On paper, Dr. Robert C. Green should be worried. An analysis of his DNA flagged a rare mutation in a gene linked to a condition that causes facial deformities at birth. But Green only has to look in the mirror to know that he does not have the disorder, Treacher Collins syndrome.
“Most likely this is not a meaningful mutation,” the Brigham and Women’s Hospital geneticist said, pulling up alarming photos on his computer screen of people with the condition. “I know this, but imagine if you’re a pregnant woman and someone reported that mutation out to you about your baby. Can you imagine?”
Green’s experience shines a light on an important truth about DNA: The facts about your genes are not necessarily facts about you. That truth has gotten lost in the hype around genome-sequencing technology.
The public tends to see DNA as holding almost-mystical power to inform us about what treatments to take or what diseases to guard against, and this belief, combined with falling prices for gene sequencing, has driven a surge in people having their genomes analyzed. But in reality, interpreting a healthy person’s DNA with the current tools and understanding of human genetics is tricky.
Now researchers and private companies are attacking the problem, by refining the technology for analyzing DNA and building curated databases that will provide a reliable reference for understanding the connections between rare mutations and disease.
Much of the underlying medical research used to make these connections has limitations, and that can lead to confusion. If the genome is the book of life, that same book could tell a different story, depending on which gene-testing laboratory is trying to make sense of the words.
Every person carries millions of alterations, called genetic variants, in the billions of “letters” that make up the genome. Think of these alterations, or variations, as typos.
The problem is that there’s no agreement on which ones are important to human health, which are damaging or contribute to a disease, and which are basically harmless. But the National Institutes of Health announced last year that it was putting $25 million toward the development of more authoritative resources to help physicians make sense of the millions of variants in their patients’ DNA.
In the journal Nature last month, a team of scientists reported on solutions proposed during an NIH workshop to the problem of figuring out what rare genetic variations mean, and warned that if it is not tackled head on, patients may be given incorrect treatments or advice by doctors.
“There’s an expectation that it is black and white, that, oh, if you just read your genes, you will determine your future, and if you have this gene or this variant it will tell you what’s going to happen to you,” said Heidi Rehm, chief laboratory director of the Partners HealthCare Laboratory for Molecular Medicine, and a leader in improving the tools used to analyze genomes. The genome is less like a dictionary in which every genetic variant is linked definitively to a disease, and much like other kinds of medical tests that are weighed in combination with a person’s symptoms and their family history.
A gene mutation that has been determined to be the root of a lifetime of medical woes for one person might be far less predictive for another person. For example, much-publicized mutations in the BRCA1 and BRCA2 genes that were initially linked to a very high risk of breast cancer at an early age have turned out to be a less clear indicator if they are detected in women without a family or personal history of disease. Those women may develop cancer later in life, or not at all.
Solving the problem requires a broader understanding of where the current limitations come from, and a large piece of that is how medical research has worked so far. The medical and scientific literature is filled with studies that can’t be used to interpret the genomes of the general population because scientists have often identified potentially disease-causing variations by studying the sickest people.
That is a productive way to study disease, but not the broader population of people who have the genetic variation.
Last year, for example, a team of scientists from the Broad Institute in Cambridge sequenced seven genes from two broad groups of people, searching for variants that previous studies implicated in a rare, inherited form of diabetes that strikes early in life. They reported in the journal Nature Medicine that 1.5 percent of 1,541 participants in the Framingham Heart Study and 0.5 percent of 1,691 participants in the Jackson Heart Study carried a supposedly disease-causing variant. Almost none had any symptoms of the rare form of diabetes that typically strikes before people turn 25, despite years of close monitoring.
Those puzzling results are a consequence of how the original studies identifying the disease-causing genes were done. Dr. David Altshuler, a founding member of the Broad Institute who led the study, compares the way genetic studies have traditionally been done to studying the patients at a lung cancer clinic to estimate how smoking increases the risk of lung cancer.
“I wouldn’t have included in my study people with smoking history who didn’t get lung cancer, and that’s exactly what we’re doing in the history of genetics,” Altshuler said. “You will get the wrong answer, in a very predictable way, if you take a set of data collected in individuals with extreme, rare disease and take a person who doesn’t have any disease and try to extrapolate what their risk would be based on people who were selected precisely because of their rare disease.”
Add to that the variability in how different laboratories analyze genomes and the situation grows more complicated. In 2012, Boston Children’s Hospital sponsored the CLARITY challenge, a competition to identify the genes causing mysterious diseases in three separate children by crunching raw DNA from their families. The teams did not all come up with the same results.
“If you give a genome to 27 different labs, you will often get different answers,” Rehm said. Her group won the CLARITY competition.
One slide Rehm uses when she is giving talks shows how widely these interpretations vary. Three genomics laboratories cataloged all the genetic variants they knew that were linked to developmental disorders called RASopathies.
Theoretically, these results should largely overlap. But the labs didn’t agree on what 53 of those genetic variants meant. And a portion of those interpretations were wildly divergent. For example, one lab asserted there wasn’t enough information to know what a genetic variant meant, while another said it was definitely pathogenic.
The genome carries powerful information about disease risk and can be an important tool in unraveling disease, and techniques and databases are already being improved. But the pace may not be as fast as the embrace of DNA as a medical tool. Scientists see a day of reckoning ahead.
“There’s going to be a massive awareness of what we don’t know and the discrepancies about what people are saying about things,” Rehm said. Then, she said, the problems will be fixed.Carolyn Y. Johnson can be reached at firstname.lastname@example.org. Follow her on Twitter @carolynyjohnson.