A machine learning framework to trace tumor tissue-of-origin of 13 types of cancer based on DNA somatic mutation

Carcinoma of unknown primary (CUP), defined as metastatic cancers with unknown cancer origin, occurs in 3‐5 per 100 cancer patients in the United States. Heterogeneity and metastasis of cancer brings great difficulties to the follow-up diagnosis and treatment for CUP. To find the tissue-of-origin (TOO) of the CUP, multiple methods have been raised. However, the accuracies for computed tomography (CT) and positron emission tomography (PET) to identify TOO were 20%–27% and 24%–40% respectively, which were not enough for determining targeted therapies. In this study, we provide a machine learning framework to trace tumor tissue origin by using gene length-normalized somatic mutation sequencing data. Somatic mutation data was downloaded from the Data Portal (Release 28) of the International Cancer Genome Consortium (ICGC), and 4909 samples for 13 cancers was used to identify primary site of cancers. Optimal results were obtained based on a 600-gene set by using the random forest algorithm with 10-fold cross-validation, and the average accuracy and F1-score were 0.8822 and 0.8886 respectively across 13 types of cancer. In conclusion, we provide an effective computational framework to infer cancer tissue-of-origin by combining DNA sequencing and machine learning techniques, which is promising in assisting clinical diagnosis of cancers.     

Distinguishing three subtypes of hematopoietic cells based on gene expression profiles using a support vector machine

Hematopoiesis is a complicated process involving a series of biological sub-processes that lead to the formation of various blood components. A widely accepted model of early hematopoiesis proceeds from long-term hematopoietic stem cells (LT-HSCs) to multipotent progenitors (MPPs) and then to lineage-committed progenitors. However, the molecular mechanisms of early hematopoiesis have not been fully characterized. In this study, we applied a computational strategy to identify the gene expression signatures distinguishing three types of closely related hematopoietic cells collected in recent studies: (1) hematopoietic stem cell/multipotent progenitor cells; (2) LT-HSCs; and (3) hematopoietic progenitor cells. Each cell in these cell types was represented by its gene expression profile among a total number of 20,475 genes. The expression features were analyzed by a Monte-Carlo Feature Selection (MCFS) method, resulting in a feature list. Then, the incremental feature selection (IFS) and a support vector machine (SVM) optimized with a sequential minimum optimization (SMO) algorithm were employed to access the optimal classifier with the highest Matthews correlation coefficient (MCC) value of 0.889, in which 6698 features were used to represent cells. In addition, through an updated program of MCFS method, seventeen decision rules can be obtained, which can classify the three cell types with an overall accuracy of 0.812. Using a literature review, both the rules and the top features used for building the optimal classifier were confirmed to be commonly used or potential biological markers for distinguishing the three cell types of HSPCs. This article is part of a Special Issue entitled: Accelerating Precision Medicine through Genetic and Genomic Big Data Analysis edited by Yudong Cai & Tao Huang.     

Likelihood models of somatic mutation and codon substitution in cancer genes

The role of somatic mutation in cancer is well established and several genes have been identified that are frequent targets. This has enabled large-scale screening studies of the spectrum of somatic mutations in cancers of particular organs. Cancer gene mutation databases compile the results of many studies and can provide insight into the importance of specific amino acid sequences and functional domains in cancer, as well as elucidate aspects of the mutation process. Past studies of the spectrum of cancer mutations (in particular genes) have examined overall frequencies of mutation (at specific nucleotides) and of missense, nonsense, and silent substitution (at specific codons) both in the sequence as a whole and in a specific functional domain. Existing methods ignore features of the genetic code that allow some codons to mutate to missense, or stop, codons more readily than others (i.e., by one nucleotide change, vs. two or three). A new codon-based method to estimate the relative rate of substitution (fixation of a somatic mutation in a cancer cell lineage) of nonsense vs. missense mutations in different functional domains and in different tumor tissues is presented. Models that account for several potential influences on rates of somatic mutation and substitution in cancer progenitor cells and allow biases of mutation rates for particular dinucleotide sequences (CGs and dipyrimidines), transition vs. transversion bias, and variable rates of silent substitution across functional domains (useful in detecting investigator sampling bias) are considered. Likelihood-ratio tests are used to choose among models, using cancer gene mutation data. The method is applied to analyze published data on the spectrum of p53 mutations in cancers. A novel finding is that the ratio of the probability of nonsense to missense substitution is much lower in the DNA-binding and transactivation domains (ratios near 1) than in structural domains such as the linker, tetramerization (oligomerization), and proline-rich domains (ratios exceeding 100 in some tissues), implying that the specific amino acid sequence may be less critical in structural domains (e.g., amino acid changes less often lead to cancer). The transition vs. transversion bias and effect of CpG dinucleotides on mutation rates in p53 varied greatly across cancers of different organs, likely reflecting effects of different endogenous and exogenous factors influencing mutation in specific organs.     

