Mechanisms of Chemoresistance in Non-Small Cell Lung Cancer: How Might We Overcome Them? by Jessie Au, Pharm.D, PhD

SLIDES & TRANSCRIPTS
Wednesday, June 14, 2000

Mechanisms of Chemoresistance in Non-Small Cell Lung Cancer: How Might We Overcome Them?
Jessie L. S. Au, Pharm.D., Ph.D.

Slide 1:

DR. SAXMAN: Our next speaker this morning is Dr. Jessie Au. Dr. Au is a distinguished professor at the Ohio State University College of Pharmacy. She is going to talk this morning about mechanisms of chemoresistance in non-small cell lung cancer, and ways that we might overcome them.

DR. AU: We have been working on the mechanism of drug resistance and sensitization. What I would like to do today is to give you a brief overview of some of the more commonly used agents in lung cancer, and the mechanisms of resistance known today, and some of the data that we have generated taking a different approach, and how we have used this data to analyze where we show that a reversal of this mechanism of resistance can eradicate well established lung cancer in mice. There is an Ohio group now taking these data and bringing it to Phase I and II trials to test out this concept in man.

Slide 2:

Here is a table summarizing some of the agents that you have heard about today, agents that are often used in lung cancer. If you look across this mechanism of resistance, first of all I should say most of these mechanisms were worked out using cell line studies, which for obvious reasons, are the way it's done. Just to contrast that, I would like to show you later why some of the mechanisms we are working on are different from what you see here.

But if you look at platinum compounds, this is mechanism of resistance -- uptake by carrier transporter, and the increased detoxification in cells in the paired DNA adducts, of course the tolerance of cells develop over time. In alkaloids you see basically the two similar types of mechanisms of resistance, mainly the MDR transporter at the efflux of the drug, and altered tubulin.

The nucleoside antimetabolite gemcitabine, we see here is that the drug is not getting activated. By using bioreductive agents you have the drug delivery problem, because the drugs are often metabolized on the way to the tumor, and there is also an activation problem.

So looking at this list, I think it's pretty clear that the general theme is that either the drug is not getting into the cells, or the drugs get through the cells too fast, or the target site has changed, or the drug simply doesn't get activated. So that's the general theme that we see over and over again for resistance.

One area that really hasn't been looked at a lot is drug delivery. In fact, I'm not going to talk about that today, but I just want to point it out because all these studies were done in monolayers; drug delivery is never an issue in the study. But drugs such as taxanes and vinca alkaloids and platinums, those are highly potent amounts of drugs. So you want to look at the diffusion of these molecules. They behave far more like a macromolecule than a small molecule. So penetration of a solid tumor, as we and other groups have shown, is always a problem for these agents. But that's the subject of another discussion.

What I would like to do now is to change and tell you what we have done, using a different system, where now, instead of taking monolayer cells, we go back to a three-dimensional solid tumor, and put the cells in contact with stromal tissues. Because we believe that the stromal tissue plays a very important role in resistance.

Slide 3:

What I will show you in the next ten slides is that we found chemoresistance a new epigenetic mechanism, due to protein factors that are expressed in the solid tumors. This is only found when you keep the tumor cells in contact with the stromal cells.

Slide 4:

Here is a system we have been using now for about 12 years. This is a very old technique. Basically, it's tumor explant. You take a tumor from patients or animals, chop it up into small pieces, put the pieces on the collagen gel matrix, and you are able to maintain growth this way. And by doing this, we are maintaining tumor cells in contact with stromal tissues. And using this system, we are able to generate a typical dose response relationship.

Slide 5:

Here is an observation made way back in the early nineties. We took some patient tumors, and in this particular case it was head and neck tumors, and we looked at 5-FU in this case, and compared primary tumor paired with lymph node tumors on the same patient. We only had three pairs at a time, but the trend is very obvious. The metastatic tumors is always more chemoresistant than the primary tumor.

That led us to the thought that perhaps the resistance in metastases is due to environmental factors. Coupling that with some of the anecdotal observations in the clinic, namely certain tumor types are more chemoresistant such lung, brain, and so forth, led us to the thinking that really the tumor environment has a major role in the way the tumor cells will react to drugs.

