Cognitive effects of deep brain stimulation across targets in epilepsy: a systematic review and meta-analysis

Drug-resistant epilepsy (DRE) affects one-third of individuals with epilepsy and increases cognitive decline. Resective surgery carries focal cognitive risks and contraindications. Deep-brain stimulation (DBS) represents a promising alternative, although its cognitive effects remain understudied.

We searched PubMed, MEDLINE, EMBASE, Web of Science, and Cochrane for pre- and post-DBS neuropsychological scores in DRE with ≥ 3 months follow-up. Fifteen studies (16 cohorts, n = 270) were reviewed; 12 (13 cohorts, n = 209) were meta-analyzed. Targets were the anterior nucleus of the thalamus (ANT, k = 7), amygdalohippocampal region (AH, k = 6), centromedian nucleus of the thalamus (k = 2), and subthalamic nucleus of the thalamus (k = 1). Robust variance estimation computed Hedges' g by domain. Bias assessed with ROBINS-I and RoB 2; certainty with GRADE.

No cognitive domain declined significantly. Executive functioning was the only domain approaching improvement (g = 0.12, 95% CI -0.08 to 0.33), most apparent with ANT (g = 0.15, CI -0.16 to 0.46). Anxiety declined significantly (g = -0.36, 95% CI -0.68 to -0.04). Memory, language, processing speed, visuospatial functions, and global intellect remained stable. Cognitive adverse events occurred in 15.8% of patients, highest with ANT (20.1%) and lowest with AH (9.24%).

Chronic DBS for DRE is cognitively well tolerated, with no decline across domains and emerging evidence of executive functioning and mood benefit, especially with ANT stimulation. Verbal memory and naming were preserved after AH stimulation, unlike dominant temporal resection. Plausible mechanisms exist for both cognitive improvement and decline, yet neither appears consistently on assessments. Whether this reflects true stability or insensitivity of current methods is unclear. A revised framework for assessment timing, domains, and metrics is recommended.