SLIDES & TRANSCRIPTS
Monday, May 12, 2003

COG Classification for Risk-Adapted Therapy

Kirk Schultz, M.D.

Jeanette Pullen and myself, as well as Mike Borowitz, Nyla Heerema, Harlan Sather and Jonathan Schuster formed a subcommittee, where we looked at designing a risk group classification based on the databases of the two centers, the two groups.

We wanted to design a group system that was best for patient care, amenable to asking biological therapeutic questions and flexible, so that it would be biology driven.

TOP

We went by a few principles in looking at the two databases, putting them together. The first one was, if we had a strong correlation between both the CCG and POG databases, we would use that as one of the risk classification criteria.

The second one was a strong correlation with one group, but limited support by the other group's data. We based our decision based on that data plus the published literature.

TOP

When we saw a strong correlation in one group and a negative correlation in the other group based on their database, we delayed a decision in utilizing that in the risk classification system that I will present to you.

Last, we felt that there were going to be some groups, such as T ALL and infant ALL that would require focused therapies and employ different risk criteria.

TOP

The classification system I will present to you primarily focuses on B precursor ALL. We combined the NCI risk classification for age and white count, used lab defined characteristics of leukemia cells, and we also looked at some of the cytogenetic criteria as well.

Then we wanted to go with four different classifications. So, we skimmed off, basically, a very high risk group and then a low risk group and then, based on what was left over, a standard and high risk.

T cell was left on its own, because we had a very hard time finding any prognostic criteria.

TOP

Based on the POG and CCG experience, we had identified the risk groups for the NCI. There is a number of trisomies that we are interested in, TEL AML1 and then some adverse translocations and hypodiploidy. Then, lastly we used the CCG experience with rapidity of response.

TOP

The first group that we did analysis to try to identify was a very high risk group. We defined that group as one that had less than a 45 percent five-year EFS.

This was based on the assumption that we wanted a 15 percent improvement in lower survival than that which you would receive with BMT and first complete remission, since that seemed to be what everybody thought was the approach that one would use with that population. So, therefore, it was basically 60 minus 15 percent.

TOP

We identified three criteria which fit, based on the CCG and POG criteria, and this was out of the Philadelphia chromosomes, to no one's surprise.

Hypodiploidy, defined as 44 or less chromosomes, and last were induction failures, and we defined induction failures as two ways.

One was if they had an M3 bone marrow 25 percent or greater at day 29, or they were at M2 at the end of induction followed by M2 which is greater than five percent, at a second time point, which was either after two weeks based on POG, or at the end of the second four weeks of therapy for CCG.

TOP

We also, then, having already defined a very high risk group, went over to the other side of the spectrum and defined a low risk group.

This group we defined as 90 percent expected five year EFS or greater, and again, based on the studies and on published literature.

There were two groups that ended up coming out of that. I will say that the TEL AML1 data, particularly, was limited at the time we were doing these analyses in 2000, 2002 because the protocols that we were evaluating then on both the POG and CGG side were still early in their accrual.

TOP

The two groups that we defined were trisomies of 4, 10, 17, a combination of the three. Further analysis may actually allow us to exclude one of those trisomies, but based on putting the two databases together, the general consensus was that the three trisomies together defined a good risk group and then TEL AML1.

Therefore, we would take a patient who was a standard risk patient and promote or put them onto the low risk, downgrade them to the low risk group.

TOP

We also defined this group by further exclusion of some of the high risk criteria, and that was, they had to have a rapid early response defined by the CCG type of criteria at day 15, no adverse translocations and no CNS or testicular disease.

TOP

Having established both the low and a very high risk group, we then moved on to define our high risk and standard risk group.

The high-risk group was defined as the group that had a standard risk that was a slow responder, or had minimal residual disease or had the MLL rearrangement. Again, if they met the very high risk criteria, they did not meet the high risk.

Then the standard risk group was essentially everybody that was left over,

TOP

and it represented a majority of patients, and they could not have any of the very high-risk criteria, or be a slow responder, which would move them into the high-risk group.

TOP

Going back, then, and taking the most recent trials, we applied the system to see how well it classified patients. This is based on the POG AlinC #16 data.

You can see the low risk group defined as a very good prognostic group, and are very high risk. So, the proof came out in the pudding, so to speak.

TOP

Then the CCG data on the most recent series, the 1950 to 1960 series, again, we saw an incremental change in outcome for those populations.

TOP

This has resulted now in a prospective classification system, based on biology, and I will briefly go through that before I finish.

At this point in time, we would have suspected leukemia that comes in. There is a consent for collection of extra marrow.

Then, at that time, at the local institution, the marrow aspirate, immunophenotyping and cytogenetics is performed.

The reference laboratories, which includes Mike Borowitz and Cheryl Willman, as well as Stephen Qualman and I am blanking on the person in Seattle, but another person immunophenotyping in Seattle, there will be immunophenotyping, DNA index, FISH for trisomies, RT PCR performed, MRD, cell banking and host polymorphisms, as part of the central lab review.

Based on that, then, we have divided the patients into high risk or standard risk, as a T cell group, a mature and an infant, and this is at the time of entry.

TOP

At the end of 29 days, we then have patients being assigned in that standard or high risk group into four groups, and that is, we divide off the very high risk and the low risk group, so they have come down and they have gone one way or another in one of four groups, with the T cells, infants, being excluded.

TOP

Just briefly, for the standard risk, they would come down and they meet the NCI's criteria, and then they are divided as the low risk and, based on those criteria, into the low risk protocol.

Then the ones that meet the standard risk will go into a non-random assignment, if they have some of the high risk features, considered as a high risk patient, or onto the regular randomized standard arm.

TOP

Then, the high risk patients, the NCI high risk group, they go onto the randomized trial, unless they have certain high risk features, which places them onto a non-random assignment.

TOP

This is just a summary of the laboratory evaluations, again, that are done at the time of induction for the B cell precursors with all of these evaluations, polymorphisms, cell banking, the high risk patients, infants, T cells, and then the mature B cells or the Burkitts. That is it.

TOP