What we discussed in our working group, of course, did resemble all the things that were discussed in all the other working groups.

There is a lot of redundancy, but the redundancy is not at the level that we discussed the same, but we discussed the same issues from a different angle. Therefore, I will quickly go through our initial slides, because it is pretty much the same that was discussed in other groups and already brought up this morning.
The fun part comes toward the end, how to prioritize things that we will take to the clinic.

The working group was mainly comprised of people who are active in cooperative group settings. So, we have discussed the role of basic scientists to improve the field, we have discussed the role of translational researchers, we have discussed the role of companies, why they don't give us all this, but we have not yet discussed which one we will choose in a cooperative group to take forward.

We have very strict requirements to do so, and we are even more strict than companies because, if you are engaging in a phase III trial, you can plan for phase III trials in stage IV or stage III melanoma every two years or so.

It is a very high investment also on the cooperative group side to engage into a trial that parallels the investment that is done in the industry. It just binds all your forces.
Therefore, we have to do this prioritization exercise in order to select drugs or modalities we will test even in cooperative groups through a process that has logical and non-logical components to it.

We were kindly provided with a title to our group and there is actually one word to that title that we did not discuss that is a two letter word right here. All the other words we just discussed.

TOP

Before even starting to discuss all those words, we did discuss what could be the potential rationale to apply emerging treatments to high risk populations. I will come to the definition of high risk later.

High risk means not the (??) tumor patients, it is clear from the beginning. So, the potential rationales that are being cited again and again is that the treatment may be more effective with smaller tumor burden, or that we need more time for the treatment to work, such as the case for immunization.

Those notions are cited again and again. They may be true, but there is no empirical evidence really supporting these notions.

What we have really to do is to seek more evidence in patients with limited metastatic disease, to try to look at their occult disease component, where there is a difference in the small metastatic lesions and the occult disease, and there are emerging techniques to do so.

Now, after having said this, the question is, what is the main goal when we discuss applying emergent treatments to a high risk population.

Our main goal is that we have to identify, through an efficient screening process drugs that hold a sufficient promise to be brought into phase III trials. That is what our obligation is.

Once we have done so, we have performed the trials. That is fine, but we have to select which one to take forward, and that is not easy.

TOP

So far, we are used to the clinical staging system to define patients at risk for something.

That is prognostic factors.
The main problem is, those prognostic factors are unlikely to tell us whether a patient will have an effect of a given treatment.

So, those prognostic factors -- and there is an important distinction -- are not predictive factors for response to any treatment.

It is totally unclear whether a given drug that has some effect in stage IV will have the same effect in stage II, it is just the different methods to detect that effect, or whether it is a differential effect in stage IV or stage II.

Usually the clinical development goes this way. Start at stage IV disease and then do slowly into earlier diseases, but in order to really accept that this is the right paradigm, we have to get more knowledge whether those prognostic factors are also predictive factors for the type of treatment we are doing.

What our hope is that, by incorporation of new modalities, such as high throughput methods, microarrays, etc., we could be able to define and better evaluate level of risk in patients, and could define predictive factors.

That is not necessarily a target for therapy, then, because the predictive factors may be totally unrelated to any mechanism, but at least it could be a predictive factor and it could help us select patients, subgroups, for certain therapeutic interventions.

TOP

One of the main problems that was discussed is that we are currently lacking efficient monitoring tools of treatment outcome in the absence of evaluable disease, without doing a phase III trial.

The question is, what end points should be thought by us in large cooperative groups, in order to select drugs that we will put forward into these phase III trials.

One possibility that has been done again and again is disease free survival, overall survival, in non-randomized trials, phase II cohorts, compared to a good selection of historical controls or concomitant controls.

The problem is how to select these controls, but at least these methods are able togenerate some hypotheses, no more, but they are just able to generate hypotheses.

Another set of patients that is a little unique to melanoma and maybe head and neck cancer is that we have patients with multiple relapse situations which you can treat by surgery.

Those are patients at highest risk and we can compare the prior course of these patients with the course after the therapeutic intervention that will give us some interesting ideas.

However, we are all aware that this patient subset is a subset of patients and the biologic behavior of the disease may differ from the biologic behavior of the disease in the vast majority of patients.

