What does your DNA really reveal about your health?

It sounded enticing: For just $99, I could spit into a tube, mail it off to a company called 23andMe, and, six to eight weeks later, I’d receive a report explaining what my DNA reveals about my risk for 120 diseases — everything from breast cancer to gout to sudden cardiac arrest.

“Knowing how your genes may impact your health can help you plan for the future and personalize your health care with your doctor,” the 23andMe Web site declares. “You’ll have access to navigational tools that enable you to explore your genome and discover a whole new world of you.”

Based in Mountain View, Calif., 23andMe is just one of several companies that sell genetic tests directly to consumers. GenePlanet, a company based in Slovenia, markets a test that costs 399 euros, or about $518, and claims to tell you the best diet for your genotype, and Genetic Testing Laboratories promises to disclose “your predisposition for cardiovascular conditions, cancers, immune system [issues], general health issues and much more” through a test that sells for $285.

While these claims may seem outlandish, most have at least a snippet of real science behind them. The tests look at a type of genetic variation called a single nucleotide polymorphism, or SNP (pronounced “snip”). Your DNA is made up of four kinds of nucleotides — adenine, cytosine, guanine and thymine (A, C, G and T) — and a SNP is a single alteration to one of these nucleotides that’s found in at least 1 percent of the population. For example, if most people have the sequence CAGGCTG at one site on the genome, those with a certain SNP might have TAGGCTG.

The notion that these tests can help you calculate your risk of disease are based on studies that compare SNPs in people with a particular condition to the SNPs of those without the disease. If a particular SNP is more common among people who have the condition or trait, this suggests that the condition and the variation may be related, but it’s not proof of a cause-and-effect relationship, says David Kaufman, director of research and statistics at the Genetics and Public Policy Center at Johns Hopkins University.

“It doesn’t mean that if you have the SNP, you’re going to get the disease and if you don’t, you’re not,” he says. Even if you do have a SNP associated with a disease, your increased risk is usually small — on the order of 10 or 20 percent more than it would be without the SNP.

While it’s clear that many diseases do have a genetic component, very few medical conditions come down to a single gene or to genetics alone, says Jeffrey Murray, a geneticist at the University of Iowa School of Medicine and president of the American Society of Human Genetics.

Identical twins, who share the same DNA, rarely end up with exactly the same medical conditions, and that tells us that genes alone cannot predict a person’s medical future, Murray says. “What you look like is almost 100 percent genetic, but what you’re going to get isn’t. There are lots of other things — random chance, environmental exposure and all kinds of stuff that we can’t control.”

If that’s the case, why get tested?

“You’re more prepared for things. It enables you to prioritize your health-care activities,” says Joanna Mountain, 23andMe’s senior director of research. “For instance, if you know you’re at higher-than-average risk for an eye disease called macular degeneration, it might prompt you to get your eyes examined more regularly. That’s where we see the value.”

She says that customers often tell her stories about having medical conditions that went undiagnosed until they found out that they had a genetic propensity for the disease. The tests can’t provide final answers, but they can help people make more-informed decisions about screening, she says.

Some companies offer another kind of test — genome sequencing, to look for specific genes, such as BRCA1 and 2, known to confer a significant risk of developing breast and ovarian cancer. The results can help people work with their doctors to develop a strategy for prevention and detection.

For an extra $75, Genetic Testing Laboratories will throw in a consult with a genetic counselor, but most companies offer only a written report. Regardless of where you get tested, you should consult with a doctor or genetic counselor to help you interpret the results and figure out their relevance to you, says Jeremy Lazarus, a psychiatrist and a board member of the American Medical Association.

Some experts worry that even with genetic counseling, people will over-interpret their test results, concluding that they can skip the sunscreen if they’re at low risk of skin cancer, for instance, or deciding that they’re doomed if the tests suggest they’re at heightened risk for heart disease. Murray cautions that the results of these genetic tests can only tell you so much: “These risks are evolving; none of these SNPs give a final answer on any disease.” He says that an apparent link between a SNP and a particular disease that turns up in one study may fall apart in a later one.

GenePlanet’s “nutrigenic” analysis promises to help people learn about their metabolism and discover their nutrient needs. But a Federal Trade Commission consumer bulletin warns, “The FDA and [the Centers for Disease Control and Prevention] say they know of no valid scientific studies showing that genetic tests can be used safely or effectively to recommend nutritional choices.”

Matjaz Petrovic, sales manager at GenePlanet, says, “We only include analyses that have the appropriate scientific base” and adds that “our company would never promote or position any of our products as a substitute for a traditional health-care evaluation. We rather see our genetic tests as a complement to traditional health care.”

Meanwhile, a study by researchers at the Cleveland Clinic’s Genomic Medicine Institute suggested that genetic tests may be less helpful than traditional means of identifying disease risks. The study compared two methods for assessing an individual’s risk for breast, prostate and colon cancer. One strategy used the person’s family history to estimate risk for these cancers; the other relied on genomic screening.

The results were quite different. The traditional approach suggested that 22 of the study’s 44 participants had an above-average risk of one of the three cancers, while the genomic test detected only one case of heightened risk. The study’s authors concluded that family history remains the standard for evaluating an individual’s cancer risk.

Where genetic testing shows exciting promise is in predicting a person’s response to certain drugs, Murray says. For example, 23andMe’s test includes SNPs that studies have associated with greater or lesser response to or side effects from more than 20 drugs, including antidepressants, statins and blood thinners.

“These are meant to inform the physician in terms of what dosage or even which drugs might be recommended,” Mountain says. Like SNPs linked to diseases, these markers can’t give conclusive answers, but they can provide additional information for making treatment decisions.

After some research, I decided not to send my DNA for testing. I wasn’t convinced I’d learn anything useful, and I balked when Mountain told me that even if I closed my 23andMe account, my DNA would remain in the company’s database indefinitely.

Not everyone shares my reluctance. Mountain says that more than 200,000 people have been tested through 23andMe, and many of them share their test results on social media or on the company’s user forum. In the age of information, one person’s privacy concern is another person’s tweet.