Summary




SLIDES & TRANSCRIPTS
Thursday, June 15, 2000

Breakout Session C Summary:  Agents that Target Cellular Pathways and New Chemotherapeutic Agents:  Therapeutic Directions
Frances A. Shepherd, MD
Slide 1:

DR. SHEPHERD:  I drew the short straw, so I'm here presenting the results of our session today.  Any molecular questions that you have, you have to direct them to Dr. Schmidt-Ullrich, because I am not a molecular scientist.


TOP
Slide 2:

We were asked to come up with strategies for the integration of new therapeutic agents into the multimodality treatment of locally advanced non-small cell lung cancer.  I really thought that we should be coming away with some very carefully designed clinical and molecular studies in locally advanced lung cancer, but we really didn't do that.

So to my way of thinking we maybe failed in our primary goal, and I'll go through a few of the reasons why we maybe failed.  As I have been sitting here this morning, I really don't even need to be here at the podium, because Paul Bunn virtually said everything that I'm going to say.  I think that the angiogenesis discussions were almost identical to our own.

TOP
Slide 3:

Why did we fail?  First of all, I think that the majority of the agents that we were given to address really aren't ready yet for incorporation into multimodality trials.  So we spent most of our time addressing some of the steps as to how we might eventually get to the important multimodality trials.  It was brought up in our session that many of the mechanisms of action for these new agents had been explored in tissue culture, but there was a feeling in our group that we needed more studies, really probably in murine models, but other preclinical models to assess the mode of action and the efficacy of these new agents. We felt that with most of these agents, the Phase I and II toxicity studies that need to be done with radiation, with chemotherapy, or with radiation and chemotherapy have not yet been done to be able to move these forward yet into Phase III trials.

TOP
Slide 4:

Another reason we failed was that we had a virtual smorgasbord of agents in front of us to discuss.  You saw the list, and many of them could have fallen into our group, some of them into the angiogenesis group.  There were as many as 60.  And I guess that's why we are here.  This is the first time in lung cancer treatment that we have really been overwhelmed with the number of new agents that we have to assess.   I imagine that's why we're here, to try and prioritize and decide how we are going to go forward with these agents.  I think if every patient with lung cancer went into clinical trials, we would have enough patients.  But we know in the real world that doesn't happen, and so we are going to have to set some priorities.

The presenters that were invited to make presentations at our session did not cover all of the new agents in our category.  So we spent a lot of time discussing farnesyltransferase inhibitors.  I think some of the problems and ways to go forward that we identified with the FTI’s  are applicable to the other agents as well.  So what I'm going to do is present some of the principles that we came up within our session.

TOP
Slide 5:

What about Phase I trials?  We really feel that we need to continue the Phase I toxicity trials, which would be customary for any agent.  We need to do the dose escalation studies.  However, the classical dose escalation to maximum tolerated dose may not be at all appropriate in this setting, and really what we have to also develop are important biological effective doses.

For instance, we may not need to go to toxicity to completely saturate an enzyme if you are looking at blocking an enzyme pathway.  What we need to know from the preclinical studies are what are the micromolar doses that are needed to block an enzyme, or to do whatever it is a new agent needs to do.  We then need to target our studies with that in mind, and we may not need to push the envelope all the way to classical toxicity.  This can sometimes apply even in chemotherapeutic agents.  We know that to be activated gemcitabine has to go through an enzymatic process.  That enzyme is saturated at relatively modest doses of gemcitabine, and yet gemcitabine Phase I studies kept going up, giving grams and grams and grams of the agent, and that probably only makes expensive urine.

So I think we have to bear in mind the biology of these agents that we are studying, as well as just looking at toxicity in our Phase I studies.  In addition, we think it's very important that the Phase I studies also have the correlative biologic and molecular studies along with them.

TOP
Slide 6:

Many of these agents actually may not be active on their own.  We may not see cytotoxicity.  We may only be seeing cytostasis.  So I think it's very important that early on in the development we do Phase I studies in combination with chemotherapy, with radiation, or both, because if we want to integrate them into multimodality trials, we don't want to be spending the rest of our lives doing sequential Phase I studies.  We want to be moving these things forward in parallel.  I think that is some of the problem that we now face.  We don't have all of the preliminary toxicity trials to be able to, with the expediency that we all want, to be able to move them into the more important trials.