Human germline and pan-cancer variomes and their distinct functional profiles

Identification of non-synonymous single nucleotide variations (nsSNVs) has exponentially increased due to advances in Next-Generation Sequencing technologies. The functional impacts of these variations have been difficult to ascertain because the corresponding knowledge about sequence functional sites is quite fragmented. It is clear that mapping of variations to sequence functional features can help us better understand the pathophysiological role of variations. In this study, we investigated the effect of nsSNVs on more than 17 common types of post-translational modification (PTM) sites, active sites and binding sites. Out of 1 705 285 distinct nsSNVs on 259 216 functional sites we identified 38 549 variations that significantly affect 10 major functional sites. Furthermore, we found distinct patterns of site disruptions due to germline and somatic nsSNVs. Pan-cancer analysis across 12 different cancer types led to the identification of 51 genes with 106 nsSNV affected functional sites found in 3 or more cancer types. 13 of the 51 genes overlap with previously identified Significantly Mutated Genes (Nature. 2013 Oct 17;502(7471)). 62 mutations in these 13 genes affecting functional sites such as DNA, ATP binding and various PTM sites occur across several cancers and can be prioritized for additional validation and investigations.     

Opposite association of two PPARG variants with cancer: overrepresentation of H449H in endometrial carcinoma cases and underrepresentation of P12A in renal cell carcinoma cases

Peroxisome proliferator activated receptor gamma (PPARgamma) is a nuclear hormone receptor that has been shown to regulate differentiation and cell growth. Studies of the differentiative effects of PPARgamma agonists on several cancer cell lines led to the hypothesis that dysfunction of PPARgamma contributes to tumorigenesis. These functional observations were strengthened by genetic evidence: somatic loss-of-function mutations in PPARG, encoding PPARgamma, in sporadic colorectal carcinomas and somatic translocation of PAX8 and PPARG in follicular thyroid carcinoma. Recently overrepresentation of the H449H variant was found in a cohort of American patients with glioblastoma multiforme. The glioblastoma multiforme data suggest that PPARG contributes common, low-penetrance alleles for cancer susceptibility. To test this hypothesis in a broader range of cancers we examined a series of carcinomas of the cervix, endometrium, ovary, prostate, and kidney for germline sequence variation in PPARG. In addition to the two common sequence variants, P12A and H449H, there were five other sequence variants. P12A alleles were underrepresented in renal cell carcinoma patients compared to country-of-origin race-matched controls (3.75% vs. 12.1%, P<0.04). In contrast, the H449H variant was overrepresented in individuals with endometrial carcinoma compared to controls (14.4% vs. 6.25%, P<0.02). These observations lend genetic evidence consistent with our hypothesis that PPARG serves as a common, low-penetrance susceptibility gene for cancers of several types, especially those epidemiologically associated with obesity and fat intake.     

Somatic mutations in mitochondrial genome and their potential roles in the progression of human gastric cancer

Background

Somatic mutation in mitochondrial DNA (mtDNA) has been proposed to contribute to initiation and progression of human cancer. In our previous study, high frequency of somatic mutations was found in the D-loop region of mtDNA of gastric cancers. However, it is unclear whether somatic mutations occur in the coding region of mtDNA of gastric cancers.

Methods

Using DNA sequencing, we studied 31 gastric cancer specimens and corresponding non-cancerous stomach tissues. Moreover, a human gastric cancer SC-M1 cell line was treated with oligomycin to induce mitochondrial dysfunction. Cisplatin sensitivity and cell migration were analyzed.

