A novel risk model based on the correlation between the expression of basement membrane genes and immune infiltration to predict the invasiveness of pituitary adenomas

Zheng Chen, Xin Sun, Yin Kang, Jian Zhang, Fang Jia, Xiyao Liu, Hongwei ZhuFrontiers in Endocrinology 2023 · DOI

Abstract

Using a variety of bioinformatics methods, we developed a novel risk model to predict the probability of PAs invasion based on the correlation between 11 BM genes and immune infiltration. These findings may facilitate closer surveillance and early diagnosis to prevent or treat IPAs in patients and improve the clinical awareness of patients at high risk of IPAs.

Introduction

Pituitary adenomas (PAs) represent a common class of endocrine tumors. While many follow a benign clinical course, a significant subset—Invasive Pituitary Adenomas (IPAs)—demonstrate aggressive behavior, infiltrating adjacent structures such as the cavernous sinus or bone. These invasive variants present substantial clinical challenges, often proving resistant to complete surgical resection and exhibiting high recurrence rates.

A critical factor in tumor progression is the degradation of the basement membrane (BM), a specialized extracellular matrix that serves as a physical barrier against tumor spread. This study explores the hypothesis that alterations in BM gene expression, in conjunction with the tumor’s immune microenvironment, can provide a molecular "fingerprint" to predict invasiveness.

Methodology

The researchers implemented a comprehensive "dry-lab to wet-lab" pipeline to identify and validate predictive biomarkers.

Flowchart showing the study process, from data acquisition to clinical validation.

Data Integration and Screening

The study integrated four transcriptomic datasets from the Gene Expression Omnibus (GEO): GSE51618, GSE26966, GSE169498, and GSE22812. These datasets provided a robust pool of samples comparing Invasive PAs with Non-Invasive PAs (NIPAs). Using the "limma" and "sva" packages, the authors identified 248 Differentially Expressed Genes (DEGs).

Network and Immune Analysis

To isolate the most relevant factors, a Protein-Protein Interaction (PPI) network was used to narrow the DEGs down to a core cluster of 11 BM-related genes. The immune landscape was then quantified using the ssGSEA algorithm, allowing for an analysis of how these 11 genes correlate with immune cell infiltration.

Model Construction and Validation

A predictive nomogram was developed to calculate individual risk scores. The model's accuracy was evaluated using Receiver Operating Characteristic (ROC) curves. Finally, the findings were validated through qPCR on 10 fresh clinical samples (5 IPA, 5 NIPA) from The First Affiliated Hospital of Xiamen University.

Results and Key Insights

The analysis revealed a complex interplay between the structural integrity of the basement membrane and the tumor's immune evasion strategies.

Heatmap showing differential gene expression across the integrated GEO datasets.

A significant finding was the strong correlation between specific BM genes and immune cell types. For instance, SDC4 showed a high positive correlation with dendritic cells (DCs), while LAMA2 was negatively correlated with T follicular helper (Tfh) cells. This suggests that BM degradation is biologically linked to immune recruitment or evasion, rather than being a purely mechanical process.

Identification of Key Biomarkers

Among the 11 identified genes (SPARCL1, GPC3, LAMA1, SDC4, GPC4, ADAMTS8, LAMA2, LAMC3, SMOC1, LUM, and THBS2), THBS2 (Thrombospondin 2) emerged as the most influential variable in the risk model. qPCR results confirmed that THBS2, SDC4, and LUM were expressed at significantly lower levels in invasive tissues compared to non-invasive ones (P < 0.05). This suggests these genes may normally act as stabilizers whose loss facilitates tumor aggression.

Predictive Performance and Drug Repurposing

The resulting risk model achieved an AUC of 0.7886, demonstrating good discriminative power. Furthermore, using the DSigDB database, the study identified 10 potential therapeutic compounds, including Dasatinib and Mifepristone, which could potentially be repurposed to treat IPAs by targeting the identified molecular pathways.

Discussion and Limitations

While the study provides a promising roadmap for early diagnosis, several limitations remain. The clinical validation was restricted to a small cohort of 10 patients, necessitating larger, multi-center studies to confirm the model's robustness. Additionally, as a retrospective bioinformatics-based study, it establishes correlation rather than causation. Future functional "wet-lab" experiments, such as gene knockdown assays, are required to elucidate the exact molecular mechanisms.

Despite these hurdles, this novel risk model offers a valuable tool for identifying high-risk patients, potentially leading to more personalized surveillance and treatment strategies for aggressive pituitary adenomas.

Summary updated

Abstract

Using a variety of bioinformatics methods, we developed a novel risk model to predict the probability of PAs invasion based on the correlation between 11 BM genes and immune infiltration. These findings may facilitate closer surveillance and early diagnosis to prevent or treat IPAs in patients and improve the clinical awareness of patients at high risk of IPAs.
BibTeX
@inproceedings{chen2023novel,
  author = {Zheng Chen and Xin Sun and Yin Kang and Jian Zhang and Fang Jia and Xiyao Liu and Hongwei Zhu},
  title = {{A novel risk model based on the correlation between the expression of basement membrane genes and immune infiltration to predict the invasiveness of pituitary adenomas}},
  booktitle = {Frontiers in Endocrinology},
  year = {2023},
  doi = {10.3389/fendo.2022.1079777},
}
Figure