Identification of reproducible gene expression signatures in lung adenocarcinoma

Gene-Set Integrative Analysis of Multi-Omics Data Using Tensor-based Association Test

MOTIVATION

Facilitated by technological advances and the decrease in costs, it is feasible to gather subject data from several omics platforms. Each platform assesses different molecular events, and the challenge lies in efficiently analyzing these data to discover novel disease genes or mechanisms. A common strategy is to regress the outcomes on all omics variables in a gene set. However, this approach suffers from problems associated with high-dimensional inference.

RESULTS

We introduce a tensor-based framework for variable-wise inference in multi-omics analysis. By accounting for the matrix structure of an individual's multi-omics data, the proposed tensor methods incorporate the relationship among omics effects, reduce the number of parameters, and boost the modeling efficiency. We derive the variable-specific tensor test and enhance computational efficiency of tensor modeling. Using simulations and data applications on the Cancer Cell Line Encyclopedia (CCLE), we demonstrate our method performs favorably over baseline methods and will be useful for gaining biological insights in multi-omics analysis.

AVAILABILITY AND IMPLEMENTATION

R function and instruction are available from the authors' website: https://www4.stat.ncsu.edu/∼jytzeng/Software/TR.omics/TRinstruction.pdf.

SUPPLEMENTARY INFORMATION

Supplementary materials are available at Bioinformatics online.

Identification of Prognostic Genes and Pathways in Lung Adenocarcinoma Using a Bayesian Approach

Lung cancer is the leading cause of cancer-associated mortality in the United States and the world. Adenocarcinoma, the most common subtype of lung cancer, is generally diagnosed at the late stage with poor prognosis. In the past, extensive effort has been devoted to elucidating lung cancer pathogenesis and pinpointing genes associated with survival outcomes. As the progression of lung cancer is a complex process that involves coordinated actions of functionally associated genes from cancer-related pathways, there is a growing interest in simultaneous identification of both prognostic pathways and important genes within those pathways. In this study, we analyse The Cancer Genome Atlas lung adenocarcinoma data using a Bayesian approach incorporating the pathway information as well as the interconnections among genes. The top 11 pathways have been found to play significant roles in lung adenocarcinoma prognosis, including pathways in mitogen-activated protein kinase signalling, cytokine-cytokine receptor interaction, and ubiquitin-mediated proteolysis. We have also located key gene signatures such as RELB, MAP4K1, and UBE2C. These results indicate that the Bayesian approach may facilitate discovery of important genes and pathways that are tightly associated with the survival of patients with lung adenocarcinoma.

Identification and analysis of key lncRNAs in malignant-transformed BEAS-2B cells induced with coal tar pitch by microarray analysis

This study aims to explore the key and differentially expressed long non-coding RNAs (lncRNAs) and elucidates their possible mechanisms in malignant-transformed Human bronchial epithelial (BEAS-2B) cells induced by coal tar pitch extracts (CTPE). BEAS-2B cells were stimulated with 2.4 μg/ml CTPE, then passaged for three times which were named CTPE1 and then passaged until passage 30 (CTPE30). The results showed that cells of CTPE30 appeared abnormal morphology. Furthermore, migration, clonality and proliferation of cells in CTPE group were significantly increased compared with those in control groups. However, the apoptosis of cells in CTPE group was inhibited. A total of 569 differentially expressed mRNAs and 707 differentially expressed lncRNAs were screened out, among which four lncRNAs were validated and were consistent with the microarray results. 32 target genes were screened out by Co-expression network. The study suggests that differentially expressed lncRNAs may play a potential role in lung carcinogenesis.

Distinct lung cancer subtypes associate to distinct drivers of tumor progression

The main non–small-cell lung cancer (NSCLC) histopathological subtypes are lung adenocarcinomas (LUAD) and lung squamous cell carcinomas (LUSC). To identify candidate progression determinants of NSCLC subtypes, we explored the transcriptomic signatures of LUAD versus LUSC. We then investigated the prognostic impact of the identified tumor-associated determinants. This was done utilizing DNA microarray data from 2,437 NSCLC patients. An independent analysis of a case series of 994 NSCLC was conducted by next-generation sequencing, together with gene expression profiling from GEO (https://www.ncbi.nlm.nih.gov/geo/).

