NHGRI’s Clinical Sequencing Exploratory Research (CSER) Consortium – Lucia Hindorf

NHGRI’s Clinical Sequencing Exploratory Research (CSER) Consortium – Lucia Hindorf


Lucia Hindorff:
I’ll take a moment, maybe, to thank Dr. Ginsburg and Dr. Manolio and the organizing committee
for inviting me here to talk about the NHGRI Clinical Sequencing Exploratory Research Program.
I am the NHGRI program officer for CSER, which means that that are number of grantees and
many, many investigators who actually do the hard work and heavy lifting. I’ll acknowledge
them at the end, but I do want to acknowledge at the outset that this is a tremendously
collaborative opportunity. So, okay, here’s my slides. Let me, right
button. I’m clicking. It’s not working. There we go. Okay, so this is a familiar slide to all of
you, so I don’t need to spend a lot of time on it, but let’s take a look at similar data
in a different way. So, in clinical sequencing what we’re trying to do is to bring the value
of clinical sequencing to the treatment of patients. And so if you consider a clinical
decision that has to be made at the level of a patient, and you look at the increasing
amount of information that’s going to be gleaned, not just from sequencing and SNPs and so forth,
but the future promises to bring in functional genetics and possibly proteomics, we’re talking
about a huge number of facts per clinical decision per person. I should mention to you
that this is a schematic. It is not data based at all, so take these numbers with a grain
of salt. They are courtesy of Dan Masys. And let me also mention that as the amount
of information that we collect in the clinical setting increases, the amount that we are
going to actually find to be clinically relevant is likely to be relatively constant, so the
realm of traditional health care will probably only consider a very small subset of this
information. And furthermore, human cognitive capacity, at least as far as we know, is really
quite limited, and so I think what we’re struggling to deal with in the realm of clinical sequencing
is this gap, right? So I think this brings at least a few implications. One that I think is that we have to be convinced
that this additional data is necessary to collect, so there has to be some clinical
utility to it. A second implication, I believe, is that there needs to be some serious filtering
going on here. I think that the ability to filter for patient context, disease context,
and so forth, is going to be critical, not to mention figuring out, at least at the sequence
level, which variants are actually actionable, right? And then a third, I think, implication
is that obviously this small set of information that’s relevant to the decision at hand will
need to be communicated in an intuitive way to the patient and to the physician, because,
obviously, they will need to be aware that all of this information is going to be assayed,
but, you know, only some small subset of information is going to be directly relevant to them. Okay, so those were just a few issues, and
obviously this is a huge area that a number of us are struggling with. In 2010, and again
in 2012, NHGRI issued a request for applications to fund grants studying the challenges inherent
to incorporating clinical sequencing into clinical care. I’ve mentioned to you that
I think the challenge really here is to look at the clinical sequencing in the clinical
workflow. We touched on a number of areas in the RFA, the generation, and the interpretation
of the genomic sequence data, both kind of the technical challenge as well as the annotation
challenge and interpretation challenge, and then obviously communicating the results back
to the patient in the correct manner and in the appropriate manner is going to be incredibly
important to study. We also, similarly to what Anastasia mentioned in the newborn program,
made the ethical and psychosocial implications a primary aspect of this program. To tell you more detail about each grant,
there are three different components, similar to the newborn sequencing program that you
just heard about. Project 1 is what we term our clinical study. There needs to be a clinical
rationale. Project 2 is the sequencing analysis and informatics pipeline. And then Project
3 is ethical and psychosocial implications. One thing that I didn’t mention to you on
the previous slide is that you may notice, actually, that nowhere in here is any particular
aim about discovery. So we recognize that discovery is an important aim, but for this
particular program we wanted to focus on common challenges which are actually, I think, more
important, and, in fact, more general for the National Human Genome Research Institute
to address. These challenges are not disease-specific. Okay, so back to the structure of each project.
