Single-cell study links an eight-gene program to LSC17-high acute myeloid leukemia

Researchers have combined single-cell and bulk RNA sequencing data to map a transcriptional program that accompanies LSC17-high acute myeloid leukemia (AML), identifying a set of eight genes that tracks the leukemia stem cell-associated signature while remaining distinct from it. The LSC17 score is a well-known measure of leukemia stem cell-related transcriptional activity in AML, but the single-cell basis of that activity and the non-LSC17 programs running alongside it in bulk LSC17-high disease had remained unclear. The new analysis, published in Discover Oncology, addresses that gap by working directly at the level of individual cells and then translating the findings back to bulk tumor datasets.

The team integrated single-cell RNA-seq data from the 16 diagnosis-timepoint AML samples in the GSE116256 dataset with TCGA-LAML and two external AML datasets. LSC17 activity was mapped across the single-cell atlas, and primitive-like cells were stratified within each eligible patient, with the upper and lower 20 percent of LSC17 scores defined as LSC17-high and LSC17-low states. Patient-aware pseudobulk analysis, enrichment analysis, and single-cell to bulk comparison were then used to identify non-LSC17 genes associated with LSC17-high primitive-like cells. LSC17 activity was found to be enriched in primitive and progenitor-like compartments, particularly HSC-like and primitive leukemic-like clusters.

Among 5,124 primitive-like cells from 16 patients, 1,024 LSC17-high and 1,024 LSC17-low cells were identified, and 14 evaluable patients contributed 1,018 cells per state to the paired pseudobulk analysis. That analysis detected 68 differentially expressed genes. LSC17-high cells were enriched for interferon alpha response, inflammatory response, and KRAS signaling, whereas LSC17-low cells were enriched for MYC targets, oxidative phosphorylation, and E2F targets. After excluding the original LSC17 genes, 17 single-cell LSC17-high-upregulated genes overlapped with bulk LSC17-high-upregulated genes, from which the researchers manually curated an eight-gene high companion core: SPINK2, HOPX, MEF2C, MSI2, BCAT1, ANGPT1, NPR3, and CD52.

The resulting companion score was elevated in LSC17-high primitive-like cells, correlated with the LSC17 score in TCGA-LAML, and was associated with poorer overall survival. The association reproduced in the external datasets GSE37642-GPL96 and GSE12417-GPL96. The authors are careful, however, to note the limits of the work: prognostic discrimination in the external datasets was modest, and the companion score is not proposed as a standalone clinical predictor. The study is an observational bioinformatic analysis of existing sequencing datasets rather than a test of any new treatment, and its conclusions rest on the samples included in those public cohorts. The authors conclude that the study identifies a single-cell-informed non-LSC17 companion program associated with LSC17-high primitive-like AML states and adverse survival, offering a cell-level view of the transcriptional landscape that accompanies bulk LSC17-high disease and a candidate gene set for future evaluation in AML research.

Subject of Research: A single-cell and bulk transcriptomic analysis of acute myeloid leukemia identifying a non-LSC17 companion gene program associated with LSC17-high primitive-like states.

Article Title: Integrative single-cell and bulk transcriptomic analysis identifies a non-LSC17 companion program associated with LSC17-high acute myeloid leukemia

Article References: Wang, J., & Li, B. (2026). Integrative single-cell and bulk transcriptomic analysis identifies a non-LSC17 companion program associated with LSC17-high acute myeloid leukemia. Discover Oncology. https://doi.org/10.1007/s12672-026-06075-w

Image Credits: AI Generated

DOI: 10.1007/s12672-026-06075-w

Keywords: acute myeloid leukemia, LSC17 score, single-cell RNA-seq, transcriptomics, pseudobulk analysis, leukemia stem cells, TCGA-LAML, gene expression, prognostic signature, bioinformatics, primitive-like cells, Discover Oncology