Expect the Unexpected
Taking stock of medical progress in the decade since the Human Genome Project
As medical mysteries go, the woman who came to Dr. Leslie Biesecker for help last year seemed like an open-and-shut case. Her HDL cholesterol was abysmal, her triglycerides were soaring, and she wasn’t responding well to statins. It was a familiar story: Her mother had a similar clinical picture and was diagnosed with Dunnigan syndrome, a less-than-one-in-a-million form of insulin resistance and abnormal fat storage caused by mutations in a single gene called LMNA. Presumably, the woman had inherited her mother’s disease. All that remained was to make the diagnosis official. But some details of the woman’s case “didn’t make sense to her and her doctor,” says Biesecker. She had come to his office at the NHGRI to have her protein-making genes sequenced as part of a study on heart disease, a common complication of Dunnigan syndrome. She wanted to know what else Biesecker could tell her.
Not only did the woman not have her mother’s disease, her mother didn’t have it either.
The answer was more than either of them expected. When Biesecker’s team sifted through the woman’s genome sequence, her LMNA gene looked fine. What didn’t look normal was a different gene called PPARG. Mutations in this gene can prevent the body from breaking down glucose, causing it to store fat improperly; in very rare cases, the effects are quite dramatic. But such mutations don’t cause Dunnigan syndrome – and, unlike the gene that does, PPARG can be targeted directly by insulin-sensitizing drugs. Not only did the woman not have her mother’s disease, her mother didn’t have it either. The initial diagnosis had not been based on genomic data – and it was wrong.
If genomics has taught its practitioners anything over the last decade, it’s that things are often not what they first seem. Since the mapping and sequencing of the first human genome 10 years ago and the subsequent sequencing of thousands more human genomes, the field has yielded a long list of surprises. Many of these findings – whether focused on rare diseases or common ones – have revealed the enormous complexity of human biology. By doing so, they’ve made clear that the road to new therapies for human diseases will be long and winding. But they’ve also started us down that road. Without them, we might not even be facing in the right direction.
Take what scientists have learned about cancer, for instance.“Historically, we’ve classified cancers according to where they come from in the body or what their cells look like under the microscope,” says Dr. William Pao, a physician-scientist at Vanderbilt-Ingram Cancer Center who is a thoracic oncologist and studies genes involved in lung cancer. At the genomic level, however, that classification doesn’t always make sense. The way a cancer cell behaves may not be mostly determined by its tissue of origin or its appearance on a microscope slide, but by the combination of mutations carried in its genome and any differing combinations in other cancer cells around it. That fact is not merely of scientific interest. There are direct clinical consequences of knowing about the relevant mutations in cancer cells, says Pao: “It influences whether a specific treatment is appropriate or not.”
Doctors are currently using genomics to “define which cancers are oncogene-addicted,” or driven to uncontrollable growth by specific mutations. Many such mutations occur in genes that encode kinases – signaling proteins that orchestrate key cellular functions. Mutant kinases are attractive therapeutic targets because they occur in cancers and not normal tissues. Within five years, says Pao, scientists hope to have comprehensively laid out which types of cancer respond to drugs that shut down kinase activity. Some cancer cells may respond to such targeted drugs at first but eventually grow resistant to them, in roughly the same way that bacteria can evolve to avoid being killed by antibiotics. But scientists can fight back by repeatedly sequencing cancer genomes to track new genomic vulnerabilities that develop as the cells divide, or by initially hitting tumors with drug combinations less likely to elicit resistance.
Genetically targeted therapies for rare diseases are also beginning to hit the market. Forbes named one of them – Kalydeco, which improves the function of a defective protein that causes some cases of cystic fibrosis – “the most important new drug of 2012.” Old drugs, too, are finding new purposes as rare diseases are genetically matched up with these drugs’ mechanisms of action. And even patients with untreatable rare conditions may benefit from genomics.
Genetically targeted therapies for rare diseases are also beginning to hit the market. Forbes named one of them... “the most important new drug of 2012.”
The NIH’s five-year-old Undiagnosed Diseases Program (UDP) has evaluated more than 500 “mystery” patients – most of whom have spent years fruitlessly seeking a diagnosis – and pinned down at least a partial diagnosis for a third of them. By combining its genomic data with studies in genetically altered model organisms, it has also discovered two entirely new diseases, says UDP director Dr. William Gahl, and connected others to “biochemical pathways that people didn’t even know about.”
Genomics is also connecting common diseases to unexpected pathways. For instance, studies of the common type of inflammatory bowel disorder known as Crohn’s disease have pointed toward two genes that influence a biological process known as autophagy, in which cells essentially eat themselves in an effort to clear bacteria from the body. Autophagy is a normal process, and not necessarily an obvious Crohn’s disease culprit. But it now appears that if genes involved in the process are defective, they may allow patients’ intestines to become colonized with harmful microorganisms. These findings could be used clinically – for instance, to determine if carriers of mutations respond better to certain drugs, such as steroids or other therapies that suppress the immune system.
Other pathways altered in common diseases are as yet more mysterious. Almost 90 percent of the genomic variants that have been genetically linked to common diseases appear to fall outside the protein-making part of the genome. Some of these variants may affect genes that function in overlapping pathways, and it is likely that they regulate the activity of protein-coding genes. But the details of how they do that are mostly unknown. “We don’t understand most of the functional elements in non-coding DNA,” says Dr. Eric Green, director of the NHGRI. “Let’s say you give me five variants from non-coding DNA and tell me, ‘One of these five is conferring risk for hypertension.’ When I try to figure out which one it is, I will scratch my head for a very long time.”
“We’ve now pushed the challenge from data generation to data analysis. It’s the biggest bottleneck in realizing the complete dream of genomics.”
Making sense of non-coding variants will require studies in cells and tissue samples, for a start. So that is what scientists are pursuing. Last month, a far-flung group announced a major biobanking project that should illuminate the so-called “dark” part of the genome. And many other geneticists are collaborating with biologists to see how their data line up with other results.
There is much still to be done. Scientists need to figure out how much of the heritability of common diseases they can explain with current methods and data, and (they are making good progress). What’s needed are “very large studies involving lots of people who are willing to share lots of information about themselves,” says Green, while protecting those people’s confidentiality as much as possible. And they need to figure out how to make sense of the data they already have. “That is Herculean,” says Green. “We’ve now pushed the challenge from data generation to data analysis. It’s the biggest bottleneck in realizing the complete dream of genomics.”
These are long-term challenges. But while scientists move closer to solutions, they will continue to churn out surprising new findings for years to come – like the ones that keep turning up in Biesecker’s office. He recalls another patient who came to him last year. “He had a Hox gene mutation,” Biesecker says, sounding a little astounded. Hox genes direct embryonic development. In fruit flies, mutations in these genes can result in animals with legs growing from their heads, and in humans, they can disrupt brain development and cause limb malformations. But this patient had no such issues. Biesecker met with him. “He said, ‘Now that you mention it, no one in my family can wear flip-flops. There’s not enough space between our toes.’ And I thought, ‘That’s it.’ You just never know what’s going to come up.” Which is why it’s important to keep asking.
Mary Carmichael is a science writer based in Boston, Massachusetts.
Plauqe illustration courtesy of Darryl Leja, NHGRI
(1) The Undiagnosed Diseases Program. National Institutes of Health.
(2) Undiagnosed Diseases Program (UDP) Discovers a New Disease. National Human Genome Research Institute.
(3) Global Alliance for Secure Sharing of Genomic and Clinical Data. Broad Institute.