| Background and Aims:The Switch/Sucrose non-fermentable(SWI/SNF)chromatin remodeling complex is a macromolecular complex composed of multiple subunits that can use the energy from ATP hydrolysis to break the contact between DNA and histones,disassemble nucleosomes and regulate gene expression.The SWI/SNF complex is essential for cell proliferation and differentiation,and the aberrant expression of subunits is closely associated with tumorigenesis.SMARCA2 and SMARCA4 are the catalytic subunits of the SWI/SNF complex.They show ATPase activity and can catalyze adenosine triphosphate(ATP)hydrolysis to produce energy,which is crucial for the SWI/SNF complex.ARID1A can localize to target genes using energy derived from the hydrolysis of ATPase subunits,which further facilitates the SWI/SNF complex binding to DNA,and the abnormal of ARID1A significantly reduces the targeting activity of SWI/SNF complex.The TCGA molecular profiling classifies gastric carcinoma into four subtypes at the molecular level,including the Epstein-Barr virus(EBV)positive subtype,microsatellite instability(MSI)subtype,genomically stable(GS)subtype,and chromosomal instability(CIN)subtype,which has greatly expanded our knowledge of the heterogeneity and molecular complexity of gastric carcinoma.Currently,abnormalities of the SWI/SNF complex subunits have been found in many types of human tumors.Still,there are few studies on gastric adenocarcinoma,especially the lack of clinical studies with large samples.The prognostic significance of the SMARCA2 subunit in gastric adenocarcinoma and the prognostic and immune infiltration significance of the ARID1A subunit in the TCGA molecular subtype of gastric adenocarcinoma have not been thoroughly studied.This study evaluated the immunohistochemical expression of five subunits of the SWI/SNF complex in 1245 patients with gastric adenocarcinoma,including SMARCA2,SMARCA4,ARID1A,ARID1B,and SMARCB1,explored the impact of SMARCA2 as an independent prognostic factor on survival,and established a new nomogram prediction model.Meanwhile,we explored the correlation between ARID1A and clinicopathological features,prognosis,and immune infiltration combined with TCGA molecular classification.Material and Methods:We screened 1245 postoperative patients with gastric adenocarcinoma,obtained pathological tissue blocks,and constructed tissue microarrays(TMA)to perform immunohistochemical staining of the above five subunits of SWI/SNF complex,as well as other important markers,including Her-2,p53,Ki-67,PD-L1,four MMR proteins(MLH1,MSH2,MSH6,and PMS2),and E-cadherin.The expression of EBV was also detected by EBER-ISH.We obtained the immunohistochemical expression of five subunits of SWI/SNF complex,which can be classified into four expression patterns:intact,reduced,lost,and heterogeneous.We first identified the SMARCA2 subunit.We performed the Chi-square test,then performed multivariate Logistic regression and Lasso regression on filtered variables,respectively.Finally,the filtered variables were intersected and ranked by importance using a random forest model.We plotted the K-M survival curves according to the immunohistochemical expression of the five subunits and compared the curves using the log-rank method.To gain a comprehensive understanding of the factors affecting the prognosis,we first performed the initial screening by univariate Cox regression analysis,followed by multivariate Cox regression analysis and Lasso regression analysis,respectively.We performed the intersection of the screened variables,and finally,the nomogram was drawn to construct a prediction model of prognosis.We comprehensively evaluated the nomogram model and compared with the traditional AJCC TNM model,including the calibration curve,C-index,ROC curve,area under the curve(AUC),decision curve analysis(DCA),net reclassification improvement(NRI),and integrated discrimination improvement(IDI).We then analyzed ARID1A subunits.We classified TCGA molecules according to the staining results of EBER,MMR,and E-cadherin.We re-classified ARID1A expression into two categories according to whether there was standard staining:ARID1A with standard staining in intact and heterogeneous expression patterns was defined as ARID1A positive;ARID1A without standard staining in the lost and reduced expression patterns was defined as ARID1A negative.To filter out variables associated with ARID1A negative expression,we first performed a Chi-square test,then performed multivariate Logistic regression and Lasso regression,intersected the filtered variables,and ranked the variables according to the importance by random forest model.To explore the association of ARID1A with survival,we