Tip:
Highlight text to annotate it
X
Iftikhar Kullo: Thank you. So if we could go to the next slide,
I wanted to thank people who contributed to this presentation, my co-chair, Gail Jarvik.
Great input from the panel members: Larry, Susan, and Lisa. I would like to thank them.
And I also got some input from the Return of Results Working Group, which the members
are listed here.
So I'm just going to briefly tell you about the evolution of ROR in eMERGE and then talk
about some future directions. So next slide, please. And next.
I think in the initial return of results deliberations in Phase I, obviously, the unique thing was
return of results in the context of EHR, and this group kind of summarized its recommendations
in the favor of genetics and medicine; looking at return of results that may vary in the
context of each site or other factors such as age; highlighting the importance of the
need to generate evidence of clinical validity and actionability before returning results;
what are the appropriate methods of return of results; and the need to -- and to obtain
opinion across diverse sites, and also input from lay community and advisory bodies. So
this group kind of started the process.
Next slide, please.
And then in Phase II, it was clearly a focus towards implementation, and this is being
done in site-specific projects at each site and also at network projects. And this is
just -- not a full but a partial listing of the genomic medicine pilot projects that are
being undertaken just to show you the flavor of these pilots. Some are including genetic
risk scores, for example, for macular degeneration or for myocardial infarction. Others are using
just single SNPs, ApoL1, for example, or HFE, or Factor V Leiden. Geisinger is doing a study
of whole genome sequencing in trios for diagnostic odysseys. And then several of the pediatric
sites are looking at returning CYP2D6 or hypothetical CYP2D6 results to patients or parents.
Next slide, please.
So, this is an example of an EHR-based genomic medicine pilot study. Myocardial infarction
is a leading killer of people in the United States. It often presents as sudden death.
We do very poorly with conventional risk factors for predicting myocardial infarction, so the
population implications are huge. And if there's any way, any way, however modest, where we
can improve the ability to refine risk stratification, that has huge implications, and therefore
we are conducting this pilot study, or the Myocardial Infarction Genes Study, of giving
patients a genetic risk score based on 28 SNPs that are related to susceptibility versus
just giving them conventional risk factor information. And this is communicated by a
genetic counselor using the electronic medical record, and they follow up with an M.D. with
it, and then we assess them at three and six months for endpoints such as LDL cholesterol,
weight, activity, diet changes, and also some other assessment of how patients understand
these results and what they do based on these results.
Next slide, please.
And these are a few examples of network-wide projects. You've heard a lot about eMERGE
PGx; I won't go into that. We are trying to assess the phenotypic correlates of copy number
variation and even larger chromosomal abnormalities, what the phenotypic correlates are on medical
record, and this was already highlighted in the project to look at hemochromatosis variants.
It illustrates the power of eMERGE in that we have a total of 1,459 individuals across
the network that have one or the other of these variants, and so we can look at the
pleiotropy and penetrance.
Next slide, please. Next.
And so for return of results, there's a perfect storm in a way because there's been so much
investment by institutions and biorepositories, and this has coincided with the HITECH Act
and the need to implement electronic medical records, and then the remarkable advances
in genome sequencing. So all of these factors are really going to toil [spelled phonetically]
us over the next many years the issue of return of results.
Next slide, please.
And so, unique -- the eMERGE network is uniquely made to address some of these issues, not
only in the context of genomic discovery, which we already discussed the many questions
we can answer in terms of pleiotropy or longitudinal phenotypes, but also in implementing genomics
in the electronic medical record, whether it's storing the data, visualizing it, linking
it to decision support, dealing with incidental findings, reinterpretation, looking at outcomes.
And then these kind of somehow merge together in the learning EHR paradigm that was alluded
to earlier, and so I think eMERGE is uniquely positioned to do both the discovery, implementation,
and some of the aspects that are in between these two paradigms.
Next slide, please.
And in the context of discovery, I think this is a huge area. I can't really summarize it
in one slide, but in the context of EHR, the questions were already discussed in the previous
presentations, documentation in the EHR, communication to family members, unique problems in the
pediatric setting, what are the patient preferences, what about consent, the mechanism and timing
of ROR, the incidental findings. So all of these are important questions, and together
with CSER, we have the CSER Cohort Consortium, we've had some initial deliberations, and
Gail is leading an effort to summarize some of the recommendations in a manuscript that
[inaudible].