Okay, so, obviously, these are interdisciplinary teams. We have a management structure for
each team, and then if you look at the overall consortium, I’ll tell you a little bit more
about these studies, but we funded nine of these interdisciplinary projects. We include
our intramural collaborators from the ClinSeq program who sit on a couple of our working
groups as they have extensive and long-standing expertise in this area. I’ll tell you a little
bit more about this, but we also funded a coordinating center to help facilitate the
high priority goals of the consortium, and then we also are fortunate to count as our
collaborators a number of grants that were funded specifically to look at return of results
from the ELSI Program and NHGRI. So, here is a summary of the U awards, the
kind of Project 1, Project 2, Project 3 organization that I showed you before. You can just take
a quick scan here. Four of these grants study cancer; three of them are doing actual sequencing
of tumor-normal pairs, three of them we are co-funded by our collaborators at the National
Cancer Institute. You’ll notice that we also have representation of pediatric diseases,
as well as healthy individuals, so, in particular, the Brigham and Women’s grant, as well as
the Kaiser grant, are actually sequencing individuals who are not showing any overt
signs of disease. The R grants, focusing on the return of results
that I mentioned to you before, are shown here. They are a combination of both sort
of empiric or data generating studies as well as normative or more philosophical approaches.
Our coordinating center is housed at the University of Washington and headed by Gail Jarvik and
her colleagues. We funded the coordinating center because we recognize that the value
of a consortium like this would be to bring the various sites together, and to identify
similarities and differences, common challenges to be addressed, and common opportunities
to really help move the field together. I think we realize that clinical sequencing
is obviously being done in a number of centers around the United States, and, in fact, around
the world, as we’ve heard. And so it was important to us to really figure
out how best to use this critical mass of individuals, and so the coordinating center
has a range of expertise. Their key activities so far have been to coordinate the working
groups in cross-consortia collaborations. As needed, they coordinate, initiate, and
even lead some high-priority CSER projects. They are synthesizing site-specific variant
pathogenicity data and gene lists; that’s one product that we hope to bring forth for
the community, is to share what each site is doing and what the consortium is doing
as a whole in terms of calling variants — of annotating variants, excuse me, and then defining
predetermined gene lists for each specific condition. And they also coordinate logistics,
and in sort of coordinating the work of the consortium, help us raise the consortium visibility. So each site is responsible for achieving
their specific aims for their specific disease context and patients, but we also rely on,
heavily, in fact, on the work of working groups. This is a similar structure to probably what
you heard from the eMERGE consortium yesterday. We have a number of working groups in a number
of areas; some of them are quite technical, such as sequencing standards or electronic
reports. Others are more sort of actionability or return of results based. I think this is
a good representation of the type of work that is being done across the consortium.
We do have a cancer working group that is just getting up and running, so we hope to
do some more specific work in that area as a consortium. So, here is what the recruitment looks like
as of December 2013; so, so far, about 1,100 patients have been approached about enrolling.
We’ve actually sequenced — oops, sorry — consented almost 500 of them, and then sequenced about
nearly 200 of them. You’ll note here that we’ve also enrolled physicians, so it’s really
important, I think, to understand the physician perspective in terms of generating the data,
interpreting the data, and returning it back to patients. And so the Project 3 individuals
are studying — the physicians — quite carefully, in fact, that a lot of them are doing audio
transcripts and so forth. I think a good analogy is that the technical side, the genome sequencing
side, is sort of, you know, mining every single variant, and then some of the clinical teams
are trying to interpret variants at sort of the sequence level. And the ELSI folks are
really doing the same thing at sort of, you know, the — each sort of word of the interview,
they’re really trying to look for themes and commonalities in terms of what physicians
want and how they respond to information, and how they best communicate results back
to the participants. So, I will just mention that we still have
a lot of work in progress. We are, I think, learning that one size does not fit all, and
I think, you know, while we hope to bring forth some best practices for the community,
we are discovering that there are a lot of different and equally valid approaches. For
example, six of the initial sites that were funded were asked about their practices of
reporting incidental findings, and you can see here that all six report the incidental
findings, half of them include the incidental findings in their primary indications, and
half actually have a separate report, and these decisions are also made by consensus
— generally made by consensus among the clinical teams, so there are many perspectives. Half
of the sites actually allow patients to opt out of receiving medically actionable incidental
findings, again for various site-specific reasons, and well-justified ones as well.
So you can see that there is some variability here, and we’re working to understand sort
of why people made these decisions, who makes the decisions. Are there — is there a clear
way going forward that we can recommend to the community, or is it more of an issue of
kind of defining what factors each group needs to consider in terms making their choices
about how information is returned? Groups are also displaying some sort of heterogeneity
in terms of how they return variants. And obviously, I think one big challenge for
the field as well that we are hoping to address within CSER is what is sufficient evidence
for pathogenicity? I think this is really a critical area that is not unique to CSER,
and we are eager to collaborate with others, I think, in this regard. Some examples of recent work from our consortia.
