【作者】 张琳；

【导师】 白元松；

【作者基本信息】 吉林大学 ， 肿瘤学硕士（专业学位）， 2023， 硕士

【摘要】 背景及目的:肺腺癌是非小细胞肺癌中最常见的亚型,在全世界范围内导致每年超过50万的死亡。尽管目前在分子诊断和靶向治疗领域取得了一些进展,肺腺癌患者的总体预后仍然不容乐观,大多数肺腺癌患者在初诊时即为局部晚期或已经发生远处转移。铁死亡是由Dixon等人最先提出的有别于自噬和细胞凋亡的一种细胞死亡,其主要特征为铁依赖和活性氧所致的细胞变化,如线粒体嵴的消失、线粒体膜的破裂和浓缩,其机制主要是膜上的脂质过氧化使得膜失去了选择渗透性。近年来,铁死亡逐渐被认为是清除恶性细胞的自适应功能,在抑制肿瘤的发生发展中发挥重要作用。本研究通过生物信息学方法分析了铁死亡相关基因在肺腺癌中的表达情况,并基于这些基因构建了能够预测肺腺癌患者预后的诺曼图,为肺腺癌患者的预后分层和个体化治疗提供新的方法和观点。方法:从公共数据库肿瘤基因图谱(TCGA)中获得肺腺癌患者的基因表达谱、生存资料及临床资料。比较铁死亡基因在肿瘤组织和癌旁正常组织中的表达量,将差异表达的基因纳入残差网络中进行训练,确定了由10个铁死亡基因构成的风险评分公式,根据得分将样本划分为高风险组和低风险组,通过Kaplan-Meier法比较两风险组间的预后差异。利用单因素COX回归和多因素COX回归确定两风险组间的风险比以及风险分组作为预后的预测因子的独立性。根据回归结果构建诺曼图,通过受试者工作特征曲线(ROC)、C指数、校准度分析曲线和临床决策曲线评价诺曼图的预测效能和稳定性。应用基因功能富集分析进一步探索两风险分组之间的生物学差异。结果:(1)在训练集中得到由CYBB,FURIN,DPP4,ETV4,RRM2,NR4A1,EPAS1,GCLC,TNFAIP3和AKR1C1构成的评分公式,根据得分将样本分为高风险组和低风险组,两组间预后有显著差异(p<0.0001),并在验证集中证实此差异(p=0.004)。(2)将风险分组和临床资料做单因素COX回归和多因素COX回归,低风险组相对于高风险组的风险更低,在单因素COX回归中的风险比为0.23(p<0.0001),在多因素COX回归中的风险比为0.26(p<0.0001)。(3)应用风险分组、p T分期、p N分期和肿瘤分期构建诺曼图,ROC曲线在训练集中在第1年、第3年和第5年的曲线下的面积(AUC)值为0.81、0.82和0.78;在验证集中第1年、第3年和第5年的AUC值为0.77、0.73和0.70。(4)通过基因功能富集分析发现高风险组在细胞增殖和侵袭的通路更加活跃,低风险组主要富集在和细胞清除及免疫反应的通路,高风险组的肿瘤纯度显著高于低风险组的肿瘤纯度(p<0.001)。两风险组间多种免疫细胞浸润程度有显著差异,肿瘤微环境差别较大。结论:(1)本研究通过生物信息学方法获得肺腺癌患者的风险分组模型,该模型是能够预测肺腺癌患者预后的独立因素。(2)从基因功能富集分析结果来看,高风险组较低风险组有更高的侵袭性,为风险分组模型提供分子生物学理论基础。(3)由风险分组及临床资料构建的诺曼图具有重要的预测价值,能够识别肺腺癌患者的风险并预测其预后。更多还原

【Abstract】 Background and objective:Lung adenocarcinoma is the most common subtype of non-small cell lung cancer,and accounts for more than 500,000 deaths annually worldwide.Despite some advances in the molecular diagnosis and targeted therapy at present,the overall survival has not improved significantly of patients with lung adenocarcinoma,and the majority of patients with lung adenocarcinoma have locally advanced disease or distant metastases at the time of initial diagnosis.Ferroptosis is a type of regulatory cell death,which is different from autophagy and apoptosis,first proposed by Dixon et al.Ferroptosis was mainly characterized as an iron-and reactive oxygen species dependent cellular changes,including mitochondria crista vanishes,mitochondrial membrane rupture and condensed mitochondrial membrane densities,and its mechanism primarily is lipid peroxidation that causes the loss of membrane selective permeability.In recent years,ferroptosis have gradually been recognized as a self-adaptation function to remove malignant cells,and serve an important role in the occurrence and development of tumors.This study analyzed the expression of ferroptosis related genes by bioinformatics in lung adenocarcinoma,and constructed a nomogram based on these genes which can predict prognosis for patients with lung adenocarcinoma,and contributed new approaches and perspectives for prognostic stratification and individualized therapy.Methods:The gene expression profiles,survival data and clinical data of lung adenocarcinoma were downloaded from The Cancer Genome Atlas(TCGA)public database.Compared the expression of ferroptosis genes in tumor tissue and tissue adjacent to tumor,used differentially expressed genes as the input of residual network for training,and determine the formula of risk score based on 10 ferroptosis related genes.Classified samples into high risk group and low risk group depending on their risk score.Kaplan-Meier survival curves were used to compare survival differences between two groups.Using univariate Cox regression and multivariate Cox regression evaluated the hazard ration between two risk groups and the independence of the risk group model as a prognostic predictor.Constructed the nomogram based on the regression result,the prognostic efficiency and stability of the nomogram were assessed by receiver operating characteristic curve(ROC),C-index,calibration curve and decision curve analysis.Performing gene functional enrichment analysis to further explore the biological differences between two risk groups.Results:(1)The formular of risk score was obtained in the training set which contained CYBB,FURIN,DPP4,ETV4,RRM2,NR4A1,EPAS1,GCLC,TNFAIP3 and AKR1C1.Based on the risk score,divided samples into high risk group and low risk group,and the prognosis of these tow groups was significant different(p<0.0001)and the difference was validated in the validation set(p=0.004).(2)Utilized patient risk group and clinical information for univariate cox regression analysis and multivariate cox regression analysis,and low risk group compared with high risk group had a lower risk,the hazard ratio was 0.23 in univariate cox regression analysis(p<0.0001),and 0.26 in multivariate cox regression analysis(p<0.0001).(3)Using risk group,p T stage,p N stage and tumor stage to construct nomogram.In the training set,the area under curve(AUC)was 0.81 at 1year,0.82 at 3 years and0.78 at 5 years.In the validation set,the AUC was 0.77 at 1year,0.73 at 3 years and0.70 at 5 years.(4)The gene functional enrichment analysis revealed that the signal pathways of cell proliferation and invasion were more active in high risk groups,and the signal pathways of cell clearance and immune responses were more active.The tumor purity was much higher in high risk group then low risk group(p<0.001).There were significant differences of several immune cell infiltration and tumor microenvironment between two risk groups.Conclusions:(1)This study using bioinformatics analysis developed a risk group model for lung adenocarcinoma,which was an independent prognosis factor for patients with lung adenocarcinoma.(2)According the results of gene functional enrichment analysis,the high risk group had a higher tendency to be aggressive than the low risk group,which provide a molecular biology theoretical basis for the risk group model.(3)The nomogram which combing the risk group and clinical data had a pivotal prognostic value,and was able to identify the risk and predict prognosis for patients with lung adenocarcinoma.更多还原