Slide 6:

So then we came up with this experiment, and this was done a few years ago. We thought that if this is the case, if this is truly the reason, it's the tumor micro environment, and we in this case we used the synergistic model, so this in a rat tumor model. We take tumor that is grown subcutaneously, letting it metastasize to the lung and to lymph nodes, and then take the tumor out from that site, and put it back in subcutaneously. The idea is if it is the environment that is causing resistance, every time we bring this back to a SQ site, the resistance should be lost. So we did this with 20 generations, and this is the data the found. Each time, when we bring the metastatic tumor back to SQ site, you can see that the resistance is gone in the primary tumor, even though it was generated from a metastatic tumor.

When the SQ tumor goes back to the metastatic site, then you get back the resistance again. So even if it's from the same generation, which means paired tumors from the same host, or from the parent-daughter pair, which is this pair, you always get the same resistance.

Slide 7:

Which led us to basically confirm the hypothesis that what we have seen here is not a genetic resistance mechanism, but an epigenetic mechanism, which is expressed in the organ site. In this case it was in the lung and lymph nodes.

Then we came up with a second hypothesis, and that is that the system is caused by extracellular factors and it is reversible.

Slide 8:

I'm going to show you a bunch of experiments that basically showed that we did some detective work. We took the conditioned medium from the tumor from different sites, primary tumor and metastases, and we looked at the proteins, separated them out, and then combined them back together. And what we found is not one protein, but two proteins causing the resistance.

I used three different agents, because they represent three important drug classes, taxane, and topo-1 inhibitor, and also the anti-metabolites. What I have shown you is the conditioned medium we collected from the tumor. When we add primary tumor, with primary tumor conditioned, medium there is no effect. There is no resistance. When we add the lymph node conditioned medium, we do see a shift of the curve to the right, indicating that we are inducing resistance. The same thing is true for lung.

Now what we also found, to again check the hypothesis, is that it is the stromal tissue and not necessarily the tumor cells that has changed. We took histocultures, the same explant you have seen, and separated them now into monolayer cultures, and let it grow out in time, going from pass 0 to pass 3. As time goes on, we lost resistance that suggests to us the protein factors that we have are decreasing in time. What this does for us, because of the quantitative differences of the different conditioned medium, it provides a system so that now we can start hunting down the proteins that are causing this problem.

Slide 9:

This is a table with a lot of numbers, but I just want to point out to you that it gives you IC50’s in case it’s not clear to you before. Obviously, let's look at the histoculture. There are some differences in IC value. They are always going higher. Lymph node is more resistant than the primary tumor, and the lung conditioned medium is even more resistant.

Again, when we plate this out in monolayers, you start seeing the numbers get closer. In time, eventually all the IC50 collapse to the same number as if there is no condition medium present.

Slide 10:

So using that system, we looked at the proteins. There are two that are really of interest to us, by comparing the different histocultures, and also the monolayer as time changes, we are able to see the changes in the protein expression, and identify the two, the 18 kilodalton as the bFGF, basic fibroblast growth factor, the 14 kilodalton, which is the aFGF, acidic fibroblast growth factor.

Slide 11:

Then we did a quantitative analysis where we looked at the protein levels and compared them to see if that agrees with what we have seen in chemoresistance. And you can see in time the protein level did go down, and there are differences in the protein levels between the different systems. As the differences disappear, so did the resistance.

Slide 12:

So if we looked at a relationship, everything agrees with the fact that these are the important proteins that cause resistance. First, the rank order of FGF concentration in the condition medium of the different systems parallels the rank order of the systems in these cultures. And secondly, as we lose FGF over time, we are also losing the resistance, and as soon as the level becomes equal, we get equal sensitivity. Our statistics tells us this is probably a very important factor.

Slide 13:

The next series of studies we did was to first inhibit the FGF’s. You inhibit FGF by antibody. This is a monoclonal antibody. The interesting thing here is with the aFGF antibodies, in both cases we saw a concentration dependent reversal of the resistance. However, for aFGF at the highest concentration we used, the highest level of reversal was only about 60%. We were never able to bring it back all the way to the original sensitivity. With bFGF antibody, however, we were able to bring it all the way back to the control. So this suggests to us the two factors both are important, but they have different roles.

Slide 14:

Some of the data I'm not showing you, just basically checking numbers, so we find out how much of the bFGF we need to reverse the resistance, and realizing the level was way too high for any clinical relevance, that naturally started us looking for a second protein, and we found aFGF. Here what we have done is to take away FGF on the conditioned medium by immunoprecipitation. Again, when we took aFGF away we cannot completely reverse the resistance, only halfway. But when we did it with the bFGF, again we see complete reversal.

