Type and degree/number of immunophenotypic abnormalities refine multiparameter flow cytometry (MFC)-based measurable residual disease (MRD) testing in adults allografted for Acute Myeloid Leukemia (AML) in morphologic remission Free

: MFC-based MRD testing is now routinely used in AML, but methodological refinements are needed to optimize assay performances. Here, we studied to what extent the type and degree/number of immunophenotypic abnormalities on leukemic blasts improves prognostication with MFC MRD testing in AML.

Methods: We examined 1,215 consecutive adults with AML (n=1,053) or myelodysplastic neoplasm (MDS)/AML (n=164) who received a first allograft in first or second morphologic remission between 1/2007 and 3/2023 at our institution and underwent 10-color/3 tube MFC as a routine clinical test during the pre-HCT work-up. MRD was identified via “difference-from normal approach”. In patients in whom MRD was identified by MFC, leukemic blasts were further classified into stem cell (“SC”: CD34^bright and CD38^dim-absent) and non-stem cell (“NSC”: all other CD34/CD38 expression patterns) phenotypes. To assess the degree of immunophenotypic abnormalities, immunophenotype scores were calculated based on deviations in the expression of specific antigens compared to normal progenitors, analogous to the Flow Cytometric Scoring System (FCSS) established for MDS. Each antigen was scored individually: aberrant T/NK-cell antigen expression: 1 point; increased or decreased antigen expression: 1 point; mildly increased or decreased antigen expression: 0.5 points; and normal expression pattern: 0 points. The SC and/or NSC scores were calculated by summing the points across all SC and NSC antigens evaluated, respectively. A total MRD immunophenotype score was then computed for each MRD^pos patient by combining points for each aberrantly expressed antigen.

Results: Among 233 patients (19%) with pre-HCT MRD, 80 (34%) had NSC-like leukemic blasts, 109 (47%) had SC-like leukemic blasts, and 44 (19%) had NSC- and SC-like leukemic blast cell populations. Between these 3 patient subsets, there were statistically significant differences regarding disease type, cytogenetic risk at diagnosis, white blood cell (WBC) count at diagnosis, remission status at HCT, residual cytogenetic abnormalities at the time of HCT, and blood counts before HCT. After multivariable adjustment, patients with NSC+SC leukemic blasts had worse outcomes than the other MRD^pos patients. Across all 233 MRD^pos patients, the total MRD immunophenotype score varied widely (median [range]: 6 [1-10]). Martingale residual plots revealed a non-linear relationship between this score and relapse-free survival (RFS), and maximally selected rank statistics empirically estimated ≤4.5 vs. >4.5 as optimal dichotomization threshold for RFS. The 63 patients (27% of MRD^pos patients) with low total MRD immunophenotype scores differed from the 170 patients (73% of MRD^pos patients) with high scores regarding age at HCT (P=0.014), WBC count at diagnosis (P=0.045), remission status (P=0.039), proportion of patients with residual cytogenetic abnormalities at the time of HCT (P=0.037), and conditioning intensity (P=0.023). Importantly, patients with a high total immunophenotype score had substantially worse post-HCT outcomes than those with a low score whereas there was no statistically significant difference for relapse risk or RFS between patients with low total MRD immunophenotype score and those without MRD. Specifically, after adjustment for age at HCT, type of disease, WBC count at diagnosis, karyotype/cytogenetic risk, remission status, cytogenetics at time of HCT, conditioning intensity, and pre-HCT blood counts, having a high total MRD immunophenotype score was associated with a significantly higher risk of relapse (HR=5.08 [3.99-6.47]; P<0.001), shorter RFS (HR=3.90 [3.17-4.81]; P<0.001), shorter OS (HR=2.99 [2.42-3.69]; P<0.001), and higher risk of NRM (HR=1.71 [1.07-2.72]; P=0.025) relative to patients without pre-HCT MFC MRD. In contrast, there was no statistically significant difference for the risk of relapse or RFS between patients with low total MRD immunophenotype score and those without pre-HCT MFC MRD (for relapse: HR=1.08 [0.68-1.71]; P=0.75; for RFS: HR=0.83 [0.57-1.22]; P=0.35).Conclusions: Considering the type and degree/number of immunophenotypic abnormalities (as done, for example, with the total MRD immunophenotype score) refines MRD testing. It identifies a significant subset of MRD^pos patients with low scores (in our cohort, 27% of all MRD^pos patients) with outcomes like MRD^neg patients that may benefit from reclassification for optimized risk assessment.