SLIDES & TRANSCRIPTS
Wednesday, February 16, 2000

Can Treatment Be Tailored to Biologic Characteristics?
Leonard Saltz, MD

Dr. SALTZ: Good morning. Jim and I were talking about this last night, and my bias is that 100 years is too long to be satisfied with and that I hope that things are going to move faster than that.

I am, I believe, an eternal optimist by nature. I am both a medical oncologist and a New York Jets fan. So I have to be, but what I would like to show you is a glimpse of where I hope we are going to be treating cancer, and I would like to say realistically within the next 5 to 10 years.

This is not the end of the story, just as the progress that we have made while encouraging is far from the end of the story. This is the beginning of the way that I think we might be able to approach individualizing care for patients. I am going to take you through some preliminary data.

TOP

Okay, let us go to the next slide, please?

This is some preliminary data for the rationale that we are exploring for looking at colorectal cancer on the basis of some selected markers, and the approach here is based on the observations that some colorectal cancers are highly sensitive to 5-FU-based therapy. There are a few patients, an unacceptably small minority, but a few patients for whom 5-FU is a really good drug.

Some people with Stage III disease are cured by 5-FU-based therapy. Some people with advanced metastatic disease have a very nice durable response to 5-FU, but it is a small percentage, and if we could identify those people we could target 5-FU therapy in those directions and, more importantly, if we could identify people for whom 5-FU is clearly not going to work, we could spare them that therapy especially now that CPT-11 and now with oxaliplatin and hopefully some other new agents coming along represents a viable alternative treatment strategy.

So, somewhat along the lines of the question that I was posing yesterday morning in discussion, in addition to being able to say who has got a good prognosis and who has got a bad prognosis, we are looking to be able to do something different about it, to be able to base some of our therapy on the basis of the information.

Preliminary data suggest that molecular markers can identify responsive or resistant tumors to these agents, and I want to take you through some of the preliminary and I want to underline the word "preliminary" data.

So, just for brief review, if I could have the next slide, please?

TOP

Everybody is familiar with 5-FU, and in a way it is a little bit scary if you think about how well this fulfills the paradigm of a drug we think we are looking for. Obviously nobody is satisfied with 5-FU, yet here you have a rationally designed drug where we understand its mechanisms of actions exceedingly well, where it really, in and of itself, created a paradigm shift in how one could treat the disease and became the standard of care for decades.

We would all like to be able to duplicate those efforts and move forward, and it has been remarkably difficult to do that. I am not trying to sing the praises of 5-FU. I am just pointing out how difficult the problem is and what a technological step it really did represent when you consider it in the context of the problem, but 5-FU as you know is metabolized primarily to Fdump which then complexes and interferes with the function of thymidylate synthase so that thymidylate synthase or TS is the primary target for 5-FU. 5-FU is deactivated primarily along the pathway where dihydropyrimidine dehydrogenates or DPD is the rate-limiting enzyme and it can, also, be activated on a salvage pathway using thymidine phosphorylase.

Next slide, please?

TOP

Some interesting data was published by Gail and Larry Leichman and Kathy and Peter Danenberg a number of years ago looking at thymidylate synthase versus response to 5-FU leukovorin in colon cancer on the basis of an RTPCR analysis of the tumor and to summarize their data what they showed is that responders all had low TS.

Non-responders could have either a high or a low TS, and what you could take away from this is if you have a low TS in the tumor you are in the game. You have a reasonable probability of response. More importantly if you have a high TS, at least in this population, 5-FU is not the drug you are looking for. It ain't going to happen.

Next slide, please?

TOP

In some follow-up analysis from this same group of 29 patients what the Danenbergs showed is looking at again response and non-response in terms of DPD versus response, again a low DPD was necessary for response to be possible, and a high DPD was inconsistent with having the response.

Next slide, please?

TOP

And looking again at thymidine phosphorylase a similar observation. If you put some of these together on the next slide, you start to increase the power of this to predict 5-FU success or failure, and here what I show you on this plot is TS and DPD, and you can see that the low TS, low DPD patients shown in the magenta here are the responders versus the non-responders in yellow high in one or both.

Next slide, please?

TOP

If we do this for TP versus TS again a similar kind of pattern is seen, and if we put it altogether --

Next slide?

TOP

-- looking at it graphically, again, for this small population of patients the overall response rate was 29 percent. If you took patients with low TS, the response rate was 57 percent. We are not increasing the response rate here. What we are doing is subselecting down. We are looking for ability to identify who the responders are going to be.

If you took TS that was low and DPD, now 92 percent of the patients are responding, and if you take TS, TP and DPD all with favorable prognostic variable, 100 percent response rate.

TOP

Of course, this is a self-fulfilling prophecy. We defined this 100 percent by this population, but what it shows you is the potential for this kind of analysis to be able to anticipate in advance which patients 5-FU is going to be a high probability drug for response and just as importantly, from a practical point of view, for which patients 5-FU isn't going to be worth the time. There is going to be toxicity, expense and delay that isn't going to benefit the patient.

Next slide, please?

TOP

Just showing you this from a graphic point of view here are the normalized curves for these patients, and you can see that the responders are all low in all three prognostic markers whereas if any of these markers are high response was not seen.

Next slide, please?

TOP

Can we do something about that? This is some data that I showed at ASCO a couple of years ago. This is the first slide that I showed you of the published data looking at 5-FU leukovorin treatment responders and non-responders against relative TS gene expression, and these are some patients that we analyzed, or that Kathy and Peter looked at for us, that were treated at Memorial Hospital on CPT-11 regimens, and when we looked back at their TS you can see that all our responders here were at or above the level of thymidylate synthase that was inconsistent with the response to 5-FU. This is evidence that not only can we predict the people with the bad outcome, but we have an alternative that is logical for these people.

