Tip:
Highlight text to annotate it
X
Thomas Ried: Thank you very much. I'm glad it worked out
finally.
So, the title of my talk is Genome and Transcriptome Dynamics in Cancer Cells. To start, the most
important slides, the members of my laboratory who, in different capacities, contributed
to what I'm going to talk about. We have active collaborations. University of Lübeck, the
Karolinska Institute in Stockholm, and the University of Göttingen, which is part of
a clinical research unit which is funded by the German research community.
So, what I will do is, the first third or half of my talk, discuss with you what we
know about chromosomal changes and patterns of aneuploidy in carcinomas. Essentially,
it's part -- it's kind of an introduction to the second topics when we -- I discuss
briefly what the consequences of this cytogenetic abnormalities are under transcriptome, under
gene expression in cancer cells, and then I will exploit what we can do in order to
translate what we have learned from these more basic research topics to translational
medicine, both in terms of genomic aberrations and alterations of the transcriptome.
What you see here are chromosomes from a patient with chronic myelogenous leukemia. And you
can see there is a single chromosomal translocation between chromosome 9 and 22, and at this point,
I have to pause and acknowledge Janet Rowley who actually discovered the balanced nature
of this chromosomal translocation. In fact, this slide was performed in collaboration
with her. She unfortunately passed away just two months ago. She was in staunch support
of the intramural research program in [unintelligible] and the Genome Institute in particular, where
she served as scientific advisor since its inception.
And, as you know, this translocation has profound effects not only in terms of diagnosis but
in for treatment because if you target the genetic event which is the fusion of bcr and
abl, which results in [unintelligible] tyrosine kinase signal with a drug, you can get these
patients into remission. And the paradigm of chromosomal translocation not only applies
to CML but to many other hematological malignancies as well, which some of them are listed here.
But I chose to show you this slide for another reason as well which is, clearly, there is
one genetic event which we now -- causes transformation, but you can also appreciate that the rest
of the genome remains normal. So, what it is that there is a single genetic event which
is sufficient to transform these hematological cells. If you look at solid tumors, carcinomas
in particular, tumors of epithelial origins, the picture changes dramatically. Here is
a normal karyotype and here is the metaphases from breast cancer cell, SkBr3. And you can
see a profound degree of chromosomal instability, not only numerically but also structurally.
Then you can see giant mark of chromosomes which are a reflection of oncogene amplification,
in this case dysemic [spelled phonetically] and the HER2/neu oncogene. Then you have other
features -- this is probably a little dark -- in cancer cells which we call aneuploidy
which is affected in an enormous degree of variability in the DNA content from one cell
to another. You can also appreciate a polar mitosis which will clearly lead to chromosome
-- to a different chromosome count in the cells. You can see phenomenon of anaphase,
which results in lost chromosome and is caused by telomere attrition.
So, pictures like this and this lead to the perception that if you look at most solid
tumors in adults, it looks like someone set off a bomb in the nucleus. And this perception
lead to the interpretation that in solid tumors, different from the hematological malignancies,
we have a cytogenetic of chaos which is induced by catastrophic mitosis -- you have seen one
-- karyotypic complexity, heterogeneity of the tumor-cell population, and ongoing chromosomal
instability. The perception of a cytogenetic chaos also led to the perception that chromosomal
aberrations in solid tumors, different from the hematological malignancies, would rather
be a consequence --would be a consequence rather than the course of the disease.
And what I will show you now is -- I will discuss whether this is indeed the case. Arguably,
we had to develop methodologies to dissect the complex aberrations in solid tumors. Most
important was comparative genomic hybridization; I will show more details later. But, in short,
it allows you to map genomic imbalance in the entire tumor-cell population. We have
spectral karyotyping, which we developed years ago, which allows you to paint all chromosomes
in different colors. And then we can used specific probes to enumerate chromosomal copy
numbers directly in non-dividing cells, psychological preparations or tissue sections, and it can
also be used to identify translocation because you see a fusion of different colors. And
now with the advent of global gene expression profiling or sequencing, we can interpret
and query how these aberrations affect the gene expression levels on a global -- in a
global sense.
So, having had these tools, we then tried to understand the dynamics of changes in the
transition from normal epithelium to invasive disease. And initially, we turned two model
systems, which is cervical carcinogenesis, colorectal carcinogenesis, mainly for the
reason because the morphological changes are very well described and they are accessible.
