Unveiling a Novel Cancer Hallmark by Evaluation of Neural Infiltration in Cancer

Immunology Cell Biology Oncology Genetics Life Sciences Medicine
neural infiltration cancer hallmark tumor microenvironment single-cell RNA sequencing immunotherapy

Cancer, as a major global public health challenge, is characterized by complex mechanisms underlying its onset and progression. For a long time, processes within the tumor microenvironment (TME)—such as immunity, inflammation, and angiogenesis—have been extensively studied and considered key determinants of tumor biological behavior. In recent years, cancer neuroscience has emerged as a new interdisciplinary field, uncovering that the nervous system not only regulates tumor development via neurotransmitters and neuropeptides, but also affects tumor growth, metastasis, and invasion through direct or indirect interactions between neurons and tumor cells. Although nerve-related phenomena such as perineural invasion (PNI) have attracted attention, systematic quantification and evaluation of the prevalence, molecular characteristics, and clinical significance of “neural infiltration” in cancer remain at an early stage. The present report addresses these scientific gaps, aiming to investigate whether neural factors can serve as a novel cancer hallmark, and provide new perspectives for cancer therapy and precision stratification.

Source and Author Information

This paper, entitled “Unveiling a novel cancer hallmark by evaluation of neural infiltration in cancer”, was authored by a team mainly from Harbin Medical University (China), spanning multiple departments such as Life Science, Bioinformatics, Pharmacy, and the “State Key Laboratory of Frigid Zone Cardiovascular Diseases (SKLFZCD).” Corresponding authors are Yunyan Gu, Haihai Liang, and Yang Hui. The article was published in 2025 in the internationally renowned journal Briefings in Bioinformatics (Vol. 26, 2025), and is open-access. The author team integrates multidisciplinary strengths from systems biology, molecular biology, pharmacology, and clinical medicine.

I. Research Background and Problem Statement

1. The Rise of Cancer Neuroscience

The interplay between the nervous system and cancer cells is being progressively revealed, involving neurotransmitters, neuropeptides (such as BDNF, NLgn3, etc.), exosomes, synaptic structures, and more. Literature indicates that neurons and the factors they release can directly promote tumor growth, invasion, metastasis, participate in mechanisms of drug resistance, and even influence patients’ response to immunotherapy.

2. The Limitations and Challenges of Perineural Invasion

Although PNI is regarded as a hallmark of malignant progression and poor prognosis—common in pancreatic cancer, head and neck tumors, etc.—the sensitivity of traditional H&E staining for PNI detection is limited. Moreover, even in tumors without detected PNI, nerve-related factors can still influence patient survival. Thus, PNI alone cannot capture the full spectrum of neural influence across different cancer types.

3. Gaps in Existing Research and Unmet Scientific Needs

Currently, there is a lack of systematic studies providing a panoramic assessment of neural infiltration in cancer. How to extract neural signals from multi-omics data, establish quantitative indicators, and clarify their relationship with tumor malignancy, molecular subtypes, immune microenvironment, and drug sensitivity is the breakthrough pursued in this research.

II. Overall Research Design and Technical Approach

Centered on the hypothesis that “neural infiltration is a novel cancer hallmark,” the authors conducted large-scale, pan-cancer, multi-layered omics data analysis, complemented by single-cell sequencing, spatial transcriptomics, and co-culture biological experiments. The overall workflow includes:

  1. Integration of neural gene sets and screening of cancer-related differentially expressed genes
  2. Development of a neural infiltration quantitative indicator (c-neural score) and its application across modalities
  3. Comprehensive multi-omics and multi-cohort data mining across cancer types, associating with clinical, survival, staging, and grading information
  4. Single-cell and multi-tumor microenvironment heterogeneity resolution and subpopulation characterization
  5. Focused spatial, subpopulation, and cell interaction analyses in key cancers (e.g. pancreatic ductal adenocarcinoma and lung adenocarcinoma)
  6. In vitro co-culture experiments verifying key pathway molecules, such as Schwann cell-driven tumor promotion
  7. Prediction of immunotherapy response and potential drug sensitivity based on neural infiltration score

Below is a detailed breakdown of each step.

1. Integration of Neural Gene Set and Screening for Differential Expression

The study first manually curated 1,889 neural-related genes from four authoritative databases and literature to form a neural gene pool. For 10 solid tumor types (covering 40 paired bulk RNA-seq datasets), the one-sided Wilcoxon rank-sum test was applied to identify differentially expressed neural genes (DEG-NS) specific to each cancer type, establishing the molecular foundation for subsequent quantification.

