Boston—"Genetics currently sucks at predicting the types of diseases that will kill most of the people in this room," complained Massachusetts General Hospital geneticist Daniel MacArthur. MacArthur's talk, "All Genomes Are Dysfunctional," opened the second day of the Fourth Annual Consumer Genetics Conference. He was a member of a panel focusing on how genomic information is actually being used by consumers and physicians. Why does genetics currently suck at making predictions? Among other reasons, simply knowing that everyone with a disease carries a particular gene variant doesn't mean that everyone with that variant has the disease.
For example, 80 percent of the people being treated in clinics for the hereditary iron overload disease hemochromatosis are homozygous (have two copies) for a specific gene variant. However, out in the world, less than 1 percent of people who are homozygous for that variant actually have the disease. Other genes are likely counteracting the effect in disease-free people or perhaps the healthy carriers have not encountered some particular environmental trigger.
In order to turn genomic information into predictive medicine, MacArthur repeated the rallying cry heard constantly at the conference, "Sequence all the genomes!" Sequencing hundreds of thousands more genomes and linking them to the electronic health records of the sequenced individuals will enable researchers to figure out which gene combinations actually produce disease. "To fully understand one genome we need to compare it to data from hundreds of thousands of genomes," said MacArthur. He urged that all of this information be publicly available, perhaps deposited in the National Center of Biological Information's ClinVar database.
No Genomic Angst
What do consumers do with genomic information and how does it affect them psychologically? Not much in either case, reported social scientist Cinnamon Bloss from the Scripps Translational Science Institute. Her group has completed a study in which several thousand volunteers took a genotype screening test that gave them some probabilistic disease risk information. The volunteers were tested for disease anxiety before and six months after the test. The study also queried the participants if they had made any changes in their lives in response to the test results. Bloss found no changes in baseline in terms of anxiety or positive changes in diet and exercise. However, at six months some 25 percent had shared their results with their physicians, and that figure increased to 36 percent after one year. A large majority of the participants also claimed to find the results useful and understandable.
Bloss noted that a Food and Drug Administration (FDA) panel asserted in 2011 "that certain types of genetic tests… should not be used without the involvement of a physician or a genetic counselor." Why? According to Bass, nearly all of the actual data finds "no measured adverse impacts of direct-to-consumer genetic testing." Besides, Bloss pointed out, it's time to consider models for providing genetic information to consumers and patients that do not involve genetic counselors and physicians since most physicians are not able to interpret genetic testing results and there are too few genetic counselors relative to consumer demand for testing. One questioner suggested that the FDA regulators seemed to be caught up in "genetic testing exceptionalism," pointing out that people can take HIV and pregnancy tests at home which yield some very consequential results.
Designer Babies for All?
That was the question in the next talk by Diana Bianchi, the executive director of the Mother Infant Research Institute at Tufts University. She focused on the blindingly fast uptake of breakthrough non-invasive testing for fetal trisomy diseases, the condition of having three copies, rather than the norm of two copies, of a given chromosome. The classic trisomy disease is Down's syndrome, which is characterized by three copies of chromosome 21, though there are also rarer trisomies of chromosomes 18 and 13. The new fetal trisomy tests rely on the fact that 10 percent of the cell-free DNA circulating in a pregnant woman's blood comes from the fetal placenta.
The tests will be a year old this month and already 70,000 women around the world have taken advantage of them; 50,000 in China, and 20,000 in the United States. The new tests are much more accurate than the old standard of care, involving serum screening, ultrasound, and invasive surgeries to obtain fetal cells for checking. The false positive rate is 0.2 percent and, according to Bianchi, there are no known examples of false positives negatives. In fact, some Chinese testing companies are so confident of the test results they offer to pay $30,000 to any woman who continues her pregnancy in reliance on the test and subsequently gives birth to a baby with Down's syndrome.
More non-invasive fetal genetic testing is on the way, including tests for Rhesus D, fetal sex, sickle cell, and thalassemia. Bianchi pointed out pregnant women have a tremendous interest in direct-to-consumer fetal testing, citing the popularity of the PinkorBlue fetal gender test. Available over the counter in drugstores, that particular test can identify fetal gender as early as seven weeks into a pregnancy.
The National Institute of Standards and Technology's Marc Salit loves measuring things. He is heading up a public/private collaboration, the Genome in a Bottle Project, to produce reference materials for validating various methods of doing whole genome sequencing. The goal is to produce extremely well characterized reference genomes that can be handed over (in a bottle) to sequencing companies and research labs to check how accurately their processes are working.
