A group of researchers from Stanford University, Emory University, and the Georgia Institute of Technology, have published a study evaluating how whole-genome sequencing could be used alongside traditional clinical evaluation in preventive medicine.
In the proof-of-principle study, published last month in Genome Medicine, the researchers sequenced the whole genomes of eight individuals from the Center for Health Discovery and Well-Being at Emory University, analyzing each individual's risk for over 100 polygenic diseases.
Lead author Chirag Patel, a postdoctoral research fellow at Stanford University, told Clinical Sequencing News that the purpose of the study was to evaluate how clinical and genetic information could be combined to estimate disease risk.
Eventually, he said, the ultimate goal is to determine whether, "when we combine both genetic and clinical information, can we better predict disease and better health outcomes?"
The researchers plan to expand the whole-genome sequencing study from the eight people in the current study to the entire 500-person cohort at the Center for Health Discovery and Well-Being, and eventually to 3,000 individuals from different centers around the US, Patel said. The researchers plan to follow the individuals long-term to study how genomic information impacts their lifestyles.
"If patients or physicians knew their underlying genetic or hereditary risk when interpreting their clinical data, how might this change patient behavior or how physicians might guide patients?" Patel added.
The eight individuals, four men and four women, were chosen from a 500-person cohort of healthy adult volunteers and were chosen to represent opposite ends of a spectrum of body mass index, percent fat, HDL cholesterol levels, and triglyceride levels.
Their genomes were sequenced on Illumina's HiSeq 2000 and genetic risk predictions were determined based on the VARIMED database, which covers over 110,000 variants in 9,700 genes associated with over 2,000 phenotypes that have been gleaned from over 6,500 epidemiology studies; as well as an analysis pipeline previously published by Atul Butte's group at Stanford University.
The researchers evaluated the participants in eight major disease categories—immunological, metabolic, musculoskeletal, cardiovascular, respiratory, cognitive, psychiatric, and oncological — covering around 100 different conditions.
The team then tested two different approaches for integrating genetic and clinical data. The first approach directly matched genome-wide association study results with individual diseases, while the second approach combined multiple clinical and genetic measures to generate an overall risk profile. The team found that the second approach was more useful, as it allowed the different metrics to be weighed according to the evidence behind them, while the first approach was often redundant.
Patel said that often the clinical and genomic data were concordant for disease risk. For instance, people at risk for cardiovascular diseases because of BMI were often at risk for the same diseases because of their genomic profile.
However, said Patel, there were also discordant cases. For instance, he said, there was one individual who had a high genetic risk for high triglycerides, but in actuality had low levels of triglycerides. In such a case, he said one possibility is to simply take a "watchful waiting approach."
The results have not yet been returned to the participants, said Patel, because that was not included in the institutional review board approval, but as the study is expanded to the entire cohort and eventually to thousands more volunteers at outside institutions, the group plans to gain IRB approval to return results in order to study how best to return such results and how physicians integrate genomic information into health management.
Currently, next-gen sequencing is primarily being used clinically to diagnose rare diseases or to sequence patient tumors to guide treatment, with more targeted sequencing being used to diagnose specific diseases.
Whole-genome and exome sequencing are rarely being used on otherwise healthy individuals that simply want to know their risk for developing common diseases, although Patel said he thinks that as the price continues to drop, this will become more common, which is why it is important to understand how best to communicate such information and how to incorporate it with other clinical markers of disease risk.
"We do believe that an individual knowing their genetic profile can induce behavioral changes that may have an overall benefit," Patel said, "but we're not there yet in terms of gauging [genomic information's] predictive value for disease."
In the Genome Medicine study, the authors suggest a method of returning results, which they dub a "riskogram," that combines an overall view of the eight health categories, including the individual's genomic and clinical scores and how they compare with the population, as well as a list of rare variants of interest. Because the group did not have IRB approval to return results to patients, the researchers did not include the rare variant findings in the study, in order to protect privacy.
"The visualization of concordance and discordance in the genetic and clinical profiles might help develop personalized health action plans in consultation with a health partner," the authors wrote.
James Evans, a professor of genetics and medicine at the University of North Carolina who was not involved in the study, told CSN that while "we certainly need studies like this that begin to explore the use of genomic data … we have a long way to go before we realize utility" for predicting risk of polygenic diseases.
Rather, he thinks that clinical sequencing efforts should first focus on genetic variants that "carry with them quite robust risk information," for example variants for Lynch syndrome, which confers an "extraordinarily high risk for colon cancer."
Evans said identifying the population that has these types of rare, but highly penetrant variants would have more immediate benefit, since these variants are often actionable.
For instance, if an individual knows his or her Lynch syndrome status, measures can be taken to either prevent or detect early signs of colon cancer, which would provide more utility than knowing about variants that "nudge somebody's risk for diabetes or heart disease by a modest amount when there's nothing specific we would do" to treat or prevent those diseases, Evans said.
Additionally, while it is still early days in terms of understanding how people will use this type of genomic information, the limited data available "suggests that the provision of genetic information doesn't motivate [people] in any lasting way," he said.
Nevertheless, Evans acknowledged that studying how whole-genome sequencing could be integrated into health management strategies is still worth studying. "We need to explore the use of this information," he said, but cautioned that it would still be years before whole-genome sequencing data could be useful for understanding risk for common diseases.
This is in part "because we don't know how to combine these factors into a useful aggregate risk, and also because there are non-genetic" contributing factors, such as behavior and environment, he said.
Patel agreed that more data is needed about how genomic information can predict disease risk, including longitudinal studies to see how accurate certain variants are at predicting disease. For instance, he said, it would be interesting to see whether a fit patient by clinical measures with risk variants for type 2 diabetes would go on to develop that disease. It would also be important to include environmental information like diet and physical activity to "see how it may modulate clinical information in the context of people's genetic risk," he said.