SLIDES & TRANSCRIPTS
Tuesday, September 14, 2000

Clinical Issues in Small Cell Lung Cancer
Paul A. Bunn, Jr., MD

DR. BUNN: Thank you very much, Scott. It is an honor to be here. I have been doing this for 25 years. One could ask whether we have accomplished very much in those 25 years.

Probably Dr. Wittes, although he is very modest, with CAV (cytoxan, doxorubicin, vincristine), made some of the biggest advances 25 years ago. I guess I got interested in lung cancer when Oleg Salari told me it was very, very interesting 25 years ago, when the NCI, in its infinite wisdom, decided it needed a branch to deal with lung cancer. Scott, what you said about John Minna actually was true. Twenty-five years ago he said the same thing, and I guess he had Dan Ivy and I to try to see if we could get anywhere, and I am not sure that we did, but I am honored to be here.

I also think two things. One is, I think it is important that this is about small cell lung cancer. Somebody asked the question about industry. The good thing and the bad thing about industry, of course, is that they follow Sutton's law and go where the money is. There are more patients with non-small cell and the pharmaceutical industry clearly recognizes that, and there are a lot more trials going on in non-small cell lung cancer. So I think it is appropriate actually for the NCI to rejuvenate clinical trials in small cell lung cancer and, of course, I also was at the meeting in Aspen and I learned a tremendous amount at the meeting in Aspen, but as Scott and I talked on the phone, this meeting does have a different purpose. The purpose of this meeting is to take what we have learned from the biology and see if we can put it into clinical trials that will be more rapid for advancing the state of the knowledge.

DR. MINNA: At the Aspen meeting, Paul, pick one or two things that you thought were ready to go into the clinic?

DR. BUNN: Oh, you know, I think that this meeting has the wrong name. It says "molecular target" and this really should have been molecular and biologic target. So the first thing is in small cell we have known for years and years and years besides dividing a little faster than other types of lung cancer, it spreads earlier and metastasizes more widely. So you know, it is a little hard to imagine why anti-angiogenesis agents would get studied in non-small cell so much before small cell. I am not sure that any of the anti-angiogenesis agents really to a certain extent are more biologic targets than molecular targets. They are not being studied in non-small cell lung cancer. Could you do a study of chemotherapy with or without anti-VEGF tomorrow? Yes, you could, and that is why we are here, and the question is, "Is that more important than studying the addition of topotecan or taxol to etoposide and carboplatin?"

That is what we are here to discuss. You know radiation, we probably haven't optimized yet, but the question is, "How many patients should we put on randomized trials with two ways of giving radiotherapy versus some of these agents?" I think many of these agents are ready for the clinic.

I guess a huge question is, "When we do these clinical trials, how do we determine the optimal biologic dose?" We are not in MTD anymore, and a huge problem that you and I know about is when you treat a patient, you can get some tumor, usually, not always in the beginning, but once you start treating a patient it is essentially impossible to get tumor anymore.

So it is very difficult to do an intervention and get tumor before and after the intervention, and so the question is, "What markers are we going to use to determine the optimal biologic dose?"

TOP

Everybody who has heard me give a talk before knows that we cannot discuss all this stuff in the absence of talking about the problem, which is the tobacco industry. It is not the people who become addicted. It is our government and our tobacco industry that sponsors all these people dying from lung cancer.

TOP

So there are some biologists, and I will give the token clinician thing. Small cell lung cancer represents about 20 percent of all the lung cancers in the United States of America, and in my 25 years I have seen one patient who was not a smoker. The person had lupus. When I ask the house staff what are the biologic sort of principles, they usually cannot identify that these tumor cells have neuroendocrine features as one of their major distinguishing things from non-small cell tumors, although non-small cell tumors actually often have neuroendocrine features as well.

They cause endocrine paraneoplastic syndromes, and they cause neurologic paraneoplastic syndromes which are very interesting. They metastasize very widely. Way back in the very early seventies, Mary Matthews recognized that if you took small cell patients and sent them to surgery and they died from operative mortality and you looked at their lymph nodes in the mediastinum, more than 90 percent of them had mediastinal lymph node involvement when they first presented and two-thirds of them had systemic visceral metastasis at the time of diagnosis, and it is sensitive to chemotherapy and radiotherapy.