Physicians Hate Consumer Genetics
Or at least they are really, really skeptical is the take-away message from the two person panel on the physician's perspective of genome data. The first panel member was Harvard Medical School physician and researcher Issac Kohane who warned consumer genetics enthusiasts to be wary of the "incidentalome" lurking in current genomic data. Kohane coined that word to highlight the proliferation of false positives in genome research. For example, in a recent article in the journal Genetics In Medicine, Kohane notes that the published genome of the co-discoverer of the double-helix structure of DNA James Watson reports that he is homozygous for gene variants previously documented to cause two different congenital diseases. In this case, Watson has inherited from both parents variants for Usher syndrome and Cockayne syndrome.
People who suffer from Usher syndrome typically become deaf and blind in early childhood, and those afflicted with Cockayne syndrome fail to grow, age prematurely, and have abnormally small heads. The 84-year-old 1962 Nobelist in Physiology or Medicine does not appear to suffer from either syndrome. As cited above, Watson's case illustrates again the fact that knowing that everyone with a disease carries a particular gene variant doesn't mean that everyone with that variant has the disease. This is because there is tremendous redundancy built into our genomes. As Kohane explained it, "one thing going wrong is not enough to take us down. That would not be evolutionarily robust."
The second panel member, Harvard University medical geneticist Michael Murray asked, "What would you do with your genome?" His answer: "For the most part you would do nothing." He then asked the audience to think about the case in which while reviewing your whole genome sequencing results you found that you had an unknown variant in the TSG11 gene, a tumor suppressor gene. What would you do with that information?
Murray temporarily set that aside to discuss some cases illustrating how genomic information might or might not be useful. In one case, about 36 percent of people of East Asian heritage have a variant ALDH2 gene which causes them to experience facial flushing, nausea, and a higher rate of heartbeats if they drink alcohol. However, if they take an H2 blocker like Zantac, they can drink and not experience these symptoms. So does taking H2 blockers solve all their problems with regard to flushing? According to Murray the answer is no, because even if they take H2 blockers people with the ALDH2 flushing variant are at a much higher risk of esophageal cancer if they drink.
As an example of how genetic information is being misused Murray cited the case of the Atlas First Sportgene Test. The company tests for variants of ACTN3 gene which it claims is related to athletic predispositions. Supposedly some carry a variant that suits them for sprint/power sports and others for endurance sports. However, Murray pointed out that a recent winner of the Boston Marathon was an East African. The so-called endurance version of the ACTN3 gene is mostly absent in East African genomes.
Murray then returned to the tumor suppressor gene variant case with which he opened. So far something like 1,300 tumor suppressor genes have been identified in the human genome. TSG11, on the basis of very thin data, has been associated with lung cancer. Murray worries that a patient, armed with this equivocal genetic information, might demand that his doctor conduct further (unnecessary) testing such as a lung CT scan. And this is where Murray and Kohane revealed their real concerns: They are mightily annoyed by the prospect that patients will have their genomes sequenced and then bedevil hardworking physicians like themselves with questions, follow-up visits, and requests for further testing. Murray actually said, "I believe that physicians should suppress information that is not useful or is uninterpretable."
During the question period, Rosalynn Gill, the chief science officer for Sciona—who also happens to be PGP#9, or Personal Genome Project Whole Genome #9—challenged Murray and Kohane to provide actual evidence that doctors are being overwhelmed by the demands of consumers of direct to consumer genetic testing. In addition, the fact that physicians are so overscheduled that they can spend only a few minutes with each individual patient suggests that the health care system is broken, not that information should be suppressed.
What Does Your Genome Say About You?
The last panel of the day was devoted to describing various efforts to interpret genome sequencing results. In a sense the main theme of the entire conference is about how to make genomic information relevant to clinical outcomes cheaply and quickly. Martin Reese, the CEO of Omicia, did a nice demonstration of how his company's Opal genomic interpretation software suite operates. Opal uses a proprietary algorithm to compare whole genomes from people suffering specific genetic diseases with reference genomes to quickly identify rare variants associated with their diseases. Next up was bioinformatics guru Michael Cariaso, the founder of SNPedia, a wiki supporting personal genome annotation, interpretation, and analysis. He is also the author the Promethease program for interpreting personal genomes. He was correctly introduced as "something of a cult figure in our field." Cariaso gave the audience a quick run through of the capabilities both SNPedia and Promethease. So far SNPedia contains some 35,000 annotated gene variants. If you're interested in how Promethease works head on over and take a look at what it tells you about my genome.
My final dispatch from the conference will report on talks focusing on using genetic results to track down your ancestry; what it's like to open up your whole genome to public scrutiny; and how genetic testing can help babies. For more background, take a look at my first conference dispatch, "Virtual Children, Genome Sequencing for Everyone, and Forget Genetic Privacy."