Komrokji Examines Differences Between the WHO and ICC MDS Criteria

Although the current classification systems for patients with myelodysplastic syndromes (MDS) were able to identify groups of patients with similar outcomes, further refinement is needed to create more consistent criteria, according to findings from an international dataset analysis of the 2 classification systems presented by Rami Komrokji, MD, during the 2023 EHA Congress.

“It’s clear that we are moving to a new era in the classification of MDS and that we have molecularly defined groups; those patients should be thought of, treated, and enrolled in clinical trials separately,” Komrokji said. “Then we have the morphologically defined groups, but we definitely need to step up, try to harmonize those [criteria], and go back into 1 classification that hematopathologists, clinicians, and clinical trialists use as a common language rather than having 2 different classifications.”

The analysis, which was conducted on behalf of the International Consortium for MDS, used datasets from Moffitt Cancer Center and the European initiative GenoMed4all:GM encompassing 7017 patients with available molecular data to validate and compare the World Health Organization (WHO) and International Consensus Classification (ICC) 2022 classification criteria. Results showed that patients with deletion 5q (del5q) in the WHO (n = 107) and ICC groups (n = 108) from the Moffitt cohort both achieved an overall survival (OS) of 75.6 months vs 82.1 months for those from GenoMed4all:GM with del5q in the WHO (n = 219) and ICC groups (n = 223), respectively.

Patients with SF3B1 mutations, which represent approximately 12% of all MDS cases, had the best outcomes compared with those with del5q and TP53-mutated disease. These patients in the WHO (n = 294) and ICC groups (n = 277) from the Moffitt cohort experienced an OS of 101.8 months and 111.6 months vs 104.9 months and 101.9 months in the GenoMed4all:GM WHO (n = 654) and ICC (n = 594) groups, respectively.

In an interview with OncLive^®, Komrokji, the vice chair of the Malignant Hematology Department and the head of the Leukemia and MDS Section at Moffitt Cancer Center in Tampa, Florida, discussed the comparative analysis as well as the next steps being taken to develop a more harmonized classification system.

OncLive: What was the rationale behind the analysis presented at the 2023 EHA Congress?

Komrokji: The classification for MDS has been evolving over the years, from FAB [French American British] in the 1980s, to different iterations from the WHO, but [for] the first time, in 2022 we have 2 classification systems: the WHO and the ICC. Both of them are an attempt to improve the classification for MDS, but also it creates a dilemma in the field and controversy regarding which one to follow. There are some differences [between the classification systems].

It was important to generate data-driven evidence to hopefully guide future harmonization of those 2 classifications. People are mixing classification vs prognostic models. Risk stratification in MDS is based on [the] Revised International Prognostic Scoring System [IPSS-R]. The classification historically is pathological classification [and] it should reflect a unique biology of the disease [as well as] unique groups that have a certain clinical phenotype [where] maybe the pathogenesis of defective hematopoiesis is similar in that group. We were trying to validate those and hopefully this is the first step in trying to come up with harmonization and [return to] 1 classification [system] in MDS.

How was the analysis conducted?

When the [ICC 2022] classification came out we looked at our database—we have a large database with more than 2500 patients fully annotated—and then we approached our colleagues in Europe. They kindly joined us with the GenoMed4all:GM MDS database, which is the MDS database for almost the whole [of] Europe.

Together, we were able to put approximately 7000 patients with MDS [into our platform], which is a very large cohort. Hopefully [it] will be a platform not only for this project, but for several other projects down the road. We reclassified all the patients according to the WHO and ICC [criteria], included only patients with fully annotated molecular data, and we looked at the questions that we wanted to address.

What were the key findings of the analysis?

The 2 classifications recognize molecularly-defined categories of MDS, namely SF3B1, del5q, as well as P53. There are slight differences in the definitions between both of them. For example, we demonstrated that SF3B1 is a unique category, approximately 12% to 13% of patients with MDS have very favorable outcomes with a median OS of more than 10 years [and] low chance of [dying from] leukemia.

We also looked at a subset of patients [with] MDS with ring sideroblasts, which is usually the phenotype we see with SF3B1, but [we examined patients] if they were SF3B1 wild-type. That’s a smaller subset, approximately only 4% of patients, but they did not have the same favorable outcomes. The WHO retained [classification of] MDS with ring sideroblasts, which is probably something that could go away down the road in the future based on this. However, we have to think of those models in terms of generalizing them. In countries where there is no access to SF3B1 [testing], ring sideroblasts are still a good marker for SF3B1.

