Unbiased vs. biased approaches to the identification of cancer signatures: the case of lung cancer

Expression profiling analysis of human cancers is a promising approach to obtain precise molecular classification of cancers, to develop stratification tools for therapeutic regimens, and to predict the biological behavior of neoplasia. Direct profiling of human cancers (herein defined as “the unbiased approach”) presents, however, intrinsic problems connected with the high genetic noise embedded in the system. This, in turn, leads to fitting of the noise in the data (the so-called “overtraining”) with consequent instability of the identified signatures, when applied on different cohorts of patients. To circumvent these problems, “biased approaches” - which exploit the molecular knowledge of cancer obtained in model systems - are being developed. Biased approaches, however, are not problem-free, in that they provide information limited to single oncogenic events, thereby failing, at least in principle, to capture the complex repertoire of alterations of human cancers. In this review, we compare the two approaches and provide a test case, from our studies, of how “integrated” strategies, which combine biased and unbiased approaches, might lead to the identification of stable and reliable predictive signatures in cancer.     

Gene expression profiles of inflammatory breast cancer reveal high heterogeneity across the epithelial-hybrid-mesenchymal spectrum

Inflammatory breast cancer (IBC) is a highly aggressive breast cancer that metastasizes largely via tumor emboli, and has a 5-year survival rate of less than 30%. No unique genomic signature has yet been identified for IBC nor has any specific molecular therapeutic been developed to manage the disease. Thus, identifying gene expression signatures specific to IBC remains crucial. Here, we compare various gene lists that have been proposed as molecular footprints of IBC using different clinical samples as training and validation sets and using independent training algorithms, and determine their accuracy in identifying IBC samples in three independent datasets. We show that these gene lists have little to no mutual overlap, and have limited predictive accuracy in identifying IBC samples. Despite this inconsistency, single-sample gene set enrichment analysis (ssGSEA) of IBC samples correlate with their position on the epithelial-hybrid-mesenchymal spectrum. This positioning, together with ssGSEA scores, improves the accuracy of IBC identification across the three independent datasets. Finally, we observed that IBC samples robustly displayed a higher coefficient of variation in terms of EMT scores, as compared to non-IBC samples. Pending verification that this patient-to-patient variability extends to intratumor heterogeneity within a single patient, these results suggest that higher heterogeneity along the epithelial-hybrid-mesenchymal spectrum can be regarded to be a hallmark of IBC and a possibly useful biomarker.     

Comparison of the Prognostic Utility of the Diverse Molecular Data among lncRNA,DNA Methylation,microRNA, and mRNA across Five Human Cancers

IntroductionAdvances in high-throughput technologies have generated diverse informative molecular markers for cancer outcome prediction. Long non-coding RNA (lncRNA) and DNA methylation as new classes of promising markers are emerging as key molecules in human cancers; however, the prognostic utility of such diverse molecular data remains to be explored.ResultsUsing the IDFO approach, we obtained good predictive performance of the molecular datasets (bootstrap accuracy: 0.71–0.97) in five cancer types. Impressively, lncRNA was identified as the best prognostic predictor in the validated cohorts of four cancer types, followed by DNA methylation, mRNA, and then microRNA. We found the incorporating of multi-type molecular data showed similar predictive power to single-type molecular data, but with the exception of the lncRNA + DNA methylation combinations in two cancers. Survival analysis of proportional hazard models confirmed a high robustness for lncRNA and DNA methylation as prognosis factors independent of traditional clinical variables.ConclusionOur study provides insight into systematically understanding the prognostic performance of diverse molecular data in both single and aggregate patterns, which may have specific reference to subsequent related studies.     

An adult tissue-specific stem cell molecular phenotype is activated in epithelial cancer stem cells and correlated to patient outcome

Recent studies have shown that embryonic stem cell-like molecular phenotypes are commonly activated in human epithelial primary tumors and are linked to adverse patient prognosis. However it remains unclear whether these correlations to outcome are linked to the differentiation status of the human primary tumours1 or represent molecular reminiscences of epithelial cancer stem cells. In addition, while it has been demonstrated that leukemic cancer stem cells re-acquire an embryonic stem cell-like phenotype, the molecular basis of stem cell function in epithelial cancer stem cells has not been investigated. Here we show that a normal adult tissue-specific stem cell molecular phenotype is commonly activated in epithelial cancer stem cells and for the first time provide evidence that enrichment in cancer stem cells-specific molecular signatures are correlated to highly aggressive tumor phenotypes in human epithelial cancers.     