Now, it is worth just thinking about that for a second. With mammography, the majority of breast cancers that we find are Stage 1 and the cure rate is extremely high.

Two-thirds of small cell lung cancers are in a different organ than when they started when we found them and 90 percent of them have lymph node metastases when we find them, and that is important for this meeting.

TOP

Every other type of cancer in the world has four stages, and lung cancer probably should as well, small cell, even, but because so many people have the disease spread when we find it, we have divided it into two stages, limited and extensive, and limited stage is actually Stage 1, 2 and 3, and extensive is Stage 4, plus people with pleural effusion in Stage 3, plus people whose tumor is too big to encompass in a tolerable radiation port.

In the old days, we used to do zillions of studies to figure out where the disease was at the time of diagnosis, and we found out it was in the brain 20 percent of the time; in the bone 30 percent of the time; in the bone marrow 15 percent of the time; in the liver 15 percent of the time; and in the adrenals 10 percent of the time at the time of diagnosis. A question for us today is, "Do we need to know this and do we need to do all these tests?"

TOP

This is a classic example of why we should be here to talk about small cell.

Obviously, probably what we should be doing is just a PET scan which can look at all those organs all at once. If you look at the literature about whether you ought to stage small cell with a PET scan versus all those other things, you won't find anything, but determining the sites of disease at the beginning is not unimportant.

Of course, whether we ought to be looking at bone marrow to find microscopic disease, submicroscopic disease is another issue.

TOP

This brings me, of course, back to 25 years ago. This is what we were asking. If you look at the drugs it is sort of interesting. You could argue that the drugs that we use today aren't any more effective in this than they were 25 years ago, and they are probably not although they are a lot less toxic.

TOP

So, you know, this is part of the debate. If you randomize people to get chemotherapy alone or chemotherapy plus radiation therapy, those patients that get both will do a little bit better if they are limited stage. When you look at some trials, the differences are much bigger, and when you look at some trials the trials are negative, and when you do a meta analysis you find a highly statistically significant difference. Although the question is, "Even with the very best radiation techniques, are we going to do any better than this or will we just reduce some toxicity?" We will probably reduce toxicity, but I don't know whether we will do better, and that is what we are here to talk about.

TOP

It is probably important the way you deliver the radiation. This was a study recently from ASCO in which patients were randomized to get the chemo and radiation together, concurrently or sequentially, and the answer was that it is better to get them concurrently.

As you can see there is a major 3-year survival advantage of 40 percent versus 21 percent, and of course, that 40 percent at 3 years translates into about 25 percent at 5 years. That means we cure 25 percent of limited stage patients.

Is that better than the slide I showed you before, where that number was about 17 percent, or is it stage migration?

TOP

Does it matter whether you give the chemotherapy and radiation therapy together early or late? There is conflicting data. There are positive and negative trials.

This happens to be one of the positive trials which I show because I believe it is important to give them both together early on, but there are negative trials on this point.

TOP

Another point is, of course, the chemotherapy— does it matter? This was a series of Southwest Oncology Group trials. The curves at the bottom of the slide were all CAV-based treatment, and the curve at the top was etoposide and platinum-based treatment given with radiation. This was not a randomized trial, but it did seem that the etoposide and platinum was a bit better, probably is, but probably not to the degree that is shown on that slide.

TOP

Then we made the striking advance that is it important to give radiation once a day or twice a day, and this ECOG intergroup study suggests that it was better to give etoposide and platinum twice a day with radiation, compared to once a day. You can see that etoposide and platinum with twice a day radiation had a 5-year survival rate of 26 percent compared to 16 percent for once a day.

