P068 | ABNORMAL CD33 EXPRESSION ON BLAST CELLS PREDICTS ELN2022 ADVERSE RISK ACUTE MYELOID LEUKEMIA

Multiparametric flow cytometry (MFC) is useful at diagnosis and during therapy by measuring minimal residual disease in acute myeloid leukemia (AML) patients (pts). However, there are few data correlating aberrant immunophenotype with prognostically relevant mutations. CD7 expression or blasts with APL-like phenotype were previously associated with specific genetics, but many different immunophenotypes still lack correlation with molecular aberrations. Our study investigated a retrospective cohort of 146 AML pts (median age 59 years, median follow-up 40 months) treated at our center with intensive chemotherapy (according to NILG AML 00-01 protocol) from 2012 to May 2024. Pts were studied with MFC, using 22 monoclonal antibodies 8-color flow cytometry panel, and with a target myeloid gene panel (Sophia Genetics) at diagnosis. 45 pts (30.8%) showed abnormally low expression of CD33 (CD33dim/-). The remaining 101 AML (69,2%) were CD33+. Abnormal CD33 expression was inversely correlated with CD34 expression: CD33dim/- was associated with CD34+ in blasts of 39/45 (86.7%) pts, while among 101 CD33+ AML, 40 (39.6%) showed CD34+ blasts, 51 (50.5%) were CD34-, and 10 (9.9%) showed bimodal CD34 expression (p <0.001). Three out of 45 CD33 dim/- AML were NPM1+, whereas this mutation was found in 47/101 CD33+ AML (6.7% vs 46.5%, p<0.001); FLT3 ITD was consistently less frequent in CD33dim/- compared to CD33+ AML (7% vs 19.8%, p = 0.013) (Fig.1A). In 45 CD33dim/- AML, 28 were characterized by adverse risk gene mutations, according to ELN2022, whereas 32/101 CD33+ carried at least one of those mutations (62.2 % vs 31.7% p <0.001). In particular, CD33dim/- pts had a higher frequency of RUNX1 (45% vs 18%, p<0.01), ASXL1 (36% vs 14%, p<0.01) and IDH2 (31% vs 13% p<0.05) mutations when compared to CD33+ pts (Fig.1A). Consequently, 32/45 CD33dim/- AML were categorized as adverse risk, whereas 35 out of 101 CD33+ pts fitted in the same high-risk category (71.1% vs 34.6%, p<0.001). Median overall survival (OS) was shorter in CD33dim/- pts compared to CD33+ (20 vs 114 months, p = 0.005) (Fig.1B). In conclusion, we showed that CD33 dim/- pts were characterized by a worse OS and enriched in high risk mutations, while being mutually exclusive with NPM1 mutations. Conversely, CD33+ pts were principally associated with favorable condition, characterized by NPM1 mutations and a better outcome. Our data suggest that MFC analysis may predict main genetic AML subgroups and pts outcome.