Researchers at the University of North Carolina, Chapel Hill, the Geisinger Health System, and the American College of Medical Genetics and Genomics have received an $8.4 million grant from the National Institutes of Health to define standards for assessing the clinical validity and actionability of genes and genomic variants.
The researchers expect the standards to help clear up confusing gene-disease associations caused by the use of multiple internally developed protocols for assessing clinical relevance, to establish standard protocols for determining clinical relevance, and to help doctors better understand the role of genetic variants in disease and how they can use this information to improve clinical care.
The grant, administered by the National Human Genome Research Institute and the National Institute of Child Health and Human Development, is one of three awards totaling more than $25 million over four years to fund the development of the Clinical Genome Resource, or ClinGen — an extension of the NHGRI's Clinically Relevant Variants Resource program that will provide authoritative information on genomic variants that are relevant to human disease and useful in clinical practice.
ClinGen's investigators have been charged to design and implement a framework of standards and informatics tools for evaluating genomic variants that are involved in conditions such as cardiovascular and metabolic disease and those that are relevant to patient care. This will include standard formats for gathering and depositing data into the National Center for Biotechnology Information's ClinVar as well as standards for determining which variants cause disease and which are medically useful, methods for curating and annotating variants, and computational tools for predicting disease risk (see related story this issue).
"It's a desperately needed resource for everyone that's doing clinical genome-scale sequencing," Jonathan Berg, an assistant professor in UNC's genetics department and one of the principal grant investigators, told BioInform.
"That's because in addition to sorting through conflicting information about variants' disease associations, researchers now have to deal with far more genetic information than they've had to in the past, Berg said. Before next-generation sequencing was born, most labs focused on curating and annotating a single gene or a couple of genes in depth. But with the availability of whole-exome and whole-genome sequencing, "you have labs facing the problem of … trying to decide whether a particular variant is clinically relevant in a gene that they may never have had any experience in," he said.
"If you simply mapped the variants that you found in your particular genome or exome sequence up against some of these databases, there would be a lot of hits that are at least putatively disease-causing, but when you look at the data behind them a little bit more carefully, its pretty clear that they are not pathogenic," he added.
The grant "is partly about starting to clean up some of that confusion about which variants are really pathogenic and which … are of uncertain clinical significance," but it's also about setting up infrastructure that other curators can use in their projects, Berg said. "We are obviously going to avoid trying to duplicate [existing] efforts," he added. "[We hope] to learn from them and try to make a generalized curation environment that can then be crowdsourced to many curators."
Berg and his colleagues have been tasked with creating standards to categorize variants as pathogenic, benign, or of unknown significance, as well as to determine their clinical actionability. For their part of the project, they will set up 12 clinical working groups composed of experts in a variety of clinical domains who will be responsible for curating genes and gene variants that are involved in cancer, cardiovascular disease, metabolic disorders, and other ailments.
These groups will evaluate existing curation efforts and explore ways to provide both resources and infrastructure to support those efforts, Berg said. They'll also be responsible for identifying gaps in the curation process as well as areas with no current curation activity and coming up ways of addressing the deficits.
The working groups will also assess the clinical validity of existing gene-disease associations in ClinVar and other publicly available resources by creating ways to categorize them based on the strength of the association. One possible classification scheme, Berg said, might be to group genes in tiers based on the strength of their disease association.
For example, genes that are well-known to cause a certain condition might be placed in the top tier of clinical validity; those that have played a causative role in a handful of cases might be placed in the middle tier of clinical validity; while genes with disease associations that are based on weak data or single cases would be in the lowest tier.
For this part of the project, the researchers are working with the developers of the Online Mendelian Inheritance in Man database, Berg said, hoping to benefit from their experience in defining these sorts of gene-phenotype relationships. They're also hoping to leverage the expertise developed by projects such as the Evidence-based Network for the Interpretation of Germline Mutant Alleles, or ENIGMA, consortium, which focuses on determining the involvement of variants of unknown significance in the BRCA1 and BRCA2 tumor suppressor genes, as well as other groups that have specialized in curating individual genes.
The UNC-led team will also develop a method of curating genes based on clinical actionability. Here, they are developing an actionability score that will be based on an expedited evidence review and a "semi-quantitative metric" developed in collaboration with the Evaluation of Genomic Applications in Practice and Prevention, or EGAPP, Berg said. This approach would provide an "updatable, transparent, evidence-based score for any given gene that would essentially describe how actionable it would be," he said. A score like this could help scientists and clinicians make decisions about how to appropriately respond to incidental findings, for example.
A final component of the grant will be to explore the best ways to make curated genetic information a part of clinical care by integrating it with electronic medical records so that physicians can access and use the data as needed. UNC's collaborators at Geisinger are coordinating efforts on this front to build modules that link the Epic Electronic Health Record system to ClinGen as well as organize pilot projects to test them internally and at other health institutions that use the same EHR software.
The team will consider a number of options for integrating the two data sources. One approach might be to actively alert physicians to genomic information that may be pertinent to their patients' care and treatment as it becomes available. Alternatively, they could simply link the EHRs to the data so that physicians have the option to access the information when they choose to do so, Berg said. This part of the grant will also focus on putting the variant information into a format that EHRs can consume and digest, he said.