first performed Kaplan-Meier survival analysis according to ARID1A expression in the entire study population and subsequently according to ARID 1A expression in each TCGA molecular subtype.To explore the association of ARID1A with immune infiltration,particularly in each TCGA molecular subtype,we randomly screened 1/2 of ARID1A negative cases.Then we matched ARID1A positive cases in a 1:1 ratio using the propensity score matching(PSM)method.We extracted pathological tissue from the tumor center(TC)and infiltration margin(IM)of formalin-fixed and paraffin-embedded(FFPE)tissue blocks to construct TMAs.We then performed multiplex immunofluorescence(MIF)analysis of CD4+T cells,CD8+T cells,and PD-L1 expression.Results:Lost expression of the five subunits was SMARCA2 lost in 233 cases(18.71%),ARID1A lost in 158 cases(12.69%),SMARCA4 lost in 84 cases(6.75%),SMARCB1 lost in 9 cases(0.72%),and ARID1B lost in 6 cases(0.48%).Of the 9 cases with SMARCB1 lost,all with concomitant loss of other subunits,5 of the 6 cases with ARID1B lost showed concomitant loss of other subunits.After the Chi-square test,multivariate Logistic regression analysis,and Lasso regression analysis,four variables associated with the lost expression of SMARCA2 were screened,ranked by significance as follows,EBER,differentiation status,and T stage.Survival analysis revealed that lost expression of both SMARCA2 and SMARCA4 caused worse survival than the present expression pattern(P=0.028 and 0.0099,respectively),whereas lost expression of ARID1A,ARID1B and SMARCB1 was not associated with survival.When stratified by TNM stage,lost expression of SMARCA2 caused poor survival in early-stage(stage Ⅰ and Ⅱ)gastric carcinoma(P=0.0034),whereas lost expression of SMARCA4 caused poor survival in advanced(stage Ⅲ and Ⅳ)gastric cancer(P=0.014).After univariate Cox regression analysis,multivariate Cox regression analysis,and Lasso regression analysis,seven independent prognostic factors were screened,ranked by variable importance as follows,N stage,M stage,T stage,E-cadherin,chemotherapy,age,and SMARCA2.Finally,based on the seven prognostic variables,we drew the nomogram and established a new prediction model of prognosis.We performed comprehensive evaluations of the nomogram prediction model,and the calibration curves showed good agreement between the predicted and actual results.The C-index was 0.797(0.786-0.808)for the nomogram model and 0.763(0.750-0.776)for the traditional AJCC TNM model.The ROC curves and time-dependent C-index curves of the two models at 1,3,and 5 years showed that the discriminatory power of the nomogram model was superior to that of the AJCC TNM model.DCA showed that the clinical benefit of the nomogram model was superior to the traditional AJCC TNM model.Compared with the conventional AJCC TNM model,the nomogram model improved the NRI by 14.5%,8.2%,and 11.5%,respectively,and the IDI by 4.3%,4.7%,and 2.6%,respectively,at 1,3,and 5 years.The distribution of TCGA molecular subtypes was as follows,70 cases(5.62%)were EBV subtype,181 cases(14.54%)were MSI subtype,331 cases(26.59%)were GS subtype,and 663 cases(53.25%)were CIN subtype.Among the four subtypes,ARID1A negative expression accounted for the highest proportion of 51.38%in the MSI subtype,followed by the EBV subtype,with ARID1A negative expression accounting for 37.14%.After the Chi-square test,multivariate Logistic regression analysis,and Lasso regression analysis,eight variables associated with RID1A negative expression were selected and ranked by significance as follows,MMR,PD-L1,T stage,p53,differentiation,E-cadherin,EBER,and M stage.Survival analysis identified that the ARID1A negative group showed worse survival than the ARID1A positive group for the entire study population(P=0.00014).After stratification by TCGA molecular subtypes,ARID1A negative expression led to worse survival in the GS subtype(P<0.0001),whereas ARID1A negative expression was not associated with survival in the other three subtypes.For the overall data,the infiltration of CD4+T cells,CD8+T cells,and PD-L1 expression was significantly higher in the ARID1A negative group than in the ARID1A positive group,both in the tumor center(TC)and in the infiltration margin(IM).After stratification by TCGA molecular subtype,the infiltration of CD4+T cells was not correlated with ARID1A in the EBV subtype of the IM,while in the remaining subtypes,the infiltration of CD4+T cells was higher in the ARID1A negative group than in the ARID1A positive group.The infiltration of CD8+T cells was higher in the ARID1A negative group than in the ARID1A positive group in the MSI subtype of the IM,while there was no correlation between the