| Objective: Gastric cancer is a highly heterogeneous malignant tumor that has accumulated a variety of genetic and epigenetic changes in the process of evolution,and the tumor biological characteristics of different patients vary greatly.The existing histopathological classification is difficult to comprehensively and objectively reflect the inherent characteristics of tumors,and cannot meet the needs of personalized treatment of gastric cancer in the era of precision medicine.With the advancement of high-throughput microarrays and next-generation sequencing technology,the molecular typing of gastric cancer has gradually matured,and the understanding and application of molecular subtype have become the basis and necessary conditions for the realization of precision treatment of gastric cancer.In 2014,the TCGA project proposed four molecular subtypes of gastric cancer,including EBV virus positive,microsatellite instability(MSI),genome stabilization(GS)and chromatin instability(CIN),which showed great significance for in-depth understanding of the complexity of the biological behavior of gastric cancer,and also attracted widespread attention.However,there is still a long way to go to guide clinical practice.In particular,the CIN subtype has the largest population and poor homogeneity,and due to the lack of in-depth research,it has not been possible to determine its exact characteristics for subdivision,and the consistency in drug efficacy and prognosis is very low.With the development of treatment,the survival of stomach cancer has been significantly prolonged,and as a result,temporal and spatial heterogeneity has become more and more recognized.The basic question of whether the subtype is a biologically independent entity or a gradual relationship has remained debated since the introduction of tumor molecular subtypes,and patient management and treatment strategies in both cases are completely different.Assuming that each static sample is equivalent to a snapshot in tumor evolution,as long as the sample size is large enough,the genetic imprint of each sample will occupy part of the development trajectory,allowing time series sampling to be simulated to construct a "dynamic" disease process from the static sample by computational methods.In this study,we aimed to construct a gastric cancer TCGA molecular subtype classifier from the molecular expression level and explore the heterogeneity of CIN subtype gastric cancer,use a large number of static samples to simulate the evolutionary process of "dynamic" CIN subtype gastric cancer,and identify the dominant population of immune checkpoint inhibitor(ICB)treatment and the feasibility of targeted drugs for patients who do not respond to ICB treatment based on the evolutionary model,and further discover biomarkers that can predict treatment benefits.Method: 1.Acquisition and pretreatment of TCGA gastric cancer training sets(n=412),validation sets consisting of 13 GEO gastric cancer chip datasets(total n=1250),and 2single-cell validation sets(total cell number 61937).2.Preliminary screening of characteristic genes of each subtype by non-parametric k-w test and Logistic-lasso regression combined with multiplex comparison.3.Predict the molecular subtype of the validation set sample and the epithelial cell cluster in the single cell validation set based on the subtype characteristic gene.4.Gene collection enrichment analysis(GSEA)with KEGG as the background gene set reflects the biological characteristics of each molecular subtype.5.Using the manifold learning algorithm,an evolutionary model of CIN subtype stomach cancer was constructed.6.Acquisition and pre-treatment of anti-PD1 for metastatic gastric cancer dataset and four other datasets of non-gastric cancers with ICB treatment.7.Use the TIDE algorithm and Submap algorithm to predict the therapeutic response of CIN subtypes of different evolutionary trajectories to ICB.8.The deconvolution algorithm cibersort,EPIC and single-sample enrichment analysis(ss GSEA)was used to calculate the immune infiltration of CIN subtypes with different evolutionary trajectories.9.Based on the calculation of interaction of ligand-recipient genes,the communication level of MHC-I and immune checkpoints between cell clusters are estimated.10.Use weighted gene co-expression network analysis(WGCNA)to identify related gene networks and core genes that evolve from early trajectory to the ICB response direction.11.Key mutations in the evolutionary trajectory of each CIN subtype were calculated using nuclear regression.12.The CCLE cell line