Estrogen-related gene INPP5D is a potential biomarker for Alzheimer’s disease

Abstract

Objective: This study aims to leverage bioinformatics methodologies to screen genes related to Alzheimer’s disease (AD) and estrogen, unravel the interplay between estrogen and Alzheimer’s pathophysiology, and furnish a foundation for the clinical treatment of AD.

Methods: Through accessing AD transcriptomics data from the GEO database, we examined differentially expressed genes (DEGs) and elucidated their functional roles via GO and KEGG analyses. Also, we constructed a protein interaction network. Furthermore, we took the intersection between genes identified by the weighted gene co-expression network and estrogen-related DEGs, then obtained key genes through machine learning and verified them through internal data sets. Finally, immune infiltration analysis was performed on the data of patients with AD.

Results: 326 DEGs were obtained from GSE118553, predominantly enriched in nucleotide diphosphorylation (biological process), nucleotide excision repair complexes (cellular composition), neurotrophic factor binding (molecular functions), and sugar metabolism (signaling pathway). Through weighted gene co-expression network analysis and the intersection of DEGs, three estrogen-related genes were screened out: INPP5D, ENO1, and NOP16. Next, INPP5D was selected as the pivotal gene by the random forest tree algorithm. The prediction model was built on INPP5D, and the AUC of the ROC curve equals 0.876. After single-sample gene enrichment analysis, the samples clustered into high- and low-immune groups. Intriguingly, there was a positive correlation between these two groups on activated NK cells and M1 macrophages, and a negative correlation between M1 macrophages and T cells CD4 memory resting.

Conclusion: The estrogen-related gene INPP5D is a potential biomarker for AD, which provides a new perspective and scientific basis for the development of AD treatment strategies based on hormone replacement therapy.