Results

We identified eight somatic mutations in the coding region of mtDNAs of seven gastric cancer samples (7/31, 22.6%). Patients with somatic mutations in the entire mtDNA of gastric cancers did not show significant association with their clinicopathologic features. Among the eight somatic mutations, five point mutations (G3697A, G4996A, G9986A, C12405T and T13015C) are homoplasmic and three mutations (5895delC, 7472insC and 12418insA) are heteroplasmic. Four (4/8, 50%) of these somatic mutations result in amino acid substitutions in the highly conserved regions of mtDNA, which potentially lead to mitochondrial dysfunction. In addition, in vitro experiments in SC-M1 cells revealed that oligomycin-induced mitochondrial dysfunction promoted resistance to cisplatin and enhanced cell migration. N-acetyl cysteine was effective in the prevention of the oligomycin-enhanced migration, which suggests that reactive oxygen species generated by defective mitochondria may be involved in the enhanced migration of SC-M1 cells.

General Significance

Our results suggest that somatic mtDNA mutations and mitochondrial dysfunction may play an important role in the malignant progression of gastric cancer.     

CanDrA: Cancer-Specific Driver Missense Mutation Annotation with Optimized Features

Driver mutations are somatic mutations that provide growth advantage to tumor cells, while passenger mutations are those not functionally related to oncogenesis. Distinguishing drivers from passengers is challenging because drivers occur much less frequently than passengers, they tend to have low prevalence, their functions are multifactorial and not intuitively obvious. Missense mutations are excellent candidates as drivers, as they occur more frequently and are potentially easier to identify than other types of mutations. Although several methods have been developed for predicting the functional impact of missense mutations, only a few have been specifically designed for identifying driver mutations. As more mutations are being discovered, more accurate predictive models can be developed using machine learning approaches that systematically characterize the commonality and peculiarity of missense mutations under the background of specific cancer types. Here, we present a cancer driver annotation (CanDrA) tool that predicts missense driver mutations based on a set of 95 structural and evolutionary features computed by over 10 functional prediction algorithms such as CHASM, SIFT, and MutationAssessor. Through feature optimization and supervised training, CanDrA outperforms existing tools in analyzing the glioblastoma multiforme and ovarian carcinoma data sets in The Cancer Genome Atlas and the Cancer Cell Line Encyclopedia project.     

Integrative Analysis of Somatic Mutations Altering MicroRNA Targeting in Cancer Genomes

Determining the functional impact of somatic mutations is crucial to understanding tumorigenesis and metastasis. Recent sequences of several cancers have provided comprehensive lists of somatic mutations across entire genomes, enabling investigation of the functional impact of somatic mutations in non-coding regions. Here, we study somatic mutations in 3′UTRs of genes that have been identified in four cancers and computationally predict how they may alter miRNA targeting, potentially resulting in dysregulation of the expression of the genes harboring these mutations. We find that somatic mutations create or disrupt putative miRNA target sites in the 3′UTRs of many genes, including several genes, such as MITF, EPHA3, TAL1, SCG3, and GSDMA, which have been previously associated with cancer. We also integrate the somatic mutations with germline mutations and results of association studies. Specifically, we identify putative miRNA target sites in the 3′UTRs of BMPR1B, KLK3, and SPRY4 that are disrupted by both somatic and germline mutations and, also, are in linkage disequilibrium blocks with high scoring markers from cancer association studies. The somatic mutation in BMPR1B is located in a target site of miR-125b; germline mutations in this target site have previously been both shown to disrupt regulation of BMPR1B by miR-125b and linked with cancer.     

Telomerase reverse transcriptase promoter mutations in cancers derived from multiple organ sites among middle eastern population

Sanger Sequencing and immunohistochemistry was employed to investigate the TERT promoter mutations and TERT protein expression with their association to clinicopathological characteristics in over 2200 samples of Middle Eastern origin from 13 different types of cancers. The TERT promoter mutations were most frequently present in bladder cancer (68.6%), followed by central nervous system tumors (28.7%), thyroid cancer (15.4%), prostate cancer (9.3%), endometrial carcinoma (3.7%), rhabdomyosarcoma (1.4%), colorectal cancer (1%), epithelial ovarian carcinoma (0.7%) and breast cancer (0.7%). No mutations were observed in other types of cancers. In bladder cancer, we found significant inverse association with metastasis and a trend to good survival in patients with TERT mutations. In gliomas, TERT promoter mutations predicted poor prognosis. In thyroid cancer, high frequency of TERT mutation was observed in poorly differentiated carcinoma. In addition, TERT promoter mutations were associated with aggressive markers and poor outcome in follicular thyroid carcinomas.     

