ALL Leukemia

SLIDES & TRANSCRIPTS
Monday, May 12, 2003

Minimal residual disease in ALL – Methods and Significance

Wendy Stock, M.D.
Jerry Radich, M.D.

Slide 1:

DR. STOCK: Jerry and I are going to split this up a little bit, in terms of I am going to review the non-transplant minimal residual disease data and Jerry is going to review the transplant related minimal residual disease data, and we both have some questions that we hope to address more in the afternoon session.

I think most of us now know that MRD does appear to be a valuable independent prognostic marker in ALL, both in pediatric and in adult, although the adult data, as you will see, are less numerous.

This has been demonstrated in multivariate analyses, and reliable quantitative methods are available for standardization of MRD quantification.

High levels in early remission are strongly predictive of relapse. Are we ready to use this for treatment stratification? I think so, but there are numerous issues that we still need to discuss, and we hope to address that, again, in the afternoon session.

Also, MRD evaluation can help us evaluate efficacy of novel treatment components, and that is being done.

Slide 2:

Just to review, the three basic methods for MRD detection include flow cytometry, PCR of the fusion genes, and PCR of the immunoglobulin and T cell receptor gene rearrangements.

The sensitivities of these various techniques varies, with flow cytometry being the least sensitive, and may have implications for certain decision making processes in MRD, whereas the PCR fusion genes are the most sensitive, and the immunoglobulin gene rearrangement somewhere in between.

All of them have good informativity. The advantages to flow are that it appears to be applicable for most. It is quite cheap, relatively so, and very rapid.
In contrast, the fusion genes are not informative in the majority of cases of ALL, but again, are very sensitive and specific, and relatively rapid.

The fusion gene approach using the immunoglobulin gene arrangements are widely applicable, very sensitive, and patient specific, but they have some disadvantages, which you will see here on this next slide.

Slide 3:

So, the disadvantages got kind of cut off here. Limited sensitivities, the flow useful in the minority of patients with the fusion genes.

The immunoglobulin genes are time consuming at diagnosis, because you have to find the individual patient clone. That is even a question we can discuss in the afternoon, whether that is really essential. It is also a little bit more expensive, the technique.

Slide 4:

In any case, this reviews the three large series from the pediatric group who have the most data in MRD monitoring.

Really, the point of this slide, which I think I am going to skip over here, is that all the groups, depending on the method, and at the time point following remission achievement, all show relatively the same thing, and that is that MRD levels do predict very accurately the outcome, and that high levels of disease down here show high relapse rate with both PCR methods, in these two studies, and flow in this method, and low levels of disease in these upper columns, show relatively low levels of relapse.

Slide 5:

These data, just to go over a couple of these studies in a little bit more detail, this is the one from the French group, from Helen Cave and associates, and what it shows is that quantification is quite important, that low levels -- this is non-quantified disease, and the outcome is intermediate, whereas patients with no residual disease, or very low levels, have a relatively good outcome, in contrast to the very poor outcome with high levels. So, this demonstrates very clearly that quantification is important.

Slide 6:

This is with flow data. These are the data from a large European series looking at the outcome with flow cytometric analysis quantification.

Again, the same thing, low levels of disease or no disease are an excellent outcome, whereas high levels of disease, as defined by flow cytometric analysis, have poor outcome.

Slide 7:

Now, to move on to the adult data, these data were published. This is really the only relatively large series that has been published in adult ALL. It is from the British group. It was published last year in JCO.

The point that I wanted to make with this slide is the fact that multiple time points were assessed here, and they all show similar things, that low level of diseases is a semi-quantitative method.

All levels of disease -- at all time heights, high levels of disease are predictive of outcome, and they are all statistically significant, but different time points show slightly different things.

In this series, the time point at six to nine months following remission was the most discriminatory for predicting outcome.

On the other hand, one has to consider that, at six to nine months, especially in our data, our high risk patients are already relapsing.

Slide 8:

TOP[There is no text associated with this slide.]

Slide 9:

This is the data, these are some of the data from the German group, the MRD in adult ALL from the Germans. This has only been published in abstract form.

What they did in their series was, they took just B lineage patients and they defined the B lineage patients' standard risk.

This study was done with only standard risk patients, that is, patients with relatively good risk markers, low white count, good achievement of remission, no cytogenetic abnormalities that would define high risk disease.

