Current Treatments Do Not Improve the Prognosis of Patients with Atypical CML and Unclassified MDS/MPN. a Joint Report from Fi-LMC, FIM, Gfch and GFM

Background

Atypical CML (aCML) is a rare myeloid neoplasm with molecular heterogeneity and overlapping features of MDS and MPN. Distinction from unclassified MDS/MPNu based on WHO criteria remains difficult, and the management of these closely related entities remains ill-codified. Most patients (pts) are managed with cytoreductive agents, but small series have reported responses to hypomethylating agents or tyrosine kinase inhibitors (TKI), notably ruxolitinib. Allogeneic stem cell transplantation (SCT) is considered the only curative option. To instruct clinical trials with novel agents in this rare and heterogeneous population, real-life cohorts must i) provide prognostic factors and molecular characterization able to stratify patients, and ii) benchmark outcomes with current treatment options.

Methods

The French National observational study of rare MDS/MPNs performed a retrospective analysis of adult patients with MDS/MPN from 35 centers. Cases were centrally reclassified according to the 2016 WHO criteria to exclude CMML, classical MPNs and CNL. All statistical analyses were done without dichotomizing continuous clinical or biological variables. The prognostic influence of treatments was analyzed considering onset of treatment as a time-dependent covariate (Mantel-Byar method).

Results

The study population included 136 pts (M/F 83/53), with a median age of 72 years. Only 43 (31.6%) met WHO 2016 criteria for aCML while the remaining 93 were classified as MDS/MPNu. At diagnosis, spleen enlargement or other tumor symptoms were present in 36% of pts, while 32% had constitutional symptoms.

Mutations in ASXL1, splice genes (U2AF1, SF3B1, SRSF2 or ZRSR2), SETBP1, EZH2, CSF3R, JAK2 and ETNK1 were present in 68.8%, 50.0%, 30.3%, 15.9%, 12.7%, 12.6% and 7.4% of 63 tested cases, respectively.

25 pts had an AML transformation. With a median follow-up of 29.8 months (0.5-276.4) median overall survival (OS) and AML-free survival (AMLFS) were 25.6 and 20.6 months, resp. Median OS was 20.2 versus (vs) 29.7 months in aCML vs MDS/MPNu, resp (log rank test p=0.2) and median AMLFS was 16.6 vs 27.4 months, resp (p=0.09).

In univariate analysis, higher WBC (p=0.003) and lower Hb level (p<10-5) predicted significantly shorter OS, while presence of splenomegaly or other tumor symptoms (p=0.08), higher proportion of IMC (p=0.06), lower platelet count (p=0.08) and dyserythropoieisis (p=0.05) tended to shorten OS. Age, gender, presence of constitutional symptoms, bone marrow or peripheral blast percentage or dysmegakaryopoiesis had no significant impact (all p>0.1). Patients with EZH2 mutations had shorter OS (median 9.9 vs 20.6 months in EZH2 wt pts, p=0.03), while other gene mutations had no significant prognostic impact (all p>0.1). All variables with p <0.1 in univariate analyses were included in a multivariate Cox model. After backward selection, only Hb levels (HR= 0.81, p<10-5) and dyserythropoiesis (HR=2.4, p=0.04) retained independent prognostic value.

At any time during follow-up, 89 (65.4%) pts received cytoreductive agents (mostly hydroxyurea), while 23 (16.9%) received TKI (ruxolitinib in 12, imatinib in 8, dasatinib in 3 pts) and 18 (13.2%) HMA, resp. Of note, 8/18 pts received HMA at transformation to AML, while all pts received TKIs prior to AML. Finally, 19 pts (14.0%) received an SCT and 35.3% received ≥2 of the above-mentioned types of treatment. Median time to SCT, HMA and TKI were 10.2, 11.9 and 4.7 months, resp.

In multivariate Cox models adjusting for baseline Hb level and dyserythropoieisis, neither cytoreduction (p=0.6) nor SCT (p=0.5) were associated with a significant OS improvement. Treatment with HMA was associated with a significantly worse OS (HR=3.0, p=0.001), but this effect was confounded when transformation to AML, as a time-dependent event, was included in the model (HMA: HR= 1.1, p=0.7, AML: HR=5.1, p<10-6). Finally, treatment with TKIs was also associated with significantly worse OS (HR=2.3, p=0.02). This TKI's detrimental effect was confirmed in a model also accounting for baseline WHO, (log-transformed) WBC count, platelet levels and bone marrow blast percentage.

Conclusions

Anemia and dyserythropoiesis are important prognostic factors in aCML and MDS/MPNu that should be used to stratify pts included in clinical trials. There is currently no standard of care for these overlap syndromes. Treatment with HMAs and TKIs should be restricted to clinical trials.