So, we -- you know the sequence, it goes from low grade dysplasia to high grade dysplasia
and eventually to invasive disease, and while it looks a little different in the colon,
it's the same. You have premalignant lesions that, if they would be removed, you would
treat -- you would cure the patients.
So, we asked -- we used a combination of molecular cytogenetic tissue microdissection techniques,
gene expression profiling, and asked the question, "Is there a non-random distribution of non-chromosomal
changes in any one of those different tumor entities? Is there a stage-specific sequence
of events? In other words, are some changes early, others late? And which additional genetic
or epigenetic changes do occur?" And it is now completely clear that the distribution
of chromosomal gains and losses is non-random. In red you see the aberration of the ***.
Let's just look at chromosome 3. On the right side are the gains; on the left side are the
losses. Chromosome 5 has gained, but the rest of the genome is relatively unaffected. In
the colon, which is displayed in green, you can see, invariably, the gains of chromosome
7, extra copies of the long arm of chromosome 8, chromosome 20, and very specific and unique
for colorectal carcinogenesis, gains of chromosome 13.
And I just wanted to point out, we and others have studied more than 500 cases of cervical
carcinomas. Here the blood is normalized for 10, and you can appreciate that all cervical
carcinomas have extra copies of the long arm of chromosome 3. In other words, this aberration
is as common as is the [unintelligible] chromosome in CML. We do know that we have risk factors.
Of course, high-risk HPV, most commonly 16 and 18, which is required for cervical cancer
to occur but it's not sufficient. And in colorectal tumors, you know that inflammation, ulcerative
colitis, Crohn's disease greatly increases the risk for colorectal tumors to occur, but
the distribution in these conditions is the same as in sporadic tumors. Conclusion: We
do not have a cytogenetic chaos but we have a stability on a different plateau of copy
number changes. It's like a speciation.
So, but, we do have the problem that we have an enormous degree of variability. So, the
question comes up, how can we reconcile aneuploidy, an ongoing chromosomal instability, which
is a fact, with a strictly conserved pattern of genomic imbalances that we observe? And
in order to answer this question, we engaged in a collaboration with Navy, at that time,
across the street, it was two streets from here, and collected patient samples where
have ductal carcinoma in situ and invasive ductal carcinoma on the same slide. And from
that we used tissue microdissection and we prepared from both of these lesions independent
slides where we had the single cell suspension here, you see, and for one case, the histology.
And then we chose -- we looked at the distribution of genomic imbalance that is specific for
breast cancer; you can, again, appreciate that it's specific, but it's different from
both colon and ***. And we chose probes, fish probes, fluorescent in situ hybridization
to enumerate the copy number of these regions together with control probes directly on these
cells that were prepared with DCIS and IDC, and that was done -- this is a technical detail
-- in repeat hybridization, but at the end of the day, we could enumerate 10 independent
loci on many of these cells that were on the slide and were prepared from DCIS and synchronous
IDC.
And we don't have to go through in detail but that allowed us to enumerate in the entire
population of the cancer copy number gains for a certain chromosome and copy number losses
for a certain chromosome. And here you can see there are many cells that have the same
pattern, and this is the DCIS and this is the synchronous IDC. You can appreciate that
here was have a major pattern that is recapitulated in the invasive component, but you can also
appreciate that we have an enormous degree of heterogeneity. In some instances, the clone
that was predominant in the DCIS actually disappeared in the IDC. And together with
our colleagues from NCBI and in collaboration -- a very fruitful and pleasant collaboration
with Russell Schwartz, Carnegie Mellon -- we could then reconstruct the dynamics that occurred
from the transition from DCIS to invasive disease. And two pathways occurred. One was
clone stability. Here you can see the DCIS on the left and the IDC on the right. You
can see the major clone, which is defined by this copy number oration, prevailed; it
was also found in this component. And here as well, and in this case, it was the same.
But much to our surprise, we also noted that in some cases the entire population of clones
that were present in pre-invasive disease disappeared. Only those that acquired extra
copy of the MYC oncogene made it to invasive disease, and that, again, told us that the
transition from pre-invasive disease in many cases is determined by the acquisition of
extra copies of the cMYC oncogene. But, you can also appreciate that in many cases that
ductal carcinoma in situ already governed by an enormous degree of chromosomal instability,
therefore, in a -- I'm not a clinician, but I just can't see that any other intervention
then surgery would remove these clones.
