Dystonia is a movement disorder defined by sustained or intermittent muscle contractions causing abnormal, often repetitive movements or postures. Mechanistically it reflects a failure of surround inhibition in motor circuits: the basal ganglia normally sharpen a motor command by suppressing competing, antagonistic muscles, and when that selectivity breaks down, agonist and antagonist co-contract and the limb twists into a fixed posture. This is why dystonia is fundamentally a disorder of network function rather than a single anatomic lesion — and why most primary genetic dystonias have a structurally normal MRI.

The 2013 consensus classification exists because older labels ('primary,' 'dystonia-plus,' 'secondary') mixed clinical description with presumed cause and aged poorly as genes were discovered. The fix was to separate what you see from why it happens across two independent axes:

• Axis I — clinical features (the phenotype: how it looks at the bedside)
• Axis II — etiology (the cause: inherited, acquired, or idiopathic)

The power of Axis I is that the phenotype itself narrows the genetic differential before any test is sent. A clinician should reflexively characterize:

• Age of onset — the single strongest predictor of an underlying gene; earlier onset and lower-limb/generalized spread point toward a monogenic cause, whereas adult-onset focal dystonia is usually idiopathic.
• Body distribution — focal, segmental, multifocal, generalized, or hemidystonia (the last should prompt imaging for a contralateral structural lesion).
• Temporal pattern — static vs. progressive, and crucially whether it is task-specific (writer's cramp) or shows diurnal fluctuation (the red flag for dopa-responsive dystonia).
• Associated features — whether dystonia is isolated or combined with another movement disorder; this distinction, more than any other, sorts patients toward specific gene families.

Dopa-responsive dystonia (DRD) earns its own section because no other genetic dystonia is so disabling when missed and so completely reversible when caught. Most cases are autosomal dominant with variants in GCH1, the gene encoding GTP cyclohydrolase 1 — the first inherited dystonia for which a causative gene was identified Ichinose et al. 1994.

Why the biochemistry matters: GTP cyclohydrolase 1 catalyzes the rate-limiting step in the synthesis of tetrahydrobiopterin (BH4). BH4 is not itself a neurotransmitter — it is the obligate cofactor for the aromatic amino acid hydroxylases, including tyrosine hydroxylase, which is itself the rate-limiting enzyme of dopamine synthesis. So a heterozygous GCH1 mutation produces a cofactor bottleneck that throttles dopamine production without killing the neurons. The nigrostriatal cells are alive and intact — they are simply starved of the cofactor they need to make dopamine.

This is the entire clinical logic of DRD. Because the deficit is a synthesis bottleneck in structurally healthy neurons (not neurodegeneration), supplying downstream levodopa bypasses the block and restores function essentially to normal — at very low doses, sustained for decades, without the dyskinesias and wearing-off fluctuations that define Parkinson disease. In PD the dopaminergic neurons are dying and the surviving terminals lose their capacity to buffer levodopa, generating motor fluctuations; in DRD the terminals are healthy and store dopamine normally, so the response is smooth and stable. A dramatic, fluctuation-free response to a tiny dose is therefore not just treatment — it is virtually diagnostic.

Why it is missed: the early picture — toe-walking, leg stiffness, an abnormal gait, brisk reflexes — looks exactly like spastic diplegic cerebral palsy or idiopathic dystonia, and many children carry a CP label for years. The diurnal pattern (worse by evening, restored by sleep) is the giveaway, but it is easy to overlook unless specifically asked about. The practical rule that follows is absolute: trial levodopa in every child with unexplained dystonia before concluding the cause is CP or 'idiopathic.'

DYT-TOR1A (previously DYT1) is the most common cause of early-onset primary generalized dystonia and the prototype monogenic dystonia. It was the first dystonia gene cloned by positional methods, and the discovery is instructive: nearly every patient — across unrelated ethnic populations — carries the identical lesion, a 3-bp in-frame GAG deletion (ΔE303) that removes a single glutamic acid from torsinA Ozelius et al. 1997. When the same tiny deletion recurs independently rather than tracking a shared haplotype, it tells you that position is uniquely intolerant of change — a structural pressure point in the protein.

What torsinA does, and why losing one residue matters. TorsinA is an AAA+ ATPase residing in the endoplasmic reticulum lumen and contiguous nuclear envelope. AAA+ ATPases use cycles of ATP binding and hydrolysis to remodel other proteins, and torsinA's job is to help organize the LINC complex that physically couples the nucleoskeleton to the cytoskeleton across the nuclear envelope. The ΔE303 deletion sits in the ATPase domain and acts as a dominant-negative: the mutant subunit poisons the oligomeric machine rather than merely subtracting one working copy, which is why a single mutant allele produces disease. The downstream consequence is impaired nuclear–cytoskeletal coupling, most consequential in striatal neurons — again a functional defect in living cells, consistent with the normal MRI.

