Electrophysiological Dose-Response to Anterior Thalamic Deep Brain Stimulation in Drug-Resistant Focal Epilepsy (ANT-DOSE)

Adults with drug-resistant focal epilepsy undergoing clinically indicated bilateral anterior thalamic DBS implantation using a sensing-enabled Medtronic Percept PC system. Research procedures include a protocol-defined stimulation amplitude titration schedule during the first 3 months, with adjustments permitted for clinical safety, and repeated acute stimulation challenges with concurrent local field potential (LFP) recording at activation and 1 and 3 months after activation.

ANT-DBS implantation using the sensing-enabled Medtronic Percept PC is standard care, independent of study participation. Study-specific procedures include scheduled chronic stimulation (145 Hz, 90 μs, cycling) at 2.0 mA from activation and 3.5 mA from month 1, with clinical adjustments permitted. At month 3, the physician selects 3.0-4.0 mA, followed by routine management. Acute challenges at activation, month 1, and month 3 test 2.0 mA; 2.0/3.5 mA; and 2.0/3.5/up to 5.0 mA, respectively. Each amplitude is repeated at least twice, with counterbalanced sequences at months 1 and 3. Each block includes a 2-minute local stimulation-OFF baseline, continuous stimulation with 60 seconds of stable LFP recording at the target amplitude, and at least 5 minutes of OFF washout between blocks. If bilateral 5.0 mA is not tolerated, the maximum common tolerated amplitude is used. Challenges may be modified or stopped for safety.

Stimulation-induced change in anterior thalamic local field potential (LFP) power in the 20-50 Hz slow-gamma band, quantified as ΔSG = ln(P_ON/P_OFF), where P_ON and P_OFF are integrated band power during 60 s of stable continuous stimulation and during the immediately preceding 2-minute stimulation-off baseline, respectively, recorded with the implanted device (BrainSense Streaming). At the 3-month visit, ΔSG is measured at 2.0, 3.5 and 5.0 mA, with at least two repetitions per amplitude in a counterbalanced order; repeats are averaged within each hemisphere and then across hemispheres. Participants who cannot tolerate bilateral 5.0 mA are tested at the highest amplitude tolerated on both sides and analyzed separately in prespecified available-dose sensitivity analyses. Zero indicates no change; negative values indicate suppression.

Intraclass correlation coefficient (two-way mixed-effects, absolute agreement, single measure) between the two repeated measurements of ΔSG (defined in the primary outcome) at the same amplitude and visit, calculated for each amplitude at each visit.

Within-participant difference in repeat-averaged ΔSG (defined in the primary outcome) between 3.5 and 2.0 mA (3.5 minus 2.0 mA), tested at the same visit as supportive replication of the primary dose-response.

Per-participant slope of repeat-averaged ΔSG (defined in the primary outcome) against stimulation amplitude across 2.0, 3.5 and 5.0 mA, in log ratio per mA.

Repeat-averaged ΔSG (defined in the primary outcome) at 2.0 mA at each visit, each normalized to its own local stimulation-off baseline, describing how the response to the same amplitude changes with cumulative chronic stimulation.

Power in an approximately 5-Hz band centered on each participant's chronic sensing frequency, recomputed offline from the same challenge recordings and expressed as ln(ON/OFF). Its agreement with ΔSG is assessed by Spearman correlation and direction concordance.

Narrow-band power stored by the device during daily life (BrainSense Timeline), normalized to the stimulation-free period after implantation; change from a prespecified early stimulation window to a window at month 3. Its relation to ΔSG is assessed by Spearman correlation.

(Baseline monthly seizure frequency - mean monthly seizure frequency in months 7-12) / baseline × 100, from seizure diaries; positive values indicate reduction. Baseline uses a prospective diary of at least 8 weeks (target 12 weeks) before implantation where available, and otherwise systematically recorded diaries or medical records. Its relation to the acute and chronic LFP measures is assessed by Spearman correlation.

Overlap of the modeled electric field (volume of tissue activated) with the anterior thalamic nucleus and with the mammillothalamic tract for each challenge condition, expressed as the percentage of each structure's volume, estimated with Lead-DBS from postoperative CT co-registered to preoperative MRI. Active-contact location and diffusion tensor imaging (DTI) structural connectivity are analyzed as additional explanatory measures. Relation to ΔSG is assessed by Spearman correlation.

Same calculation as the main seizure outcome, restricted to focal-to-bilateral tonic-clonic seizures, in participants who have them at baseline; prespecified sensitivity anchor

Same calculation, restricted to disabling seizures witnessed or confirmed by a cohabiting or primary caregiver; prespecified sensitivity anchor

Number of electrographic seizure-like events or epileptiform transients per hour of analyzable recording (per analyzable segment for event-triggered recordings), using prespecified criteria. Where possible, events are adjudicated independently by two clinical neurophysiologists blinded to clinical outcomes. Calculated only for participants with raw time-domain or event-specific LFP recordings; agreement with diary-based seizure outcomes is assessed.

Challenges with two valid repeats at every planned amplitude, divided by challenges planned

Reasons for not reaching it (adverse event, seizure worsening, clinical judgment or technical limitation) are reported

Recordings meeting prespecified signal-quality criteria in both hemispheres, divided by recordings performed

Participants whose chronic narrow-band data meet prespecified coverage in both hemispheres

Changes made for clinical safety during the first 3 months after activation

Graded with the Common Terminology Criteria for Adverse Events (CTCAE) version 5.0

Range 0-30; higher scores indicate better cognition

Range 0-100; higher scores indicate better quality of life

Log ratio ln(ON/OFF) computed as for ΔSG; band limits defined in the statistical analysis plan

Difference between ON and OFF in the aperiodic exponent/offset of the parameterized LFP power spectrum

Difference between left and right hemispheres in device-recorded spectral power over time

Deviation from the participant's usual narrow-band power around events marked with the patient programmer

Months 7-12 compared with baseline

Same calculation as the main seizure outcome, using months 10-12

All disabling seizures recorded in the diary, regardless of corroboration

Longest run of days without a diary-recorded seizure