So, let's start with the first breakout session, which is the one that I was in charge of.

We talked about capitalizing on existing targets for therapy. Now, we had two sessions on targets, one of which was entitled, existing targets, and the second of which was entitled, new targets.

There was a lot of discussion between the two groups about what our individual purviews were supposed to be.

The way we anticipated this would break out was that existing targets would include an understanding of the genetics as it exists today, and we will talk about that.

New targets, we thought, might encompass new technology for the discovery of targets. It is difficult to talk about things the details of which are unanticipated, but it would include the application of profiling and array technologies, how we would validate targets in the future, what pathways might be important.

As it turned out, I think our two groups talked about some of the same things, and you will see that when I present these results of our discussions, and Lynn Schuchter presents the second group.

TOP

So, we started by stating the main problem of the field as we saw it. This was the consensus of the group and there was really very little disagreement about this.

First of all, we all agree that good targets for therapy already exist. Those targets, as we will show you in a moment, have met the criteria for validating.

Now, we are talking about biological targets here. It is also quite clear from our discussions, in the two target sessions, that the word target shifts back and forth to go from the biologic target to the agent or the drug that is being used.

We are talking about the biological target here, and the pharmacologic or immunologic agents that we might use to approach them would vary, but good targets clearly exist, and we will go through what we think they are in a moment.

In many cases, good agents directed at the targets do not, and we will highlight that as a major problem of the field.

In many cases, we do not understand enough of the biology to completely be able to prioritize how we should test those, and I think that has implications for Lynn's discussion.

In many cases, we do have enough of an understanding of the biology, and that is a function of the criteria for validation that we will talk about in a second.

TOP

So, we began our discussion by reviewing the status of the field with regard to therapy. You all know that there are only two therapeutics agents, but one of them being an immunotherapeutic agent and the other being a cytotoxic, it seemed a good departure point to reconsider the role that cytotoxic or target directed agents might have in melanoma.

I think we all appreciate that this has been relatively underrepresented in our thinking about the treatment of advanced and adjuvant disease.

It is interesting that, when we reviewed the data, the history of immunotherapy and the history of chemotherapy are very roughly in parallel.
The numbers of the initial trials, the response rate and the enthusiasm about the use of these agents is very closely paralleled.

For IL-2, we know the initial studies showed relatively high response rates in the 40 percent range and then, with the broader application of IL-2 and similar therapies to wider patient populations with more relaxed eligibility criteria, response rates fell to the 15 percent level that we know exists now, with complete responses in the range of about five percent, with durable complete responses being three percent in a selected patient population.

That is really roughly what happened with chemotherapy in the 1980s. Initial studies of polychemotherapy demonstrated response rates in the 40 to 50 percent range in a single institution, selected populations.

Finally, evolving to randomized studies, polychemotherapy has response rates that range from anywhere just above nine percent to the high 20 percent range, depending on which randomized trial you look at. Yet, we have discarded targeted and cytotoxic therapies.

Nonetheless, having reviewed that, we came to several conclusions about the similarities and differences.

TOP

We asked the question, is melanoma curable by either of these methods, and the consensus of the group was that probably yes, there are a few real cures for IL-2 and immunotherapy.

In the NCI's latest publication of their 374 patients they have treated, there are 10 patients who have durable complete responses over 30 months. Mike Atkins talked about those data a little bit.

That is about three percent of the patients, most of whom don't have brain mets, so maybe it doesn't represent the entire population, but we still think that those are real.

Polychemotherapy or chemotherapy probably also has a cure rate. People have seen chemotherapeutic cures, but it is very difficult as our emphasis on immunotherapy has shifted, to draw conclusions to compare the two modalities of therapy, because more modern methods of staging, more modern methods of care have not largely been applied to chemotherapeutic populations.

We do not know if the patients responding are the same in the two instances. There is really little way of deciding this.

There have not been randomized studies that directly compare immunotherapy alone to chemotherapy alone, so that we could stratify the patient characteristics.

There have also not been real sequential studies which may be, it turns out, a better way of deciding which patients respond to what, but some of the characteristics of responses are the same in the two groups. Skin only disease does better in both.

Ultimately we decided that there is an opportunity for cytotoxic and targeted therapy that might parallel at least the successes we have had with immunotherapy and that the opportunity to do genetics in conjunction with the clinical opportunities was strong, and that that would depend on our ability to acquire tissues, as we move forward.

TOP

We set out -- I should mention, Jamie Zweibel was our co-director. Jamie led a discussion about what are the considerations that allow us to validate a target, and Jamie also led a discussion about individual targets that do exist now.

First of all, we need clear evidence that a biologic target is altered in melanoma in some way. We defined altered as either having human mutations -- human mutations being broadly defined to include point mutations, deletions, aberrations of chromosomal structural expression.