Next slide, please.
In the context of implementation, obviously there's a question of what could be returned.
And you could think of it as listed there, or in the [unintelligible] paradigm suggested
by Burg [spelled phonetically] and colleagues, we could return copy number variation, recessive
mutation, single nucleotide variants that are relevant to disease susceptibility, or
pharmacogenomics, or genetic risk scores as they mentioned. And then, of course, the whole
issue of sequencing and what comes out of sequencing, actionable variants as well as
incidental findings. Sequencing can be genome-wide sequence, or all exome, or targeted. There's
a fair bit of effort, then, to ascertain the clinical validity of these findings, and so,
jury concept where medical experts and others decide on that aspect, the need for this to
be clear, certified if it's going to be in the medical record. And sometimes we will
also need to resort to statistical modeling and -- to really create the correct statistic
for what the genetic variant or the collection of variants implies for risk. And this has
to be integrated in the EHR, and then we have to deal with the ELSI issues. And some of
the legal aspects that were highlighted were actually summarized in the paper we wrote
in the EHR team issue, and I will refer panelists to that manuscript. The question of storage
and reinterpretation, again, very important when we're looking in the context of implementation
and the clinical decision support issues, as well as then trying to assess outcomes
of all of this effort, and perhaps, initially, just as implementation outcomes but, of course,
longer term and be able to do clinic outcomes as to -- that are patient-centric.
Next slide, please.
And so these are some ideas that were thrown around for developing a framework where we
could assess some of these challenges. For example, let's talk about whole genome or
whole exome sequencing as Debbie mentioned, and the unique aspects would be the multiple
phenotypes that we could look at to correlate with this data. The issues of penetrance,
pleiotropy, pediatric considerations, pathogenicity; all of these eMERGE could unique address.
We could think about targeted sequencing. This could be the 56 ACMG genes. Again, issues
related to pathogenicity, informing kin, et cetera. There's this very widely-used genetic
testing in the clinics which is usually doing candidate gene panels. For example, at our
institution, we do a whole lot in terms of cardiomyopathies, hypertrophic or dilated.
We talk about aneurysms, aortic aneurysm syndromes, sudden death syndromes, pediatric syndromes.
So these are often very expensive, they take months to come back, and perhaps where eMERGE
would contribute in focused genomic medicine pilot projects. And, of course, we shouldn't
forget high-density genotyping which has been our forte for eMERGE, and many of the sites
not only have common variants, but also with the more affordable rare variant CHiPs, there's
a fair bit of data on rare variants as well, so correlating these to EMR and the right
phenotypes would be quite useful.
Next slide, please.
And I think I'm going to quickly go this because these were very nicely discussed. Within the
framework of EHR implementation, we have to address participant privacy and potential
vulnerability to adverse social consequences, and therefore appropriate consent to include
genomic data in the EHR has to be there. We discussed recontact to ascertain preferences
and over time, and how this could be done electronically, I think, is, again, an area
ripe for investigation.
Next slide.
And we, of course, need to learn more about how stakeholders perceive all of this activity,
whether they're patients, parents, guardians, family members, care providers, laboratorians,
investigators, or biorepository scientists.
Next slide, please.
And I would submit that the eMERGE consortium as a whole, and within eMERGE, the ROR working
group really is the final transducer of many of these activities for us and to be at the
leading edge of implementing genomics, whether we interact with payers or regulatory bodies,
with ACMG or EGAPP, with other NHGRI activities such as ClinGen, ROR, or CSER, and with other
entities that are looking at making genomic information scalable machinery that will -- and
thereby -- and try to be integrated into the medical records. So I think we have a unique
potential strategically positioned in this area.
And my last slide is to summarize the unique features of eMERGE to address these knowledge
gaps and challenges, the linkage to EHR with deep and diverse phenotypes, the diversity
of clinical settings and electronic medical records, the diversity of genomic information
ranging from sequencing to high-density genotypes, the ability to look at best practices for
implementation, and finally a cohort at this point of more than 50,000 that includes pediatric
patients. So I'm going to stop there.
Male Speaker: And Larry, you can continue, even with the
same microphone.