That, again, is a notion. We don't know whether that is true, but we have to address that notion. Of course, we would love to have surrogate markers that tell us whether we are doing what we want to do, and we are all working on that, on different levels.

In theory, they are much more efficient to tell us whether we are doing what we want to do, and I think we are at the stage -- and that holds true for monitoring now days as well as other modalities -- that at least we have defined the surrogate markers well enough that we ask the right questions with them, that they tell us what we ask them to tell us.

However, it needs validation whether what they tell us is meaningful for the patients, and there is a gap there. Something is lacking.

TOP

So, state of the field, I am just taking these titles and trying to work with these titles, although sometimes it is difficult.

To define high risk populations is clinical staging. That is giving us a statistical prognosis, and that is AJCC, but there are new markers, they are listed here.

All of them give us a statistical prognosis. That is, genetic predisposition of a given tumor for any prognosis. It is, again, statistical prognosis.

It is high throughput genomics. It is, again, at this point, statistical prognosis. This may turn into a predictive prognosis, but we have to look at all of these stages whether we can have predictive information on the efficacy of any sort of treatment, even in a small subset of patients. We just have to be able to think how to define a subset.

So, that was brought up, that sometimes the statistics of clinical trial statisticians may be supplemented by these statistics that are used by people interpreting microarrays. They are used to interpret hundreds of thousands of pieces of information just on a few samples, and that is maybe helpful, if that is what we want in order to generate hypotheses.

There are emerging techniques to identify occult disease and to follow occult disease, and that could develop into surrogate markers in the future that may be on the edge of being developed into surrogate markers.

TOP

The interesting discussion arose here, when we asked the question, based on the knowledge that approximately 150 compounds are currently in preclinical or clinical testing in oncology, which one would we put forward into clinical trials.

That question we have to ask today, that we did not have to ask 10 years ago. I will start with chemotherapy and recapitulate the story of colorectal cancer.

Colorectal cancer is a disease that does not respond to chemotherapy. That is true for 95 percent of the agents. There are three exceptions. Ten years ago, it was 5FU. Six years ago it was CPT-11, and then we knew that colorectal cancer does not at all react to cisplatin or carboplatin.

Then came a new platinum compound called oxaliplatin and, in the most recent randomized trial, that was the most effective drug, and that was to the surprise of many people.

So, I am not looking forward only to the Glevec of melanoma, but we should have open eyes to look forward to the oxaliplatin of melanoma.

We should really screen the new chemotherapeutic agents that come along in small trials. Then, what would the hurdle -- where would we set the bar with which to go on for further testing.

We came up with different numbers, but they were in close proximity, and we have basic agreement that in phase II trials, we have at least 30 percent response rate, meaningful response duration and there is controversy whether it should be a complete response fraction in that 30 percent response rate. Only then we will take a cytotoxic agent further in melanoma.

This number is open to any sort of criticism, but it was defined in the first group, and our scrambling, we came up with the same number without my influence, which sometimes can be strong.

The next one is the anti-angiogenesis field. We know approximately 10 agents that are antiangiogenic. Which one would you take forward in clinical development?

The hurdle is different because we know the mechanism of action is different. The toxicity pattern is different. So, we came up with the notion that, with the evidence of prolonged stabilization rates and approximately 10 percent response rate, we will take a compound further.

So, we would expect to see responses in metastatic disease patients, occasional responses, but then prolonged stabilization is something that would be important because that is close to the mechanism that we think those agents could have.

We also dared to ask the question of group C, what would make us take a specific immunotherapy agent forward into clinical trials.

We have so many antigens now, we are not short of antigens. Will we test them all in clinical trials? Most likely not.

We have to recapitulate what was the story of monoclonal antibodies. Thirty years ago, there were these people who had their antibody. Great, fantastic tools, an antibody and they were tried in clinical trials.

Of course, they all didn't work because we learned now that an antibody has to undergo a series of refinements, molecular engineering and then, in the end, you have a reagent that is optimally doing what it ought to do.

It is not this crude protein that is doing something close that you want to do and in certain instances you can make it do the things that you want to do, but by refinement, those antibodies now are working in the clinic.

So, what do we expect from an immune therapy agent? Almost all the antigens will, in certain instances, give rise to specific immunity. You can vaccinate with almost all the antigens.

