pathClass: an R-package for integration of pathway knowledge into support vector machines for biomarker discovery.

UNLABELLED: Prognostic and diagnostic biomarker discovery is one of the key issues for a successful stratification of patients according to clinical risk factors. For this purpose, statistical classification methods, such as support vector machines (SVM), are frequently used tools. Different groups have recently shown that the usage of prior biological knowledge significantly improves the classification results in terms of accuracy as well as reproducibility and interpretability of gene lists. Here, we introduce pathClass, a collection of different SVM-based classification methods for improved gene selection and classfication performance. The methods contained in pathClass do not merely rely on gene expression data but also exploit the information that is carried in gene network data. AVAILABILITY: pathClass is open source and freely available as an R-Package on the CRAN repository at http://cran.r-project.org.

Circulating miRNAs are correlated with tumor progression in prostate cancer.

Circulating miRNAs have recently been indicated as practicable and promising biomarkers for noninvasive diagnosis in various tumor entities. However, cell-free miRNAs have not been found to correlate with clinicopathological variables in epithelial carcinomas. To learn more about the potential clinical relevance of circulating miRNAs in prostate cancer, we screened 667 miRNAs in serum samples from patients with metastatic (n = 7) and localized prostate cancer (n = 14). Various miRNAs were highly abundant in the sera of patients with metastatic disease, and five upregulated miRNAs (miRNA-375, miRNA-9*, miRNA-141, miRNA-200b and miRNA-516a-3p) were selected for further validation. In the first validation study (n = 45), selected miRNAs were analyzed in a prospectively collected serum set taken from different prostate cancer risk groups. Most of the selected miRNAs were significantly correlated with adverse risk factors when different clinicopathological variables were analyzed. Circulating miRNA-375 and miRNA-141 turned out to be the most pronounced markers for high-risk tumors. Their levels also correlated with high Gleason score or lymph-node positive status in a second independent validation study (n = 71). In addition, the expression levels of miRNA-375 and miRNA-141 were monitored in 72 prostate tissue samples (36 tumor vs. 36 benign). Both miRNAs were highly expressed in all samples and significantly upregulated in the tumors compared to normal tissues. Overall, our observations suggest that miRNA-375 and miRNA-141 expression is enhanced in prostate cancer specimens and their release into the blood is further associated with advanced cancer disease.

Integration of pathway knowledge into a reweighted recursive feature elimination approach for risk stratification of cancer patients.

MOTIVATION: One of the main goals of high-throughput gene-expression studies in cancer research is to identify prognostic gene signatures, which have the potential to predict the clinical outcome. It is common practice to investigate these questions using classification methods. However, standard methods merely rely on gene-expression data and assume the genes to be independent. Including pathway knowledge a priori into the classification process has recently been indicated as a promising way to increase classification accuracy as well as the interpretability and reproducibility of prognostic gene signatures. RESULTS: We propose a new method called Reweighted Recursive Feature Elimination. It is based on the hypothesis that a gene with a low fold-change should have an increased influence on the classifier if it is connected to differentially expressed genes. We used a modified version of Google's PageRank algorithm to alter the ranking criterion of the SVM-RFE algorithm. Evaluations of our method on an integrated breast cancer dataset comprising 788 samples showed an improvement of the area under the receiver operator characteristic curve as well as in the reproducibility and interpretability of selected genes. AVAILABILITY: The R code of the proposed algorithm is given in Supplementary Material.

TMPRSS2-ERG -specific transcriptional modulation is associated with prostate cancer biomarkers and TGF-ß signaling.

BACKGROUND: TMPRSS2-ERG gene fusions occur in about 50% of all prostate cancer cases and represent promising markers for molecular subtyping. Although TMPRSS2-ERG fusion seems to be a critical event in prostate cancer, the precise functional role in cancer development and progression is still unclear. METHODS: We studied large-scale gene expression profiles in 47 prostate tumor tissue samples and in 48 normal prostate tissue samples taken from the non-suspect area of clinical low-risk tumors using Affymetrix GeneChip Exon 1.0 ST microarrays. RESULTS: Comparison of gene expression levels among TMPRSS2-ERG fusion-positive and negative tumors as well as benign samples demonstrated a distinct transcriptional program induced by the gene fusion event. Well-known biomarkers for prostate cancer detection like CRISP3 were found to be associated with the gene fusion status. WNT and TGF-ß/BMP signaling pathways were significantly associated with genes upregulated in TMPRSS2-ERG fusion-positive tumors. CONCLUSIONS: The TMPRSS2-ERG gene fusion results in the modulation of transcriptional patterns and cellular pathways with potential consequences for prostate cancer progression. Well-known biomarkers for prostate cancer detection were found to be associated with the gene fusion. Our results suggest that the fusion status should be considered in retrospective and future studies to assess biomarkers for prostate cancer detection, progression and targeted therapy.

