Dystonia: GPi DBS, Cervical Dystonia, and Genotype

Evidence status. Evidence is strongest for isolated generalized and segmental dystonia and established cervical dystonia cohorts. Genotype-outcome data are mostly observational and should guide counseling, not create absolute exclusions. Acquired dystonia sits outside that evidence base and is treated separately below.

1.Confirm phenotype and reversibility

Document mobile versus fixed postures, phasic movements, tremor, pain, gait, speech/swallowing, task specificity, and diurnal variation. Review video over time. Exclude mimics and treat dopa-responsive dystonia, Wilson disease, medication-induced syndromes, functional movement disorder, and orthopedic or spine pathology as appropriate.

Optimize botulinum toxin for focal/segmental disease, oral medication, rehabilitation, sensory strategies, and pain care. DBS is considered when disability, pain, injury, or loss of independence remains substantial and realistic goals can be measured.

2.Target choice

Posteroventral GPi is the standard target, with broad evidence across generalized, segmental, and cervical dystonia. Benefits often accumulate over weeks to months; tonic postures may improve more slowly than phasic movements. STN DBS can produce faster effects and lower energy use in some cohorts. GPi has longstanding randomized evidence in generalized/segmental and cervical dystonia; STN evidence now also includes the 2025 randomized, double-blind RELAX trial in isolated generalized or segmental dystonia. This does not establish equivalence across phenotypes or make STN part of the US dystonia device label. The randomized PASTS-CD trial comparing the two targets in cervical dystonia has published its protocol, not its results. Both targets can cause overlapping adverse effects. GPi stimulation may produce bradykinesia or gait impairment, and STN stimulation may induce dyskinesia; these are possible complications, not diagnostic signatures of target location.

Trajectory and contact location should span the sensorimotor pallidum while respecting optic tract, internal capsule, and vascular anatomy. Programming begins with a broad monopolar survey and then becomes a negotiation. Dystonia programs are energy-hungry, so generator choice and the case for a rechargeable device belong in the preoperative conversation rather than the year-two conversation. Stimulation-induced parkinsonism, meaning bradykinesia, micrographia, hypophonia, freezing or festination emerging weeks to months after a good dystonia response, is the characteristic late programming problem, and it is frequently missed because it arrives long after the settings that caused it. Re-examine the patient and review medication, device function, and lead imaging as indicated. Programming options include reducing amplitude, testing lower frequency, narrowing or steering the field, or selecting a more dorsal contact; the choice and sequence depend on the examination, anatomy, and system. Assess each change against recurrence of dystonia and the patient’s functional goals. If benefit and parkinsonism cannot be separated, reassess lead location, medication, disease evolution, and the benefit–harm tradeoff; this alone does not prove lead malposition.

3.Etiology sets the ceiling before genotype refines it

The isolated dystonias, inherited or idiopathic, with no other neurologic sign but tremor, carry the evidence base, and everything on this page is calibrated to them. Acquired dystonia is a different conversation. Dyskinetic cerebral palsy is the most common acquired referral in pediatric practice; pallidal stimulation there produces real but markedly smaller motor gains than in isolated disease, much of the residual disability is fixed rather than dystonic, and the honest goal is often pain, positioning, care burden, and seating rather than a scale change. Tardive dystonia, by contrast, is among the more responsive acquired forms and should not be lumped with the rest. Heredodegenerative and metabolic dystonias sit between, with the underlying disease rather than the dystonia setting the trajectory. Say which of these categories your patient is in before you quote any number to them.

4.Selection beyond the neck angle

Refractory cervical dystonia can respond substantially to GPi DBS, including severity, disability, and pain. Confirm high-quality toxin trials with correct muscles, doses, guidance, and intervals. Characterize head tremor, anterocollis/retrocollis, shoulder elevation, sagittal shift, pain generator, and any fixed spine disease.

Pooled individual-patient data from 208 patients across published GPi and STN cohorts put TWSTRS total improvement near sixty percent at a mean of about two years, with severity and disability moving together and pain improving somewhat less. Read those figures for what they are: open-label, unblinded, and pooled across centers with different technique. The classic randomized sham-controlled GPi evidence includes generalized/segmental and cervical dystonia trials with extensions, and the blinded three-month effect is considerably smaller than the open-label figure. The difference is the natural history of a therapy whose benefit accrues over months, plus the expectation effect that open follow-up cannot remove. Quote the pooled number if you also say which kind of number it is. Discuss stimulation-induced parkinsonism, gait or speech change, infection, revision, and the need for prolonged programming.

5.Measure what matters

Use blinded standardized video when possible. In cervical dystonia report TWSTRS severity, disability, and pain separately; in generalized and segmental dystonia the Burke-Fahn-Marsden Dystonia Rating Scale movement and disability subscales are the instrument nearly every published number you will meet is expressed in, so record both at baseline. Add patient-selected activities, analgesic use, sleep, work, driving, and botulinum toxin requirements. A visible postural improvement with persistent pain is not complete success; pain relief without perfect alignment may be life-changing.

6.Genetics refines probability

The rows summarize reported response signals; they are not a validated ranking across genes. Locus numbers such as DYT1, DYT6, and DYT11 are retired but persist in the older literature and in board questions, so expect to read either convention. Every row rests on small, selected, largely open-label series; sample sizes and follow-up vary substantially by gene, with publication and selection bias.

7.The phenotype can overrule the gene

A favorable genotype does not reverse contracture, severe scoliosis, or advanced bulbar disease. An uncertain genotype does not preclude benefit when the dominant disability is mobile, medically refractory dystonia and the network target is plausible. Reanalyze older genetic tests as panels and classifications evolve, but avoid delaying urgent treatment of life-threatening status dystonicus when the clinical case is coherent.

8.Timing and fixed disability

Longer disease duration and skeletal deformity often predict less complete functional recovery, especially in childhood-onset generalized disease. This duration effect is a generalized-dystonia finding; in adult-onset cervical dystonia, pooled data have not identified duration or any other reliable predictor of response. Earlier surgery may preserve education, mobility, and independence, but pediatric hardware burden, growth, repeated anesthesia, and decades of device management belong in the decision.

9.Longitudinal care

Dystonia DBS is not a one-visit operation, and prospective follow-up now extends beyond a decade. The ten-year extension of the original randomized trial was open-label; the publication’s use of “controlled” does not mean patients remained randomized to sham for ten years. In the prospective multicenter pallidal cohort followed to a decade, motor scores remained about 56 percent below baseline, with stable disability, mood, and cognition. But about a third of the original cohort were nonresponders at ten years: a small number never responded, and a larger number responded and then lost it; suboptimal lead placement accounted for roughly half of these failures, which is a surgical lesson as much as a programming one. Secondary loss of benefit, not primary nonresponse, is the more common late failure. Plan programming intervals, medication and toxin tapering, rehabilitation, orthopedic review, psychiatric support, transition from pediatric to adult care, hardware surveillance, and a rescue pathway for battery depletion. A patient who is doing well at year two has not finished being at risk.