Epidemiology and Prognostic Assessment of Soft Tissue Sarcomas

SLIDES & TRANSCRIPTS
Monday, June 17

Molecular Pathogenesis of Soft Tissue Sarcomas

Marc Ladanyi, MD

Slide 1:

Good morning. I thought this morning that I would focus on biological issues mostly, not so much practical issues. So, I actually don't have an answer on whether RT-PCR is better than FISH and so on.

Coming from my perspective, which is the perspective of somebody who has been interested in sarcomas with translocations, my view of sarcoma molecular biology is perhaps a little bit slanted by that experience. So, I will try to outline what I think are some of the interesting developments that have occurred in the past decade in sarcoma molecular pathology, and to try to raise some questions, throw out some ideas, some of which may be completely wrong, but just to kind of stimulate some more thinking on the molecular biology of sarcomas.

Slide 2:

I think I should start with a disclaimer, an apology. I know the focus of the meeting is supposed to be adult soft tissue sarcomas. In terms of the molecular biology, we can't really separate them from pediatric sarcomas and we can't separate them from bone sarcomas. So, my apologies in advance -I will mention some pediatric tumors and some bone sarcomas.

Now, I would like to start with, I think, perhaps kind of a pet peeve of surgical pathologists and molecular pathologists, and that is the issue of lumping of sarcomas. I think that the lumping of soft tissue sarcomas in various prognostic studies really was driven by the relative uniformity of available treatment options.

Slide 3:

This kind of led to this minimalist view of sarcoma classification, looking more at soft tissue sarcomas by grade, location and so on, than either pathologic classification or, as you will see, by molecular pathology classification. Now, this approach, I think, has worked fine for studies that have looked purely at clinical markers, clinical prognostic markers.

I think, in studies that have looked for biological prognostic markers, I think this has really had a detrimental effect, because I think we have been overlooking some of the biological distinctiveness of individual sarcoma subtypes. In turn, I think that has kind of delayed the identification of novel prognostic markers, potentially now therapeutic markers, and fed into kind of a continuing paucity of new treatment options or new prognostic options in sarcomas.

Slide 4:

So, in terms of what I would like to propose as the two major subtypes of sarcomas, at the level of cancer genetics or molecular pathology, as you might expect, one large subgroup of sarcomas is what I would describe as sarcomas with specific genetic alterations and relatively simple karyotypes. So, these are sarcomas in which we have a pretty good idea of what is the key genetic lesion, possibly the initiating genetic lesion. In this category, we have the sarcomas with reciprocal translocations and I would also add now, the sarcomas with specific point mutations, obviously, KIT mutations and GIST.

The second group is basically the rest, and it is kind of a group that we really don't have very good models for at the moment. These are sarcomas that do not seem to have any recognizable, specific or initiating genetic mutation, and often have relatively complex karyotypes.

Slide 5:

Now, focusing on the first group, I think we are familiar now with a series of gene fusions in sarcomas that have been cloned over the past decades. I am sure you are familiar with most of these. I think we have identified just about all of them. There are a few more out there that are in the process of being cloned.

Perhaps Jonathan Fletcher will mention the aneurysmal bone cyst translocation. I think there is a translocation in the angiomatoid type of MFH that probably should be added to this list.

We have basically cloned, I think, most of the major translocations in sarcomas. They really define a group of tumors that I think has a special biology that should not be combined with tumors that lack translocations.

Now, in terms of diagnostic markers, evidently these are powerful diagnostic markers, as Chris mentioned. The specificity of these markers has made them very useful in molecular diagnosis.

I would submit that, at least for clinical studies of these sarcomas, we should include molecular testing in these studies, both in terms of data collection and in terms of objective confirmation of the diagnosis.

Slide 6:

Another way of looking at this list of translocations is to organize them into the translocations that involve the EWS family of genes. EWS was originally identified through the cloning of the Ewing sarcoma translocation in 1992.

So, by now, there are 11 different EWS family gene fusions in five different sarcomas. It is very interesting to note that various family members have been identified that are highly homologous to EWS and can replace EWS in different fusions and subsets of sarcomas, such as myxoid liposarcomas and extraskeletal myxoid chondrosarcomas

Slide 7:

Then there is also the so-called non-EWS family of gene fusions, along with other sarcomas characterized by different gene fusions, and this includes over 10 fusions in seven different sarcoma types.

So, there are quite a few sarcoma types characterized by translocations, at least 12, probably closer to 14 or 15 by now.

