Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2024 May 30:16:1352681.
doi: 10.3389/fnagi.2024.1352681. eCollection 2024.

Identification and validation of a novel Parkinson-Glioma feature gene signature in glioma and Parkinson's disease

Hengrui Zhang  1   2 , Jiwei Wang  1   2 , Nan Su  1   2 , Ning Yang  1   2 , Xinyu Wang  1   2 , Chao Li  1   2
Affiliations

Identification and validation of a novel Parkinson-Glioma feature gene signature in glioma and Parkinson's disease

Hengrui Zhang et al. Front Aging Neurosci. .

Abstract

Introduction: The prognosis for glioma is generally poor, and the 5-year survival rate for patients with this disease has not shown significant improvement over the past few decades. Parkinson's disease (PD) is a prevalent movement disorder, ranking as the second most common neurodegenerative disease after Alzheimer's disease. Although Parkinson's disease and glioma are distinct diseases, they may share certain underlying biological pathways that contribute to their development.

Objective: This study aims to investigate the involvement of genes associated with Parkinson's disease in the development and prognosis of glioma.

Methods: We obtained datasets from the TCGA, CGGA, and GEO databases, which included RNA sequencing data and clinical information of glioma and Parkinson's patients. Eight machine learning algorithms were used to identify Parkinson-Glioma feature genes (PGFGs). PGFGs associated with glioma prognosis were identified through univariate Cox analysis. A risk signature was constructed based on PGFGs using Cox regression analysis and the Least Absolute Shrinkage and Selection Operator (LASSO) method. We subsequently validated its predictive ability using various methods, including ROC curves, calibration curves, KM survival analysis, C-index, DCA, independent prognostic analysis, and stratified analysis. To validate the reproducibility of the results, similar work was performed on three external test datasets. Additionally, a meta-analysis was employed to observe the heterogeneity and consistency of the signature across different datasets. We also compared the differences in genomic variations, functional enrichment, immune infiltration, and drug sensitivity analysis based on risk scores. This exploration aimed to uncover potential mechanisms of glioma occurrence and prognosis.

Results: We identified 30 PGFGs, of which 25 were found to be significantly associated with glioma survival. The prognostic signature, consisting of 19 genes, demonstrated excellent predictive performance for 1-, 2-, and 3-year overall survival (OS) of glioma. The signature emerged as an independent prognostic factor for glioma overall survival (OS), surpassing the predictive performance of traditional clinical variables. Notably, we observed differences in the tumor microenvironment (TME), levels of immune cell infiltration, immune gene expression, and drug resistance analysis among distinct risk groups. These findings may have significant implications for the clinical treatment of glioma patients.

Conclusion: The expression of genes related to Parkinson's disease is closely associated with the immune status and prognosis of glioma patients, potentially regulating glioma pathogenesis through multiple mechanisms. The interaction between genes associated with Parkinson's disease and the immune system during glioma development provides novel insights into the molecular mechanisms and targeted therapies for glioma.

Keywords: Parkinson’s disease; gene signature; glioma; machine learning algorithms; prognosis.

PubMed Disclaimer

Conflict of interest statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Figures

