Meta-analysis of several gene lists for distinct types of cancer: a simple way to reveal common prognostic markers

Our plain-English summary of Meta-analysis of several gene lists for distinct types of cancer: a simple way to reveal common prognostic markers, by X. Yang, X. Sun (BMC Bioinformatics, 2007).

The study

• Authors: X. Yang, X. Sun
• Journal: BMC Bioinformatics
• Published: 2007;8:118
• Cited by: 24 articles, as counted by NIH iCite
• Read it: PubMed · Publisher · Free full text

Abstract

Background: Although prognostic biomarkers specific for particular cancers have been discovered, microarray analysis of gene expression profiles, supported by integrative analysis algorithms, helps to identify common factors in molecular oncology. Similarities of Ordered Gene Lists (SOGL) is a recently proposed approach to meta-analysis suitable for identifying features shared by two data sets. Here we extend the idea of SOGL to the detection of significant prognostic marker genes from microarrays of multiple data sets. Three data sets for leukemia and the other six for different solid tumors are used to demonstrate our method, using established statistical techniques.

Results: We describe a set of significantly similar ordered gene lists, representing outcome comparisons for distinct types of cancer. This kind of similarity could improve the diagnostic accuracies of individual studies when SOGL is incorporated into the support vector machine algorithm. In particular, we investigate the similarities among three ordered gene lists pertaining to mesothelioma survival, prostate recurrence and glioma survival. The similarity-driving genes are related to the outcomes of patients with lung cancer with a hazard ratio of 4.47 (p = 0.035). Many of these genes are involved in breakdown of EMC proteins regulating angiogenesis, and may be used for further research on prognostic markers and molecular targets of gene therapy for cancers.

Conclusion: The proposed method and its application show the potential of such meta-analyses in clinical studies of gene expression profiles.

In plain English

This statistical methods study analyzed microarray gene data, not asbestos or any exposure. The authors extended a method, Similarities of Ordered Gene Lists (SOGL), to several lists. They applied it to nine published data sets on leukemia, breast, lung and prostate cancer, glioma and mesothelioma of the pleura, comparing good- and poor-outcome groups. One breast comparison used lymph-node status; lung groups used neuroendocrine C2 subtype, which the article inconsistently links to worse and better survival. Among six solid-tumor data sets, 21 of 57 comparisons showed significant similarity. Lists for mesothelioma survival, prostate cancer recurrence and glioma survival were significantly similar (p = 0.024). When these three were combined, SOGL selection gave an area under the ROC curve of 0.747 (95% CI 0.709–0.774), versus 0.665 (95% CI 0.634–0.702) for highest t-score selection. Applied to 125 lung cancer patients, a 5-transcript signature from these genes gave a hazard ratio of 4.47 (p = 0.035); median survival was 26.7 versus 71.5 months. A 17-transcript set was not significant. The authors concluded that poor prognosis sometimes develops independently of original cancer type, and called for larger numbers.

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: Computational meta-analysis of previously published microarray gene expression data sets, extending the SOGL method from two to multiple gene lists
• Who was studied: Nine published Affymetrix microarray data sets: three leukemia (adult T-cell and pediatric), two breast cancer, lung adenocarcinoma (126 cases), prostate, glioma, and mesothelioma. Per Table 1, the mesothelioma data set had 31 samples, 17 with outcome information (8 good, 9 poor). Its good-outcome patients survived more than seventeen months and its poor-outcome patients less than six months. The groups for one breast comparison were defined by lymph-node status. The lung data set had no outcome classification for each case, so its poor-outcome group was the 9 neuroendocrine C2 adenocarcinoma patients. The article is inconsistent about how that subtype relates to survival: the Methods say less favorable survival, while the Results say good outcome.
• Main results:
  • Among the six solid-tumor gene lists, 21 of the 57 possible comparisons showed significant (p < 0.05) similarity
  • The gene lists for mesothelioma survival, prostate recurrence and glioma survival were significantly similar, with an empirical p-value of 0.024
  • When the three solid tumor studies (glioma, mesothelioma, prostate) were combined, the area under the ROC curve was 0.747 (95% CI 0.709–0.774) for SOGL gene selection and 0.665 (95% CI 0.634–0.702) for selection by the highest t-score
  • In the three leukemia studies, the area under the ROC curve was 0.73 (95% CI 0.64–0.76) for SOGL using the z-statistic as effect size and 0.69 (95% CI 0.63–0.71) for the highest t-score
  • The genes driving the mesothelioma–prostate–glioma similarity numbered 17 transcripts when fold-change was used as effect size and 5 transcripts when the z-statistic was used
  • Four of the similarity-driving genes belong to the extracellular matrix (ECM)-receptor interaction pathway, and the authors reported that up-regulation of 7 ECM-related genes was associated with poor cancer outcomes
  • In 125 lung cancer patients divided into two groups using the 5-transcript signature, the hazard ratio was 4.47 (p = 0.035)
  • Median survival after therapy was 26.7 months in the lung cancer poor-prognosis subgroup and 71.5 months in the good-prognosis subgroup
  • The hazard ratio from the 17-transcript set in the lung cancer patients was not significant (p = 0.25)
• The authors' conclusion: The authors concluded that significant similarities span several kinds of cancer, implying that poor prognosis sometimes develops independently of the original cancer type
• Limitations the authors state:
  • The authors stated that larger numbers of samples are required to ensure the findings are reproducible
  • The similarity was not as significant when more than 4 studies were compared, which the authors said shows that similarities across tumor tissues are limited
  • SOGL was not strongly superior to the highest t-score; it only tended to be more accurate
  • The individual genes were not each tested against survival outcomes
  • The authors did not expect strong power to predict lung cancer outcomes

Topics

Medical terms in this study, explained in our Lexicon: Meta-Analysis, Biomarker, Mesothelioma, Lung Cancer, Hazard Ratio, Angiogenesis.

Related studies

• Prognostic and Clinicopathological Significance of BAP1 Protein Expression in Different Types of Cancer-A Meta-Analysis — Z. Wang, X.Y. Wang, et al. (2018)
• Comprehensive analysis reveals prognostic gene set that could impacts the response to chemotherapy and immunotherapy outcomes in malignant pleural mesothelioma — J. Li, H. He, et al. (2025)
• Different prognostic roles of tumor suppressor gene BAP1 in cancer: A systematic review with meta-analysis — C. Luchini, N. Veronese, et al. (2016)
• Genome-wide analysis of abdominal and pleural malignant mesothelioma with DNA arrays reveals both common and distinct regions of copy number alteration — A.C. Borczuk, J. Pei, et al. (2016)
• Prognostic Implications of the Complement Protein C1q in Gliomas — A. Mangogna, B. Belmonte, et al. (2019)
• Proteome screening of pleural effusions identifies galectin 1 as a diagnostic biomarker and highlights several prognostic biomarkers for malignant mesothelioma — F. Mundt, H.J. Johansson, et al. (2014)

Browse return to medical research.