What they found was that, again, in this very defined group, high levels of disease predicted outcome, low levels of disease predicted a poor prognosis.

When they combined two time points, one following induction therapy and one time point anywhere between 10 and 52 weeks of their intensification therapy, they found a group where there were no relapses.

They have used these data now to design a study where they are actually decreasing intensity of treatment, which I will review for you in a minute, based on these data.
In this study, they actually took patients that we really know nothing about and were able to discriminate a different subset of patients, both good and poor risk.

Slide 10:

These are our CALGB data. We have preliminary data. Again, these are on our patients that were recently treated on our most recent study, the 19802 study where, during induction, patients received intensified daunorubicin.

We measured MRD at the end of induction, when all patients who were in morphologic remission were included in this study. Our data are rather preliminary in terms of the numbers of patients.

Slide 11:

These are our data, showing again that -- this was using real time PCR of the immunoglobulin gene rearrangements and T cell receptor gene rearrangements.

We used a consensus 5’ primer, a patient specific 3’ primer and a consensus probe, using the light cycler.

These are our data at the end of induction therapy at the time of remission and we, again, are able to discriminate pretty clear two prognostic groups that we hope will be confirmed in a larger series of patients. This had 40 patients in it.

Slide 12:

Currently, these are the studies that are ongoing, where MRD interventions are being made. Clinical interventions are being made on the basis of MRD.

These are European studies. In the BFM, they are using early transplant for high risk MRD patients, and reduction of the re-intensification for the patients who have good risk disease.

In the GM ALL in the adult study, based on the data that I showed you, the standard risk patients who had low MRD at all time points, had a zero percent risk of relapse, and they are actually not doing consolidation therapy in this group of patients.

The patients with high risk disease in the German study, they are going to move on to early transplant. So, these are the first studies, and they are going on in Europe, where MRD-based interventions are being done.

Slide 13:

The U.S. trials right now, we are mostly using it to evaluate the efficacy of our therapies. We are just about to embark on a study in CALGB where we are going to evaluate the efficacy of Campath 1H as a post-remission module in terms of modulating minimal residual disease.

CALGB and SWOG have a combined intergroup study that has just started to evaluate the effect of Gleevec on MRD during allo and auto transplant for PH+ ALL, and the pediatric groups are about to embark, I think -- and we can talk more this afternoon -- on these two studies, where MRD will be used as an end point.

Slide 14:

TOP[There is no text associated with this slide.]

Slide 15:

These are some of the questions I think we need to address, and we will go over this in more detail in the afternoon.

Should MRD be used to treatment assignment, intensification therapy for the high MRD patients, reduction of therapy, perhaps, for the MRD negatives, what methods, what level is important for intervention and can MRD be used as a surrogate marker for comparing therapies. These are just some of the questions.

Slide 16:

DR. RADICH: Good morning. I have been asked to say a few words on minimal residual disease in bone marrow transplant. Minimal disease in bone marrow transplant is bad. Thank you.

Compared to the adult literature, and the pediatric literature and conventional chemotherapy, in transplants, there is really only a handful of papers. This is going to be pretty quick.

The summary is basically this. Residual disease at the time of transplant is important. So, if you actually look at patients who come to transplant and measure their burden of residual disease, it actually makes an impact on how they do afterwards.

So, if they have no residual disease, that is best, and in those patients you can get a 60 to 70 percent disease free survival.

If you have some, you kind of get mediocre, and if you have lots, it is poor, and that is by either PCR assays or by flow assays.

Residual disease after transplant very strongly predicts relapse. So, if you give a qualitative test, usually the relative risk of undergoing relapse subsequently is five to 10 fold, and in general, the time of residual disease detection to relapse is somewhere between 30 and 90 days.

There is some data that, if you look at PCR positive ALL, p190 and p210 are a little bit different disease, wherein the relative risk associated with p190 detection is about a 10-fold risk of relapse, with a relative risk of p210 about a ten-fold.

Slide 17:

Just to kind of go through this quickly, this is one paper looking at qualitative assessment of PCR post-transplant in about 70 to 80 positive ALLs, which shows that GVHD makes a slight impact, a modest effect on subsequent relapse. In the relative risk here, being PCR positive for any kind of break point, you get about a 400 fold increase of risk compared to patients who are PCR negative throughout their transplant.

This P value ends up being P .00013. When you do a multivariate analysis, basically the only thing that really makes a difference is your PCR status.

