Identification of glycolysis genes signature for predicting prognosis in malignant pleural mesothelioma by bioinformatics and machine learning

Our plain-English summary of Identification of glycolysis genes signature for predicting prognosis in malignant pleural mesothelioma by bioinformatics and machine learning, by Y. Xiao, W. Huang, L. Zhang, et al (Frontiers in Endocrinology, 2022).

The study

• Authors: Y. Xiao, W. Huang, L. Zhang, H. Wang
• Journal: Frontiers in Endocrinology
• Published: 2022;13:1056152
• Cited by: 8 articles, as counted by NIH iCite
• Read it: PubMed · Publisher · Free full text

Abstract

Background: Glycolysis-related genes as prognostic markers in malignant pleural mesothelioma (MPM) is still unclear. We hope to explore the relationship between glycolytic pathway genes and MPM prognosis by constructing prognostic risk models through bioinformatics and machine learning.

Methods: The authors screened the dataset GSE51024 from the GEO database for Gene set enrichment analysis (GSEA), and performed differentially expressed genes (DEGs) of glycolytic pathway gene sets. Then, Cox regression analysis was used to identify prognosis-associated glycolytic genes and establish a risk model. Further, the validity of the risk model was evaluated using the dataset GSE67487 in GEO database, and finally, a specimen classification model was constructed by support vector machine (SVM) and random forest (RF) to further screen prognostic genes.

Results: By DEGs, five glycolysis-related pathway gene sets (17 glycolytic genes) were identified to be highly expressed in MPM tumor tissues. Also 11 genes associated with MPM prognosis were identified in TCGA-MPM patients, and 6 (COL5A1, ALDH2, KIF20A, ADH1B, SDC1, VCAN) of them were included by Multi-factor COX analysis to construct a prognostic risk model for MPM patients, with Area under the ROC curve (AUC) was 0.830. Further, dataset GSE67487 also confirmed the validity of the risk model, with a significant difference in overall survival (OS) between the low-risk and high-risk groups (P < 0.05). The final machine learning screened the five prognostic genes with the highest risk of MPM, in order of importance, were ALDH2, KIF20A, COL5A1, ADH1B and SDC1.

Conclusions: A risk model based on six glycolytic genes (ALDH2, KIF20A, COL5A1, ADH1B, SDC1, VCAN) can effectively predict the prognosis of MPM patients.

Copyright © 2022 Xiao, Huang, Zhang and Wang.

In plain English

This study used public gene-activity data from malignant pleural mesothelioma (MPM) tumours to test whether glycolysis genes (genes for a sugar-burning energy pathway in cells) could help predict survival. It did not study asbestos exposure. Comparing 55 MPM and 41 normal tissue samples (GSE51024), the authors found 17 glycolysis genes whose activity differed. In 84 MPM cases from The Cancer Genome Atlas, 11 were linked to overall survival. Six of them (COL5A1, ALDH2, KIF20A, ADH1B, SDC1 and VCAN) were combined into a risk score. Patients with high scores had significantly lower survival than those with low scores, and the score's area under the curve (a measure of accuracy) was 0.830. In a separate set of 33 MPM cases (GSE67487), the low-risk group again survived significantly better (area under the curve 0.782). A machine-learning method ranked five genes as most important for telling MPM tumour tissue apart from normal tissue. The authors concluded that the six-gene model can effectively predict MPM prognosis. They noted the data came mainly from North American populations and laboratory studies are still needed.

We wrote this plain-English summary from the full article with the help of AI, and a second AI model checked each sentence against the article. It is not the authors' wording; the abstract and the article are their own account.