But, anyway, despite this chromosomal instability, if we then look at all these clones that we
have analyzed in copy number changes, we can conclude that chromosome 1Q, which is frequently
gained, is very rarely lost, and chromosome 8Q, MYC oncogene, which is frequently gained,
is very rarely lost. And those chromosomes, here P53 for instance, which is frequently
lost, is rarely gained, despite this enormous degree of chromosomal instability. And this
makes it then consistent with the genomic imbalances that are specific for copy number
changes. So, despite chromosomal instability, what is the pressure for the [unintelligible]
selection is the maintenance of the genomic copy number changes that are the defining
features for any one of those carcinomas.
This now, of course, [unintelligible] regressions, we have changes, mostly whole chromosome arms,
or, in other instances, whole chromosomes in the colon, gain of entire chromosome 7.
This now brings us, I think, a fundamental question: how does it affect the transcriptome?
You could ask the question, you could test the hypothesis. The expression of all or most
genes located on a chromosome is affected by chromosomal gain or loss. Or, the expression
of only a few genes whose reduced or increased expression is critical for tumor genesis is
the target of chromosomal aneuploidy during tumor genesis, and this was not clear. And
in order to identify -- so, the conclusion so far: chromosomal aneuploidies are a defining
feature of carcinomas. The distribution of genomic imbalances is cancer-specific to an
extent that you don't have to have any other information, but the distribution of genomic
imbalances to say it's a cervical carcinoma or a colorectal carcinoma. Specific genomic
imbalances occur before the transition to invasive disease, which is very important
if you want to convert that to a diagnostic test, which I will show you later. And, as
I said before, there is no cytogenetic chaos but a stability on a different plateau of
genomic copy number changes.
So, as I said, we then asked, what are the consequences of chromosomal aneuploidy of
the transcriptome?" which I will discuss briefly in the next few slides, and then I will turn
to how we can use this knowledge to improve the diagnosis of cancer and pre-malignant
lesions.
So, in order to address the question, what are the consequences on the transcriptome,
we use an old technique which is called microcell-mediated chromosome transfer to generate artificial
chromosomal trisomies, and this is how it looks. Here we have a normal karyotype, and
when we do this manipulation, we can put on -- put in an extra chromosome, in that case,
chromosome 3. But you can see -- which is present in three copies. But you can see that
the rest of the karyotype is unaffected. Then we collected RNA from the control and RNA
from the plus-3 cell and performed global gene expression profiling. Here, again, the
control, some genes up, some genes are down, which is what you would expect. But, if you
look at the plus-3, you can appreciate that most, if not all, genes go up in expression
levels. And this, by the way, is a sign if you look at the constitutional chromosomal
trisomies -- 13, 18, 21 -- where this increase is also not restricted to a few genes but
affects most of the genes. And interestingly, on a site, it's tolerated for chromosome 13,
18, 21 because these are the gene-poorest chromosomes, despite they have larger size.
Thirteen, of course, it's larger than chromosome 20.
So, aneuploidy results in a significant increase of average message levels of genes on the
affected chromosomes. The degree of increase closely follows genomic copy number, therefore,
chromosomal aneuploidies not only target a few specific genes but results in a massive
and complex deregulation of the cancer transcriptome. And just contrast that to what we see as a
consequence of the [unintelligible] chromosome where you have one aberration, now you have
a thousand genes that go up, and you also have to appreciate, of course, there are numerous
transcription factors, with then, in turn, affect genes and other chromosome, so it's
really a massive deregulation which we have not understood in its complexity.
This is not only the case in our model system but also in real tumors. Here, a control,
a normal DNA copy number, and this is the gene expression level, which is normal as
well, but if you have copy number decreases or copy number increases, the gene expression
goes down in this region, goes up in these regions. It's completely clear: function follows
form. And this is established not only from our laboratory but now accepted in the scientific
community.
