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

Summary

This multi-omics case-control study identifies PI3K–Akt pathway-related gene signatures (e.g., IL6, IL6R, CHRM2, NOS3, NGF up; IGF1 down) in peripheral blood as predictive/diagnostic biomarkers for postoperative delirium, validated by RT-qPCR and scRNA-seq across immune cell subsets. Machine learning models achieved discriminatory performance, supporting translational potential.

Key Findings

• Bulk RNA-seq identified upregulation of CHRM2, IL6, NOS3, NGF, IL6R and downregulation of IGF1 in POD versus controls.
• RT-qPCR in an independent cohort (n=60) and scRNA-seq across T cells, B cells, NK cells, dendritic cells, and monocytes validated these signatures.
• Machine learning and ROC analyses showed these genes have predictive/diagnostic value for POD.

Clinical Implications

If prospectively validated, preoperative or early postoperative blood testing for these immune gene signatures could stratify delirium risk, guide prophylaxis, and tailor monitoring or anti-inflammatory strategies.

Limitations

• Case-control design with modest sample size and potential confounding
• Lack of external prospective validation and calibration in diverse perioperative populations

Future Directions

Prospective, multicenter validation with preoperative sampling, integration with clinical risk scores, and interventional trials testing biomarker-guided delirium prevention.