A systems biology-based gene expression classifier of glioblastoma predicts survival with solid tumors

Identification of potential key genes associated with glioblastoma based on the gene expression profile

Gliomas are serious primary brain tumors. The aim of the present study was to identify potential key genes associated with the progression of gliomas. The GSE31262 gene expression profile data, which included 9 glioblastoma stem cells (GSCs) samples and 5 neural stem cell samples from adult humans, were downloaded from Gene Expression Omnibus (GEO) database. limma package was used to identify differentially expressed genes (DEGs). Based on STRING database and Pearson Correlation Coefficient (PCC), a co-expression network was constructed to comprehensively understand the interactions between DEGs, and function analysis of genes in the network was conducted. Furthermore, the DEGs that were associated with prognosis were analyzed. A total of 431 DEGs were identified, including 98 upregulated DEGs and 333 downregulated DEGs. Genes including PDZ binding kinase, topoisomerase (DNA) II α (TOP2A), cyclin dependent kinase (CDK) 1, cell division cycle 6 and NIMA related kinase 2 had a relatively high degree in the co-expression network. A set of genes including cyclin D1, CDK1 and CDK2 were significantly enriched in the cell cycle and p53 signaling pathway. Additionally, 69 DEGs were identified as genes involved in glioblastoma prognosis, such as CDK2 and TOP2A. The genes that had a higher degree and were associated with cell cycle and p53 signaling pathway may play pivotal roles in the progress of glioblastoma.

Use of the cell cycle progression (CCP) score for predicting systemic disease and response to radiation of biochemical recurrence

BACKGROUND

Determining the optimal treatment for biochemical recurrence (BCR) after radical prostatectomy (RP) is challenging.

OBJECTIVE

We evaluated the ability of CCP score (a prognostic RNA expression signature) to discriminate between systemic disease and local recurrence in patients with BCR after RP.

METHODS

Sixty patients with BCR after RP were selected for analysis based on: 1) metastatic disease, 2) non-response to salvage external beam radiotherapy (EBRT), and 3) durable response to salvage EBRT. CCP scores were generated from the RNA expression of 46 genes. Logistic regression assessed the association between CCP score and patient group.

RESULTS

Passing CCP scores were generated for 47 patients with complete clinical and pathologic data. CCP score predicted clinical status when comparing patients with metastatic disease or non-responders to salvage therapy to patients with durable response (p = 0.006). CCP score remained significantly predictive of clinical status after accounting for time to BCR, PSA level at BCR, and Gleason score (p = 0.0031).

CONCLUSIONS

Elevated CCP score was associated with increased risk of systemic disease, indicating that CCP score may be useful in identifying patients with BCR who are most likely to benefit from salvage radiation therapy.

Combined expressional analysis, bioinformatics and targeted proteomics identify new potential therapeutic targets in glioblastoma stem cells

Glioblastoma (GBM) is both the most common and the most lethal primary brain tumor. It is thought that GBM stem cells (GSCs) are critically important in resistance to therapy. Therefore, there is a strong rationale to target these cells in order to develop new molecular therapies.

To identify molecular targets in GSCs, we compared gene expression in GSCs to that in neural stem cells (NSCs) from the adult human brain, using microarrays. Bioinformatic filtering identified 20 genes (PBK/TOPK, CENPA, KIF15, DEPDC1, CDC6, DLG7/DLGAP5/HURP, KIF18A, EZH2, HMMR/RHAMM/CD168, NOL4, MPP6, MDM1, RAPGEF4, RHBDD1, FNDC3B, FILIP1L, MCC, ATXN7L4/ATXN7L1, P2RY5/LPAR6 and FAM118A) that were consistently expressed in GSC cultures and consistently not expressed in NSC cultures. The expression of these genes was confirmed in clinical samples (TCGA and REMBRANDT). The first nine genes were highly co-expressed in all GBM subtypes and were part of the same protein-protein interaction network. Furthermore, their combined up-regulation correlated negatively with patient survival in the mesenchymal GBM subtype. Using targeted proteomics and the COGNOSCENTE database we linked these genes to GBM signalling pathways.

Nine genes: PBK, CENPA, KIF15, DEPDC1, CDC6, DLG7, KIF18A, EZH2 and HMMR should be further explored as targets for treatment of GBM.

