The pathogenesis of neuroblastoma with bone or bone marrow metastasis (NB-BBM) and its complex immune microenvironment remain poorly elucidated, hampering the advancement of effective risk prediction for BBM and limiting therapeutic strategies. Feature recognition of 142 paraffin-embedded hematoxylin-eosin-stained tumor section images was conducted using a Swin-Transformer for pathological histology to predict NB-BBM occurrence. Single-cell transcriptomics identified a tumor cell subpopulation (NB3) and two tumor-associated macrophage (TAM) subpopulations (SPP1⁺ TAMs and IGHM⁺ TAMs) closely associated with BBM and highlighted transketolase (TKT) as a key molecular marker for metastatic progression in NB. This extensive multi-omics investigation into NB-BBM enhances our understanding of single-cell transcriptional dynamics in NB beyond existing research, outlining the evolution from in situ carcinoma through tumorigenesis to bone marrow metastases. Furthermore, exploration of the immune microenvironment identified specific subpopulations of TAMs crucial in promoting NB-BBM, presenting new avenues for immunotherapy. These insights enhance our understanding of the metastatic process from NB to BBM and facilitate the development of more effective diagnostic and therapeutic strategies for this aggressive pediatric cancer.

Neuroblastoma (NB), which is the most common pediatric extracranial solid tumor, varies widely in its clinical presentation and outcome. NB has a unique ability to spontaneously differentiate and regress, suggesting a potential direction for therapeutic intervention. However, the underlying mechanisms of regression remain largely unknown, and more reliable prognostic biomarkers are needed for predicting trajectories for NB. We performed scRNA-seq analysis on 17 NB clinical samples and three peritumoral adrenal tissues. Primary NB displayed varied cell constitution, even among tumors of the same pathological subtype. Copy number variation patterns suggested that neuroendocrine cells represent the malignant cell type. Based on the differential expression of sets of related marker genes, a subgroup of neuroendocrine cells was identified and projected to differentiate into a subcluster of benign fibroblasts with highly expressed CCL2 and ZFP36, supporting a progressive pathway of spontaneous NB regression. We also identified prognostic markers (STMN2, TUBA1A, PAGE5, and ETV1) by evaluating intra-tumoral heterogeneity. Lastly, we determined that ITGB1 in M2-like macrophages was associated with favorable prognosis and may serve as a potential diagnostic marker and therapeutic target. In conclusion, our findings reveal novel mechanisms underlying regression and potential prognostic markers and therapeutic targets of NB.