infiltration of CD8+T cells and ARID1A expression.The expression of PD-L1 was higher in the ARID1A negative group than in the ARID1A positive group in any TCGA subtype.We analyzed the relationship among PD-L1,CD4,and CD8 according to the data from mIF.We found a positive correlation between these three markers when ARID1A was negative,while PD-L1 was not associated with CD4/CD8 when ARID1A was positive.Conclusions:The loss of SMARCA2 was closely related to advanced T stage,worse differentiation,and EBV infection,and was an independent adverse prognostic factor of gastric adenocarcinoma.According to T stage,N stage,M stage,E-cadherin,SMARCA2,age,and chemotherapy,we constructed a new nomogram prediction model,which performed better than the traditional AJCC TNM model.The ARID1A negative expression was significantly higher in EBV and MSI subtypes than in GS and CIN subtypes.The CD8 expression caused by ARID1A negativity was more due to the indirect effect,while the PD-L1 and CD4 expression caused by ARID1A negativity was more due to the direct effect.The increase of CD8 caused by ARID1A negativity was accompanied by the increase of PD-L1 expression,which may induce adaptive immune resistance that can be resisted by inhibiting PD-1/PD-L1.Background and Aims:Undifferentiated gastric carcinoma is a histopathologically rare,highly aggressive malignancy composed of cells showing no specific cytologic or architectural differentiation,including no gland formation or mucin production and no neuroendocrine,squamous,or sarcomatoid differentiation.It is called dedifferentiated carcinoma if undifferentiated carcinoma contains a minor differentiated component.The SWI/SNF chromatin remodeling complex is an evolutionarily conserved complex that exhibits dysfunction in many tumors,especially undifferentiated carcinomas.Cancer stem cells(CSCs)are a particular class of undifferentiated cancer cells with stem cell-like properties,which play important roles in tumor cell proliferation,invasion,and metastasis.The association of the SWI/SNF complex with clinicopathological features,CSC phenotypes,and prognosis in undifferentiated gastric carcinoma is not fully understood.Material and Methods:We collected gastric adenocarcinoma patients who underwent surgical resection at Weihai Municipal Hospital between January 2014 and December 2020.We first performed morphological screening by H&E section and then screened by immunohistochemical indicators(CK-pan,LCA,CD56,CgA,Syn,NSE,and Vimentin),and finally identified 21 cases of undifferentiated/dedifferentiated gastric carcinoma.We next performed immunohistochemistry staining for the five subunits of the SWI/SNF complex(ARID1A,ARID1B,SMARCA2,SMARCA4,and SMARCB1),and four mismatch repair proteins(MLH1,PMS2,MSH2,and MSH6),as well as other markers such as p53,PD-L1,and cancer stem cell(CSC)markers(SOX2,SALL4).Loss of any of the five SWI/SNF complex subunits was defined as deficient,and the existence of all five subunits was defined as intact.The correlation of the SWI/SNF complex status with clinicopathological features was analyzed by Fisher’s exact test(sample size n<40);The impact of SWI/SNF complex,differentiation status,stem cell phenotype,SMARCA2,SMARCA4,and ARID1A on survival was analyzed by Kaplan-Meier survival analysis;Independent prognostic variables were selected by univariate and multivariate Cox regression analysis.Results:The lost expression of the SWI/SNF complex subunits was as follows:12 cases(57.14%)showed SMARCA2 lost,5 cases(23.81%)showed ARID1A lost,3 cases(14.29%)showed SMARC4 lost,and no cases showed ARID1B or SMARCB1 lost.We observed SMARCA2 loss in 12 cases(57.14%),ARID1A loss in 5 cases(23.81%),and SMARCA4 loss in 3 cases(14.29%);however,we did not find any case showing SMARCB1 or ARID1B loss.Taken together,14 cases(66.67%)showed any one of the SWI/SNF complex subunits loss,including 3 cases with ARID1A and SMARCA2 co-loss and 3 cases with SMARCA4 and SMARCA2 co-loss.The correlation analysis showed that the CSC phenotype was significantly associated with SWI/SNF complex deficiency(P=0.0158).Survival analysis revealed the SWI/SNF complex,deficiency,undifferentiated state,CSC phenotype,SMARCA2 lost expression,and SMARCA4 lost expression led to worse survival.Univariate and multivariate Cox regression analyses identified three independent variables associated with worse prognosis:undifferentiated status,the SWI/SNF complex deficiency,and lymph node metastasis.Conclusions:The SWI/SNF complex deficiency was more likely to result in a CSC phenotype and worse survival and was an independent prognostic factor in undifferentiated/dedifferentiated gastric carcinoma. |
PDF Full Text Request