To summarize, we have a normal cell, karyotypically stable. Then, we have chronic inflammation,
for instance, in the colon, HPV infection. We have, for whatever reason, a depressed
hyperproliferation, or have just a cell cycle accident. What happens then that we have extra
copies of a given chromosome, and this chromosome is tissue-specific; that can occur with or
without additional mutation. They're what call nuclear aneuploidies, very low so there
is not much difference in the DNA content from one cell to another. The trisomies affect
global gene expression levels, and as you will see later, these events are the basis
for global expansion of these early lesions. If this is not being treated, then we arrive
at a cancer cell where we have additional mutations -- as we know, P53 occurs relatively
late in colorectal tumor genesis. We have an enormous degree of the variability in the
DNA content from one cell to another, but the genomic aneuploidy persists; that's why
CGH was successful, as I have shown you before. What is the selective pressure is the maintenance
of the specific genomic imbalances.
So, now I -- after this introduction I want to show you how we convert it to translational
application. Three topics; I hope I can cover them all. First, I will discuss with you the
role of genomic instability in the prognosis of breast cancer, then we look at cervical
carcinoma, and, if time, how we use transcriptional profiling to predict treatment response in
patients with *** cancer.
So, breast cancer's usually present in two flavors. One have a relatively stable genome,
which are called diploid, and others are genomically instable, have high degree of aneuploidy.
You can -- if you -- for a take-home lesson, this is the Washington Monument and this is
the Manhattan skyline. And many years ago, our collaborator Gert Auer at Karolinska -- I
was not on that paper at the time -- discovered that patients who have a genomically stable
tumor have a far better prognosis than patients who have an aneuploid tumor. Profound differences.
And then, in 2002, and as a sect [spelled phonetically] many other papers followed.
Van de Vijver published a paper in a New England journal, who showed that there is, by gene
expression, a good prognosis signature and a poor prognosis signature. They even came
to the same conclusion. And if you look at these curves, you are -- you cannot argue
that they are very similar.
So, we can try to understand the nature of the similarity and perform gene expression
profiling of a set of diploid tumors and a set of aneuploid tumors. And we could easily
separate the aneuploid ones from the diploid ones, and this separation was based on a signature
of 12 genes, which perfectly separated -- almost perfectly separated to stable from the instable
breast cancers. And then we asked the question, can we use this aneuploidy-specific gene expression
signature to predict -- to recapitulate the published data sets? And that worked in all
instances and the numbers are very high with very high statistical significance.
And then the next question was whether we can ask the prognostic signatures that were
derived from these data sets to predict the degree of genomic instability in our cases,
and that worked perfectly well. The oncotype DX, which is used in the clinic, as you know,
and the good prognosis signature of the oncotype DX predicted the tumors were diploid, and
the poor prognosis signatures predicted that the tumors are aneuploid. And you can see
that works very well. Even -- and the same occurred for the MammaPrint. The good prognosis
signature predicted genomic stability, the poor prognosis signature predicted aneuploidy.
And just an aside, a Luminal A is another gene expression signature which you know is
associated with a good prognosis, predicting genomic stability, and the basal signature
and the HER-2 new [spelled phonetically] signature associated with poor prognosis predicted genomic
instability.
In summary, the degree of genomic instability and gene expression signature of poor prognosis
are linked. The degree of genomic instability is the nature biologically determinant of
poor prognosis, and it is not only the degree of biological -- of instability if you look
at the tumor cell, but it's the fact that you have many multiple clones which provide
the tumor with the nimbleness to react to environmental challenges including therapy.
The next application which we developed for over -- more than 10 years is the identification
of individual progression risk in cervical dysplasia. And here you remember what I showed
you, that there is invariably a gain of chromosome 3 -- the long arm of chromosome 3 in cervical
carcinomas, and this gain never occurs in normal cells. So we developed, in collaboration
with Abbott [spelled phonetically], a probe set that targets these regions, that these
are just control probes which help us to -- just control probes which help us to enumerate.
You have the same scenario if you test for HER-2 new amplification, you'll also use a
central map for chromosome 17 as a control. And then we applied this to a routinely-collected
pap stain cervical smears.
And I want to really make the point, in cervical cytology, the challenge is not to diagnose
cancer, because we all actually do not want to diagnose cancer because it's too late.
We want to diagnose early patients which can be cured by surgery. The challenge in cervical
cytology is that -- there are two challenges. I mean, everyone can say, "Well, this looks
different than this," and here it's actually nice, you have a normal cervical cell which
is defined by a small nucleus and a large cytoplasm, and you can easily discern them
from these cancer cells. But the distinction from here to here is much harder. Even the
distinction from here to here is essentially impossible. What is even more important is
that only 15 percent of the low-grade dysplastic lesions would progress. Therefore, it would
require treatment, despite the fact that about 90 percent of these low-grade lesions are
already positive for HPV. So, HPV does not really help. It only helps if it's negative
because then you know there won't be cervical carcinoma, but most of them are positive,
so it doesn't really help.