Slide 8:

So, these fusion genes usually produce aberrant transcription factors. As you know, transcription factors regulate the expression of downstream target genes.

So, I think we can conceive of these transcription factors as providing multiple hits in a single genetic event. I think we have to start looking at how the target genes of these aberrant transcription factors provide the multiple hits that allow the sarcoma precursor cell to become fully malignant, and that is something that should be a goal of the next decade.

The other observation that is kind of inherent in this data set is that there is really tremendous tumor type specificity. Fred Barr has proposed that the specificity reflects a dynamic relationship between the gene fusion and the cellular environment in which it occurs. Basically, the gene fusion can only occur in a susceptible cell, a cell that is likely to be transformed by that particular fusion oncogene at that particular developmental stage. In turn, the gene fusion then modifies the phenotype of the cell in which it occurs.

Incredible specificity for a particular cell type also accounts for the difficulty in developing transgenic mouse models of this class of sarcomas.

To my knowledge, there is only one transgenic mouse model of a sarcoma translocation, and that is a TLS-CHOP mouse that was made by a Spanish group. That is in spite of many unpublished attempts to make other models of other translocation-associated sarcomas.

Now, the key role of these sarcoma fusion genes has been supported by many studies that have shown that, if you block the function of the fusion products in the corresponding sarcoma cell lines, you block in vitro growth.

Another way that the central role of these fusions has been shown, as has been mentioned already this morning, is the impact of relatively minor structural variability in these fusions can correlate with either phenotype or clinical course in certain sarcomas.

The observation that the karyotypes of this class of sarcomas are relatively simple argues against telomere dysfunction as a pathogenic mechanism, and I will get back to that later.

Slide 9:

In terms of the biological significance of fusion structure variability, I will not show any survival curves here. You have already seen some of them.

There are two sources of fusion structure variability. There are cytogenetic variants, where you get an alternative translocation with a related gene from the same gene family.

This has been shown to be prognostically significant in synovial sarcomas, and in alveolar rhabdomyosarcomas in a paper that just came out in JCO from Poul Sorensen and Fred Barr.
In other tumor types, it has been looked at and it has not been found to correlate with any clinical parameters.

The second source of structural variability is molecular variants, basically a different exon composition of the fusion genes due to different breakpoints at the genomic level. This has been shown by two groups for EWS-FLI1 , our group and Heinrich Kovar's group. It has also been studied in TLS-CHOP, but was found not to be significant. This was a study from our group.

Slide 10:

Now, the nice thing about this group of sarcomas is that we have a pretty good idea of what causes them, what is the key lesion. We know what to look for, for specific targets.

Clearly, the fusion genes themselves would make wonderful targets. Obviously, it is much more difficult to develop ways to block directly the function of the fusion genes in the clinical setting than in the in vitro setting.

The hope has actually shifted to the possibility that, by looking for key target genes, key downstream target genes of these transcription factors, we might have pharmacologically more accessible targets.

So, there is quite a bit of interest, especially now with the development of specific kinase inhibitors, to identify downstream targets which are involved in signaling.

One thing that I would draw your attention to is that, even though it has been almost 10 years that some of these translocations have been cloned, the list of target genes here is still very, very small. It is an area where we really need much more effort.

Slide 11:

Now, this group of sarcomas, then, is amenable to a fairly simple model in which we have basically some kind of precursor cell or stem cell. In some cases, we have a pretty good idea what lineage this cell might be. In other cases, it is a little bit unclear.

A chromosomal translocation occurs. It is thought to be a random event that happens to hit the right cell at the right time, causes the transformation of the cell. Then the transcription factor itself is, of course, important for the maintenance of the phenotype. It impacts on the behavior of the sarcoma and, in some cases, you get secondary genetic lesions, which I will get back to a little bit later.

Slide 12:

The second group of sarcomas are a little bit more problematic. We really don't have good models for what is going on in these sarcomas.

Slide 13:

I will put in this category the following sarcomas, including the tumors formerly known as MFH. So, these are sarcomas which typically show very complex karyotypes. It is hard to come up with any consistent alteration. There are some recurrent losses, some recurrent gains.

Slide 14:

I think it is interesting to compare these sarcomas to recent data that have emerged from epithelial cancers, from the Ron DePinho group, looking at the role of telomere erosion in generating karyotypic complexity in tumors.

