Phenotypic shift from primitive to monocytic immunophenotype is unique to patients treated with azacitidine + venetoclax (AZA/VEN) and associated with worse outcomes in AML

Abstract Background: Azacitidine (AZA) with Venetoclax (VEN) has improved outcomes in older or unfit patients with newly diagnosed acute myeloid leukemia (AML), yet most ultimately relapse. Outcomes for those with relapsed/refractory disease after AZA/VEN remain poor. A deeper understanding of VEN resistance is essential for improving outcomes. Prior studies have shown that monocytic differentiation is associated with VEN resistance and inferior outcomes. While phenotypic shift from primitive at diagnosis to monocytic at relapse has been reported anecdotally in AZA/VEN-treated patients, it has not been systematically characterized, nor is it known whether this occurs in patients treated with intensive chemotherapy (IC). We conducted a multi-institutional analysis to define the frequency, molecular correlates, and clinical impact of diagnostic to relapse phenotypic shifts in AML. Methods We retrospectively analyzed 210 adult AML patients across four academic centers treated with either IC or AZA/VEN who had paired diagnostic and relapse samples available for immunophenotypic review. French–American–British (FAB) classification was assigned at both timepoints using histopathology and flow cytometry with central hematopathology adjudication. FAB M0/M1/M2 AML were designated as “primitive”, and FAB M4/M5 as “monocytic”, with CD14, CD11c, and CD64 expression used as confirmatory markers. FAB M6 and M7 subtypes were excluded. Patients were categorized: (1) primitive to primitive (P-P), (2) primitive to monocytic (P-M), (3) monocytic to primitive (M-P), or (4) monocytic to monocytic (M-M). Clinical and genomic variables were compared using chi-square, Fisher exact, ANOVA, and Kruskal-Wallis tests. Survival outcomes were calculated using Kaplan-Meier methods. Results Of 210 patients, 122 (58%) received IC and 88 (42%) received AZA/VEN. Patients in the AZA/VEN cohort were older (median 72.5 vs. 50.3 years; p<0.0001) with lower diagnostic WBC count (38.3 vs. 13.8×10⁹/L; p=0.0290), and higher frequency of ELN 2022 adverse-risk disease (64.4% vs. 40%, p=0.0230). AZA/VEN patients were also enriched for mutations in ASXL1 (15.9 vs. 4.9%; p=0.0074), RUNX1 (18.2% vs 4.1%; p=0.0008), and TP53 (25% vs. 4.1%; p<0.0001). Baseline rates of primitive phenotype were similar between AZA/VEN (n=63; 71.6%) and IC (n=77; 63.1%). Phenotypic shift from P-M occurred in 5 of 210 patients (2.4%) in the study population, all of whom had received AZA/VEN (5/88; 5.7%; p=0.0124). In the AZA/VEN cohort, compared to primitive patients who did not shift (P-P), P-M shifters were older (median 81 vs. 73 years; p=0.0935) and were significantly more likely to harbor RUNX1 (80% vs. 14.5%; p=0.0035) and IDH1 or IDH2 mutations (80% vs. 14.5%; p=0.0035). Two of five P-M shifters carried RAS-pathway mutations (40%), and none had TP53 mutations. In the AZA/VEN cohort, patients with P-M shift had numerically shorter median overall survival, 8.1 months versus 17.0 months with other phenotypic trajectories (p=0.1448). Median progression-free survival (PFS) was also numerically shorter (5.4 vs. 8.5 months; p=0.4442). The median survival after relapse for P-M shifters was 1.25 months (95% CI, 0.23-12.07). In contrast, M-P shift occurred in 25/210 (12%) patients; n=16 (13.2%) in the IC cohort and n=9 (10.2%) in the AZA/VEN cohort. Among IC-treated patients, M-P shift was associated with longer PFS (14.8 vs. 8.2 months; p=0.0556); among AZA/VEN patients, there was no PFS difference (8.8 vs. 8.4 months for M-P vs other; p=0.9924). Conclusions Phenotypic evolution from primitive to monocytic at relapse occurred exclusively among AZA/VEN-treated patients and was not observed with IC. This phenotypic shift occurred in nearly 8% of AZA/VEN-treated patients and was associated with worse outcomes. The associations of this phenomenon with RUNX1 and IDH1/2 mutations in this subgroup is a novel observation and warrants further study. Conversely, M-P shift occurred with both treatment modalities, and was associated with improved PFS in IC-treated patients but not AZA/VEN. These findings suggest AZA/VEN applies distinct selective pressure favoring monocytic differentiation, and this represents a unique mechanism of resistance. Prospective studies incorporating early clonal tracking and therapeutic strategies to prevent monocytic evolution are needed to mitigate this form of relapse.