So, we argued that if all these carcinomas have a gain of chromosome 3Q but none of the
normal have it, and there's only a fraction of the low-grade lesions that progress and
would require treatment, those that progress already have the aberration that defines the
invasive disease. And that those 85 percent that grew spontaneously to regress are cytogenetically
normal. And we then explored that, as you can see here with this probe set for three
copies: 2-2, 2-2, 2-2. This is normal. Here we have a cervical intraepithelial neoplasia
grade two, which is a model of dysplasia. You can see here one, two, three copies; one,
two, three copies of chromosome 3Q. The rest of the genome is still normal. And in some
of the carcinomas you can see clouds of 3Q of the TERC oncogene which we choose as a
probe that is reminiscent of the amplification of HER-2 in breast cancer.
And then, this is unfortunately very difficult to see. I come back to the -- what I said,
it is the basis for clonal expansion. Here you can see a pap smear which was stained.
We then de-stained it. Here, a normal cell, two copies. And here we have a nest of cells
which were collected. They all have three copies of chromosome 3Q, but they are next
to each other, meaning that they originated from each other. So, once a cell has acquired
three copies, that is the basis for clonal expansion. So, not only do we have a diagnostic
test, but we can also visualize the emergence of cancer by looking at the copy number changes
of that particular chromosome.
We confirmed that in many different studies, this was one in collaboration with Karolinska,
where women were randomly -- had pap smear randomly, and those that had a suspicious
pap smear then underwent colposcopy and biopsy so we could correlate the findings on the
pap smears with the histology with a gold standard, and the conclusion is that the gain
of 3Q, the human telomerase gene which is on there, has the highest combined sensitivity
and specificity for the detection of histologically-confirmed high-grade lesions. And we are now conducting
an even larger study to extend it to low-grade lesions and to [unintelligible].
We conducted another study which was aimed at validating 3Q as the molecular marker of
progression, which, as I showed you at the beginning, is the most important. And this
is how it was done. We collected a group of patients, the numbers are now much higher,
which were diagnosed with severe dysplasia, and before, they had a low-grade dysplasia.
Then we had another group of women which had the same low-grade dysplasia but returned
to normal. We had another group of women which were diagnosed despite being involved in an
active surveillance program with severe dysplasia or even carcinoma, but the pap smear before
was normal, so that should not happen.
So, we then hypothesized that those -- we know already these are positive, these are
all negative, these are positive, but we hypothesized that those low-grade lesions that show progression
are positive for 3Q and those that show regression are negative for 3Q. And this worked, essentially,
with the sensitivity of 100 percent and the specificity of 95 percent. We can discuss
it later. It's a particular feature of HPV, but you can see that the point of no return
in the progression risk for individual lesion is the acquisition of the specific cytogenetic
abnormality that then defines the cancer entity. We also were -- expected but were a little
bit shocked that, in some cases, where the pap smear was assessed as being normal, and
after a relatively short latency, the woman was diagnosed with a carcinoma, we could detect
extra copies in a third of the cases in the cases that were cytologically assessed as
being normal, and I just show you some examples here. Here there is a CIN1 lesion which has
regressed. Two copies. Two copies. So, there would not have been any other treatment required.
Here we have a CIN2 lesions, and you can see here, for instance, these two cells -- these
two cells, and they have both three copies. This lesion is progressed to high grade dysplasia
and carcinoma. And this is the case that I mentioned to you before which was assessed
as being normal, but you can see we have four copies of chromosome 3Q in the set of cells,
and, again, they are next to each, indicating that the acquisition of aberration is the
basis for cloning expansion.
I shall also emphasize that -- so, I talked to a cytologist as Hopkins, and they come
in the training program twice a year. And every time at the beginning the one diagnosis
go up and then they learn a little and it goes down, because it is obviously relatively
difficult to unambiguously assess the morphology. And when I go to cytology meetings, there
are whole days that are -- where they discuss on how to best repeat the reading of a pap
smear to avoid false negatives. And the fact that it is relatively often repeated is an
acknowledgement of the fact that an individual pap smear is relatively ambiguous. But you
cannot argue that there are 4 copies. So, it's a binary -- I mean, you can ask your
6-year-old to -- probably younger -- to count to four and make the statement, "This is not
two." Therefore, they cannot even say -- make the statement as cervical carcinoma, they
can even say, "There is a woman which -- whose low-grade dysplastic lesion will eventually
be invasive carcinoma."
