Gene Signature Identifies High-Risk Neuroblastoma Cells Missed by Older Models

Neuroblastoma is the most common solid tumor found outside the brain in children, but why some children respond well to treatment while others do not has remained poorly understood. Now, St. Jude Children’s Research Hospital scientists and their collaborators have now built one of the most comprehensive datasets of neuroblastoma cells to date, using it to define the tumor’s cell types and identify a gene signature marking a malignant cell population linked to poor prognosis. The team also created patient-derived models that capture this population, something older laboratory cell lines had missed. The findings, published in Cancer Cell, give researchers a new framework and a set of tools for studying high-risk neuroblastoma.

Neuroblastoma cells exist in two states: adrenergic cells, associated with low-risk disease, and mesenchymal cells, associated with high-risk disease, treatment resistance, and poor outcomes. Directly identifying cancer-related mesenchymal cells in patient tumors has been difficult, however, because the body also contains healthy mesenchymal cells that can be mistaken for their malignant counterparts. To find a better way to identify these treatment-resistant cancer cells, researchers examined 54 tumors from 50 patients, building a dataset that incorporated single-cell RNA sequencing and spatial omics. By using patient-derived xenografts, tumor cells grown within mice, the team isolated a clear gene expression profile connected to malignant mesenchymal cells, since healthy mesenchymal cells from patient tumors do not grow in xenografts. 

These xenografted tumor cells, generated through the St. Jude Childhood Solid Tumor Network, were essential both to confirming that these cells exist and to defining their molecular and cellular features. “With this new gene expression signature for mesenchymal cells in neuroblastoma, we can dig into why outcomes are so varied between different children with this cancer,” said co-corresponding author Michael Dyer.

To confirm that the new signature outperformed previous methods, the team tested it against an independent database of patient tumor RNA sequencing data, finding that it significantly improved outcome predictions, while a pre-existing signature based on decades-old cell lines did not. The team validated the signature using organoids, cell lines, and xenografts, applying spatial transcriptomics, spatial proteomics, electron microscopy, and chromatin profiling, and found that it consistently distinguished between the two cell types, which also showed distinct shapes, internal spatial organization, and behaviors. “I’m just endlessly fascinated by the signature itself, because regardless of the test, it’s been very robust,” said first author Anand Patel. “These markers are going to be powerful tools to understand neuroblastoma better.”

With the signature validated, the researchers created additional in vitro models and new cell lines from xenografts derived from patient samples with varying proportions of mesenchymal to adrenergic cells. These models are available upon request through the Childhood Solid Tumor Network, offering a resource that could reinvigorate research into high-risk disease by helping scientists understand tumor cell states, identify vulnerabilities, and evaluate treatment strategies. “Our new models are a launching point for neuroblastoma research,” said co-corresponding author Elizabeth Stewart. “Using this dataset as our foundation, the field is poised to explore the biology of this pediatric cancer more deeply, and ultimately, move us toward improving our patients’ outcomes.”