Tip:
Highlight text to annotate it
X
Stephen Chanock: Good morning. It's certainly a pleasure to
be here. I'm Stephen Chanock, and one of the co-directors of the Center for Cancer Genomics.
And Lou Staudt, the other acting director, will speak in just a few minutes.
It's really a tremendous pleasure to be here, and to be able to be part of this particular
meeting at a very important juncture in the time the lifespan of TCGA and what will come
after that, and I want to talk about two things briefly today. One is the tremendous value
of TCGA, particularly in the realm where I come from, which is more germline and asking
questions of the relationship between germline and somatic alterations, and I think Lou is
going to give you some very exciting things that he's seen. And then I'm going to intersperse
a few comments about what some of the discussions are right now, as we're going forward and
thinking about exactly what kinds of things the NCI is going to support, and in what ways
we'd like to do it, and those discussions are very active at this time.
So, let me just start with a very exciting set of analyses that came out of one of the
first TCGA analyses: an ovarian cancer analysis, where there was a paper in JAMA not so long
ago, where they were looking at not only the germline, but the somatic alterations, and
particularly asking the question of BRCA1 and BRCA2. And these generated what I think
are very important key preliminary findings, and this is really, to me, the tremendous
value of TCGA in establishing hypotheses and establishing findings, that then larger data
sets will clearly go ahead and hopefully replicate and be able to extend and potentially push
into clinical implications. And I think this is an example where the findings between the
TCGA study and a follow-up study that I was very intimately involved in certainly suggests
that if we look at the BRCA1 germline status and probably also the somatic alterations
as well, this may have a tremendous implication with respect to survivorship in ovarian cancer
and may be a factor to consider in the design of studies and particularly certain kinds
of trials that are going to potentially go forward.
So I think this is a very nice paradigm of thinking about where TCGA goes from discovery
to characterization to eventually some, you know, version of clinical implications. And
we know that those clinical implications, as much as we like to say that's where we
want to go, those are hard and arduous, and they're going to take time, and they're going
to take a multitude of studies and various activities to really put those in place.
So, the first paper is this paper that came out in JAMA at the beginning of 2012, where
many of the authors are sitting here, so it's a little embarrassing to present the data
to those authors, but what they looked at, basically, was looking at the clinical data
that was available and asking the question of survivorship. Did the initial BRCA1 or
BRCA2 mutation status have anything to do with separating those individuals on a survival
curve compared to those individuals who don't have an underlying BRCA1, BRCA2 mutation.
You could see an overall survival shift between BRCA status and, so to speak, the individuals
not having mutations there. Certainly progressions-free survival saw this same trend, and then the
platinum-free survival, and this raised some very interesting questions about mechanisms
of platinum therapy itself and where the BRCA genes and mutations in those genes may be
important, in a funny way, in sort of protecting those individuals. They then went a step further
and looked at some of the very interesting somatic alterations, asking the question of
the expression of BRCA, particularly BRCA1, and the MRNA, and they've also looked at -- looking
at, specifically, signatures of expression in methylation.
So this was, in our minds, a very exciting finding that another group that I'm part of,
the OCAC, a large Ovarian Cancer Association Consortium, pulled out a much larger number
of individuals, and as opposed to just less than 500 which had been the initial study,
now taking this study and looking across nearly 3,900 individuals who we had very good clinical
information and where we knew quite a bit about sort of the clinical histories and the
risk factors, and then to look, can we see those same curves in place. And, in fact,
when we went forward, we certainly could, and we could actually see a separation between
BRCA1 and BRCA2, which the initial study was not quite well enough powered to do that.
But in a subsequent larger study, we could see that separation.
And these are things that had clearly been suggested in the literature before in very
small underpowered studies, but here, the power of really larger team science, and this
is the other key issue I want to really emphasize, in moving from initial hypotheses and preliminary
observations, to going to large amalgamated data sets, were tremendously valuable to be
able to start to separate these things. Because it's really those kind of agnostic statistical
criteria that I think that we eventually want to achieve in looking at our large data sets.
And if we look at the five year overall survivorship by BRCA status, it was very impressive. When
we go look at the Kaplan-Meier curves, we, again, see this same separation. And interestingly
enough, we can actually look and see that individuals who received platinum had a different
degree of response compared to those who didn't. Again, seeing very similar findings to what
we had seen in that first paper.
