Unveiling the Immune Landscape of Delirium through Single-Cell RNA Sequencing and Machine Learning: Towards Precision Diagnosis and Therapy.

Abstract

Background: Postoperative delirium (POD) poses significant clinical challenges regarding its diagnosis and treatment. Identifying biomarkers that can predict and diagnose POD is crucial for improving patient outcomes.

Methods: To explore potential biomarkers for POD, we conducted bulk RNA sequencing (bulk-seq) on peripheral blood samples from POD patients and healthy controls. The expression levels of genes downstream of the phosphatidylinositol 3-kinase/protein kinase B (PI3K-Akt) signalling pathway were analysed. We then validated the expression of these genes using quantitative real-time polymerase chain reaction (RT-qPCR) in an independent cohort of 30 healthy controls and 30 POD patients. Receiver operating characteristic (ROC) analysis and six machine learning models were used to evaluate the predictive and diagnostic value of these genes. Additionally, single-cell RNA sequencing (scRNA-seq) was performed to validate gene expression in specific subsets of peripheral blood mononuclear cells (PBMCs), including T-cells, B-cells, natural killer (NK) cells, dendritic cells (DCs), and monocytes.

Results: Bulk-seq revealed increased expression of genes downstream of the PI3K-Akt signalling pathway, specifically CHRM2, IL6, NOS3, NGF, and IL6R, in the peripheral blood of POD patients compared to healthy controls. Conversely, the expression of IGF1 was significantly decreased. RT-qPCR validation confirmed these findings. ROC analysis and machine learning models indicated that these genes are useful for predicting and diagnosing POD. scRNA-seq further validated the expression of these genes in specific PBMC subsets, including T-cells, B-cells, NK cells, DCs, and monocytes, with results consistent with the bulk-seq and RT-qPCR data.

Conclusions: The abnormal activation of the PI3K-Akt signalling pathway in T-cells, B-cells, NK cells, DCs, and monocytes may serve as potential biomarkers for predicting and diagnosing POD. These findings could inform the development of novel therapeutic strategies for managing POD.