Integrated analysis of genome-wide DNA methylation and cancer-associated fibroblasts identified prognostic biomarkers and immune checkpoint blockade in lower grade gliomas

Jiawei Dong, Fang Wang, Xin Gao, Hongtao Zhao, Jiheng Zhang, Nan Wang, Zhihui Liu, Xiuwei Yan, Jiaqi Jin, Yixu Ba, Shuai Ma, Jianyang Du, Hang Ji, Shaoshan HuFrontiers in Oncology 2023 · DOI

Abstract

Our study reveals characteristics of cancer-associated fibroblasts (CAFs) in lower-grade gliomas (LGGs), refines the direct link between epigenetics and tumor stroma, and provides clinical implications for guiding tailored anti-CAF therapy in combination with immunotherapy for LGG patients.

Introduction

Lower-grade gliomas (LGGs) are slow-growing brain tumors, but they are notoriously unpredictable, often progressing into highly aggressive glioblastomas. A major driver of this progression is the tumor microenvironment (TME), specifically cancer-associated fibroblasts (CAFs). These cells act as the "architects" of the tumor's surroundings, promoting growth, drug resistance, and immune evasion.

While DNA methylation (an epigenetic "off switch" for genes) plays a role in glioma behavior, the specific link between these epigenetic changes and the presence of CAFs was poorly understood. This study sought to bridge this gap by identifying genes that are both aberrantly methylated and differentially expressed in CAF-rich environments to create a more reliable clinical toolkit for prognosis and treatment selection.

Methodology

The researchers employed an integrated computational and experimental workflow to develop a robust prognostic signature.

Study Flow Chart: The integrated workflow used to identify DMEGs and build the prognostic model.
  1. Stratification: LGG samples from public databases (TCGA-LGG, CGGA, GTEx) were grouped into "high-CAF" and "low-CAF" categories based on cellular abundance scores.
  2. Multi-omics Integration: Researchers identified genes that were both Differentially Expressed (DEGs) and Differentially Methylated (DMGs) between these groups. These overlapping genes (DMEGs) represent the intersection of epigenetic regulation and stromal influence.
  3. Signature Building: Using Univariate Cox and LASSO regression, the list was narrowed to a robust five-gene signature: GSAP, EMP3, LATS2, SWAP70, and SLC2A10.
  4. Validation: The model's predictive power was tested across multiple independent datasets and verified via Immunohistochemistry (IHC) on patient tissues and in vitro cell line experiments.

Key Results and Biological Insights

The five-gene risk score proved to be a highly significant predictor of survival. In the training set, the Area Under the Curve (AUC) for predicting 1-year survival was 0.86, and it remained strong in validation sets (AUC ~0.73–0.79).

Biological Pathways and PCA: PCA shows clear separation between LGG and normal tissue, while the chord diagram highlights key GO terms like Wnt signaling and cell adhesion.
  • Epigenetic-Stromal Crosstalk: The study reinforces the idea that the tumor's epigenetic state is significantly correlated with the density of CAFs in the surrounding stroma. Identified DMEGs were heavily involved in the Wnt signaling pathway and Epithelial-Mesenchymal Transition (EMT), hallmarks of cancer progression.
  • Peripheral Invasion: IHC staining showed that the five signature proteins were more highly expressed in the tumor periphery than the tumor center, suggesting these genes are actively involved in the tumor's "front line" as it invades healthy brain tissue.
  • Immune Landscape: High-risk patients had higher Tumor Mutational Burden (TMB) and more "tumor-promoting" immune cells (like M2 macrophages). Paradoxically, while these tumors look "hotter," they are more immunosuppressed, making them prime candidates for specific immunotherapy combinations.
  • Therapeutic Sensitivity: Low-risk patients showed higher sensitivity to Immune Checkpoint Blockade (ICB). The study also identified significant differences in sensitivity to Axitinib, GDC-0941 (a PI3K inhibitor), and Bleomycin between high- and low-risk groups.

Discussion and Future Directions

While the results are promising, the authors acknowledge limitations, including the use of standard glioma cell lines that do not perfectly replicate human LGG architecture and the need for clinical validation of predicted drug sensitivities. Future research will focus on in vivo co-culture models to further unravel the mechanical and chemical signaling that drives this epigenetic-stromal crosstalk. This five-gene signature serves as a potential roadmap for identifying patients who might benefit most from combined immunotherapy and targeted anti-CAF treatments.

Summary updated

Abstract

Our study reveals characteristics of CAFs in LGGs, refines the direct link between epigenetics and tumor stroma, and might provide clinical implications for guiding tailored anti-CAFs therapy in combination with immunotherapy for LGGs patients.
BibTeX
@inproceedings{dong2023integrated,
  author = {Jiawei Dong and Fang Wang and Xin Gao and Hongtao Zhao and Jiheng Zhang and Nan Wang and Zhihui Liu and Xiuwei Yan and Jiaqi Jin and Yixu Ba and Shuai Ma and Jianyang Du and Hang Ji and Shaoshan Hu},
  title = {{Integrated analysis of genome-wide DNA methylation and cancer-associated fibroblasts identified prognostic biomarkers and immune checkpoint blockade in lower grade gliomas}},
  booktitle = {Frontiers in Oncology},
  year = {2023},
  doi = {10.3389/fonc.2022.977251},
}
Figure