There have been two recent working group papers, one focused on summarizing the similarities
and differences, and defining challenges in terms of identifying actionable genes within
the consortium. So this is a summary and descriptive paper from the first six CSER sites in terms
of actionable genes. There is also a working group paper on approaches to integrating genomic
information into the electronic healthcare record. We have a number of other papers that
are published or co-authored by subsets of the consortia or individual sites, and I’m
just highlighting a few here. There was an article written by a number of folks in response
to the ACMG recommendations on incidental findings, a exercise to actually annotate
incidental findings in NHLBI exome sequencing project, and then one example of the discovery
study in cancer where a whole exome sequence — oh, sorry, this is — this is describing
the clinical pipeline — clinical sequencing pipeline from one of our sites that has extensive
experience, and then here’s the discovery study identifying several EFR mutations in
hormone-resistant breast cancer. So we are doing some critical discovery as well. So, I could not give this talk were it not
for the contributions of numerous people. I’m showing you here the U grants, the first
phase that were funded, the three that were funded in the second phase, as well as our
— the PIs of our grantees. Also, I’m fortunate to count among our collaborators a number
of people at NHGRI that have helped us really lead and champion this effort, in particular
Brad Ozenberger, who is now at Washington University in St. Louis, and then our collaborators
at NCI as well. I’m happy to take any questions at this point.
Thank you for your time. [applause] Male Speaker:
So how does your study differ from the TCGA? Lucia Hindorff:
So, my understanding is TCGA is — they’re collecting samples from a number of cancer
— individuals with cancer, but it’s not in the setting of their routine clinical care.
And the purpose of CSER, first of all, is much — it goes beyond cancer. It’s — the
lessons are not cancer specific, although we do have cancer sites, but it’s really to
figure out how to generate the information, how to interpret it and return it within the
context of the clinical workflow. So I don’t know if anybody else from NHGRI wants to comment. Male Speaker:
[inaudible]. It’s all discovered; it stops there. Lucia Hindorff:
It’s all — right, so it’s all discovery. Male Speaker;
Can I ask maybe any impressions or data on patients and providers, and how adaptable
sequencing has been to the provider community in particular, and the — you know, I don’t
know if there’s a little discordance between the number of people, why people haven’t agreed
to do this, if there’s any insights into that as well. Lucia Hindorff:
I’m going to try and tease some of that apart. So maybe the acceptability of sequence to
physicians and their — Male Speaker:
Their — you know, how readily they are, you know, really prepared to actually utilize
it in their practice, and what do we need to do to really facilitate that. And also
the patients, what are the most common reasons for them not to want to participate? Lucia Hindorff:
Okay, so I think in terms of the physicians, we are fortunate to have a pretty big diversity
of, I think, clinical domains represented within the CSER consortium, and some of you
that are involved in CSER may want to comment on this, but my understanding is, you know,
it really depends on the — not just on the clinical community in terms of cancer versus
pediatrics. It really depends on one’s institution, what sort of the climate is for being receptive
to clinical sequencing. So NHGRI’s visited, we’ve had the fortune to visit a number of
CSER sites, and, you know, many physicians really are on board because they see this
as sort of the next sort of step in personalized health care. Many, through, I think, recognize
that there are challenges, like really, really critical challenges, where they don’t have
a lot of time; the clinical utility, perhaps, is not as evident for genomic sequencing yet,
but I think, you know, for these nine sites at least, there’s a lot of promise there,
I think. Many individuals sort of see this as a research endeavor, which it absolutely
is. It’s clinical research, and I think we’re learning the lessons in order to move the
field forward. So I have to say that at least in the nine
sites that we’ve heard about so far that we fund, you know, the physicians have been receptive.
I think it does take some effort to recruit some patients in light of their busy workload,
but many of them are wanting to do it. I can offer one particular example, the pediatric
cancer study from Baylor, for example. The physicians there are used to enrolling their
patients in clinical trials, and so, you know, the value of clinical sequencing in that context
I think is seen as much higher, and there are similar examples throughout the CSER consortium. And then in terms of patients not participating.
In general, actually, we have heard that a lot of patients are really receptive to this.
