Abstract 1085: A bioinformatics approach to evaluate the involvement of CDH1/E-cadherin in solid tumors and to identify breast cancer biomarkers

Abstract

Abstract Cancer is a complex disease characterized by abnormal proliferation of different cell types among which solid tumors (ST) stand out due to their high incidence and mortality rates. Of many ST, breast cancer (BC) is the 2nd most common cancer worldwide and the most frequent among women (∼1.67 million new cases in 2012). Ductal breast cancer (DBC) and lobular breast cancer (LBC) are the most frequent BC types. The human gene CDH1 encodes Epithelial Cadherin (ECad), a calcium-dependent cell-cell adhesion molecule defined as tumor suppressor since its absence prevents adherens junctions and associates to tumor progression and poor patient outcome. In cancer research, a great deal of data is available in the literature, providing an abundant source of knowledge for biomedical research, although the information is disaggregated. The use of bioinformatics tools (BT) dedicated to text mining (TM) and network biology has become very useful to study cancer pathology from a global perspective since integrate the vast amount of data available. In this study, we have implemented BT (DisGeNET, Cytoscape, COSMIC, Hippie, PANTHER, String) to systematically assess the relationship between CDH1/ECad and ST and to identify potential biomarkers for DBC and LBC. To identify CDH1 related diseases, a gene-disease network analysis was performed using DisGeNET/Cytoscape. In a query restricted to curated databases, CDH1 was associated (biomarker/genetic variation/ therapeutic target) to 29 diseases among which 20 were ST. When all databases (curated, predicted and literature) were included, the terms related to CDH1 expanded to 173 diseases (98 ST). A manual term-curation related to CDH1 (80 ST) was performed, with a total of 286 publications reviewed. In 199/286 (70%) reports, the association between CDH1 and ST was confirmed (67/286 (23%) inconclusive, 20/286 (7%) unconfirmed). As a result, 75/80 ST terms were validated as related to CDH1. An analysis done with the COSMIC BT led to the identification of 376 somatic mutations on the CDH1 gene in 11/20 ST terms from the DisGeNET output (curated databases). For BC, 3045 genes were found using TM tools as related to this term (CDH1: biomarker/genetic variation). For DBC, 42 genes were related with proven evidence (high score: BAG1, CDH1, ATF4, CLDN4, PTGER1, SERPINB5). To analyze protein-protein interaction (PPI), the HIPPIE BT was used. A selected DBC biomarkers were identified from a PPI network of 448 nodes, among which HDAC1, SKP2, CFTR, HSPA8, SMAD7, TGM2, ATF2, MDM2 were listed as putative biomarkers based on the ´Linkage method´. For LBC, 28 genes were found as associated (3 with proven evidence) and generate a PPI network of 216 nodes; ERBB2IP, SRC, EGFR, EP300 y HDAC1 were listed as potential biomarkers. Altogether, BT allowed a systematic assessment of current knowledge of CDH1/ECad and led to the identification of novel putative biomarkers for DBC and LBC. Citation Format: Maria F. Abascal, Maria J. Besso, Evangelina Aparicio, Marina Rosso, Victoria Mencucci, Laura I. Furlong, Monica H. Vazquez-Levin. A bioinformatics approach to evaluate the involvement of CDH1/E-cadherin in solid tumors and to identify breast cancer biomarkers. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 1085. doi:10.1158/1538-7445.AM2015-1085