So, it took us a long time to convince gynecologists and pathologists to embrace that because,
I mean, the procedure for the collection of this material would not even have to be changed
because there is now -- in particular, with liquid-based cytology, there is always leftover.
But, now finally Crest Diagnostic has embraced the tests and is offering it in its portfolio.
So, now I want to switch gears a little bit. I don't know what the time is.
Male Speaker: [inaudible]
Thomas Ried: I'm fine. And just talk a little bit how we
not only use genomic aberrations, but aberrations of the transcriptome and injected genomic
information into a problem of treatment response in patients with *** cancer. And you know
it better than I do, there are many treatment options in solid tumors. It can be surgery
alone, surgery plus adjuvant chemo, radiation or chemotherapy in colon carcinomas, or neoadjuvant
chemotherapy followed by surgery, or other approaches, or, in some instances, depending
on the mobility of the patient, maybe chemoradiation alone.
And the problem in *** cancer is a real clinical problem, because, based on a large
study which was conducted from the terminal *** cancer group and has been adopted now
in most of Europe, I know it's not quite completely adopted in the United States, the standard
treatment of locally advanced *** cancer is neoadjuvant chemoradiotherapy, mostly based
on 5-FU radiation, followed by surgery and chemotherapy. And the reason for that is that
they found out that if pre-surgery chemoradiotherapy is administered, there is a higher rate of
R0 resection reception, there was a higher rate of sphincter preserving receptions, and
there was a reduction of local occurrences, significant from, I believe, 14 to 6 percent.
The profound clinical problem is that the response to neoadjuvant chemoradiotherapy
is very heterogeneous. You can have complete pathological response in that case, where,
after surgery, there is not a single tumor cell left. Or, you have essentially complete
resistance where the tumor just doesn't bother being treated with 5-FU and radiations.
So, we generated, together with a former post doc of mine, Michael Ghadimi, who is now the
Chair of Surgery at the University in Göttingen, Germany, a clinical research unit -- you can
look it up -- in order to address that problem, and we aimed at identifying predictors of
response and identify targets of that and understand mechanisms that could explain this
profound difference in response resistance. And in order to do so, we initially designed
this pilot study with 30 patients. They all had pre-treatment clinical staging based on
*** ultrasound. We then collected tumor biopsies and performed gene expression profiling.
Obviously can barely see from here. But then they went to this neoadjuvant scheme, which
is chemotherapy followed by surgery and four rounds of 5-FU, and then we had the long-term
follow-up. And pathological staging is shown here, and, as I said, we performed gene expression
profiling, and then evaluated the local and distant reoccurrences after median follow-up
times of 44 months.
And the gene expression profiling worked actually fairly well. Here we had -- I forgot how many
-- over 30 cases. Approximately half of them had non-responding tumors. Fewer had tumors
that showed a complete response, and a gene expression profiling could actually fairly
well discern these two groups. But then we looked at the genes that were unregulated
in the tumors that were resistant. And among those were the transcription factor TCF4,
and this obviously rings a bell the transcription factor of TCF4 is the major effector of WNT
signaling in colorectal tumor -- of the major effect of WNT signaling, of course, not only
in colorectal tumors. And WNT signaling is intricately involved in colorectal tumor -- effect
on mutations in the adanomatous polyposis instability and resulted in stability of catenin,
which then actually increases the expression of the transcription factor of TCF4, which
in turn turns on cyclin, D1, c-MYC, and other notorious oncogenes, and therefore transforms
the cells.
So, that was a very interesting lead. So, we hypothesized that silencing of a gene that
is overexpressed in resistant tumors would increase sensitivity to chemoradiotherapy.
And we did the silencing with the approach of using RNA interference, and I don't go
into details here, but if you inject small interfering RNAs into cells, you can reduce
either protein synthesis or you can leave the transcript which results in a loss of
function, and you can do that essentially for any gene of interest. So, we used small
interfering RNA against the transcription factor TCF4, and asked the question if the
silencing of TCF4 would increase sensitivity of colorectal cancer cells to chemoradiotherapy.