Secondly, epigenetic silencing or activation, in other words, promoter control or amplification, other ways that don't involve point mutation, but clearly target an individual genetic locus.

A supportive model system is very helpful evidence for different targets that we will talk about in a minute. There are different levels of murine model support or lower organism model support.

When the model support goes in the same direction as the human mutation genotypes, that very much strengthens our belief that those are important targets.

Finally, pathway effects. We know that there are certain pathways that are repeatedly involved in malignancy and we will discuss those in a moment.

When a gene is found to be on one of those pathways, that, too, is supportive evidence that that is an important player.

For example, this is BRAF's position downstream of ras in a pathway already known to be important in melanoma. That immediately lent credibility to the mutation rates that we were seeing.

So, all those things taken together -- in other words, the context of the alterations, the specificity of the alterations and replication of the alterations in other systems are important aspects of what it takes to validate a target as a biological target for melanoma.

Now, functional assays are important. Probably the most important functional assay is if you can demonstrate the conferring the malignant phenotype on a melanocyte or melanocytic cell with the alteration, reversing it by reversing the alteration is important.

There are obviously many biological strategies to approach this type of experiment, ranging from genetic deletion methods to small inhibiting RNAs to pharmacological inhibition.

It is quite clear that, if you can reverse the alteration and you reverse the phenotype, that is important supportive evidence that there is a potential pharmacologic approach to the target in question.

Finally, in some circumstances, we do have early human trial data, almost immediately upon the discovery of a potential target.

I think this is going to be increasingly true when the pharmacological industry, big pharma, has a lot of agents in hand, testing them in important genes and important pathways.

It is going to be the usual case that an unexpected player in one of those pathways turns out to have relevance to melanoma, and agents may already be in the pipeline with regard to targeting that pipeline.
So, in some cases, you will have trial data right off the bat, and that was true of BRAF as well, and it may be true down the line for other targets.

TOP

So, we set out to discuss the targets that already exist. Many of them were identified in our lectures on Monday morning.

This is sort of in the order that we considered them, and we ended up, because BRAF is so current, talking about BRAF almost as a case study.

The most important pathway right now for melanoma is BRAF, ras and PTEN. So, BRAF is immediately downstream of ras in the receptor tyrosine kinase signalling pathway. PTEN is down an additional pathway that ras has function with. We didn't hear much about them on Monday morning, but ras also binds up PI3 kinase, ultimately affecting its action through the AKT PKB, and PTEN is an inhibitor of that pathway. So, it was felt that BRAF, ras and PTEN had definite import for melanoma.
PI3 kinase and AKT are also on the PTEN pathway. Concurrent to the alterations in the receptor tyrosine kinase signalling pathway, are alterations in cell cycle control.
p16 is the most clearly implicated of these genes, CDK4 as well and, of course, the p16 locus is co-localized with the p14 ARF gene on the p53 pathway.

So, together, CDK4, p16 and p14 are the genes that are most frequently mutated and the RB1 and the p53 loci are the genes that are really the central affecters of those pathways.

To sum that up, the best genetic evidence for the causality of melanoma involves the ras pathway, by both of its arms of effect, and the RB and p53 pathways.

In general, most melanomas, especially in cell lines, demonstrate alterations of those pathways. So, those are the best targets.

That is not unique to melanoma, of course. p53 is a central culprit in most malignancies, as is RB1. The problem of generating targets for tumor suppressor genes and specific targets for oncogenes is not unique to melanoma obviously.

We then turned our attention to additional sets of targets. The first and most important of these are the apoptosis genes, particularly in BCL2.
BCL2 is controlled by microphthalmia, as you heard from David Fisher, and microphthalmia having its effect in melanoma cells being variable. BCL2 expression is variable as well.

Nonetheless, BCL2 has been targeted by an agent G3139, and that clinical trial is done. We do not know the results of that randomized study with DTIC, but I think that will be an especially relevant target in the next few months, when those data become available.

We talked a lot about the melanin biosynthesis pathway. Individual targeting of any of the melanin biosynthesis targets is unlikely to have a controlling effect on the growth or proliferation of melanoma.

However, there may be ways to modulate the expression of these, especially in the context of immunotherapy.

For instance, we heard about MSH signalling and how MSH could enhance MITF signal transduction and, ultimately, enhance the expression of the tumor specific antigens that we all use as targets.

So, melanin synthesis pathway may not be a cytotoxic target, but may be something that we can target down the line in conjunction with immunotherapy.
Microphthalmia itself, Dr. Fisher was pessimistic about its being a target, but the rest of the group was not.

A number of individuals felt that if we could augment microphthalmia's activity, that we could push cells to differentiate, and the complete lack of microphthalmia activity is lethal for melanocytes.