dataset(n=37)and the compound sensitivity datasets PRISM and CTRP were obtained.13.Based on compoundsensitive AUC values of cell line and CMAP tool to predict the compound-sensitive information of training set and validation set samples.14.Screening of MDK highly expressed cell lines and evaluation of knockdown efficiency of si RNAs using q RT-PCR and western blotting.15.Evaluate the relationship between MDK and the m TORC1 pathway using GSEA and Western blotting.16.Evaluate the proliferative viability and sensitivity of cells to AZD8055 using MTS method and colony formation experiments.17.Effects of different concentrations of AZD8055 on m TORC1 pathway proteins before and after MDK knockout were detected using western blotting.Results: 1.Construct a TCGA molecular subtype classifier,in which CIN subtype gastric cancer is heterogeneous in molecular expression level,which can be further divided into 3 CIN subtypes,and the prognosis and biological characteristics of each subtype are significantly different.(1)The 57 characteristic genes screened had good classification of EBV,MSI,GS,and CIN subtypes in the training set.K-means consistent clustering further divides the CIN subtype into three subtypes(CIN1/CIN2/CIN3)under optimal classification.(2)A comparison of gene expression levels characterized by subtypes showed that the six-molecule subtype was consistent in the classification of the training set and the validation set.GSEA showed that the EBV and GS subtypes were characterized by immune activation and cell adhesion motility,respectively,the MSI and CIN2 subtypes were closely associated with cell proliferation,and the CIN1 and CIN3 subtypes were closely linked to metabolic pathways.The MSI and CIN3 subtypes with better prognosis have a low proportion of tumor recurrence,and the GS,CIN1 and CIN2 subtypes with significantly worse prognosis increase in the proportion of advanced tumors and the probability of tumor recurrence is high,among which the GS subtype dominated by diffuse gastric cancer mainly recurs as peritoneal implantation and ascites metastasis,the CIN2 subtype dominated by gastrointestinal cancer tends to recur in the liver and intraabdominal lymph nodes,and the CIN1 subtype has a higher recurrence rate for each part.2.At the level of molecular expression,the evolutionary process of CIN subtype gastric cancer is from metabolic changes to rapid proliferation.The evolutionary models constructed from CIN subtype tumor samples and epithelial cells using manifold learning showed that each CIN subtype had a gradual relationship in evolution,mainly reflected in the evolutionary process from metabolic changes to rapid proliferation,but at the same time,independent branches could be developed in evolution.From the normal epithelium to the end of the proliferative subtype CIN epithelium,the differentiation of tumor differentiation gradually decreases.The CIN3 subtype has a good prognosis and a low probability of distant recurrence because it is mostly in the early stages of evolution and has a strong desmosome junction with other high differential gastric epithelial cells.3.CIN subtype of gastric cancer evolves to form a branch similar to the tumor with MSI characteristics,and this evolutionary trajectory is more likely to benefit from ICB treatment.(1)When evolving from metabolic CIN tumors to the branching point of proliferative CIN tumors,a small number of CIN subtype samples tend to acquire genetic characteristics of MSI subtype gastric cancer and form independent evolutionary branch C1.Analysis of the expressed genes showed that the evolutionary trajectory from early clonal C3 to C1 clade was predicted to be more beneficial from ICB treatment than the proliferative CIN gastric carcinoma terminal branch C2.(2)Evaluation of tumor microenvironment(TME)showed higher levels of MHC-I and immune checkpoint receptor communication between C1 branch epithelial cells and CD8+ T cells in the TME of C1 branch samples,activated CD8+ T cells and M1 macrophages prior to depletion.The CD8+ T cells in the C2 branch sample TME were mainly depleted CD8+ T cells.