医学信息2026，Vol.39Issue(10)：32-39,8.DOI:10.3969/j.issn.1006-1959.2026.10.005
基于机器学习的肥胖相关基因特征预测胃癌患者的预后和治疗反应
Obesity-related Gene Features Based on Machine Learning in Predicting the Prognosis and Treatment Response of Patients with Gastric Cancer
高梓扬 1赵崇玉 2成龙 3蓝煜 4陈承业 1陈俊良1
作者信息
- 1. 肇庆市第一人民医院肿瘤科,广东 肇庆 526000
- 2. 暨南大学附属第一医院外科,广东 广州 510630
- 3. 广州医科大学附属惠州医院外科,广东 惠州 516000
- 4. 中山大学肿瘤防治中心内科,广东 广州 510000
- 折叠
摘要
Abstract
Objective To explore the obesity-related genes associated with the prognosis of gastric cancer,and to construct a prediction model to evaluate the prognosis and treatment response of patients.Methods Based on the transcriptome data of gastric cancer in TCGA and GEO databases and obesity-related genes in GeneCards,obesity genes associated with gastric cancer were screened,and 58 prognostic genes were identified by univariate Cox regression analysis.The 101 machine learning models(including Enet,RSF,etc.)were used to construct a prognostic prediction model through 10-fold cross-validation,and the Enet[alpha=0.1]model was selected as the best solution.Kaplan-Meier analysis was used to evaluate the survival differences between the high-risk group and low-risk group.The GSE84437 data set was used for model validation,and Cox regression analysis further assessed the independence of the risk score.Immunotherapy response was evaluated by tumor mutation burden(TMB)and immune response score(TIDE).Drug sensitivity assessment was used to compare the response of the two groups to common chemotherapy drugs.The differential expression of obesity genes in gastric cancer cells was verified by cell experiments.Results A total of 239 obesity-related genes were screened,and 58 obesity-related genes were identified for the prognosis of gastric cancer.Based on these genes,the constructed Enet[alpha=0.1]model divided patients into high-risk group and low-risk groups,and the survival time of the high-risk group was significantly shorter than that of the low-risk group.The AUC of the model for predicting the 1-,3-,and 5-year overall survival rates of gastric cancer patients was 0.713,0.705,and 0.712,respectively,showing excellent performance.The prognostic nomogram C index integrating clinical factors was 0.706,and the predictive ability was good.Immunological analysis showed that the TMB of the low-risk group was higher than that of the high-risk group,and the TIDE score was lower,suggesting that the low-risk group was more sensitive to immunotherapy.Drug sensitivity assessment showed that high-risk groups might be more likely to benefit from chemotherapy and targeted therapy.Cell experiments verified the high expression of CARTPT and APOE in gastric cancer cells,while the expression of PDSS1 was down-regulated.Conclusion The constructed obesity gene characteristic model can effectively predict the prognosis,immune response and treatment response of patients with gastric cancer,and can provide support for clinical decision-making.关键词
胃癌/肥胖/机器学习/预后Key words
Gastric cancer/Obesity/Machine learning/Prognosis分类
医药卫生引用本文复制引用
高梓扬,赵崇玉,成龙,蓝煜,陈承业,陈俊良..基于机器学习的肥胖相关基因特征预测胃癌患者的预后和治疗反应[J].医学信息,2026,39(10):32-39,8.
