Skip to main content
Premium Trial:

Request an Annual Quote

With More Than a Million Records, ClinVar Value Grows as Variant Classification Resource


NEW YORK – Since its launch in 2013, ClinVar has not only amassed a wealth of genetic variant information, but the lab and clinical genetics community in the US are readily referencing it when they have questions about the clinical significance of variants.

A few healthcare systems have recognized the importance of sharing data on variants as genetic testing becomes increasingly incorporated into the care of patients and have begun ordering genetic testing from labs that deposit information on detected variants in ClinVar. And at least two insurers are vetting labs for coverage based on whether they deposit variant data into the public variant repository.

ClinVar, funded by intramural US National Institutes of Health funding and maintained by the National Center for Biotechnology Information, was launched seven years ago as a public resource for tracking the clinical significance of genetic variants. The effort was initiated just before the US Supreme Court invalidated the patent eligibility of naturally occurring genetic sequences, which meant that labs could no longer enforce their gene patents and maintain testing monopolies for entire genes. After this decision, the genetic testing market grew rapidly, but as more patients were tested, more variants were identified. Many of these variants had never been seen before and their role in disease was unclear, and labs were interpreting them differently, sometimes incorrectly. 

To improve the field's ability to interpret the clinical significance of genetic variants, it became evident that health systems, researchers, and commercial labs share data on detected variants and deposit them to a publicly accessible repository that would maintain up-to-date information as new evidence emerged. ClinVar was positioned to be that resource. The database has now amassed more than 1 million records on genetic variants from more than 1,400 submitters.

However, when the database was first launched, there were detractors who felt that like prior databases, it could end up being a dumping ground for out of date classifications. Some labs refused to publicly disclose the variant interpretations, while others declined participation criticizing it for having conflicting classifications between labs.

The proponents of ClinVar, such as Heidi Rehm at Massachusetts General Hospital, would attribute its success to date to the power of public transparency. Competing labs submitting to ClinVar can plainly see how their classifications differ from their competitors', as can providers, genetics experts, and patients who receive test results from these labs.

Meagan Farmer, genetic counseling business manager at the health technology firm My Gene Counsel, noted that she and other genetics professionals tend to consult ClinVar when they encounter test reports with variants that are classified likely pathogenic, variants of unknown significance, or pathogenic variants with reduced penetrance. "In these instances, they may be looking to see how multiple labs are classifying a variant and looking for some consensus among labs they trust," she said.

When looking up variants, Farmer is cognizant of the fact that classifications in ClinVar, even if some labs update classifications regularly, can get out of date and there may be a lag in terms of when the most recent information is uploaded. In those circumstances, Farmer calls up labs to ask how they classified a variant and whether they internally reviewed that variant more recently. 

Invitae Chief Medical Officer Robert Nussbaum said that the lab receives several inquiries daily from providers and genetic counselors when they have questions about a genetic test result. For example, members of a family with the same variant may have been tested by different labs and gotten varying interpretations. Or they may have looked up a variant found in a patient in ClinVar and seen that labs are classifying it differently.

Invitae submitted the millionth record in ClinVar and has made the most submissions, more than 300,000. "This is not surprising to me, since Invitae has massively dominated the market and genetic testing in general," said Rehm, principal investigator of ClinGen, a NIH-funded sister effort to ClinVar that aims to define the clinical relevance of genes and variants in precision medicine and research.

Invitae has estimated that it will accession more than 500,000 samples for genetic testing in 2019 and generate $220 million in revenue, and in 2020, exceed 1 million samples accessioned and bring in $500 million in revenues. The company has grown its market share in genetic testing by pushing down test pricing, but Invitae also committed early on to publicly sharing its variant classification data as a way to improve the quality of testing in the field.

Variant interpretations can differ between labs based on the evidence they've gathered and how they weigh it. But seeing the discrepancies in ClinVar and fielding more questions from the community about them has motivated labs to work together to tackle the variant classification challenges together. "ClinVar enables a forum where clinicians and laboratories can look and see what other people think about a variant," Nussbaum said. "It makes information available for discussion and for coming to some kind of consensus."

In the early days of ClinVar, an oft-cited figure among detractors was that 17 percent of around 12,900 variants interpreted by two or more submitters had conflicting classifications. This figure, as experts involved in ClinVar have pointed out, includes all classification discrepancies, not just those that would impact clinical management of patients. Today, there are approximately 159,230 variants in ClinVar with multiple submissions though only around 4.5 percent have a medically significant conflict. 

