Data_Sheet_4_Identification of Subtype-Specific Three-Gene Signature for Prognostic Prediction in Diffuse Type Gastric Cancer.pdf

Gastric cancer (GC), with high heterogeneity, can be mainly classified into intestinal type and diffuse type according to the Lauren classification system. Although a number of differences were reported between these two types, no study on the Lauren subtype-specific multi-gene signature for evaluation of GC prognosis has been conducted, and the molecular mechanism underlying its poor prognosis has still remained elusive. Therefore, this study aimed to explore subtype-specific multi-gene signature for prognostic prediction in different subtypes of Lauren classification. With combination of the least absolute shrinkage and selection operator (LASSO) algorithm and the Akaike information criterion (AIC), the 3-gene subtype-specific prognostic signature was successfully established in diffuse type GC using GSE62254 dataset. Following the calculation of risk score (RS) based on 3-gene signature, the nomogram models were established to predict 1-, 3-, and 5-year overall survival in diffuse type GC. Moreover, the prognostic predictive nomogram model of diffuse type GC was also proved to be effective for validation of GSE1549 dataset and by a Gene Expression Omnibus (GEO)-based meta-analysis. In the analysis of the correlation between RS and clinical-pathological characteristics, RS and two genes of the 3-gene signature (EMCN and COL4A5) were found to be positively correlated with peritoneal metastasis. Furthermore, EMCN and COL4A5, rather than CCL11, were proved to be able to enhance the adhesion ability of MKN45 and NUGC4 cells to peritoneal mesothelial cell line HMR-SV5. Eventually, it was proved that COL4A5 promoted peritoneal metastasis by activating Wnt signaling pathway, whereas the upregulation of integrin family genes mediated by FAK-AKT/ERK/STAT3 signaling pathway activation is involved in peritoneal metastasis promotion function of EMCN. Taken together, our study identified the subtype-specific 3-gene signature in diffuse type GC, which could effectively predict the patients' OS and might explain the molecular mechanisms in presence of its poor prognosis.

胃癌（Gastric cancer, GC）具有高度异质性，根据劳伦分型（Lauren classification）系统，其主要可分为肠型与弥漫型两个亚型。尽管已有研究报道了这两种亚型间存在诸多差异，但目前尚无针对劳伦亚型特异性多基因特征标记物用于胃癌预后评估的相关研究，其不良预后背后的分子机制仍尚不明确。为此，本研究旨在探索适配劳伦分型不同亚型的亚型特异性多基因特征标记物，以用于预后预测。本研究结合最小绝对收缩和选择算子（Least absolute shrinkage and selection operator, LASSO）算法与赤池信息准则（Akaike information criterion, AIC），利用GSE62254数据集成功构建了弥漫型胃癌的3基因亚型特异性预后特征标记物。基于该3基因特征标记物计算风险评分（risk score, RS）后，本研究进一步构建了列线图（nomogram）模型，用于预测弥漫型胃癌患者的1年、3年及5年总生存期（Overall Survival, OS）。此外，通过GSE1549数据集以及基于基因表达综合数据库（Gene Expression Omnibus, GEO）的荟萃分析验证，本研究构建的弥漫型胃癌预后预测列线图模型均展现出良好的预测效能。在风险评分与临床病理特征的相关性分析中，研究发现风险评分以及3基因特征标记物中的EMCN与COL4A5两个基因均与腹膜转移呈正相关。进一步功能实验证实，EMCN与COL4A5（而非CCL11）能够增强MKN45与NUGC4细胞对腹膜间皮细胞系HMR-SV5的黏附能力。最终研究证实，COL4A5可通过激活Wnt信号通路促进腹膜转移；而EMCN的促腹膜转移功能，则依赖于FAK-AKT/ERK/STAT3信号通路活化所介导的整合素家族基因上调。综上，本研究成功鉴定出弥漫型胃癌的亚型特异性3基因特征标记物，该标记物可有效预测患者的总生存期，并可为阐释其不良预后的分子机制提供理论依据。