By Aaron Krol

October 8, 2013 | As personalized genomic testing has grown from a niche market into a mainstream service – with 2013 witnessing the millionth customer to purchase a consumer genetic test – questions about the clinical usefulness of all this genomic information, and the ethics of how pathogenic variants are reported, have become matters of real practical concern. Two large-scale surveys that peered into consumer responses to genomic test results are now delivering insights into the mindsets and health behaviors of patients who seek out whole genome sequencing. Their findings may help point the way to best practices in reporting genetic risk factors to patients and their clinicians.

The Impact of Personal Genomics Study (PGen), led by Dr. Robert Green of Harvard Medical School and Dr. Scott Roberts of the University of Michigan School of Public Health, surveyed 1,000 customers of the genomic testing companies 23andMe and Pathway Genomics. PGen was launched in early 2012 through a grant from the NIH, and its data is currently being analyzed for publications. Commercial companies like 23andMe form “a natural experiment in incidental findings,” said Dr. Green, speaking at the Consumer Genetics Conference (CGC) in Boston on Sept. 25 – and their customers can provide a direct look at how those findings impact people’s lives.

The Coriell Institute, meanwhile, has been running the Coriell Personalized Medicine Collaborative (CPMC) since 2007, under the direction of president and CEO Dr. Michael Christman. CPMC delivers its own sequencing, and surveys participants on a voluntary basis. Over 8,000 people have enrolled in CPMC, through partners ranging from universities and medical centers to the U.S. Air Force – and about half of those have followed up with surveys. CPMC continues to enroll and survey new participants, and with a wealth of data already available, they are beginning to discuss the patterns in the results.

If there is a single takeaway from these studies for clinicians, it is that a substantial percentage of people who seek out genomic testing want to take action in response to their results. “They actually see this in a way as educating doctors,” said Dr. Christman at CGC. Ninety percent of participants in both studies expressed an intention to share their tests with their physicians, although only half as many actually did. A few took more concrete steps. After 6 months, 11% of PGen participants had followed up with additional tests, exams or clinical procedures. After 12 months, as many as a third of CPMC participants reported some change in their lifestyle. And months after receiving their results, respondents had little trouble recalling which diseases they felt themselves to be at highest risk for.

The Genome-to-English Translation

Still, while CPMC and PGen are seeing many of the same patient behaviors, the testing services they examine are not perfectly comparable. PGen, by partnering with 23andMe and Pathway Genomics, gets a glimpse at how consumers react to the test reports currently on the market; while CPMC is able to design its own custom reports with the participants’ comprehension in mind. The differences between these reports are subtle, but important. Crucially, 23andMe delivers estimates of absolute risk based on a customer’s genetic profile and lifestyle information (i.e., a 10% chance of developing Type 1 Diabetes); Pathway Genomics reports total risk on a relative gradient (i.e., an increased susceptibility to breast cancer); and CPMC’s report does not estimate total risk at all, but shows relative risk broken down by cause (i.e., a moderately elevated risk for coronary artery disease based on genetic factors, but a sharply elevated risk from smoking). Consumers can therefore see the same genetic data through very different lenses depending on the services they choose.

Dr. Christman believes that relative risk is the responsible approach. “The only way you could actually tell someone their absolute risk is by measuring groups that do or don’t have a given genetic component and following them for life,” he told Clinical Informatics News. In the absence of such a concerted study, any estimate of absolute risk must rely on incomplete information. Dividing risk into its constituent causes also prevents CPMC from exaggerating the genetic component. “When we tell you your genetic risk for cardiovascular disease, we tell you a lot of more familiar information that you’re also able to put into context,” said Dr. Christman.” You see for example that your body mass index being high is maybe your biggest risk for heart disease, or a bigger risk than the genetic component.”

A Sample Risk Report from CPMC

Dr. Green also has his concerns about the reliability of absolute risk predictions, or consumers putting too much faith in their genetic profiles. Still, he suspects the benefit of these reports outweighs the harm. “Particularly the high-level companies, I think, have [calculated risk] in a very responsible manner,” he said at CGC, “with lots of communication that you don’t get in a clinical encounter.” With 23andMe remaining the market leader, the merits of absolute vs. relative risk reports may have a real bearing on health outcomes.

As long as the clinical behaviors reported by the two studies are so similar, it might seem unimportant precisely how risk is calculated. But whole genome sequencing as a clinical tool is in its infancy, and most physicians are still uncertain how or whether to take action based on the information at their disposal. “You don’t have too many physicians who think this isn’t useful information,” said Dr. Christman, but few are changing their clinical practice. If doctors start to take their patients’ genomic data as important cues for action, comprehension of that data could have much more profound consequences.

Both PGen and CPMC took the dangers of misunderstanding very seriously, asking questions like, “On a scale of 1 to 5, what do you think your chance of developing heart disease in your lifetime is?” and comparing participants’ responses to the opinions of genetic specialists. PGen also delved deeply into users’ levels of anxiety, and found some perplexing results. For instance, users with the highest overall disease risks did not report significantly greater anxiety over their results than those with the lowest risks. But if distress scores were instead stratified by the number of conditions participants believed themselves to be at increased risk for, a small but real correlation emerged. “What people walk in the door with in terms of risk perception, risk anxiety and distress seems to have a very potent effect as a filter on how they see the information that they’re getting back,” said Dr. Green.

So far, CPMC has not detected the same pattern. “Participants really understand their results pretty well,” said Dr. Christman at CGC. “We survey genetic knowledge at baseline, and then we do it also following various reports. And people at all education levels have a pretty reasonable understanding.” While it’s impossible to draw firm conclusions about risk reporting from these two methodologically distinct studies, they do serve to highlight the need to establish clinical standards for genomic data before these tests become ingrained in everyday medical practice. The decision to report either relative or absolute risk will certainly be a key question as these standards are developed.

Genetic counseling might also be advocated to clarify the meaning of genetic results, but neither PGen nor CPMC found a high level of interest in counseling – even though it was available as a free service in both studies. Fewer than 10% of participants in PGen consulted with any type of genetic specialist; CPMC, which offers genetic counseling on every page of its risk reports, still finds that only between 10% and 12% of respondents take advantage of the option.

Demographically, CPMC and PGen found similar groups sought out genome sequencing whether it was offered on the private market or through research institutions. Users of genomic testing span a wide range of ages, education levels and self-reported health, but currently skew toward higher incomes and are less racially diverse than the general public. Women made up 60% of both studies’ populations, demonstrating a much higher inclination than men to pursue these services.