We also really do not have a good idea about the scheduling of these agents.  Do they need to be given concurrently with chemotherapy?  Should we be bringing them in after chemotherapy?  Do they need to be given before chemotherapy?  Or does the chemotherapy have to be given before the agents?  Some of these new ones are totally inactive in one schedule, but very active in other schedules.  So that has to be looked at early on in the development, maybe not in Phase I, but maybe even more in the preclinical murine models.

TOP
Slide 7:

What about Phase II?  You've already heard a great deal of discussion about the pros and cons of Phase II.  I must say, when we were evaluating the metalloproteinase inhibitors, I was really very supportive of almost leap frogging over the Phase II studies and going to Phase III.  I think with some of these agents though, we have to be somewhat more cautious.  I think I'm now convinced that we absolutely can't abandon our Phase II studies.  As I think I said earlier, they are going to provide us with some more long-term toxicity data, and we have seen some unexpected toxicities particularly arising out of some of the VEGF agent inhibitors.

I think we need to do Phase II, maybe because our patient resources are limited, and because I’m getting older and I want to see some of these trials finish before I retire.  I would like to have some strong consideration for this concept of randomized Phase II, rolling into Phase III to move these things along a little bit faster, and to make the maximum use of the small patient resource that we have.

For the combination studies that we do, how have we decided which ones to do?  We have generally looked at a tumor type and said gemcitabine is active in pancreas cancer.  gemcitabine/platinum in lung, or taxol/carboplatin in lung.  So we want to add to those, and I think that's appropriate.  We don't all of a sudden want to go back to evaluating 5-FU in lung cancer,  to add a VEGF inhibitor to that.  So I think our primary studies should be in combination with the agents that we know are activity in our tumor types.

TOP
Slide 8:

What about patient selection for the trials?  This hasn't been addressed yet this morning, but I think we have also learned a little bit about patient selection.  We have learned from the breast cancer trials that if you are want to do trials of herceptin, you want to have HER-2 over-expressers.  But we know from the randomized breast cancer trials that just expressed HER-2 may not be enough, and that the patients who really benefited from those studies were the ones that were 3+ on immunohistochemistry.

We are now doing the same thing in lung cancer, and we don't have all that many patients who are 3+ on immunohistochemistry for the expression of HER-2.  So clearly patient selection will be important.  However, is it always going to be important?  The farnesyltransferase inhibitors were developed because ras oncogene, to be activated, has to be farnesylated.  So the initial thought was that we must choose tumor types that have ras gene mutations.

The first studies of the FTI’s were done in pancreas cancer that has 80-90 percent mutation rates.  But we are now realizing as we are studying these more in depth, looking at the downstream markers, et cetera, that it may not be necessary to have ras oncogene abnormalities, because ras itself pays such a pivotal role.  That if we inhibit ras, we may be inhibiting the downstream markers.

It is very important for us, when we are doing our correlative studies, not to just focus on that one molecule.  If we had limited ourselves in the FTI’s to only ras abnormalities, we would be left with 30 percent of adenocarcinomas, and the vast majority of our lung cancer wouldn't be eligible.  Whereas it may be equally important in other non-small cell lung cancers. 

TOP
Slide 9:

What about surrogate markers?  And we spent a great deal of time discussing this in our session.  Just as you heard from Dr. Bunn, it's really very difficult with lung cancer patients to do multiple biopsies.  The biopsies on the whole are invasive.  We don't usually have skin nodules.  We sometimes do, but not usually, and the patients don't always come with nice neck nodes that we can access.  So to do deep biopsies in the lung or in the liver is not always that easy.

One of the investigators in our group had proposed a study looking at pleural fluid, which is fairly readily accessible.  He said that his study was turned down, and I find that a little surprising.  I think we should be rethinking looking at pleural fluid.

I  had even thought of doing a p53 gene therapy study a few years ago looking at pleural fluid and our ability to have gene expression in the cells—pleural fluid is easy to get at.  And I believe that we should be able to sort out the difference between mesothelial cells and pleural cells and cancer cells.  Pleural fluids still might remain a proper source to evaluate for proof of principle and basic science correlative studies.

