SLIDES & TRANSCRIPTS
Tuesday, September 14, 2000

Targets for Immunotherapy of Lung Cancer
Paul Chapman, MD

DR. SAXMAN: We structured the breakout sessions this afternoon into three groups, grouping three molecular targets or groups of molecular targets—the molecular genetics or the genetics of the cancer cell, the signal transduction or the receptors that are involved with the malignant transformation or the malignant process, and factors that are external to the cell itself, angiogenesis types of factors, immunotherapeutic types of targets.

To introduce those topics and to sort of get everyone thinking along that line in anticipation of the breakout sessions this afternoon, I have asked four individuals to speak about each of those issues and put those molecular targets into a clinical scenario or discuss the clinical implications.

The first speaker this morning is Dr. Paul Chapman. Dr. Chapman is an associate attending physician at the Clinical Immunology Service at Memorial Sloan-Kettering Cancer Center, and he is going to speak this morning on targets for immunotherapy of lung cancer.

DR. CHAPMAN: I am flattered to have been asked to speak here. In the interests of honesty I must admit that I come here through the melanoma field and that my interest in small cell lung cancer has arisen almost by accident.

So what I will do today, very briefly, is to give an overview of what I know about some antigenic targets in small cell lung cancer, and hopefully that will stimulate some other interest.

The whole idea of immunotherapy really has arisen from the field of transplantation biology, and this really has started in the thirties and forties when the inbred strain of mouse had become available. I note actually that this month Linda Gross died, one of the pioneers of this field. It is interesting how this field has matured.

The idea here was that, once we had inbred strains of mice, it became possible to develop methylcholanthrine-induced sarcomas in this case, and to develop a whole series of cell lines, many of them developed at the NCI. It became obvious that you could implant these cells into the mice, in geneic mice, and the tumors would grow. So, you could excise this tumor from the mouse. If you re-challenged the mouse with the same tumor line, the mouse could reject the cells and no tumors would grow.

If, however, you took that same mouse and challenged it with a different sarcoma line, although also from the same strain of mouse, that tumor would grow, and this was really one of the first observations that suggested that the immune system first of all had the power to reject syngeneic tumor cells and had a specificity that was quite remarkable where we could have the mouse recognize an identical tumor type. There were certainly epigenic differences. It has been the challenge of the tumor immunology field since then to try to identify what it is that the animal can see and to try to identify these so-called rejection antigens.

TOP

In a broad sense, there are three types of antigens on tumors, the unique antigens which tend to be mutations. Ras is one example. Unfortunately a lot of the unique antigens which have been identified, at least in melanoma, seem to be individual mutations and are not antigens that are probably broadly exploitable.

There are four antigens that tend to be either alloantigens or viral antigens. Some of these are beginning to be exploited now in clinical trials, for example, in cervical cancer, looking at E7 antigens. Most of the antigens that have been identified in melanoma are differentiation antigens, that is to say molecules that are normal, non-mutated molecules found on tumor cells but that are also found on certain differentiated normal cells.

TOP

So what becomes the challenge is to distinguish inducing tumor immunity from autoimmunity, and a lot of us do think of these two as being on the same immunological spectrum because if we are inducing tumor immunity we are inducing immunity against a self-antigen. So with respect to small cell lung cancer, I chose to really focus on carbohydrate antigens, and that is not because protein antigens aren't interesting; they are. It is just that we know more about carbohydrate antigens, at least with respect to small cell lung cancer.

TOP

This is a cartoon that Phil Livingston put together for a different talk, but I like to use it. This is a diagram of the plasma membrane, and here we have a variety of carbohydrate antigens that we and others have been interested in trying to target.

Shown here are various gangliosides, GD3, GM2, GD2, fucosol GM1, other carbohydrate antigens such as Lewis Y blood group type antigens, global H and then a variety of glycoproteins, and this is a large mucin molecule which has fairly monotonous protein backbone on which are hung a variety of potentially immunogenic sugars.

This represents a traditional type of glycoprotein. In small cell lung cancer, there is an antigen called polysialic acid, which is really just a string of sialic acids which also seem to be potentially immunogenic but also relatively specific for tumor cells.