TP53 and mutations in human cancer

TP53 is the most frequently mutated gene in human cancer, with a predominance of missense mutations scattered over 200 codons. In many cancers, specific mutation patterns can be identified, which are shaped by site-specific mutagenesis and by biological selection. In tobacco-related cancers (lung, head and neck), organ-specific patterns are observed, with many mutations compatible with the ones experimentally induced by tobacco carcinogens. In several other cancers, such as squamous cell carcinoma of the oesophagus or hepatocellular carcinoma (HCC), mutation patterns show geographic variations between regions of high and low incidence, suggesting a role for region-specific risk factors. HCC from high-incidence regions showing also a high prevalence of a specific Ser-249 TP53 mutation is one of the most striking examples of a mutagen fingerprint. All such assessments are useful to generate clues on the mutagenic mechanisms involved in human cancer. Moreover, it has been shown that DNA retrieved from plasma can be successfully used for detection of TP53 mutations, which gives hope for earlier more accurate detection of human cancers.     

肝癌基因组变异与分子分型

肿瘤是机体在各种致癌因素刺激下,基因组发生变异导致细胞失去正常生长调控而异常增殖的一种恶性疾病.肿瘤具有维持细胞增殖信号、逃避生长抑制、抗细胞凋亡、无限复制、诱导血管生成、激活侵袭和转移、能量代谢的重编程和免疫逃避等特点.原发性肝癌是一种高致死性的癌症类型,在中国发病率高,约占全世界发病人数的一半.肝细胞癌是原发性肝癌中的主要组织学亚型,与乙型和丙型肝炎病毒感染、酒精刺激、肥胖以及饮食污染等有关.遗传学和表观遗传突变事件的研究有助于理解肝癌的发病机制并对患者进行分子分型,而分子分型则可以指导临床个体化治疗和预后判断.     

Assessing the contribution of tumor mutational phenotypes to cancer progression risk

Cancer occurs via an accumulation of somatic genomic alterations in a process of clonal evolution. There has been intensive study of potential causal mutations driving cancer development and progression. However, much recent evidence suggests that tumor evolution is normally driven by a variety of mechanisms of somatic hypermutability, which act in different combinations or degrees in different cancers. These variations in mutability phenotypes are predictive of progression outcomes independent of the specific mutations they have produced to date. Here we explore the question of how and to what degree these differences in mutational phenotypes act in a cancer to predict its future progression. We develop a computational paradigm using evolutionary tree inference (tumor phylogeny) algorithms to derive features quantifying single-tumor mutational phenotypes, followed by a machine learning framework to identify key features predictive of progression. Analyses of breast invasive carcinoma and lung carcinoma demonstrate that a large fraction of the risk of future clinical outcomes of cancer progression—overall survival and disease-free survival—can be explained solely from mutational phenotype features derived from the phylogenetic analysis. We further show that mutational phenotypes have additional predictive power even after accounting for traditional clinical and driver gene-centric genomic predictors of progression. These results confirm the importance of mutational phenotypes in contributing to cancer progression risk and suggest strategies for enhancing the predictive power of conventional clinical data or driver-centric biomarkers.     

Somatic inactivation of the VHL gene in Von Hippel-Lindau disease tumors.

Von Hippel-Lindau (VHL) disease is a dominantly inherited disorder predisposing to retinal and CNS hemangioblastomas, renal cell carcinoma (RCC), pheochromocytoma, and pancreatic tumors. Interfamilial differences in predisposition to pheochromocytoma reflect allelic heterogeneity such that there is a strong association between missense mutations and risk of pheochromocytoma. We investigated the mechanism of tumorigenesis in VHL disease tumors to determine whether there were differences between tumor types or classes of germ-line mutations. Fifty-three tumors (30 RCCs, 15 hemangioblastomas, 5 pheochromocytomas, and 3 pancreatic tumors) from 33 patients (27 kindreds) with VHL disease were analyzed. Overall, 51% of 45 informative tumors showed loss of heterozygosity (LOH) at the VHL locus. In 11 cases it was possible to distinguish between loss of the wild-type and mutant alleles, and in each case the wild-type allele was lost. LOH was detected in all tumor types and occurred in the presence of both germ-line missense mutations and other types of germline mutation associated with a low risk of pheochromocytoma. Intragenic somatic mutations were detected in three tumors (all hemangioblastomas) and in two of these could be shown to occur in the wild-type allele. This provides the first example of homozygous inactivation of the VHL by small intragenic mutations in this type of tumor. Hypermethylation of the VHL gene was detected in 33% (6/18) of tumors without LOH, including 2 RCCs and 4 hemangioblastomas. Although hypermethylation of the VHL gene has been reported previously in nonfamilial RCC and although methylation of tumor-suppressor genes has been implicated in the pathogenesis of other sporadic cancers, this is the first report of somatic methylation in a familial cancer syndrome.     

