Single-Cell Atlas Reveals How Sonic Hedgehog Medulloblastoma Comes Back

Medulloblastoma is the most common malignant brain tumor in children, and the sonic hedgehog (SHH) subtype accounts for roughly one-third of all cases. Even when initial treatment succeeds, SHH medulloblastoma relapses frequently, and once the disease returns, the prognosis is dismal. Now a team at the University of California, San Francisco, led by Abhinav Jain, Bohyeon Yu, Jangham Jung, and senior author Aaron A. Diaz, has produced the most detailed molecular portrait yet of how this tumor evolves from first diagnosis to recurrence. The study, published in Genome Medicine, combines single-nucleus RNA sequencing, single-nucleus ATAC sequencing, and spatial transcriptomics into a longitudinal atlas of SHH medulloblastoma, and it nominates several candidate therapeutic targets that could reshape how relapsed disease is treated.

The technical scope of the work is unusual for pediatric brain tumor genomics. The researchers profiled 22 SHH medulloblastoma specimens, 16 primary tumors and 6 recurrent ones, including four matched primary–recurrent patient pairs. Sixteen samples underwent single-nucleus RNA sequencing, which captures gene expression in individual tumor and stromal cells; thirteen underwent single-nucleus ATAC sequencing, which maps which regions of chromatin are open and therefore potentially regulatory; and six underwent spatial transcriptomics, which preserves the physical layout of the tumor so gene programs can be linked to anatomical neighborhoods. The team integrated these layers with exome-derived variant calls from 25 specimens and validated key findings in bulk RNA sequencing data from 48 specimens representing 24 primary–recurrent pairs. Because the modalities overlap across samples, the authors could cross-check cell-type assignments, regulatory programs, and mutation calls rather than relying on any single technology.

One of the clearest signals to emerge concerns TP53, the most frequently mutated gene in human cancer. The analyses showed evidence of positive selection for nonsynonymous tumor mutations, meaning that tumors carrying certain protein-altering changes were overrepresented in a way consistent with a growth or survival advantage. Among these were TP53 mutations clustered in the DNA-binding domain and the tetramerization domain, the two regions that determine how p53 recognizes damaged DNA and how four p53 subunits assemble into the functional transcription factor. That pattern matters clinically because TP53 status is already known to influence outcomes in SHH medulloblastoma, and the new data place those mutations within a broader network of downstream failures.

That network, the study found, involves base-excision repair. TP53-mutant tumors showed reduced expression of base-excision repair genes and reduced chromatin accessibility at the cis-regulatory regions linked to them. In practical terms, when p53 function is lost, the tumor not only loses its ability to police genomic damage but also appears to dial down a repair pathway that fixes small chemical lesions in DNA every day. The regulatory signature suggests that the suppression is coordinated at the level of chromatin, not merely a passive consequence of mutation, which raises the possibility that restoring or bypassing base-excision repair could be a viable therapeutic strategy in TP53-mutant disease.

Recurrent tumors told a different story. Compared with primary tumors, they showed increased tumor mutational burden, a measure of how many mutations the tumor has accumulated over time. That finding fits with the idea that relapsed disease has been through more rounds of cell division and evolutionary pressure, often under the selective stress of chemotherapy and radiation. The longitudinal design, with matched pairs from the same patients, allowed the researchers to distinguish changes that arose during treatment from features that were present from the start.

The recurrence-associated expression programs were equally striking. Recurrent tumors showed activation of an endoplasmic reticulum stress program, a cellular response triggered when proteins fold incorrectly inside the endoplasmic reticulum and accumulate faster than the cell can process them. This unfolded protein response is a known vulnerability in some cancers, and drugs that push stressed cells past the point of recovery are being explored in several tumor types. In this cohort, the ER stress signature was associated with poorer overall survival, making it both a potential therapeutic target and a prognostic marker for relapsed SHH medulloblastoma.

The tumor microenvironment also shifted between diagnosis and relapse. Recurrent tumors showed increased expression of angiogenic genes, reflecting the growth of new blood vessels that supply the tumor with oxygen and nutrients. Alongside this vascular remodeling, the researchers inferred increased immunosuppressive myeloid signaling, pointing toward tumor-associated macrophages and related immune cells that dampen antitumor immune responses. Spatial transcriptomics helped anchor these findings in tissue, showing where angiogenic and myeloid programs localize within the tumor architecture rather than treating the microenvironment as a uniform soup.

Together, these findings define recurrence-associated cellular and regulatory programs in SHH medulloblastoma and nominate candidate therapeutic targets across several axes: base-excision repair in TP53-mutant tumors, ER stress pathways in relapsed disease, angiogenesis, and immunosuppressive myeloid signaling. The study also delivers a single-cell and spatial atlas of SHH medulloblastoma that other researchers can mine for additional targets, biomarkers, and cell-type-specific signatures. The atlas is openly available, consistent with the open-access publication of the paper in Genome Medicine.

The clinical implications are significant because recurrent SHH medulloblastoma currently has few effective options. Standard medulloblastoma treatment, which combines surgery, craniospinal radiation, and chemotherapy, has improved survival for many children, but relapsed disease remains one of the hardest pediatric oncology challenges. SHH medulloblastoma has been an attractive target for hedgehog pathway inhibitors, drugs that block the sonic hedgehog signaling cascade driving the tumor, but resistance develops, and these inhibitors have limited utility in some patients, particularly those with TP53-mutant disease. The new atlas suggests alternative angles of attack that do not depend on the hedgehog pathway itself.

The work also demonstrates the value of longitudinal single-cell and spatial profiling in pediatric neuro-oncology. By following the same molecular features across primary and recurrent tumors from individual patients, the researchers could trace tumor evolution at cellular resolution rather than inferring it from bulk data alone. The integration of gene expression, chromatin accessibility, and spatial context provides a template for how future studies of relapsed brain tumors might be designed. As the authors note, the data define prognostic markers and nominate therapeutic targets, offering a resource that could inform both clinical trial design and the development of new treatments for children whose medulloblastoma returns.

Subject of Research: Genome Medicine

Article Title: Longitudinal single-cell and spatial profiling of sonic hedgehog medulloblastoma identifies novel therapeutic targets

Article References: Jain, A., Yu, B., Jung, J., Rosatti, E., Phillips, J. J., & Diaz, A. A. (2026). Longitudinal single-cell and spatial profiling of sonic hedgehog medulloblastoma identifies novel therapeutic targets. Genome Medicine. https://doi.org/10.1186/s13073-026-01803-x

Image Credits: AI Generated

DOI: 10.1186/s13073-026-01803-x

Keywords: Longitudinal, single-cell, spatial, profiling, sonic, hedgehog, medulloblastoma, identifies, novel, therapeutic, targets, scientific research