Slide 18:

This is looking at patients pre-transplant now and determining whether your p190 or p210 types of bcr-abl at pre-transplant makes a difference. It does. If you compare the patients who had no residual disease, this is going to be people who, at pre-transplant have both p190 and p210. Their relative risk is five. If they only had p210, their relative risk is 2.9. Again, I don't know where these figures went to, but it shows that if you are p190 and have qualitative positive PCR after transplant, your relative risk is nine, p210, two, and if you expressed both, five.

Slide 19:

This looks at minimal residual disease detection both at the beginning of transplant and after. So, these are patients who are assessed just before the transplant regimen.

Patients who are in clinical relapse do badly. People who are in complete remission but are negative for minimal residual disease do pretty well. People who are in complete remission but have detection of bcr-abl do sort of in the middle.

So, the actual amount of leukemia burden is really very, very critical in these patients. This is just the detection of residual disease after transplant if you are positive versus if you are negative. So, you definitely don't want to be positive.

Slide 20:

Things just to discuss at our next session, because we are running out of time here is, can you even devise any trials to intervene on minimal residual disease after transplant.

Are there easier ways to quantitate minimal residual disease, and is there a way to detect new markers?

Slide 21:

For treating residual disease -- we should just do this in the breakout session, but the only thing I will definitively say is there is lots of evidence probably shouldn't be offered in cases.

You can talk about decreasing immunosuppression, interferon therapy, especially in the Philadelphia positive cases. That works in CML. Imaginative therapy and possibilities of chemotherapy. Somehow we have lost the rest of the slides.

Slide 22:

One of the other ways that you might think about detecting residual disease faster would be to come up with some sort of new system for diagnostics.

Right now, you have to screen for translocations, if you are doing VDJ rearrangements. You have to do consensus primers or do a number of PCRs to look at what families of V and D genes are used.

So, one of the things that we have been playing around for fun -- and this is a very loose definition of what is fun -- is developing leukemia chips, where you can actually build oligos with specific translocations, put them on a chip and then imagine, when a patient comes in for diagnosis, labeling the mRNA, hybridizing on the chip and sort of bypassing cytogenetics and figure out what translocation they have.

You can imagine doing the same thing with V and D families. You can just basically array all the possible combinations of V and D utilizations, hybridize and basically know very quickly what rearrangement you have to go fishing for. That would be something that would be fun for people to get involved in.

Slide 23:

We could use arrays to pick up genes that might be novel targets for minimal residual disease. Much like WT1 now has been found to be a pretty good marker for AML, you might imagine trying to find genes that are up in ALL, but really not up in normal bone marrow or peripheral blood.

So, we just sort of did an in silico experiment looking at Todd Golub's original data set using the AFI gene sets that have like 3,000 genes, and compared the pediatric AML or ALL expression to several pools representing 22 bone marrows.

When you do that, when you ask the question, do I want to look at genes that are completely off in bone marrow but are up in leukemias, we actually find a subset of genes. So, out of the 3,000, there were eight genes that were way up in AML but way off in bone marrow, 12 genes in ALL, and there are 24 that are up in both AML and ALL.

This was a bar gram showing just an example of mucin, which is way up in ALL and AML, but not up at all in bone marrow.

Out of this we found a number of genes, now, when we validate by quantitative PCR that might be interesting targets. That is kind of with the old data.

We, like many other people, are doing lots of arrays in ALL. One of the things you can do is, once you have got the data down, you can then go and compare them to many different groups.

So, what this was going to show is, imagine now, using arrays that use 25,000 genes, if you have a number, a library, essentially, of ALL cases, depending on what controls you are using, you can actually start making interesting comparisons.

So, we have now constructed controls of normal bone marrow cells, normal CD34 cells, normal lymphocyte populations.

You can imagine going from here is your ALL expression, you basically do the Venn diagram of ALL and bone marrow, and then, with CD19 and 22 cells, you are looking for genes that are only expressed in ALL but not in those other sets, to try to find the new markers that might be a marker for flow, or might be minimal residual disease by PCR.

If you are starting about 25,000 genes, you get down to a small set in a hurry, but that is what you really want.

You really want something -- if you are one post doc with one gene model, you really need to get down to 10 or 20 genes. So, we can talk about that in the breakout sessions, and I will bring my own computer, which I know works. Thank.