I am going to focus basically on two antigens. I am going to mention a little bit about fucosol GM1 since this is some new data, and then I am going to focus the rest of the talk on GD3, a ganglioside that we have done more of our work on.

TOP

First, I will talk about the GD3. In general, GD3 and GM1 are both gangliosides. A ganglioside is a glycolipid in which there is a ceramide, a very hydrophobic portion, which anchors this molecule into the plasma membrane. The antigenic moieties are sugars. The ganglioside has a variety of different sugars, and that determines what type of ganglioside it is.

It turns out that the immune system both in mice and in people can distinguish very subtle differences in the sugar structures.

TOP

So these are data which are taken from our recent fucosal GM1 trial in small cell lung cancer. This was carried out by Stefan Grant and Phil Livingston and a variety of other people at Sloan-Kettering. In this trial, patients with small cell lung cancer who had completed initial therapy of chemotherapy and radiotherapy were immunized with fucosal GM1, 30 micrograms with the adjuvant QS21. This is serological data where the top graph is the IgM response against fucosal GM1. The bottom graph is IgG response. The Y axis in each case is the reciprocal of the titer, and I don't think you can see it but up here are these little blue arrows that show when the patients were immunized. The point of this slide is to show you that of the 10 patients that are evaluable, all of them developed antibodies against fucosal GM1.

You cannot see that from this slide because this is a composite, but it shows that both IgM antibodies and IgG antibodies were induced, and it also gives you a sense of the titers, around 1 to 160, around 1 to 640. These are typical titers for responses against carbohydrates, as opposed to protein antigens which tend to generate much higher titers.

TOP

This shows that this fucosal GM1 molecule is quite immunogenic. These are results from complement-mediated lysis, and this shows the pre-vaccination is the open. Post-vaccination is the closed circle. The point of these data is to show that virtually everybody developed antibodies that could fix complement.

So this identifies fucosal GM1 as one potential antigenic target in small cell.

TOP

Let me go on to GD3. These are data actually from melanoma patients. These are 20 melanoma tumors in which the gangliosides were extracted and identified by immuno-thin layer—these are just thin layer chromatography. The point here is to show that GD3 is a very common ganglioside in melanoma. It formed one of our important targets in the melanoma field.

We were not interested in GM3, which you cannot read. But this top line is GM3, and that is primarily because it turns out that GM3 is on every cell in the body, and that seemed like an uninteresting target for us.

In melanoma we have, also, been interested in GM2 and GD2. That is a different story, although with respect to small cell those gangliosides are, also, on small cell lung cancer, but in melanoma we focused on GD3 partly because it was so abundantly expressed.

TOP

So we were interested to see whether GD3 is expressed in small cell lung cancer. At the time, which was several years ago, we were hampered by the same things that were just discussed, that is, the unavailability of tissue, and yet Stefan Grant was able to pull together eight specimens of either cell lines or tissue from Sloan-Kettering and got three more from ImPath and showed that there was GD3 expression on these cells. Then subsequently, Fuentes and colleagues showed that there was a variety of cell lines that also expressed GD3, and so, these data suggest to us as we expected that GD3 was expressed on small cell.

TOP

We were interested in GD3 because of this initial experience in melanoma where patients with metastatic melanoma were treated with monoclonal antibody against GD3, called R24. This is a variety of trials reported representing 103 patients from at least four different institutions, and the point of this slide is to show you that there are responses in metastatic melanoma with a monoclonal antibody against GD3.

The response rate is not very high, and yet it is really one of the highest rates in a solid tumor for a monoclonal antibody and one of the few where we have been able to show responses at more than one institution, although admittedly this is not a knock-your-socks-off result. But it showed us that antibodies against GD3 might have some biological effect in melanoma and, by implication, perhaps in small cell lung cancer, too.

TOP

So we thought, instead of infusing antibodies against GD3, perhaps if we immunized patients against GD3 that would be even better, but there were limitations in gangliosides as an antigen in vaccines in general. First of all, generally IgM antibodies are induced, although I showed you data already from the fucosal GM1 that if you do some tricks by conjugating the antigen to KLH which I didn't talk about or use the right adjuvant you can induce IgG antibodies.