We also confirmed that [the patient population with] del5q is a favorable group, [constituting] approximately 5% of patients [with MDS]. Bi-allelic TP53 [is found in] approximately 10% of patients [with MDS] and unfortunately they have very poor outcomes. The ICC proposed a different group of P53, monoallelic. If the blasts were more than 10%, they would consider that MDS/AML with P53. Those patients had worse outcomes and part of it is driven by the increased blast [percentage]. In the future, that category may go away, and we’ll probably stick more to the bi-allelic [characterization of] P53.

In the morphologically defined [criteria], we showed that MDS ring sideroblasts without SF3B1 [mutations] are not different. The ICC proposed that the number of lineage dysplasia, whether we have multilineage vs single lineage dysplasia, does matter [and we showed that] multilineage dysplasia did worse.

Then we tried to tackle the issue of the blast [count] and that’s probably one of the most difficult [as] many of those cut offs are arbitrary. It’s clear when the myeloblasts are more than 5% that patients have worse outcome. In the Moffitt database there was no difference in the survival, [but in] the European database there was.

Both of them showed difference in the leukemia-free survival. Part of this could be explained by the modalities of treatment where there were more transplants in Europe and more [use of] hypomethylating agents.

The message is that increased blasts or myeloblasts is not good. Where’s the exact cutoff? The cutoff between MDS/AML is not clear. The ICC proposed this MDS/AML category, 10% blast or more, and that’s also something that we’ll have to come to a consensus on because it has some advantages [and disadvantages as well]. It allows the continuation with AML, may allow access to drugs and trials, but it has not been validated in a systemic fashion. There are molecular subentities in that group. Do you treat all those patients [as having] AML––which is not always the right answer––or do you treat them as having MDS? And which response criteria [do you] use [with] that group? Based on that, we are proposing hopefully a harmonization.

What are some of the next steps for this research?

We have several next steps, [including] performing things such as hierarchical clustering, to see where those subgroups reside in a gene or cytogenetic mapping, and then coming up with more [of a] consensus on the nomenclature for those groups.

We have to think of this in a conceptual way [as we have 3 conceptual classifications of MDS]. We have a chronic phase, less aggressive MDS, [and] those are the patients that don’t have excessive blasts, may have hypoplastic MDS that could have therapeutic application, [and have] del5q/SF3B1.Then we have a group that’s almost like an accelerated MDS phase, or [forms] that are more aggressive when the blasts are increased [and/or] when there is biallelic TP53. From the data, we were able to show that presence of fibrosis in the bone marrow is also in that group. Finally, we have the [difference between] AML and MDS, whether that be by prior history of MDS [with] genetic defining [or] cytogenetic defining for MDS. That blast cutoff is yet to be determined.

[Looking ahead], there are several plans and proposals coming from the people involved. We are going to [take] at deeper dive into some categories, like P53. We’re going look at the IPSS-M [Molecular International Prognostic Scoring System] in the context of the classifications. We could start looking into the implications of treatments among each of those categories and validating other proposals.

Looking back at the EHA, what data stood out to you?

For the first time in 1 meeting, we had 2 positive phase 3 clinical trials [for patients with] lower-risk MDS. At the plenary session we heard the [phase 3] COMMANDS [trial NCT03682536] data [which will] probably [result in] moving luspatercept-aamt [Reblozyl; to] the upfront treatment setting. We also heard about imetelstat which [proved to be] a very active drug in a positive phase 3 trial, that hopefully will be available for our patients by next year. That’s very encouraging.

For higher-risk MDS, we heard a little bit of a phase 2 trial [investigating] a total oral therapy [regimen] using oral decitabine with oral venetoclax [Venclexta]. It is an early study, but it’s promising for the future and maybe we’ll be able to move to total oral therapy in that group of patients.

Reference

Komrokji RS, Ball S, Maggioni G, et al. Myelodysplastic neoplasms (MDS) classification from WHO 2017 to WHO 2022 and ICC 2022: an expanded analysis of 7017 patients on behalf of the international consortium for MDS (ICMDS). HemaSphere. 2023;7(suppl 3):e616260d. doi:10.1097/01.HS9.0000967592.61626.0d