内科

目的筛选与预后相关的SNHG家族lncRNA，并构建胃腺癌患者生存预测列线图模型。方法从TCGA数据库获取STAD队列377份癌组织及34份正常组织RNA-seq数据及临床随访资料，以BioSysDB数据库提供的SNHG家族注释为基础，纳入31个SNHG家族lncRNA为候选分子集，采用R软件limma包筛选胃腺癌组织与癌旁正常组织间差异表达的SNHG家族lncRNA。依次通过Kaplan-Meier生存分析（log-rank检验）和单因素Cox回归分析筛选潜在预后影响因素，再将其纳入多因素Cox回归分析（逐步回归法），最终确定胃腺癌的独立预后因素。基于上述独立预后因素，采用R软件rms包构建预测1年、2年、3年总生存率的列线图模型；采用R软件survivalROC包绘制1年、2年、3年时间依赖受试者工作特征曲线并计算曲线下面积（AUC）以评估区分度；Bootstrap自抽样（1000次）绘制校准曲线以评估校准度。结果从31个候选SNHG家族lncRNA中共筛选出19个在胃腺癌组织中差异表达的SNHG家族lncRNA，其中14个上调、5个下调。经Kaplan-Meier生存分析（log-rank检验）和单因素Cox回归分析，确定年龄、SNHG1、SNHG3、SNHG4、SNHG12、SNHG20均为胃腺癌潜在预后影响因素（均P<0.05）。进一步行多因素Cox回归分析显示，SNHG1（HR=0.469，95%CI：0.288～0.856）、SNHG3（HR=0.505，95%CI：0.298～0.855）、SNHG4（HR=0.425，95%CI：0.252～0.717）、SNHG12（HR=0.369，95%CI：0.211～0.646）、SNHG20（HR=0.371，95%CI：0.217～0.636）高表达均是胃腺癌患者总生存率的独立保护因素（均P<0.05）。基于上述5个SNHG家族lncRNA构建的列线图模型预测1年、2年、3年总生存率的AUC分别为0.797、0.808、0.830，校准曲线显示预测生存概率与实际总生存率一致性尚可。结论基于5个SNHG家族（SNHG1、SNHG3、SNHG4、SNHG12、SNHG20）lncRNA构建的列线图模型对胃腺癌患者1年、2年、3年总生存率具有良好的预测效能，可为胃腺癌患者的个体化预后评估提供新型量化工具。

Objective To screen prognosis-related SNHG-family lncRNAs and construct a nomogram model for survival prediction in patients with gastric adenocarcinoma. Methods RNA-seq data of 377 tumor tissues and 34 normal tissues, as well as clinical follow-up data, of the STAD cohort were obtained from the TCGA database. Based on the annotations of the SNHG family provided by the BioSysDB database, 31 SNHG-family lncRNAs were included as the candidate molecular set. The limma package of R software was used to screen differentially expressed SNHG-family lncRNAs between gastric adenocarcinoma tissues and normal adjacent tissues. Potential prognostic factors were screened sequentially by Kaplan-Meier survival analysis (log-rank test) and univariate Cox regression analysis, which were further included in multivariate Cox regression analysis (stepwise regression method) to identify independent prognostic factors for gastric adenocarcinoma. Based on the abovementioned independent prognostic factors, the rms package of R software was applied to construct a nomogram model for predicting 1-, 2-, and 3-year overall survival rates. The survivalROC package of R software was adopted to plot time-dependent receiver operating characteristic (ROC) curves at 1-, 2- and 3-year time points and calculate areas under the curves (AUCs) for discrimination assessment. Bootstrap resampling (1000 iterations) was performed to generate calibration curves for calibration evaluation. Results A total of 19 differentially expressed SNHG-family lncRNAs in gastric adenocarcinoma tissues were screened out from 31 candidate SNHG-family lncRNAs, including 14 up-regulated and 5 down-regulated lncRNAs. Kaplan-Meier survival analysis (log-rank test) and univariate Cox regression analysis identified age, SNHG1, SNHG3, SNHG4, SNHG12, and SNHG20 as potential prognostic factors for gastric adenocarcinoma (all P<0.05). Further multivariate Cox regression analysis revealed that high expression of SNHG1 (HR=0.469, 95%CI: 0.288-0.856), SNHG3 (HR=0.505, 95%CI: 0.298-0.855), SNHG4 (HR= 0.425, 95%CI: 0.252-0.717), SNHG12 (HR=0.369, 95%CI: 0.211-0.646), and SNHG20 (HR=0.371, 95%CI: 0.217-0.636) were independent protective factors for overall survival in patients with gastric adenocarcinoma (all P<0.05). The AUCs of the nomogram model constructed based on the abovementioned 5 SNHG-family lncRNAs for predicting 1-, 2-, and 3-year overall survival were 0.797, 0.808, and 0.830, respectively. Calibration curves showed acceptable consistency between predicted survival probability and actual overall survival. Conclusion The nomogram model built upon 5 SNHG-family lncRNAs (SNHG1, SNHG3, SNHG4, SNHG12, SNHG20) exhibits favorable predictive performance for 1-, 2-, and 3-year overall survival in gastric adenocarcinoma patients, and it can serve as a novel quantitative tool for individualized prognostic assessment of gastric adenocarcinoma.