Cellular immunity is generally not even discussed in non-protein antigens, and GD3, unfortunately, is not very immunogenic in melanoma patients, which is where we have the data.

TOP

These are three different clinical trials at Memorial carried out primarily by Phil Livingston in which patients were either GM2, GD3 or GD3 conjugated with KLH. These are the results in terms of the antibody responses. You can see GM2 is the most immunogenic ganglioside. Basically everybody develops antibodies against GM2, if you immunize them.

GD2 has been our next most immunogenic antigen, but GD3 is just very poorly immunogenic, and that has been other peoples' experiences as well. Despite this, we still wanted to immunize against GD3, and that is where the anti-id approach came that Paul Bunn introduced.

TOP

We had an antibody, R24, that was against GD3, and the idea was if we could develop an anti-idiotypic antibody against R24, an antibody that would bind specifically to the GD3 binding site of R24, that this molecule might mimic in some structural way the actual antigen. I have drawn it here showing that the antigen-binding site here of our molecule which we have called BEC2 mimics GD3,

TOP

and just to show you that there are some data supporting this, these are results which have been published a while ago now in rabbits, showing rabbits immunized either with BEC2 or with BEC3, which was another anti-id that I made, or a control antibody. The closed circles are the pre-immune sera. The open circles are the post-immune sera, and along the Y axis shows the anti-GD3 antibody response as measured by absorbence in the ELISA, and this is a serum dilution.

These data showed that immunization with BEC2 induced antibodies in these rabbits against GD3, whereas BEC3 did not seem to, and of course, the control, I suspect didn't either.

TOP

And if you looked at the antibodies induced in these mice—this is an immunodot blot in which various purified gangliosides were spotted onto a nitrocellulose strip and then incubated with the rabbit serum and then developed with just a peroxide second antibody.

You can see that the antibodies bound to GD3, and there was no cross reactivity with other gangliosides. So this preclinical data led us to develop a variety of clinical trials in melanoma initially,

TOP

and this is just a list to show you what some of the kind of parameters that we were looking at, perhaps speaking to what kind of clinical trial design we should be discussing.

The approach at Memorial has been to pilot our vaccines with the primary end point of immune response. The idea is that we did not assume that we knew the dose, the schedule, the formulation or the adjuvant that would best induce an immune response, and so in these trials we tested BEC2 at different routes of administration, SQ, IV, and intradermal, and we looked at different adjuvants, QS21, BCG which is not listed here. We had a trial with alum, and we also looked at a narrow range of different doses and the bottom line is in this last column. This shows the number of patients who developed anti-GD3 antibodies, and basically what we have learned so far is that the best adjuvant for us is BCG. Approximately 20 percent of the patients developed anti-GD3 antibodies, and these were all basically at the same dose. So the dose that we have used in these trials was really an extrapolation from the rabbit data which was an extrapolation from the mouse data which in other words is a complete guess. So what we are doing now in this trial, which has almost finished accrual, is looking at BEC2/BCG at a range of BEC2 doses over five logs.

Again, the assumption is not that more is better. There are data that less may be better in terms of vaccines. I don't have any data yet, but this is an ongoing trial.

So while we were doing these trials, one of the fellows in the lab, who subsequently became an attending, Stefan Grant, was interested in small cell lung cancer. With the data that I showed you that he generated, GD3 was expressed on small cell, he carried out a clinical trial, a pilot trial in small cell. I am sorry, I am getting ahead of myself one slide, just to show you what kind of antibodies we can generate in patients.

TOP

These are melanoma patients where the Y axis again is a titer of the anti-GD3 responses and the X axis is time.

The arrows are the times of immunization, again, showing that you can immunize patients, and the titers that we can generate are similar to what I showed you with the fucosal GM1, around 1 to 60, 1 to 640, typical for a ganglioside or carbohydrate antibody responses in general.

