The study, called COMEDD – Chronic Outcome Monitoring in Dystonia with Deep Brain Stimulation – takes advantage of a technological shift that has only recently become possible. Modern sensing-enabled neurostimulators, such as the Medtronic Percept devices used in the study, can record local field potentials, the summed electrical oscillations of neurons surrounding the implanted electrode, while simultaneously delivering therapeutic stimulation. In Parkinson’s disease, this capability has already proven clinically decisive: beta band oscillations around 13 to 35 hertz in the subthalamic nucleus track bradykinesia, the slowness of movement, and beta-based adaptive stimulation systems that automatically suppress this activity are now in clinical use. Dystonia researchers believe the same principle can work in the pallidum, but the biomarkers there are different and far less systematically mapped.

Previous work has identified a promising candidate. Low-frequency activity in the theta and alpha range, roughly 5 to 13 hertz, correlates with dystonia symptom severity, and effective stimulation suppresses it. Intriguingly, chronic stimulation has also been observed to increase beta band activity in dystonia patients, who simultaneously develop bradykinesia – a stimulation-induced side effect that mirrors the transdiagnostic role beta plays in Parkinson’s disease. Gamma band activity, which has recently emerged as a marker of effective therapy states in Parkinson’s disease, has also been described in the dystonic pallidum. What has been missing is any systematic, long-term characterization of how these oscillations evolve over the first year of stimulation, precisely the period in which clinicians must make their most consequential programming decisions without electrophysiological guidance.

COMEDD aims to fill that gap with a two-pronged data collection strategy applied to the same patient cohort over 12 months following surgery. The researchers expect to recruit 25 to 30 patients with isolated dystonia, aged 5 to 80 years, who receive pallidal electrodes and choose a sensing-enabled Percept neurostimulator. The first strategy relies on continuous chronic recording: the device tracks peak biomarker activity around the clock, storing one mean value every 10 minutes at an investigator-selected peak frequency. In one hemisphere, low-frequency activity is tracked as the leading candidate biomarker for dystonic symptom severity; in the other, beta activity is monitored to capture any stimulation-induced increase over time. Over a full year, this yields up to 365 days multiplied by 144 mean values per day – an extraordinary longitudinal dataset of the living brain under therapy.

The second strategy complements the continuous stream with high-resolution monthly recordings in the hospital’s electrophysiology laboratory. During these approximately one-hour visits, local field potentials are sampled at 250 hertz during rest and during voluntary finger tapping, recorded with an accelerometer, and crucially during both stimulation ON and OFF conditions. The protocol also includes monopolar reviews, in which stimulation amplitude is increased step-wise per hemisphere, allowing the researchers to map how different stimulation settings alter the pallidal electrical landscape. In cervical dystonia patients, bilateral electromyography of the sternocleidomastoid muscle is continuously recorded, and when present, the geste antagoniste – the voluntary counter-posture many dystonia patients instinctively use to relieve their symptoms – is performed during both ON and OFF stimulation.

Correlating brain signals with lived experience requires measuring that experience with equal rigor. Every week, patients complete a 15-question patient-reported outcome questionnaire covering general wellbeing, mood, energy, pain, sleep quality, hand and neck function, walking ability, self-care, therapy effect, and side effects, rated on a 9-point Likert scale. The questions were refined through a patient engagement program piloted by the same group, ensuring the instrument reflects what patients actually notice. Alongside the questionnaires, participants record video diaries: cervical dystonia patients film a shortened protocol modeled on the Toronto Western Spasmodic Torticollis Rating Scale, while those with generalized dystonia or their caregivers record whole-body videos at rest. These videos will be rated by blinded clinicians and analyzed with automated video-based kinematic assessment, providing objective severity measures that can be matched against both the subjective reports and the electrophysiological data.

Clinical characterization anchors the entire dataset. At the first and last recording visits, and at each monthly in-hospital session, the team assesses the Burke-Fahn-Marsden Dystonia Rating Scale and the Toronto Western Spasmodic Torticollis Rating Scale for dystonic motor symptoms, and the Unified Parkinson’s Disease Rating Scale part III for bradykinetic symptoms. Non-motor dimensions – pain, disability, depression, apathy, and obsessive-compulsive behavior – are captured with validated instruments including the Beck’s Depression Inventory and the Starkstein Apathy Scale. The primary outcomes are specific and falsifiable: the researchers expect suppression of low-frequency activity over time to correlate with clinical improvement on the dystonia scales, the patient-reported outcomes, and the automated video analysis; and they expect the increase in beta band activity under chronic stimulation to correlate with deterioration in the bradykinesia items of the UPDRS-III and with slowing in the monthly accelerometer recordings of finger tapping velocity.

