Connectomic Profiling Identifies Responders to Vagus Nerve Stimulation

Area of Science:

• Neuroscience
• Medical Imaging
• Machine Learning

Background:

• Vagus nerve stimulation (VNS) is a treatment for intractable epilepsy.
• Current methods lack preoperative prediction of VNS response.
• Identifying responders is crucial for patient outcomes and resource allocation.

Purpose of the Study:

• To develop a predictive algorithm for VNS response using connectomic profiling.
• To compare connectomic prediction with clinical covariates.

Main Methods:

• Utilized diffusion tensor imaging and resting-state magnetoencephalography in 56 children.
• Developed a support vector machine classifier based on white matter microstructure and functional connectivity.
• Validated the classifier on independent cohorts and compared it to clinical predictors.

Main Results:

• Responders showed increased fractional anisotropy in specific white matter tracts and enhanced functional connectivity in key brain regions.
• The connectomic classifier achieved 89.5% accuracy in cross-validation and 83.3% in external validation.
• Connectomic prediction significantly outperformed prediction using clinical covariates alone (AUC 0.93 vs. 0.57).

Conclusions:

• This study presents the first multi-institutional, multimodal connectomic algorithm for predicting VNS response.
• The findings offer new insights into the mechanism of action of VNS.
• Accurate prediction can optimize VNS therapy, minimize surgical risks, and improve healthcare efficiency.