Now, there is some debate about the meaning of this because the biologic effect of twice-daily radiation was greater than once-daily. The question is, "Should we have a trial of once a day with a higher dose versus twice a day, and where will that lead us?" So we put a bunch of patients on that type of a trial. The people from the North Central Group want us to know that if you do this same study where you give the once a day versus twice a day in the fourth cycle instead of the first cycle, it doesn't make any difference whether it is once a day or twice a day.

So, again, my belief is that you should give them both early on at the same time, and it does matter, and probably the dose of radiation matters. Whether the schedule of radiation matters we could discuss later,

TOP

but really probably more important is, "Should we be putting lots of resources in terms of patients into these trials?"

Clearly I believe that those differences are clinically relevant though as well as statistically different.

TOP

The surgeons can cure Stage 1A small cell lung cancer like they can cure any T1N0 tumor. The problem is we don't find T1N0 tumors, and if we take tumors that have lymph nodes and randomize patients to chemo and radiation versus chemo, radiation and surgery, surgery doesn't help because the disease has spread.

TOP

So, the challenge, if we are going to talk about surgery, is to find patients at an earlier stage rather than incorporate surgery into the treatment of people who have widely spread disease. This was the state of the art 20 years ago when platinum and etoposide came along and we wanted to know whether it was better than cytoxan, adriamycin and vincristine or whether alternating the regimens to prevent drug resistance would be better. You can see from these trials that there was no difference in complete response rate or survival in any of these regimens.

You could say, "Where did we get with platinum and etoposide versus CAV?" Really all we got was less toxicity, and the reason that we have platinum and etoposide is largely less toxicity.

TOP

This is perhaps another way of showing that from 15 years later. We randomized people to CAV or carboplatin and etoposide. Actually, in this particular study, in extensive stage disease the differences in survival, as well as the differences in response, were statistically significant in terms of carboplatin and etoposide over CAV, although clearly we have a long way to go. Again, I think what we have with carboplatin and etoposide is a more tolerable regimen, and

TOP

for 15 years or so, actually the whole 25 years, the questions have been, "Are two drugs better than one? Are three better than two? Are four better than three?" In lung cancer it doesn't matter whether it is small cell or non-small cell. There are lots of trials where two is better than one, and there are very few trials where three is better than two, and as far as I know, there are no trials where four is better than three or two. The question here is, this was a study adding ifosfamide to etoposide and platinum, and in this particular Hoosier Oncology Group study, the addition of ifosfamide made a very teeny, tiny improvement in survival. It is interesting on the basis of the study that the three-drug combination did not become the winner.

The reason for that was twofold. One is the addition of ifosfamide to etoposide and platinum adds a lot of toxicity, and the other thing was there were some negative trials. But we are always left, of course, with the issue of whether three will be better than two,

TOP

and we never give up. So I must admit I was involved in this, which was adding the newer agent taxol. The first thing about adding taxol was that it was like ifosfamide in that to give a therapeutic dose of taxol with etoposide and platinum required the addition of G-CSF, if you wanted to give full doses of each of the drugs.

TOP

The Southwest Oncology Group treated 100 patients with this 3-drug regimen. This is the survival curve. You can wonder whether 11 months is a major advance.

TOP

There are a couple of other trials that get almost identical results with this particular three-drug regimen, and it is better than the Southwest Oncology Group had been doing with other things recently. So what are we doing now?

TOP

That raises this question for these new biologic agents: "Do we need a new paradigm for determining A) what dose to use in clinical trials and B) how to figure out which drugs and what doses to put into clinical trials?"

Certainly, with standard cytotoxic drugs, we think that if there is a response rate of more than 25 percent (50 percent shrinkage in more than 25 percent of patients), we think that the drug is active and worthy of further study.

If we take people who failed first line treatment—that means failed, didn't respond or who relapsed within 8 weeks—a response rate of 10 percent we think is some magic number, and if you take sensitive relapsed patients, that number might be 15 or 20 percent. But what do we do about agents that affect blood supply or agents that prevent invasion? What do we do about those agents? How do we determine which ones are winners? That is hopefully what we are going to discuss in the breakouts, because for molecularly-targeted agents the question is, "Are these end points appropriate or not?"

