Skip to main content
Premium Trial:

Request an Annual Quote

Quest Highlights Benefits of Public/Private Data-Sharing Model but Experts Disagree on Best Approach


NEW YORK (GenomeWeb) – In a recent publication, Quest Diagnostics and its collaborators made a case for sharing knowledge about BRCA and other genetic variants through a public/private collaborative model, in which commercial labs pay fees that support curation and other evidence gathering activities, while researchers, doctors, and patients have free access.

But this approach butts up against the view of other experts in the life sciences community who maintain that while multiple databases can coexist and serve different functions, there needs to be a single, coordinated effort for collating information on genetic variants and the database containing this information should be entirely open access.

In a recent paper published in Human Mutation, experts led by Charles Strom, VP of genetics and genomics at Quest Diagnostics, reported that since launching the BRCA Share variant database last July, more than 1,100 users have registered to the site and the database has been queried on average 17,000 times each month. Moreover, since Laboratory Corporation of America joined BRCA Share, with founders Quest and the French National Institute of Health and Medical Research (Inserm), the number of variants in the database has grown by nearly 30 percent to around 6,250.

Strom and colleagues also compared the data in BRCA Share against other large, publicly accessible databases containing BRCA variants — including NIH's archive of genotype/phenotype associations called ClinVar and ARUP Laboratories' repository — and found 72 percent agreement in classifications between ClinVar and BRCA Share; 81 percent agreement between BRCA Share and ARUP's database; and 60 percent agreement between ClinVar and ARUP's database. 

A number of studies now have noted discrepancies between how genetic testing labs are classifying variants and the discordance across databases that contain information on these variants. Strom and colleagues also pointed out that ClinVar and BRCA Share have a large number of variants not present in the other, which "indicates that no existing collection has yet to uncover all the variants in the human population."

While for some this suggests the need to align variant data collection efforts between groups, Strom believes that BRCA Share's public/private collaborative model is the best way to go about it, because it can better support the curation necessary to provide clinically useful data, whereas public repositories are largely limited to research use and dependent on government funding. "Research-grade data and clinical-grade data are not equivalent," Strom told GenomeWeb over e-mail.

Quest and Inserm launched BRCA Share last year with the hope that labs will submit variant data, Inserm will curate it employing various approaches (using family studies, prevalence information in other repositories, and functional analysis), and the database will become a resource for BRCA variant classifications that labs and healthcare providers can use in patient care. Unlike freely accessible databases housing BRCA variant data, such as ClinVar and ARUP's repository, BRCA Share allows researchers, physicians, and patients free access, but participating commercial labs must pay an annual fee based on a sliding scale. So far, LabCorp and Quest, the two largest reference labs in the country, are the only commercial labs in the effort. 

"We believe that as the model continues to demonstrate its value, other commercial entities will seek to engage as paying members," Strom said. In Human Mutation, Strom and colleagues wrote that freely accessible databases such as ClinVar "would undermine the BRCA Share model," but the groups are discussing whether some data can be shared between the repositories.

Meanwhile, ClinVar, because it doesn't charge any fees for submissions or data access, also has been growing and genetic testing firms are increasingly submitting to the repository. Nearly 600 submitters have deposited more than 234,000 variant records. 

Of the top five submitters to ClinVar, three are commercial labs: GeneDx, Invitae, and Ambry Genetics. These labs have described the resources they've put toward submitting data in an iterative fashion and reviewing internal classifications based on discrepancies between labs in ClinVar. While the time, expertise, and funds needed to submit to a public database can be significant, these labs have said this is part of the cost of running a genetic testing lab. 

"Laboratories do a lot of work that's required by the College of American Pathologists or CLIA regulations that they have to work into their budget. If it improves patient quality, it has to be done," said Christa Martin, director of the Autism and Developmental Medicine Institute at Geisinger Health System.

"I don't see data sharing as any different than another measure, now that we have the capability to do it on a broad scale, that would help to improve the quality of patient care through better variant interpretation," said Martin, who is involved in ClinGen — an effort, in collaboration with ClinVar, to build a genomic knowledge base that can be used in patient care.

For smaller groups that don't have the bioinformatics resources to easily submit to ClinVar, Martin added that there are efforts underway within ClinGen to form a group that can help labs format the data and get it into the database. In the future, there will be tools, like a "one-button solution," that labs can use to easily export their data from analysis software they are using to ClinVar, Martin added.

