DR. ROBERT BELL: A question for Marc. Marc, we really obviously enjoyed your paper and you discussed this a little bit; but I wonder if you could carry on, perhaps, with the discussion about molecular classification in your presentation around telomere erosion.

It sort of suggests there is a lot of randomness that may be developing within the majority of soft tissue sarcomas that may preclude molecular classification, because of the random change that are occurring within a single tumor.

DR. LADANYI: I think that is a question best answered by Paul, or perhaps Matt van de Rijn in the audience.

I think we are already seeing that the sarcomas with specific translocations, specific genetic mutations, such as GIST, are very efficiently classified by microarray studies.

The remaining sarcomas -- it is perhaps still a little unclear how efficiently they will be classified. I think Paul is optimistic that they will be sharply defined.
I am perhaps still a little bit concerned that, because of their somewhat random karyotypic alterations, this will be a more challenging task.

DR. MELTZER: I think it is fundamentally a numbers game, having sufficient numbers of samples to represent the heterogeneity that is there in the patient population. I think if we have the numbers, the pattern will tend to emerge.

DR. BORDEN: Paul, let me ask a follow-up question to that. Leiomyosarcoma -- if you just take well-differentiated versus poorly differentiated, high-grade versus low-grade leiomyosarcoma, are there any array studies to suggest how patterns will emerge from a molecular standpoint?

DR. MELTZER: There are really very few studies in sarcomas of varying grade that I think really look at this issue adequately with enough samples. At this point, it is an act of faith to a certain degree. If you look at other cancers where larger numbers have been studied, then I think these separations are tending to be found. I think the assumption is that you will eventually find it. My worry would be whether some of these tumors have too much heterogeneity within the tumor tissue and we would really have to go to microdissection in order to adequately see the differences between subtle things in the population, but I don't think we can answer the question yet.

DR. VAN DE RIJN: I just wanted to add to that, that we found a subgroup of leiomyosarcomas that we call the calpone-positive group, and it is still very preliminary data on a very few samples.

I think that we will find subdivisions. It is only logical. If you look at 30,000 genes, you see some things that you don't see by histology or immunohistochemistry. So, I think that will continue on.

There are some sarcomas, I think, that will be very clearly distinguished from all others, like DFS, like you were looking at now, fibromatosis. They all live on very distinct branches.

I think cases where you histologically have a problem -- like with MFH in making up your mind what it is -- that will also be difficult with this technique.

That is where I think Dr. Meltzer is completely correct when he says that we would need very large numbers of cases to approach that. So, if you give the genes that are there and do make a difference a chance to speak up and separate the tumors in two different groups.

DR. LADANYI: I have a question for Dave Parkinson. In terms of target identification, you mentioned profiling, proteomics, expression profiling. It seems to me that, so far, the best targets have been targets that have an underlying genetic alteration. Do you think that that is going to continue to be the case, or do you think that even targets that are not genetically altered will be good targets?

DR. PARKINSON: We very clearly believe that targets that are not genetically altered are good targets. Somebody showed a slide earlier -- I can't remember whose talk it was -- but it strongly suggested that targets are most relevant in terms of their individual context. So, we have molecules under development for targets that we know are not genetically altered, but we know are critical to the physiology of the pathways that are driven by the genetically altered targets.

Now, it may be that the particular target is not druggable. By druggable, we mean the ability to develop small molecules that can actually inhibit a molecular activity.

For example, the kinases were thought to be not druggable in the late 1980s and now we know that they are an eminently druggable target. The current problem is with these molecular interactions that represent large protein interfaces, and they are not druggable. The whole industry is struggling with this, because there are many, many wonderful targets. If there are large surfaces, it is not possible to develop small molecules that interfere with that binding that actually can move into the cells. So, we have this huge disconnect.

DR. CHRIS FLETCHER: Actually, probably the person who can answer the question is Murray, because the truth is he is the one who is best at raising money.

We can all sit and do this for two days. How are we going to work out how to bring this technology to patients as a whole? We are ending up with so many tiers of care, you have to remember right now half the pathology labs around the world don't even have immunohistochemistry right now, let alone sit there and fantasize about cDNA microarrays, which we all think we have access to; but how are we going to extrapolate those patterns of gene expression in reality? Will it all have to be through immunohistochemistry?

