Integrative multi-omics and machine learning identify CALR as a diagnostic and therapeutic target in aneurysmal subarachnoid hemorrhage

Background

Aneurysmal subarachnoid hemorrhage (aSAH), caused by the rupture of intracranial aneurysms, is a devastating cerebrovascular event with high mortality and disability. However, effective diagnostic biomarkers and therapeutic targets remain limited.

Methods

We combined 4D label-free quantitative proteomics with transcriptomic datasets (GSE122897, GSE36791, GSE73378) for integrative analysis. Weighted Gene Co-expression Network Analysis (WGCNA) identified key gene modules. Functional enrichment (GO, KEGG) and GeneMANIA interaction networks were constructed. A diagnostic model was built using 113 machine learning algorithm combinations and validated across multiple datasets. Immune infiltration was evaluated by CIBERSORT. Gene Set Enrichment Analysis (GSEA) explored underlying biological processes. A nomogram was developed using the “rms” package. Experimental aSAH models were established in vivo and in vitro to validate candidate gene expression and function via qPCR, Western blot, immunofluorescence, and immunohistochemistry. AAV-mediated gene modulation and primary cortical neuron cultures were used for mechanistic validation.

Results

CALR was identified as a key diagnostic biomarker through WGCNA and machine learning. The diagnostic model demonstrated high accuracy (AUC > 0.85). CALR was significantly associated with immune cell infiltration and endoplasmic reticulum stress. In vivo, AAV-mediated CALR overexpression attenuated neuronal damage in SAH mice. In vitro, CALR exerted neuroprotective effects on primary neurons under SAH conditions.

Conclusion

CALR serves as a promising diagnostic and therapeutic target in aneurysmal subarachnoid hemorrhage. This study highlights the role of multi-omics and machine learning in uncovering novel mechanisms and targets in cerebrovascular diseases.