38 Spasmodic torticollis

A fanciful interpretation of fossil records has suggested the existence of torticollis in dinosaurs (Kaiser 1954). It is uncertain when the disorder was first recognized in man but Michelangelo seems to depict this (Fig. 38.1). Rabelais, the 16th century French physician, priest, and satirist, is credited with first using the term ‘torty colly’ from which torticollis is derived. Involuntary neck movement was described by Wepfer in 1727, but attempts at surgical correction are recorded as early as 1641 (Finney and Hughson 1925). By the end of the 19th century the disorder was well known and several excellent papers by French and British physicians date from this era (Gowers 1893, Brissaud 1895, Redard 1898, Cruchet 1907). At this time there was a tendency to regard many, if not most, of these cases as being due to hysteria (Fig. 38.2). By the 1920s to the 1940s the pendulum had swung the other way and most cases were thought to be organic. In 1943 Patterson and Little clearly outlined the major features in a clinical study of 103 patients with spasmodic torticollis. Six years later, in another large series of patients, the clinical, electromyographic, and surgical aspects were well described (Herz and Glaser 1949, Herz and Hoefer 1949, Putnam et al. 1949).

Torticollis is a disorder of posture in which the neck is held turned or flexed to one side. Spasmodic torticollis is a condition in which the neck intermittently and involuntarily adopts this position. It is associated with tonic or clonic spasms of the neck muscles. There is frequently associated flexion (antecollis) or extension (retrocollis) of the neck, either of which can occur without torticollis. Spasmodic torticollis may be divided into idiopathic (primary) or symptomatic (secondary) varieties. The secondary form can be caused by a wide spectrum of pathologies, including most disorders that cause dystonia. Some symptomatic varieties are shown in Table 38.1. In idiopathic spasmodic torticollis there is no apparent cause. Idiopathic torticollis may occur as part of generalized primary dystonia (dystonia musculorum deformans) or as a focal or segmental dystonia. In the latter case there may be associated cranial or upper limb dystonia (with or without tremor) (see Chapters 37 and 39). We have restricted the term ‘spasmodic torticollis’ to primary focal or segmental varieties and further discussion is limited to these. Torticollis associated with idiopathic torsion dystonia is discussed in Chapter 35, ‘Generalized Primary Dystonia’, and symptomatic causes of dystonic torticollis are covered in chapter 41 in the section that deals with secondary or symptomatic dystonias, especially Chapter 43. As antecollis and retrocollis are usually clinical variants rather than separate disorders, they are included in the discussion on torticollis. Some authors have defined the direction of torticollis as being the side to which the occiput turns, while others have used the direction of rotation of the chin. We have used the latter.

There have been many neuropathology investigations in which torticollis has been associated with dystonia musculorum deformans or other diseases. There are, however, few examples of primary spasmodic torticollis and these have shown no consistent anatomical abnormality (Tarlov 1970, Holton et al. 2008). Some radiological reports have suggested structural changes. Historically, pneumoencephalograms were reported as showing frontal, parietal, or temporal cortical atrophy in 10 out of 13 patients examined by Kaste et al. (1981) and two also had dilatation of lateral ventricles. Generalized brain atrophy was noted in two patients submitted to pneumoencephalography by Patterson and Little (1943). Hassler and Dieckmann (1970) found a ‘significant difference in size’ between the cerebral hemispheres in 85% of patients with spasmodic torticollis submitted to this radiological investigation. Similar findings, however, have not been reported with CT and MRI brain scan (Schneider et al. 1994) and they are of dubious significance. More advanced imaging techniques, however, suggest changes in patients with torticollis compared to controls, such as differences of fractional anisotropy and mean diffusivity in the corpus callosum, both putamina, the caudate, pre-frontal cortical areas, and supplementary motor areas in cervical dystonia patients when applying diffusion tensor imaging (Colosimo et al. 2005, Fabbrini et al. 2008).

Thorough neuropathological examination of the spinal cord, nerve roots, and peripheral nerves has not been undertaken in this disorder.