And this experiment here defines the importance of the two factors. What we have done here is take both of them away this time, and now add back one of them at a time. If you take both of the FGF’s away and add aFGF back, we are never able to get any resistance, which suggests to us that aFGF is not required for the resistance.

Now if you add back bFGF at the level that we found in the conditioned medium, we were able to get partial resistance as we see in the conditioned medium, but not fully. We add back both, we get complete restoration of the resistance. This suggests that bFGF is the key molecule, that is the FGF is required, but aFGF is certainly important as well, because that's what brings it back, the resistance at a level of bFGF that is clinically relevant.

Slide 15:

This is now taking the condition medium out, and just using recombinant proteins, and showing that the recombinant proteins did exactly the same as we were seeing with the conditioned medium. The level we see in the conditioned medium is 0.1 nanogram per ml of aFGF, and 0.9 nanogram per ml bFGF, and at that concentration we get complete restoration of the resistance, telling us that those are the only two proteins in the condition medium that are causing resistance.

Slide 16:

What I had shown you before, the data all are from the deoxyuridine uptake, only looking at anti-proliferative effects, so to speak. But we also wanted to see whether the same thing happened with apoptosis. We were looking at LDH release given that the index of cell lysis. And you see that again, the story agreed with what we were seeing before. And the same observation is seen for all three drugs. Every time we add more FGF or the conditioned medium, we reduce the resistance to apoptosis as well, or cell kill anyway.

Slide 17:

So with that we took it to the next level. I know this drug suramin also elicits a very negative response, so I ask you to look at that as just a small molecule that we chose to use to inhibit the FGF, and not necessarily that it is the best drug there is. But that is something we could regularly get our hands on and use. Let me point out that suramin itself in this particular cell line system, the IC50 is about 100 micromolars. What we are using is 10-15 micromolars. We fully recognize at that high concentration this drug would never see the light of day in the clinic. So it's important to use a concentration where the suramin is not toxic.

So the level we used -- and you see here when we add suramin we are able to reverse the resistance in a concentration-dependent manner -- at 10-15 micromolar we were able to completely reverse this resistance. I can come back to that in minute. The concentration of suramin is going to be extremely important, and that's something that we will watch for in the clinical trial.

Slide 18:

The next slide, it shows you the in vivo study that we have done. This is a lung metastasis model where we put IV tumor cells in tail vein. PC3 is a prostate cell line. They give us 100% metastases. Normally we had a group of 20 animals and we take two animals randomly and just open them up. When the tumor is established, and by that we mean having at least five nodules larger than 1 millimeter in diameter, we start treatment.

This is saline control. You can see the nodule in the control animals. And this is a microscopic view of that. Suramin alone really did not do a whole lot. You still see a lot of tumor nodules and tumor cells. The tumors look microscopically about the same as the control. Doxorubicin alone showed some activity. There is reduction in the number of nodules, as well as the tumor mass. When we combined the two, what we see is a pinhead-sized tumor left. And what is left, as I will show you in the next slide, are mainly apoptotic cells or cells that no longer appear healthy.

Slide 19:

One way we analyze this data is to take the percent tumor free. When we combine suramin to reverse the resistance or enhance the effect of doxorubicin, we are able to eradicate tumors in 42% of animals we have seen. By that we mean no visual tumor nodules on the surface or in five random microscopic sections throughout the lung. You can look at the percent of lung surface area. Again, you see a very drastic improvement of the reduction of the tumor mass. Just simply count the number of apoptotic cells by the usual standard, the TUNEL or methodological changes. Here, when we add suramin to doxorubicin, we are able to nearly triple the percent of apoptotic cells left in the tumor.

The problem with looking at apoptosis is that you are measuring something that is no longer there, because the cells have disintegrated and disappeared. So we also measured the tumor cell density, what is left appearing nonapoptotic. And again, you see that suramin drastically improved the activity of doxorubicin. In terms of body weight loss, suramin did not enhance the toxicity of doxorubicin. We have basically done the same experiment now with taxanes and had the same observations. Taxane is much better drug in this particular tumor model, and we are able to produce a similar type of data with much less body weight loss, less than 10%.