The Impact of Age and Sex in DLBCL: Systems Biology Analyses Identify Distinct Molecular Changes and Signaling Networks

Potential molecular alterations based on age and sex are not well defined in diffuse large B-cell lymphoma (DLBCL). We examined global transcriptome DLBCL data from The Cancer Genome Atlas (TCGA) via a systems biology approach to determine the molecular differences associated with age and sex. Collectively, sex and age revealed striking transcriptional differences with older age associated with decreased metabolism and telomere functions and female sex was associated with decreased interferon signaling, transcription, cell cycle, and PD-1 signaling. We discovered that the key genes for most groups strongly regulated immune function activity. Furthermore, older females were predicted to have less DLBCL progression versus older males and young females. Finally, analyses in systems biology revealed that JUN and CYCS signaling were the most critical factors associated with tumor progression in older and male patients. We identified important molecular perturbations in DLBCL that were strongly associated with age and sex and were predicted to strongly influence tumor progression.

Cardiomyocyte cytokinesis score: a potential method for cardiomyocyte proliferation

One of the most important indicators of myocardial regeneration is cardiomyocyte proliferation. However, it is difficult to distinguish cardiomyocytes in the regenerating stage from binucleated or multinucleated myocytes by conventional morphometric techniques. As cell cycle progression (CCP) scores have been successfully applied to the evaluation of the proliferation of cancer cells, we sought to establish a multi-gene score to evaluate cardiomyocyte proliferation in this study. Given the disturbances of nuclear division without cell division that occurs in cardiomyocytes, ten cytokinesis-correlated genes (Anln, Aurkb, Cenpa, Kif4, Kif23, Prc1, RhoA, Spin1, TACC2, and CDC42) were chosen to establish the cardiomyocyte cytokinesis score (CC-Score). The expression levels of these genes in H9C2 rat cardiomyoblast cells, the proliferation of which were stimulated or inhibited, were detected using qRT-PCR. To confirm the feasibility of the CC-Score system, four conventional methods for evaluating cardiomyocyte proliferation, including the MTT assay, BrdU assay, immunofluorescence, and flow cytometry analysis, were used in each group. The results of the CC-Score in the assessment of the proliferation of H9C2 cells were consistent with those of four commonly used proliferative assay methods. We conclude that the CC-Score can be used to assess the proliferation status of H9C2 cells, and suggest that the CC-Score may be a potential method for the assessment of cardiomyocyte proliferation in myocardial regeneration. However, validation studies utilizing primary cultured rat cardiomyocytes and heart tissue are warranted.

An integrated mRNA and microRNA expression signature for glioblastoma multiforme prognosis

Although patients with Glioblastoma multiforme (GBM) have grave prognosis, significant variability in patient outcome is observed. The objective of this study is to identify a molecular signature for GBM prognosis. We subjected 355 mRNA and microRNA expression profiles to elastic net-regulated Cox regression for identification of an integrated RNA signature for GBM prognosis. A prognostic index (PI) was generated for patient stratification. Survival comparison was conducted by Kaplan-Meier method and a general multivariate Cox regression procedure was applied to evaluate the independence of the PI. The abilities and efficiencies of signatures to predict GBM patient outcome was assessed and compared by the area under the curve (AUC) of the receiver-operator characteristic (ROC). An integrated RNA prognostic signature consisted by 4 protective mRNAs, 12 risky mRNAs, and 1 risky microRNA was identified. Decreased survival was associated with being in the high-risk group (hazard ratio = 2.864, P<0.0001). The prognostic value of the integrated signature was validated in five independent GBM expression datasets (n = 201, hazard ratio = 2.453, P<0.0001). The PI outperformed the known clinical factors, mRNA-only, and miRNA-only prognostic signatures for GBM prognosis (area under the ROC curve for the integrated RNA, mRNA-only, and miRNA-only signatures were 0.828, 0.742, and 0.757 at 3 years of overall survival, respectively, P<0.0001 by permutation test). We describe the first, to our knowledge, robust transcriptome-based integrated RNA signature that improves the current GBM prognosis based on clinical variables, mRNA-only, and miRNA-only signatures.

