Integrative transcriptomics identifies shared aging-associated inhibitory-neuron states and prioritizes RGL2 across postoperative delirium and Alzheimer's disease.

Postoperative delirium (POD) and Alzheimer's disease (AD) are increasingly recognized as related neurocognitive conditions, but the aging-associated cell states that may connect them remain poorly defined. Here, we integrated two brain transcriptomic discovery datasets analyzed at single-cell/single-nucleus resolution, including a POD-related cohort (GSE291019) and an AD cohort (GSE129308), together with two independent peripheral-blood bulk transcriptomic datasets (GSE163943 and GSE63060), to identify aging-associated cellular programs and prioritize convergent molecular candidates. Across 184,168 high-quality cells, inhibitory neurons showed the most consistent aging-associated perturbation across the POD- and AD-related datasets. Re-clustering further identified three inhibitory-neuron subtypes, Inh_Neurons2, Inh_Neurons3, and Inh_Neurons5, with relatively high aging-related gene activity and preferential localization to later pseudotime states. Cross-platform integration of aging-associated inhibitory-neuron genes with a shared bulk DEG set identified four convergent candidates: RGL2, AKT1, SYK, and TNFSF13B. Among them, RGL2 emerged as the leading candidate, with the strongest downstream support concentrated in AD-related analyses. In two-sample Mendelian randomization, genetically predicted higher RGL2 expression was associated with increased AD risk, whereas the estimate for the delirium genome-wide association study proxy used for POD-related analyses was not significant. Pathway analyses further linked higher RGL2 expression to complement/coagulation and innate immune-inflammatory programs in AD-related analyses. These findings suggest a model in which POD and AD may partially intersect through aging-vulnerable inhibitory-neuron states and identify RGL2 as a prioritized candidate for downstream mechanistic investigation.