TOP

So Stefan formulated a pilot trial in small cell lung cancer using BEC2, and the hypothesis was that immunization against GD3 might ultimately improve survival in patients with small cell lung cancer following their induction chemotherapy.

TOP

The objectives were to evaluate the ability to induce an immunological response to BEC2/BCG in small cell patients. I was interested in this for two reasons. One, I had never immunized another patient population other than melanoma, and I wanted to make sure that it could be translated to other patients.

Also, we had never immunized a patient population that had been treated with chemotherapy and/or radiation therapy, and there was this mythology that has grown up that chemotherapy is immunosuppressive, for which there really is not very good data to support. In fact, there is data to suggest that in any way that we can really measure the immune response, most patients who have had chemotherapy really seem to be pretty much like naive patients.

So we wanted to see if we could immunize these patients. I have already given it away with the fucosal GM1 data I showed you already that those patients could all make fucosal GM1 antibodies, although we didn't know that at this time. We wanted to see toxicity, and we were going to just observe the outcome, but this was really just a pilot trial.

TOP

In order to be eligible for this pilot trial, you had to have had small cell lung cancer, of course. You had to have had your initial therapy, which was chemotherapy, and if you had limited disease at Memorial you also got radiation therapy. So you had to have had a major response to your initial therapy. You had to have at least a partial response. You couldn't have subsequently progressed. You had to still have your response, and you couldn't have had an antibody or already been treated with mouse antibodies.

TOP

These are the patients who were treated on this trial. There were seven with limited stage, eight with extensive stage, a mixture of men and women. Most of the patients had had a partial response to initial therapy although there were a couple of patients, three, who had had a complete response, and there were a number of patients who had negative prognostic markers as you can see.

Generally these represent for our group, and I am speaking just from what I have been told by Mark Griss and Stefan Grant, these are typical small cell lung cancer patients at Memorial.

TOP

These are the serologic results from these 15 patients. I have listed them in terms of overall survival, with the longest survivors at the top and the shortest survivors at the bottom, and this is the status as of last fall, and this indicates whether or not those patients had made anti-GD3 antibodies that we could detect after they had received their immunization. As you can see, a lot of these people did die of small cell lung cancer, although one patient died of a car accident. A couple of patients died of other smoking-related diseases. They were free of disease in terms of small cell but died of COPD or non-small cell.

What was interesting to me is that the longest survivors tended to have GD3 antibodies. Of the five, five out of 15 of the patients developed anti-GD3 antibodies which is about a 33 percent response rate which is very similar to what we have seen in melanoma, and so this answers the question are they immunosuppressed. Well, they are not any more immunosuppressed than melanoma patients who had never received chemotherapy, and four out of the five were among the long-term survivors. So that was interesting.

TOP

If you plot the data by Kaplan Myer, the white survivals are the limited disease patients. Remember there are only seven, and the red are the eight extensive disease patients.

Again, this was updated as of last year when the paper was submitted. So six of seven of the limited disease patients were alive as of last year.

TOP

So this was interesting but admittedly a very small number of patients. Actually our conclusions were that all patients, 100 percent of the patients, developed an immune response to BEC2, in the sense that they developed antibodies against BEC2 although only 33 of them developed antibodies that actually cross reacted to GD3.

Immunization with BEC2 BCG was feasible and safe in small cell. I haven't discussed the toxicities, but they were minimal. Small cell patients, some of them immunized with BEC3/BCG have had long survival, and we thought that a randomized Phase III trial was necessary, and this generated a lot of discussion among ourselves and among people in the lung cancer community. We thought that in terms of mustering the resources that were available that rather than go on to an intermediate-size trial that the question that we really wanted to answer required a randomized Phase III trial, and so that is what is going on now through the EORTC and, a variety of institutions participating in North America, Australia and New Zealand.

TOP

This is a trial for patients with limited disease small cell lung cancer. Patients are stratified by institution, performance status, the primary treatment and whether or not they had a CR or PR with primary therapy. They have initial therapy with chemotherapy, chemoradiotherapy and some countries always do cranial radiation. Some countries never do it. So that is optional, and then if they develop a partial or a complete response they are eligible to be randomized to either standard therapy which is observation after their initial therapy or to be immunized with BEC2/BCG with the exact same dose and schedule as was used in our pilot trial.