Reduced penetrance is the counseling headline. Only about 30% of ΔE303 carriers ever become symptomatic, so most people who inherit the deletion remain well for life. This decouples genotype from phenotype: an unaffected parent can transmit the variant, and predictive testing of relatives will identify carriers who may never develop dystonia — facts that must frame any conversation about cascade testing. When dystonia does appear, it characteristically begins in a leg or arm in childhood (median age 9 years; about 5% after 30) and often spreads; cranial onset is uncommon, and about 20% remain focal.

Combined dystonias are defined by dystonia appearing alongside another movement disorder — myoclonus, parkinsonism, or chorea — and that companion sign is diagnostic gold. Once you classify a dystonia as combined rather than isolated, you have already eliminated most of the genome: the question shifts from 'which of dozens of dystonia genes?' to 'which gene produces this specific pairing?'

Myoclonus-dystonia (DYT-SGCE) illustrates the reasoning. The phenotype is lightning-quick, alcohol-responsive myoclonus of the arms and trunk with only mild dystonia, frequently accompanied by OCD and anxiety. Its pedigree pattern is the teaching point: SGCE is maternally imprinted, so the maternal allele is methylated and silenced and only the paternally inherited copy is expressed. A pathogenic variant therefore causes disease almost only when transmitted by the father; the same variant passed from an affected mother yields unaffected children who are nonetheless carriers. This produces apparent generation-skipping in a pedigree that is genuinely autosomal dominant — a pattern that looks like incomplete penetrance but is actually a deterministic consequence of imprinting.

Why the complex label drives urgency. When dystonia comes with other neurological or systemic signs, the dystonia is complex rather than combined, and two groups matter most. Wilson disease (ATP7B) is a treatable emergency and must be excluded in every young person with dystonia, because copper chelation halts it, often with substantial recovery, while untreated disease causes irreversible damage; the pairing to look for is dystonia plus liver disease, psychiatric change, or Kayser–Fleischer rings. The NBIA disorders (PKAN/PANK2, PLAN/PLA2G6, BPAN/WDR45) instead pair dystonia with iron deposition visible as globus pallidus T2 hypointensity — the 'eye of the tiger' of PKAN — and, while not reversible, their imaging signature is so specific it short-circuits the workup. Recognizing the pattern is thus not academic taxonomy; it is what separates a treatable diagnosis from a missed one.

The governing principle of the dystonia workup is treatability before taxonomy: exclude the conditions you can fix before settling on a primary genetic label, because a months-long gene panel turnaround is unacceptable when a levodopa trial or copper chelation could be started today. Two diagnoses sit at the front of every childhood- or young-adult-onset evaluation — dopa-responsive dystonia and Wilson disease — precisely because both are common enough to encounter, devastating if missed, and dramatically reversible. The phenotype (isolated vs. combined, focal vs. generalized, with or without diurnal fluctuation) directs the order in which the remaining treatable causes (GLUT1 deficiency, NPC, glutaric aciduria type 1, biotin-thiamine-responsive basal ganglia disease, AADC deficiency) are pursued.

Why genetic testing has become panel-first. The genetic dystonias are strikingly heterogeneous and phenotypes overlap, so for most generalized presentations a broad NGS panel is more efficient and cheaper than sequential single-gene tests. The exceptions are the cases where the phenotype is nearly pathognomonic — the Ashkenazi child with classic limb-onset generalization (test TOR1A first), the eye-of-the-tiger MRI (PANK2), the alcohol-responsive myoclonus (SGCE) — where targeted testing confirms a strong clinical hypothesis fastest.

Treatment logic, from least to most invasive. Oral therapy is built around mechanism. Levodopa is both a treatment and a diagnostic test and is mandatory in all childhood-onset dystonia. Trihexyphenidyl (anticholinergic) is the oral agent with the best trial evidence in generalized dystonia but must be titrated slowly because cognitive side effects limit dosing, especially in adults. Botulinum toxin is first-line for focal or segmental disease, chemically denervating the overactive muscles. When oral agents fail in generalized disease, GPi deep brain stimulation is the escalation — and its rationale is specific: dystonia is a disorder of aberrant pallidal output, and high-frequency stimulation of the internal globus pallidus normalizes that output. DBS works best in exactly the genetic forms discussed above — TOR1A and KMT2B among the strongest responders — and outcomes are better with earlier surgery, before fixed contractures and skeletal deformity become irreversible, which is why identifying a DBS-responsive genotype can itself justify earlier referral.