Abstract Body

The management of high-risk MDS has been mostly based on allogeneic transplantation and hypomethylating agents for almost 15 years. However, only a small proportion of patients are actually allografted, making HMA the cornerstone of treatment. Nevertheless, the limits of HMA have been recognized, since only half of the patients respond, and the median survival of 2 years remains short.

We have tried to improve these results by combining drugs with HMA. Many have been evaluated, in combination with HMAs in phase 1-2 trials, including HDACs, Imid and recently TP53-inhibitors, NEDD-inhibitors and TIM3-inhibitors among others. Most of these molecules have shown results that have been qualified as promising in non-randomized studies, based on results showing higher response rates than those expected and response duration, if not survival, that appeared promising. However, the optimism of the analyses of these early phase studies has recently been tempered by the results of randomized trials, comparing these combinations to HMA alone. Indeed, the US-study did not show any advantage to the addition of Revlimid or Vorinostat; the French-study did not show any benefit to the addition of idarubicin, lenalidomide or valproate either; more recently studies also failed to show any improvement with the addition of APR246 in TP53^m patients; or the addition of Pevonedistat or Saboatolimab compared to HMA alone in HR MDS.

One can therefore legitimately wonder about the reasons for such a series of failure and I propose here some thoughts to explain this.

1/ Of course, the inefficiency of the studied drug is a possible reason, but the repetition of these failures makes this hypothesis unlikely, at least for all of them.

2/ The toxicity of the studied drug combined with HMAs is likely, in some patients, to lead to complications that prevent continuation of treatment. This was suggested in the US-trial where, because of toxicities in the experimental arms, patients were withdrawn from the trial and therefore were possibly under-treated.

3/ From one study to another, using the usual prognostic criteria, there is a large degree of heterogeneity in the populations studied, making it difficult to replicate small phase 2 studies on a larger scale in phase 3 trials.It is hoped that new prognostic tools, especially molecular ones, can help us to better select patient candidates for a specific treatment. An example of this variability was recently given by the comparison of the populations included in the STIMULUS-MDS-1 and 2 trials: with similar inclusion criteria, the populations studied were in fact different, particularly in terms of very high-risk patients. This discrepancy might lead to different outcomes in the 2 studies.

It would therefore be useful either to better select/stratify patients using more precise inclusion criteria (IPPS-M), or, to overcome heterogeneity, to include a larger number of patients in the trials. Indeed, it is not very surprising that no survival advantage is detected in studies that include a limited number of patients, especially since the calculations of the number of patients to be enrolled are often a bit optimistic in the supposed advantages of the study drugs.

4/ Finally, one can also question the primary endpoints of the studies, which are rarely survival, but rather (supposedly) surrogate markers, such as response or EFS. While this type of endpoint may seem acceptable in early phase trials, they are more difficult to justify in registration trials.

Despite these past pitfalls, the MDS community is awaiting several major randomized studies, evaluating venetoclax, sabatolimab, magrolimab and tamibatorene. While these studies are not entirely free of the criticisms mentioned above, it is hoped that the efficacy of these treatments will be sufficient to overcome them.