Think about the drug stuff. I mean, Gleevec -- most patients with GIST right now, whether we like it or not, don't get Gleevec. It depends where you live whether you get Gleevec. It depends whether you are good at the Web. It depends whether you are in the United States or Europe, but they don't. How are we going to make these fabulously targeted therapies affordable? At the end of the day, we have to lobby to a much broader constituency, politically, federal government -- it doesn't matter what it is. We can sit here and play with ourselves about the joys of sarcoma biology for two days and go away all excited, but most patients ain't going to benefit unless we do something about it. We shouldn't delude ourselves they will benefit. All the things David's company can do aren't going to reach most sarcoma patients while any of us are alive.

DR. BRENNAN: I actually have a practical suggestion about that. It always amazes me that the patient walks in with relatively limited information and you can say you are going to live or die. Having cancer isn't the problem. Dying of cancer is the problem. I would think the focus is, in fact, on the patients who have the pathologies that you know will result in a bad outcome. Your resources should be focused using molecular diagnoses to define those populations, but you are really interested in targets in which there is a systemic risk that they will die; and that should be way ahead of all the -- I can't say it politely -- all the ways that we address the biology of tumors: that living with them is irrelevant. You do just fine, thank you very much.

DR. PARKINSON: As someone who deals with constituencies in 140 countries around the world, life is not fair. So, we have to decide what we want to deal with today. Do we want to deal with the technical problems that we need to develop new therapeutics? Or do we want to deal with the social issues related to dissemination of that technology? I feel both are important, but I don't think we can solve both at the same time. All I know is I have yet to go to a country around the world where patients are not getting, for example, Gleevec for GIST. There is a huge understanding. It is remarkable how the democratization of information causes information to be spread rapidly.

Mr. Scherzer is here from Life Raft. We had the same phenomenon with CML -- that often the patients know far, far, far more about what is going on in a particular, narrow therapeutic niche than their physicians do. So, the information is there.

Dealing with it equitably around the world is an enormous social challenge, but I think first we need to discuss the technology. I am a firm believer that technology starts expensive and ends up inexpensive. If it is necessary, we can ultimately develop those kinds of therapeutics and diagnostics relatively inexpensively. In the short term, it is enormously expensive, complex, and inefficient.

DR. SORENSON: A question for Dr. Parkinson and maybe the panel in general: it seems to me, by having studied individual genetic lesions in tumors for many years, one of the real challenges is to relate that genetic change to how secondary or complementary pathways are altered in a particular tumor. For example, if we identified gene fusion, then how does this relate to turning on of a survival signal or a cell-cycle progression pathway or various other so-called "hallmarks of cancer"?

I guess my question is, in your heart of hearts, do you believe that down the road the real way to treat a particular malignancy is to know the specific disregulation of all these pathways in a particular patient -- in other words, to chronicle p53 status, telomerase activity, PI3 kinase/Akt status, ras, MAP-k status in a particular patient -- and then combine whatever drugs are available to target those pathways is really the way to go? So, you are dealing with individual patient samples and whatever the pathways are disregulated in that particular pathway, as opposed to looking at pathways in a wide range of tumor samples of a particular tumor subtype. Do you think that really what we are going to be doing in the future is taking one biopsy from one patient, determining pathway status of all these relevant pathways that we have agents to, and then focusing our treatment based on that information?

DR. PARKINSON: With these kinds of tumors?

DR. SORENSON: Yes.

DR. BRENNAN: No, I don't think that is true. It is about high-risk groups. It is not about every sarcoma. It is about high-risk groups. That is where the biological reward is. That is where the treatment patient reward is. Leave the surgeons like me to fool around with the things that are relevant and worry about preservation of function. That is a technical exercise.

DR. FLETCHER: Jon Fletcher from Boston. Marc, you made a comment that intrigued me, which is that we had a sense of the progenitor cell in at least some of the sarcomas. I have always wondered about that, so this is a question for the panel generally. Do we know what the initial cell is, the non-neoplastic cell that is transformed for any sarcoma type? Obviously, it would be important at some level in modeling of sarcomas and understanding what these things really arise from, what they are. The final phenotype -- as we have learned from leukemias and various other entities where the spectrum of differentiation is fairly well catalogued -- it doesn't necessarily tell us anything at all about the initial, perhaps primitive, mesenchymal cell that began the ball rolling. The question is -- take liposarcomas, leiomyosarcomas, what not, or benign tumors that are very well differentiated. Do you know for any of these what the original cell is which is transformed?

DR. LADANYI: The short answer is no. The truth is that the folks in the leukemia world have these wonderful diagrams of hematopoietic differentiation. They can assign different subtypes of leukemias to different stages and different lineages and so on. Obviously, mesodermal differentiation is probably the same -- as complicated, if not more -- but the level of knowledge is much less.