2. Development and Multi-omics Adaptation of Neural Infiltration Score

The team proposed a new indicator, the Cancer-related Neural Infiltration Score (c-neural score).
- At the bulk RNA-seq level, gene set variation analysis (GSVA) computes enrichment scores for DEG-NS, and the difference between upregulated and downregulated scores is used.
- At the single-cell RNA-seq level, the Ucell algorithm calculates cell-level c-neural scores, enabling precise profiling of neural signals per cell.
This generalizable indicator is compatible with multiple data platforms, allowing cross-cohort and cross-sample type quantitative comparisons.

3. Pan-cancer Analysis Associating with Clinical Parameters

Using samples from TCGA and multiple public databases, along with information on PNI status, staging, recurrence/metastasis, tumor size, survivorship, etc., the study systematically analyzes the relationship between c-neural score and these multidimensional clinical factors.

4. Single-cell Resolution Analysis of Cellular Heterogeneity

The study incorporated 55 single-cell RNA-seq datasets across 10 solid malignancies, encompassing 28 major cell types. Standardization and batch synchronization were conducted using R packages Seurat and Harmony. Ucell scores were applied at the single-cell level, and the proportion, distribution, and enrichment analyses of high/low scoring populations were performed within each cell type to dissect TME neural signal heterogeneity.

5. In-Depth Investigation of Pancreatic and Lung Adenocarcinoma Models

  • For Pancreatic Ductal Adenocarcinoma (PDAC): Special analysis of Schwann cell-enriched regions, spatial transcriptomics, and tumor epithelial subpopulations. Combined with CopyKat inference of copy number variation, Cytotrace-derived stemness, and AUCell-evaluated EMT status, a precise portrait is built for “high-neural-score” epithelial cells (epi-highCNS), highlighting their oncogenic features.
  • For Lung Adenocarcinoma (LUAD): Using integrated single-cell atlases, the study differentiates tumor subpopulations with high neural scores and their interactions with CD8+ T cells, as well as correlation with differentiation status and metabolic pathway enrichment.

6. Analysis of Cell-cell and Protein-protein Interaction Networks

By integrating CellChat and the Pathway Commons database, the research explored key signaling pathways, ligand-receptor pairs, and PPIs (protein-protein interactions) between subgroups (e.g., epi-highCNS and Schwann cells), identifying critical communicative axes such as FN1-related and collagen-related signaling.

7. In Vitro Biological Function Validation

PANC-1 (pancreatic cancer) and A549 (lung cancer) cells were co-cultured with Schwann cells in a Transwell system. VDAC1 (Voltage-dependent Anion Channel 1) was knocked down using siRNA, observing subsequent changes in tumor cell proliferation, migration, invasion, and EMT. Some experiments also used Schwann cell-conditioned medium to distinguish between direct contact and secreted factor effects.

8. Neural Infiltration Score for Predicting Immunotherapy Response and Drug Sensitivity

Multi-omics data pre-immunotherapy were analyzed to compare c-neural scores across responders and non-responders. Drug sensitivity was predicted using Oncopredict and pharmacogenomics databases, with model validation assessing the correlation between neural infiltration context and susceptibility to cytotoxic/targeted therapies.

III. Detailed Presentation of Main Findings

1. Neural Infiltration Score Strongly Indicates Tumor Malignancy

Among 10 cancer types, high c-neural scores were significantly associated with PNI positivity (e.g., head and neck squamous cell carcinoma), high grade/stage, tumor recurrence, metastasis, and poorer survival outcomes. High-score groups also showed greater tumor purity, tumor cell content, and larger tumor size, suggesting that neural signaling is a novel biological marker of malignant behavior.

2. Heterogeneity of Neural Signals in the TME and Features of epi-highCNS

Large-scale single-cell analyses showed that in the vast majority of cancers, tumor epithelial subpopulations were significantly enriched in high c-neural score groups. Further exploration in PDAC revealed that epi-highCNS is associated with elevated CNVs, higher stemness (lower differentiation), stronger oxidative phosphorylation (OXPHOS), and EMT activity, suggesting this subpopulation represents “nerve-dependent” highly malignant tumor cells. Spatial transcriptomics confirmed that Schwann cell-enriched and high c-neural score regions corresponded to PNI high-expression areas.