SUITOR: Selecting the number of mutational signatures through cross-validation

For de novo mutational signature analysis, the critical first step is to decide how many signatures should be expected in a cancer genomics study. An incorrect number could mislead downstream analyses. Here we present SUITOR (Selecting the nUmber of mutatIonal signaTures thrOugh cRoss-validation), an unsupervised cross-validation method that requires little assumptions and no numerical approximations to select the optimal number of signatures without overfitting the data. In vitro studies and in silico simulations demonstrated that SUITOR can correctly identify signatures, some of which were missed by other widely used methods. Applied to 2,540 whole-genome sequenced tumors across 22 cancer types, SUITOR selected signatures with the smallest prediction errors and almost all signatures of breast cancer selected by SUITOR were validated in an independent breast cancer study. SUITOR is a powerful tool to select the optimal number of mutational signatures, facilitating downstream analyses with etiological or therapeutic importance.     

A Predictive Phosphorylation Signature of Lung Cancer

Background

Aberrant activation of signaling pathways drives many of the fundamental biological processes that accompany tumor initiation and progression. Inappropriate phosphorylation of intermediates in these signaling pathways are a frequently observed molecular lesion that accompanies the undesirable activation or repression of pro- and anti-oncogenic pathways. Therefore, methods which directly query signaling pathway activation via phosphorylation assays in individual cancer biopsies are expected to provide important insights into the molecular “logic” that distinguishes cancer and normal tissue on one hand, and enables personalized intervention strategies on the other.

Results

We first document the largest available set of tyrosine phosphorylation sites that are, individually, differentially phosphorylated in lung cancer, thus providing an immediate set of drug targets. Next, we develop a novel computational methodology to identify pathways whose phosphorylation activity is strongly correlated with the lung cancer phenotype. Finally, we demonstrate the feasibility of classifying lung cancers based on multi-variate phosphorylation signatures.

Conclusions

Highly predictive and biologically transparent phosphorylation signatures of lung cancer provide evidence for the existence of a robust set of phosphorylation mechanisms (captured by the signatures) present in the majority of lung cancers, and that reliably distinguish each lung cancer from normal. This approach should improve our understanding of cancer and help guide its treatment, since the phosphorylation signatures highlight proteins and pathways whose phosphorylation should be inhibited in order to prevent unregulated proliferation.     

A fuzzy gene expression-based computational approach improves breast cancer prognostication

Early gene expression studies classified breast tumors into at least three clinically relevant subtypes. Although most current gene signatures are prognostic for estrogen receptor (ER) positive/human epidermal growth factor receptor 2 (HER2) negative breast cancers, few are informative for ER negative/HER2 negative and HER2 positive subtypes. Here we present Gene Expression Prognostic Index Using Subtypes (GENIUS), a fuzzy approach for prognostication that takes into account the molecular heterogeneity of breast cancer. In systematic evaluations, GENIUS significantly outperformed current gene signatures and clinical indices in the global population of patients.     

CENP-A overexpression promotes aneuploidy with karyotypic heterogeneity

Chromosomal instability (CIN) is a hallmark of many cancers. Restricting the localization of centromeric histone H3 variant CENP-A to centromeres prevents CIN. CENP-A overexpression (OE) and mislocalization have been observed in cancers and correlate with poor prognosis; however, the molecular consequences of CENP-A OE on CIN and aneuploidy have not been defined. Here, we show that CENP-A OE leads to its mislocalization and CIN with lagging chromosomes and micronuclei in pseudodiploid DLD1 cells and xenograft mouse model. CIN is due to reduced localization of proteins to the kinetochore, resulting in defects in kinetochore integrity and unstable kinetochore–microtubule attachments. CENP-A OE contributes to reduced expression of cell adhesion genes and higher invasion of DLD1 cells. We show that CENP-A OE contributes to aneuploidy with karyotypic heterogeneity in human cells and xenograft mouse model. In summary, our results provide a molecular link between CENP-A OE and aneuploidy, and suggest that karyotypic heterogeneity may contribute to the aggressive phenotype of CENP-A–overexpressing cancers.     

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.     

SPARCoC: A New Framework for Molecular Pattern Discovery and Cancer Gene Identification

It is challenging to cluster cancer patients of a certain histopathological type into molecular subtypes of clinical importance and identify gene signatures directly relevant to the subtypes. Current clustering approaches have inherent limitations, which prevent them from gauging the subtle heterogeneity of the molecular subtypes. In this paper we present a new framework: SPARCoC (Sparse-CoClust), which is based on a novel Common-background and Sparse-foreground Decomposition (CSD) model and the Maximum Block Improvement (MBI) co-clustering technique. SPARCoC has clear advantages compared with widely-used alternative approaches: hierarchical clustering (Hclust) and nonnegative matrix factorization (NMF). We apply SPARCoC to the study of lung adenocarcinoma (ADCA), an extremely heterogeneous histological type, and a significant challenge for molecular subtyping. For testing and verification, we use high quality gene expression profiling data of lung ADCA patients, and identify prognostic gene signatures which could cluster patients into subgroups that are significantly different in their overall survival (with p-values < 0.05). Our results are only based on gene expression profiling data analysis, without incorporating any other feature selection or clinical information; we are able to replicate our findings with completely independent datasets. SPARCoC is broadly applicable to large-scale genomic data to empower pattern discovery and cancer gene identification.     

