Prognostic impact and timing considerations for allogeneic hematopoietic stem cell transplantation in chronic myelomonocytic leukemia

Dear Editor, Chronic myelomonocytic leukemia (CMML) is a clonal disorder of aging hematopoietic stem cells characterized by overlapping features of myeloproliferation and myelodysplasia, with a median overall survival (OS) of ≤36 months. Hypomethylating agents (HMAs) have an overall response rate of about 30–40%; however, these agents are ineffective in altering the natural disease biology due to inability to prevent acquisition of molecular abnormalities and transformation to acute myeloid leukemia (AML). Allogeneic hematopoietic stem cell transplantation (alloHCT) is a potentially curative option, with 20–50% patients achieving long-term remissions. However, as the median age of presentation is 73 years, only a fraction (<20%) of CMML patients are eligible for alloHCT. Prognostic models such as the Mayo Prognostic Model (MPM), Mayo Molecular Model (MMM), Groupe Francais des Myelodysplasies, and the CMML-specific prognostic scoring system (CPSS) are important tools to identify patients at high risk for disease progression and death. Several retrospective analyses of outcomes in alloHCT patients have identified adverse cytogenetics, blast percentage, HCT-comorbidity Index, time to alloHCT, disease control at the time of alloHCT and acute and chronic graft vs. host disease (GVHD) as factors influencing OS and AML-free survival (LFS). In the absence of randomized controlled trials, the questions of optimal timing of alloHCT in CMML, prealloHCT use of HMA vs. cytotoxic chemotherapy, and the selection of patients who should be treated upfront with alloHCT remain unanswered. We performed this study to assess the outcomes and therapeutic impact of alloHCT in patients with CMML.

In the absence of randomized controlled trials, the questions of optimal timing of alloHCT in CMML, pre-alloHCT use of HMA vs. cytotoxic chemotherapy, and the selection of patients who should be treated upfront with alloHCT remain unanswered. We performed this study to assess the outcomes and therapeutic impact of alloHCT in patients with CMML.

Study population
After Mayo Clinic Institutional Review Board approval, medical records of 406 consecutive CMML patients (age ≤ 75 years at diagnosis) from January 1990 to December 2018 were reviewed (75 years being the upper age limit for alloHCT in our institution). Disease and alloHCT-related data were retrospectively collected. Next-generation sequencing (Supplementary Table 1S) for myeloid-relevant mutations was performed on bone marrow mononuclear cells at CMML diagnosis, or at first referral (within 6 months of diagnosis). Response assessment was documented as per the 2015 International Working Group (IWG) myelodysplastic/myeloproliferative neoplasm overlap neoplasm criteria 13 . The CPSS, MPM, MMM, and the Mayo-French Models (MFMs) were employed for risk stratification. A 1:1 propensity score matching (PSM) analysis for age and MPM risk stratification (low, intermediate, and high risk) was used to determine the impact of alloHCT in patients who did and did not undergo alloHCT. High-risk cytogenetics included complex and monosomal karyotypes, and low risk included normal, sole -Y, and sole der ( Of the 46 patients transplanted in CP, 31 (67%) patients received prior therapies such as HMA (48%), AML-like induction chemotherapy (11%), or investigational agents (7%). Of the 24 patients transplanted in BT, 13 (54%) patients received prior AML-like induction and 10 (42%) received HMA followed by induction chemotherapy prior to alloHCT. There was no statistically significant difference in day 100 mortality in patients who received pretransplant HMA vs. induction therapy in both CP (p = 0.5) and BP (p = 0.1) CMML patients. Of the 70 CMML patients who received alloHCT, only 7 were untreated prior to receiving conditioning therapies. There was no difference in Kaplan-Meier estimate of median OS in untreated patients vs. those who received pretransplant cytoreduction (HMA, AML-like induction therapy, and investigational agents; log-rank test, p = 0.3). Twenty five (57%) and 20 (87%) patients transplanted in CP and BT, respectively, met criteria for complete response (CR) or optimal marrow response at the time of alloHCT. The conditioning regimens (myeloablative vs. reduced intensity) and donor sources (matched vs. mismatched, related vs. unrelated) were evenly matched in both the CP-and BT-transplanted CMML patients (Table  1). Peripheral blood stem cells were the favored donor source in both groups (85% in CP vs. 87% in BT, p = 0.76). Sixty six (94%) patients had sustained donor engraftment. Similarly, rates of acute (40% vs. 60%, p = 0.12) and chronic (65% vs. 46%, p = 0.24) GVHD were not significantly different between CMML patients transplanted in CP and BT ( Table 1). None of the transplanted patients received posttransplant HMA therapy.
Among the CP alloHCT recipients, median OS was not reached in the CPSS intermediate 1 (95% CI 3-NR) and 2 (95% CI 21-NR) risk groups, and was 12 (95% CI 2-67, p = 0.02) months in the high-risk group ( Supplementary  Fig. 6S). Likewise, post-alloHCT median OS was not reached in the intermediate-and low-risk MPM categories in CP patients, vs. a post-alloHCT median OS of 36 (95% CI 8-189) months in high-risk patients (Supplementary Fig. 7S). These posttransplant OS trends were mirrored when CP patients were risk stratified as per the MFM (Supplementary Fig. 8S).
The composite end point of median GVHD-free/ relapse-free survival was 3.5 (95% CI 2-7) and 7 (95% CI 5-21) months in the BT and CP, respectively [p = 0.02, Supplementary Table 7S]. Further, LFS and OS in the alloHCT group did not differ in CMML alloHCT patients with vs. without chronic GVHD (Supplementary Table  8S). In the transplant eligible (age ≤ 75 years) group (n = 406), 200 (49%) patients were classified as proliferative CMML, whereas 203 (50%) were classified as dysplastic CMML (information not available for 3 patients). The Kaplan-Meier estimate median OS in the proliferative subtype was significantly lower when compared to the dysplastic subtype (20 vs. 32 months, log-rank p < 0.001). In the proliferative subtype, the median OS for transplant Umbilical cord blood 2 (3) 1 (2) 1 (5) group was higher than the non-transplant group (50 vs. 19 months, log-rank p < 0.0001). In dysplastic subtype, the median OS for transplant and non-transplant group was not significantly different (41 vs. 37 months, log-rank p = 0.5). The Kaplan-Meier estimate of median OS in patients with CR or optimal pre-transplant blast % (defined as BM blast% < 5) was higher than those without CR or optimal blast% (50 vs. 27 months); however, this difference was not statistically significant (log-rank test, p = 0.2).

Propensity score matched analysis
Forty-eight patients in alloHCT and non-alloHCT groups were matched for age and MPM using 1:1 PSM analysis. Median OS in the PSM-matched alloHCT group was higher compared to non-alloHCT group [40 months, (95% CI 26-NR) vs. 23 months, (95% CI 10-37), In summary, within limitations of a retrospective analysis, our study confirms the survival benefit conferred by alloHCT in CMML, especially in CP disease. AlloHCT was able to achieve a 5-year OS of 51% in CP-CMML vs. 19% in BT-CMML, underscoring the importance of early alloHCT, especially in higher risk patients. This observation was also validated with the help of a propensity scorebased comparison (Fig. 1b). The survival advantage of alloHCT was somewhat offset by a GFRFS of only 7 months, indicating that in CMML, alloHCT can be associated with significant morbidity. We also show that intermediate to high-risk cytogenetic abnormalities by MFM are predictive of post-alloHCT relapse and inferior OS, highlighting the need for better pre-alloHCT therapies.