3. Schwann Cells Promote Tumor Progression through the FN1 Axis and VDAC1

Cell-cell interaction analyses identified intensive communication between epi-highCNS and Schwann cells via the FN1 (fibronectin) signaling axis and collagen ligand-receptor pairs, with overlapping protein interaction networks mainly enriched in extracellular matrix–receptor interaction pathways. Patients with high expression of key gene pairs (such as APOD_VDAC1) had worse survival. In vitro experiments showed that Schwann cell-driven promotion of tumor growth, migration, and EMT relies on high VDAC1 expression; this molecule is primarily expressed in epithelial cells, with higher VDAC1 levels observed in PNI-positive patients.

4. Neural High-score Tumor Cells Promote Antitumor Immune Responses; c-neural Score Predicts Immunotherapy Efficacy

Analysis of pre-immunotherapy clinical samples (e.g., PD-1/PD-L1 therapy) suggested that responders generally exhibit higher c-neural scores. Both single-cell and bulk data from multiple cohorts revealed that high scorers have increased CD8+ T cell, activated memory CD4+ T cell, and M1 macrophage infiltration. Immune checkpoint molecules (PD-L1, HLA genes) positively correlated with c-neural score, indicating a strong link between neural signaling and antitumor immune benefit. The c-neural score also outperformed many established biomarkers in predicting immunotherapy response (as measured by AUC).

5. Novel Marker for Guiding Personalized Drug Selection and Prediction

By mining drug sensitivity databases and Oncopredict models, the study found multiple drugs whose efficacy correlated closely with the c-neural score: high-score patients may be more sensitive to Axitinib and Olaparib, while Afatinib and Trametinib showed lower IC50 values among high-score malignant melanoma samples, indicating that patients with strong neural signaling may derive extra benefit from specific chemotherapies and targeted agents.

IV. Conclusions and Scientific/Application Value

1. Scientific Significance

  • Neural infiltration already features characteristics of a “novel cancer hallmark”, with its quantification able to predict malignancy, metastasis, prognosis, and immunotherapy response across cancer types, enhancing molecular subtyping and precision oncology.
  • Single-cell neural signal profiling reveals TME diversity and novel oncogenic subpopulations (epi-highCNS), opening new directions for studies on tumor heterogeneity and mechanisms of carcinogenesis.
  • The novel Schwann cell–epithelial–VDAC1 axis finding elucidates a new mechanism of neural-tumor interaction, serving as a paradigm for deeper histological and functional investigations.
  • Coupling between tumor neural signal and the immune microenvironment opens new windows for understanding immune evasion mechanisms and predicting therapeutic benefit.

2. Application Prospects

  • Early diagnosis, classification, and prognosis assessment. The scalability of the c-neural score makes it highly promising for clinical subtyping and risk stratification.
  • Prediction of immunotherapy and targeted therapy sensitivity. It guides the development of individualized treatment plans to improve patient outcomes.
  • Development and validation of new drugs/targets (e.g., VDAC1, FN1 pathway, etc.).
  • A paradigm integrating spatial multi-omics and functional validation, driving systematic integration and translation of complex TME signaling in cancer research.

V. Highlights and Innovations of the Study

  • First systematic scoring system quantifying neural infiltration across cancer types;
  • Single-cell-resolution, multi-layered, and cross-sample enrichment and heterogeneity analysis of neural signals;
  • Neural infiltration score shows multimodal predictive clinical value: disease progression, immunotherapy benefit, and drug sensitivity;
  • Clear proposal and experimental validation of the Schwann cell–VDAC1 oncogenic pathway;
  • A fully integrated multi-omics, spatial, and in vitro functional research paradigm.

VI. Additional Valuable Information

  • All data sources in this study are from public databases, and the algorithmic code and scoring parameters should be theoretically reproducible and transferable to other analysis pipelines.
  • Ethically compliant; all patient samples were approved and obtained with informed consent.
  • The paper provides recommendations for future research in spatial multi-omics, in vivo functional validation, and more complex cell–cell interaction networks, encouraging peers to further expand research at the intersection of neuroscience, oncology, and immunology.

Summary

This study refines the mechanisms of neuro-tumor interaction, revealing at molecular, cellular, and clinical levels the central role of neural infiltration in cancer occurrence, progression, and therapeutic response. Its innovative c-neural score system and multimodal experimental design open up new avenues for foundational and clinical cancer research, providing important impetus for future precision oncology and interdisciplinary medical research.