Spatially resolved T cell receptor tracking reveals γδT cell localization to tumor-rich regions in high-risk neuroblastoma: A Report from the Children’s Oncology Group

Abstract

High-risk neuroblastoma (HRNB) is a leading cause of pediatric cancer death. Current therapies center on intensive multimodal treatment including anti-GD2 therapy, with growing interest in harnessing T cell-mediated immunity. How T cells and their receptors (T-cell receptors, TCRs) are spatially organized and function within tumors remains poorly defined. To assess whether intratumoral location influences clonotype-specific T cell states, we profiled TCR repertoires across blood and tumor samples from 37 patients with HRNB using longitudinal bulk TCR sequencing. In a nested subset of 5 patients with paired pre- and post-therapy tumors, we integrated spatial transcriptomics with in situ TCR profiling. Across all tumors, T and B cells preferentially co-localized in immune-rich regions and showed reduced proximity to neuroblast cells. Despite this compartmentalized architecture, γδT cells were more evenly distributed across tumor sections and showed greater proximity to neuroblast-rich regions than other T cell subsets. Within TCR clonotypes, spatial location was associated with distinct transcriptional states, with immune-rich regions supporting more progenitor-like programs. These findings identify spatial context as a key determinant of phenotype clonotype-specific T cell phenotype and highlight γδT cells cells as a spatially distinct population with potential roles in neuroblastoma tumor-immune interactions.

Repository Structure

neuroblastoma_spatial_tcr/
├── README.md
├── requirements.txt
├── manuscript_sourcedata/ # source data in Excel format
├── data/
│   ├── README.md
│   └── processed/ # source data for figure generation in .csv format
└── scripts/
    ├── 1.process_and_integrate.ipynb
    ├── 2.subcluster_and_visualize_metadata.ipynb
    ├── 3.calculate_celltype_proportions_and_neighborhoods.ipynb
    ├── 4.correlate_distance_and_genelists.ipynb
    ├── 5.find_domains.ipynb
    ├── 6.tcr_bulk/
    │   ├── 1_run_mixcr_command_example.sh
    │   ├── 2_immunarch_tcrs.R
    │   └── 3_make_sankey_plot.ipynb
    └── 7.find_tcr_clonotypes.ipynb

Overview of Notebooks and Scripts

1. process_and_integrate
   Preprocessing, integration, and harmonization of spatial transcriptomics data.

2. subcluster_and_visualize_metadata
   Subclustering and metadata-based visualization of spatially resolved cell populations.

3. calculate_celltype_proportions_and_neighborhoods
   Computes cell type proportions and spatial neighborhood metrics/enrichment.

4. correlate_distance_and_genelists
   Spatial distance calculations and correlation with gene signatures.

5. find_domains
   Identification of latent spatial domains using scimap.

6. tcr_bulk/
   CapTCR-seq processing and clonotype anlaysis pipeline including MiXCR4, Immunarch and TCR probe visualization.

7. find_tcr_clonotypes
   Clonotype-aware spatial analysis.

Software Requirements

Python Environment

See requirements.txt for Python package dependencies.

External Tools

MiXCR v4.6.0 (https://github.com/milaboratory/mixcr)

Xenium Explorer v4.1

R v4.4.0

• immunarch v0.10.3
• ggplot2 v4.0.0
• dplyr v1.1.4

Reproducibility

This repository supports two levels of reproducibility.

1. Full analysis reproduction from the original .h5ad file.
   The .h5ad file is not included because it contains controlled-access patient-derived data that will be used in a future manuscript, but it is available from the authors upon reasonable request.

2. Figure reproduction from processed CSV files.
   All processed .csv files required to regenerate the manuscript figures are provided in data/processed/.

The manuscript source data Excel files are provided in manuscript_sourcedata/. These files are intended for journal source-data submission and reader inspection. The canonical code-readable files are the .csv files in data/processed/.