Does stringent cytoreduction improve survival in myeloproliferative chronic myelomonocytic leukemia? a correlative study on the dacota trial.

Abstract Background. Myeloproliferation at diagnosis is an adverse prognostic factor in CMML. Cytoreduction with hydroxyurea (HY) is considered palliative but decitabine (DAC) failed to improve survival over HY in advanced proliferative CMML in the randomized DACOTA trial (PMID 36455187). Whether resolution of myeloproliferation improves prognosis independently of bone marrow (BM) response is unknown. Multiparameter flow cytometry (MFC) reveals pathognomonic accumulation of CD14+/CD16- classical monocytes (cMo,) and variably, of CD45+CD66b+CD15+CD14- immature granulocytes (iGRAN) at CMML diagnosis. These flow-defined cell populations have not yet been studied as biomarkers of treatment response. Objectives. To inspect the prognostic value of WBC and absolute monocyte counts (AMC) and of MFC-defined cMo and iGRANs at the 3- and 6- cycle evaluations in the DACOTA trial (NCT02214407). Methods. MFC assessment of cMo and iGRAN was done centrally as published (25852055, 39545419). Survival analyses were done using landmarks at the 3- or 6-cycle evaluation visit. Main results of DACOTA were previously reported (36455187) Results. Of 170 patients (pts) with advanced proliferative CMML randomized in DACOTA, 120 (DAC n=63, HY n=57, M/F 83/37, median age 73y) were evaluated after 3 cycles with BM aspiration and complete blood count (CBC), including 79 with centralized MFC. Median WBC (DAC 32.7 x109/L; HY 31.2 x109/L), CPSS risk (int2/high; DAC 54%, HY 51%) and overall survival (OS) from the 3-cycle landmark (DAC 22.0 months, HY 20.1 months) were comparable between arms. Across arms, 58% and 34% had monocytes > 1 x109/L and WBC > 10 x109/L at the 3-cycle evaluation respectively (resp); 52% and 31% had AMC > 1 x109/L and WBC > 10 x109/L at the 6-cycle evaluation resp. There were marked differences between arms, with 76% HY pts having AMC > 1 x109/L after 6 cycles compared to 33% DAC pts, and 51% HY pts with WBC > 10 x109/L compared to 14% DAC pts. In univariable analyses, while differences were not significant after 3 cycles, both an AMC > 1 x109/L (p=0.017) and a WBC > 10 x109/L (p=0.00021) predicted shorter OS after 6 cycles. Adjusting for arm, CPSS risk at baseline (higher [int-2/high] risk vs else) and BM blasts ≥5% at the 3- and 6- cycle evaluations, an AMC > 1 x109/L was associated with a higher hazard of death at both landmarks (3-cycle, HR=1.83, p=0.044; 6-cycle, HR=3.46, p=0.009), whereas a WBC > 10 x109/L only significantly increased this risk at the 6-cycle evaluation (HR=4.50, p=0.0004). Across treatment arms, the 56% of pts with AMC > 1 x109/L OR WBC > 10 x109/L after 6 cycles had poorer OS (HR=5.38, p=0.0003) irrespective of treatment, CPSS risk at baseline and persistence of BM blast excess at the 6-cycle evaluation. This criterion was predictive of death in both HY (p=0.049) and DAC (p=0.014) arms. Across arms, pts with AMC > 1 x109/L OR WBC > 10 x109/L after 6 cycles had a median OS from landmark of 13.9 months vs 34.6 months for those with AMC ≤ 1 x109/L AND WBC ≤ 10 x109/L. We first integrated MFC data in the 79 pts assessed at 3 cycles into CMML-like (67%), inflammatory-CMML-like (17%) and normal-like (15%) profiles. When adjusting for treatment arm, baseline CPSS, and BM blasts at evaluation, normal-like phenotype at 3 cycles had no impact on OS. We next integrated MFC and CBC data to compute absolute cMo and iGRAN counts and dichotomized patients based on published cutoffs (25852055, 39545419). At the 3-cycle landmark, both an absolute cMo count > 0.94 x109/L (HR= 2.52, p=0.014) and an iGRAN count > 0.4 x109/L (HR=2.38, p=0.009) predicted poorer OS independent of arm, CPSS risk, and persistence of BM blasts. There was no significant interaction between treatment arm and cMo (p=1.0) or iGRAN excess (p=0.42). Median OS from the 3-cycle landmark was 15.3 months in the 72% pts with cMo > 0.94 x109/L OR iGRAN > 0.40 x109/L compared to 35.1 months in the 28% pts with cMo ≤ 0.94 x109/L AND iGRAN ≤ 0.40 x109/L (p=0.013). Conclusion. Biomarkers integrating CBC and MFC data may predict CMML prognosis irrespective of treatment after limited (3 months) duration. The improved survival of HY-treated pts achieving complete WBC and AMC control warrant randomized investigation of stringent vs liberal cytoreduction in CMML.