Network signatures of survival in glioblastoma multiforme

To determine a molecular basis for prognostic differences in glioblastoma multiforme (GBM), we employed a combinatorial network analysis framework to exhaustively search for molecular patterns in protein-protein interaction (PPI) networks. We identified a dysregulated molecular signature distinguishing short-term (survival<225 days) from long-term (survival>635 days) survivors of GBM using whole genome expression data from The Cancer Genome Atlas (TCGA). A 50-gene subnetwork signature achieved 80% prediction accuracy when tested against an independent gene expression dataset. Functional annotations for the subnetwork signature included “protein kinase cascade,” “IκB kinase/NFκB cascade,” and “regulation of programmed cell death” – all of which were not significant in signatures of existing subtypes. Finally, we used label-free proteomics to examine how our subnetwork signature predicted protein level expression differences in an independent GBM cohort of 16 patients. We found that the genes discovered using network biology had a higher probability of dysregulated protein expression than either genes exhibiting individual differential expression or genes derived from known GBM subtypes. In particular, the long-term survivor subtype was characterized by increased protein expression of DNM1 and MAPK1 and decreased expression of HSPA9, PSMD3, and CANX. Overall, we demonstrate that the combinatorial analysis of gene expression data constrained by PPIs outlines an approach for the discovery of robust and translatable molecular signatures in GBM.

Glioblastoma multiforme (GBM) is the most common and aggressive brain tumor in adults, and, while the median survival time for treated patients is approximately one year, subgroups of patients respond differently to the same treatments, with some patients showing little improvement and other patients living far longer than expected. These differences in treatment response indicate that the tumors may show molecular differences that we can harness to tailor cancer therapy. To this end, we sought to identify biomarkers of patient survival in GBM. To improve the applicability of our molecular markers to other patient groups, we constrained our markers using maps of protein-protein interactions, and we also employed a unique computational strategy that incorporates patient-to-patient molecular variability into the results. We identified a set of 50 genes comprising a subnetwork signature that successfully separated GBM patients by their survival times. Our approach to identifying this subnetwork signature also improved our ability to identify its protein products in an independent cohort of patients. In the ongoing search to improve cancer detection and treatment, our work represents a successful strategy for identifying reproducible biomarkers that can more efficiently lead to the discovery of druggable protein targets.

A fourteen gene GBM prognostic signature identifies association of immune response pathway and mesenchymal subtype with high risk group

BACKGROUND

Recent research on glioblastoma (GBM) has focused on deducing gene signatures predicting prognosis. The present study evaluated the mRNA expression of selected genes and correlated with outcome to arrive at a prognostic gene signature.

METHODS

Patients with GBM (n = 123) were prospectively recruited, treated with a uniform protocol and followed up. Expression of 175 genes in GBM tissue was determined using qRT-PCR. A supervised principal component analysis followed by derivation of gene signature was performed. Independent validation of the signature was done using TCGA data. Gene Ontology and KEGG pathway analysis was carried out among patients from TCGA cohort.

RESULTS

A 14 gene signature was identified that predicted outcome in GBM. A weighted gene (WG) score was found to be an independent predictor of survival in multivariate analysis in the present cohort (HR = 2.507; B = 0.919; p<0.001) and in TCGA cohort. Risk stratification by standardized WG score classified patients into low and high risk predicting survival both in our cohort (p = <0.001) and TCGA cohort (p = 0.001). Pathway analysis using the most differentially regulated genes (n = 76) between the low and high risk groups revealed association of activated inflammatory/immune response pathways and mesenchymal subtype in the high risk group.

CONCLUSION

We have identified a 14 gene expression signature that can predict survival in GBM patients. A network analysis revealed activation of inflammatory response pathway specifically in high risk group. These findings may have implications in understanding of gliomagenesis, development of targeted therapies and selection of high risk cancer patients for alternate adjuvant therapies.

Weighted frequent gene co-expression network mining to identify genes involved in genome stability

Gene co-expression network analysis is an effective method for predicting gene functions and disease biomarkers. However, few studies have systematically identified co-expressed genes involved in the molecular origin and development of various types of tumors. In this study, we used a network mining algorithm to identify tightly connected gene co-expression networks that are frequently present in microarray datasets from 33 types of cancer which were derived from 16 organs/tissues. We compared the results with networks found in multiple normal tissue types and discovered 18 tightly connected frequent networks in cancers, with highly enriched functions on cancer-related activities. Most networks identified also formed physically interacting networks. In contrast, only 6 networks were found in normal tissues, which were highly enriched for housekeeping functions. The largest cancer network contained many genes with genome stability maintenance functions. We tested 13 selected genes from this network for their involvement in genome maintenance using two cell-based assays. Among them, 10 were shown to be involved in either homology-directed DNA repair or centrosome duplication control including the well-known cancer marker MKI67. Our results suggest that the commonly recognized characteristics of cancers are supported by highly coordinated transcriptomic activities. This study also demonstrated that the co-expression network directed approach provides a powerful tool for understanding cancer physiology, predicting new gene functions, as well as providing new target candidates for cancer therapeutics.

