NEW YORK (GenomeWeb) Genomics-driven wellness firm Arivale said this week that it is making some changes to the service it offers to its customers with the launch of several new polygenic profiles that are integrated in its analysis and coaching interface.
The firm's business involves whole-genome sequencing to look for variants that affect health and wellness metrics. The results are reported to clients, coupled with close monitoring of biomarkers like blood sugar and lipid levels, which coaches use to guide them in improving or maintaining their health.
Whereas Arivale's initial reports have only referenced individual variants, the company is now rolling out a new set of polygenic profiles (PGPs), which predict traits like higher bodyweight or risk for high lipid levels based on the combined influence of hundreds of SNPs.
The firm added new PGPs linked to body mass index and waist circumference in January, and just followed those up with additional profiles for lipids (LDL cholesterol, HDL cholesterol, total cholesterol, and triglycerides). Next on deck is polygenic assessments of traits that have to do with physical exertion and exercise, Arivale CEO Clayton Lewis said in an interview this week.
The idea behind polygenic profiles is that the combined effect of all the common genetic variants a person carries might offer more insight into genetic predisposition for a certain trait than any single genetic variant alone.
For example, the company said in a recent description of its new PGPs that scientists have identified several hundred variants that affect BMI. "No one has all these variants, but everyone has some subset of them. And, most likely, no other person has the exact same subset," the company wrote. "Furthermore … some common variants might slightly raise your BMI, while others might lower it. In this manner, virtually everyone has a unique set of variants contributing to their unique polygenic profile for BMI."
Constructing PGPs can also allow the weighted consideration of loci that might not have achieved genome-wide statistical significance on their own, but nevertheless contribute incrementally.
"The SNPs we were reporting were ones where there was a lot of evidence and they had relatively strong effects," Arivale's chief translational science officer Jennifer Lovejoy said. "But a single SNP, no matter how good, is only going to explain a very tiny amount of variance in an overall trait. If you can combine hundreds, you can explain a lot more, so it's a better potential predictor."
To create its PGPs, Arivale is using public data from peer-reviewed published studies, mainly very large GWAS from international consortia.
"We pull data from our [customers'] whole-genome sequences but the basis of the [PGP] algorithm is existing peer-reviewed public data," Lovejoy explained.
The firm's lipid polygenic profiles, for example, calculate customers' potential predisposition for high LDL cholesterol, low HDL cholesterol, or high triglycerides based on the presence or absence of roughly 800 genetic variants. Those variants come from a GWAS meta-analysis conducted by the Global Lipids Genetics Consortium, which analyzed a total of 188,000 people and over 2.4 million common genetic variants.
Its BMI and waist circumference polygenic profiles, meanwhile, come from data published by the Genetic Investigation of Anthropometric Traits (GIANT) consortium. Depending on the particular profile, these incorporate about 400 different variants, Lovejoy said.
In addition to the more direct PGPs, like those for BMI or for high triglyceride risk, the company is also reporting scores that speak to gene-environment interactions, like the impact of specific types of food or exercise on weight.
For example, Lewis said, Arivale is collating variants that it believes affect participants' propensity to gain weight on three different types of diet, or if they don't exercise, or that predict their attraction to sweets or bitter foods.
So far, there has not been much data collected on the predictive power of polygenic risk scores, or how they improve risk or trait prediction over strategies that just measure individual variants.
Despite this, Arivale is not the only entity implementing them. Researchers from the NHGRI-funded MedSeq project, for example, also developed methods to summarize participants' polygenic risk for complex diseases in whole-genome sequencing reports. They published a report on their strategy in late 2014.
Arivale makes clear to customers that the science on wellness trait loci — not to mention the greater field of common complex disease genetics — is still evolving. For example, the company says on its website that variants with small effect sizes or very low frequency may not yet be known, and that the way variants interact with each other is far from understood.
However, the company says that it believes its PGPs are based on the "best available research" right now.