Having said that, we thought that perhaps there may be other tumor types from which it is easier to get tissue, and those are basically the circulating tumors -- leukemia or myeloma.  In lung cancer we may have to make do with larger scale studies of proof of principle of basic science correlates being done in those tumors and hoping that that would apply to lung cancer.

Now we realize that is far from ideal.  Because of the difficulties, we may not be able to have large patient populations to do these correlative studies.  However, there may be other surrogate markers, and buccal mucosal scrapings were suggested as one possible alternative, easy to get.  Peripheral blood mononuclear cells may in some instances also be surrogates for actually looking at the tumor tissue.  Clearly, they are not ideal, but other possibilities for surrogate testing were raised in our session.

TOP
Slide 10:

What about Phase III trials with these new agents?  I think we are getting close to being able to do them, with some of the new agents.  I think we all felt strongly that for these agents to be accepted as primary therapy, they must have an important endpoint.  Just because you can p53 into a tumor and have it expressed, may not be important if that doesn't result in increased response rate, longer time to progression, and longer survival. So we emphasize that the important large Phase III randomized studies must be done.  We can't just be looking at surrogate endpoints only.  We figure that this is going to be costly therapy, and there must be important survival benefit at the end of it.  Then the motherhood statement that, whenever possible, these larger randomized studies should be supported by correlative research, maybe in a subset, if not all patients.

I think that's it.  Remember, the molecular questions to Dr. Schmidt-Ullrich, the clinical questions to me. 

TOP
Slide 11:

DR. SAXMAN: You said at the beginning that there was a large list of agents, but was the group able to prioritize any of these classes of agents, or individual agents at all?  Or did they feel that they were sort of all at the same level, and you weren't even able to begin to prioritize?

DR. SHEPHERD:  We didn't really attempt to prioritize them.

DR. SCHMIDT-ULLRICH:  If I may comment on that?  I think it is a problem.  I think in a way it depends on how you structure a session like this, if you have presentations that focus on certain molecules, that's what kind of happened.  We got stuck in a discussion that brought up some very important points, but kind of distracted from structured discussion of the individual agents.

I think what came out of this is interesting in relationship to the previous presentation. If you look at signaling molecules that are responsive to upstream effects, and have downstream effects and downstream effectors as well, there is the likelihood of getting into this difficulty of developing some even at the level of the molecule and our investigation, a correlation between inhibition of that molecule.  The cellular effect is clearly greater than going to molecules that may be at the top of cascade, where you only have a fairly defined signal input, and then look at downstream effects.

I think it was interesting that the previous group identified mostly membrane receptors.  Those are clearly targets where you have clear signal input, and you don't have a lot of upstream noise, which have both with ras, if you look at farnesyltransferase inhibitors, and even worse with any of these multiple PKC molecules and variants.

So in a way, the course of the discussions were kind of directed by these presentations, and the difficulties that were heavily discussed within the groups.  So at that point we felt that we had to take a direction, and this is the direction we took, rather than going through a catalogue of the different reagents.  But it is clear from the previous discussion that we will certainly heavily support that tyrosine kinase receptors, in addition to some of the immediate downstream effectors like ras and PKC are very worthwhile molecular targets to examine.

DR. STEVENS:  What kind of intermediate markers, if any, would you use for a drug like tirapazamine or other hypoxic cell toxins?

DR. SHEPHERD:  David, you are about to study tirapazamine.  What immediate markers would you like to see studied?

DR. GANDARA:  Paul Gumerlock, what intermediate markers would you see?  We have been debating this.  I showed a slide yesterday of some of the proposed correlative studies.  One was functional imaging, but after Ned Patz this morning, maybe we are not ready for prime time.  We did have as part of our consortium, a program project grant in functional imaging with Peter Kant at USC, so we intend to incorporate functional imaging with at least two different imaging agents.

For the molecular studies we are working with the group at Stanford, Martin Brown who developed tirapazamine.  They have looked at a whole series of hypoxia-related genes and other potential markers, both in tissues and in animal models.  At least one which could be utilized in patients and also done serially is PAF-1, plasminogen activating factor-1, which they measured in serum specimens from their patients on a Phase I trial with head and neck cancer, and correlated with the effects of tirapazamine in response.  So we are still struggling with this.  We would be happy to take suggestions.  But the group at Stanford I think has provided us with at least some leads.

