The background is that over the past few years there have been multiple publications about the persistence of disparity of outcome of Black patients compared to White patients with AML but in terms of what factors they are related to it has not been very clear because some publications have shown there are differences in prevalence of certain genomic abnormalities where others didn’t show those same differences. Also in terms of impact of different genomic abnormalities in AML in the different races, also there have been mixed results. But because a lot of those studies have a very limited number of patients we wanted to contribute to the data by analysing a large dataset of patients treated uniformly in a clinical trial, large [??] clinical trials at ECOG-ACRIN.

What was the study design?

It was a retrospective analysis of about 3,800 patients enrolled in clinical trials that were all interventional trials, phase II and phase III, testing different regimens in AML patients over the course of about four decades. I think they started somewhere in the 1980s and all the way through to 2019. What we did, we analysed those patients and looked at different factors, at their association with outcome. So we had a multivariable analysis that we did and all the analysis was stratified by protocol because, as you may imagine, when you have protocols involving patients over the course of decades there’s going to be variability between the different protocols. So we also stratified by protocol and we stratified by clinical characteristics that we know they have an impact on outcomes like level of anaemia, [??], age of the patient as well as blasts down in the bone marrow, in the blood, performance status, all these different things.

What were the key results?

What we have found is that for patients treated on those trials, the race base, Black-White disparity, actually persisted and Black race was an independent predictor of inferior survival. We had cytogenetics available for about two thirds of the patients and when we actually look at prevalence of different cytogenic risk groups, like favourable versus intermediate versus adverse, we didn’t see any difference in the prevalence of the risk score. We also looked at what impact the risk score had on outcome and we didn’t really see any difference in the impact in terms of overall survival.

Then we also had mutations for a selected number of patients, about 20-something percent of patients we had mutations available for them. When we looked again at prevalence of those mutations we didn’t see any difference that was with race, so the prevalence was very similar between Black and White. But what we did see is that for one mutation, NPM1 mutation, we found that Black patients who had that mutation actually had very poor survival. This mutation is typically considered a favourable risk mutation and the White patients did much better than the Black patients and the difference was statistically significant.

What do you think should be done with these findings?

There are a couple of things that I would think, off the top of my head, that can be done. Number one is for clinicians, when we are in the clinic evaluating a patient and offering them treatment, I think we should pause when we’re evaluating a Black patient and acknowledge to them that this disparity exists and we don’t fully understand the reason. So we may have to have, in terms of how we monitor, we should acknowledge the limitation of what we’re applying to them because we don’t really know why they have this disparity and also encourage patients to participate in clinical trials so we can functionally improve the available data so we can have a more diverse dataset so that can be better applicable to patients. So that’s one.

Then, on the research side, I think not only do we need to encourage participation in clinical trials but we also, as clinical trial leaders, we also should make every effort to actually minimise barriers to participation because it goes both ways. It’s like the willingness to participate is one thing but also the ability to refer patients to participate, it may be associated with how feasible it is for them. So removing or minimising barriers to participation for most under-served patients, that’s also very important because in order to move the science forward we need to have trials in populations that are representative of the general population so that the science can be better applicable to them.

Is there anything that you would like to add?

One thing I would want to add, when we were trying to understand the reason for those differences, not only did we look at the genomics, we also looked at treatment response, treatment tolerability, where we didn’t see any difference. But we also looked at different patterns of treatment and you would expect in a clinical trial each parallel treatment should be very similar. But we saw a difference in the percentage of Black patients going to allogeneic stem cell transplant compared to White patients. We also don’t know why is that, it could be because of donor availability, because usually donor selection is very strict when it comes to clinical trials for transplant on trial. So this an area of disparity that also should be addressed.