Development and Validation of a Prognostic Signature for Malignant Pleural Mesothelioma

Our plain-English summary of Development and Validation of a Prognostic Signature for Malignant Pleural Mesothelioma, by J.G. Zhou, H. Zhong, J. Zhang, et al (Frontiers in Oncology, 2019).

The study

• Authors: J.G. Zhou, H. Zhong, J. Zhang, S.H. Jin, R. Roudi, H. Ma
• Journal: Frontiers in Oncology
• Published: 2019;9:78
• Cited by: 24 articles, as counted by NIH iCite
• Read it: PubMed · Publisher · Free full text

Abstract

Introduction: Dysregulated genes play a critical role in the development and progression of cancer, suggesting their potential as novel independent biomarkers for cancer diagnosis and prognosis. Prognostic model-based gene expression profiles are not widely utilized in clinical medicine. We investigated the prognostic significance of an expression profile-based gene signature for outcome prediction in patients with malignant pleural mesothelioma (MPM). Methods: The gene expression profiles of a large cohort of patients with MPM were obtained and analyzed by repurposing publicly available microarray data. A gene-based risk score model was developed with the training dataset and then validated with the TCGA-MESO (mesothelioma) dataset. The time-dependent receiver operating characteristic (ROC) curve was used to evaluate the prognostic performance of survival prediction. The biological function of the prognostic genes was predicted using bioinformatics analysis. Results: Three genes in the training dataset (GSE2549) were identified as significantly associated with the overall survival (OS) of patients with MPM and were combined to develop a three-gene prognostic signature to stratify patients into low-risk and high-risk groups. The MPM patients of the training dataset in the low-risk group exhibited longer OS than those in the high-risk group (HR = 0.25, 95% CI = 0.11-0.56, P < 0.001). Similar prognostic values for the three-gene signature were observed in the validated TCGA-MESO cohort (HR = 0.53 95% CI = 0.33-0.85, P = 0.008). ROC analysis also demonstrated the good performance in predicting 3-year OS in the GEO and TCGA cohorts (KM-AUC for GEO = 0.989, KM-AUC for TCGA = 0.618). The C-statistic for the 3-gene model was 0.761. Validation with TCGA-MESO confirmed the model's ability to discriminate between risk groups in an alternative data set with fair performance (C-statistic: 0.68). Functional enrichment analysis suggested that these three genes may be involved in genetic and epigenetic events with known links to MPM. Conclusions: This study has identified and validated a novel 3-gene model to reliably discriminate patients at high and low risk of death in unselected populations of patients with MPM. Further larger, prospective multi-institutional cohort studies are necessary to validate this model.

In plain English

This study used public gene activity data from tumors of people with malignant pleural mesothelioma (MPM) to build and test a survival-prediction score. It did not study asbestos or any other exposure. The authors built the score from one dataset, GSE2549, which had 39 tumor samples with survival data plus nine normal pleura and lung samples. They then tested it on a second dataset, TCGA-MESO, with 84 tumors. From 8 candidate genes, three were kept: LSM6, GZMB and HJURP. In the first dataset, patients scored low-risk had a median survival of 622.5 days versus 243 days for high-risk patients (HR 0.25, 95% CI 0.11 to 0.56). In the test dataset the figures were 512.5 versus 347 days (HR 0.53, 95% CI 0.33 to 0.85). A nine-gene score built from genes named in earlier studies showed no statistically significant survival difference in the test dataset (P = 0.732). The authors concluded that the three-gene model discriminated high- and low-risk patients, and that larger prospective studies are needed. They noted small samples, missing tumor stage data, and different laboratory platforms between the two datasets.

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: Retrospective bioinformatics analysis of publicly available gene expression data, developing a prognostic gene signature in one dataset and validating it in another
• Who was studied: Training set GSE2549: 40 MPM surgical specimens (39 used after removing one without complete survival data), 5 normal pleura and 4 normal lung specimens. Validation set TCGA-MESO: 84 tumor tissues. Locations and dates of the patients are not stated.
• Exposure: No asbestos or other environmental exposure was studied. The factor examined was tumor gene expression, combined into a risk score from three genes (LSM6, GZMB, HJURP). Patients were split into high- and low-risk groups at the median score.
• Main results:
  • In the training set, 22 genes were associated with overall survival at P of 0.01 or less. A false discovery rate filter reduced these to 8 candidate genes.
  • Stepwise selection produced a three-gene model: LSM6, GZMB and HJURP.
  • Training set (GSE2549): low-risk patients had a median overall survival of 622.5 days versus 243 days for high-risk patients (HR = 0.25, 95% CI = 0.11 to 0.56, P less than 0.001).
  • Validation set (TCGA-MESO, 42 low-risk and 42 high-risk patients): median overall survival was 512.5 days for the low-risk group versus 347 days for the high-risk group (HR = 0.53, 95% CI = 0.33 to 0.85, P = 0.008).
  • A nine-gene model built from genes reported in earlier literature separated the risk groups in the training set (HR = 0.36, 95% CI = 0.17 to 0.77, P = 0.020).
  • The nine-gene model showed no statistically significant survival difference in the TCGA-MESO set (median 470 days versus 458 days, HR = 0.84, 95% CI = 0.53 to 1.34, P = 0.732).
  • For predicting 3-year survival, the three-gene model had KM-AUC values of 0.989 (GSE2549) and 0.618 (TCGA-MESO). The nine-gene model had 0.921 and 0.457.
  • The C-statistic for the three-gene model was 0.761 in the training set and 0.68 in TCGA-MESO, which the authors described as fair performance.
  • Functional enrichment analysis found 61 GO terms and 10 pathways, including the cell cycle and DNA replication.
• The authors' conclusion: The authors concluded that they had identified and validated a three-gene model to discriminate MPM patients at high and low risk of death. They stated that larger, prospective multi-institutional studies are needed to validate it.
• Limitations the authors state:
  • The findings must be validated in a prospective study with independent patient samples.
  • The clinical data lacked TNM stage and other information, which limited adjustment of the signature's predictive power.
  • TCGA-MESO had no normal tissue, so the signature was identified only from the GEO dataset. Other signatures, such as lncRNA or microRNA, might have been missed.
  • Small sample size limited testing of the signature.
  • The training data came from a microarray and the validation data from sequencing, which the authors said might bias results.

Topics

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

Related studies

• Identification of a novel prognostic gene signature in pleural mesothelioma: a study based on The Cancer Genome Atlas database and experimental validation — X. Wang, Y. Yang, et al. (2025)
• Development and validation of a prognostic nomogram for patients with malignant peritoneal mesothelioma — X. Wang, R. Liu, et al. (2025)
• A gene signature linked to fibroblast differentiation for prognostic prediction of mesothelioma — J. Liu, Y. Lu, et al. (2024)
• Genomic and transcriptomic analyses identify a prognostic gene signature and predict response to therapy in pleural and peritoneal mesothelioma — N.U. Nair, Q. Jiang, et al. (2023)
• Identification of prognostic and bone metastasis​-related alternative splicing signatures in mesothelioma — R. Huang, Z. Zheng, et al. (2021)
• Comprehensive analysis to identify PUS7 as a prognostic biomarker from pan-cancer analysis to osteosarcoma validation — B. Dong, B. Wang, et al. (2024)

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