Slide 20:

That brings us to another study. We asked then “is there any clinical relevance to our work?” We have archived a number of tissue from patients. In this case we had about 100 patient tumors where we tested activity with taxanes. Mainly we are looking at IC50 for the DNA incorporation index. So we are looking at anti-proliferation effect. We are also looking at overall effect, which is anti-proliferation plus apoptosis.

Of all the things we looked at -- the tumor pathological screening, tumor stage, grade, labeling index, p53 expression, bcl-2, and so forth, and PGP of course is also one of the important ones, because that's a major mechanism of resistance -- bFGF came out to be the best predictor of the response to the paclitaxel in these 96 patient tumors culture, three dimensional histoculture. Now aFGF was not as high in the ranking, but if you combined the two, that's only second to bFGF plus stage. We don't know why stage is important, but certainly the two FGF's combined is a very good predictor of response. And the same thing for the overall effect. So there seemed to be some agreement with the clinical specimen at least -- the FGF's are important in predicting response.

Slide 21:

The other observation is, if you ask the question how do know that suramin really worked by inhibiting the FGF pathway? What we have done is add antibody together with suramin. We found that if we added additional antibody, we did not further enhance the sensitization, suggesting that suramin and the FGF antibody share the same mechanism, and that the FGF inhibition by suramin at the concentration we were using was maximally inhibited.

We have found the same observations were made in multiple cells lines, including a lung cell line that is used on the panel. We also had done some work looking at the downstream events of the FGF that may be causing the resistance.

Slide 22:

But now back to how are we going to use this data for therapeutic gains. We have designed Phase I and Phase II trials. It just received a concept approval from the NCI about 10 days ago, and we are now submitting a protocol for review. What we are doing here is using paclitaxel, the 200 milligram/m² dose, and carboplatin with an AUC 6, and we are using suramin at a very low dose. Some of you have been involved in a suramin trials and you recognize this is a low dose.

The first change is a low dose. At this dose we expect to get somewhere between starting out at 90 micromolar right after an infusion. What we want to see is 10-15 micromolar at 72 hours, because we believe if we go too high we are going to run into a lot of the effects that we don't want. Another change here from the conventional treatment that we have been using on suramin is we are only giving suramin at the same time as the chemotherapeutic agents are given for obvious reasons, because we are only using suramin as a sensitizer, and not as a chemotherapeutic agent as it has been used in the past.

Because there is a lot of work done already on suramin, we can calculate the dose the we need, and we believe the Phase I portion of this study will be done very quickly. We don't really expect any surprises in terms of the dose that we have chosen in what we are going to see in patients. Hopefully we will move on to the Phase II portion very quickly. There are two groups of patients we are going to look at. One is chemotherapy-naive patients, and the second group will be patients who have been treated with paclitaxel and carboplatin, but have not responded. The endpoint here, we are looking for a response rate.

A correlative study, recognizing that there is always a possibility that it doesn't work, and one thing I am most worried about is the FGF level we use was based on a rat tumor model. The animal study was done in a mouse model. We don't really have a good understanding of how the FGF expression works in humans; whether the FGF level we have seen in animals is reflective of what happens in humans.

So it's very possible that the suramin level we are using may not be sufficient to inhibit what is going on in humans. So we need to determine FGF levels in patient tumors and in effusion fluid, and then take this level and go back to an in vitro system where we can then look at the pharmacokinetics/pharmacodynamic relationship and ask if the concentration of suramin we obtained in that particular patient is sufficient to inhibit the FGF that is expressed in the tumor. I am very interested to see if some of the genes have been identified in this pathway is also changed in patients.

I have used up my time, and thank you for your attention.

[Applause.]

Slide 23:

DR. SAXMAN: We have a few minutes for questions.

DR. GIACCONE: The downstream genes that come after bFGF that you identified, which are those?

DR. AU: We looked at the microarray to look at those. A lot of them are new. You don't see the typical p53 RB type of genes. We didn't find those. They are obviously different. We have quite a few of those and we sequenced those, and now we are just about to test whether they are actually important. The microarray had a lot of pitfalls as well.

DR. GIACCONE: Suramin has a lot of targets, and a lot of toxicity as well. Are you sure you are going to answer the question you are asking?