So, these are also potential targets, even though they have not been explored pharmacologically.

We heard about the ski gene. The ski protein is a protein that is a regulator of TGF beta transduction. There is less broad consensus about its importance, but there is clearly evidence emerging from several laboratories that, by virtue of its binding to the SMAD proteins, that it is an important controller of TGF beta signalling.

We also heard a little bit about NF kappa B and its importance in melanoma, also not as broadly accepted and as validated as the other levels of target consideration, but potentially of importance in melanoma.

I think, again, NF kappa B is something that is not important only to melanoma. It has implications for a broad variety of malignancies.

So, these are the targets we talked about. It is not an exclusive list. There are other potential targets, I am sure, and you will see, when we go through the NCI portfolio in a second, that many of the targets that are potentially targeted now are not on this list. For instance, the proteosome, we didn't talk about the proteosome, we didn't talk much about stromal considerations as well.

TOP

So, the first level of validation are the first genes that I talked about. These have several levels of evidence.

They are either mutated or altered in human melanoma.

There are murine models that support their use, and there is already evidence in some cases of therapeutic effect of interferons with their signalling pathways.

BRAF is at the top of our list. There is a tremendous amount of enthusiasm now in the melanoma-wide community about potential treatment through BRAF.
Some caveats emerged, and those include the lack of understanding of the effect of this gene right now, of this protein.

The BRAF protein has pleotropic effects. There are several isotypes. There are also several raf molecules, but BRAF is clearly the one important in melanoma.

The exact biology of it makes it difficult to predict what is going to happen when we interfere with it.

It is a downstream player of ras, but ras has several effects, and I have already alluded to one of them being the PTEN AKT pathway.

It is not clear whether inhibition of raf alone will be treatment for melanoma. You saw from Richard Mule's work that it is not as potent an inhibitor of cell growth as ras is itself.

So, there is a lot of empiric information that has yet to be gleaned about the importance of BRAF and how we can best therapeutically approach it.

One of the cautions that you will see on a slide in a moment, but I want to make it here, is that the committee very strongly felt that lack of evidence of utility of an individual pharmacologic agent does not invalidate the approach to a target.

In other words, if we start with the Bayer drug, for instance, and it doesn't work for BRAF, that doesn't overturn the wealth of accumulating evidence that this is an important gene in the pathogenesis of melanoma.

Other approaches that are trying to inhibit or trying to interfere with its action are still valid, even if a single pharmacology drug fails.

I think for emerging players like BRAF, we are more likely going to have multiple potential agents that are targeted. That is clearly true for BRAF. I think there are at least four pharmaceutical companies now that have BRAF inhibitors in various stages of testing.

That is not always true for some of the genes that have been around longer. It is clearly true, partially because it is an oncogene with a specific mutation site, which lends itself more to pharmacological intervention.

So, it is going to be possible to test multiple agents. The failure of one does not imply that others will fail.

Ras is an important gene in melanoma, but it is only mutated in 10 to 15 percent, at most, of melanomas. The pharmacologic industry has spent an enormous amount of money trying to inhibit this gene.

Finding specific inhibitors of ras has been very difficult. We have all done probably farnesyl transferase inhibitor trials, but those agents are not very specific for ras.

I think it is important to point out that in human melanomas, almost all the mutations are in NRAS. There are three ras forms that differ at their carboxy terminus.

Although, for instance, the mouse model uses an HRAS, in humans, NRAS is almost completely specifically involved, and there is not a whole lot known about what that implies for how this gene works in melanoma.

The p16 pathway is important. There are genes out there that are cycle independent kinase inhibitors, but p16 itself, again, is a tumor suppressor that is lost. Replacing a lost gene is a different problem from down-regulating an over-regulated gene.

There are no good mimics, for instance, of p16 right now, but modulating that pathway clearly is felt to be a priority, given that this is the major predisposition gene for melanoma.

BCL2 I have already alluded to. We do have an agent out there, an antisense G3139, an 18mer that targets the first six amino acids of the cDNA. We are waiting for those data.

The PTEN pathway, the other arm of ras, also appears to be routinely involved in melanoma. AKT over-expression or over-activity is probably an important component of the pathogenesis of the disease, but there are not good inhibitors of PI3 kinase for cancer or AKT yet.

All those genes have multiple levels of evidence, from human mutation to mouse models, and evidence that reversing each of their actions reverses the melanocytic malignant phenotype.

TOP

The second level, or the other genes we talked about, microphthalmia, is a pleasing target. It seems like there might be something in going after it, but the evidence is less firm.

In particular, there are no clear cut mutations or alterations in microphthalmia. It is relatively melanoma specific and, in fact, it is an immunohistochemical marker for melanoma.

Some of the other criteria that we talked about for validation are not met. That is true for NF kappa beta and ski. These are not genes that are altered or mutated.

