Personalized Genomics Informed Medicine - Rex Chisholm

Rex Chisholm: All right. Well, thanks very much, Tim. My
job here is to actually now set the stage for what it is we're going to be doing for
the next day and a half or so. And, of course, the reason we're all here is this whole idea
of personalized, genomics-informed medicine. Eric didn't -- if you noticed on Eric's slide
that showed the strategic plan, that final right-hand piece of it, there were just a
few dots in the 2011-2012 range. So the folks here are really at the bleeding edge. And
what Eric didn't say is that at the, for example, the retreat that they held to develop the
final version of the strategic plan, there was actually a fairly robust debate about
whether it was even time for us to be doing this.
And I think, to address the point of whether it's time or not, I happened to turn on the
television to look at the news this morning as I was getting up, and lo and behold, there
was an ad on the TV from a local organization that talked about using genomics to personalize
the medicine at some local health care organization that I'm not even familiar with, not from
this area. The fact that that was already on the TV, just this morning, says that the
time is really right for us to actually be here and thinking about how to do it right,
because if we don't think about how to do it right, there's going to be a lot of people
out there doing it, and what we're going to have to deal with is cleaning up the mess
of hype that doesn't actually deliver what we need. So what we do here today is really
much more important than just a few dots at the bleeding edge of the NHGRI strategic plan.
So, as you heard from Eric, we have really three goals for the meeting: to collect potentially
-- how do we think about collecting potentially clinically relevant variants; how do we decide
whether they're actionable or not; and then how do we provide this information for clinical
use? So I just want to pose a few questions for us to be considering as we hear all of
the talks today and tomorrow, and we've built a lot of time into the schedule for discussion.
And you'll notice that several of the sessions are a moderated discussion that we hope to
actually really engage the whole group in the room in a real discussion to think about
how we'll come up with answers to these questions. Now, while it would be great to actually answer
all those questions in the next day and a half, that's probably a tall order. But if
we can at least get some clarity about the questions, about the scope of the questions
and some outlines for how they get answered, that would make this a very productive meeting.
So, first of all, what are the criteria that need to be met to consider a genetic variant
or, I must say, I think a lot these days about patterns of genetic variants where, you know,
we've got these whole GWAS studies that show a SNP that has a, you know, 20 percent increase,
a 20 percent or 1.2 relative risk associated with it; how do we use that? But if you actually
put it together with, I don't know, 33 other SNPs, each of which have some significance,
you might actually end up with something that produces a very usable risk score. So we really
need to think about what criteria. The other thing that we need to think about is a paradigm
shift: are we going to be able to actually do a randomized control trial for every single
one of these variants? So we really need to think hard about the criteria, and how do
we do that.
Do we really have available adequate datasets? Tim showed in the U.K. they're thinking about
how to hold this; we're going to hear a whole session where we talk about what kinds of
resources are currently available. Are those adequate? Do they need to be modified? How
do they need to be improved in order to be able to provide both datasets and accessibility
through databases that are open and available? And that provide, I think, importantly, the
supporting data that gives the evidence for why things might be clinically actionable.
So that's another important question for us to consider.
When is it necessary to integrate that data, those bits of evidence, into an electronic
health record and then present it to a physician in a way that's meaningful? That itself is
a huge task. Physicians already report what they call "alert fatigue" from too many things
being popped up in the screen of their electronic health record. So how do we provide this additional
information in the context of environment where physicians are stressed in terms of
time and they're stressed in terms of what kind of data they need to transmit in a typical,
seven to 15-minute interaction?
How do we create a dynamic loop that recognizes that this is really still an area of very,
very active research? As we see more and more of these DNA sequences being done for both
research purposes and for clinical purposes, how do we make sure there's a feedback loop
that adds that new information back into whatever datasets exist, version it so that the users
at the end of the process actually know what version they're looking at and whether that's
actually got the updated evidence and information.
What kind of decision support needs to be available to physicians, and what kind of
education do we need to be thinking about in order to educate the physicians to use
this data? If you're going to talk to most practicing physicians and ask them what they
would do if somebody brought in a particular genetic variant that they had, that they said,
"I have no idea." And that’s a big hurdle, to go from there to somebody that might come
in with a whole genome sequence and a list of variants that they have to deal with.
So we've got a big challenge ahead of us in terms of thinking about all of those. So these
are really topics for discussion today and tomorrow in this meeting. And I'll throw out
one that's not a topic for discussion but one that we've already heard in both Eric
and Tim's brief presentations, and that's that as we have more and more of these sequences
appearing that were done in a clinical setting, where does the line between clinical action
and clinical activity and research exist? And I think the blurring of that line is going
to be a critical factor that we're going to have to deal with. And actually NHGRI has
some other meetings going on, one Monday and Tuesday that I think some of you in the audience
will be at, which will also be thinking about that question of how do we deal with there
no longer being a bright line between clinical activity and research.
So what we're going to embark on, then, is a series of workshops. And again, I want to
emphasize that these are really meant to be interactive, so we really hope everybody will
give us the best ideas they have, that provide perspectives on the existing genetic variation
resources that exist, how well do they meet our needs, are there things that need to be
added to that? Could approaches for identifying variants for clinical use, how well do we
do with that; when you actually sit down and try to think, what variants you might actually
want to act on clinically, it's a pretty challenging problem. The answers are not so clear, and
we're going to spend a lot of time trying to work that through. Developing consensus
for actually not only figuring out whether they're actionable but figuring out, are they
almost actionable, are they certainly actionable, is going to be a really important topic for
us. How do we create that loop? How do we make sure that the benefits of all of the
wonderful research that's being done actually gets fed back in some kind of consistent feedback
loop into this collection of actionable variants? And then, how do we actually present that
in a clinical setting and use it effectively?
So, a lot of big topics. I think it's going to be a really exciting meeting. Just want
to emphasize again the importance of everybody to participate and give us your best ideas.
So, with that, I think we'll begin our first session, which is looking at existing resources.
And I'm going to invite Matthew Hurles from the Sanger Insitute, who's going to moderate
the first session.