Whole miRNome-wide differential co-expression of microRNAs

Co-regulation of genes has been extensively analyzed, however, rather limited knowledge is available on co-regulations within the miRNome. We investigated differential co-expression of microRNAs (miRNAs) based on miRNome profiles of whole blood from 540 individuals. These include patients suffering from different cancer and non-cancer diseases, and unaffected controls. Using hierarchical clustering, we found 9 significant clusters of co-expressed miRNAs containing 2–36 individual miRNAs. Through analyzing multiple sequencing alignments in the clusters, we found that co-expression of miRNAs is associated with both sequence similarity and genomic co-localization. We calculated correlations for all 371,953 pairs of miRNAs for all 540 individuals and identified 184 pairs of miRNAs with high correlation values. Out of these 184 pairs of miRNAs, 16 pairs (8.7%) were differentially co-expressed in unaffected controls, cancer patients and patients with non-cancer diseases. By computing correlated and anti-correlated miRNA pairs, we constructed a network with 184 putative co-regulations as edges and 100 miRNAs as nodes. Thereby, we detected specific clusters of miRNAs with high and low correlation values. Our approach represents the most comprehensive co-regulation analysis based on whole miRNome-wide expression profiling. Our findings further decrypt the interactions of miRNAs in normal and human pathological processes.

Increased immune gene expression and immune cell infiltration in high-grade astrocytoma distinguish long-term from short-term survivors

Survival in the majority of high-grade astrocytoma (HGA) patients is very poor, with only a rare population of long-term survivors. A better understanding of the biological factors associated with long-term survival in HGA would aid development of more effective therapy and survival prediction. Factors associated with long-term survival have not been extensively studied using unbiased genome-wide expression analyses. In the current study, gene expression microarray profiles of HGA from long-term survivors were interrogated for discovery of survival-associated biological factors. Ontology analyses revealed that increased expression of immune function-related genes was the predominant biological factor that positively correlated with longer survival. A notable T cell signature was present within this prognostic immune gene set. Using immune cell-specific gene classifiers, both T cell-associated and myeloid linage-associated genes were shown to be enriched in HGA from long-term versus short-term survivors. Association of immune function and cell-specific genes with survival was confirmed independently in a larger publicly available glioblastoma gene expression microarray data set. Histology was used to validate the results of microarray analyses in a larger cohort of long-term survivors of HGA. Multivariate analyses demonstrated that increased immune cell infiltration was a significant independent variable contributing to longer survival, as was Karnofsky/Lansky performance score. These data provide evidence of a prognostic anti-tumor adaptive immune response and rationale for future development of immunotherapy in HGA.

Why is there a lack of consensus on molecular subgroups of glioblastoma? Understanding the nature of biological and statistical variability in glioblastoma expression data

INTRODUCTION

Gene expression patterns characterizing clinically-relevant molecular subgroups of glioblastoma are difficult to reproduce. We suspect a combination of biological and analytic factors confounds interpretation of glioblastoma expression data. We seek to clarify the nature and relative contributions of these factors, to focus additional investigations, and to improve the accuracy and consistency of translational glioblastoma analyses.

METHODS

We analyzed gene expression and clinical data for 340 glioblastomas in The Cancer Genome Atlas (TCGA). We developed a logic model to analyze potential sources of biological, technical, and analytic variability and used standard linear classifiers and linear dimensional reduction algorithms to investigate the nature and relative contributions of each factor.

RESULTS

Commonly-described sources of classification error, including individual sample characteristics, batch effects, and analytic and technical noise make measurable but proportionally minor contributions to inconsistent molecular classification. Our analysis suggests that three, previously underappreciated factors may account for a larger fraction of classification errors: inherent non-linear/non-orthogonal relationships among the genes used in conjunction with classification algorithms that assume linearity; skewed data distributions assumed to be Gaussian; and biologic variability (noise) among tumors, of which we propose three types.

CONCLUSIONS

Our analysis of the TCGA data demonstrates a contributory role for technical factors in molecular classification inconsistencies in glioblastoma but also suggests that biological variability, abnormal data distribution, and non-linear relationships among genes may be responsible for a proportionally larger component of classification error. These findings may have important implications for both glioblastoma research and for translational application of other large-volume biological databases.

Prognostic value of an RNA expression signature derived from cell cycle proliferation genes in patients with prostate cancer: a retrospective study

BACKGROUND

Optimum management of clinically localised prostate cancer presents unique challenges because of the highly variable and often indolent natural history of the disease. To predict disease aggressiveness, clinicians combine clinical variables to create prognostic models, but the models have limited accuracy. We assessed the prognostic value of a predefined cell cycle progression (CCP) score in two cohorts of patients with prostate cancer.