TOP

Second line treatment—we have again made such huge advances. We have a new drug, topoisomerase-1 inhibitor, topotecan, and we compare that in second line to our old gold standard CAV, and what do we get? We don't get really anything any different, and in this case we don't even get less toxicity. Yet because there is such a dearth of new things, topotecan gets approved by the FDA for tying CAV.

TOP

Nonetheless, despite those negative comments, the new drugs really haven't been studied very much in small cell lung cancer, which is undoubtedly another reason we are here.

Topotecan and taxol, two brand new drugs which have very high response rates—in this particular study 100 percent of the patients responded. Of course, it was a very small study, but we don't often find a 1-year survival rate of 80 percent. Certainly we worry about whether this is really patient selection or a better agent, but you know there is not a lot of study of combinations like this, and there is a whole bunch of drugs which are very interesting.

TOP

Tirapazamine is an interesting drug to combine with radiation. It really has received very little study, oxaliplatin almost no study in small cell.

The topo-1 inhibitors are clearly active in small cell. Their role is completely undefined, and new drugs like MTA (multi-targeted antifolate) get studied in gazillions of non-small cell patients and don't even hit the radar chart for small cell. I guess, Bob, for your comments about industry this is an issue.

Here is a very interesting drug. No studies in small cell are being conducted. Phase III trials are being conducted in non-small cell lung cancer with a drug that hasn't even been studied in small cell.

TOP

So biologics came along, and Dr. Minna got excited a while ago, and you know, where do we stand with biologics? The biologic that has been studied the most has been interferon because that was the first one that came along, although like in anything there are a few positive trials with interferon in small cell. The bulk of the trials, including this one conducted by the Southwest Oncology Group, are largely negative. One of the issues that these experiences raise is, "What is the best trial design?"

We will see in a minute a trial design in which you give chemotherapy and radiation and then randomize patients to the agent that you intend to study.

Another trial design is to take the agent that you intend to study and just give it with chemo and radiation, and we can debate which is the better strategy. The difficulty with this strategy is the n (number of patients required), you can see here is approximately 60. The trial started with an n in the 300 range, and when you give these treatments at the end there is a tremendous amount of drop-off because people don't respond. People progress, and after getting treated with chemo and radiation, people don't want to get something else.

TOP

So I like the typo in this slide. This is supposed to be new classes of biologic agents for small cell, but it came out new classes of small cell for biologic agents. There is a lot, and this is the reason for John's excitement, and it is clearly justified.

We have anti-angiogenesis agents, and if there is any tumor that develops a blood supply quickly, it is small cell. If there is any tumor that invades and metastasizes early, it is small cell. It is hard to imagine how any agent in this class should not be studied in small cell first. It is the most widely metastatic tumor that we have. Antigrowth factors, we actually know that peptides make these tumors grow, and we actually, thanks to our Japanese colleagues and the NCI and VA, have known that peptides are growth factors for 20 years, and it is time to probably get off the dime and actually study some drugs.

Gene replacement treatment, we know the gene. The problem is we don't know how to deliver them, and actually for anti-sense and ribozyme it is largely the same thing. We know the targets. The problem is delivery.

Vaccines—Paul Chapman is going to talk about this, but I will show a little bit of data. Tyrosine kinase inhibitors are showing activity in CML, and they might show activity in small cell as well. Cell cycle inhibitors—ras is not mutated in small cell, and yet if you look at the FTI inhibitors they may work better in small cell than in non-small cell, which probably shows that the FTI inhibitors aren't working on mutated ras but need to be studied. Other agents include retinoids, COX inhibitors, prostaglandin inhibitors, methylation inhibitors.

TOP

So do we have any biologic evidence that these things will work? This is MMP (matrix metalloproteinase), and if you express in small cell lung cancer lots of MMP, then you have a worse prognosis than if you don't have much MMP. So MMP inhibitors make sense.

One of the issues is, "Does it matter which inhibitor you choose? Does it matter which ones you want to inhibit?" That is actually a huge issue because what happens if the first trial that is done is negative? If the first trial that is done with an MMP inhibitor is negative, does that mean that is the end of MMPs or does it mean that you picked the wrong MMP, and two trials are done?

