence to spot variants linked to a disease. It starts from a simple blood or saliva sample. The laboratory then analyzes a few targeted genes, the exome (the coding part, about 1 to 2% of the genome) or the whole genome.

The term covers two very different uses. A diagnostic genetic test confirms a disease already suspected in a person who has symptoms. A predictive genetic test, by contrast, is aimed at a healthy person and estimates a future risk. It is this second use that raises the highest expectations, and the most misunderstandings.

You also need to separate the DNA you inherited, fixed for life, from the way your genes are expressed. That expression varies with age and lifestyle: this is the field of epigenetics and its levers against aging. A standard genetic test does not measure these epigenetic marks.

Mechanism: what a DNA test actually reads

A DNA test does not detect diseases; it detects variants whose effect on risk ranges from massive to almost nil. Genetics distinguishes three families of variants, and the value of a result depends entirely on which one it belongs to:

• high-penetrance monogenic mutations: a single alteration (BRCA1, BRCA2, Lynch syndrome genes, the LDLR gene of familial hypercholesterolemia) multiplies the risk substantially;
• susceptibility variants with an intermediate effect, the textbook example being the ε4 allele of the APOE gene for Alzheimer’s disease;
• thousands of common variants with a tiny effect, added up in a polygenic risk score;
• pharmacogenetic variants, which change the response to certain drugs rather than the risk of disease.

A rare, powerful mutation behaves like an on/off switch. A polygenic score is more like a slider that moves a few notches. The table below summarizes these differences with figures from the main studies.

What the science says about predicting disease with a DNA test

The studies converge: a DNA test predicts monogenic diseases well and common multifactorial diseases poorly. The figures below come from the largest published cohorts and meta-analyses.

Rare mutations: a solid prediction

Following 9,856 BRCA mutation carriers, Kuchenbaecker and colleagues (JAMA, 2017) put the cumulative risk of breast cancer by age 80 at 72% for BRCA1 and 69% for BRCA2. The risk of ovarian cancer reaches 44% and 17%. In the general population, about one woman in eight develops breast cancer during her lifetime.

Familial hypercholesterolemia illustrates the other benefit of genetic screening. In a US health system, sequencing 50,726 people found one carrier in 256 (Abul-Husn, Science, 2016). Only 24% of these carriers met the clinical criteria for the diagnosis. Genetic testing therefore revealed at-risk people whom a lipid panel alone had missed.

APOE ε4: a risk factor, not a verdict

Farrer’s meta-analysis (JAMA, 1997), covering 5,930 patients and 8,607 controls, estimated that the risk of Alzheimer’s was multiplied by 3.2 with one copy of APOE ε4 and by 14.9 with two copies, in people of European ancestry. Reference sources usually retain a factor of 3 to 4 for one copy, and MedlinePlus stresses that carriers inherit a higher risk, not the disease itself. Yet many carriers never develop the disease, and many patients carry no copy at all.

Polygenic scores: promises to put in perspective

In 2018, Khera’s team (Nature Genetics) showed that a polygenic score identifies 8% of the population with at least a tripled coronary risk. This result launched the idea of genetic screening for everyone. Wald and Old (Genetics in Medicine, 2019) pointed out that a factor associated with a disease is not necessarily a good tool for individual prediction.

The analysis by Hingorani (BMJ Medicine, 2023) settled the question across 926 scores covering 310 diseases. For a 5% false-positive rate, these scores detect a median of only 11% of future patients. The 2.5% of people with the highest coronary score account for only 7% of heart attack cases. The authors conclude that performance is poor and the promises disproportionate.

Put another way: even with the least favorable score, about 89 people in 100 will not have a heart attack within ten years. A genetic test shifts the probability; it does not turn it into a certainty.

Longevity: a weak genetic inheritance

The family trees of hundreds of millions of people analyzed by Ruby and colleagues (Genetics, 2018) place the real heritability of lifespan well below 10%. Earlier estimates of 15 to 30% were inflated by people choosing partners with similar lifestyles. No genetic test can therefore predict your longevity with any precision.

In practice: when a DNA test is useful, and when it is not

Genetic testing provides useful information when a family history points to a specific mutation. Several breast or ovarian cancers among female relatives, colorectal cancer before 50 or very high LDL cholesterol from childhood are examples. The doctor then refers the patient to a genetics consultation, which oversees the prescription and the disclosure of the result.