The melanin synthesis pathway, again, is a pathway effect that is specific, but not shown to be specifically mutated and pathogenetically necessary, true also for MSH.

Now, one of the things that is clear is that one of the criteria, we did not talk about testing very much, but when we talk about these mutations being pathogenetic, if the loss of the mutation or the alteration allows the malignant phenotype to continue, it is not very useful.

In other words, the sustained or maintained alteration or mutation is necessary, and for this level of validation, we don't know that to be true.
So, the reversibility of the phenotype with the reversibility of the alteration is an important criteria for validation.

TOP

I made this point earlier but, again, failure of a candidate drug does not indicate that insufficiency of biological target and should not stop development of that target, in our opinion.

TOP

Jamie then took us through some of the NCI portfolio right now. These are agents that are already in the clinic.

You will see that some of the targets that they are aimed at are not targets that we considered on our list, especially stromal targets, like the alpha beta 3 integrin, ccelingitide and meti 522. There is no way out for celingitide right now. There are no trials up and running.

There are VEGF antibodies, the PS341 agent for melanin, which is a proteasome inhibitor, and you can read down the list.

Some of these are aimed at targets that we talked about, some of them are not. CTLA, obviously, is an immunological molecule as well.

So, the way targets have been developed so far, most of these agents were not developed specifically to look at their utility in treating melanoma.

Many of them -- PS-341 comes to mind -- had broad applicability and are often tested in populations with larger patient populations than melanoma.

I think that the feeling of the committee was that that is unfortunate. I think melanoma's reputation as a drug development vehicle has slipped in the last years, and we would like to make it clear to pharmaceutical companies that there is a compelling rationale for testing new agents in melanoma, or designing agents specifically to treat melanoma.

TOP

So, what are the problems that we identified? The overwhelming problem is the generation of new agents. Some of these genes give rise to proteins that are lost in the transforming process and, as I said, replacing a lost protein is very difficult to do.

So, we are really starved for new agents that address some of these pathways, and that is something that the melanoma community obviously does not have a whole lot of control over.

The NCI can help us here. The NCI can help foster the opening of the pipeline to the melanoma community, but I think that is the best we can do right now to solve that particular problem.

A major issue that we identified is the lack of multicorporate collaboration. To give you one example, if BRAF inhibition by itself does not work, a potential hypothesis for why that might be is that concurrent inhibition of parallel pathways needs to be initiated to get an effect.

It is hard to inhibit two drugs at the same time in the pharmacologic environment as it stands right now, because we can't get drugs from different companies that might be useful.

You can imagine treating melanoma with a farnesyl transferase inhibitor and flavopiridol, for instance, one of which would attack the p16 CDK4 pathway, and the farnesyl transferase inhibitor that might inhibit ras.

That is perfectly consonant with Lynda Chin's model that you saw the other day that shows that ras and p16 together collaborate to make melanoma.

We can't do that study because the drugs are made by different companies and they are very unlikely to allow us to collaborate directly on a phase I/II trial.

So, this is a big problem and we have to find a way to break down these barriers.

Waiting for an individual agent to get to the phase II stage or approved before putting it in combination with another is a huge impediment to progress.

Understanding biology, this is a problem, and it varies with the level of validation. Clearly, for some of the genes that we have talked about, and their protein products, we know a lot about how their biology contributes to melanoma, but I would have to say that we don't know enough to rationally prioritize which ones we should go after first, and with the most vigor.

It might be that inhibiting CDK4 is the best way to go, even though only a few percent of melanoma patients have CDK4 mutations.

Maybe that is the best target for many other criteria that we don't understand right now, but we don't understand enough about the biology to be able to make those determinations.

So, largely we are driven on frequency of mutations, how common an alteration is in melanoma and other relatively subjective criteria about what we think about drug development aimed at that target.

Patient eligibility is an important aspect of this. By patient eligibility, I mean, we talked a lot about whether, if you have a targeting agent, you need to prove, a priori, that a patient going on a clinical study has an alteration that is hit by that target.

So, when we are designing our BRAF trials, do we really need to demonstrate BRAF mutations before putting people on a study, and is that true for all the other targets that we are looking at.

It has not been true, for instance, in the BCL2 trial. Nobody looked at BCL2 over-regulation as a criteria for entry into the BCL2 study. I think that is an open question.

For very frequent mutations, putting patients on study without looking means that only a few percent of the patient will be misapplied but, for rare mutations, we have markedly reduced the potential patient population for accrual, and yet the rationale for treating those patients lacking mutation is lacking.

As a consequence, I think that that is an issue that we are going to have to deal with and, in consonance with that, we have to have good assays for target effects.