So, that is not really enough to serve as a ground for further clinical development, not immunogenicity alone. We need some correlation with cause of disease, and we need some correlation with a surrogate end point with those responses, and we were pretty confident that group C would solve this problem.

Gene targeting therapy is the fourth class of agent, the one that we place a lot of hope in. It is a different one because there we really need, in order to go forward, the proof of concept that was raised early this morning.

We need a gene targeting drug that, in the first few patients, is doing what it ought to do. So, if there is a dramatic effect on the target and it alters in the target in a way, or the effects of the target in a way we want it to do, we go forward.

If it does not do so, really, we should not go forward with clinical development. We should look for better compounds that really have a dramatic effect on the target, a dramatic effect to go forward.

What are the areas of controversy? Well, the main area of controversy was really the selection of specific vaccines with promise for efficacy in occult disease.

TOP

That is where the controversy is and that we did not solve in our group and we stopped that discussion after some time.

TOP

Future directions. We thought it would be important to foster consensus mechanisms. It could be a conference, it could be a different kind of mechanism to define high risk populations by molecular immunologic methods based on an integrated approach. That would be helpful and important, and to look for predictive factors, as well as prognostic factors.

One important issue here is that we all have our banks. They were offered earlier this morning for analysis. Sometimes we find out that those banks that we have, have been collected the wrong way for a given assay.

So, it is very important to define a tool book, how do we need to collect our samples to make them suitable for analysis with the modern techniques.

That is a process that should involve both sides. The clinicians need to know how to collect the samples. What is important? Does it need to be in liquid nitrogen for five minutes or five hours? What is critical here.

Also, the basic scientists need to be taught by us, clinicians, that they make their assays adapt to the reality of clinical samples.

It is fantastic to have an assay that works on a sample 10 seconds after it has been taken out of a patient, but that is never going to be meaningful for any clinical development. So, the basic scientists have to adapt their methods to the clinical reality, wherever possible.

One approach to this is that the STP that has been cited, that has the monitoring workshop a year and a half ago, will have another workshop this fall in early November at the NCI again, to discuss the high throughput methods, and how they can be used to ask questions in the biological therapy of cancer.

I would suggest that, after that session, the melanoma people should get together and discuss melanoma specific issues, because if we do not discuss them, I would say, this year, we will waste another five years where we collect clinical samples in the wrong way, or create banks that are useless for the modern techniques.

The goal is to validate identification of prognostic markers through retrospective, if possible, and preferably prospective collection of strategically collectible materials, tissue and blood sample bank, of course.

It is a retrospective issue because, of course, we need to utilize our banks now. It is very likely that those banks will go down in value over the next five years because new methods are coming up and they may not be useful for the tissue that has been collected right now.

We should really make every effort to make our banks available now for these techniques because, retrospectively, we can only generate hypotheses that then need to be prospectively tested.

So, once an assay system is robust enough to be applied to a larger bank, then efforts should be taken to actually do this, and not wait another five years.

Also, we should encourage the identification of surrogate markers and their validation in relation to a given treatment.

These surrogate markers may be treatment specific or cancer specific, and that we have to take in mind, for every treatment modality that we are discussing.

TOP

What are the major impediments to the progress? Well, they are grouped in three groups. One is tumor related.
It is difficult to obtain tumor to study. This is where we discussed a lot the lack of accessibility to the primary tumor. That is important.

On the other hand, the clinicians know that the primary tumor is out of the patient. It is in formalin or nitrogen, whatever. Both agents are quite effective against the melanoma. It is out. That is not the problem.

What the problem is for the patient is the occult disease that will eventually progress and kill the patient. So, it will be interesting to study the primary melanomas, although that is not what will kill the patient in the end.

Another impediment is the chaotic genome of the tumor. That we cannot change. It is the lack of routine access to routine residual disease. That was felt quite strongly in our committee, and there are emerging techniques to sort single tumor cells, to study their biology.

There are groups that have not been fully represented at this meeting that take great effort to develop their techniques and have highly interesting results on the biology of occult disease, and we have to compare occult disease to metastatic disease to see what the differences are, and then see whether that is important and meaningful for the patients.

One method is increased use of biopsy fine needle aspirate. The other method is to sort occult disease from sources like blood and bone marrow, and other alternatives can be discussed. The goal is to investigate more the nature of minimum residual disease.