DR. SHEPHERD:  Paul, do you have anything else to say?


DR. GUMERLOCK:  I don't think so.  The PAF is really the one that looks most attractive at this point, and we are able to measure it when patients are in CR.  Using a proteonomics approach they also looked for changes in proteins in serum.  That is what fell out initially.  So I think that's the best first candidate.

DR. SHEPHERD:  Have you any thoughts yourself?  Sometimes when people ask questions, they have some thoughts behind the question.  So do you want to elaborate?

DR. STEVENS:  I actually have a couple of things that we are looking at.  We have some technetium-labeled misonidazole as one potential agent.  But there are several other hypoxic cell markers that are nuclear medicine studies.  Amersham has a technetium-labeled compound and others.  I was wondering what you thought of the current status of those?

DR. SHEPHERD:  It's far from an area of my expertise.  Anyone working with the hypoxic agents that wants to expand on this?

DR. COLEMAN:  I can just make a general comment that there are a lot different hypoxia markers, and we showed one EPR marker yesterday that are now being evaluated.  The real problem, is there a gold standard for hypoxia?  The oxygen electrode has been used, and whether that is perfect or not, it is not clear.

For tirapazamine, as you know, you can use the common assay to do small tumor biopsies, and measure the extent of double strand and other DNA strand break damage you are creating, to at least see if you doing more than you would with radiation alone.  I don't know whether you can do that on circulating DNA.  I guess someone mentioned that yesterday. I had never even thought about that.  But whether you can look at DNA profiles from the serum after you give hypoxic marker and see if there is a difference, that may be something curious to do.

DR. CURRAN:  Getting back to the issue of the Phase II design, and those Phase II issues.  You may have come down just as strongly as Paul did on the idea that somewhere in the Phase II evaluation there has to be an efficacy demonstrated sufficiently strong for everyone to put such resources as are required in the Phase III study.

Clearly, we need toxicity data.  We would like to have some intermediate marker of efficacy of these agents.  But the real key is whether your breakout session felt strongly that before Scott and his group approve a Phase III study, do we need not just disease-specific, but even stage-specific, and multimodality-specific data on incorporating these agents into the management of locally advanced disease, whether we are talking about induction therapy before surgery, or biological agents with chemoradiation?

DR. SHEPHERD:  We didn't get that specific in our group.  I don't think our group came to any final recommendation about Phase II.  We recognized the importance of Phase II mainly for toxicity, and for providing longer-term toxicity than comes from many of the Phase I studies, because of the patient populations inherent in Phase I.  I think we have all become sensitized to this because some of the more recent toxicities that we are seeing in some of the Phase II trials that have been done for studies that are about to go to Phase III.  I still have a great deal of difficulty about the efficacy in Phase II.  And I think that maybe what you want to do is make sure that it is not worse than what you would expect with the standard treatment alone.

But I don't know how you really can come down in Phase II studies, which are often single center, and we know how they never really correlate well with the Phase III studies.  I would argue against trying to trap ourselves into defining response or survival levels from a Phase II study that would give us the green light to ahead to Phase III.

I think we are going to box ourselves into a corner if we do that.   I think we are going to hamstring our ability to move forward to Phase III.  We have not put the chemotherapeutic agents through this kind of rigorous high hurdles that they have to get over.  There wasn't a great deal to suggest that taxol, etoposide, and platinum would be much better than etoposide and platinum in small cell lung cancer, and yet it went forward to a Phase III study.

I would like to issue a word of caution to all of you that we don't set hurdles so high that no one is going to be able to leap over them.  That we don't just put back the research by years, and that we consider more strongly the randomized Phase II rolling into Phase III, so that we don't waste all of that time.  We might learn to define the endpoints at the end of the Phase II portion of a Phase III study that are going to be important for us to look at, looking for differences in toxicity, making sure that there is no trend to inferiority, and then allow the study to go on.  I just don't want to hold up the research process forever.