(3)The use of WGCNA to identify gene modules that are significantly positively correlated with the pseudo-time changes of C3 to the C1 branch and not related to the evolution to the end of C2,and the key gene networks in the modules are significantly enriched in immune pathways,showing that gastric carcinomas of the CIN subtype that respond to the evolutionary trajectory of ICB treatment have higher inflammatory activating molecular signatures.Among them,the network core gene NFKB1 was significantly positively correlated with the expression of multiple immune checkpoint genes in CIN subtype gastric carcinoma,and the expression of NFKB1 and network core gene signature in C1 branch epithelial cells was also significantly higher than that of C2 branch.4.ARID1 A and LRRK2 are identified as the characteristic mutation genes that reflect the evolution of CIN subtype gastric cancer to benefit from ICB treatment.Tumors accumulate mutations in evolution,tumor mutational burden(TMB)levels increase with the progression of pseudo-time series,and TMB rises to a higher degree but is not significant compared with the evolutionary trajectory of C3 to C1 branches to the end of C2.Nuclear regression analysis showed that ARID1 A and LRRK2 gene mutations formed two fates after initial cloning to the evolutionary branch point,and the mutation probability gradually increased in the C1 evolutionary trajectory predicting benefit from ICB,while the mutation probability decreased with evolutionary time in the C2 terminal branch that did not respond to ICB.NFKB1 is significantly higher expressed in CIN subtypes of gastric cancers with ARID1 A or LRRK2 mutations.5.Gastric carcinoma of the CIN subtype with low response to ICB therapy is sensitive to the m TOR inhibitor AZD8055 and has a higher sensitivity with tumor evolution.After validating the feasibility of predicting gastric cancer sample drug-sensitive AUC using Ridge Regression based on gastric cancer cell line drug-sensitive AUC information,analysis of C3 and C2 samples with low response to ICB treatment showed that the C2 end samples were generally sensitive to m TOR inhibitors,of which AZD8055 had consistent results in the prediction of training sets and validation sets based on PRISM and CTRP drug-sensitive datasets,respectively,and compared with other compounds,AZD8055 also had lower drugsensitive AUC in C3.At the same time,the algorithm based on gene expression perturbation supports that all CIN subtypes are sensitive to AZD8055.6.MDK can promote the formation and evolution of gastric carcinoma of CIN subtype,and its high expression suggests that CIN subtype gastric carcinoma is sensitive to AZD8055.(1)MDK expression increased significantly after the branching point of the initial evolutionary branch of CIN subtype gastric cancer samples and epithelial cells and was expressed in more than half of the CIN subtype gastric cancer epithelial cells.Survival analysis showed that MDK was a significant prognostic factor for gastric cancer and even pan-solid tumors.(2)In the CIN subtype gastric cancer samples,MDK positive correlation genes were significantly enriched in proliferation and m TORC1 pathways,and MDK expression levels were significantly and negatively correlated with the drug-sensitive AUC values of AZD8055.In the CIN subtype cell line,the proliferative viability of tumor cells after MDK knockdown decreased significantly.Western blotting showed that m TORC1 pathway related proteins had higher levels of phosphorylation and were more sensitive to AZD8055 in MDK highly expressed gastric cancer cell lines.MTS detection of cell proliferation viability showed a decrease in sensitivity of CIN subtype gastric cancer cells to AZD8055 after MDK knockdown.After acting on the same concentration of AZD8055,Western blotting showed that the phosphorylation inhibition ability of AZD8055 on the m TORC1 in MDK knockdown cells was weakened,thereby reducing the sensitivity of cells to AZD8055.Conclusion: 1.At the transcriptional level,CIN subtype gastric cancer is heterogeneous and can be reclassified into two metabolic subtypes and one proliferative subtype with significantly different prognosis and biological characteristics.The evolutionary process of CIN subtype gastric cancer is mainly manifested from metabolic changes to rapid proliferation.2.A small number of CIN subtypes of gastric cancers have evolved in the direction of benefiting ICB treatment,and their TME is more inflammatory and activated,and the evolutionary trajectory is accompanied by a greater degree of TMB rise.Identification of ARID1 A and LRRK2 as characteristic mutations reflecting the evolution of evolution of gastric cancer of the CIN subtype towards benefit from ICB therapy.3.AZD8055 is a viable targeted drug for gastric cancer of the CIN subtype that does not respond to ICB,and CIN subtype gastric cancer is more sensitive to AZD8055 when the MDK,an important gene that promotes the occurrence and evolution of gastric carcinoma of the CIN subtype,is highly expressed. |
PDF Full Text Request