Another group of impediments are trial related, lack of surrogate markers of unequivocal clinical evidence. Studies often designed to test chemotherapeutic agents may not efficiently test biologic approaches that we know are not specific for melanoma, but we should not put too much emphasis on this because we have in mind the prioritization strategies we would use in order to take a drug forward.

Then there are regulatory impediments that are two sided. We have discussed the limited markers of melanoma, and proof of efficacy for regulatory approval is a high hurdle, and that may discourage the development of high risk novel therapeutic approaches for the disease.

We have to face another problem, and that is product development protection. Lots of those peptides we have will not be patented. So, a company will not have the product development protection possibility at this point, and their mechanism has to be discussed.

This, to a large level, is almost melanoma specific because the vast majority of peptide vaccines have been developed in melanoma.

The ECOG is taking one forward to a phase III trial currently. So, that is an issue that is not melanoma specific, but in the melanoma field, there is the lead in that discussion.

TOP

What recommendations can we make? We have to better define higher risk using molecular markers, and look which one is prognostic factor, predictive factor.

We have to standardize tissue and blood collection for high throughput analysis, by creating a tool book that is not a written Bible for the next 2,000 years, but that will be a constant process once these assays change.

We have to investigate the materials already available to generate hypotheses. These will all be retrospective or some whatever studies, but no matter what word you create, it will only be to generate hypotheses, because all those hypotheses will have to be tested in prospective fashion, and that is where I will stop.

TOP

PARTICIPANT: Actually, one of the things that we did stress in that meeting, too, Kelly McMasters had to leave, so I will bring it up for him.

Knowing what, as clinicians, we need to do with tissue specimens when we gather them is extremely important. A good example would be the Sun Belt Trial, where they have gathered serial blood samples in people prior to surgery, after surgery and then serially as they follow them until they recurred.

That would be a great resource for people to have, except that they extracted the PBMC layer and threw everything else away. So, there is no serum, there is no nothing. It is just the cells they thought the tumor was in. They didn't know to keep those things, or have the resources to do so.

The same thing with lymph nodes. The tissues were processed. They took out the RNAs and then threw all the residual tissue away, essentially.

So, if somebody would have come along and said, oh, well here is a basic set of things that you should do with your tissues -- so, for instance, you might not want the serum, but bank it, because there is a bunch of things that we could do with it.

You might not want the residual tissue because you might think there is nothing in it, but bank it this way, and this is how you should collect it.

For instance, can we do proteomics on serum samples when they have been processed with heparin? Should you put it in heparinized tubes or should you put them in citrate tubes? What should you do with those?

So, just kind of telling the groups what to do with their specimens and how to handle them and what should be collected and how it should be collected is an extremely important point.

DR. KEILHOLZ: That is what the tool book means, that we need to have constant contact with the basic scientists, once they develop new methods of looking at a tumor or another item of the patient. We need to know how that is done, what is the requirement for sample collection, and we need to tell the scientists whether this is realistic or unrealistic.

PARTICIPANT: Just one other point that we discussed again yesterday is the notion that there has been an extraordinary resource amassed in large pharmas, where some of the biggest bioinformatics groups now reside, in the belly of the Pfizers, the GSKs and so on.

They also have many of the high throughput technologies, since they have been in the business of doing high throughput drug screening for quite a long while.

The question is, is there any way by which we can forge alliances where we are truly in alignment, where we can take advantage of some of those technologies.

I am not suggesting that this is a trivial undertaking, but I think we might be able to use our substantial clinical reach, which is unavailable to them, by using technologies that they do have, for other uses.

DR. KEILHOLZ: You are right, that is a very important issue, especially seeing that the focus of the companies may not be melanoma. They may throw things away we would happily take.

So, that is something where the NCI could be instrumental, as far as I understand the system in this country.

DR. SONDAK: I just want to stress the sort of synergy between techniques to identify minimal residual disease reliably, and testing adjuvant therapies.

In melanomas, as you mentioned, with the staging system, we have large groups of patients with a low risk, smaller groups of patient with a high risk, tiny groups of patients with an extremely high risk.

If we could cull from all of those all the patients who had minimal residual disease and eliminate from our trials all the people who didn't, not only would we spare those non-tumor bearers the toxicity of therapy, but we would enrich our population and be able to test some things potentially in a group where almost you didn't need a control group.