What the study did and found

• Design: Bioinformatics analysis of publicly available gene expression datasets. Methods were gene set enrichment analysis, Cox regression to build a prognostic risk model, validation in a second dataset, and machine learning (support vector machine and random forest).
• Who was studied: Malignant pleural mesothelioma (MPM) tissue data from public databases. The Cancer Genome Atlas supplied 84 MPM cases, after 3 cases with missing survival or sequencing data were removed. GEO dataset GSE51024 supplied 55 MPM and 41 normal samples, and GEO dataset GSE67487 supplied 33 MPM samples (per Table 1). The authors state that the databases mainly included North American populations. Dates of sample collection were not given.
• Exposure: Asbestos exposure was not studied. The study examined the expression levels of glycolysis-related genes in tumour tissue: 312 genes from 5 glycolysis pathway gene sets. Asbestos is mentioned only in the introduction, as background from other studies.
• Main results:
  • In GSE51024 (MPM versus normal tissue), all five glycolysis-related gene sets differed significantly and were positively correlated with MPM tissue. Of their 312 genes, 17 were differentially expressed between MPM and normal tissue.
  • In TCGA MPM cases, 11 glycolysis-related genes were significantly associated with overall survival in single-factor Cox analysis (P < 0.05). Six genes (COL5A1, ALDH2, KIF20A, ADH1B, SDC1, VCAN) were then included in the multi-factor risk model.
  • In the multi-factor analysis (Table 2), ALDH2 had a hazard ratio of 0.777 (95% CI 0.559-1.081, P = 0.134). This was not statistically significant, although the gene was kept in the model. The other five genes had P values from 0.008 to 0.049. Examples: COL5A1 HR 1.627 (1.107-2.393), KIF20A HR 1.401 (1.092-1.798), VCAN HR 0.666 (0.462-0.961).
  • In TCGA MPM cases, the high-risk group had significantly lower survival than the low-risk group (P < 0.05). The risk score's area under the ROC curve was 0.830. Survival was also significantly lower in the high-risk group within age and tumour-stage subgroups.
  • In the separate GSE67487 MPM dataset, the low-risk group had significantly higher survival than the high-risk group, with an area under the ROC curve of 0.782 (P < 0.05).
  • In classifying MPM tumour tissue versus normal tissue, the random forest model was more accurate than the support vector machine (AUC 0.957 vs 0.776).
  • The random forest gene-importance ranking reflected each gene's contribution to classifying MPM tumour tissue versus normal tissue. The five most important genes, in order, were ALDH2, KIF20A, COL5A1, ADH1B and SDC1.
• The authors' conclusion: The authors concluded that a risk model based on six glycolytic genes (ALDH2, KIF20A, COL5A1, ADH1B, SDC1, VCAN) can effectively predict the prognosis of MPM patients. They also stated that the model is reference information for clinicians and does not represent an absolutely accurate prognosis.
• Limitations the authors state:
  • The databases used mainly included North American populations, so the model's validity outside North America needs further validation.
  • Further laboratory (in vitro) functional analysis in MPM cell lines is needed to understand the role of the identified genes.
  • The model is reference information for clinicians and does not give an absolutely accurate prognosis.

Topics

The National Library of Medicine indexes this study under subjects our site covers: Mesothelioma.

Medical terms in this study, explained in our Lexicon: Prognosis, Pleural Mesothelioma, Overall Survival.

Related studies

• Cancer-Associated Fibroblasts Influence Survival in Pleural Mesothelioma: Digital Gene Expression Analysis and Supervised Machine Learning Model — S. Borchert, A. Mathilakathu, et al. (2023)
• The use of a next-generation sequencing-derived machine-learning risk-prediction model (OncoCast-MPM) for malignant pleural mesothelioma: a retrospective study — M.G. Zauderer, A. Martin, et al. (2021)
• Prediction of Radiation Pneumonitis With Machine Learning in Stage III Lung Cancer: A Pilot Study — M. Yakar, D. Etiz, et al. (2021)
• Distant Metastatic Pattern and Its Prognostic Significance in Malignant Pleural Mesothelioma: A Population-Based Study Based on a Machine Learning Model — J. Yu, C. Peng, et al. (2025)
• Cancer Risk Prediction Using Machine Learning for Supporting Early Cancer Diagnosis in Symptomatic Patients: A Systematic Review of Model Types — F. Pennisi, S. Borlini, et al. (2025)
• Near-infrared spectroscopy combined with machine learning for plasma-based discriminant diagnosis of malignant mesothelioma: a retrospective study — Y. Gu, R. Wang, et al. (2025)

Browse more research in Frontiers in Endocrinology, or return to medical research.