A blood transcriptomic signature anchored to central nervous system pathology enables noninvasive detection of multiple sclerosis.

BACKGROUND

Multiple sclerosis (MS) lacks noninvasive biomarkers anchored to central nervous system (CNS) pathology. This study aimed to identify a blood transcriptional signature anchored to CNS lesion biology for noninvasive MS detection.

METHODS

We analyzed bulk RNA-seq data from MS white matter lesions across three independent cohorts and integrated paired CNS and blood transcriptomic data to identify genes with concordant directional dysregulation. A consensus multi-algorithm machine-learning framework was applied to derive a candidate gene signature. Validation included an independent external cohort, single-nucleus RNA-seq analysis, and an experimental autoimmune encephalomyelitis (EAE) mouse model. Immune cell composition in blood was estimated using deconvolution algorithms.

RESULTS

We identified 277 lesion-associated differentially expressed genes, of which 66 exhibited concordant directional dysregulation between the CNS and peripheral blood. A 14-gene transcriptional signature effectively discriminated patients with MS from controls (training area under the curve (AUC) = 0.954; external validation AUC = 0.917). Models trained on genes with discordant expression patterns failed to generalize (validation AUC = 0.556), highlighting the importance of cross-tissue directional consistency for robust biomarker identification. Immune deconvolution revealed expansion of plasmacytoid dendritic cells and depletion of naive B cells in MS blood. Single-nucleus analysis confirmed the characteristic downregulation of these 14 biomarkers in lesion-associated microglia and T cells, mirroring their peripheral expression profiles. EAE validation demonstrated cross-tissue concordant dysregulation of representative biomarkers (Sort1, Pi4kb, and Csde1) in both CNS and peripheral blood.

CONCLUSIONS

This study identifies a candidate blood-based transcriptional signature anchored to CNS pathology with potential for noninvasive MS detection. The cross-tissue concordance framework offers a generalizable strategy for biomarker discovery in neurological diseases. Prospective validation in independent cohorts is warranted to confirm its clinical utility.