Acute lymphoblastic leukemia (ALL) is the most common childhood cancer and remains a leading cause of cancer death in children. Large scale studies examining the genomic landscape of ALL using bulk tumor samples have defined multiple new subtypes, genomic drivers, risk classifying genes and therapeutic targets. However, there are few studies of ALL using single-cell RNA-seq technology to study heterogeneity and the surrounding tumor microenvironment (TME). Previous studies, such as those described here, indicate that single-cell RNA seq studies of AML can provide new insights in the tumor intrinsic and extrinsic factors driving tumor behavior and relapse. Gene expression profiling (GEP) using the 3’ 10x platform of small numbers of matched diagnosis and relapse samples have shown enrichment of a CSF1R signature in the TME at relapse. Single-cell profiling of ALL-stromal cocultures identified a resistant ALL cell population undergoing epithelial mesenchymal transition. Mutational profiling of stem and progenitor populations from leukemia samples was shown to map tumor initiating lesions to developmental stage, indicating that mutational driver and cell of origin is a key determinant of leukemia lineage, and T cell profiling identified autoreactive T cells directed to fusion oncoprotein and mutational neoepitopes. Here, we expand upon previous single-cell studies of ALL, using single-cell RNA seq to profile gene expression, mutational diversity, immunophenotype, TME composition and T cell repertoire in ALL subtypes representative of standard and high risk disease in 95 patients: ETV6-RUNX1-like, KMT2A-rearranged, Ph+, Ph-like, ZNF384-rearranged, B/myeloid mixed phenotype acute leukemia, DUX4-rearranged, MEF2D-rearranged, TCF3::PBX1, hyperdiploid, low hyplodiploid and near haploid ALL. All samples are subject to 5’ 10x single-cell GEP of the tumor, TME and T cell compartments, and B-cell ALL. Simultaneous single-cell cell surface protein sequencing and RNA-seq is incorporated for a subset of tumors with lineage ambiguity. When available, single-cell GEP of relapse samples was obtained. Complementary studies include profiling full length RNA-seq (SMART-seq HT and PacBio) of blast and progenitor cell populations to integrate fusion/mutational profile, expression, and cell of origin.

Leukemia is the most frequently diagnosed pediatric cancer, accounting for nearly one third of childhood cancer diagnoses. Although outcomes for pediatric leukemia patients have improved significantly with advents in molecular diagnostics and targeted treatments, a number of patients nonetheless have poor outcomes including treatment-resistant recurrence of their disease. In this dataset, we utilized 10X v2 single-cell RNA sequencing to profile the transcriptomes of 25 pediatric leukemia samples collected from patients at Children’s Mercy Kansas City at single-cell resolution. Viably preserved blood and bone marrow samples collected at initial diagnosis, remission, or recurrence were banked in the Children’s Mercy Tumor Bank biorepository, then profiled for this study. Represented leukemias include B-cell Acute Lymphoblastic Leukemia (B-ALL), T-cell Acute Lymphoblastic Leukemia (T-ALL), and Acute Myeloid Leukemia (AML). Profiled samples include those with well-characterized subtypes (e.g. ETV6::RUNX1 B-ALL) as well as those where a subtype was not determined by standard molecular and cytogenetic methods. Characterization of these samples at single-cell resolution provides a unique opportunity to study the heterogeneity of these cancers, identify subpopulations of interest (e.g. leukemic stem cells), and explore transcriptional similarities between patients with unknown leukemic subtypes and those with a confident diagnosis. Furthering our understanding of the cellular environment of these leukemias is vital for improved diagnosis and treatment selection for pediatric patients afflicted by these cancers.

Samples were multiplexed across three pools. Demultiplexing to create sample-specific FASTQ files was performed via demuxlet based on patient-specific VCF files generated from whole genome or whole exome sequencing. These demultiplexed sample-specific FASTQ files were treated as individual libraries for the ScPCA-nf workflow.