For instance, if we consider the HER2/neu story, you have already heard several times that if patients had not been specifically entered into the trial with HER2/neu overexpression, the effect of HER2 inhibition may have been lost. That actually drove the development of a better HER2/neu assay.

In many of the pathways that we are talking about, we do not have good assays and, in particular, getting the tissue is sometimes difficult.

So, biopsies for many of these patients, pre and post therapy, may be necessary. That is something that we are not used to be doing.

It is relatively facile for patients who have skin metastases only, but it is not always easy. As a consequence, if we are going to really approach specifically targeted therapy, we are going to have to understand the genetics of the tumors involved.

TOP

So, all those things go together to demonstrate that we feel there is a tension between the need for understanding how these agents work, and the empiric push for clinical benefit.

Again, to go back to the Genasense trial, the G3139 trial, here is a drug that is designed to down-regulate BCL2 expression. In vitro, it does this very well to the tune of about 70 percent in 48 hours in most assays.

As a consequence, it would be nice to demonstrate in the clinic, that when you give the drug, that that is what you see in some target cells, especially melanoma, but you could also use a surrogate target.

In fact, I think the companies, when they design these agents, are more interested in seeing a clinical effect.

If they can't demonstrate a biological effect and yet they see clinical benefit, they don't really care. That tension of cost reduction and ease of carrying out of trial is in direct competition for our need to understand better how some of these agents work.

Especially if we are going to consider multiple agents that are aimed at a single target, we would like to know, when we do a study, if it doesn't work, why it doesn't work.

Is it actually achieving its purposed biologic effect, or is it an issue of pharmacokinetics or pharmacology.

Those are questions that we can only answer if we have good end points. So, the committee felt strongly that this tension exists, but we would really like to come down in favor of understanding the biology and pushing to biopsy, when we can.

The final problem that we can't surmount right now is the emergence of resistance. This is most clearly demonstrated in the Glevec story, that once you treat patients that are specifically targeted at an oncogene, the tumors will evolve.

They have plastic genomes and there is a very strong selection pressure for that plasticity to result in natural selection of resistance.

So, the emergence of resistance is an important problem, and it is something that is probably best handled with concurrent therapy rather than monotherapy.

TOP

So, possible solutions, we talked about our clinical trial structure. We are in relative agreement that clinical trials must focus on advanced disease, that is, patients with measurable disease, so that we can measure response.

I know this is controversial, and I expect different working groups discussed this in different contexts and came up with different agents.

I know that there is a strong feeling that we should start many agents in the adjuvant setting, but I would have to say it was not unanimous but the consensus was that advanced disease would be where we would start, that we do need biological end points -- I have already discussed that -- and that minimal residual disease might be best approached later if we could demonstrate an effect.

We need help from the NCI in getting new agents, and the experimental procurement of drugs might be a potential solution for that.

By that I mean, if we can't get a drug that is commercially in some phase of testing, it might be possible to synthesize the drug and use it in an experimental setting with no hope of commercial applicability, so that we can understand it better.

In practice, I think that would be very difficult, but we discussed it at length because in some instances we are able to do this.

For immunologic reagents, for instance, we don't need to get a peptide from a company that supplies it. We can synthesize it on our own and test its effects.

It is maybe a little bit more difficult for some complicated small molecules but, nonetheless, it is a potential strategy.

We talked a lot about the patent implications of that and so forth, but I think that is something that is going to generate more discussion in the future when we are trying to look at potential combination therapies and to circumvent legal barriers to understanding that.

TOP

Finally, the recommendations. The first one was easy. Test BRAF and test it quickly. That is obviously happening.

There are clinical trials up and running and I think the example set by the rapid movement of understanding from the basic bench to the clinic with the BRAF story is probably going to be exemplary for what happens in the future.

So, if we do this well and we do this with some understanding, if we understand the mutations or lack thereof in patients that we are treating, if we get some sense of the biological end points and the extent to which we are able to achieve them with therapy, it will be useful.

The first BRAF drug there is the Bayer drug, obviously, but there are going to be other agents that we can test, and I think there are going to be opportunities for a variety of combination therapies, even in combination with conventional chemotherapy.

We also very clearly voiced the opinion that we need to foster melanoma's role in drug development. Melanoma is a genetic disease, as are all cancers, and it has been clear that over the last years that our emphasis on immunotherapy has resulted in a lack of emphasis on melanoma as a substrate for drug development.

There is a very strong sense in the basic science community here that the biology of melanoma is relatively underrepresented in our thinking about the treatment of advanced disease.

I think we would like to rectify that. I think we need to think about target pathways in melanoma. We need to think about ways that we can make it clear to the pharmaceutical industry that melanoma should be a substrate for their testing.

We need to facilitate biological end point testing. Our trials, especially at the cooperative group level, where large numbers of patients are entered into clinical studies, that we have to think about getting tissue samples. It is not that we aren't thinking about this, but we are re-emphasizing the idea.