Gaining insights into relevance across cancers based on mutation features of TP53 gene

The tumor suppressor gene TP53, one of the most frequently mutated genes, is recognized as the guardian of genome and can provide a significant barrier to neoplastic transformation and tumor progression. Traditional theory believes that TP53 mutations are equal among cancer types. However, to date, no study has explored the TP53 mutation profile from a holistic and systematic standpoint to discovery its relevance and feature with cancers. Mutation signature, an unbiased approach to identify the mutational processes, can be a potent indicator for exploring mutation-driven tumor occurrence and progression. In this research, several features such as hotspots, mutability and mutation signature of somatic TP53 mutations derived from 18 types of cancer tissues from cBioPortal were analyzed and manifested the organizational preference among cancers. Mutation signatures found in almost all cancer types were Signature 6 related to mismatch repair deficiency, and Signature 1 that reflects the natural decomposition of 5-methylcytosine into thymine associated with aging. Meanwhile, several signatures of TP53 mutations displayed tissue-selective. Mutations enriched in bladder, skin, lung cancer were associated with signatures of APOBEC activity (Signature 2 and 13), alkylating agents (Signature 11), and tobacco smoke (Signature 4), respectively. Moreover, Signature 4 and 29 associated with tobacco smoking or chewing found in lung, sarcoma, esophageal, and head and neck cancer may be related to their smoking history. In addition, several digestive cancers, including colorectal, stomach, pancreatic and esophageal cancers, showed the high correlation in context and mutation signature profiles. Our study suggests that the tissue-selective activity of mutational processes would reflect the tissue-specific enrichment of TP53 mutations and provides a new perspective to understand the relevance of diverse diseases based on the spectrum of TP53 mutations.     

A new deep learning technique reveals the exclusive functional contributions of individual cancer mutations

Cancers are caused by genomic alterations that may be inherited, induced by environmental carcinogens, or caused due to random replication errors. Postinduction of carcinogenicity, mutations further propagate and drastically alter the cancer genomes. Although a subset of driver mutations has been identified and characterized to date, most cancer-related somatic mutations are indistinguishable from germline variants or other noncancerous somatic mutations. Thus, such overlap impedes appreciation of many deleterious but previously uncharacterized somatic mutations. The major bottleneck arises due to patient-to-patient variability in mutational profiles, making it difficult to associate specific mutations with a given disease outcome. Here, we describe a newly developed technique Continuous Representation of Codon Switches (CRCS), a deep learning-based method that allows us to generate numerical vector representations of mutations, thereby enabling numerous machine learning-based tasks. We demonstrate three major applications of CRCS; first, we show how CRCS can help detect cancer-related somatic mutations in the absence of matched normal samples, which has applications in cell-free DNA–based assessment of tumor mutation burden. Second, the proposed approach also enables identification and exploration of driver genes; our analyses implicate DMD, RSK4, OFD1, WDR44, and AFF2 as potential cancer drivers. Finally, we used CRCS to score individual mutations in a tumor sample, which was found to be predictive of patient survival in bladder urothelial carcinoma, hepatocellular carcinoma, and lung adenocarcinoma. Taken together, we propose CRCS as a valuable computational tool for analysis of the functional significance of individual cancer mutations.     