This work led us to identify 69 distinct tumor prognostic determinants, which impact on LUAD or LUSC clinical outcome. These included key drivers of tumor growth and cell cycle, transcription factors and metabolic determinants. Such disease determinants appeared vastly different in LUAD versus LUSC, and often had opposite impact on clinical outcome. These findings indicate that distinct tumor progression pathways are at work in the two NSCLC subtypes. Notably, most prognostic determinants would go inappropriately assessed or even undetected when globally investigating unselected NSCLC. Hence, differential consideration for NSCLC subtypes should be taken into account in current clinical evaluation procedures for lung cancer.

Identification of a robust subpathway-based signature for acute myeloid leukemia prognosis using an miRNA integrated strategy

Acute myeloid leukemia (AML) is a heterogeneous disease, and survival signatures are urgently needed to better monitor treatment. MiRNAs displayed vital regulatory roles on target genes, which was necessary involved in the complex disease. We therefore examined the expression levels of miRNAs and genes to identify robust signatures for survival benefit analyses. First, we reconstructed subpathway graphs by embedding miRNA components that were derived from low-throughput miRNA-gene interactions. Then, we randomly divided the data sets from The Cancer Genome Atlas (TCGA) into training and testing sets, and further formed 100 subsets based on the training set. Using each subset, we identified survival-related miRNAs and genes, and identified survival subpathways based on the reconstructed subpathway graphs. After statistical analyses of these survival subpathways, the most robust subpathways with the top three ranks were identified, and risk scores were calculated based on these robust subpathways for AML patient prognoses. Among these robust subpathways, three representative subpathways, path: 05200_10 from Pathways in cancer, path: 04110_20 from Cell cycle, and path: 04510_8 from Focal adhesion, were significantly associated with patient survival in the TCGA training and testing sets based on subpathway risk scores. In conclusion, we performed integrated analyses of miRNAs and genes to identify robust prognostic subpathways, and calculated subpathway risk scores to characterize AML patient survival.

An individualized gene expression signature for prediction of lung adenocarcinoma metastases

Our laboratory previously reported an individual‐level signature consisting of nine gene pairs, named 9‐GPS. This signature was developed by training on microarray expression data and validated using three independent integrated microarray data sets, with samples of stage I non‐small‐cell lung cancer after complete surgical resection. In this study, we first validated the cross‐platform robustness of 9‐GPS by demonstrating that 9‐GPS could significantly stratify the overall survival of 213 stage I lung adenocarcinoma (LUAD) patients detected with RNA‐sequencing platform in The Cancer Genome Atlas (TCGA; log‐rank P = 0.0318, C‐index = 0.55). Applying 9‐GPS to all the 423 stage I‐IV LUAD samples in TCGA, the predicted high‐risk samples were significantly enriched with clinically diagnosed metastatic samples (Fisher's exact test, P = 0.0015). We further modified the voting rule of 9‐GPS and found that the modified 9‐GPS had a better performance in predicting metastasis states (Fisher's exact test, P < 0.0001). With the aid of the modified 9‐GPS for reclassifying the metastasis states of patients with LUAD, the reclassified metastatic samples presented clearer transcriptional and genomic characteristics compared to the reclassified nonmetastatic samples. Finally, regulator network analysis identified TP53 and IRF1 with frequent genomic aberrations in the reclassified metastatic samples, indicating their key roles in driving tumor metastasis. In conclusion, 9‐GPS is a robust signature for identifying early‐stage LUAD patients with potential occult metastasis. This occult metastasis prediction was associated with clear transcriptional and genomic characteristics as well as the clinical diagnoses.

Gene Expression Signature Differentiates Histology But Not Progression Status of Early-Stage NSCLC

Advances in molecular analyses based on high-throughput technologies can contribute to a more accurate classification of non-small cell lung cancer (NSCLC), as well as a better prediction of both the disease course and the efficacy of targeted therapies. Here we set out to analyze whether global gene expression profiling performed in a group of early-stage NSCLC patients can contribute to classifying tumor subtypes and predicting the disease prognosis. Gene expression profiling was performed with the use of the microarray technology in a training set of 108 NSCLC samples. Subsequently, the recorded findings were validated further in an independent cohort of 44 samples. We demonstrated that the specific gene patterns differed significantly between lung adenocarcinoma (AC) and squamous cell lung carcinoma (SCC) samples. Furthermore, we developed and validated a novel 53-gene signature distinguishing SCC from AC with 93% accuracy. Evaluation of the classifier performance in the validation set showed that our predictor classified the AC patients with 100% sensitivity and 88% specificity. We revealed that gene expression patterns observed in the early stages of NSCLC may help elucidate the histological distinctions of tumors through identification of different gene-mediated biological processes involved in the pathogenesis of histologically distinct tumors. However, we showed here that the gene expression profiles did not provide additional value in predicting the progression status of the early-stage NSCLC. Nevertheless, the gene expression signature analysis enabled us to perform a reliable subclassification of NSCLC tumors, and it can therefore become a useful diagnostic tool for a more accurate selection of patients for targeted therapies.