PARTICIPANT: Vern, aren't they all long-term non-progressors, the ones you just defined, rather than non-tumor bearers?

DR. SONDAK: We don't know.

PARTICIPANT: And we don't have adequate assays to really fundamentally address that issue.

DR. SONDAK: I think, again, what I am saying is that the synergy, as we develop more understanding of minimal residual disease, either markers of it or predictors of progression, it would strengthen our ability to do trials and would actually potentially alter the very nature of the way we designed and conducted those trials.

PARTICIPANT: If I may make a remark regarding the current theme as to how to organize translational research, we must realize that the way we are sitting here, 60 people, we are not representing an organization or cooperative group or anything that actually has a working modality.

Many of you are participating in cooperative group mechanisms. The translational research item comes back and back and back basically on each and every point and each and every group that has reported here.

The EORTC is trying, or has tried to solve this issue to better methodologically accompany the process of improving clinical trial designs, by creating a translational research unit inside the EORTC, which is headed by a PhD, and which has its own staff.

On top of that, we have a translational research advisory committee. So, all protocols that are being submitted inside the EORTC mechanism are going to be handed -- and actually are already handed to the translational research advisory committee people.
The translational research elements that are part of the trial designs are weighed and are judged, and there are comments for approval. If obvious chances are missed, they will be pointed out. If not the best labs are actually involved in doing the translational research, that will be pointed out.

This translational research unit and that whole mechanism and this organization will be published in the Journal of Translational Research that Franco Marincola is spear-heading.

The point here is that you are part of cooperative group mechanisms. I think this is the type of model that you should all install in your cooperative group mechanisms, because it could have prevented -- well, we are missing a great opportunity in the Sun Belt Trial.

It could have presented a number of other greatly missed opportunities in cooperative group mechanism trial. It is actually a simple model to do it this way. You need to find the specialists, the judge, the protocols. You can only do this in an organized way.

We are not organized, the way we are sitting here. So, this initiative should be taken back, I think, to your cooperative group mechanisms. That is, in and by itself, a starting point in addressing what has now been mentioned in four consecutive reports.

DR. SCHUCHTER: Just a couple of comments on your slides about what to take forward in the adjuvant setting. I am not going to quibble with the numbers, but to go back to the colon cancer analogy, a lot of the activity in metastatic disease in some of these agents was in the 10 to 15 percent range, and it is an extremely active adjuvant therapy. That is one comment.

DR. KEILHOLZ: As adjuvant therapy, it is in the five percent range. It is not extremely active as an adjuvant therapy. It is in the lower five percent range.

DR. SCHUCHTER: The same for adjuvant breast and node negative, it is in that same response benefit range.

The second issue is that, for some of the agents we are talking about, like BRAF inhibitors, these may cause disease stabilization, just like you showed for anti-angiogenesis agents.

I think that second bullet should be expanded to think about other agents that cause disease stabilization that you might consider taking in the adjuvant setting.

So, they may not be cytotoxic and may be cytostatic. So, prolongation, disease stabilization is an important end point, but is probably appropriate for more than just antiangiogenesis agents.

DR. KEILHOLZ: Of course. Just targeting genes, we thought, should meet the end point. So, a BRAF inhibitor should inhibit BRAF and lead to the consequences we think that should have.

If it does not do that, we should discard it, and look for another inhibitors. If it does do that, then we will take it forward to the next step.

That argues for translational research in the first set of patients in order to see whether a compound is doing what it ought to do.

PARTICIPANT: I have a question about the identification of patients at high risk. Right now we have a list of groups of patients with positive sentinel lymph nodes and with also already primary tumors, in which they are definitely identified as being at high risk.

In the Melanoma Sun Belt protocol, those are the ones that are being treated with interferon. Are you suggesting these patients?

DR. KEILHOLZ: In order to develop new compounds, we should look at prognostic factors. We have clinical prognostic factors.

We should integrate the ones that we know have some prognostic implication. I have listed them, and we should integrate the high throughput methods in order to come up with a new set of factors that may be predictive for response to a given treatment.

Just having prognostic factors does not really help that much. It takes the assumption that the treatment effect would only be in a certain prognostic subgroup.

PARTICIPANT: But you know that those patients are at high risk. So, those are the ones that should be treated, isn't it?

DR. KEILHOLZ: What is high risk?

PARTICIPANT: Those are the ones that are referred.