DR. MABRY:  This is actually  related to that issue, because I was in Group A, and the discussion about proof of principle, as you call it, in other tumors was addressed.  I walked away from our session saying that that would be inadequate.  Would someone like to comment on that from the session, or either session?

DR. SHEPHERD:  I put forward my own thoughts, and the thoughts of our group, that at the end of the day, in addition to a principle, we have to prove superiority of the important endpoints in cancer therapy, which are survival basically, or cure rate in early stage disease.  So just to be able to manipulate a gene or to reduce vascularity, if that doesn't then translate to an important endpoint, I don't see that as a useful step forward in cancer therapy overall.  I'm speaking a little bit for myself here.

DR. GANDARA:  Frances, I agree completely with your statements regarding the limitations of Phase II trials, and also the problem with really setting criteria where for the first time we have a host of agents that really are independent of the targets of chemotherapy and radiation.  We need to employ them in the best way possible to see if we are going to improve survival, but we don't want to hold them to really unattainable standards.

I'm not sure that our statisticians are going to buy into a randomized Phase II rolling into Phase III.   I would suggest that we have a Phase III design where we do an interim analysis for toxicity at an early point, if that's the issue, because those really would be almost the same, giving you an early look to make sure there isn't something unexpected with your regimen, again, based on some Phase II data, but really not having to say that we want four years of follow-up in a Phase II trial in case somebody's shoes fall off at year four.

DR. SHEPHERD: In fact, there is a Phase III trial being designed in limited small cell lung cancer, looking at adding erythropoietin.  That study is being designed with an interim analysis that has a slightly different bent to it, because the interim analysis is not going to look at “whether this is statistically better, and so we should close the study.”  Or “is it statistically worse, and from a safety point of view we should close the study.”  It's going to also look at whether it's about the same, and therefore is there any need to keep this study open?  If we add another 300 patients to it, we are still going to end up with some kind of marginal difference that isn't going to be important to us, and so we are not going to continue the study.

So I think there are various ways of doing our interim analyses, and I think that we should charge the statisticians with maybe changing their thinking.  Let's hear from the statisticians, now that I've fired them up.

DR. KIM:  I think we are dealing with a very complex issue here, obviously.  From a methodologically perspective, we are struggling with how to monitor large scale studies like Phase III studies using an efficacy endpoint.  When you start to mix different aspects of the clinical manifestation of the treatment, it really becomes a very complex and difficult problem.  That is one of the main reasons why a lot of the statisticians would rather stick with the separation of Phase II studies, looking at specific early treatment efficacy question or even toxicity.  Then, once that question has been adequately addressed, we would rather move into the Phase III to focus on a specific question of efficacy.

That is primarily the reason.  We don't really have a good methodology to address all of these questions simultaneously.  That's just the nature of the complexity of the problem.  Personally, I can understand the clinician's frustration in terms of how long it takes to get the study developed through the IRB’s and the review process and things like that.  But I often wonder whether by rushing into these Phase II/Phase III studies do you really gain much?  I have some fundamental questions about that.

DR. SHEPHERD:   I don't know why the statisticians should be let off the hook so easily.  These clinicians are always having to balance this.  We have to balance the quality of life, the symptom control, the toxicity, the response, and the survival.  That's looking at cancer, and you cannot decide on any treatment based on one of those things in isolation.  So why should you guys have it so easy, when we have it so tough?

The next statistician has a chance to speak now.

DR. KORN:  I think you have perhaps misrepresented a little bit how trials are now done with standard chemotherapy in that if there are one-sided questions, we practically always build in a futility analysis in the interim monitoring.  So partway through the trial if it looks like the new agent is never going to work better than say the placebo, we call it quits.  We don't ask for it be significantly worse before we stop.  That would obviously be unethical.

I think the data monitoring committees and the people writing trials are now aware of that, and write that into the trial.  So you could write that in even earlier if you were interesting in a looking, stopping earlier, looking at futility.  But it obviously can't be too early, because you won't have enough information.

I think the problem that many of us have with the Phase II trials is not with the Phase II trials per se, but they have been designed poorly.  We have seen a lot here today, randomized Phase II trials with 30 patients in each arm looking at a survival endpoint.  That's ridiculous.  You really can't compare things with 30 patients in each arm, looking at a survival endpoint.  Nor can you compare in a one-armed 30 patient trial survival versus historical controls, and get really any information.