TOP

This was the trial design, which is a trial design which is reasonable. It takes a little bit longer. So patients who responded with small cell were randomized in a couple of studies sponsored by British Biotech (Marimastat) or in several studies sponsored by Bayer (Bayer MMP inhibitor). These trials are completed and done, and if anybody in the room knows what the answer is they should let us know. Presumably it is not a home run because nobody is talking about it, but you know one of the issues is, "Should we study more of these if the initial trials are negative? Should we take a new agent that inhibits different MMPs?"

TOP

So in the standard cartoon the blood vessels develop after invasion, and this is what we all want to achieve.

TOP

We give some treatments and the tumor shrinks. The first question is, "Will angiogenesis inhibitors make tumors shrink on their own in humans, but if we keep giving them the tumor never comes back?" That is what we all want.

TOP

Now, in lung cancer, I think there is pretty compelling evidence. Actually, the pathologists for years have been trying to find things that have prognostic relevance in lung cancer and microvessel density. Even though you have different pathologists counting microvessels in different ways, essentially every study where any pathologist counted microvessel density there was a highly statistically worse prognosis in patients who had lots of vessels, and the number of vessels that is high versus low varies between the studies. This is obviously pathologists counting a little bit differently,

TOP

but certainly in lung cancer there is good evidence that too many blood vessels, i.e., too much angiogenesis, is bad. Of course we have this potpourri of agents which inhibit blood vessel formation, and I am sure the reason that Scott brought us here is to figure out which ones of these we should put in clinical trials first and what type of trial design.

The simplest thing is that you just take any that come along, and you say anti-VEGF should be studied in small cell lung cancer, which actually I think you should. The question is, "What is the priority for anti-VEGF in small cell versus non-small cell and versus radiation or versus adding another cytotoxic drug?"

TOP

So we have these tumor suppressor genes, some of which are incredibly important, and we have the biologist who discovered these here. Retinoblastoma protein is non-functional, absent in 100 percent of small cell tumors. So, why not just add RB? Has anyone seen any clinical trials adding RB into small cell lung cancer patients? No, and that is why we are here.

TOP

Paul Chapman's data has been around for a while—making monoclonal antibodies to monoclonal antibodies often will give you anti-idiotype, and if you immunize patients with an anti-id they develop antibodies—self-antibodies that recognize the anti-id and, also, recognize the tumor.

TOP

Again, this is Paul Chapman's thunder, but this data in a very few, small number of patients is—I don't know, Paul, how many years old is this?—8 years old?

DR. CHAPMAN: No, about 3 or 4.

DR. BUNN: Three or four years old, but how come we don't know the answer as to whether that is real?

TOP

My last comment is what I have been doing for 20 years, and that is looking at autocrine growth factors for small cell lung cancer, and we know that there is a list of peptides that stimulate small cell lung cancers.

TOP

We also know a lot more about their signal transduction pathways, and that allows us to develop lots of compounds. There are many compounds that inhibit either the simplest place at the receptor or interfere with signaling pathways inside the cell. This is work that began with Enrique Rozengurt who is here. Gary Johnson largely helped Lynn Heasley, who is here, define the intracellular pathways.

We actually have compounds. This part of the signaling pathway stimulates proliferation, and this part of the pathway stimulates apoptosis. If you disrupt the balance in terms of this part of the pathway, as opposed to that part of the pathway,

TOP

you actually get very potent anti-apoptosis agents, one of which is a peptide linked at the terminal portion so that you have a dimer. The length of the chains is separated. This molecule is actually being developed in a RAID grant with the NCI largely because to make 10 grams of this compound costs $50,000.

It would be impossible in any grant mechanism other than a RAID mechanism, and we need to find out if these work.

TOP

They also inhibit small cell and non-small cell tumors in athymic nude mice. The problem is, to treat 50 of these mice is prohibitive in terms of production, and that is why the RAID mechanism exists.

