ICML 2021 | Developing new prognostic scores in Hodgkin Lymphoma

What we have done is that we know that clearly, metabolic tumor volume is a really good marker of disease burden at diagnosis in Hodgkin lymphoma, and that it predicts prognosis very well. So what we wanted to do was to try and combine PET characteristics, including metabolic tumor volume and the traditional clinical factors that we use to assess prognosis, and see if we could come up with a new prognostic score that might improve on what we currently use, which is the IPS to select patients’ therapy for first-line treatment.

So what we did was we looked at various PET characteristics and the components of the IPS, but also we looked at bulk B symptoms and performance status, and put these all together to try and come up with a model or a new score that would predict treatment failure for patients first line and we examined this in the UK RATHL patients. I think we had about 780 patients.

And the model that we came up with that predicted for both a positive interim PET scan, relapse from a Hodgkin lymphoma after a negative PET scan, or death from Hodgkin lymphoma during the first two cycles, which we called a Hodgkin lymphoma event. The three factors that came out as being prognostic in our model were the metabolic tumor volume, the lymphocytes as a percentage of the white cell count, and also stage four disease. What we decided was we looked at various cut points first of all in our data to work out what we thought would be acceptable risks for giving patients more intensive therapy at diagnosis than ABVD and we decided on that basis to put the 30% of patients with the highest score into a high risk group, and the 70% of patients with the lower scores into our standard risk group.

And using this particular cutoff, we found that the chances of having a Hodgkin lymphoma event were 2.4 times greater in the high risk compared to the standard risk group. So then we thought, well, we need to compare this with the current standard, which is the IPS. And we found that using this new score, it was more sensitive, it was more specific, and it correctly assigned more patients to the high risk group than using the IPS. So we were pretty pleased with that, but obviously that’s only in our own data. So what we then did was we reached out to colleagues in the Southwestern Oncology Group in the U.S., and we evaluated the score and it’s association with a Hodgkin lymphoma event in the patients that they had included in their 0816 trial, which had a really similar design to the RATHL study.

And gratifyingly, I can tell you that it did validate. If we consider patients, for example, in the standard risk group, then in the RATHL data, they have a five-year progression-free survival around about 85%. And what’s interesting about this is that this is the same as the five-year progression-free survival reported in the AHL 2011 study, where patients instead of being started on ABVD and then deescalated to AVD or escalated to BEACOPP, which is what happened in the RATHL study.

In the AHL 2011 study, it’s the other way around. In the PET driven arm, patients received escalated BEACOPP as first-line therapy, carried on if they had a PET positive interim scan or were deescalated to ABVD treatment on the basis of a PET negative scan. So it seems that at the moment that the prognosis using this new score in the standard risk group is very similar to using the alternative approach of starting with first line escalated BEACOPP.

So we’re quite excited about this, because we think this new prognostic score, which incorporates a metabolic tumor volume or PET imaging could potentially identify patients in whom more intensive therapy than ABVD should be used at diagnosis whilst avoiding escalation for patients who have a high chance of cure with ABVD alone. The other thing that’s exciting about this score is that it is a continuous score. And so the cut point that you use to divide patients into high and standard risk groups could be changed depending on the proposed therapy and the risk reward balance for over or under treatment and it could, of course, be personalized for the individual patient depending on their individual risk factors.