Multiomic Single-Cell Sequencing of Mixed Phenotypic Acute Leukemia (MPAL) Reveals Complex Immunophenotypic, Transcriptional, and Genetic Heterogeneity

Introduction

Mixed phenotypic acute leukemia (MPAL) is characterized by leukemic blasts with both lymphoid and myeloid cell-surface markers. Due to the relative rarity and heterogenous nature of MPAL, the optimal therapeutic strategy remains uncertain, although emerging data suggest further sub-classification of MPAL may facilitate therapeutic decision making. Using single cell (SC) multiomic profiling, we sought to further characterize the heterogenous transcriptional, genetic, and immunophenotypic landscape of MPAL.

Methods

We analyzed samples from 14 patients with newly diagnosed MPAL (11 B/myeloid; 3 T/myeloid). We performed high-throughput SC DNA sequencing (Mission Bio Tapestri) and SC RNA sequencing (Fluent BioSciences PIPseq) with simultaneous cell surface immunophenotyping.

Results

Using SC DNA sequencing, we analyzed a total of 51,847 cells (4,196 cells/patient); using RNA sequencing, we analyzed a total of 71,579 cells (6,143 cells/patient).

Multiomic SC sequencing revealed striking intra-patient heterogeneity, with multiple immunophenotypically, genetically, and/or transcriptionally distinct populations within individual patients. Furthermore, for individual patients, “mixed” phenotype was not always correlated with genetics or gene expression. In 5/14 patients, immunophenotype correlated with both genotype and transcriptional signature in at least one subpopulation. For example, in one patient, 32% of biphenotypic blasts demonstrated monocytic features (CD11b, CD14, and CD64 expression). Unlike blasts without monocytic features in the same specimen, this population was IDH2^mut and had increased differential expression of LYZ and S100A8/9. For 7/14 patients, however, phenotypic populations did not correlate with a clear genetic driver. In one patient, for example, we observed two blast populations, one with an immature immunophenotype characterized by elevated CD117 and CD34 expression, and one with a monocytic immunophenotype characterized by elevated CD14 and CD64 expression. Although these two populations were both TP53^mut, they had different transcriptional signatures, with the immature population showing increased SOX4, FLT3, and VPREB expression and the monocytic population with increased LYZ, S100A8/9, and MNDA expression. Finally, 2 patients had immunophenotypic heterogeneity but we observed identical genetic and transcriptional signatures within the immunophenotypically distinct subpopulations.

Across this cohort, we identified a recurrent transcriptional signature associated with immature blasts, characterized by increased CD34 and CD117 expression. Found in 7 patients with either B or T/myeloid MPAL, these immature subpopulations were enriched for expression of genes associated with lymphocyte development (DNTT, VPREB, PAX5), malignant myelopoiesis (SOX4, FLT3, SPINK2), and cell-cycle regulation (TUBB, UHRF1) compared to more differentiated blasts from the same patients.

Phylogenic analysis revealed the effect of mutation acquisition on development of a malignant immunophenotype. In 3 patients, we observed both heterozygous and homozygous TP53^mut populations. In all 3 patients, immunophenotypically normal T cells were identified within the heterozygous TP53^mut population, consistent with initial acquisition of this mutation within a stem/progenitor population. In contrast, homozygous TP53^mut cells all expressed aberrant immunophenotypes, consistent with leukemic blasts. Within each individual patient, relative to the heterozygous TP53^mut cells, homozygous TP53^mut cells demonstrated significantly increased CD33, CD34, and CD117 expression.

Conclusions

Multiomic SC sequencing allows for direct measurement of immunophenotypic, genetic, and transcriptional clonal architecture. We demonstrated considerable intra- and inter-patient heterogeneity, in which both genotype and transcriptional signature, transcriptional signature alone, or neither were found to correlate with immunophenotype. Despite this heterogeneity, in half our cohort, we identified a recurrent gene expression signature associated with immature immunophenotype, which correlated with genetic drivers of immunophenotype in some cases. Future multiomic studies are needed to completely subtype MPAL and elucidate the impact of these subtypes on treatment outcomes.