And so I think these kinds of supersizing are very important message for us to consider
in thinking about what we're going to do with our TCGA data; where we're going to take it,
what kinds of ways we can grab on to, whether it's germline, or somatic, or interactions
between those, and take them to additional data sets that would be the basis of new grants,
new approaches, and most importantly, potentially clinical approaches towards specifically treating
these diseases. And when we were able to look specifically at the types of mutations in
BRCA1 and BRCA2, we could see in BRCA1 that, perhaps, and again I use the word perhaps
here because I'm not sure the statistics here are quite as -- are strong enough to say that
we can, you know, unequivocally say that the type of mutation you have in BRCA1 and where
it is may be important in determining this, and again, this now tells us, as we get to
a granular level, we're going to need larger and larger studies as we really move forward.
So really, I think the story here I've already sort of projected is that BRCA1 and BRCA2
carriers potentially, I mean, do show substantially improved survivorship compared to non-carriers
in ovarian cancer, okay? And this is using primarily retrospectively or collected case
series and cohort case studies. We see a slight distinct change in the clinical course between
BRCA1 and BRCA2, and that there's some survivorship that may be very importantly mapping to the
actual types of mutations in BRCA1.
So, these are the sort of the next hypotheses that this community is now going after, and
I think TCGA was very important in empowering and sort of moving this to a very important
step. And then, of course, the implications for this for clinical trials are things that
won't be realized for two, three, or four years from now, as those studies have to be
suitably designed, integrating this information, accumulate it, and then, most importantly,
analyze once we have those particular studies far enough along.
So I sort of use this as a paradigm to where I think we're thinking about where we'd like
to see the TCGA data going. And this, again, comes from where I'm much more comfortable,
in my world of germline, but I think they're very important interaction here with the somatic
alterations that clearly will be followed up.
So when we think about this, we now have to sort of take a step back. We know TCGA is
moving along as a spectacular train, as Kenna and others, you know, have clearly been able
to move this along. There have been spectacular advances in really beginning to reach those
milestones and develop the datasets that are extremely useful for applying the hypotheses
and testing and analyzing in the ways in which we want. And it's really the Center for Cancer
Genomics that's relatively recently been established that Barbara Wold had really been instrumental
in cementing within the NCI structure. And then now, Lou and myself sort of have the
fun of taking this over for a period of time to establish what's the vision, what are the
next set of studies that are going to come. And to do that, really, you know, we have
to think about what kind of mission we have. This is a slide that Barbara had initially
developed that I think sort of characterizes where we see the Center for Cancer Genomics
going, to, you know, to develop and apply these cutting-edge genomic scientific tools
to improve cancer prevention -- a very important thing -- cancer care, and cancer detection.
And so these discoveries will then go in many different ways, as suggested here on this
particular graph, and clearly, there'll be iterations and revisitations of these data,
and these data should form really the nucleus of establishing new hypotheses and new studies
that clearly are going to need to go forward. So the discovery really is a key element,
and we know that we really have to think about what comes next.
So, in our minds, over the next three to six months, I think there's going to be a very
active dialogue and a very active discussion of putting in place what kinds of studies
are going to be lined up behind TCGA. Because we know TCGA is on track for 2013, 2014, you
know, realizing the milestones and our goals, and we have to exhort everyone to meet those
deadlines, get to where they need to get the samples there, get the analysis going, get
the PanCancer analyses, all those things that we know are going to be very fruitful. Because
behind that, we are already considering what are we going to line up, and we know that
it takes time.
As Kenna just showed the timelines from the point of identifying the study, getting the
permissions, getting everyone to buy in, getting the samples, analyzing -- there are very specific,
you know, timeframes that have to be met, so we are now very actively thinking, what
is it that's going to be put into the TCGA-like pipeline, and how that's going to be configured
as something that's a very active discussion at this moment. And as we hear the exciting
presentation and the discussions in the hall today, I think we'll be in a very good position
to really try and develop those, and it's really going to be a dialogue between those
of us at the NCI and NHGRI and the community as to really what makes sense and where the
opportunities are right at hand for the next year, the next two years, the next four years.
