In mid-September, the International Collaboration for Clinical Genomics (ICCG), a partnership between the International Standards for Cytogenomic Arrays (ISCA) Consortium and members of the genetic sequencing community, was awarded funding from the National Genome Research Institute (NHGRI), a division of the federal National Institutes of Health (NIH), to continue its efforts to improve the quality of genomic healthcare through data-sharing and collaboration.
The work of the ICCG constitutes a major part of a new NIH-funded resource known as the Clinical Genome Resource, or ClinGen. Through ClinGen, ICCG is collaborating with other groups to design and implement a framework for evaluating which variants play a role in disease and which are relevant to patient care. We will work closely with the National Center for Biotechnology Information (NCBI) of the National Library of Medicine (NLM), which will distribute this information through its ClinVar database.
Shortly after the announcement, HCI Editor-in-Chief Mark Hagland spoke with Heidi L. Rehm, Ph.D., associate professor of pathology at Brigham and Women’s Hospital and Harvard Medical School, and director of the Laboratory for Molecular Medicine at the Partners Healthcare Center for Personalized Genetic Medicine, in Boston, and Christa Lese Martin, Ph.D., director and senior investigator in the Lewisburg, Pa.-based Autism and Developmental Medicine Institute at the Danville, Pa.-based Geisinger Health System, regarding the important work of the ICCG, and its implications for patient care going forward. Below are excerpts from that interview.
This announcement signals yet another element in progress taking place on the leading edge of medicine and science.
Christa Lese Martin, Ph.D.: Yes, this NIH-funded project has brought together three different efforts. Heidi and I are focused on facilitating the deposition of variants involved in clinical laboratory testing. There are clinical genetics laboratories all over the world that do the majority of genomic analyses; and currently, with the exception of our ongoing effort, which this new effort will help to facilitate, the knowledge and information from such genomic analyses remain pretty much siloed in individual labs and aren’t shared. So this sharing will allow multiple labs to have access to these rare variants, and thus, it will help the interpretive process to determine which of these variants are clinically relevant and actionable, versus those that are benign.
What is a rare variant?
Heidi L. Rehm, Ph.D.: One of the best-known variants is in breast cancer. So the breast cancer testing myriad looks at two genes, BRCA1 and BRCA2. And there’s a small subset of breast cancer that’s inherited and passed from generation to generation, perhaps 5 percent. So if a lot of people in your family have breast cancer, you might worry that it’s caused by a variant, a mutation, in one of those genes. We tend to use the term variant rather than mutation, because of the bad connotation of mutation. So Christa and I, as lab geneticists, try to determine whether a variant, a change in the DNA, is good or bad. Pathogenic versus benign variants.
Is this a communications channel?
Rehm: All these labs are doing these genetic tests, and are finding these rare genetic variants or mutations, because most genetic diseases are caused by very rare mutations. So if one lab finds a particular variant, they may report it in a report that says, we’ve found this variant, but don’t know what it means; it’s inconclusive. And the patient and physician will say, wow, we don’t know what this means. So if an oncologist orders a test and gets this test back that says “inconclusive,” and there’s a variant, sometimes that physician might misinterpret that and think it’s causing a disease. So our goal is, if we can get everyone to share disease information with each other, we can help people do analyses. We’re focused on rare variants causing genetic diseases; but over time, we’ll try to tackle variants causing common diseases, as well. So we’re asking labs to share this data into a central database.
How many labs are involved in this initiative?
Rehm: At the start, we have letters of commitment from two different groups. Christa has been leading a project working with cytogenetic labs connecting data on structural variants, and that group includes 190 labs.
Martin: We’re hoping that this grant will increase the number. So you have to agree to submit the data, and then have the sharing occur.
Rehm: And then we have about another 50 labs that do sequencing. So altogether, over 200 laboratories have agreed.
How will the information be shared? What are the mechanics of that data-sharing process?
