DR. COBRO: Jay Cobro. The only thing is, when you design this study, you have got to have phenomenal numbers to make sure you are able to analyze by each one of these subgroups, if you do it prospectively, which no study has such a design. Usually it is retrospective. When you define these subgroups, it is a phenomenal task.

DR. SCHULTZ: Yes, and the COG will accrue 1,900 ALLs per year, which should allow the numbers that you are suggesting to prospectively analyze the features and modify.

DR. COBRO: You need about 5,000 patients.

DR. SCHULTZ: Per year?

DR. LARSON: How similar is this risk stratification to what the Europeans are currently using?

DR. SCHULTZ: It depends on what you mean by the Europeans. If you look at the BF classification, a lot of it is based on giving prednisone for seven days before the patient is placed onto the standard induction.

So, already, the premise of how you assign risk is quite different from any of the North American studies. So, things are different, but a number of the factors are the same. For instance, their very high risk features, excluding steroid response, is essentially the same.

DR. LARSON: Do you think, at the end of the day, you will be able to compare your outcomes with other large international groups?

DR. SCHULTZ: I do, although I would like to think the level of sophistication here is good, and there may be another layer there.

DR. VAN DER JAGT: Have you considered looking at a response adapted type therapy as opposed to a risk adapted therapy?

DR. SCHULTZ: That is probably the next generation. Certainly, MRD studies are being done, as you saw in earlier presentations, in all of our prospective studies.

So, MRD as part of that -- already we are doing response because of the CCG-based day 15 bone marrow and that, already, we are assigning risk based on that.

We hope for a higher level of sophistication by using MRD evaluations, which are being done by -- probably immunophenotypically is the best way when you are trying to do it early on, and you don't need something that sensitive.

PARTICIPANT: When you tried to apply your risk model to this subsequent study, I wasn't impressed with the differences except for the very high risk group, which is an obvious one.

DR. SCHULTZ: You have to also remember that the high risk and standard risk patients, there are differences there, but they have already had some adjustment of their therapy. Especially on CCG, if they are a slow responder or a high responder, they are put into the higher risk groups.

There is an incremental change, but it is not 20 percent. It is more in the 10 percent range.

DR. ANDREEFF: An ideal classification would identify targets, I believe. Is there any correlation to biology that can be exploited?

DR. SCHULTZ: Hopefully. I think the low risk group in particular, I am hoping that that would be. Certainly with TEL AML1, if targets come up -- and I don't know if anything is specifically a target directed study, in a very high-risk study, Gleevec is being evaluated, and maybe a farnesyl transferase inhibitor or something else like that might be as well.

Certainly, for instance, in the T cell protocol, which is not presented here, and we haven't been able to identify good risk classification except to say there is T cell, the AraG prodrug Compound 506U78 is being evaluated.

So, if you have something that is targeted therapeutics, it will be, and we hope we have built something flexible enough to allow us to do that.

TOP