She and others involved in ClinVar and ClinGen have had discussions with CAP about requiring labs to submit to public databases for accreditation. Meanwhile, the US Food and Drug Administration and payors are already supportive of advancing public variant databases, and have issued draft regulatory guidances and coverage policies that further prod labs toward sharing data in open repositories, such as ClinVar. At least one national insurer, Aetna, is requiring all newly contracted labs to submit their variant interpretations to ClinVar as a condition of participation in its network for BRCA testing.

"Given the Precision Medicine Initiative, which ClinVar and ClinGen directly feeds into, and discussions we're having with the FDA to get [public] databases in FDA-compliant mode for precision health, I think they're going to remain a high priority at the federal level," Martin said.

This support from government agencies and other groups has gone far in getting labs that hadn't submitted to ClinVar in the three years it has been around, to do so. ARUP, for one, is planning to submit all variants it has tested and classified, including BRCA1/2 variants, into ClinVar, and is getting ready to make its first submission. "It will be nice to have all variants in a central repository rather than in separate databases that are not talking to each other," said Rong Mao, section chief of molecular genetics and genomics at ARUP.

According to Strom, Quest also began to submit BRCA variant data to ClinVar over the summer. As of July, the firm has submitted nearly 500 variants to ClinVar.

But combining data between BRCA Share and ClinVar, which have different aims, may prove a bigger challenge, in Strom's view. "In theory, there may be value to sharing the data across the databases," he said. "The practical reality, however, is that these databases are managed and curated (or not curated) in different manners, undermining the ability to ensure quality and consistency when combining databases."

He noted that public databases often lack appropriate curation, and don't have people manually interpreting data deposited by submitters. "Research databases, including those funded by government entities, often consider pathogenicity assessment based on prevalence in a given population, compared to prevalence in an unaffected control population," which works when classifying common variants but not for rare variants, observed Strom.

"That sort of robust, clinical-grade genetic data is generally compiled by large commercial laboratories, with databases that include factors beyond prevalence, such as family segregation studies, co-inheritance, in silico prediction, and functional studies," he continued. "And large-scale databases require money and resources that may not be readily available to the academic scientist dependent on government funding."

ClinVar has gotten similar criticism, and labs not submitting to it have highlighted classification discrepancies in the repository. But those involved in the effort say the whole point is to shed light on these differences, so the community can work to resolve them. Some ClinGen expert groups are already starting to do this work, and have submitted more than 6,200 variant classifications to ClinVar, but given that the database currently contains 171,678 records on unique variants, this work will be labor and time intensive.

Within BRCA Share, though it focuses on only two genes, Strom and colleagues highlighted some advantages of the model, since experts at Inserm are in charge of curating the data, flagging discrepancies, and using various approaches to provide a consensus classification whenever possible. In BRCA Share, Inserm experts have tackled 148 BRCA1 variant discordances between submitting labs, and for most they were able to arrive at a consensus classification based on changes in available evidence.

Of the 37 remaining discordant classifications, 17 were due to policy differences among labs for classifying synonymous variants and would have "minimal impact" on clinical decision making. The other 20 are classified as variants of unknown significance in BRCA Share until more evidence emerges allowing a more definitive interpretation. Inserm experts are currently reviewing BRCA2 discrepancies. Discrepant classifications are marked by an asterisk, so users know to review them more carefully.

Within BRCA Share funding has also been set aside for conducting functional studies on variants of unknown significance, and experts are figuring out the best way to prioritize which variants should have functional studies performed on them. "One of the unique aspects of BRCA Share is that it’s the first time that a group fully committed to open sharing of BRCA data is conducting functional studies, a critical gap in current BRCA research," Strom said.

In the 18 months since launching BRCA Share, experts have identified and reclassified hundreds of variants in these two genes, which in Strom's view shows that the public/private partnership model is working. He added that his group has had discussions with other commercial labs interested in joining, and is planning to eventually launch Cancer Risk Share, a broader database of cancer risk gene variants. 

"Quest opted to collaborate with Inserm because we believe their approach surpasses others in terms of quality of curation," Strom said. "This is critical to ensuring the data is robust enough for clinical use."

Heidi Rehm, who is playing a leading role in ClinVar and ClinGen, is generally supportive of BRCA Share and its focus on gathering and curating evidence on BRCA pathogenicity. But she acknowledged that its separation from other databases does make it harder to organize a community wide effort around variant classification.

"It is really important if we want to interpret variants accurately that we aggregate interpretations together, what I call share-and-compare, so that we can identify where there are differences," said Rehm, director of the Laboratory for Molecular Medicine at Partners HealthCare Personalized Medicine. "Right now, with some of the [BRCA] data being separated between ClinVar and BRCA Share, it's not easy to identify where there are differences in interpretation so we can as a community work to resolve them."