METHODS

We measured the expression of 31 genes involved in CCP with quantitative RT-PCR on RNA extracted from formalin-fixed paraffin-embedded tumour samples, and created a predefined score and assessed its usefulness in the prediction of disease outcome. The signature was assessed retrospectively in a cohort of patients from the USA who had undergone radical prostatectomy, and in a cohort of randomly selected men with clinically localised prostate cancer diagnosed by use of a transurethral resection of the prostate (TURP) in the UK who were managed conservatively. The primary endpoint was time to biochemical recurrence for the cohort of patients who had radical prostatectomy, and time to death from prostate cancer for the TURP cohort.

FINDINGS

After prostatectomy, the CCP score was useful for predicting biochemical recurrence in the univariate analysis (hazard ratio for a 1-unit change [doubling] in CCP 1·89; 95% CI 1·54-2·31; p=5·6×10(-9)) and the best multivariate analysis (1·77, 1·40-2·22; p=4·3×10(-6)). In the best predictive model (final multivariate analysis), the CCP score and prostate-specific antigen (PSA) concentration were the most important variables and were more significant than any other clinical variable. In the TURP cohort, the CCP score was the most important variable for prediction of time to death from prostate cancer in both univariate analysis (2·92, 2·38-3·57, p=6·1×10(-22)) and the final multivariate analysis (2·57, 1·93-3·43; p=8·2×10(-11)), and was stronger than all other prognostic factors, although PSA concentration also added useful information. Heterogeneity in the hazard ratio for the CCP score was not noted in any case for any clinical variables.

INTERPRETATION

The results of this study provide strong evidence that the CCP score is a robust prognostic marker, which, after additional validation, could have an essential role in determining the appropriate treatment for patients with prostate cancer.

FUNDING

Cancer Research UK, Queen Mary University of London, Orchid Appeal, US National Institutes of Health, and Koch Foundation.

Cell cycle and aging, morphogenesis, and response to stimuli genes are individualized biomarkers of glioblastoma progression and survival

BACKGROUND

Glioblastoma is a complex multifactorial disorder that has swift and devastating consequences. Few genes have been consistently identified as prognostic biomarkers of glioblastoma survival. The goal of this study was to identify general and clinical-dependent biomarker genes and biological processes of three complementary events: lifetime, overall and progression-free glioblastoma survival.

METHODS

A novel analytical strategy was developed to identify general associations between the biomarkers and glioblastoma, and associations that depend on cohort groups, such as race, gender, and therapy. Gene network inference, cross-validation and functional analyses further supported the identified biomarkers.

RESULTS

A total of 61, 47 and 60 gene expression profiles were significantly associated with lifetime, overall, and progression-free survival, respectively. The vast majority of these genes have been previously reported to be associated with glioblastoma (35, 24, and 35 genes, respectively) or with other cancers (10, 19, and 15 genes, respectively) and the rest (16, 4, and 10 genes, respectively) are novel associations. Pik3r1, E2f3, Akr1c3, Csf1, Jag2, Plcg1, Rpl37a, Sod2, Topors, Hras, Mdm2, Camk2g, Fstl1, Il13ra1, Mtap and Tp53 were associated with multiple survival events.Most genes (from 90 to 96%) were associated with survival in a general or cohort-independent manner and thus the same trend is observed across all clinical levels studied. The most extreme associations between profiles and survival were observed for Syne1, Pdcd4, Ighg1, Tgfa, Pla2g7, and Paics. Several genes were found to have a cohort-dependent association with survival and these associations are the basis for individualized prognostic and gene-based therapies. C2, Egfr, Prkcb, Igf2bp3, and Gdf10 had gender-dependent associations; Sox10, Rps20, Rab31, and Vav3 had race-dependent associations; Chi3l1, Prkcb, Polr2d, and Apool had therapy-dependent associations. Biological processes associated glioblastoma survival included morphogenesis, cell cycle, aging, response to stimuli, and programmed cell death.

CONCLUSIONS

Known biomarkers of glioblastoma survival were confirmed, and new general and clinical-dependent gene profiles were uncovered. The comparison of biomarkers across glioblastoma phases and functional analyses offered insights into the role of genes. These findings support the development of more accurate and personalized prognostic tools and gene-based therapies that improve the survival and quality of life of individuals afflicted by glioblastoma multiforme.