| Background and purposeGastric cancer is one of the most common malignant tumors in the world,and according to the latest statistics,the number of new cases of gastric cancer is the sixth highest in the world.China is the high incidence area of gastric cancer in the world,the morbidity and mortality rate account for the second and third place in the tumor spectrum,respectively.Because of the low rate of initial diagnosis of gastric cancer and it’s easy to metastasize,it leads to high mortality and poor prognosis of gastric cancer patients.Therefore,the individualized treatment of gastric cancer is particularly important.The establishment of the prediction model of prognosis risk of gastric cancer can better help clinical screening of applicable populations,provide guidance for individualized treatment and have a certain judgment on the prognosis of patients.In recent years,several studies have confirmed the predictive role of genomic information in tumor prognosis,and can better understand the development process of tumor development by analyzing genomic information,and have certain guiding significance for patient’s prognosis.The first part of this paper is to construct a genomic risk prediction model related to the prognosis of gastric cancer by analyzing the genomic information of gastric cancer from the TCGA public database.With the in-depth study of the molecular basis of tumors,the birth of molecular target drugs has provided a new therapeutic direction for gastric cancer.Human epidermal growth factor 2(HER-2)is used widely.The HER-2 positive rate of gastric cancer is about 12%-20%.Due to the large heterogeneity of HER-2 positive gastric cancer,only some patients can benefit from it.How to select a more suitable treatment population is still the focus of our research.The establishment of a prognostic model will help us to screen the dominant population,provide guidance for individualized treatment of patients,while predicting the patient’s prognosis.The second part of this paper is to establish a prognostic nomogram in HER-2-positive gastric cancer patients treated with trastuzumab by analyzing the clinicopathological characteristics of patients with HER-2 positive gastric cancer who have received trastuzumab.Method and Materials1.Part Ⅰ:Download the data of 375 cases of gastric cancer tissues and 32 cases of gastric cancer adjacent tissues via the Cancer Genome Atlas(TCGA),select functional annotations as protein-coding genes and extract their gene expression quantification to complete construction of gastric cancer tissue genomics database.Differential analysis of genes in 375 cancer tissues and 32 adjacent tissues were performed by R language.Screening was performed using the t test’s P value and threshold,and the selection criteria were up-regulation or down-regulation fold difference>2 and P<0.001,and differentially expressed genes were obtained by screening.Cox proportional hazards model was used to conduct univariate and multivariate analysis to evaluate the impact on the prognosis of gastric cancer,build a predictive risk prediction type based on genomic information,and use the receiver operating characteristic(ROC)curve and the area under the curve(AUC)to assess the ability of prognostic models to identify risks.2.Part Ⅱ:Retrospective analysis of the clinical characteristics of 75 CASES of HER-2 positive gastric cancer patients treated with trastuzumab in Zhengzhou University Cancer Affiliated Hospital,Zhengzhou University Cancer Affiliated People’s Hospital and Zhengzhou University First Affiliated Hospital from January 2015 to May 2019.The Cox regression model was used to analyze the clinicalpathological factors that may affect the prognosis,to determine the independent prognostic factors.R language was used to establish a prognostic nomogram,draw a progression-free survival calibration curve and compared with the actual observations.Bootstrap method was used for internal verification,and the consistency index(C-index)was calculated to evaluate the accuracy of the model.External verification:Randomly draw existing samples,draw 50 samples for verification,using R language to draw ROC curves,and obtain the ability of AUC to assess the risk of prognostic models.ResultPart Ⅰ1.A total of 375 cancer tissues and 32 adjacent tissues were included in the TCGA database.Gene expression levels of protein-coding genes were analyzed in R.3.5.1.A total of 1632 DEGs were screened out(FDR<0.01,|log2FC |>1).There are 855 DEGs with high expression and 777 DEGs with low expression among them.2.The 1632 DEGs screened were analyzed by univariate analysis using the COX function of R 3.5.1.and the results showed that 38 DEGs were related to the OS of gastric cancer(P<0.05)and 10 DEGs with P<0.001 were selected.These 10 DEGs were included in the COX multivariate analysis,and 7 DEGs(RDH8,CGB8,MATN3,FAM9B,CREB3L3,CYP19A1,MMRN1)were obtained.Finally,the 7 DEGs were selected to build a gastric cancer prognosis risk assessment model,RS=(0.00048*RDH8 mRNA)+(0.00201*CGB8 mRNA)+(0.00031*MATN3 mRNA)+(0.00208*FAM9B mRNA)+(0.00018*CREB3L3 mRNA)+(0.00091*CYP19A1 mRNA)+(0.00034*MMRN1 mRNA).Part Ⅱ1.We enrolled a total of 75 patients with HER-2 positive gastric cancer treated with trastuzumab,and analyzed their clinicalpathological characteristics.After univariate analysis,there were 14 factors related to PFS(P<0.2).They are gender,age,D-dimer.CEA,CA724,LNR,number of organ metastases,liver metastases,M stage at first diagnosis,TNM stage at first diagnosis,whether the primary lesion is removed,and combined chemotherapy Frequency,whether maintenance treatment,local treatment.2.The above 14 factors related to PFS(P<0.2)were analyzed by multivariate analysis.Finally,there were 7 independent prognostic factors(P<0.05)of PFS,which were age,CEA,LNR,liver metastasis,and initial diagnosis,TNM staging,frequency of combined chemotherapy,and whether it is treated locally.3.Based on the results of multivariate analysis,independent prognostic factors of PFS for trastuzumab in the treatment of HER-2 positive gastric cancer were established.A prognostic nomogram was drawn using R language to predict progression-free survival of trastuzumab in treatment of HER-2 positive gastric cancer.And using internal verification,the consistency index is calculated to be 0.883,suggesting that the model has a good discrimination.Draw a calibration curve to show that the average absolute error between the predicted value and the true value obtained through this nomogram model is 0.020.The graph shows that the prediction and actual observed progression-free survival rate have a strong consistency.The nomogram model has better resolution.External verification,50 samples were taken for verification,and the ROC curve was drawn using R language.The AUC was 0.883,which shows that the nomogram has higher accuracy in predicting the prognosis of gastric cancer.ConclusionsPart Ⅰ1.According to the statistical analysis of the protein genes encoding gastric cancer in the TCGA database,seven DEGs related to the prognosis of gastric cancer were obtained:RDH8,CGB8,MATN3,FAM9B,CREB3L3,CYP19A1,MMRN1.2.Based on these 7 DEGs,a risk model related to the prognosis of gastric cancer was established,RS=(0.00048*RDH8 mRNA)+(0.00201*CGB8 mRNA)+(0.00031*MATN3 mRNA)+(0.00208*FAM9B mRNA)+(0.00018*CREB3L3 mRNA)+(0.00091*CYP19A1 mRNA)+(0.00034*MMRN1 mRNA).Part Ⅱ1.After COX univariate and multivariate analysis,there were 7 independent prognostic factors(P<0.05)related to PFS in trastuzumab-treated HER-2 positive gastric cancer patients,which were age,CEA,LNR,liver metastasis,and initial diagnosis TNM staging,frequency of combined chemotherapy use,and whether it was treated locally.12.A nomogram model was established based on the results of multivariate analysis of COX regression. |
PDF Full Text Request