In the spring of 2014, my colleague John Garcia discussed the potential benefits of genetic data sharing. He hypothesized that as researchers and laboratories share more and more of their findings, these data will result in a greater understanding of genetic variation within the community and, ultimately, better patient care.
A year later, how do things look? Even better than we had imagined.
At the 2015 annual meeting of the American College of Medical Genetics and Genomics (ACMG), we presented data from a recent validation study we completed in collaboration with scientists and physicians at Stanford University and Massachusetts General Hospital. (A manuscript on this study is now in press.)
One of the principal findings was that in 99.8 percent (973 of 975) of cases, Invitae’s geneticists came to the same clinical interpretation as scientists who had published prior results. Additionally, in genes for which there was sufficient data to make such comparisons, Invitae’s tests had only a marginally higher rate of variants of uncertain significance (also know as a VUS rate).
These are very good numbers. But beyond the numbers, Invitae’s variant interpretations were produced based solely on the aggregate of broadly available, nonproprietary data such as clinical/research publications and public databases like ESP, ExAC, and ClinVar, while the prior interpretation came from a single independent laboratory that relied on a proprietary database based on over a decade of experience. It appears, remarkably, that in just a few short years, the data-sharing community has amassed a knowledge base comparable in quality and size to the proprietary database that a single laboratory took many years to build.
While we were pleased to see the 99.8 percent statistic, we were frustrated by the 2 cases that resulted in a disagreement. We were frustrated not because we disagreed (occasional disagreement among labs is expected for challenging variants) but because we couldn’t assess why we disagreed. Could it be that the Invitae geneticists found additional evidence the prior lab didn’t consider? Perhaps we looked at the same evidence but weighed it differently? Without knowing the rationale behind the previous interpretation, we cannot assess which interpretation was more accurate and, most importantly, determine how to reconcile similar cases in the future.
Last year, while this study was still in progress, we submitted variants identified in this cohort and their clinical interpretation to ClinVar. With the recognition of the importance of the interpretation rationale, we have now updated those submissions to provide that additional information. In fact, Invitae recently submitted to ClinVar over 1,200 clinically observed variants and their interpretations. For the vast majority of them, we also provided detailed rationale for our interpretation and cited the sources of evidence. We hope that the additional data we provide can be used to critically assess the quality of our work and lead to better clinical management of patients, regardless of which diagnostic laboratory conducts a test.
We believe that genetic information is more valuable when it’s shared. Our community is already beginning to see the tremendous value of data sharing; just take a look at this April 2015 editorial in Nature. We look forward to playing an important role by continuing to contribute data in the coming months and years.