Bioinformatic analysis and experimental validation of hub autophagy-related genes as novel biomarkers for type 2 diabetes mellitus and Alzheimer's disease.

Background & Objectives: Alzheimer's disease (AD) and type 2 diabetes mellitus (T2DM) share considerable similarities in their proposed patho mechanisms. Autophagy, an intrinsic cellular process involved in the degradation of dysfunctional organelles and abnormal proteins, has been implicated in the pathogenesis of both AD and T2DM. This study aims to identify potential shared biomarkers related to autophagy in AD and T2DM by analyzing hub differentially expressed autophagy-related genes (DEARGs) and examining their potential functions. Methods: Gene expression profiles for AD and T2DM were acquired from the Gene Expression Omnibus (GEO) database (training sets: GSE109887 for AD and GSE104674 for T2DM; validation sets: GSE122063 for AD and GSE64998 for T2DM). Autophagy-related genes (ARGs) were extracted from multiple databases. DEARGs were identified and integrated with module genes derived from weighted gene co-expression network analysis (WGCNA) to determine key shared ARGs. Then, the STRING database was used to construct a protein-protein interaction (PPI) network, from which hub genes were identified. These hub genes were validated using independent microarray datasets through differential expression analysis, and ROC curves were generated to assess their diagnostic value. Moreover, the expression of the hub genes was validated in brain tissues of T2DM mouse models using qRT-PCR. Results: A total of 33 shared DEARGs were identified, among which 12 were designated as hub genes (ANXA5, CCND1, MAP2K1, HSPB1, BNIP3, BAG3, YAP1, MET, FBXW7, CCL2, PFKFB3, CDKN1A) in both AD and T2DM patients. Validation using other datasets confirmed that ANXA5, BAG3, and CDKN1A remained significantly upregulated, while MET remained downregulated in both AD and T2DM patients. Additionally, PFKFB3 showed an inverse expression pattern between the two diseases. The diagnostic performance of these five hub genes was assessed using ROC curves, with all five exhibiting values of area under the curve (AUC) exceeding 0.7 for T2DM in both training and validation sets. However, only MET and PFKFB3 demonstrated good diagnostic efficacy in AD patients. In animal models, qRT-PCR analysis revealed that the expression of ANXA5, BAG3, and MET was consistent with the bioinformatics results. In contrast, the expression of PFKFB3 and CDKN1A did not differ significantly between db/db model mice and db/m control mice. Conclusions: Our integrated bioinformatics analyses, supported by preliminary experimental validations, identified several hub ARGs shared between AD and T2DM. Among these, ANXA5, BAG3, and MET exhibited consistent expression trends across datasets and experimental models, while CDKN1A and PFKFB3 showed inconsistent expression patterns. These findings underscore the complexity of autophagy-related crosstalk in AD-T2DM comorbidity and highlight the need for further research to clarify their diagnostic and therapeutic potential.