Autopsy studies of brain biochemistry have not been reported in spasmodic torticollis, and biochemical studies of cerebrospinal fluid (CSF) have been inconclusive. Curzon (1973) found normal levels of dopamine and serotonin metabolites [homovanillic acid (HVA) and 5-hydroxyindolacetic acid (5HIAA)] in lumbar CSF, while Papeschi et al. (1972) reported HVA values in ventricular

CSF intermediate between controls and patients with Parkinson's disease. A single patient investigated by Johansson and Roos (1974) had low HVA and 5HIAA levels. CSF gamma aminobutyric acid levels have been normal (Abbott et al. 1983). Abnormalities hypothesized on the basis of pharmacological responses are mentioned below under ‘Pathophysiological mechanisms’.

Radiological abnormalities suggesting underlying brain atrophy have been mentioned above. It has been postulated that such damage, particularly to the putamen and adjacent areas, may occur early in life but remain undetected until later decompensation produces torticollis (Hassler and Dieckman 1970).

Release of an underlying postural rotational abnormality has been proposed by Stejskal and Tomanek (1981) who found asymmetry in calorically induced nystagmus (see later under ‘Investigations’). Nystagmus with the slow phase in the direction of the torticollis was of longer duration, higher velocity, and greater amplitude. They also noted a correlation with preferred direction of turning, which was greater to the side of the torticollis. Similar directional preponderance of vestibular nystagmus has been found by others (Matthews et al. 1978, Bronstein and Rudge 1988). It has not, however, been definitively demonstrated that asymmetry of the vestibular system causes the torticollis. Arguments against such an aetiology include the observation that ocular fixation is able to compensate for this directional preponderance, which decreases with increased disease duration (Bronstein and Rudge 1988, Stell et al. 1989). In addition, when torticollis does arise from an eighth nerve lesion, the direction of head rotation in relation to the side of the vestibular abnormality is inconstant.

The abnormal head position is probably not responsible for the asymmetry in nystagmus, as duplication of this position in normals does not produce these changes and head straightening in spasmodic torticollis does not eliminate them (Bronstein and Rudge 1988, Stell et al.  1989). Overall, it seems more likely that the torticollis and changes in nystagmus are both caused by the underlying abnormality (Bronstein and Rudge 1988).

Lesions of the brainstem in experimental animals have been more successful in producing torticollis than cerebral or corpus striatum damage. Electrolytic lesions of the midbrain, including the region of the red nucleus, brachium conjunctivum, medial reticular formation, and medial longitudinal fasiculus, produce torticollis in monkeys and cats (Carpenter 1956, Foltz et al. 1959, Denny-Brown 1962, Malouin and Bedard 1982). A patient with a lesion in this region due to multiple sclerosis has been reported to develop spasmodic torticollis, suggesting possible relevance of these structures in humans (Plant et al. 1989) (see Chapter 43). In animals rotation occurs so the chin is directed towards the side of the lesion and it may be associated with head tilt. As in man this rotation is often spasmodic and may be hardly detectable when the animal is calm, although it becomes obvious during excitement. Complex tonic and clonic contractions may occur and appear largely due to sternomastoid activity (Foltz et al. 1959). Although the precise structures involved are uncertain the interstitial nucleus of Cajal (Fig. 38.3) and its connections may be important in this (Malouin and Bedard 1982). This nucleus is intimately connected with the vestibular apparatus and interference with its function has been postulated to underlie the head tilt seen as part of the ocular tilt reaction caused by meso-diencephalic lesions in humans (Halmagyi et al. 1990). Lesions of the vestibular nuclei produce temporary head tilting in animals (Tarlov 1969, Malouin and Betard 1982), but unilateral eighth nerve or labyrinthine damage has not been clearly demonstrated to cause torticollis in adults (Tarlov 1970). Intermittent labyrinthine dysfunction has been suggested as a cause of paroxysmal torticollis in infancy (Snyder 1969) and occasionally spasmodic torticollis has been reported in association with unilateral eighth nerve lesions, but more extensive neurological damage has also been present (Bronstein et al. 1987).