French law strictly regulates this test. The analysis is only allowed on medical prescription, by court order or within a research project (the general principles of genetic testing are explained by MedlinePlus). Ordering a genetic test online for medical purposes exposes a person in France to a fine of 3,750 euros (article 226-28-1 of the French Penal Code). Yet in 2019 the French research institute Inserm estimated that 100,000 to 200,000 French people used them every year.

Kit reliability is the most underestimated issue. The Ambry Genetics laboratory checked 49 samples whose consumer genetic raw data flagged a worrying variant. In 40% of cases, the variant did not exist (Tandy-Connor, Genetics in Medicine, 2018). Conversely, a chip that tests only three BRCA variants can falsely reassure someone who carries a different mutation.

For Alzheimer’s disease, knowing your APOE status does not change any medical decision today. The prevention levers remain the same for everyone, as our article on Alzheimer’s prevention according to science explains. For cardiovascular risk, a standard lipid panel remains the first tool, and our feature on cholesterol and longevity explains how to read it. All the mechanisms that damage DNA over the years are gathered on our genomic instability pillar page.

Advice: how to approach a DNA test in an informed way

A useful genetic test starts with a precise medical question, not with curiosity. Before taking any step, these pointers drawn from the studies help you decide:

• map your family’s medical history over three generations, with the age at diagnosis: it often predicts better than a polygenic score;
• talk to your primary care doctor, who will judge whether a genetics consultation is warranted;
• make no medical decision based on an online kit without confirmation by an accredited laboratory;
• anticipate the psychological and family impact of a result, because it also concerns your brothers, sisters and children;
• keep in mind that a reassuring result does not cancel the screenings recommended for your age.

Your lifestyle weighs at least as much as your genes. In four cohorts totaling 55,685 participants, Khera and colleagues (NEJM, 2016) observed 46% fewer coronary events in people at high genetic risk who had a favorable lifestyle. In the ARIC cohort, the 10-year incidence fell from 10.7% to 5.1%. The criteria were simple: not smoking, avoiding obesity, exercising every week and eating a balanced diet.

The same holds for dementia. Among 196,383 adults over 60 in the UK Biobank, a high genetic risk combined with a favorable lifestyle went with 1.13% dementia over 8 years, against 1.78% with an unfavorable lifestyle (Lourida, JAMA, 2019). These are observational associations, not proof of causation.

Knowing your genetic risk is not enough to change your habits, either. The meta-analysis by Hollands (BMJ, 2016), covering 18 studies, found no effect of disclosing a DNA-based risk on quitting smoking, diet or physical activity. Prevention habits are worth adopting without waiting for a genetic test. To follow these topics, our weekly health newsletter summarizes the studies that matter.

Frequently asked questions about DNA tests and disease risk

Can a DNA test predict whether I will get cancer?

Genetic testing can detect a high-risk mutation, such as BRCA1 or BRCA2, which raises the risk of breast cancer to 72% and 69% by age 80 (Kuchenbaecker, 2017). These mutations are rare. For most cancers, polygenic scores flag only a small share of future patients. A negative result never exempts you from recommended screenings.

Is it legal to take a DNA test online in France?

No, not for medical purposes. In France, genetic testing is only allowed on medical prescription, by court order or within regulated research. Ordering a kit from abroad exposes you to a fine of 3,750 euros under article 226-28-1 of the French Penal Code. Inserm estimated in 2019 that 100,000 to 200,000 French people used them every year.

What is the difference between a medical DNA test and a consumer kit?

A medical genetic test fully sequences the targeted genes in an accredited laboratory, with a genetics consultation before and after. A consumer kit uses a chip that reads only certain positions in the genome. A 2018 study showed that 40% of the variants reported in the raw data of these kits were false positives.

Does the APOE4 gene mean I will get Alzheimer’s?

No. One copy of APOE ε4 multiplies the risk of Alzheimer’s by about 3, two copies by about 15, according to Farrer’s meta-analysis (1997). Many carriers never develop the disease. No guideline justifies this test in a person without symptoms, because the result does not change prevention, which remains the same for everyone.

Can a high genetic risk be offset?

Partly, according to observational studies. In people with the highest genetic coronary risk, a favorable lifestyle is associated with 46% fewer heart attacks (Khera, NEJM, 2016, 55,685 participants). The same profile is associated with less dementia in the UK Biobank. Not smoking, staying active and eating a balanced diet remain the best documented levers.