So these strategic lessons are really crucial in our mind, and capitalizing on these structures
are really crucial. And as I already alluded to, there will clearly be a continued fruitful
partnership between NCI and NHGRI as we clearly go forward.
So we want to build upon these strengths of these pipelines, particularly for the processing
and characterization, but also the analytic tools; we know the hardest part about this
is really the analyzing of the data and having to revisit it and look at it in new, creative
ways, and the data-sharing challenges are really quite, quite, interesting. And I think,
you know, the TCGA team has done a spectacular job in getting the data out and making it
available to people, and how and in what way the policies at the NIH are going to evolve
to take into account new technical opportunities, et cetera, are clearly a very active discussion
right now.
We also kind of think -- foresee that the kind of projects that will be supported will
be both top-down and clearly the value of bottom-up, and I made that allusion earlier.
I think the importance of taking the findings from these studies and generating RO1s and
generating institution or PI-specific grants is very important in sort of fitting in and
building the larger analyses, you know, and that's something that's clearly going to be
very important in the NCI portfolio, and I'm sure in the NHGRI as well. We have a major
transition towards wanting to be clinically important here, and that transition is really
crucial, but we know that we've got to be very careful, that this is very arduous, it's
filled with landmines; it's not as easy as it is to say to actually do it. And at the
same time, we know that really the next clinical opportunities are going to come out of very
basic discoveries, and so we can't lose sight of the value of basic discovery and of looking
at the datasets in the most effective, intelligent ways.
So really, what kinds of large-scale questions? Two or three more slides. Unraveling cancer
biology, in our minds, is very important, getting at this question of drivers and mutations,
drivers and passengers, when do passengers change places with drivers, and the like.
I mean, this temporal nature is something that's going to be very important, and we're
going to need numbers, and we're going to need very clever studies to be able to put
those kinds of things together, sort of developing a kind of somatic molecular epidemiology.
Larger studies, clearly, and we're going to have to paste together different kinds of
things, asking questions of clonality and progression, a very hot topic in what kinds
of coverages are necessary and what kinds of study designs clearly are going to be very
important.
The value of epidemiology and germline, and now asking questions of susceptibility, you
know, when we want to look at adeno CA [spelled phonetically] of the lung, and really in terms
of understanding what kinds of smoking and other types of exposures and in much more
exquisite details to be able to understand questions of risk that are both individual
and very importantly public health questions of risk. And these are two very different
venues. The contribution to somatic events in the relationships between germline drivers,
so to speak, of cancer, and the treatment stratification in pharmacogenomics are certainly
going to be there, particularly as we look at response, toxicity, and, of course, outcome.
So we sort of think about genome-related trials. And now these genome-informed trials, where
we're going to learn things from things that have been actually collected; the value of
prospective collections going forward is going to be very important, but they will be reaped
for a period of time, but the question is how and what do we do with future clinical
trials and how do we start to integrate into these kinds of analyses. Genome-driven trials,
there will be a lot of discussion about those, about where we actually have very remarkable
phenotypes and what can we learn in genomic characterization there. And then, certainly,
the genomic analysis, not as a part of a trial. Using the archive samples, the gene environment
analyses, the epidemiologic studies; these are all things that are going to be part of,
I think, portfolio that's going to have to be built. It's not going to be done in any
one particular venue, but probably a cross-section of the kinds of things that I've just outlined.
So let me finish by saying that the current TCGA goals in really, I think, very important
achieving those milestones per cancer site with the timely publications is really essential.
These are the coin of the realm that we have to be sure we keep up with, conducting the
PanCancer analyses, the technical pilots. We're all very charged up and excited about
seeing these to their completion and, particularly, to the thorough completion to know what the
next steps are going to be. And then, most importantly, the forging of new solutions
related to the integration, storage, and sharing of data, things that only become more difficult
as we generate more information.
And lastly, the fortification of the collaborative spirit, because I think there is a very great
American president, Woodrow Wilson, who made a very important statement, and this is my
sort of final comment to the TCGA community, and it should be in the first person plural.
"We not only use all the brains that we have but all that we can borrow." And so I think
the idea of really collaborating and being able to count on and work with others is really
a central component of not only TCGA but what comes out of TCGA and what directions it goes
in terms of large top-down as well as bottom-up studies.
So let me stop there, and bring Lou up. Lou, do you have your computer over here, or? Okay.
[applause]