Destruction of the zone of termination of interstitio-thalamic fibres in the internal segment of the oral part of the ventrolateral nuclear complex of the thalamus may relieve spasmodic torticollis in humans (Hassler and Diekman 1970, Bertrand et al. 1978). It has been proposed that such lesions interrupt a pathway subserving head rotation originating in the vestibular nuclei and passing via the interstitial nucleus of Cajal to the contralateral thalamus and thence to frontal eye fields in cortical area 8 (Bertrand et al. 1978). The prestitial nucleus lies in the prerubral region, rostral and ventral to the interstitial nucleus and dorsal and caudal to the mamillary bodies. Stimulation of this produces extension of the neck (Jung and Hassler 1960). The mechanisms involved in head turning, however, are complex and utilize other structures, including the superior colliculus and entopeduncular nucleus (Montanelli and Hassler 1964). Non-dopaminergic fibres from the substantia nigra pars reticulata which send axon colaterals to these structures and the thalamus may be involved. (See earlier in ‘Anatomy’ sections Chapters 1 and 2). In addition, involvement of dopaminergic mechanisms in head turning is suggested by torticollis in monkeys with unilateral destruction of the nigrostriatal pathway (Crossman and Sambrook 1978). Although head turning can be produced by damage of a number of brainstem structures, there is no direct evidence that involvement of these is the cause of human idiopathic spasmodic torticollis. The mechanisms underlying this disorder have perhaps been clarified by the finding that abnormalities of the blink reflex are present in patients with spasmodic torticollis and are similar to, but less marked than, those with cranial dystonias (see Chapter 37 and below in ‘Investigations’). This suggests that the interneurons subserving some brainstem reflexes are in a hyperexcitable or disinhibited state. While abnormalities of the blink reflex could not directly be responsible for the torticollis, they may reflect a subclinical tendency to cranial dystonia, which is known to be more common in spasmodic torticollis than normal. They also suggest that a similar state of disinhibition may be present in those brainstem and spinal neuronal circuits involved in head rotation, probably involving structures mentioned above.

Earlier workers postulated that the basic abnormality was at a spinal or nerve root level (Cruchet 1907). While this now seems unlikely it must be admitted that this possibility has not been fully explored. Some authors have considered the disorder is due to psychological disturbance. This possibility was emphasized by early French workers (Redard 1878, Brissaud 1895, Cruchet 1907). The movements have been regarded as having particular meaning for patients and have been interpreted as symbolizing an erect penis (Abse 1966) or a rejection with turning away (Whiles 1940). A less psychoanalytic approach has been taken by others who have claimed that patients with spasmodic torticollis show obsessional personality traits (Meares 1971[a]), a tendency to deny and suppress emotion, unresolved conflict, and evidence of marital or sexual disturbance (Meares 1971[b],[c]). Although Herz and Glaser (1949) claimed abnormalities of ‘basic personality patterns’ in over 50% of patients, they did not satisfactorily define these and concluded that psychological factors were not of major importance in the onset of the disorder. In a retrospective and uncontrolled review of 220 cases of isolated and idiopathic spasmodic torticollis  Rondot et al. (1991) felt there was evidence of antecedent psychopathology in 58%. In 33% it was termed moderate, consisting mainly of anxiety character traits, and in 25% it was regarded as major, including schizophrenia and other psychotic conditions, bipolar disorders, severe depression, and any disorder leading to psychiatric hospitalization. Other workers have found no evidence of psychological cause (Matthews et al. 1978). Cockburn (1971) made a detailed evaluation of the psychological features in 46 patients and matched controls. There was no significant difference between these groups with regard to Maudsley Personality Inventory N and E scores, frequency of neurotic symptoms in childhood, previous psychiatric treatment, previous psychological trauma, employment stability, marital status, marital stability, and the incidence of premorbid anxiety, depression, suicidal attempts, or alcoholism. The only difference between patients and controls was an increased incidence of premorbid obsessionality in the latter. The doubtful significance of this finding led Cockburn to conclude that the study provided ‘no evidence to support the hypothesis that spasmodic torticollis is either a psychogenic condition, or occurs in patients with abnormal premorbid personalities’.