We would like to encourage the translational and basic participation in the cooperative groups. I think many of the basic scientists here would like to take a more active role in thinking about clinical trial planning, in thinking about if we are doing immunotherapy trials or other large trials that provide us the availability of potential biologic samples, that we work in the basic science community for enhancing the availability of those samples.

I think there is an opportunity here to enhance the dialogue that will only be continued by Meenhard's meeting and others like it, where we can foster a greater interaction.
So, that is it for our group and we are open for discussion now.

TOP

JEFF SOSMAN: A couple of things. One is, I certainly wouldn't look at the Bayer study up to now as a success story.

In fact, I think it is an example of the difficult hurdles. The people in Great Britain actually went to Bayer before the paper was published, and they had no success.

Many of us went to them right after the paper got accepted and, even in the face of their preclinical development where they showed it had activity, they resisted going ahead with it.

I actually think it is a model of failure in melanoma. What we have to address is that they would rather go ahead and do a large phase III study in lung cancer with carbo/taxol in BRAF than do an outright study in melanoma where I felt that if they had good phase II data, just like GIST, they didn't need a control arm.

We all know that the NCI, or at least my view, is that the ODAC and the FDA would rather have a placebo control for most of our new treatments.

DR. HALUSKA: So, Jeff, what was their resistance to getting involved in melanoma right away?

DR. SOSMAN: I think it was the disease. I think they felt there were a lot of failures previously, that it is a rapidly progressive disease, that the market was not great.

People don't live very long. They can't take it for a long time.

Obviously, it is a corporation that is under a great flux as well. I even had less success with Pfizer. They just wouldn't even move, and they have an oral MEK inhibitor.

I know Lee Allen -- I know we shouldn't use names, but they had a preclinical person who was very interested in this. She really wanted to push it.

The clinical program had no interest at all in developing it. We talked about phase I's in melanoma. We talked about making them totally directed toward tumor biopsies and really trying to learn if this drug has any effect. There was absolutely no interest.

I think that, really, this barrier is a huge one. I think that is one message. I know, from my discussions with our cancer center director, and hearing Dr. Von Eschenbach last night, I think that is something we need to push, this importance of trying to form some more objective way of evaluating new drugs where there is some buy in from pharmaceutical so that they are protected, and yet, we can do the obvious experiments.

PARTICIPANT: I share most of your concerns, but I just want to come out a little bit on the side of Bayer. I mean, they did eventually agree to a trial, and there are two trials going on at the moment and more planned.

While I agree that it is very difficult to get over them, but we shouldn't beat Bayer with a stick here, because we are in a kind of dangerous position of really pissing them off and then they just pull out.

DR. SOSMAN: I agree, but actually, Richard, I wouldn't characterize these as two real new trials. I mean, this was something that got sort of somehow twisted into this picture.

I don't think it is just Bayer. I mean, if you talk to George Demitri and Chuck Blanke, there is no question that Glevec would never, ever have been looked at in GIST if it was up to Novartis. Novartis resisted for years until someone had the money to buy a component of Novartis or Sandoz, and basically get the drug for himself, so he could respond to it. This is a theme that goes way beyond Bayer, I agree.

DR. HALUSKA: Jeff, what I am hearing is two issues here. One is the perception of the biology of melanoma is not appropriate for the testing of new cytotoxic agents, and the second is market share.

We can't do anything about market share except continually hold up the Glevec story and evidence of it demonstrating that a small market can have an enormously disproportionate salutary effect for a company if you are successful.

If we have 8,000 metastatic deaths a year, that is our market for this therapy. Yet, we have to hold out that the specificity of the genetic alterations make it worth doing.

I think an undercurrent of this committee's deliberation is that there is this perception that melanoma is only a therapeutic model system and it is not something we should be testing these drugs on.

I think that is what you are hearing from these groups. I don't think melanoma is that much of a worse disease than lung cancer.

DON MORTON: I just wanted to introduce a bit of history into this meeting. For those of us who were around in the 1960s and 1970s when the only drug available was DTIC, the NCI, which developed DTIC and did the clinical trials, actually distributed it for a long, long time. I think it was into the 1980s, even though the NCI did all the work, before there was a pharmaceutical company that was willing to put it through the FDA and sell it.

The facts are that the pharmaceutical company does not see melanoma as a big market. I had the same problem with my vaccine.

I knocked on every major pharmaceutical company's door in the country, in the world, actually, for seven years, and basically they just discounted it. They figure this is a small market.

That is the problem. I don't think this problem for melanoma is going to be solved until the NCI has its own drug development program and decides that they are going to take these things through and show that they work.