DR. KEILHOLZ: High risk is the risk of a melanoma patient to progress and die. What Vern was pointing out is that we have primary melanoma patients, 20 percent of them are at high risk, 80 percent are at low risk, but the same number of low risk patients will die as the number of high risk patients.

If a small percentage of this 80 percent will die and a high percentage of the 20 percent, the absolute numbers in both groups will be identical. We do not think that this is the only solution to the problem to identify high risk.

We should look at a more molecular level with other methods. We should investigate how we can better define patients at high risk, in addition to their clinical stage.

PARTICIPANT: Ulrich, I am not sure if this was part of the purview of your group, but it was part of the discussion yesterday morning, and that is whether there are any thoughts you could give on chemoprevention strategies, identifying a different group of high risk patients, those at high risk for getting the disease and whether there are any criteria that were discussed about how one should bring forward agents in that arena.

DR. KEILHOLZ: We did discuss that very briefly because everybody on the table felt very uneasy about this and were not able to make anything close to a recommendation or proposal at this point in time.

DR. SONDAK: I would just say that, in that regard, the recommendations about choosing the drug, the agent, for testing based on response rate and some biologic outcomes, all of that is modified by the toxicity profile and the nature of the group being studied.

So, to take it to its extreme, you might have a super-biologically active agent that was highly plausible, would block the progression from a dysplastic nevus to a melanoma.

Yet, even if it had a moderate degree of toxicity, that might be unacceptable, despite tremendous activity, whereas it might be a no brainer to test in a patient who had had a resected metastasis.

PARTICIPANT: I just wanted to comment on the methodologies that are currently available to do some of the studies that you propose.

You can actually go and use the genomic DNA of the primary tumors of the studies that have already been performed and perform whole genome scans now, in order to identify which patients of the low risk group fared worse, and which ones of the high risk group did better, in the individual arms and compare that.

That can be done retrospectively. Obviously, as you say, that would generate hypotheses that would need to be validated, but it can be done right now.

DR. KEILHOLZ: That is what we suggest, is to do those things right now, in the very near future.

PARTICIPANT: Just one other thing that I think we touched on a little bit, but I think we talked about a little bit more yesterday and would be important in trying to dissect this important issue, that it would be a small percentage of a much more frequent group that may equal a bigger percentage of a less frequent group, but also the role of host factors.

I think the Sosman and Sondak dissection of the Malacine data is an excellent example of that, and that is being taken now into prospective trial.

Everything we have talked about, almost, has been about tumor factors, but the host factor, especially in the immuno and angiogenesis approaches, are going to be a big, big part of this.

DR. KEILHOLZ: Maybe also in the genetic field, looking at all the genetic polymorphisms.

PARTICIPANT: I think what Boris said is important, and this gets back to maybe some of the stuff Jim said.

We had a very simple paradigm in SWOG. We had looked at metastatic samples and looked at some spectral FISH. Basically, all we tried to do was get the initial samples from an earlier trial, the paraffin embedded samples.
In a year and a half, we have five samples. So, these kinds of issues -- and Vern is very well aware of the issues of, do we have to go back and have the patient sign another consent.

There have been so many barriers, and I think that is something that I think was mentioned in other people's discussions. We need to get this defined -- Jim, I think, brought this up mostly -- for all of us.

I think the IRBs will be rational eventually, and at least clarify for us. It is always better to get it prospectively, but there should be an easier way to also get it retrospectively.

PARTICIPANT: I think it is possible under the current IRB regulations, which get definitely more and more difficult, to really uncouple the patient identity from the tissue specimens, which clearly significantly facilitates the efforts.

You can just leave the crucial clinical information for the investigators who would do the molecular studies.

They don't need to know anything else. Then you are exempt and you actually don't need to get additional consent from the patient.

PARTICIPANT: This is just a small point along those lines, because I don't think Lyn Duncan is here, but she had some interesting remarks about that yesterday in group D.

At her institution, they have been able to get patients willingly to sign sort of a global consent that conforms with the HIPAA requirements that assigns patient material to essentially a unique patient identifier which remains with them forever.

Obviously, patients' willingness when they are alive from this horrible disease is generally high and, if they are dead, you are exempt from -- you can get the IRB to waive any requirements. So, it shouldn't be that difficult to get it done.

TOP