A network-based signature to predict the survival of non-smoking lung adenocarcinoma.
A network-based signature to predict the survival of non-smoking lung adenocarcinoma.
复制标题
基于网络的签名来预测非吸烟肺腺癌的生存率
DOI：
10.2147/cmar.s163918
复制
发表时间：
2018
影响因子：
3.3
通讯作者：
Xu L
中科院分区：
文献类型：
--
作者：
Mao Q;Zhang L;Zhang Y;Dong G;Yang Y;Xia W;Chen B;Ma W;Hu J;Jiang F;Xu L
在过去的十年中，非吸烟肺腺癌（LAC）患者数量的大幅增加引起了广泛的关注。然而，可以指导精确治疗的有效生物标志物在识别高风险患者方面仍然有限。在这里，我们提供了一个基于网络的签名来预测非吸烟LAC的生存。材料和方法从The Cancer Genome Atlas和Gene expression Omnibus下载基因表达谱。通过加权基因共表达网络分析，鉴定出显著基因共表达网络和枢纽基因。利用基因本体分析了共表达网络的潜在机制和途径。通过惩罚Cox回归分析构建预测特征，并在两个独立数据集上进行测试。结果在4个基因表达Omnibus数据集中，两个不同的共表达模块与不吸烟状态显著相关。基因本体揭示了蓝色模块的核分裂和细胞周期途径是主要机制，绿松石模块中的基因参与淋巴细胞活化和细胞粘附途径。以最佳lambda值从枢纽基因中选择17个基因，构建预后特征。预后特征区分非吸烟LAC患者的生存（训练：风险比[HR]=3.696, 95% CI: 2.025 ~ 6.748， P<0.001；检验：HR=2.9, 95% CI: 1.322 ~ 6.789， P=0.006； HR=2.78, 95% CI: 1.658 ~ 6.654， P=0.022），在训练和验证数据集中具有中等预测能力。结论非吸烟LAC患者的预后指标具有良好的预测价值，有助于临床实践和精准治疗。
Background A substantial increase in the number of non-smoking lung adenocarcinoma (LAC) patients has been drawing extensive attention in the past decade. However, effective biomarkers, which could guide the precise treatment, are still limited for identifying high-risk patients. Here, we provide a network-based signature to predict the survival of non-smoking LAC. Materials and methods Gene expression profiles were downloaded from The Cancer Genome Atlas and Gene Expression Omnibus. Significant gene co-expression networks and hub genes were identified by Weighted Gene Co-expression Network Analysis. Potential mechanisms and pathways of co-expression networks were analyzed by Gene Ontology. The predictive signature was constructed by penalized Cox regression analysis and tested in two independent datasets. Results Two distinct co-expression modules were significantly correlated with the non-smoking status across 4 Gene Expression Omnibus datasets. Gene Ontology revealed that nuclear division and cell cycle pathways were main mechanisms of the blue module and that genes in the turquoise module were involved in lymphocyte activation and cell adhesion pathways. Seventeen genes were selected from hub genes at an optimal lambda value and built the prognostic signature. The prognostic signature distinguished the survival of non-smoking LAC (training: hazard ratio [HR]=3.696, 95% CI: 2.025–6.748, P<0.001; testing: HR=2.9, 95% CI: 1.322–6.789, P=0.006; HR=2.78, 95% CI: 1.658–6.654, P=0.022) and had moderate predictive abilities in the training and validation datasets. Conclusion The prognostic signature is a promising predictor of non-smoking LAC patients, which might benefit clinical practice and precision therapeutic management.