Taxol is another example, one of the most successful chemotherapeutic drugs in history. It would never have been developed if not for the NCI.

You know, my vaccine, if there wasn't -- if the peer review system hadn't provided grant support to start this trial and get to where we are now, it would never have seen the light of day.

I think you have to face the realty of the economics. These are businessmen, and these big corporations, there are multiple conflicting groups that have different priorities.

You know, you are in marketing and I am in drug development or something, and you know, we may not get along. So, anything you bring along, I am going to -- it is like a big political bonanza.

We got some drug companies to be interested in the vaccine, but the warfare between the groups, it has to be a consensus to go forward.

If we have to wait for corporate development to get effective drugs for melanoma, I am afraid we are never going to get there.

DR. BAR-ELI: I want to shift from this discussion. If you can go back to your list of targets?

DR. HALUSKA: The NCI list or my list?

DR. BAR-ELI: Your list. I was not in this committee but is seems to me that you missed some very important targets there.

DR. HALUSKA: Like I said, we weren't exhaustive.

DR. BAR-ELI: I think it is worth mentioning, like in the category of adhesion molecules, we need to mention the MUC-18 MCAM which is over-expressed in advanced metastatic melanoma cells which, in my opinion, is a very good target to treat melanoma, especially after surgery.

Indeed, we recently developed fully humanized antibodies against this molecule and, in preclinical studies, it is very promising and it is now in phase I clinical trials, and maybe we should discuss it in the immunotherapy session.
O

ther pathways that we should add there is certain transcription factors like CREB/ATF1. This pathway is over-expressed in advanced metastatic melanoma cells.

CREB/ATF1 regulating MMP2 and MUC-18 and it confers resistance of melanoma cells to chemotherapy and apoptosis.

If we can interfere and down regulate the over-expression of this transcription factor, this is another way of targeting.

The other pathway is the EPI2 transcription factor which is being lost in metastatic melanoma cells and it is the major cause of why c-kit is lost in metastatic melanoma cells.

If we can find a way to re-express -- because the gene is there -- re-express the EPI2 in advance of metastatic melanoma cells, we can maybe change and reverse the phenotype.

The other pathway that we should consider is the angiogenic factor. I know it is not specific, but melanoma cells express tons of IL-8, for example. It is probably the result of NF kappa B. If we can target this or if we can neutralize IL-8 or neutralize the receptor for IL-8 on melanoma cells by small molecules, because IL-8 works as an autocrine to affect also vascular endothelial cells, if we can interfere with angiogenesis by interfering with the function of IL-8, this is another target that we should consider.

DR. HALUSKA: I have to say, those are all potential targets, but they kind of fall into the category of ski and NF kappa B, in that they don't meet all the validation criteria. Most of them are not mutated and lost and they are relatively less broadly studied in the field. That is not to undercut the potential validity, but they certainly need more study.

PARTICIPANT: I would just like to add one more thing which hasn't been discussed here.

DR. HALUSKA: Can I say that we probably shouldn't just list everyone's pet genes. What the committee did was try to come up with criteria for validation. What the group thought was that these are the ones that are at the top.

There are other genes that are over or underexpressed in melanoma, but I don't think it right now serves us to everyone just list them.

PARTICIPANT: There is quite a bit of literature on some more genes that are not listed here.

DR. HALUSKA: I agree with that. There is a lot of literature out there.

PARTICIPANT: For example, Renato Baserga has considerable data on IGF-1 receptor as a target. Dorothy Becker has quite some data on FGF receptor as a target.

There are studies coming out now with STAT3 as a target in melanoma, because it is constituitively active. These are just a few, and this study is from the University of Southern Florida. There are several others on the adhesion and stroma field, which might be too much right now.

FRANK: I just wanted to comment on the melanin synthesis a bit. You somewhat limited it to immunological considerations.
I would like you to consider the fact that melanin and melanosomes are unique to melanocytic cells and, therefore, offer a target vis-à-vis very little toxicity perhaps to the rest of the host.

I will just refer you to an article by Farmer in Pigment Cell Research this month, in which we summarize our five years worth of data of developing a strategy to melanin because of the appearance of a functional quinone amine entity during the transformation process.

It is really a totally different way of looking at this and I wouldn't limit it to melanin or melanosomal issues to immunology.

DR. HALUSKA: The committee considered that and the major issue, the major criterion that melanin doesn't meet is its necessity for the malignant phenotype. There are many amelanotic melanomas. They do just fine.

PARTICIPANT: Amelanotic melanoma has lots of eumelanin in it.

DR. HALUSKA: That was the committee's deliberation.

PARTICIPANT: It is a very common misnomer that amelanotic melanoma does not have melanin. It has lots of melanin.