TOP

Lots of other things—Lynn Heasley is here, and not giving a talk, so I thought I would show some of his work. Certainly many of the prostaglandin and COX inhibitors have an effect on small cell, as well as non-small cell and should be considered for study.

TOP

Another one of those, of course, is exisulind which is not a typical COX inhibitor but which is being studied in breast, prostate and colon cancer and has probably as much activity in lung cancer as it does in breast, prostate and colon cancers.

TOP

Question and Answer

I am actually a few minutes early so that I can have questions, and I would be happy to answer any questions. Hopefully I have set the stage for some interesting discussion in the next day and one-half.

DR. BUNN: Come on, John, first one? There we go.

DR. MINNA: I think I was sitting here drawing up a laundry list, but just to focus on a couple of points here of potential interest. The first is it would certainly be useful to have a large archive of pathology materials with clinical data tied to it. I mean your microvascular count showed that, let us say, we find that a certain matrix metalloproteinase is the one that we have got to target, and we have good immunologic reagents. It is just going to be difficult for any one center to try to get all of the necessary samples to do that.

So, I think if there is some way, and if you look in the literature a lot of the series where they have done these correlations between histology or various DNA abnormalities are mainly in non-small cell, and there are very, very few studies in small cell where this has been done. So, I think that is one important consideration.

DR. BUNN: Before you go on, keep your comments. I mean that is a huge issue, and that actually PET (Positron emission-tomography) is going to make worse. You know, when we were at the NCI, one of John's great things that he did was when we first went there he said, "We are going to get tumor from every patient." And John brought Adi Gazdar along with him to try to grow tumor specimens. If you want to know how these things got called NCI-H, it was HT, human tumor, and we had HT1 and HT2 on down the line, and those were just patients from whom we got tumor samples. But in those days, in small cell, we were staging them, and so we did liver biopsies and bone marrow biopsies and pleural effusion taps. I mean we put a needle in every single place that you could get a needle, and those tumor samples went right down to the lab to try to grow a cell line.

Now it is a huge problem. Patients don't get staged anymore. So if they have small cell they don't get surgery, and nobody puts a needle in anything other than to get the first cytologic specimen where there are 10 cells. This is a huge problem and even in our own institution where we try pretty hard, we don't have that many small cell samples that have good clinical correlations, and that is a huge issue and PET scanning is going to make it worse.

DR. MINNA: And I would guess most of it, like at our institution, is being diagnosed with FNAs now. So, you are ending up with very limited amounts, but I think that is one thing we need to think about because as we come with new more rationally based therapies, it would be great to be able to do immunohistochemical studies on even 500 samples. That brings me to the second point in terms of the NCI small cell lung cancer panel. There are other lung cancer panels that have been started in Japan and Denmark and Holland with small cell lung cancer. I think it would be very important to know how good those lines are both in vitro and in xenographs as models of disease sensitivity to therapy.

My prejudice is that they are actually very good. I think you showed some of the preliminary data, and I think if we are talking about proof of principle or preclinical studies, I think we probably would do a lot with those.

Now, we have a lot of the clinical data that has been published on those lines. Adi, right now are there 200 small cell lines deposited, something like that that are available?

DR. GAZDAR: About 15.

DR. MINNA: But if we get even more, there are probably over 100 easily available, is that right?

DR. GAZDAR: There are enough lines around.

DR. MINNA: Yes, there are enough lines around, and I think that it is really striking if you compare the in vitro sensitivity or xenograph sensitivity of those to non-small cell lung cancers that many of those before any treatment are quite sensitive. You used several in your studies, and because it is going to be so hard to get these materials if we knew that these things really were good models for bringing up new therapies, and they won't be perfect for everyone, I think these could be another source that we could try out in xenographs.

DR. BUNN: One question there is, "Should you take those 200 cell lines and do DNA array analysis and spend a whole lot of time just looking at the genes in those 200 cells lines and pick your targets on the basis of a very small number of samples that are going to give you a huge amount of data though?"