Here you can see the results. It was a very effective silencing, here shown in the reduction
of protein. We repeated that with many different constructs, and when the results were then
subjected to radiation and not shown here, but also to chemoradiation -- this is a log
scale -- you can see here the control, and you can see here a profound sensitization
after silencing of the transcription factor TCF4, which, in primary tumor samples, was
overexpressed in those that were resistant to treatment. Conclusion: Reduced expression
of TCF4 leads to a sensitization to radiation.
We are now exploring whether we can recapitulate using small molecule inhibitors of WNT signaling,
and together with this clinical research in Göttingen, we are about to design clinical
trials to test that impatience.
So, in summary, we have performed gene expression profiling, which I have shown you, in ***
carcinomas that are resistant or sensitive to neoadjuvant chemoradiotherapy. We have
also queried many different level of the genome using CGH to see whether we have specific
genomic imbalances that may explain the sensitivity. We have performed microRNA expression profiling.
We explored KRAS and BRAF mutation, which, by the way, did not explain different sensitivity
to radiation. We looked at the methylation status in collaboration with Stephen Chanock.
We used 1 million Illumina platform to look for single nucleotide polymorphisms in the
group of patients that respond differentially to see -- to explore whether certain haplotypes,
for instance, in genes that involved in direct metabolism, could explain the profound clinical
differences in the clinical course.
We do all this in order to address the problem that all drives us, which is that patients
coming to the clinic with different clinical features they have a different genetic makeup
and a different tumor biology, and, in the case of *** carcinomas, they also have
a different treatment toxicity. And this is not only true for *** carcinomas but a
different genetic makeup of, for instance, pre-malignant lesions in the *** could
be summarized here. However, despite these acknowledged differences which we do not understand
to the extent we should, all these patients receive identical therapy. The goal of all
of us, of course, is to learn enough from all these different techniques -- I have not
listed next-generation sequencing here, which is important as well -- to learn enough about
the genetic makeup so that we can assign patients to therapies from which they would benefit
the most.
With that, I am at the end, and I thank you for your attention. I'm happy to take your
questions.
[applause]
Male Speaker: [unintelligible] comments or questions I have.
One, [unintelligible] said that all biology is only understandable by natural selection
and evolution. And is it possible that by using chemotherapy [unintelligible] we encourage
resistant tumors to grow?
Thomas Ried: I think it is. I think it is. I think -- I
mean, tumor heterogeneity was already -- I don't have it on the slide, unfortunately,
but what we are now doing, we look at these *** carcinomas that are complete responders
and those that are resistant, and ask the question whether in those that are then resistant,
there are minor clones that expand during chemoradiotherapy. So, I think it's completely
possible, but I can also say, I mean, it's now being more and more acknowledged that
we cannot think, if we look at the bulk of the tumor that we have understand -- understood
the tumor.
Male Speaker: So, this is a signal to noise question, really.
Other comments or questions? Yes.
Male Speaker: In the breast and colon cancer, are there
also any abnormalities that the normal tissue in the same patient, like, those that you
described in cervical cancer.
Thomas Ried: The question was whether in -- and let's just
look at cervical carcinomas or colorectal carcinomas, whether those cytogenetic abnormalities
that are the early events in tumor genesis can already be detected in the adjacent normal
epithelium. Not on the chromosomal level, but we are now exploring the possibility that
on those chromosomes that are early gained in any one of those diseases, that they are
genes which are already higher -- more highly transcribed, so that the initial impetus to
gain these chromosomes would be a physiological one. So, but on the chromosome normal tissue
is -- has no chromosomal aberrations.
Male Speaker: Compared to normal person without cancer,
is there any changes at all?
Thomas Ried: No. No.
Male Speaker: Other comments or questions? Yes.
Female Speaker: This is on a little bit different tangent.
That was a very nice talk. What do you feel is the role, if any, of long non-coding RNA
in the maintenance of genomic stability? And sort of a consequent question is do you feel
long non-coding RNA could be used as a way to manipulate the transcriptional profile
in cells where additional genomic aberrations have been obtained?
Thomas Ried: So, the question was, "What is the role of
non-coding RNAs, whether they are short or long, in the maintenance of genomic instability?"
and the answer is I don't know. The answer is nobody knows. Obviously, the [unintelligible]
Genome Institute and initiatives, the ENCODE Project, we will learn much more about it,
hopefully through other initiatives as well. But the role of RNAs in the genome is not
yet established.
