阿尔茨海默病失巢凋亡相关基因挖掘与验证OA

Aim To identify the expression character-istics of anoikis-related genes in Alzheimer's disease(AD)using machine learning algorithms and evaluate their diagnostic value.Methods The GEO dataset GSE44770 was used as the training set to screen for anoikis-related differentially expressed genes(DEGs).Consensus clustering was applied to classify AD patients into C1 and C2 subtypes.Weighted gene co-expression network analysis(WGCNA)was em-ployed to identify AD-related feature genes.Diagnos-tic models were constructed using Random Forest,Support Vector Machine,eXtreme Gradient Boosting(XGBoost),and Generalized Linear Model.The opti-mal model was selected,and key genes most corre-lated with AD clinical traits were extracted.Correla-tion analysis between these key genes and anoikis-related DEGs was conducted to identify crucial anoikis-related DEGs.Experimental validation was performed using APP/PSI transgenic AD model mice through HE staining,Western blot,and RT-qPCR.Results A total of 20 anoikis-related DEGs were identified.AD patients were classified into C1 and C2 subtypes,with the C2 subtype exhibiting enhanced glycolysis,acti-vated inflammation,and suppressed mitochondrial function.WGCNA identified 243 AD-related genes.The XGBoost model demonstrated the best perfor-mance,and five key genes(including OLFM1 and ITPKB)significantly associated with clinical traits were identified.Systematic correlation analysis be-tween these key genes and anoikis-related DEGs ulti-mately screened UCHL1,TNFRSF10B,and ITGB5 as crucial anoikis-related DEGs.Animal experiments re-vealed significant hippocampal neuronal damage,ab-normal p-Tau protein accumulation,and consistent ex-pression trends of the key anoikis-related DEGs with bioinformatic predictions.Conclusion Anoikis plays an important role in the pathogenesis of AD,and its key genes may serve as potential diagnostic biomarkers and therapeutic targets.