Next slide, please?

TOP

And so this is the current prospective trial that we are running at Memorial Hospital and a few other centers have now joined us in this, and the clinical work is being done primarily on the East Coast. The laboratory work is being done in the Danenberg lab.

The specimens are obtained, Fed Ex'd, and we are getting the PCR analysis results up on my e-mail the next day, which is really rather remarkable. I am getting PCR results a little faster than I often get CBCs, but that is important in terms of practical applicability. This is not delaying patient therapy. We are getting this information in a very rapid real-time practical manner which is attributed both to the ability to shrink the world and move specimens quickly and to enormous efforts and dedication on the part of the Danenberg laboratory.

Now, in this study I said metastatic colorectal cancer biopsy frozen. I am going to show you in a moment that we don't need to do that anymore, but when we started this trial we were obtaining fresh frozen tissue from a metastatic site on all patients, most of them by core needle biopsy. We actually are able to do this as well from cytologic aspirates. We are doing FNAs as well and patients were having a treatment decision made on the basis of their TS. We said, okay, the preliminary data look good here. Let us go with it for TS. We will know the TS, and if the patient has a low TS we will treat them with 5-FU leukovorin. If the patient has a high TS, we will say that those people are low probability for benefit, and we will treat them with CPT-11 right out of the starting gate, and we, the clinical folk are blinded to this data, what the DPD and TP status are until after the final response data for the patients are known, so that we can look at whether or not, in fact, we can refine and identify with a higher probability those patients who are going to respond to 5-FU.

Our anticipation is that the high response rates will be the patients that are low TS, low DPD, low TP and that the low TS 5-FU failures will be accounted for on the basis of one or both of these markers. Now that is the hypothesis. The study is ongoing. I don't have data from this confirmatory trial to show you at this point, but that is where we hope that we are going to be taking this kind of technology. Obviously if we are able to confirm this sort of thing it would have interesting applicability for the adjuvant setting.

May I have the next slide, please?

TOP

This is to show you relative TS gene expression in matching frozen and fresh frozen paraffin embedded colon, both normal and tumor tissue. The N sample is normal. The T sample is tumor just showing you that the correlation is very strong, and we are now comfortable offering this kind of analysis to patients on the basis of paraffin-embedded tissue.

This now has practical applications for patients who have already had a biopsy done in the course of their management, and it, also, opens up the huge possibility for some retrospective analysis where we have clinical databases, and we can go back and obtain the tissue.

Next slide, please?

TOP

These are some of the putative determinants we are looking at for 5-FU. We are looking for TSTP. We are looking at p53 status DPD. For CPT-11 we are on a bit of a fishing expedition. We are looking at Topo-1. We are looking at the excision repair gene, ERCC 1 which correlates with cisplatin activity. Peter will show you that data later in some upper GI malignancies. We are looking at p53 as well as bcl-2/bax status. Again, we don't know if any of these markers are going to be useful for CPT-11, but our hope is that we would be able to identify some markers which may be useful in identifying those patients most likely to benefit or just as important as we now have newer agents coming along, those patients for whom it is not going to be a productive undertaking, and we can spare them the expense and the toxicity.

Next slide, please?

TOP

Now, these are some data that are fresh off the press and are extremely preliminary, some slides that Kathy Danenberg sent me earlier this week looking at some patients that are 5-FU refractory and who received 5-FU/oxaliplatin therapy in a salvage setting and again looking at response on the basis of some of these predictive markers.

We looked at ERCC 1 because of its ability to predict response to cisplatin, and we see a trend here. I think we are going to need larger numbers to see whether this is going to be of any real usefulness or not, but again the responders appear to be lower than the stable and the progressing ones, higher here. This may have some usefulness.

TOP

The next slide was a bit of a surprise to us, and I don't have a good explanation for this yet. This is looking at the relative TS expression versus response to 5-FU/oxaliplatin for the same population and here all of the responders were low TS, and in fact the highest TS patients here did not respond to this oxaliplatin-based therapy.

It is clear that oxaliplatin is salvaging patients who were 5-FU refractory in this data as in many other sets. The explanation here is not going to be as simple as we would ideally like it to be, and this is going to require a little more digging, and obviously it is going to require much larger data sets to interpret.

Next slide, please?

TOP

In conclusion, preliminarily, and I really should underline in bold that word, data suggest that the relative gene expression of TSTP and DPD can be used to accurately predict sensitivity or resistance to 5-FU.

Our hypothesis is that elevated expression of Topo-1 as well as p53 bcl-2/bax ratios and ERCC 1 status may predict sensitivity to CPT 11, and that we may be able to develop markers that would be useful for oxaliplatin prediction as well.

Next slide?

TOP

I think that is my last one there. So, I am sorry, these preliminary data, if verified could provide a powerful tool for selection of therapy, reducing unwarranted toxicity in advanced disease and, also, most importantly we are looking at the applicability of this for moving this rapidly into the adjuvant setting, if we can show that we could perhaps make a rational selection of chemotherapy on the basis of analyzing the tumor that the surgeon has taken out.

That could potentially let us improve the cure rate for local regional disease following surgery. We are looking very carefully now at the correlations between primary and metastatic disease. There are going to be some differences, and we need to sort that out.

Obviously it is the metastatic disease that we are treating in the post-surgical setting, and we will have to see whether things that can be learned from the primary could be applicable for adjuvant therapy on that basis.

With that I will stop. I think we will get the other speakers up and then take questions.

Thank you.

(Applause.)

TOP