Over the years, Rehm has tried to encourage labs to come together and hash out the discrepant classifications and try to arrive at consensus interpretations. The labs that have engaged in such exercises have had some success at decreasing conflicting classifications. Rehm noted that this work is continuing among commercial labs and new data will soon be published showing that the discrepancies are very low among labs that collaborate in this way. 

But much of the current approaches to variants classification are slow and painstaking work. For example, there are currently 675,600 unique variation records in ClinVar, but only around 10,300 have been reviewed and classified by expert panels. Given the growth in genetic testing, Nussbaum noted the need to do variant interpretation at scale with the help of artificial intelligence and other methods.

Toward this end, Invitae last year acquired Jungla, which uses a machine learning platform to integrate functional, structural, and computational data and create predictive models for variants interpretation. However, these type of approaches need to be trained and validated. This is where ClinVar is also proving useful. 

For example, in a study led by University of Washington's Jay Shendure, which employed saturation genome editing to determine the function of thousands of BRCA1 variants, researchers compared the assay's functional determinations against variants in ClinVar interpreted by an expert panel, and showed strong agreement between the two. "They used ClinVar as a source of truth in order to validate what I think is a powerful assay," Nussbaum said. "This [validation approach] is being replicated by lots of other people." 

Invitae will validate the use of its machine learning platform in variant classification using its internal data but also expert consensus classifications in ClinVar. "If you're validating a variant interpretation system, you need a truth set. You need a gold standard to measure your method by," Nussbaum said. In his view, the closest thing the field has to a truth set right now in variant classification may be the expert consensus variant interpretations in ClinVar.

Moreover, a public repository allows the broader genetic community to keep tabs on the quality of the data in it. For example, in 2018, Amalio Telenti and Craig Venter led a research team that used variant frequencies in whole-genome sequences to determine that likely pathogenic and pathogenic classifications may be inflated within ClinVar. But their analysis also showed that over time, as more evidence was accumulated in the database, a large proportion of pathogenic or likely pathogenic variants were being reclassified as benign or indistinct risk groups, as would be expected. 

As this analysis suggests, the quality of the data in ClinVar stands to improve as more labs share their interpretations. To encourage this, Rehm has continued to educate the broader healthcare community on the importance of publicly sharing variant data.

When Rehm joined the Massachusetts General Hospital's Department of Medicine and its Center for Genomic Medicine in 2018 as chief genomics officer, she made data sharing a priority and enforced a policy that external labs that MGH uses for genetic testing submit variant data in ClinVar. MGH does make exceptions, Rehm said, when a patient's insurer will only reimburse for testing from a lab that doesn't share in ClinVar. 

Geisinger has a similar policy, and according to Amy Sturm, codirector of the health system's MyCode Genomic Screening and Counseling Program, it hasn't been difficult to enforce. "It is truly our culture and our belief that we only want to work with labs that free the data," she said. Within Geisinger's program, a genetic counselor or variant classification scientist reviews each variant that is reported after confirmatory testing in ClinVar before uploading it into the electronic medical record and providing it to a patient.

With most major commercial genetic testing labs in the US now submitting to ClinVar, there is growing acceptance in the broader healthcare community, including payors, that transparency around variant classifications is a public health good and can help distinguish quality labs in an increasingly crowded genetic testing market. 

In 2017, Rehm spearheaded the creation of a lab list within ClinGen that meet submission requirements, in order provide ClinVar users with a level of assurance that variant data of a certain standard was being deposited and that labs were keeping the evidence underlying the variants up to date. 

In order to be on the list, labs have to receive one star from ClinVar, indicating they have outlined the criteria they use to make assertions about variants submitted to the public database. Labs also have to submit new variants and update reclassified variants to ClinVar at least once a year; submit all categories of variants they return to patients; attest to submitting at least 100 variants and at least 75 percent of them in the past year; and have CLIA certification or meet equivalent standards in another country. 