DR. AU: Not for sure. Actually, you point out an area that we have been working very hard at for a number of months with not much success. We have been trying to find more molecules that we can screen. Obviously, suramin is a bad drug to use from that standpoint. It is not a specific inhibitor. We are looking at the possibility of developing humanized antibodies to bFGF for example. We talked to Pharmacia Upjohn, because I know some people there. We are hoping to find through their link with Sugen, some of the compounds Sugen may have that are worth our screening. So they are not coming up with any compounds. I just found a link to Germany where there is a professor who has synthesized over 800 analogues of suramin. So we are hoping that he will provide some of that to us so we can screen them

But I think this proof of concept in humans is important. I think in a few months we will have some clue as to whether this has any merits at all in humans. Then hopefully some companies then will get more interested in it.

DR. PLUDA: Your data actually suggests that suramin is acting by binding recirculating FGF and preventing it from having a paracrine effect on the tumors. It is known that suramin does bind free FGF. There have already been clinical trials with much higher doses of suramin in combination with drugs like doxorubicin. One would assume if the mechanism is really binding a free FGF, that the high concentrations would bind it just as well as the low concentrations. Yet I'm not aware that any of those combination studies with chemotherapy in suramin were positive.

DR. AU: That's a very good question, and we have the same skepticism going in, knowing that suramin is going to raise a lot of eyebrows. But now you are touching on another area that we are very interested in, and that is combination therapy. You heard this morning there are all sorts of ways of combining things. I'm coming back to suramin in a minute. But what we have learned from paclitaxel, as you know, there is sequence dependence in combining different drugs, and that's how we are starting this area of research.

What we have found is that if you combine agents that block cells in different parts of the cell cycle, what you are doing is not necessarily getting synergy. It's very easy to understand conceptually. Take suramin -- if you go about 50-75 micromolar, you block cells in the G1 phase. Then you combine it with paclitaxel, which works on an M phase.

Now the suramin half-life is longer than 21 days. So it's going to hang around much longer than paclitaxel ever will. So the cells are blocked now in the G1 phase. You can't go to M phase to get the cytotoxic insult from paclitaxel. And we have shown this mathematically. We have developed a mathematical model to test this and prove that in cell lines this is the case.

If you combine drugs that work on different parts of the cell cycle, and by just simply blocking it before, if you use one drug to block it, and before the second drug can come in to show its effect, you are going to get either antagonism or you're going to get only the effect of the single agent. I think if you look back at the clinical trials, all the data shows that the effect that you get is no better than a single agent if you combine suramin. So I think what happened in the past, the MTD concept is flawed in that particular case, because you push everything so high, that you were actually blocking cells in G1 phase, and therefore you wipe out all the benefit you are going to get from blocking the bFGF.

DR. PLUDA: I guess one way to look at that in your system would be to instead of using 15 micromolar suramin, to actually use 100 or 200 micromolar, and see if at the higher concentrations of suramin you no longer have the effects that it has in combination with FGF. And if you still saw it, you would have to rethink that hypothesis.

DR. AU: We have done the experiment looking at cell cycle block. What you mentioned is true. It should be easy to do. But if you look at the assays one has in the laboratory, you can either do bromodeoxyuridine incorporation, you can do MTD, you can do SRB. All you are measuring is the effect. You don't know where it is coming from. This is why it's important to look at it downstream of the event. Then we can start looking at that. Am I answering your question all right? You cannot really see it. You cannot tell where it is coming from. Is it blocking the FGF, or suramin is doing a G1 block, so therefore you are going to get some effect from suramin itself.

DR. PLUDA: But you would then see the exact same thing with the suramin alone as you saw with the suramin with FGF then. You would lose any effects of the FGF. You wouldn't see changes. Everything would always be flat, no matter what you did. And you would be able to at least have some idea of whether or not at the higher concentrations, the suramin alone was having all the effects, or whether or not the FGF was still able to have some effects.

DR. AU: We can talk more about this later, but what you are talking about is how to analyze synergy. Analyzing synergy is a very complex problem. It's been debated over two decades, and there is really no good method. I did a combination index analysis, and you will find that the synergy that we see with suramin and doxorubicin will be highest at lower concentrations, and slowly you lose it. I still cannot conclude that suramin is not doing what it did. All I can say is that synergy slowly decreases as you increase the suramin concentration, my thinking is that because now we are actually looking at the suramin cell cycle blocking effect more than we are looking at the doxorubicin effect. So therefore, you start losing your synergy. The combination index, when you go back to one, you actually don't have synergy any more.