Christa: The data-sharing will be centralized and accessible to all the participating laboratories via the National Center for Biotechnology Information, or NCBI, within NIH. ClinVar is the new database that NCBI is developing, and we’re working with NCBI to customize and develop that so that it’s useful for clinical lab personnel and others, to easily search the data and also classify the level of curation that the data has gone through. Is the information from one laboratory, or multiple laboratories? Has it gone through a structured peer review? That will be important information to know.
In other words, it will involve an online query process?
Yes, the ClinVar database is available online. And you start with a query based on either the gene or a particular variant that is known to be associated with that gene, or a genomic region, depending on the type of variance you’re looking at.
When did the ClinVar database go live?
Martin: It went live in April.
What’s new in all this, with the September announcement?
Rehm: Well, we now have funding for the project. Christa had a grant before, to get this kind of data for structural variant, and there are over 30,000 cases with structural variants that have been deposited and are available to the community. More labs will be depositing data, and then we’ll also have sequenced data.
Martin: We had a previous NIH grant for more of the structural data. We now have this broader grant, with a program called the Clinical Genome Resource, or ClinGen.
Rehm: It involves four groups. Three are the funded organizations, and the fourth entity is NCBI, with the ClinVar database. So we’re all working together to generate resources to support the interpretation of patient genomes. And that’s why we call it the Clinical Genome Resource Program.
Martin: And ultimately, we want to connect this with electronic health records [EHRs].
When is ultimately?
Martin: Well, it’s a four-year grant to start with. And Heidi, you can comment, too. But Heidi and I were involved with ACMG, the American College of Medical Genetics and Genomics—we already know there are some variants, when they’re detected, and that is something clinically relevant to patients’ care. Some of that data is already out there, so how to efficiently and effectively incorporate that into the patient health record, is the challenge—there’s still a lot of curation and evidence-based processes involved before we get to that point.
Rehm: We are already depositing genetic structure and data into the EHR. And through reverse patient data registries—as at Partners, we have one, that is basically a replica of the clinical ERH, but it can be searched and mined. And we actually have a repository effort to bring phenotypic and genotypic data together and associate that with
So one of the things that NHGRI has done a very good job of doing is bringing together groups working in overlapping areas. So already the fact that they’re bringing together the four groups that initiated this project is evidence that they’re bringing together groups that can complement and enhance each other.
There’s also a group called EMERGE, focused on getting phenotype and genotype into and out of the medical record. So our grant doesn’t necessarily have a very specific time for getting the EHR hooked up, it’s something we’re looking at, to see how we can consider bringing that data into different environments. The biggest challenge we have with this grant is that it’s not focused on a single institution. It’s not as though Christa and I are focused just on Geisinger and Partners. We’re working with hundreds of organizations. So it’s not a simple solution for us to get everybody’s data and get it integrated into the EHR. However, we do have an effort that involves people at Geisinger, Partners, and Boston Children’s Hospital, in a phenotyping effort.
We have a phenotyping work group chaired by David Miller at Boston Children’s. And the goal of that work group is to come up with approaches to enhance the collection of patient phenotyping. Sometimes, just getting a better paper form in which to get better checkbox information on the phenotype of the patient, as well as using NLP to do text-based gathering of data, that should be an improvement. However, that’s a low-tech-based effort. We’re also working on other methods, with several collaborators working on structured phenotyping, to develop new methods to do this, along with other people involved in bringing data into and out of the EHR.
We want to put genetic data back into the electronic health record to support clinical decision support. And in fact one of the grants given to Jonathan Berg’s group, and Christa's group, with Mark Williams, at Geisinger, is leading a project related to providing clinical decision support with genetic information, at Geisinger. We want to get phenotypes out of the EHR, but also want to put genetic results and clinical indications back into the EHR to support care to patients. But you can’t do that without collecting better data on these variants. You have to start there before you can say, well, now we know everything about this variant. Because right now, we don’t have enough information about what the variant means, so there’s nothing to deliver.