So, the complex unbalanced karyotypes in this subgroup of sarcomas may reflect a process that has been called the chromosomal fusion bridge breakage cycle. This has been shown in some osteos and some MFHs as well. This is something that is due to advanced telomere erosion. This suggests -- the complexity that is observed in these sarcomas suggests that telomere dysfunction may be part of the pathogenesis of these sarcomas.

The basic idea is that, as cells reach a so-called senescence checkpoint, due to progressive telomere erosion, the cells that have p53 alterations can bypass that checkpoint, can survive that checkpoint, and the telomeres continue to shorten.

Eventually, when telomeres shorten to a critical length, this results in the telomeres being sticky, sticking to heterologous telomeres.

So, you get these kind of dicentric structures which, at the next cell division, simply break in random fashion and then refuse at the following cell cycle. So, you have the cycles of fusion, bridge breakage. This generates terrific gains and losses in the cells, aneuploidy, non-reciprocal translocations and an unstable complex karyotype.

Because of the karyotypic similarities of some of these sarcomas, can we try to make a similar model for this class of sarcomas ?

Slide 15:

This is what that would look like. Basically, we would have a precursor cell that is somehow subjected to inappropriate proliferative signals and loss of cell cycle control, possibly by RB pathway inactivation, autocrine signaling abnormalities and so on. This would cause, as this telomerase negative cell would continue dividing, telomeres to shorten, eventually it would hit so-called early crisis or senescence checkpoint.

There would be a selection for those subsets of cells that have p53 pathway inactivation. Then, this would allow further cell division and further karyotypic instability, and eventually there would be a selection for cells that have reactivated telomerase or have developed the alternative lengthening of telomeres mechanism to stabilize their karyotypes.

At this point, you would see -- this is one of the additional amplifications that you often see in sarcomas, that might be arising, as well as further deletions.

These would be high-grade tumors, possibly low-grade tumors, maybe somewhere in this spectrum, and this portion of the biological history might be preclinical.

I think this is a model -- it may be completely wrong, but I think that we can test it. The data are just not out there right now to really even begin to figure out whether this might be a likely scenario or not. We would need studies that compare p53 status, telomere phenotype and karyotypic complexity. There are just very few studies out there like that.

Slide 16:

Now, p53 alterations are a key component of this model. I would submit that, in this group of tumors, the group without translocations, that the alterations of the p53 pathway are present in the majority of cells.

I show this tremendously simplified schema of the p53 pathway, just to remind you of the different ways that the pathway can be inactivated in different tumors.

Of course, there are always p53 point mutations or loss. There is, famously, the CDKN2A gene which encodes both p14ARF and p16. As we know , p14ARF regulates the p53 pathway. So, that is another way of inactivating the p53 pathway. So, that is 9p21 deletion.

Then, something that is often seen in sarcomas is amplification of the 12q13 region. Interestingly, this causes co-amplification, in most cases, of MDM2 and CDK4, and I show you an example taken from osteosarcomas.

Slide 17:

So, there are several ways to inactivate p53 in these sarcomas.

Now, I would like to compare these two biological groups of sarcomas. I think it is interesting to really draw your attention to the differences. The sarcomas with specific translocations then have simple karyotypes compared to the more complex karyotypes of the non-specific sarcoma group. These translocations are reciprocal and specific, whereas these tend to be non-reciprocal translocations and non-specific. So, non-reciprocal translocations are really a mechanism for chromosomal gains or losses. The French cytogeneticist, Bernard Dutrillaux actually published an interesting article in the French literature where he was making this kind of case, and he actually calculated the average age of diagnosis in these two groups. So, this subgroup of sarcomas tends to occur in younger patients, but not exclusively. We know that extraskeletal myxoid chondrosarcomas can occur at a fairly advanced age.

As I will show in a minute, the p53 pathway alterations are relatively rare in this group of sarcomas, but I believe are much more frequent in this subset. As a corollary, the prognostic impact of p53 pathway alterations is strong in these sarcomas, but weak in these.

The incidence in p53 transgenic mice: these sarcomas commonly occur, these, rarely if ever. I don't think there is a single Ewing's sarcoma or synovial sarcoma that has ever occurred in a p53 transgenic mouse.

Likewise, in other kinds of situations, such as Li-Fraumeni syndrome or familial RB, you rarely, if ever, get translocation-associated sarcomas. The same thing for radiation-induced sarcomas. Quite rarely do you get translocation-associated sarcomas, whereas you get typically the other type.

