A Simple Prediction Model of Outcomes after Allogeneic Hematopoietic Stem Cell Transplant (HCT) in Myelodysplastic Syndromes Using HCT-Comorbidity Index, Cytogenetic Risk, and Platelet Count

Introduction: Allogeneic hematopoietic stem cell transplant (HCT) is currently the only potentially curative option for patients (pts) with myelodysplastic syndromes (MDS). However, many pts relapse after HCT. While the International Prognostic Scoring System-Molecular (IPSS-M) is useful, next generation sequencing is not always available particularly in resource limited settings. Hence, we aimed to develop a prediction model utilizing readily available pt and HCT characteristics of post-HCT outcomes in pts with MDS treated with hypomethylating agent (HMA) therapy pre-HCT.

Methods: Pts with MDS treated with HMA (azacitidine or decitabine) who received an HCT at Dana-Farber Cancer Institute from 01/2014-12/2020 were included. The primary outcome was progression free survival (PFS) post-HCT, defined as time from transplant to disease relapse or progression. Univariable analysis (UVA) and multivariable analysis (MVA) were used to identify predictors of PFS and build a model. Variables were pt age, sex, donor type (matched related donor (MRD) vs. matched unrelated donor (MUD) vs. 7/8 MUD vs. haploidentical), HCT-comorbidity index (HCT-CI), CMV status of donor/recipient, MDS subtype, graft versus host disease prophylaxis (GVHD ppx), HMA type, number of HMA cycles, conditioning intensity (myeloablative vs. reduced intensity), cell source (bone marrow vs. peripheral blood), response to HMA assessed by IWG 2023 criteria (Zeidan Blood 2023), IPSS-Revised (IPSS-R) pre-HCT, and cytogenetic risk (very good, good, intermediate, poor, very poor) at diagnosis and pre-HCT.

Results: A total of 148 pts with MDS received HMA and underwent subsequent HCT. Median age was 64 years (range 26-79) and 61.5% were men. Pts were treated with 7 days azacitidine (54.1%), 5 days decitabine (39.2%) and other schedules (6.7%) and received a median of 4 HMA cycles (range 1-20). Prior to HCT, IWG 2023 responses were: CR (15.5%), CRbi (14.9%), CRuni (19.6%), CRh (0.7%), PR (1.4%), HI (9.5%), and no response (38.5%). Most donors were MUD (60.8%) or MRD (18.2%), and most pts received reduced intensity conditioning (74.3%). GVHD ppx was tacrolimus (tac)/methotrexate (MTX) (54.7%), tac/MTX/sirolimus (19.6%), cyclophosphamide/mycophenolate mofetil/tac (15.5%), and other (10.1%). Median HCT-CI was 2 (range 0-13); 48.6% had HCT-CI ≥3. Median follow-up time among ongoing survivors was 48.3 months (range 5.5-101.3). The only statistically significant variables associated with post-HCT PFS in both UVA and MVA (with Cox MVA results shared here) were HCT-CI (≥4 vs. 0-3: HR 1.94, 95% CI 1.18-3.19, p=0.009), platelet count pre-HCT (platelets <40 vs. >70: HR 2.46, 95% CI 1.33-4.54, p=0.004; platelets 40-70 vs. >70: HR 1.84, 95% CI 1.02-3.32, p=0.044) and cytogenetic risk pre-HCT (very poor vs. other: HR 2.52, 95% CI 1.43-4.44, p=0.001). HCT-CI impacted non-relapse mortality (NRM) and not relapse whereas very poor cytogenetics impacted relapse and not NRM. A model predicting post-HCT outcomes was built with the following risk scores: HCT-CI: 0-3 (0), ≥4 (1.5); platelet count pre-HCT: >70 (0), 40-70 (1), <40 (2); cytogenetic risk pre-HCT: not very poor (0), very poor (2). Based on the sum risk score, patients were defined as low (0-1), intermediate (int) (1.5-2) or high (≥2.5) risk. The risk classification was significantly associated with post-HCT outcomes. Four-year post-HCT PFS was 68% (95% CI 56-78%), 36% (95% CI 22-51%) and 15% (95% CI 5-31%) for low, int and high risk, respectively (p<0.00001) (Figure A). Accordingly, 4-year cumulative incidence of relapse was 23% (95% CI 14-33%) for low, 54% for int (95% CI 38-68%) and 62% for high (95%CI 39-78%) risk scores (p=0.00005). Four-year overall survival was 70% (95% CI 58-80%), 45% (95% CI 29-59%), and 19% (95% CI 7-36%) (p<0.00001), for the same respective risk groups (Figure B). In contrast, NRM did not differ between risk categories (low: 9%, int: 10%, high: 23%; p=0.1). The predictability of this simple risk model (C-index: 0.72) was found to be superior to IPSS-R pre-HCT (C-index: 0.61) and IWG 2023 response pre-HCT (C-index: 0.617).

Conclusion: We developed a simplified prognostic model of post-HCT outcomes for pts with MDS treated with HMA based on HCT-CI, platelet count, and cytogenetic risk post-HMA therapy and pre-HCT. We expect that this model will be particularly useful in resource-limited settings without easy access to next generation sequencing.