Liu Wenhao, Dong Ruihong, Gao Shuai, Shan Xiaodi, Li Mian, Yu Zhaoyan, Sun Liang
College of Artificial Intelligence and Big Data For Medical Sciences, Shandong First Medical University & Shandong Academy of Medical Sciences, Jinan, Shandong, China.
Beijing Mentougou Hospital of Traditional Chinese Medicine, Beijing, China.
J Oncol. 2022 Jul 31;2022:2151396. doi: 10.1155/2022/2151396. eCollection 2022.
Lung adenocarcinoma (LUAD) is the commonest of the subtypes of lung cancer histologically. For this study, we intended to analyze the expression profiling of the immune-related genes (IRGs) from an independently available public database and developed a potent signature predictive of patients' prognosis.
Gene expression profiles and the clinical data of lung adenocarcinoma were gathered from the Gene Expression Omnibus database (GEO) and The Cancer Genome Atlas (TCGA), and the obtained data were split into a training set ( = 226), test set ( = 83), and validation set ( = 400). IRGs were then gathered from the ImmPort database. A prognostic model was constructed by analyzing the training set. Then the GO and KEGG analysis was performed, and a gene correlation prognostic nomogram was constructed. Finally, external validation, such as immune infiltration and immunohistochemistry, was performed.
The 110 genes were significant by univariate Cox regression analysis and randomized survival forest algorithm for the training set and showed a good distinction between the low-risk-score and high-risk-score groups in the training set ( < 0.0001) by screening for four prognosis-related genes (HMOX1, ARRB1, ADM, PDIA3) and validated by the test set GSE30219 (=0.0025) and TCGA dataset (=0.00059). Multivariate Cox showed that the four gene signatures were an individual risk factor for LUAD. In addition, the genes in the signatures were externally verified using an online database. In particular, PDIA3 and HMOX1 are essential genes in the prognostic nomogram and play an important role in the model of immune-related genes.
Four immune-related genetic signatures are reliable prognostic indicators for patients with LUAD, providing a relevant theoretical basis and therapeutic rationale for immunotherapy.
肺腺癌（LUAD）是肺癌组织学亚型中最常见的类型。在本研究中，我们旨在分析来自独立可用公共数据库的免疫相关基因（IRGs）的表达谱，并开发一种有效的预测患者预后的特征。
从基因表达综合数据库（GEO）和癌症基因组图谱（TCGA）收集肺腺癌的基因表达谱和临床数据，并将获得的数据分为训练集（n = 226）、测试集（n = 83）和验证集（n = 400）。然后从免疫数据库收集IRGs。通过分析训练集构建预后模型。然后进行基因本体（GO）和京都基因与基因组百科全书（KEGG）分析，并构建基因相关性预后列线图。最后，进行外部验证，如免疫浸润和免疫组织化学。
通过单变量Cox回归分析和随机生存森林算法，训练集中110个基因具有显著性，通过筛选四个与预后相关的基因（血红素加氧酶1（HMOX1）、β- arrestin 1（ARRB1）、肾上腺髓质素（ADM）、蛋白二硫异构酶A3（PDIA3）），训练集中低风险评分组和高风险评分组之间显示出良好的区分度（P < 0.0001），并在测试集GSE30219（P = 0.0025）和TCGA数据集（P = 0.00059）中得到验证。多变量Cox分析表明，这四个基因特征是LUAD的个体危险因素。此外，使用在线数据库对特征中的基因进行了外部验证。特别是，PDIA3和HMOX1是预后列线图中的关键基因，在免疫相关基因模型中起重要作用。
四个免疫相关基因特征是LUAD患者可靠的预后指标，为免疫治疗提供了相关的理论基础和治疗依据。