Slide 18:

Now, I just want to dwell a little bit on the p53 story. This subgroup of sarcomas shows a very strong effect of p53 alterations. They are present in very small subsets of cases, but they have a dramatic impact on survival. I would like to contrast that with the other group. This also shows similar data for myxoid liposarcoma.

Slide 19:

Contrast that to the other group of sarcomas, and actually Murray Brennan has already shown this graph this morning, looking at the impact of p53. We are not observing a good prognostic impact in this subset of sarcomas.

Slide 20:

Now, this is shown very nicely in a study from a German group which looked at the impact of p53 status. When they looked at all liposarcomas, including myxoid, pleomorphic and well-differentiated, it was statistically significant. When they focused on the myxoid liposarcoma subset alone, the prognostic significance actually increased, in spite of the smaller numbers.

Slide 21:

Likewise, if you look at alternative ways of knocking out the p53 pathway, you get the same effect. So, this is in Ewing's sarcoma and the patients with p16/p14 deletion do extremely poorly, and the survival curve basically parallels that of p53 mutated patients.

Slide 22:

So, I propose that, in this group of sarcoma, the p53 pathway is at least partly functional in most cases and is acting as some kind of safety valve, an apoptotic or senescence safety valve on the cellular effects of the fusion oncogene. So, the p53 inactivation is uncommon and it is usually preclinical, but it is strongly prognostic.
In this subset of sarcomas, we get p53 pathway inactivation either as an early event or it is selected, in most cases, to overcome this telomere erosion checkpoint that I mentioned.

Slide 23:

So, where are we now? I think we have come to recognize the heterogeneity in sarcomas, obviously pathologically, but also at the molecular level.

We recognize the specificity of these different associations between fusion genes and histology. I think we are beginning to appreciate the impact that that has on looking at other prognostic markers in sarcomas. I think I mentioned some of this.

Slide 24:

I just wanted to briefly mention a parallel evolution in cancer genetics thinking. I think at the beginning of the 1990s the mantra, the paradigm, was multiple hits. It didn't matter which hits they were, but we were always looking for multiple hits.

I think we have moved from that to a much more functional viewpoint of cancer, and it was very nicely presented by

Hanahan and Weinberg in their recent paper, in which they identified the six key hallmarks of a cancer cell, the six pathophysiologic processes that a cancer cell needs to achieve. I think we need to begin to look at sarcomas in this same framework.

Another observation that they brought out in this paper is that the hits actually may come in any sequence and, in some cases, different alterations may provide one or more of these different hallmarks to the cancer cell.

Slide 25:

So, what do we need to do to move forward? Again, this is a fairly idiosyncratic list. For the sarcomas with specific translocations, I think we really, at this point, need more comprehensive target gene analysis, and that is, I think, going to be happening through new techniques like cDNA microarray analyses and chromatin immunoprecipitation analyses.

We need to link these types of genes to the hallmarks of cancer. Again, how did these genes contribute to evading apoptosis, autocrine growth and so on.
We need to understand better the interaction between these oncogenes, these fusion oncogenes, and the p53 pathway.

We need to understand the telomere phenotype of this set of sarcomas, again, as very little is known. Do these sarcomas arise from telomere-positive stem cells or do they also activate telomerase in some different way?

Slide 26:

In terms of resources, I should note there are very few cell lines for certain sarcomas and none for others.

In terms of the other major subgroup of sarcomas, the one with non-specific alterations, I think we really need to define the p53 pathway in single series of these sarcomas in a comprehensive manner.

We really are just kind of going about it in a very piecemeal fashion at the moment. Again, we need to relate these alterations to the hallmarks of cancer. We need to understand better the telomere phenotype of these sarcomas. Does the model that I presented hold, or is there something else going on?

I was going to mention very briefly the findings on p53 transgenic mice with shortened telomeres, but I think I will breeze through this.

Slide 27:

In terms of the timing and relationship of common oncogenic events, again, we really don't know if p53 alterations are early or late -are they sufficient or not?

One way of getting to that question is to look at the sarcomas that arise in familial retinoblastoma and Li-Fraumeni syndrome, look at the other gene, not the gene that is mutated in the germline, but the other gene, the p53 gene, in the familial retinoblastoma cases, and the RB pathway in Li-Fraumeni cases.

Finally, we need to better understand why some tumors seem to activate the p53 pathway preferentially, using, for instance, 12q13 amplification, like in atypical lipomatous tumors, whereas other soft tissue sarcomas prefer, seemingly, to just mutate p53, and what set of functional differences are a consequence of that. I think I will stop there.