Figure 1
Figure 1
Identification of Parkinson-Glioma feature genes. GSE7621: 9 normal brain samples and 16 PD brain samples GSE20141: 8 normal brain samples and 10 PD brain samples GSE49036: 8 normal brain samples and 20 PD brain samples. (A) The boxplot illustrated the distribution of gene expression before the removal of batch effects; (B) Principal Component Analysis (PCA) demonstrated expression patterns before the elimination of batch effects; (C) The boxplot depicted the distribution of gene expression after batch effect removal; (D) PCA displayed expression patterns following batch effect removal; (E) Box plots of sample residuals from the eight algorithms were presented. The x-axis represented the quantile of outliers, with the red dot indicating the mean; (F) Eight different algorithms identified the top 10 significant genes, resulting in the discovery of 30 Parkinson-Glioma feature genes; (G) Reverse Cumulative Distribution Maps of model residuals were constructed using RF, SVM, XGB, GLM, Elastic Net, stepLDA, PLS, and msaenet. The y-axis represented the outlier percentile.
Figure 2
Figure 2
Construction and validation of the prognostic signature. TCGA-Glioma: 631 glioma patients, 315 cases in the high-risk group and 316 cases in the low-risk group CGGA-693: 618 glioma patients, 365 cases in the high-risk group and 253 cases in the low-risk group CGGA-325: 306 glioma patients, 178 cases in the high-risk group and 128 cases in the low-risk group GSE16011: 249 glioma patients, 171 cases in the high-risk group and 78 cases in the low-risk group The significance of the survival curve was evaluated using the log-rank test. (A) Univariate Cox analysis identified 25 prognostic genes; (B,C) Coefficient profiles of the 19 prognostic PGFGs obtained through Lasso-Cox regression analysis. The Lasso regression model revealed the partial likelihood deviance of variables. Red dots represented the partial likelihood of deviance values, and gray lines represented the standard error (SE). The two vertical dotted lines on the left and right symbolized optimal values based on minimum criteria and 1 − SE criteria, respectively; (D) Meta-analysis demonstrated the heterogeneity and consistency of the signature across TCGA-Glioma, CGGA-693, CGGA-325, and GSE16011 datasets; (E–H) KM curves illustrated the differences in OS between the high-risk and low-risk groups in the TCGA-Glioma, CGGA-693, CGGA-325, and GSE16011 datasets; (I–L) Calibration curves denoted the accuracy and specificity of the signature. ROC curves displayed the 1-, 2-, and 3-year OS in the TCGA-Glioma (M), CGGA-693 (N), CGGA-325 (O), and GSE16011 (P) datasets.
Figure 3
Figure 3
Internal validation of the signature. The univariate (A) and multivariate (B) Cox regression analyses demonstrated the independent prognostic value of the signature for Glioma patients (p < 0.001); (C) The time-dependent C-index indicated the accuracy of the signature; (D) The DCA curves emphasized the potential clinical benefits of the signature for Glioma patients.
Figure 4
Figure 4
Immune infiltration profiles. The main methods involved were Wilcoxon rank-sum test for difference analysis and Spearman correlation analysis. (A) Eight immune infiltration software tools showed varying quantities of immune cells between the high-risk and low-risk groups; (B) Correlation between immune cells and risk scores; (C–J) Correlation between immune cells and genes incorporated in the model.
Figure 5
Figure 5
Mining of network modules using WGCNA. The main method involved was Pearson correlation analysis. (A) Heatmap of the network for all genes; (B) Heatmap showing the correlation between module eigengenes and risk traits; (C) Correlation and significance of the red module with the low-risk group of Glioma patients; (D,E) GO and KEGG analysis of genes in the red module; (F) Correlation and significance of the magenta module with the high-risk group of Glioma patients; (G,H) GO and KEGG analysis of genes in the magenta module.
Figure 6
Figure 6
Exploration of potential risk mechanisms. The main methods involved were Wilcoxon rank-sum test for difference analysis, GSEA for enrichment analysis, and Spearman correlation analysis. (A) Comparative analysis of four distinct categories of immune-related genes, namely Immunoinhibitor genes, Chemokines, Immunostimulator genes, and Human Leukocyte Antigen, between high- and low-risk groups; (B) Bar charts represented the GSEA results; (C) Butterfly plots vividly illustrated the correlation between risk score, TIP score, eight immunotherapy scores, and various tumor signaling pathways.
Figure 7
Figure 7
Chemotherapy and immunotherapy. The main methods involved were Wilcoxon rank-sum test for difference analysis and Spearman correlation analysis. (A–H) The scores of eight different types of immunotherapies between high-risk and low-risk groups; (I–L) The IC50 of four chemotherapy drugs between high-risk and low-risk groups; (M–P) The IC50 values of the four chemotherapy drugs showed a significant negative correlation with the risk score.
See this image and copyright information in PMC

Similar articles

See all similar articles

References

    1. Al-Astal H. I., Massad M., AlMatar M., Ekal H. (2016). Cellular functions of RNA-binding motif protein 3 (RBM3): clues in hypothermia, Cancer biology and apoptosis. Protein Pept. Lett. 23, 828–835. doi: 10.2174/0929866523666160628090340, PMID: - DOI - PubMed
    1. Aran D., Hu Z., Butte A. J. (2017). xCell: digitally portraying the tissue cellular heterogeneity landscape. Genome Biol. 18:220. doi: 10.1186/s13059-017-1349-1, PMID: - DOI - PMC - PubMed
    1. Armstrong M. J., Okun M. S. (2020). Diagnosis and treatment of Parkinson disease: a review. JAMA 323, 548–560. doi: 10.1001/jama.2019.22360 - DOI - PubMed
    1. Becht E., Giraldo N. A., Lacroix L., Buttard B., Elarouci N., Petitprez F., et al. . (2016). Estimating the population abundance of tissue-infiltrating immune and stromal cell populations using gene expression. Genome Biol. 17:218. doi: 10.1186/s13059-016-1070-5, PMID: - DOI - PMC - PubMed
    1. Becker C., Brobert G. P., Johansson S., Jick S. S., Meier C. R. (2010). Cancer risk in association with Parkinson disease: a population-based study. Parkinsonism Relat. Disord. 16, 186–190. doi: 10.1016/j.parkreldis.2009.11.005 - DOI - PubMed

Grants and funding

The author(s) declare financial support was received for the research, authorship, and/or publication of this article. This research was funded by National Natural Science Foundation of China (82203760) and Shandong Provincial Natural Science Foundation (ZR2020QH182, ZR2022MH155, and ZR2020MH158).