Rationale:
Deep brain stimulation (DBS) of the anterior nucleus of the thalamus (ANT) can reduce seizures in drug-resistant epilepsy, but seizure freedom remains rare and no objective early-response biomarker exists. Clinicians rely on patient-reported seizure rates for long term evaluation of efficacy of therapy, which is subjective and makes iterating through DBS parameter space for optimal seizure control impossible. A short-term marker of therapeutic response could expedite DBS parameter optimization and improve DBS outcomes.
Methods:
We investigated short-term ANT stimulation (minutes) in nine drug-resistant epilepsy patients undergoing intracranial EEG monitoring, comparing standard high-frequency periodic versus a novel non-periodic pulse pattern. Effective connectivity modifications were assessed by testing for significant deviation in response to cortico-cortical evoked potentials (CCEPs) immediately prior to and following 18 minutes of ANT stimulation. Responses were decomposed into phase and amplitude components to separate out effects on the response 1) timing and 2) potentiation to explore potential mechanisms. Logistic mixed effects models were created to explore significant spatial and temporal components of the response modifications. Finally, feature reduction in the form of L1 regularized logistic regression detected significant factors covarying with the response modifications. Removal of category specific factors within a logistic model pertaining to stimulation, anatomy, patient, and timing related their relative contributions to whether the response modification occurred.
Results:
Stimulation intensities of 3mA altered the first 150ms of CCEPs in thalamo-cortical circuits, with the anterior cingulate and superior frontal cortex showing the most consistent changes. Notably, non-periodic stimulation selectively modified early CCEP components relative to periodic stimulation with these changes being more related to changes in response phase. These findings highlight a pattern-specific modulation of connectivity between the ANT and spatially restricted cortical areas. When modeling what factors are most associated with CCEP changes, inclusion of significant factors determined by feature reduction yielded a pseudo-R2 value of 0.67. Removal of parameters related to the type of stimulation and location of stimulation were more predictive than patient-specific features (reduction of pseudo-R2 values from the full model of 0.15, 0.10, and 0.08 respectively).
Conclusions:
CCEP changes are quickly elicited, making them a potential treatment response biomarker for ANT-DBS epilepsy patients while in clinic. Prospective validation is required to determine if this biomarker can guide clinicians towards ideal parameters for ANT-DBS mediated seizure suppression.
Funding:
Neurosurgery Research and Education Foundation