Pathway-based gene signatures predicting clinical outcome of lung adenocarcinoma

Lung adenocarcinoma is often diagnosed at an advanced stage with poor prognosis. Patients with different clinical outcomes may have similar clinico-pathological characteristics. The results of previous studies for biomarkers for lung adenocarcinoma have generally been inconsistent and limited in clinical application. In this study, we used inverse-variance weighting to combine the hazard ratios for the four datasets and performed pathway analysis to identify prognosis-associated gene signatures. A total of 2,418 genes were found to be significantly associated with overall survival. Of these, a 21-gene signature in the HMGB1/RAGE signalling pathway and a 31-gene signature in the clathrin-coated vesicle cycle pathway were significantly associated with prognosis of lung adenocarcinoma across all four datasets (all p-values < 0.05, log-rank test). We combined the scores for the three pathways to derive a combined pathway-based risk (CPBR) score. Three pathway-based signatures and CPBR score also had more predictive power than single genes. Finally, the CPBR score was validated in two independent cohorts (GSE14814 and GSE13213 in the GEO database) and had significant adjusted hazard ratios 2.72 (p-value < 0.0001) and 1.71 (p-value < 0.0001), respectively. These results could provide a more complete picture of the lung cancer pathogenesis.

Identification of regulatory SNPs associated with genetic modifications in lung adenocarcinoma

Background

Although much research effort has been devoted to elucidating lung cancer, the molecular mechanism of tumorigenesis still remains unclear. A major challenge to improve the understanding of lung cancer is the difficulty of identifying reproducible differentially expressed genes across independent studies, due to their low consistency. To enhance the reproducibility of the findings, an integrated analysis was performed to identify regulatory SNPs. Thirty-two pairs of tumor and adjacent normal lung tissue specimens were analyzed using Affymetrix U133plus2.0, Affymetrix SNP 6.0, and Illumina Infinium Methylation microarrays. Copy number variations (CNVs) and methylation alterations were analyzed and paired t-tests were used to identify differentially expressed genes.

Results

A total of 505 differentially expressed genes were identified, and their dysregulated patterns moderately correlated with CNVs and methylation alterations based on the hierarchical clustering analysis. Subsequently, three statistical approaches were performed to explore regulatory SNPs, which revealed that the genotypes of 551 and 66 SNPs were associated with CNV and changes in methylation, respectively. Among them, downstream transcriptional dysregulation was observed in 9 SNPs for CNVs and 4 SNPs for methylation alterations.

Conclusions

In summary, these identified SNPs concurrently showed the same direction of gene expression changes with genetic modifications, suggesting their pivotal roles in the genome for non-smoking women with lung adenocarcinoma.

Electronic supplementary material

The online version of this article (doi:10.1186/s13104-015-1053-8) contains supplementary material, which is available to authorized users.

Identification of gene expression biomarkers for predicting radiation exposure

A need for more accurate and reliable radiation dosimetry has become increasingly important due to the possibility of a large-scale radiation emergency resulting from terrorism or nuclear accidents. Although traditional approaches provide accurate measurements, such methods usually require tedious effort and at least two days to complete. Therefore, we provide a new method for rapid prediction of radiation exposure. Eleven microarray datasets were classified into two groups based on their radiation doses and utilized as the training samples. For the two groups, Student's t-tests and resampling tests were used to identify biomarkers, and their gene expression ratios were used to develop a prediction model. The performance of the model was evaluated in four independent datasets, and Ingenuity pathway analysis was performed to characterize the associated biological functions. Our meta-analysis identified 29 biomarkers, showing approximately 90% and 80% accuracy in the training and validation samples. Furthermore, the 29 genes significantly participated in the regulation of cell cycle, and 19 of them are regulated by three well-known radiation-modulated transcription factors: TP53, FOXM1 and ERBB2. In conclusion, this study demonstrates a reliable method for identifying biomarkers across independent studies and high and reproducible prediction accuracy was demonstrated in both internal and external datasets.