Real-life CD34+CD38- leukemic stem cell (LSC) flow quantification in adult AML at diagnosis independently predicts survival of intensively but not AZA-VEN treated patients: A study of prospective observational ALFA-PPP acute leukemia french association

Abstract

Abstract Introduction CD34+CD38- leukemic stem cell (LSC) is a key prognostic marker in AML at diagnosis and follow-up in intensive clinical AML trials (Ngai et al, Haematologica 2025, Plesa et al, Annual ASH Meeting 2024). LSC frequency has been reported as a predictor marker for OS in patients (pts) less-intensively treated with azacitidine (AZA) or decitabine without venetoclax (VEN) (Reuvekamp et al, Leukemia 2025). However, there are no data evaluating the clinical impact of LSC frequency at diagnosis in pts treated with AZA-VEN. In this study, we aim to evaluate the prognosis value of flow LSC at diagnosis in pts enrolled in the prospective real-life observational ALFA-PPP study (NCT04777916). The flow data were obtained from 6 French laboratories using a standardized multicentric approach as previously reported. Methods The quantification of LSC was based on at least one aberrancy in CD34+CD38- fraction using most relevant LSC markers as Mix (CLL1/TIM3/CD97) or CD45RA or CD123 +/- CD90(Thy-1). Multiparametric flow cytometry (MFC) was performed on fresh bone marrow (BM) using a 2-tubes panel with 8 minimal mandatory markers by tube (Canto8C/Navios10C/Lyric12C/DxFlex12C). A backbone of CD34/CD38/CD45/CD117 was completed by HLADR,CD7,CD19,CD56,CD13,CD33,CD36. At least 500,000 to 1 million cells were acquired. The rare events populations (LSC,nHSC,MPP,LMPP”like”) were evaluated using a common gating strategy, independently reviewed by the MRD flow network coordinators. Results Between April 2022 and August 2024, 668/767 pts (87%) registered in 12 centers of the ALFA-PPP were studied by flow LSC at diagnosis:median age,64 years;ECOG-PS 0/1/2/3/unknown,180/328/102/42/16; HCTCI comorbidity index <2 (347) vs ≥2 (321); median WBC, 7.4 G/L; de novo AML, 541 (81%); ELN-2022 risk: 127 (19%) fav, 128 (19.2%) int, 383 (57.3%) adv, 30 (4.5%) unknown. A total of 378 pts were treated with intensive chemotherapy (ICT). At diagnosis, using a 1% of CD45+/ssc BM blast cells cut-off, 140 (37%) were LSC-high and 238 (63%) were LSC-low. LSC-high pts were older than LSC-low pts (median age, 62.5 vs 50 years; p= 0.005). They were 84/160 (52.5%), 32/93 (34.4%) and 19/103 (18.4%) in the adv, int, and fav ELN-2022 risk group respectively (p<0.001). With a median follow-up of 1.9 year, overall survival (OS) was significantly shorter in the LSC-high group (median OS,2.4 years vs not reached; p<0.001). Survival analyses were thus adjusted for age, gender, ECOG-PS, HCTCI, WBC ≥30G/L, de novo vs secondary/therapy-related AML, ELN-2022 risk group, and allogeneic HSCT in CR1 as a time-dependent covariate in multivariate Cox regression models. Interestingly, the negative impact of LSC was observed in the ELN-2022 adverse-risk group (adjusted HR, 1.76 [95% CI, 1.06-2.94]; p= 0.030) while not in the other ELN-2022 risk groups and LSC remained an independent prognostic marker for OS overall (adjusted HR,1.52[1.01-2.28];p= 0.046). Among the remaining 290 pts,170>60years were less-intensively treated with AZA-VEN. Here, using the same 1% of LSC cut-off, 90 (53%) pts were LSC-high and 80 (47%) pts were LSC-low at diagnosis. LSC-high pts were 74/130 (56.9%), 12/24 (50.0%) and 3/12 (25.0%) in the adv, int, and fav ELN-2022 risk group respectively (p= 0.12). Contrary to what observed in ICT-treated pts, no difference in OS was observed here between the LSC-high and LSC-low groups (median OS, 0.86 [0.57-1.14] vs 0.99 [0.43-1.98] years; p= 0.46) even in the adjusted multivariate model (adjusted HR, 0.89 [0.59-1.34]; p= 0.58). These results could be linked to: (i) the specific genomic profile observed in this poor group largely enriched in TP53-mutated AML pts (37/170 ; 24/37 LSC-high ; 13/37 LSC-low);(ii) the biological heterogeneity in the LSC compartment with different stemness signature according to MPP-like vs LMPP-like, and clonal hematopoiesis profile;(iii) the sensitivity of LSC to VEN by disrupting the metabolic machinery (Polyeea et al, Nature 2018). Conclusions In this prospective real-life AML cohort, we confirmed that flow CD34+CD38- LSC at diagnosis allowed better defining the prognosis of intensively-treated AML pts, independently of the ELN-2022 risk classification and especially in the adverse risk group, while this impact was not observed in pts treated with AZA-VEN. Further studies combining LSC flow MRD monitoring, MAC-Score and refined ELN2024 risk stratification should be conducted to clarify the clinical impact and interconnection of these markers in AZA-VEN treated pts.