Relative Prognostic Value of Flow Cytometric Measurable Residual Disease before Allogeneic Hematopoietic Cell Transplantation for Adults with MDS/AML or AML

Background: Measurable residual disease (MRD), assessed by multiparameter flow cytometry (MFC) before allogeneic hematopoietic cell transplantation (HCT), is strongly and independently associated with poor post-HCT outcomes in adults with acute myeloid leukemia (AML). On the other hand, the relative prognostic value of MFC-based MRD testing is unknown in patients with myelodysplastic neoplasm (MDS)/AML, a disease entity newly defined by the 2022 International Consensus Classification (ICC) that encompasses cases of MDS with 10-19% blasts and recognizes the diagnostic continuum between MDS and AML.

Methods: To assess the relationship between pre-HCT, disease type, and post-HCT outcomes, we retrospectively analyzed 1,262 adults ≥18 years with MDS/AML (n=150) or AML (n=1,112), based on the ICC 2022 criteria, who received a first allograft while in first or second morphologic remission at a single institution between 4/2006 and 3/2023.

Results: The MDS/AML and AML cohorts differed significantly regarding patient and disease-specific characteristics, including age at HCT (60 vs. 56 years in MDS/AML and AML patients, respectively, P<0.001), HCT-CI score (P<0.001), adverse cytogenetic risk according to ELN 2022 criteria (39% vs. 26%, P<0.001), intensive therapy prior to allogeneic HCT (69% vs. 97%, P<0.001), time from last remission (76 vs. 98 days, P<0.001), first remission status (97% vs. 77%, P<0.001), pre-HCT MRD by MFC (35% vs. 19%, P<0.001), and peripheral blood stem cells as graft source (89% vs. 80%, P=0.015). There were no differences in donor source or conditioning intensity. With a median follow-up of 5.11 years (interquartile range [IQR]: 2.38-9.66) in survivors, there were 384 relapses, 590 deaths, and 258 non-relapse mortality (NRM) events contributing to the estimates for relapse, NRM, relapse-free survival (RFS), and overall survival (OS). At three years, NRM (21% [14-28%] for MDS/AML vs. 16% [14-19%] for AML, P=0.19), relapse (29% [22-36%] vs. 29% [27-32%], P=0.94), RFS (50% [42-59%] vs. 54% [51-58%], P=0.25), and OS (52% [44-61%] vs. 60% [57-63%], P=0.09) were similar between the two groups. After multivariable adjustment, relapse rates for patients with MDS/AML vs. those with AML were similar (hazard ratio [HR]=0.77 [0.50-1.17], P=0.2), as were NRM (HR=1.41 [0.91-2.18), P=0.12), RFS (HR=0.99 [0.73-1.35], P>0.9) and OS (HR=1.17 [0.85-1.61], P=0.3). Importantly, a significant interaction was found between pre-HCT MRD status and disease type (MDS/AML vs. AML) for relapse (P=0.009), RFS (P=0.013), and OS (P=0.048). The interaction models indicated that the HRs for the association between pre-HCT MRD and post-HCT outcomes were lower in patients with MDS/AML (for relapse: HR=1.73 [0.96-3.11] in MDS/AML vs. HR=3.56 [2.90-4.38], in AML; for RFS: HR=1.60 [1.03-2.48] vs. HR=2.67 [2.25-3.16]; for OS: HR=1.53 [0.97-2.39] vs. HR=2.34 [1.96-2.78]). Pre-HCT MRD was associated with a higher risk of relapse in both MDS/AML and AML patients but to a lesser extent in the MDS/AML patients. In these patients, relapse at three years was 37% [24-51%] for those with pre-HCT MRD vs. 24% [16-33%] for those without (P=0.068), translating into lower RFS (39% [28-56%] vs. 56% [47-67%], P=0.038) and OS (41% [29-58%] vs. 58% [48-70%], P=0.066) in MDS/AML patients with pre-HCT MRD. For comparison, in patients with AML, relapse at three years was 22% [20-25%] for those without MRD vs. 56% [50-62%] for those with (P<0.001), translating into lower RFS (27% [22-31%] vs. 61% [58-64%], P<0.001) and OS (35% [29-42%] vs. 65% [62-68%], P<0.001) in those with pre-HCT MRD. In line with these findings, C-statistics showed that pre-HCT MRD was not as predictive in MDS/AML patients in comparison to AML patients (for relapse: 0.57 in MDS/AML patients vs. 0.62 in AML patients; for RFS: 0.57 vs. 0.59; for OS: 0.56 vs. 0.58).

Conclusion: Our data suggest that, overall, patients with MDS/AML or AML in first or second morphologic remission have similar outcomes following allogeneic HCT. However, MFC-based pre-HCT MRD testing is prognostically substantially less informative for MDS/AML than AML patients, possibly because of increased rates of misclassification. These findings support the rationale to examine the prognostic value of molecular MRD testing in patients with MDS/AML.

Figure. Cumulative incidence of relapse stratified (A) by disease type at diagnosis and (B) by disease type at diagnosis and pre-HCT MRD.