In December 2020, San Francisco-based population genomics firm Helix received the first FDA approval for a whole-exome sequencing platform. The Helix Laboratory Platform offers coverage of approximately 20,000 human genes and 300,000 non-coding regions, they deem “informative.” Two months ago, a Helix-led research team published a breakthrough study in Genetics in Medicine entitled “Positive predictive value highlights four novel candidates for actionable genetic screening from analysis of 220,000 clinicogenomic records.” (Schiabor Barrett KM et al. Genetics in Medicine, Aug. 13, 2021.)
That study identified four potential candidate genes for preventive health screening in a large-scale genomic study, conducted by researchers at Helix and Renown Institute for Health Innovation in Reno, NV.
Currently, the CDC identifies only three conditions as “having significant potential for positive impact on public health based on available evidence-based guidelines and recommendations.”
- Hereditary Breast and Ovarian Cancer Syndrome (HBOC), due to mutations in the BRCA1or BRCA2 genes.
- Lynch syndrome (LS), which is associated with colorectal and uterine cancers and involves variants in MLH1, MSH2, MSH6, PMS2, and EPCAM.
- Familial hypercholesterolemia (FH), associated with atherosclerotic cardiovascular disease via mutations in LDLR, APOB, and certain gain of function variants in PCSK9.
The Helix/Renown team analyzed exome and medical record data from more than 220,000 people across two patient cohorts with different demographics—the UK Biobank (189,495 people) and Healthy Nevada Project (28,423 individuals)—to find variants associated with conditions for screening that could be clinically justified.
They found dozens of potentially significant gene disease associations across almost 30 genes. Of these, seven had a positive predictive value (PPV) of at least 30% in both cohorts. Three are already used in population screening (see above). The other four are novel candidates for population screening: diabetes mellitus, beta-thalassemia minor and intermediate, polycystic kidney disease, and cataracts. The associated genes are, respectively: GCK, HBB, PKD1, and MIP.
The authors concluded that “the best candidates to expand genetic screening programs are those rare variants that predispose individuals to common diseases. Compared to common variants, rare variant associations are much more penetrant, resulting in direct and often more severe phenotypic effects that are also often relevant across ethnicities.”
Helix was cofounded in 2015 by Scott Burke, Justin Kao, and James Lu, MD, PhD, who is now the company’s CEO. It was started as a personal genomics firm, with $120 million from Illumina and others. The company completed a $200 million Series B financing in 2018, intended to advance development of its “marketplace” for DNA-powered products across categories that include ancestry, health, wellness, and entertainment. In June, Helix closed on a $50 million Series C co-led by Warburg Pincus, DFJ Growth, Kleiner Perkins Caufield Byers, the Mayo Clinic, and Temasek.
Since its founding, Helix has shifted its business model to population genomics based on its platform to integrate genomic data into clinical care—their “Sequence Once, Query Often” model.
Besides its proprietary Exome+ assay, the company also operates one of the world’s largest CLIA-certified and CAP-accredited next generation sequencing laboratories, which supports both its Exome+ and COVID-19 viral sequencing efforts.
GEN Edge spoke to Helix’s William (Will) Lee, PhD, VP Science to find out how the company was able to make this progress.
GEN Edge: What is Helix’s business focus?
William Lee: Helix is deploying Genomics at population scale – screening healthy individual for actionable variants. We partner with health systems and provide that kind of information. Ideally, you could have a genomic dataset that is generated early that you can then query over time. Germline genome data is, after all, one of the few health measurements that will not change over your lifetime.
Right now, like a lot of companies, we are tied up with a lot of COVID-19 related work, but there is an assumption that soon health systems will be turning to other things. But our primary research focus is on deploying new applications for the CDC Tier 1 screening.
GEN Edge: What are the challenges with predictive screening?
Lee: From a scientific perspective there is no issue, we are ready to go. But every health system is different and not all specialists who could use this data are ready for it. We need the provider, the payer, and the patient all aligned for this to be successful.
GEN Edge: What were some of key features of this study?
Lee: The partner in this paper was the Healthy Nevada Project from Renown Health Systems. One of the things you need to do in these studies is make sure the participants know what they are getting into—informed consent. Also, in these types of screenings, most people are negative, but we needed to make sure that any positives were handled in a responsible manner. In other words, what was the care pathway for those few people who did have a positive finding?
Also, our platform, Exome+, is unique in a couple of ways. It has all the features that people ask for today. We have boosted it for regions of clinical significance but also spiked in several thousand non-coding sites that give you a backbone of genomic data similar to a genotyping array. That is because we realized that the ability to do GWAS and polygenic risk scores are important to many.
What we generate for every patient is not just the variant calls for everywhere there is sufficient coverage. Because we have these SNPs, we can impute to sites in the reference genomes and from there you can get your genome-wide association score (GWAS). With the data generated this way, the clinical piece is still the #1 priority, but you can learn more about the genetics over time. The more data you have to start with, the more you can learn about it and the more applications you have.
Another advantage was that by having two cohorts [UK BioBank and Renown Health System], we were able to meet the threshold in independent cohorts. The population distribution within these cohorts is different. One of the challenges now in genetics is that the majority of data is from [people of] European descent. Neither of these cohorts is yet as diverse as we would like, but they are diverse from each other.
Finally, we used a type of analysis that is different from what is typically used, it’s called gene-based collapsing analysis. The main advantage is that it lets you incorporate rare variants that are predicted to be high impact. In a typical GWAS analysis, you are focused on common variants. But this approach leverages the power of the exome to find the strongest signal. (See: Cirulli ET et al. Nature Communications, Jan. 2020.).
GEN Edge: How did you get FDA approval for your platform?
Lee: It was very involved. Previously, they had not approved an open-ended assay, rather they were working on variant-by-variant validation. We worked with them to develop guardrails for what we call “representative sampling validation approach.” We divided up the exome into many subsets by variant type, length in the case of indels, by GC context, by proximity to short tandem repeats, by region of genome. Then we did robust validations studies to show that our performance would meet predetermined landmarks.
GEN Edge: What are the key implications of your recent study?
Lee: CDC Tier 1 just covers three conditions right now, and those involve multiple genes. We found four genes associated with four conditions that could impact a significant number of people. The bar we set was high—a 30% positive predictive value. So that means there is a good likelihood that when there is a positive result, that person could go to a specialist and find out what that means.
But generating the data is now the easy part. We are working with the health system on how to deploy that information. If you cannot follow up, you cannot report it.