7042 Background: Surface antigen expression evaluation is part of the standard work-up at acute myeloid leukemia (AML) diagnosis. The biological & prognostic implications of surface antigen expression patterns in normal karyotype (NK) AML patients (pts) remain unknown. Methods: The diagnostic antigen expression patterns of mononuclear cells in bone marrow (BM) of 111 NK-AML pts were assessed using a standard flow cytometric panel. At diagnosis common AML gene mutations (mut) & expression levels were analyzed. Pts received stem cell transplantation (SCT, 98% allogeneic, 2% autologous; median age 63 years [y, range 26-74y]) after induction therapy at our institution. Median follow up was 3.3y. With R’s gplot package unsupervised hierarchical clustering of surface antigens was performed & revealed 4 distinct clusters. Results: Pts in cluster 1 (n = 36) had higher expression of immature, in cluster 2 (n = 31) of thrombocytic/T-cell/erythroid, in cluster 3 (n = 24) of monocytic & in cluster 4 (n = 20) of myeloid surface antigens. All 4 clusters associated with distinct clinical & molecular features. At diagnosis, compared to all others, pts in cluster 1 had a higher CD34+/CD38- cell burden ( P< .001), higher blood blasts ( P< .03) & BM blasts ( P< .06) by trend. They had less NPM1 mut ( P< .001) & DNMT3A mut ( P= .02), were more likely to be EVI1 positive ( P= .03) & had higher EZH2 ( P= .02), RUNX1 ( P= .009), BAALC ( P< .001), ERG ( P= .02) & MN1 ( P< .001) expression. Compared to all others, pts in cluster 1 had a higher cumulative incidence of relapse (CIR, P= .002, at 1y 41% vs 15%) & shorter event-free survival (EFS, P= .02, at 1y 50% vs 69%). In multivariate analysis, cluster 1 pts had a significantly higher CIR (Hazard Ratio [HR] 5.4, P= .01) after adjustment for FLT3-ITD & shorter EFS (HR 2.1, P= .02) after adjustment for FLT3-ITD, age & disease status at SCT. Conclusions: Pts in cluster 1 had high expression of immature surface antigens (eg CD34, CD117, CD13), genes involved in stem cell renewal & worse outcome. Our data indicate a relationship between easily accessible surface antigen expression patterns at diagnosis, molecular disease features & aggressiveness of the NK-AML phenotype.