Systemic Lupus Erythematous and Malignancy Risk: A Meta-Analysis

Background

Pilot studies have estimated cancer incidence in patients with systemic lupus erythematous (SLE). However, the results have been inconclusive. To ascertain the correlation between SLE and malignancy more comprehensively and precisely, we conducted a meta-analysis.

Methods

PubMed, the Cochrane Library and Embase databases through June 2014, were searched to identify observational studies evaluating the association between SLE and malignancy. The outcomes from these studies were measured as relative risks (RRs). A random or fixed effects model was chosen to calculate the pooled RR according to heterogeneity test. Between-study heterogeneity was assessed by estimating I² index. Publication bias was assessed by Egger’s test.

Results

A total of 16 papers, including 59,662 SLE patients, were suitable for the meta-analysis. Of these papers, 15 reported RRs for overall malignancy, 12 for non-Hodgkin lymphoma (NHL) and lung cancer, 7 for bladder cancer, 6 for Hodgkin lymphoma (HL) and leukemia, 5 for skin melanoma, and liver and thyroid cancers, 4 for multiple myeloma (MM), and esophageal and vaginal/vulvar cancers and 3 for laryngeal and non-melanoma skin cancers. The pooled RRs were 1.28 (95% CI, 1.17–1.41) for overall cancer, 5.40 (95% CI, 3.75–7.77) for NHL, 3.26(95% CI, 2.17–4.88) for HL, 2.01(95% CI, 1.61–2.52) for leukemia, 1.45(95% CI, 1.04–2.03) for MM, 4.19(95% CI, 1.98–8.87) for laryngeal cancer, 1.59 (95% CI, 1.44–1.76) for lung cancer, 1.86(95% CI, 1.21–2.88) for esophageal cancer, 3.21(95% CI, 1.70–6.05) for liver cancer, 3.67(95% CI, 2.80–4.81) for vaginal/vulvar cancer, 2.11(95% CI, 1.12–3.99) for bladder cancer, 1.51(95% CI, 1.12–2.03) for non-melanoma skin cancer, 1.78(95% CI, 1.35–2.33) for thyroid cancer, and 0.65(95% CI, 0.50–0.85) for skin melanoma. Only the meta-analyses of overall malignancy, NHL, and liver and bladder cancers produced substantial heterogeneity (I², 57.6% vs 74.3% vs 67.7% vs 82.3%). No apparent publication bias was detected except for NHL studies.

Conclusions

Our data support an association between SLE and malignancy, not only demonstrating an increased risk for NHL, HL, leukemia, and some non-hematologic malignancies, including laryngeal, lung, liver, vaginal/vulvar, and thyroid malignancies, but also a reduced risk for skin melanoma. Although an increased risk of MM, and esophageal, bladder and non-melanoma skin cancers was identified from the accumulated data in these studies, this observation requires confirmation.     

The mRNA related ceRNA–ceRNA landscape and significance across 20 major cancer types

Cross-talk between competitive endogenous RNAs (ceRNAs) through shared miRNAs represents a novel layer of gene regulation that plays important roles in the physiology and development of cancers. However, a global view of their system-level properties across various types of cancers is still unknown. Here, we constructed the mRNA related ceRNA–ceRNA interaction landscape across 20 cancer types by systematically analyzing molecular profiles of 5203 tumors and miRNA regulations. Our study highlights the conserved features shared by pan-cancer and higher similarity within similar origin cell type. Moreover, a core ceRNA network was identified. Function analysis identified a common theme of cancer hallmarks, however they exhibit phenotype-specific connectivity patterns. Besides, we found a marked rewiring in the ceRNA program between various cancers, and further revealed conserved and rewired network ceRNA hubs in each cancer, which were tensely competitive interactions to constitute conserved and cancer-specific modules. By providing mechanistic linkage between known cancer miRNAs, their mediated ceRNA–ceRNA interactions, and the associations with known cancer hallmarks, the inferred cancer ceRNA–ceRNA interaction landscape will serve as a powerful public resource for further biological discoveries of tumorigenesis.     

