Expression of intra-tumoral necrosis-associated cytokine pattern correlated with prognosis and immune status in glioma

Hongtao Zhao, Jiawei Dong, Jiheng Zhang, Nan Wang, Zhihui Liu, Xiuwei Yan, Fang Wang, Hang Ji, Shaoshan HuFrontiers in Molecular Neuroscience 2023 · DOI

TL;DR

Researchers have identified an 11-gene cytokine signature associated with intra-tumoral necrosis (ITN) in glioma. This molecular "fingerprint" not only predicts significantly shorter patient survival but also reveals an immunosuppressive tumor environment that, paradoxically, may be more responsive to specific chemotherapies and PD-1 immunotherapy.

What Problem Does This Paper Address?

Glioma is a devastating form of brain cancer, and the presence of intra-tumoral necrosis (ITN)—areas of dead tissue within the tumor—is a well-known red flag for poor prognosis. Traditionally, pathologists identify ITN by looking for "pseudopalisades" (crowded cells surrounding dead zones) under a microscope.

However, histological diagnosis has limits: a small biopsy might miss necrotic zones entirely, and visual inspection doesn't explain the underlying molecular chaos. There is a significant gap in understanding how the molecules released by dying cells—specifically cytokines—reshape the tumor's biology and the patient's immune response. The authors sought to define a robust molecular signature of necrosis to better predict patient outcomes and therapeutic needs.

Methods

The study utilized a multi-stage approach combining animal modeling with large-scale human data analysis:

  1. The Mouse Model: The researchers used the GL261 syngeneic mouse model, which is unique because it develops necrosis over time. They compared "non-necrotic" tumors (Day 10) against "necrotic" tumors (Day 20) to isolate the changes specifically triggered by tissue death.
  2. Transcriptomics & Enrichment: They performed whole-tumor mRNA sequencing on the mouse samples. Using Gene Ontology (GO) enrichment analysis, they identified differentially expressed genes (DEGs) related to cytokine activity.
  3. Human Validation: They cross-referenced these mouse findings with human data from The Cancer Genome Atlas (TCGA).
  4. Clustering: Using the identified cytokine pattern, they performed consensus clustering to divide 616 human glioma patients into a High-Expression Group (HEG) and a Low-Expression Group (LEG).
  5. In Silico Prediction: They used algorithms like pRRophetic to predict chemotherapy sensitivity and submap to predict response to immune checkpoint blockades (ICB).

Datasets

  • Primary Experimental Data: Original RNA-sequencing data from GL261-C57BL/6J mouse glioma models (Day 10 vs. Day 20).
  • Human Validation Data: 616 glioma samples retrieved from The Cancer Genome Atlas (TCGA), including transcriptomic, pathological, and clinical survival data.

Key Insights and Ideas

  • The "Necrosis Fingerprint": The study narrowed down a signature of 11 cytokines: nine pro-inflammatory (including IL6, CXCL10, and CCL8) and two anti-inflammatory (IL1RN and IL10). This suggests that necrosis isn't just "passive" cell death; it's an active signaling event.
  • Invasive Transformation: Histology showed that as necrosis appears (Day 20 in mice), the tumor border transforms from a clean line to a "disturbed" boundary with "tumor islands," suggesting necrosis acts as a catalyst for invasion.
  • The Immunological Paradox: Patients in the High-Expression Group (HEG) had more infiltrated immune cells (T-cells, macrophages, etc.), but lower actual anti-tumor activity. The necrotic environment appears to recruit immune cells only to "exhaust" or suppress them via high expression of immune checkpoints like PD-1 and CTLA-4.
  • Molecular "Stealth" Necrosis: The authors suggest this 11-gene signature could identify high-risk patients even when a physical biopsy fails to capture necrotic tissue, acting as a molecular proxy for tumor grade.

Key Results

  • Survival Gap: The median survival for the HEG (High-Expression Group) was significantly shorter at 1.09 years, compared to 1.90 years for the LEG.
  • Clinical Correlation: HEG was strongly associated with high-grade tumors (Grade 4/GBM), IDH-wildtype status (a deadlier subtype), and older age (>45 years).
  • Therapeutic Silver Lining: Despite the poor prognosis, the HEG was predicted to be more responsive to several chemotherapies (Temozolomide, Paclitaxel, Bortezomib) and specifically to PD-1 blockade therapy. This suggests that the very inflammation driven by necrosis might make the tumor more "visible" to certain drugs.

Limitations and Future Work

The authors acknowledge that while the GL261 mouse model is a reliable tool for studying ITN, it is still a single model. Gliomas are notoriously heterogeneous, and these cytokine patterns may vary across different genetic subtypes.

Furthermore, the therapeutic responsiveness results are based on computational predictions (in silico). The authors emphasize that future work must involve prospective clinical trials and single-cell sequencing to confirm which specific cells within the tumor microenvironment are producing these cytokines. Understanding the "who's who" of cytokine production could lead to targeted therapies that flip the switch from immunosuppression to active tumor killing.

Summary updated

Abstract

Intra-tumoral necrosis (ITN) is reported to be an independent prognostic factor in glioma. However, knowledge of ITN is mainly limited to pseudopalisadwe, while its other aspects were neglected. Therefore, a deeper understanding of ITN could be valuable for understanding its exact role in glioma. The only reliable ITN model was time-dependently achieved with the GL261 syngeneic mouse model. The ITN-associated expression pattern was enriched from RNA sequencing. TCGA glioma samples were clustered into a high-expression group (HEG) and a low-expression group (LEG) based on their pattern and their association with prognosis, clinical status, immune status, and therapeutic responsiveness were compared. Mouse glioma with ITN demonstrated invasive histology. Cytokine signaling was significantly enriched in necrotic mouse glioma compared with non-necrotic glioma tissues. Nine pro-inflammatory (IL6, PPBP, IL1A, TNFSF11, CXCL11, CXCL9, CXCL10, CXCL3, and CCL8) and two anti-inflammatory cytokine (IL1RN and IL10) genes were found to be related to ITN-associated cytokine patterns. Comparative analysis showed that HEG had a significantly shorter survival time, five differentially distributed clinical statuses, more infiltrated immune cells, greater expression of immune checkpoints, and better therapeutic responsiveness than LEG. In conclusion, the ITN-associated cytokine pattern is characteristically expressed in glioma with ITN and might indicate necrosis missed in histology diagnosis. Its expression pattern could predict the prognosis, tumor grade, immune status, and therapeutic responsiveness of glioma patients.
BibTeX
@inproceedings{zhao2023expression,
  author = {Hongtao Zhao and Jiawei Dong and Jiheng Zhang and Nan Wang and Zhihui Liu and Xiuwei Yan and Fang Wang and Hang Ji and Shaoshan Hu},
  title = {{Expression of intra-tumoral necrosis-associated cytokine pattern correlated with prognosis and immune status in glioma}},
  booktitle = {Frontiers in Molecular Neuroscience},
  year = {2023},
  doi = {10.3389/fnmol.2023.1117237},
}
Figure