So, ideally, if we had a more detailed knowledge, we could begin to assign different developmental stages from which particular precursor cells are derived. I think the microarray studies are contributing to elucidating those precursor cells. Paul may argue a little bit; but I think, for instance, the kit results on the GIST microarray analysis -- the observation that kit was very high on the list -- I think was very intriguing.

Even if you didn't know what the cell of origin of GIST was, once you saw kit at the top of the list, you would make the connections. Hopefully, that same process will occur with other sarcomas. Some sarcomas, we have a pretty good idea. Rhabdomyosarcomas, we have a pretty good idea. Ewing's, we have a more or less pretty good idea. Synovials, we have pretty much no idea what the precursor cell is. Hopefully, I think expression profiling might begin to shed light on that.

DR. MELTZER: If I could just comment, I think Jon raises a really important point in the fundamental biology of sarcomas, which is our poor understanding of mesodermal stem cell biology. I think this explains why -- as Marc pointed out -- that it has been so difficult to make mouse models: because no one knows what cell you should put that fusion gene into to allow it to survive and develop into a tumor.

It is true, I think, that arrays are helping us to get a bit of a picture of the histotype-specific genes that are markers for particular lineages; but what needs to be done is to extend this to the normal stem cell development, and look in developing embryos and developing animals to really define the specific compartments where those genes are expressed.

So there is a huge amount of developmental biology and stem cell biology that needs to be done if you really want to understand all this.

DR. SAXMAN: David, let me take the prerogative of the last question. I really appreciate what you showed us, the stepwise fashion through development of these things and figuring out what these are targeting, and I agree with you totally. I think at least some significant part of the corporate culture is interested in skipping most of these steps and moving as rapidly as possible through the clinical testing into the phase III testing. As soon as you begin to talk about biology and things, people become glassy-eyed, and they want to move as rapidly as possible to what they consider definitive phase III testing.

There is a great deal of pressure on clinicians, on Cooperative Groups, on cancer centers, to skip those steps, actually. Do you have any advice on, or thoughts, about that, and how to --

DR. PARKINSON: Let me point out that most of those steps are regulated. Nobody is interested in skipping those steps. I think what you are talking about is, once there is an actual molecule that is in the clinic, what the strategy for clinical development should be. Of course, there is the homerun school of clinical development, which is often used in the United States, and there is a direct relationship with the smaller the size of the company, the more likely they are to go for the homerun school of registration, which is, use accelerated strategies.

So, single arm phase II studies, response as an endpoint, not survival. That is what every company would like to do. The reality is, in cancer, most of the time it fails, and there are some major examples in the newspaper day by day that you can examine. However, I would point out that the paradigm I am suggesting, where you begin to try to isolate individual, biologically homogeneous groups of patients, starts to allow for the possibility of getting binary answers--yes or no--to particular drugs and particular niches. So, if you ask me, for example, does Gleevec work in chronic myelomonocytic leukemia? Yes. Does it work in dermatofibrosarcomas, in a couple of case we looked at, yes.

So, if we can isolate out individual entities where the biology is relatively straightforward and relevant to the therapeutic target, then actually the registration strategies become quite straightforward. Unfortunately, it is for that particular niche.

What you are talking about is the huge pressure within big pharma companies. One, never to develop agents against small niches because of the cost, the resources, and the fact that there is no commercial return at the other end. That is the pressure at the one end.

Two, once you have an agent, in some companies still, there is a huge pressure to not want to biological subset because you might end up with a drug that has these labels that restrict reimbursement. Don't forget, the gatekeeping for drug usage now is not the FDA often. It is the insurer, the provider, which is why we spend so much time trying to get a label that actually allows for reimbursement.

So, my advice to this community: define the biology of the relevant high-risk populations where there will be benefit from the use of these therapeutics. It wouldn't, obviously, be in the context of low risk because you already have good approaches and the risk is not there and you couldn't answer the clinical trials in any period of time, in any case. Define the populations, define them biologically. Find those targets that represent the targets already under development by the industry by publicizing or, if they believe the opportunity is great enough that it carries over to other tumors, they will come. Find the target, they will come, to paraphrase. That is the way therapeutics are going to get developed in this particular area.

DR. SAXMAN: I believe we will stop there. Thank you very much. I am sure there are many more questions and ideas and the breakout sessions will be an opportunity to discuss these further.

Our next speaker this morning is Dr. Sam Singer from Memorial Sloan-Kettering, who is going to talk to us about primary management of soft tissue sarcoma -- current approaches and challenges.

TOP