An ileal Crohn's disease gene signature based on whole human genome expression profiles of disease unaffected ileal mucosal biopsies

Previous genome-wide expression studies have highlighted distinct gene expression patterns in inflammatory bowel disease (IBD) compared to control samples, but the interpretation of these studies has been limited by sample heterogeneity with respect to disease phenotype, disease activity, and anatomic sites. To further improve molecular classification of inflammatory bowel disease phenotypes we focused on a single anatomic site, the disease unaffected proximal ileal margin of resected ileum, and three phenotypes that were unlikely to overlap: ileal Crohn's disease (ileal CD), ulcerative colitis (UC), and control patients without IBD. Whole human genome (Agilent) expression profiling was conducted on two independent sets of disease-unaffected ileal samples collected from the proximal margin of resected ileum. Set 1 (47 ileal CD, 27 UC, and 25 Control non-IBD patients) was used as the training set and Set 2 was subsequently collected as an independent test set (10 ileal CD, 10 UC, and 10 control non-IBD patients). We compared the 17 gene signatures selected by four different feature-selection methods to distinguish ileal CD phenotype with non-CD phenotype. The four methods yielded different but overlapping solutions that were highly discriminating. All four of these methods selected FOLH1 as a common feature. This gene is an established biomarker for prostate cancer, but has not previously been associated with Crohn's disease. Immunohistochemical staining confirmed increased expression of FOLH1 in the ileal epithelium. These results provide evidence for convergent molecular abnormalities in the macroscopically disease unaffected proximal margin of resected ileum from ileal CD subjects.     

CAERUS: predicting CAncER oUtcomeS using relationship between protein structural information, protein networks, gene expression data, and mutation data

Carcinogenesis is a complex process with multiple genetic and environmental factors contributing to the development of one or more tumors. Understanding the underlying mechanism of this process and identifying related markers to assess the outcome of this process would lead to more directed treatment and thus significantly reduce the mortality rate of cancers. Recently, molecular diagnostics and prognostics based on the identification of patterns within gene expression profiles in the context of protein interaction networks were reported. However, the predictive performances of these approaches were limited. In this study we propose a novel integrated approach, named CAERUS, for the identification of gene signatures to predict cancer outcomes based on the domain interaction network in human proteome. We first developed a model to score each protein by quantifying the domain connections to its interacting partners and the somatic mutations present in the domain. We then defined proteins as gene signatures if their scores were above a preset threshold. Next, for each gene signature, we quantified the correlation of the expression levels between this gene signature and its neighboring proteins. The results of the quantification in each patient were then used to predict cancer outcome by a modified naïve Bayes classifier. In this study we achieved a favorable accuracy of 88.3%, sensitivity of 87.2%, and specificity of 88.9% on a set of well-documented gene expression profiles of 253 consecutive breast cancer patients with different outcomes. We also compiled a list of cancer-associated gene signatures and domains, which provided testable hypotheses for further experimental investigation. Our approach proved successful on different independent breast cancer data sets as well as an ovarian cancer data set. This study constitutes the first predictive method to classify cancer outcomes based on the relationship between the domain organization and protein network.     

肿瘤突变特征与病理分型的关联研究

准确评估肿瘤的病理亚型对诊断、治疗和预后至关重要。以往病理亚型的诊断主要依赖HE染色法和免疫组织化学法,而随着测序技术的不断发展,对患者进行基因型和表型特点的个体分析成为可能,将肿瘤病理分型与基因分型结合用于疾病分型、诊治选择和疗效判断的精准医学研究逐渐兴起。不同病理亚型的肿瘤细胞来源、致癌因素和临床表型均不尽相同,其在基因组上会留下特异“印迹”,即突变特征。本研究通过整合癌症基因组数据库(The Cancer Genome Atlas, TCGA)中肾癌、肺癌和食管癌的外显子测序数据,分别对3种肿瘤通过肿瘤基因突变特征进行肿瘤病理分型聚类和预测。首先通过非监督聚类方法将3种肿瘤分别按照24种突变特征进行聚类分析,其次通过随机森林法从24种突变特征中进一步选择对于区分不同病理亚型有显著性的突变特征并进行聚类分析,构建突变特征对3种肿瘤病理亚型的分型模型。在肾癌中,该模型准确率达到了100% (95% confidence interval (CI): 0.93~1.00),肺癌和食管癌中分别达到了78% (95% CI: 0.66~0.86)和84% (95% CI: 0.60~0.97)。以上研究结果表明,突变特征作为新型分子标记物,对肿瘤的病理分型、诊断,尤其是早诊具有一定的参考意义。