Skip to Content

Vagus Nerve Stimulation: Developing a New Predictive Model for Pediatric Epilepsy Treatment Response

June 16, 2023

Predicting which pediatric patients with drug-resistant epilepsy (DRE) will respond to vagus nerve stimulation (VNS) presents a host of challenges for clinicians.

However, newly published research from a UPMC Children’s Hospital of Pittsburgh research team could lay the foundation for a new approach using a predictive clinical model they designed and tested.

The work was published as the Editor’s Choice selection in the May 2023 edition of the Journal of Neurosurgery: Pediatrics.

The study was led by Nalli Muthiah, MD, MS, under the direction of Taylor J. Abel, MD. William Welch, MD, of Child Neurology was a collaborator. Dr. Abel is the Chief of Pediatric Epilepsy Surgery at UPMC Children’s, and Dr. Welch is the director of Epilepsy Neurostimulation at UPMC Children’s. The two are frequent research collaborators across multiple lines of investigation involving epilepsy treatment approaches and neurostimulation. Dr. Muthiah is a recent graduate of the University of Pittsburgh School of Medicine student and is the study’s first author.

About the Study

Close to a half million children in the United States suffer from some form of epilepsy, with a significant proportion facing the rigors of DRE. For some individuals with DRE, VNS can provide an alternative treatment pathway, but deducing who those individuals may be presents numerous challenges.

The clinical model developed by Drs. Muthiah and Abel, which aims to predict patient response to VNS, was based on an 11-year retrospective cohort analysis involving children with DRE who had undergone VNS.

The model takes into account multiple patient variables and characteristics, including the duration of epilepsy prior to VNS, age at seizure onset, number of pre-VNS antiseizure medications, whether VNS was the patient's first therapeutic epilepsy surgery, and the type of seizures the patient primarily experienced. The model was trained on 70% and validated on the remaining 30% of the patient sample.

Important Findings

The key findings indicated that a shorter duration of epilepsy (under 4 years) prior to VNS and a diagnosis of focal motor seizures were linked to a more favorable response to VNS therapy. However, despite the large sample size, the team found that the clinical prediction score alone wasn't as accurate in predicting VNS response as desired. The area under the curve (AUC) – a measure of the model's ability to predict outcomes – scored 0.7013 for the "fitted" sample and 0.6159 for the "validation" sample. In other words, the model was able to discriminate which patients would respond to VNS with 70% accuracy for the patients on which it was trained, and 62% accuracy for patients on which it was not trained.

While the clinical prediction score was able to predict VNS response slightly better than chance, it fell short of full accuracy, suggesting the limitations of relying on clinical variables alone.

The team notes that this finding underscores the pressing need for developing more sensitive biomarker-based tools to aid in more accurate predictions. If successful, such tools could significantly augment the predictive power of the current model, making it a valuable tool in the outpatient evaluation of seizure control options for pediatric patients with DRE.

The findings also suggest that VNS could be beneficial for a wider range of patients than previously believed. The study showed that patients with focal seizures or motor seizures were more likely to respond favorably to VNS. This broadens the potential scope of VNS treatment beyond patients with focal seizures.

The results of this study provide a promising stepping-stone in the search for more tailored, effective treatments for pediatric epilepsy. By continuing to refine and improve this clinical model, it could eventually better predict which children with drug-resistant epilepsy will respond positively to VNS, thereby improving individualized care and patient outcomes while potentially avoiding a costly and invasive procedure in some individuals who may not realize meaningful clinical benefit or seizure reductions.


Muthiah N, Mallela AN, Vodovotz L, Sharma N, Akwayena E, Pan E, Welch W, Ibrahim GM, Abel TJ. Development of a Clinical Model to Predict Vagus Nerve Stimulation Response in Pediatric Patients with Drug-Resistant Epilepsy. J Neurosurg Pediatr. 2023; 31(5): 463-468.

Learn More About:

Pediatric Epilepsy Center at UPMC Children’s Hospital of Pittsburgh

Division of Pediatric Neurosurgery

Division of Child Neurology

Dr. Abel

Dr. Welch