Redefining Clinical Response Criteria in Myelodysplastic Syndromes to Better Forecast Post-Transplant Outcomes

Background: The International Working Group (IWG) 2023 response criteria are the new standard by which treatment responses in myelodysplastic syndromes (MDS) are evaluated. However, these criteria have not yet been validated to predict outcomes in pts undergoing allogeneic hematopoietic stem cell transplantation (HCT). We set out to determine the significance of pre-transplant response to hypomethylating agent (HMA) therapy based on IWG 2023 criteria in predicting transplant outcomes in MDS. In addition, we propose a novel prognostic tool based on redefined pre-transplant-specific response criteria to better predict HCT outcomes.

Methods: We analyzed pts with higher-risk MDS who were treated with HMA followed by HCT at 2 large academic centers: Dana-Farber Cancer Institute and Moffitt Cancer Center. Pre-HCT responses were estimated from marrow specimens sampled closest to HCT and from phlebotomy specimens drawn closest to conditioning therapy. Responses evaluated by IWG 2023 criteria were defined as complete remission (CR), CR with bi-lineage (CRbi), CR with unilineage (CRuni) or CR with partial (CRh) hematological recovery. Composite CR (cCR) was defined as CR+CRbi+CRuni+CRh. The primary outcome was post-HCT progression-free survival (PFS), defined as time from HCT to relapse, progression or death.

Results: A total of 280 pts underwent HCT. 62% were male. At diagnosis, 50.7% presented with MDS IB-2, 34.3% with MDS IB-1 and ~15% with MDS-LB. The median age at HCT was 65 yrs (range 22-79). Pts received a median of 4 cycles of HMA prior to HCT. Based on baseline IPSS-M risk, 251 pts were classified as very low/low/moderate low (16.8%), moderate high/high (42.2%), and very high-risk 30.7 %). Pre-HCT IWG 2023 responses were: CR (24.6%), CRbi (19.6%), CRuni (13.6%), CRh (3.2%), PR (1.4%), HI (7.1%) with 27.1% exhibiting no response (NR) and 2.9% with progressive disease (PD). The mOS for the cohort was 52 mos (95%CI 32-74) with a median f/u time among survivors of 62 mos. 5yr PFS was 43% (95% CI: 37-49%) and 5yr OS was 48% (95% CI: 42-54%). 5yr PFS based on IWG 2023 were: 56% (CR), 35% (CRbi), 35% (CRuni), 56% (CRh), 38% (PR), 60% (HI), 36% (NR) and 38% for PD (p=0.24). IWG 2023 failed to accurately predict post-HCT PFS (c-index 0.525, p=.51). Similarly, when IWG 2023 responses were dichotomized, 5y PFS was 45% with cCR and 41% without cCR (p=0.1, c-index 0.525). 5y PFS was 58%, 45% and 29% for low, high and very high IPSS-M groups, respectively (p=.0012, c-index 0.574).

As there was no prognostic distinction among CR categories, we evaluated the response components and found that only Hb and PLT significantly impacted HCT outcomes, while ANC and marrow blast % did not. 5y PFS was 38% for pre-HCT Hb<10 vs 49% for Hb≥10 (p=.03), and 35% for PLT<100k vs 52% for PLT≥100k (p=.002). 5y PFS for pts with <5%, 5-9% and ≥10% blasts were 42.5%, 53% and 35% respectively (p=.61). 5y PFS was 45% for ANC<1 vs 43% for ANC≥1 (p=.76).

We then used recursive partitioning to identify cutoff values for blasts, PLT, and Hb in the pre-HCT setting. The model produced 3 segregated PLT categories with distinct PFS: <40k, 40k-160k, and ≥160k (5y PFS: 18%, 42%, 63% respectively, p<.0001, c-index: 0.593) and 3 segregated categories for Hb <10, 10-13, ≥13 (5y PFS: 38%, 45%, 68% respectively, p=.01, c-index: 0.5595). Lastly, blast % was dichotomized into <10% and ≥10% categories with 5yr PFS of 44% and 35% respectively. We then assigned a score to each category: 3 for PLT <40k, 2 for blasts ≥10% OR Hb <10, 1 for blasts <10%, PLT 40k-160k OR Hb 10-13, and 0 for PLT ≥ 160k OR Hb ≥13. Based on the summed score from these categories we established 3 discrete pre-HCT response strata: favorable (1-2), intermediate (3-4) and poor (≥5). 5y PFS for these response categories was 70%, 40% and 24% (p<.0001, c-index 0.61), a notable improvement over IWG 2023 (c-index 0.526) and IPSS-M (c-index 0.574).

In multivariable analysis after adjusting for patient, HCT, and disease-related factors, the HRs for PFS were 2.9 (95%CI 1.7-5.1, p=.0001) for the intermediate group and 5.7 (95%CI 3-10.9, p<.0001) for the poor-risk group (c-index: 0.692).

Conclusion: The IWG 2023 response criteria do not adequately predict outcomes for pts treated with HMA followed by HCT. To address this, we developed a risk model based on pre-HCT specific response criteria, which provides a more accurate prediction of HCT outcomes. To ensure applicability, this model will require validation in prospective settings.