ClinGen's list launched initially in July 2017 with 10 labs that met minimum submission requirements: Ambry, ARUP, Center for Mendelian Genomics/University Medical Center Ljubljana, Counsyl, EGL Genetics (Emory), GeneDx, Illumina, Invitae, Partners Laboratory for Molecular Medicine, and the University of Chicago. Later that same year, Athena Diagnostics and Nichols Institute San Juan Capistrano (which are part of Quest Diagnostics), the Center for Pediatric Genomic Medicine at Children's Mercy Hospital and Clinics, Integrated Genetics/Laboratory Corporation of America, and Phosphorus Diagnostics made it on the list. Color and GeneKor joined in 2018, and Baylor Genetics was added in 2019.

Labs must continue to meet the criteria to remain on the list, but if they fail to do so, may lose their place in it. ClinGen plans to update the list again at the end of January.

This list has garnered the interest of insurers increasingly looking for ways to pay only for medically necessary, quality genetic testing. For example, inclusion on the ClinGen lab list is among the many requirements genetic test providers must meet in order to be in UnitedHealthcare's preferred lab network.

UHC launched the preferred lab network last year and announced that it had selected seven labs that had agreed to provide testing at a lower average cost and offer services aiming to improve patient outcomes. The four genetic testing labs included in this network are GeneDx, Invitae, LabCorp, and Quest, which are also in ClinGen's lab list for meeting data sharing standards in ClinVar. 

Before UHC, Aetna several years ago began requiring that labs submit their variant interpretations to ClinVar as a condition of participation in its lab network for BRCA1 and BRCA2 genetic testing. 

Over the last six months, Rehm has been speaking to other insurers and raising awareness of ClinGen's lab list. "I was surprised that most of them had never heard of it," she said, but noted that now that more insurers know about it, they're starting to think about whether they want to use the ClinGen lab list as a measure of lab quality.

My Gene Counsel's Farmer has also noted a lack of awareness of ClinVar among non-genetics providers. "I don’t think many non-genetics providers know about ClinVar or that they may need to look beyond the lab’s test report," she said.

Despite ClinVar's steady growth, there are still labs not willing to share variant data. "This is something I find disappointing, but rare," Nussbaum said.

Myriad Genetics does not submit to ClinVar, citing concerns about the quality of the data in the database. However, reproductive genetics firm Counsyl submitted variants it detected via its Foresight carrier screening test and Myriad, which acquired the firm in 2018, will continue to support submissions specifically from this test in ClinVar. Counsyl's name in ClinVar is even slated to be changed to Myriad Women's Health to reflect the acquisition.

ClinVar now also has a few more international submitters, suggesting that some genetics experts outside the US are also using the repository or at least know of it. For example, Systems Biology Platform, Zhejiang California International NanoSystems Institute was the first ex-US submitter. The University of Khartoum in Sudan, the Institute of Human Genetics at Cologne University in Germany, and several institutes in India have also submitted a limited number of variants. "We're going to see more growth [in ClinVar submissions] in coming years from the international space," Rehm assured.

Going forward, ClinVar must make efforts to increase the variant data from individuals of non-European ancestries, Nussbaum reflected. "We need to have a much better understanding of rare variants and their clinical impact in other ethnicities around the globe," he said. Invitae has ongoing projects to investigate the clinical significance of variants in South African, East African, and Latin American populations, and the company plans to deposit the data into ClinVar.

Lastly, one of the major milestones for ClinVar came with the US Food and Drug Administration's recognition in 2018 of the ClinGen Expert Curated Human Genetic Data. In recognizing ClinGen expert panel submissions to the database, the agency determined that the procedures used to classify variants "constitute valid scientific evidence that can be used to support clinical validity of genetic tests in future premarket submissions." As a result, ClinVar now also contains FDA-recognized variants that have been interpreted by experts within ClinGen.  

However, the utility of this FDA-recognized variant data in easing the regulatory burden for test developers is difficult to track, since most labs aren't pursuing the FDA regulatory pathway for their tests.

Nonetheless, Rehm is involved in submitting an investigational device exemption application for the NIH's All of Us program, which will return certain clinically relevant genetic test results to participants. The IDE application will have to describe how experts will interpret and classify detected variants using ClinVar and other resources.

"But we're not just using the expert classifications in ClinVar directly. We will check every variant [that will be reported to All of Us participants] and we could disagree with what expert panels have reported out if new evidence has emerged," Rehm said.

If ClinVar has shown anything, it's that variant classification is "complex and ever changing," she said.