Leveraging external knowledge on molecular interactions in classification methods for risk prediction of patients.

Classification of patients based on molecular markers, for example into different risk groups, is a modern field in medical research. The aim of this classification is often a better diagnosis or individualized therapy. The search for molecular markers often utilizes extremely high-dimensional data sets (e.g. gene-expression microarrays). However, in situations where the number of measured markers (genes) is intrinsically higher than the number of available patients, standard methods from statistical learning fail to deal correctly with this so-called "curse of dimensionality". Also feature or dimension reduction techniques based on statistical models promise only limited success. Several recent methods explore ideas of how to quantify and incorporate biological prior knowledge of molecular interactions and known cellular processes into the feature selection process. This article aims to give an overview of such current methods as well as the databases, where this external knowledge can be obtained from. For illustration, two recent methods are compared in detail, a feature selection approach for support vector machines as well as a boosting approach for regression models. As a practical example, data on patients with acute lymphoblastic leukemia are considered, where the binary endpoint "relapse within first year" should be predicted.

Serum miR-142-3p is associated with early relapse in operable lung adenocarcinoma patients.

The outcome of resectable non-small cell lung cancer (NSCLC) is critically determined by metastatic spread: About 30-50% of early-stage NSCLC patients encounter tumour recurrence within 5 years after surgery. A biomarker-driven stratification of early-stage lung cancer with a high risk of recurrence may improve therapy management and patient care. The aim of this study was to identify microRNAs (miRNAs) in serum of patients associated with early relapse in pulmonary adenocarcinoma. Serum samples were collected from 204 patients before surgery. miRNA screening was done using qRT-PCR based low-density arrays (664 miRNAs) comparing adenocarcinoma patients (n=40) with and without recurrence 24 months after surgery. Selected miRNAs associated with disease recurrence were validated in an independent patient cohort (n=114). miRNAs were also measured in advanced adenocarcinoma patients (n=29), and individuals with benign pulmonary diagnosis (n=21). Circulating miR-142-3p (p=0.005) and miR-29b (p=0.01) were identified in the screening and confirmed in the validation cohort to be increased in sera of early-stage adenocarcinoma patients suffering from recurrence within 24 months. Elevated miRNA levels were exclusively observed in the group of high-risk patient diagnosed for operable adenocarcinoma compared with benign diagnosis or advanced tumour disease. The differentiation between pulmonary adenocarcinoma patients with low and high risk for recurrence was improved by accounting for both miR-142-3p levels and tumour stage (p=0.007; AUC=0.78). In conclusion, circulating miR-142-3p was found to be associated with a high risk of recurrence in early-stage lung adenocarcinoma patients, and a putative serum marker for risk assessment.

Early detection of lung cancer by molecular markers in endobronchial epithelial-lining fluid.

INTRODUCTION: Early detection of malignancies in the lung by less-invasive methods aims at achieving efficient intervention and subsequently a reduction of the high mortality rate. We investigated whether biomarker analysis in endobronchial epithelial-lining fluid (ELF) collected by bronchoscopic microsampling (BMS) may be useful for a definitive preoperative diagnosis. METHODS: ELF was collected from subsegmental bronchi close to the indeterminate pulmonary nodule, which was detected by computed tomography, and from the contralateral lung. Diagnosis was confirmed by transbronchial biopsy or surgery. The study includes 142 ELF samples from 51 non-small-cell lung cancer patients and 20 benign cases. Microarray analysis was done with a patient subset (n = 15) to narrow down genes associated with a malignant phenotype. Thirteen potential biomarkers have been further analyzed by quantitative real-time polymerases chain reaction in an independent patient cohort (n = 56). RESULTS: All patients underwent BMS without complications. Gene-expression analyses by microarrays and quantitative real-time polymerases chain reaction could be reliably applied to ELF samples, and resulted in potential biomarkers for malignant pulmonary nodules. Four genes (tenascin-C, [C-X-C motif] ligand 14, S100 calcium binding protein A9, and keratin 17) were found to be upregulated in ELF of non-small-cell lung cancer patients with adenocarcinoma or squamous cell carcinoma. Combined analysis of tenascin-C expression and the nodule size improved the prediction of malignancy in this patient cohort. CONCLUSIONS: Our study suggests that the analysis of specific biomarkers in ELF collected by BMS could be a potentially useful adjunct to other diagnostic techniques aiming at the preoperative diagnosis of malignant pulmonary nodules.