The statistical design reflects the rarity of the disease. Because electrophysiological studies in dystonia have historically shown significant effects in samples of only 6 to 27 patients, the team performed a power analysis by simulating a longitudinal mixed model in R, drawing on previous findings that changes in the Burke-Fahn-Marsden scale related to low-frequency activity with a marginal R-squared of 31 percent. Simulating weekly means of 35 measurements per patient across the year, accounting for dropouts and missed data, the analysis determined that 18 patients would be sufficient to detect a significant relationship at a 5 percent alpha level with 80 percent power. Aiming for 25 to 30 recruits to buffer against attrition, with at least 18 cervical dystonia patients for a planned sub-analysis of the most common dystonia type, the study balances feasibility against statistical rigor. Two patients have already completed the full protocol with 92.3 percent appointment adherence, underlining that the demanding schedule of monthly hospital visits and weekly home monitoring is genuinely workable.

Exploratory outcomes reach beyond the primary biomarkers. The team aims to determine whether stimulation-entrained gamma activity correlates with the clinically effective stimulation contacts, which could directly support DBS optimization, and plans to explore biomarker stability and anatomical localization across the longitudinal dataset. Analysis will be performed in Matlab and Python using established open-source toolboxes including FieldTrip, the perceive toolbox, specparam for separating periodic and aperiodic spectral components, and the ReTap algorithm for quantifying finger tapping from accelerometer data. The researchers also hope to electrophysiologically characterize non-responders, whose pallidal signatures may differ from those of patients who benefit, and, if possible, to identify distinct pallidal patterns for different dystonia subtypes. A modular study design allows participants to reduce recording frequency without leaving the study, and a patient meeting is planned at the end of recruitment to discuss results with participants directly.

The implications extend well beyond the clinic. If chronic low-frequency suppression indeed tracks the delayed clinical effects of stimulation, biomarkers could inform and accelerate the often strenuous programming process, and ultimately steer adaptive algorithms that adjust stimulation automatically based on the brain’s own signals. The home monitoring concept, combining subjective questionnaires with objective video kinematics, could also be applied beyond DBS, for instance in patients treated with botulinum toxin. The study is registered as NCT07244549, and its progress can be followed in the public registration log. Approved by the ethics committee of Charité under the identifiers EA1/164/23 and EA2/163/25, the protocol introduces no additional clinical risk, since it uses the in-situ therapeutic devices for recording and non-invasive measures of motor activity, with negligible battery drain from the rechargeable implant. What the Berlin team is really building is a bridge between the electrical whispers of the basal ganglia and the lived reality of dystonia patients – a bridge that could one day let the brain itself tell clinicians exactly how to heal it.

Subject of Research: The chronical outcome monitoring in dystonia with deep brain stimulation (COMEDD) study protocol: a longitudinal evaluation of chronic electrophysiological biomarkers in patients with dystonia and deep brain stimulation therapy

Article Title: The chronical outcome monitoring in dystonia with deep brain stimulation (COMEDD) study protocol: a longitudinal evaluation of chronic electrophysiological biomarkers in patients with dystonia and deep brain stimulation therapy

Article References: Feldmann, L. K., Kaplan, J., de Almeida Marcelino, A. L., Schulze, D., & Kühn, A. A. (2026). The chronical outcome monitoring in dystonia with deep brain stimulation (COMEDD) study protocol: a longitudinal evaluation of chronic electrophysiological biomarkers in patients with dystonia and deep brain stimulation therapy. Dystonia, 5, Article 16304. https://doi.org/10.3389/dyst.2026.16304

Image Credits: AI Generated

DOI: 10.3389/dyst.2026.16304

Keywords: chronical, outcome, monitoring, dystonia, deep, brain, stimulation, COMEDD, protocol, longitudinal, evaluation, chronic