Cofounded in early 2014 by Lewis, along with Lee Hood and Nathan Price from the Institute for Systems Biology, Arivale was initially inspired by a pilot study that was part of the ISB's 100K Wellness Project.
So far, that pilot data is the only evidence of the impact of Arivale's approach on patients' health — the clinical utility or value of what it is selling.
In that study, Hood and other ISB researchers sequenced the genomes of 107 individuals and tracked their metabolites, protein markers, and gut microbiome, as well as fitness activity, sleep, and dietary patterns. As in Arivale's commercial service, a nutrition-trained coach then provided participants with individualized health recommendations to optimize their wellness and help them avoid disease.
Lewis said that the company plans to publish detailed results from that study soon, and is also analyzing results now from its commercial cases, which it hopes will demonstrate that participants do see improvements in their overall health. Without a control group to test against, this still won't speak to whether the program improves outcomes over what might have happened to these individuals in its absence.
The company is also working with an independent health economist to analyze what these improvements might translate to in terms of savings in healthcare costs and long-term impact on public health.
"Of roughly 1,300 clients that we have signed up [in our first six quarters] a little north of 800 have gone through a six-month blood draw, so Jennifer and the scientific team are looking at that now," Lewis said.
Though it hasn't done so yet, Lewis also said that Arivale plans to do clinical trials that provide more persuasive evidence of the impact of its genetics-based coaching.
Some previous research has suggested that the impact of genetics on individuals' behavior is quite limited.
For example, researchers from the Impact of Personal Genomics (PGen) project reported last year on an analysis of 1,042 patients who received cancer risk estimates for breast cancer, prostate cancer, and colorectal cancer.
When patients were surveyed to determine whether they had made any changes to their diet, exercise, vitamin and supplement intake, or cancer screening behaviors, the researchers found that most who received elevated risk results did not significantly change their behaviors.
An earlier analysis by other researchers, published in 2011 in the New England Journal of Medicine, looked at changes in anxiety, diet, exercise behaviors, and follow-up screening tests and similarly found no evidence of negative or positive effects.
According to Lovejoy, that's not surprising considering some of Arivale's anecdotal experiences so far, but the company believes that the comprehensive coaching it offers can fill the gaps that genetic analysis alone leaves.
"Getting the lab markers, the combination of lab testing and analyzing the genomics, is really what brings the power to the program," she said.
"We have had individuals interested in the genetic and biomarker data who didn't want to make lifestyle changes, but often when they see deterioration in clinical lab tests or [a lack of] improvement with modest changes in behavior, it can be a sobering moment for them and they start to be more engaged," Lewis said.
"Let's say you have really brilliant genetic predisposition for low LDL but you get your lab results back and you actually have high LDL," he added. "That might give the coach an opportunity to discuss lifestyle decisions … there is an opportunity there."
Robert Green, co-principal investigator of the PGen project and director of the Genomes 2 People Research Program at Brigham and Women's Hospital, the Broad Institute, and Harvard Medical School, said that he and colleagues have no new updates to their most recent findings yet, though they are conducting additional research. However, he said, the fact that behavior is very hard to change, and that coaching regimens do help, are both well recognized in the field.
"I deeply respect the science that … is going into the Arivale offering, but it’s a tough area in which to prove efficacy, because behavior is notoriously difficult to change, and because without control groups and long-term follow-up, it may be difficult to ascertain the efficacy of a combined intervention like Arivale’s multi-omics information combined with health coaching," he said.
Although Arivale sequences customers' whole genomes, it does not report medical genetic results that predict an increased risk for a specific disorder outside the narrow rubric of wellness.
For example, while it might report mutations associated with a disorder that affects absorption of a particular nutrient, it will not report mutations in the BRCA genes that confer elevated risk for breast cancer.
Users, however, own their data. "We can put VCF files on their dashboard if they want or send them the whole thing," Lovejoy said. However, only about 40 customers have done so thus far.
"It's not that frequent because it's not that easy to do very much. You can't walk into your doctor, or even a medical geneticist, and hand them your BAM file and say 'hey, what do I do with this," she said.