Acute myeloid leukemia with mixed phenotype is characterized by stemness features with limited lineage plasticity

Mixed phenotype (MP) in acute leukemias poses unique classification and management dilemmas and can be seen in entities other than de novo mixed phenotype acute leukemia (MPAL). Although WHO classification empirically recommends excluding AML with myelodysplasia related changes (AML-MRC) and therapy related AML (t-AML) with mixed phenotype (referred to as “AML-MP”) from MPAL, there is lack of studies investigating the clinical, genetic, and biologic features of AML-MP. We report the first cohort of AML-MP integrating their clinical, immunophenotypic, genomic and transcriptomic features with comparison to MPAL and AML without MP. Patients with AML-MP share similar clinical and genetic features to its AML counterpart but differs from MPAL. AML-MP harbors more frequent RUNX1 mutations than AML without MP and MPAL. RUNX1 mutations or complex karyotypes did not impact the survival of MPAL patients. Unsupervised hierarchal clustering based on immunophenotype identified biologically distinct clusters with phenotype/genotype correlation and outcome differences. Furthermore, transcriptomic analysis showed an enrichment for stemness signature in AML-MP and AML without MP as compared to MPAL. Lastly, MPAL but not AML-MP often switched to lymphoid only immunophenotype after treatment. Expression of transcription factors critical for lymphoid differentiation were upregulated only in MPAL, but not in AML-MP. Our study for the first time demonstrates that AML-MP clinically and biologically resembles its AML counterpart without MP and differs from MPAL, supporting the recommendation to exclude these patients from the diagnosis of MPAL. Future studies are needed to elucidate the molecular mechanism of mixed phenotype in AML. Overall design: Multiparameter flow cytometry was performed on bone marrow aspirates at diagnosis. Briefly, up to 1.5 million cells from freshly drawn bone marrow aspirate were stained with 10-color panels, washed, and acquired on a Canto-10 cytometer (BD Biosciences, San Jose, CA). The results were analyzed with custom Woodlist software (generous gift of B. L. Wood, Children's Hospital Los Angeles). Cryopreserved bone marrow aspirate specimens collected at diagnosis from our institutional biospecimen bank were thawed and flow sorted into myeloid blasts (CD33+, CD7-) and T lymphoblasts (CD33-, CD7+) using FACS-Aria Fusion cell sorter (BD Biosciences). Samples were prepared for RNA-seq using the SMARTer Stranded Total RNA-Seq Kit (Takara Bio) and sequenced at the MSKCC Integrated Genomics Core. FASTQ files were subsequently processed using the Center for Epigenomics Research pipeline consisting of adapter and quality trimming using TrimGalore! (Version 0.6.4), alignment to hg19 using STAR (version 2.7.10b) in two pass mode, and quantification using HTSeq (version 0.9.1).