Programming of deep brain stimulation of the centromedian nucleus of the thalamus for drug‐resistant epilepsy: A meta‐analysis and proposed programming framework

Abstract

Abstract Objective Although the centromedian nucleus of the thalamus (CM) is an increasingly considered deep brain stimulation (DBS) target for drug‐resistant epilepsy (DRE), there is significant variability in programming practices, which may contribute to heterogenous outcomes. We systematically described stimulation parameters for CM‐DBS and their association with seizure outcomes to inform a standardized, evidence‐informed programming framework. Methods We performed a Preferred Reporting Items for Systematic reviews and Meta‐Analysis (PRISMA)–guided systematic review and meta‐analysis of published studies pertaining to CM‐DBS including all dates through March 2026. Eligible studies included patients with DRE treated with CM‐DBS, extractable seizure outcomes, and reported stimulation parameters. Study‐level data and extractable patient‐level data were synthesized. The primary outcome was percent seizure reduction at the last follow‐up. Using random‐effects meta‐analysis we summarized study‐level efficacy and performed regression analyses to assess associations with diagnosis, seizure semiology, and stimulation parameters. Results Thirty studies including 269 individuals met inclusion criteria. Median seizure reduction was 71.5% and 200 participants (74.6%) with seizure outcome data were responders (>50% seizure reduction), which likely represents an upper bound on true efficacy given the predominance of non‐randomized studies at serious‐to‐critical risk of bias. Generalized tonic–clonic seizures were associated with greater seizure reduction. Tonic seizures were associated with worse outcome, yet this was attenuated after accounting for study clustering. Stimulation parameters varied substantially across studies. No single nominal stimulation parameter demonstrated a robust clinically actionable association with outcome; however, estimated total electrical energy delivered was associated with greater seizure reduction on multivariable modeling, yet this was attenuated when accounting for clustering between studies. Significance These findings map the landscape of programming of CM‐DBS, while emphasizing the need for standardized reporting of stimulation parameters, contact localization, and stimulation exposure to enable prospective optimization. On this basis, we propose a pragmatic, evidence‐informed programming framework for CM‐DBS that emphasizes postoperative contact localization, standardized initial stimulation settings, and stepwise titration of stimulation exposure.