PARTICIPANT: I don't have any pet genes. I think the big issue, though, is strategy. I think we all know, whether it is BRAF or NF kappa B, which is my pet gene or at least Anne's, but you know, we need a strategy in which to develop these in melanoma.

I don't want to bring up the F word, but we need, whether it is a consortium, whether it is some different approach, so that we can get the trials done, so we can interpret the results, so we can figure out why the drugs fail, and we know many of them will fail.

Then we can move forward.

Drug development, I think there has been a lot of cytotoxic drug development in melanoma. It has obviously been a failure.

My own interpretation is that that is not so much immunology. That is a false sense that we have made progress with biochemotherapy or chemotherapy. I think it is chemotherapy itself that has inhibited the development of other chemotherapy, more than anything.

I think we have got to develop the field in a way that we sort of more quickly address those questions, do the studies in a way that is not six of 13, or three of 12, so that we really have an answer.

I think the BCL2, Ray Worrell is not here, which is not a good sign. We all know that. I think that study is an important -- obviously, if Genasense does prove effective, then they have made the right decision.

If you look at what they based that decision to move ahead on, it was an incredibly small population. The inhibition of BCL2 in that population certainly didn't look very impressive.

There was even no correlation between response and inhibition of BCL2, and they did a 450-patient study. Now, they could win. The study design has turned out to be right on, I think, but if they win, it is not based on good strategy of developing a drug. It is luck.

DR. HALUSKA: I would rather be lucky than good in that circumstance. You can criticize them skipping a large phase II that would have taken the same amount of time, but they will know the answer when they are done. They won't be five years from knowing the answer.

Actually, whether the results are positive or not, they did the correct trial design and got a good answer, but I agree with the thrust of your remarks. That is what the committee felt, too, that we need a good strategy to do this. Other thoughts? We only have a couple more minutes.

DR. SAXMAN: Not so much a question, I guess, as a comment. I think that the discussion that you are having really highlights one of this group's recommendations, and that is that there needs to be a closer collaboration between the biologists and the clinicians here.

I mean, the clinicians are willing to test whatever you all think are the important things to test. I think that is obvious here.

It is much more difficult -- I showed John's slide the first day from the review paper about all the genetic alterations in melanoma, and there are 35 things there.

When everyone lists these and more, throwing a dart at one of them is really the wrong way to approach this. So, I think this kind of effort, continuing on, this collaborative cross talk, and what are the most important things to target, is really valuable, and that is something we can also assist with as well.

I would also say, Frank -- and I agree, by the way, with what you are saying, that just because you do a study with a drug that doesn't work against a target does not invalidate the target.

That doesn't mean that other agents against the same target won't work. The reality is that a negative trial has very serious implications. The idea that, well, if this one is negative, then we can just do another one with a different agent, if that one is negative, we can do another one with a different agent, probably is not actually true.

The reason I say that is, I think that makes it more imperative, as Jeff was saying, that we are very thoughtful, particularly before we move into large scale clinical trials, that we are testing the right agent in the right circumstances in the right group of patients.

Particularly in melanoma, we don't have the luxury, fortunately, of having 200,000 patients per year that get this disease.

I think we need to be very, very considerate and cautious about moving forward with a particular drug, because I think in many circumstances we are really going to get one chance, not just because the pharmaceutical companies are going to lose interest if one study turns out to be negative, but I think investigators will, too. That is just more of a comment, I guess.

DR. MORTON: It just occurred to me, a suggestion as to how we may get corporate partners more interested in developing drugs for melanoma.

I think that the answer may be in the mechanisms that already exist for NIH grants to for-profit companies. I guess it is small for-profit companies in terms of technology transfer, but it seems to me that it would be possible to set some of that money aside and say, we are going to spend $100 million this year in collaborations with corporate partners to test new drugs for melanoma.

If, in fact, it was a partnership where the NCI, because this is in the public good to develop new drugs for melanoma, will partially underwrite this, through this mechanism that already exists, I think that may be a practical way to get corporate partners more attuned.

For them, everything is the in total rate of return, the bottom line. For one major pharmaceutical company, my vaccine got all the way up to the board of directors, and they chose to go ahead with a Viagra look alike rather than my vaccine, because the internal rate of return was five percent higher.

PARTICIPANT: That was against melanoma?

PARTICIPANT: The question has not come up, how, if the pharmaceutical company is hesitating about going into melanoma as a disease in a major effort, except very few drugs, one may consider to develop academic centers that go into drug discovery.

Actually, the NCI is supporting drug discovery laboratories, sort of core laboratories throughout the country. I think there are four or five which could play at least a partner role.

I think this subject of how now to develop strategies should come up in a discussion in more detail of how to go about it and how to combine that with the current knowledge in biology.

DR. HALUSKA: All right, let's draw this to a conclusion and turn it over to Lynn Schuchter with the second group.

TOP