We are looking to randomize 560 patients, and that is where we stand here at small cell lung cancer, and I would be happy to take any questions or comments that anyone has.

(Applause.)

TOP

Question and Answer

DR. MABRY: How is the accrual?

DR. CHAPMAN: The accrual is slow. Partly it was because it took a while. It has been taking a while for centers to come on line, partly because of constraints imposed by the European investigators that patients had to be accrued at diagnosis prior to beginning any therapy. That restriction has changed now, and so we will be able to accrue patients if they are currently getting induction therapy or if they have just finished it.

DR. BRAUN: Have you had a chance to look at heterogeneity of the expression in different patients and do you have any inclination as to what the impediments are for developing antibodies in those that don't do—

DR. CHAPMAN: By heterogeneity do you mean from cell to cell?

DR. BRAUN: Exactly.

DR. CHAPMAN: I think that is very important. We don't have that information in small cell lung cancer because we are working as everyone has noted with FNAs or cell lines. We would love to get a piece of tumor. We have that data in melanoma, and certainly there can be quite a lot of heterogeneity from cell to cell within a melanoma tumor.

DR. DENNIS: Have you looked at the levels of the enzymes that actually synthesize gangliosides to see if that might provide better protein antigens than carbohydrates?

DR. CHAPMAN: That is a good question. No, we have not looked at that.

DR. SPIRIDONIDIS: What is the chemoradiation regimen used for the study?

DR. CHAPMAN: This is changing, initially the trial required one of three different chemotherapies. They are all platinum-containing regimens with etoposide or I think they were also going to use CAV. But I think they have changed this now to now allow almost any type of reasonable regimen because remember in order to be eligible for randomization they have to have a PR anyway.

DR. ELIAS: Have you looked at minimal residual tumor of either initial samples for heterogeneity and/or vaccination?

DR. CHAPMAN: No, we have not. It was hard enough for us to get tissue to begin with, but I guess you are saying that if there are any residual mediastinal densities?

DR. ELIAS: Bone marrow.

DR. CHAPMAN: No, we have not looked at that at all.

DR. MABRY: I have a question about your original presentation. Most of the patients, if I understand it correctly, who had titers to GD3 were patients with limited disease. Is that correct?

DR. CHAPMAN: Yes, most of them were.

DR. MABRY: So, is there any possibility that your first data set that your survival was based on the fact that they had limited disease and that GD3 titers were a surrogate for that?

DR. CHAPMAN: It is possible, although we have so few patients in that initial trial, I would hate to make a statement like that. I would love to think that that is true, but I am not willing to say that with such a low number of patients. We will find out.

DR. DMITROVSKY: Have you built into the larger trial the GD3?

DR. CHAPMAN: There was discussion on that, and it was felt to be logistically too difficult to do. So, they are not doing that.

DR. BUNN: But you are measuring antibody response?

DR. CHAPMAN: Yes, absolutely.

DR. BUNN: So you will note that?

DR. CHAPMAN: We will know if antibody response correlates. That is correct.

DR. BUNN: And what are you doing, I mean only one-third of the people actually developed anti-GD3 antibody. So you must be doing some other things to try to increase the percentage of patients who have an immune response.

DR. CHAPMAN: We are. We are doing two things, one looking at different doses of BEC2, as I mentioned and, also, an upcoming trial after that is going to look at immunization with anti-id BEC2, alternating with antigen itself, GD3 KLH. The other thing that we are doing is classic oncology. We are just looking for a different parameter.

It may be that antibodies are not the best surrogate marker, and we are looking at other things, too, which we can talk about in the breakout session.

DR. LANGER: Paul, what kind of survival advantage are you looking for?

DR. CHAPMAN: This is powered for an improvement in median survival, I believe of 20 or 30 percent.

DR. SPIRIDONIDIS: And what is the control expected median survival?

DR. CHAPMAN: Dr. Bunn, what would you say is the control median—

DR. BUNN: Eighteen months, 16 months.

TOP