DR. MINNA: Yes. As an example of that, Ed Gabrielson already has some really exciting data that we will probably hear about later that looked at a couple of lines in a couple of microdissected samples. I think that that could be expanded, but I think that it would be important to know if those are good models.

DR. TRAVIS: I am Bill Travis, a pathologist from AFIP (Air Force Institute of Pathology). One of the reasons I came to this meeting is because I really think there is a place to develop a tumor registry for pathology samples. I am interested not just in small cell lung cancer but in some of the related tumors, atypical carcinoid or endocrine carcinoma, but because of the problem that you are raising I think there would very much be a place to try to develop a tumor registry for surgically resected small cell lung cancers. We have just put together a series of 100 such specimens from the AFIP that we are in the process of completing analysis on, and I would just suggest that sometime during the next day or so that this type of proposal be discussed.

DR. BUNN: I think that is incredibly important. You know even though it is only probably less than 5 percent of the patients, those patients are not being entered onto any clinical trial. What happens is, those patients go to surgery. They have a small nodule. The nodule gets taken out, and it is small cell. Then they get treated with chemotherapy or chemo and radiation at the choice of their local physician, but there is no clinical trial for those patients, and so they are gone. That is the end of it, and their tissue is gone.

So those tissues, you know, could be captured and it would have to be national because you know, each of us sees two of those cases a year. But you know, that is what we are here for. So I think that was a great comment.

DR. SMITH: John brought up the microarray work, and I think this is an important new technology that we have to take account of. But there is a problem with the cell lines. The expression that you are going to see in the cell lines is going to be quite distinct from the expression you are going to see in the primary tumors. You are going to get a tremendous amount of things that come from the tissue culture media, and this is going to be a major problem in small cell lung cancer. You are going to have to use something like laser capture microdissection and maybe universal amplification and characterize the expression profiles in your primary tumors if you really want to have something that is important in vivo.

DR. MABRY: Mack Mabry, Sugen. In industry there are a number of companies who have a number of small molecules that might make sense to test in patients with small cell lung cancer. When we have gone and looked at trying to build trials we find that it is very difficult to find investigators who can generate enough patients to generate a trial that has statistical power to be conclusive. So, in fact, I think we need to think about, as we are talking about getting 5 percent of small cell lung cancer patients, think about the practice patterns. Paul was talking about basically doing away with having tissue samples. I think the question in terms of clinical trials with patients with small cell lung cancer is, "Where will we get the patients to do the tests?"

How many patients with small cell have you seen in Hopkins in the past year?

DR. ETTINGER: I think the numbers appear to be going down. To raise a different issue, we are seeing a lot of non-small cells with neuroendocrine features, and I wonder, "Are they getting confused?" I remember when Joe Atcheson was alive, Mary Matthews talked about fusiform cells, which are larger cells, and the classic oat cell, and not being a pathologist they took out the artifactual and actually looked at it, and it didn't make a difference clinically. But now with the younger pathologists and with less tissue I just wonder what we are seeing. I don't know about the feeling among pathologists, but we are seeing less and less small cell. We are seeing more non-small cell with neuroendocrine features.

DR. BUNN: Part of that may be that that is our problem though. You know, we have all these trials in non-small cell and so patients get referred, and patients know. They come to us because they know we have got something new, and small cell, you know, we haven't tried to get people. It is true we can do a Phase II trial at my own institution in non-small cell in about a year and one-half. With small cell it would take us four years, as a single institution, to do a study. That is what Mack is talking about and the question is, "How do we get around that?" I think taking advantage of our community networks is what we are going to have to do, but that is another point, for sure.

David, and then I will get off the stage.

DR. D. JOHNSON: Your presentation in a sense was a bit seminal for a Phase II study.

DR. BUNN: No, not Phase III trials. The question is whether we should be bringing these new things into them right away. I think the gene array people have already found that actually small cell and non-small cell have much more gene expression things in common than small cell and carcinoids. I think carcinoid is coming from a completely different cell type and that doesn't surprise me. But I think that gene array people are going to confirm that, and there are lots of similarities between small cell and non-small cell.

TOP