I think once they get passed the idea that, you know, you’re sequencing the human genome,
you’re sort of actually learning about an individual at sort of the nucleus high level,
once you get past that scientific explanation, a lot of them really do see sequencing as
perhaps an end to their diagnostic odyssey. Some of these individuals have been going
from visit to visit, test to test, trying to get an answer to, you know, what is causing
their condition, so that’s one example. I think we haven’t seen so far people decline
participation due to the fact — due to return of results. Actually, a lot of them — or
results being placed in their medical records. I think once people — once the researchers
explain to them what the value of the information potentially is and how it will be interpreted,
many of them really don’t see many concerns with that approach. So, so far, early signs
are that it’s reassuring. Female Speaker:
Yeah, but you have fewer than half of patients approached enroll. Lucia Hindorff:
That’s true. Female Speaker:
That does seem low. Lucia Hindorff:
That’s — so we do have — so we do have a pipeline. I think what I’m showing you is
that there is a process by which people are consented. They are putting in the sequencing
pipeline. They have to come back to have their results returned. Female Speaker:
So they may be in a purgatory. Lucia Hindorff:
And in fact — and the other important thing that I want to add is that in the early days,
especially at each site, there is a process for determining which variants to return,
right? That’s not a trivial process, and I see Heidi back there, so maybe she can comment
on really how long it takes to actually call a variant, but, I mean, that is a process
that takes a substantial amount of time. I think it’s shortening, but that has to be
factored into as well in terms of who actually gets sequenced and the results returned. Heidi Rehm:
And I just do want to comment that we have had patients decline because they do not their
result in the medical record. So we do have a proportion of patients that, upon enrollment,
once they go through and find out that a criteria for enrollment is that their results will
be placed in the medical record, they have declined for that reason. Not — you know,
the majority do continue to enroll, but I just wanted to make that clarification. Lucia Hindorff:
It’s not universal, yeah, that’s a good point. Heidi Rehm:
And on the physician enrollment side, they are required to undergo six hours of training,
two that are didactic, and then four hours of self study, case modules that they have
to take quizzes on. They get CME credit, but there is, you know, the time commitment that
is the major barrier for the physician involvement, though I think it’s usually mainly just a
time commitment question and not interest in general in the topic. Female Speaker:
So, I guess what is returned at one site might be considered incidental to the phenotype
of interest at another site, or what? Lucia Hindorff:
So, each site does make their own decision about what constitutes an incidental finding,
so you could find your — you know, there are differences in terms of what’s returned,
so it’s possible that what one site that considers a variant crucial for primary phenotype is
considered an incidental finding in another study because they’re not recruiting patients
from the same phenotype. Female Speaker:
And I guess you are not trying to make people harmonize their procedures at this point,
but are they at least keeping tracks of which findings are returned and which genes are
considered incidental in some public place where we — Lucia Hindorff:
Yes — Female Speaker:
— could access that information? Lucia Hindorff:
So it’s not yet public, we’ve just now started to collect the gene list that are actually
the diagnostic gene list, as well as the incidental findings. So I think, again, I think it remains
to be seen whether or not, you know, there is a consensus. I think there are also plenty
of other organizations and societies that want a way in that are probably more appropriate
that this research consortium, but certainly we are looking at that very critically and
plan to share that information as well. Male Speaker:
As a participant in the Philadelphia site, I can answer Mary by saying, you know, if
you’ve seen one CSER site, you’ve seen one CSER site, that they are very different, and
hopefully the working groups will amalgamate some results. I just want to emphasize a couple points that
you’ve mentioned. One is that we are very concerned about why people don’t participate
and are screening them — querying them as carefully as we are the participants, and
the second thing is that we found one of the reasons that parents opt not to participate
is that this is not a substitute for regular medical care, and so our patients with congenital
hearing loss, let’s say, are getting cytogenomic arrays and full panels of known hearing loss
genes, and then this is an add-on, and oftentimes, parents say, “Look, you’ve done what you can,
that’s enough. We don’t want to invest the additional time.” Lucia Hindorff:
So, right, so the extra time burden of participating. Male Speaker:
Thank you. Lucia Hindorff:
Okay, thank you.

You May Also Like

About the Author: Oren Garnes

Leave a Reply

Your email address will not be published. Required fields are marked *