Immunological characterization and diagnostic models of RNA N6-methyladenosine regulators in Alzheimer's disease

Abstract

AbstractAlzheimer's disease (AD) is the most prevalent form of dementia, and it displays both clinical and molecular variability. RNA N6-methyladenosine (m6A) regulators are involved in a wide range of essential cellular processes. In this study, we aimed to identify molecular signatures associated with m6A in Alzheimer's disease and use those signatures to develop a predictive model. We examined the expression patterns of m6A regulators and immune features in Alzheimer’s disease using the GSE33000 dataset. We examined the immune cell infiltration and molecular groups based on m6A-related genes in 310 Alzheimer's disease samples. The WGCNA algorithm was utilized to determine differently expressed genes within each cluster. After evaluating the strengths and weaknesses of the random forest model, the support vector machine model, the generalized linear model, and eXtreme Gradient Boosting, the best machine model was selected. Methods such as nomograms, calibration curves, judgment curve analysis, and the use of independent data sets were used to verify the accuracy of the predictions made. Alzheimer's disease and non-disease Alzheimer's groups were compared to identify dysregulated m6A-related genes and activated immune responses. In Alzheimer's disease, two molecular clusters linked to m6A were identified. Immune infiltration analysis indicated substantial variation in protection between groups. Cluster 1 included processes like the Toll-like receptor signaling cascade, positive regulation of chromatin binding, and numerous malignancies; cluster 2 included processes like the cell cycle, mRNA transport, and ubiquitin-mediated proteolysis. With a lower residual and root mean square error and a larger area under the curve (AUC = 0.951), the Random forest machine model showed the greatest discriminative performance. The resulting random forest model was based on five genes, and it performed well (AUC = 0.894) on external validation datasets. Accuracy in predicting Alzheimer's disease subgroups was also shown by analyses of nomograms, calibration curves, and decision curves. In this research, we methodically outlined the tangled web of connections between m6A and AD and created a promising prediction model for gauging the correlation between m6A subtype risk and AD pathology.