That's not to say you couldn't do a 70 patient Phase II trial at certain institutions where they have a good historical database, and get a fairly good feel of whether things look promising or not.  I don't think anyone is demanding that you want to see things be statistically significant to 0.05 level at a Phase II trial before going onto a Phase III trial.  But it wouldn't be unreasonable to perform a Phase II trial with 70 patients, compared to a good historical control database and say, does this look promising or not?  Especially when we have a lot of agents to choose from, why not pick one that has some promise?

DR. SHEPHERD:  It takes time.

DR. ABRAMS:  It does take time, but that should be, one might hope, the advantage of cooperative groups, which have focused a lot on Phase III trials, but could also focus on doing these multi-institutional Phase II whether randomized or not.  It would be one of the advantages that the cooperative groups could bring to this effort.

Given the plethora of drugs that are available, I think Ed's suggestion is a good one.  Whether you do them in randomized Phase II, or if that's difficult because of the pharmaceutical issues, just larger Phase II’s being compared against some historical database that all agree to use, might be a very productive way of finding out which agents to bring forward to Phase III.

DR. SHEPHERD:  We also have the difficulty, and no one has really addressed this, of now having multiple agents that are somewhat similar, EGFR receptor antagonists, the VEGF antagonists, et cetera, that are similar, but coming forward from multiple pharmaceutical firms.  And each one of those is expected to go through a randomized trial.

Would there be any role for the NCI to be bringing all of the pharmaceutical firms to the table, and saying, all right, we have three MMPI’s.  We want to do a big four-arm of control and MMPI A, B, C, and D.  That's where you could have the control to bring these people together, and that would be a much, much more efficient use of our patient resources, and we would get comparative data at the same time.  There are going to be some pharmaceutical winners and losers out of that, and I don't know how the pharmaceutical firms would feel about it.  But it would certainly make things easier at the end of the day for the clinicians.  We would really know where to go.

DR. CURRAN:  Following-up on what Jeff Abrams said for a disease different than what we are talking about today.  In the RTOG we were somewhat frustrated by three decades of negatives trials for malignant glioblastoma Phase III trials, and decided we would only open a Phase III trial if we had the exact type of Phase II data Jeff referred to.  We had a large database we could classify into six categories, so that if we had an uneven distribution of "better" patients than other studies, we could account for that.  I'm a little concerned if we don't think about that in a disease like locally advanced non-small cell lung cancer, which is far more common than glioblastoma and for which we should have more patients to do that.

Our enthusiasm for new mechanisms of actions could result in lots of negative trials over the next decade.  The standard against which we are comparing multimodality therapy is not like it was in 1980, where median survival time was 9-11 months.  We are up in a year and half kind of range.  And to improve upon that, we may need to be more disciplined than we used to be.

DR. GANDARA:  I agree.

DR. JETT:   Frances, we already have trouble with the NCI not having access to a lot of drugs, that companies will take them and run their own studies, run their own Phase II’s, run their own Phase III’s, and bid out to their own institutions.

Some of the discussion here today reminds me of Church Lady on "Saturday Night Live," a holier than thou attitude.  I think the higher we raise the bar, the more difficult it is to get to where we do the Phase III trial through CTEP approval, the less likely that we are going to have access to the drugs.  We already have some problems with that right now.

So I would caution you -- and I know CTEP is meeting later today B if we raise the bar so high, I think all the companies are going to opt out, and we're all going to be negotiating single institutions with the companies to be one of the players.

DR. SHEPHERD:  I said myself that I think we are raising the bar too high.  So I think you and I are saying the same thing.

DR. SAXMAN:  Any other comments or questions? 

I want to again thank Rick Cumberlin, who helped put this meeting together, the speakers, the moderators, all of you as participants.

I forget yesterday, many people commented about this molecular target sheet that was in your folder.  I can't take any credit for that whatsoever.  Janet Dancey and her colleagues in the Investigational Drug Branch put that together, and I forgot to mention that yesterday.

So I hope you all have a safe trip, and thank you very much for coming.

TOP