Clinical Significance in Oral Cavity Squamous Cell Carcinoma of Pathogenic Somatic Mitochondrial Mutations

Somatic mutations affecting the mitochondrial DNA (mtDNA) have been frequently observed in human cancers and proposed as important oncological biomarkers. However, the clinical significance of mtDNA mutations in cancer remains unclear. This study was therefore performed to explore the possible clinical use in assessing oral squamous cell carcinoma (OSCC) of pathogenic mtDNA mutations. The entire mitochondrial genome of 300 OSCC with their matched control DNAs was screened by direct sequencing and criteria were set to define a pathogenic somatic mutation. The patients'' TP53 R72P genotypes were determined by polymerase chain reaction-restriction fragment length polymorphism. The relationships between pathogenic somatic mutations, clinicopathogical features, TP53 R72P genotype and clinical prognosis were analyzed. Overall, 645 somatic mtDNA mutations were identified and 91 of these mutations were defined as pathogenic. About one quarter (74/300) of the OSCC tumor samples contained pathogenic mutations. Individuals with the TP53 R allele had a higher frequency of pathogenic somatic mutation than those with the PP genotype. Kaplan-Meier analysis indicated that TP53 R allele patients with pathogenic somatic mutations demonstrated a significant association with a poorer disease-free survival than other individuals (HR = 1.71; 95% CI, 1.15–2.57; p = 0.009) and this phenomenon still existed after adjusting for mtDNA haplogroup, tumor stage with treatment regimens, differentiation and age at diagnosis (HR = 1.59; 95% CI, 1.06–2.40; p = 0.03). Subgroup analyses showed that this phenomenon was limited to patients who received adjuvant radiotherapy/chemo-radiotherapy after surgery. The results strongly indicated that pathogenic mtDNA mutations are a potential prognostic marker for OSCCs. Furthermore, functional mitochondria may play an active role in cancer development and the patient''s response to radiotherapy/chemo-radiotherapy.     

Most human carcinomas of the exocrine pancreas contain mutant c-K-ras genes

Using in vitro gene amplification by the polymerase chain reaction (PCR) and mutation detection by the RNAase A mismatch cleavage method, we have examined c-K-ras genes in human pancreatic carcinomas. We used frozen tumor specimens and single 5 micron sections from formalin-fixed, paraffin-embedded tumor tissue surgically removed or obtained at autopsy. Twenty-one out of 22 carcinomas of the exocrine pancreas contained c-K-ras genes with mutations at codon 12. In seven cases tested, the mutation was present in both primary tumors and their corresponding metastases. No mutations were detected in normal tissue from the same cancer patients or in five gall bladder carcinomas. We conclude from these results that c-K-ras somatic mutational activation is a critical event in the oncogenesis of most, if not all, human cancers of the exocrine pancreas.     

Associating somatic mutation with clinical outcomes through kernel regression and optimal transport

Somatic mutations in cancer patients are inherently sparse and potentially high dimensional. Cancer patients may share the same set of deregulated biological processes perturbed by different sets of somatically mutated genes. Therefore, when assessing the associations between somatic mutations and clinical outcomes, gene-by-gene analysis is often under-powered because it does not capture the complex disease mechanisms shared across cancer patients. Rather than testing genes one by one, an intuitive approach is to aggregate somatic mutation data of multiple genes to assess their joint association with clinical outcomes. The challenge is how to aggregate such information. Building on the optimal transport method, we propose a principled approach to estimate the similarity of somatic mutation profiles of multiple genes between tumor samples, while accounting for gene–gene similarities defined by gene annotations or empirical mutational patterns. Using such similarities, we can assess the associations between somatic mutations and clinical outcomes by kernel regression. We have applied our method to analyze somatic mutation data of 17 cancer types and identified at least five cancer types, where somatic mutations are associated with overall survival, progression-free interval, or cytolytic activity.     

E-cadherin's dark side: Possible role in tumor progression

In the context of cancer, E-cadherin has traditionally been categorized as a tumor suppressor, given its essential role in the formation of proper intercellular junctions, and its downregulation in the process of epithelial-mesenchymal transition (EMT) in epithelial tumor progression. Germline or somatic mutations in the E-cadherin gene (CDH1) or downregulation by epigenetic mechanisms have been described in a small subset of epithelial cancers. However, recent evidence also points toward a promoting role of E-cadherin in several aspects of tumor progression. This includes preserved (or increased) E-cadherin expression in microemboli of inflammatory breast carcinoma, a possible "mesenchymal to epithelial transition" (MET) in ovarian carcinoma, collective cell invasion in some epithelial cancers, a recent association of E-cadherin expression with a more aggressive brain tumor subset, as well as the intriguing possibility of E-cadherin involvement in specific signaling networks in the cytoplasm and/or nucleus. In this review we address a lesser-known, positive role for E-cadherin in cancer.