Male Speaker: From a practical point of view, there are
tumor markers that I get in my office [spelled phonetically], the CA125 in ovarian cancer
or something like that. What does have to do with the genetics of the tumor? I mean,
is that an indicator that you're getting increased risk of genetic defects, or do you look at
that, or it has something totally different?
Thomas Ried: So, the question is, "To which extent established
serum markers would reflect the risk of developing specific aberrations." I cannot -- there are
no -- I cannot answer that. I mean -- and the serum markers are -- I mean, you have
PSA [spelled phonetically], but -- in the ovarian cancer, but it's -- there are obviously
problems, and many people have looked for colon and other tissues, pancreas, but there
is no direct correlation. It could be, also, factors that are unrelated to the presence
of certain genomic imbalances which is whether you, you know, have a blood vessel that goes
through the tumor and you can check cells. So, what I believe will become a possibility
in the future for earlier detection is the detection of circulating tumor cells.
Male Speaker: The reason I ask that is I have a lady that
with metastatic ovarian cancer who is in total remission and her scans are normal, and she
just has an increase in her CA125. So as a non-oncologist, you know, how am I supposed
to deal with that? Or just let the oncologist handle her? I mean, because she comes to me
and asks me the questions. .
[laughter]
Thomas Ried: I'm not going to make a suggestion. [laughs]
Male Speaker: Okay.
Male Speaker: So, can you comment on the tempo of translation
of this exciting science to the clinic? [unintelligible] practice, the question I have in my head,
is this something I'm going to have to face tomorrow, a year from now, or 10 years from
now? It seems like it's pretty soon.
Thomas Ried: For the cervical carcinomas, it's being implemented
now. There are several laboratories around the world who actually use it for samples
that are ambiguous. They -- the goal would be to all the cases that are HPV positive
should get this genetic test, because then you know what's going to happen. And I usually
restrain from such statements in biology or genetics, but this is black and white. It's:
You have it, you progress. You don't, you do not. Period.
So, this is probably the most advanced we are now, as I mentioned to you before, we
worked with the Karolinska Institute where we have 9,000 archived breast cancer cases,
from all of which we have to clinically follow up. The Swedes are very good with that: homogenous
population, not so much mobility. And we have to clinically follow up and we have the DNA
content, so we are now in the process of correlating, in a large definitive dataset, the degree
of aneuploidy with prognostication. So that's going to be next. We have just received funding
in Göttingen for additional prospective trials where we can try -- where we can evaluate
in larger numbers the value of gene expression profiling for prediction of response.
Male Speaker: I recall reading a note from Vince DeVita
who was a former [unintelligible] at the Cancer Institute that we were entering a phase where
in order to treat cancer, you need a very skillful doctor, and this stuff is really
beginning to bring that home. A pathologist isn't enough; we need more. Other comments
or questions? Yes.
Female Speaker: Well, right now, before the age of 30, we
don't get HPV unless there's atypical cells, unless the patient [inaudible].
Thomas Ried: I didn't --
Female Speaker: Presently, with Crest and LabCorp, you can't
order a high-risk HPV initially. It's only done as a reflex when you first start seeing
abnormal cells.
Thomas Ried: Yeah.
Female Speaker: Is that likely going to change?
Thomas Ried: I know that there were actually several studies,
I believe they suffered from budget cuts and sequestration, to exactly address the question
whether HPV should be the first test, because it, essentially, makes sense, because those
that are negative do not get cervical cancer. I completely agree, but I have not seen the
result of any -- there was one out of Seattle, there was one that was initiated by Mark Schiffman
at the NCI and others, but there is no -- I don't want to step on anyone's toes, but maybe there is also a priority resistance
by the cytopathologist to change -- or by the gynecologist to change something which
has been successful. And, I mean, one should be very careful to change something that is
successful into something where you first have to validate it on a large basis. But,
from a biological point and a genetic point of view, there would be no reason not to change
the practice to what you just suggested.
Male Speaker: It is a fact that we clinicians stumble along
behind investigators, and it takes a while to translate, I think. Other comments or questions?
If you have them, please come down. Thank you very much.
Thomas Ried: Thank you.
[applause]