It has also been claimed that anxiety, neuroticism, and marked marital or sexual disturbances are more common in patients who recover from spasmodic torticollis and that analysis of such features may allow prognostication (Meares 1971[c], Gundel et al. 2003). Others have been unable to confirm this (Matthews et al. 1978). Jahanshahi and Marsden (1988[a]) assessed 100 torticollis sufferers and compared them with 49 patients with cervical spondylosis, using measures of personality, obssessiveness, anxiety, marital status/harmony, employment, and events preceding the onset of the illness. The only significant differences noted were that higher proportions of patients with torticollis were in the permanently sick category of employment status and single. There was no difference in marital harmony between the two groups. In a follow-up study these authors (Jahanshahi and Marsden 1988[b]) found the prevalence of depression to be the same in patients with spasmodic torticollis as in those with cervical spondylosis, but the former had higher scores on the Beck Depression Inventory. Self-referent negative cognitions, such as self-blame, self-accusation, self-punitive thoughts, and negative body image, emerged as the prominent component of depression in torticollis. Disfigurement was a major predictor of depression in torticollis (Jahanshahi and Marsden 1990[a]) and over a 2-year follow-up the extent of depression and poor body concept seemed to mirror the severity of the physical disability (Jahanshahi and Marsden 1990[b]). There was improvement of the depression along with disability following botulinum toxin injections which improved torticollis (Jahanshahi and Marsden 1992). These findings were interpreted as supporting the notion that the depression was the result of the torticollis and not its cause. Such an interpretation was favoured by Rondot et al. (1991) in their retrospective review of 220 patients. Gundel et al. (2003) compared 48 patients with torticollis and 48 patients with alopecia areata who were matched for age, sex, and body image dissatisfaction. Psychiatric diagnoses were based on a structured psychiatric interview (SCID-I). Results of patients were also compared with a matched sample of the representative normal population. The authors found that odds ratios to develop psychiatric comorbidity for patients with the dystonia group were significantly increased throughout nearly all assessed DSM-IV categories compared with patients with alopecia areata. The authors concluded that the high psychiatric comorbidity in cervical dystonia is thus unlikely to be a mere consequence of chronic disease and disfigurement.

There is thus considerable conflict as to the role of psychological factors in the causation of this disorder. It seems likely, however, that many of the reported psychiatric changes may have been the result, rather than the cause, of the torticollis. Although occasional examples of hysterical torticollis and psychologically induced tics may occur, the vast majority of patients with spasmodic torticollis appear to have an organic disorder, although the cause of this has yet to be defined.

The exact incidence and racial predilection of spasmodic torticollis are unknown. The prevalence of torticollis of sufficient severity as to prevent employment has been estimated at 0.4/per 100,000 in Finland (Nuutila and Wickstrom 1971). This must, however, be a gross underestimate of the total number of sufferers in the community. The prevalence in Mayo Clinic studies has been put at 89 per million with an incidence of 11–12 per million person years, making it the most common type of focal dystonia (Nutt et al. 1988, Claypool et al. 1995). Cervical dystonia was also the commonest type of dystonia in an epidemiological study from North England, accounting for 42.7% of all primary dystonias (Duffey et al. 1998). The sexes are affected equally in some reports (Patterson and Little 1943, Jahanshahi and Marsden 1988[a]), although in others there is a predominance of females compared with males in ratios of about 1.5:1 (Jankovic et al. 1991, Rondot et al. 1991), 1.9:1 (Chan et al. 1991, Soland et al. 1996), and 2.2:1 (Duffey et al. 1998). In two Mayo Clinic studies, women were affected approximately four times as often as men, but relatively small numbers of patients were involved (Nutt et al. 1988, Claypool et al. 1995). Onset may occur as early as the first decade of life, but in the majority of patients it is delayed until the fourth or fifth decades. Patterson and Little (1943) found the average age of onset in 103 cases was 38 years and in 347 cases reported by Duane (1988) it was 43 years (Fig. 38.4). In other large series, the mean age of onset has similarly been in the fifth decade of life (Chan et al. 1991, Jankovic et al. 1991, Rondot 1991). Only 3.5% of Duane's (1988) patients were non-right handed, which is lower than the anticipated frequency of 10%. This has not generally been a feature and some workers have shown the proportion of non-right handers to be as expected (Rondot et al. 1991).

A family history of dystonic neck and other movements may occur in primary generalized dystonia and apparently isolated torticollis may really be a manifestation of this condition. It has been suggested that patients with a family history of dystonia tend to have onset of torticollis at a younger age (Chan et al. 1991). A family history of dystonia was reported in only 1.3% or cases of spasmodic torticollis studied by Rondot et al. (1991), but it was found to occur in 4.2% by Jahanshahi et al. (1990) and 11.7% by Chan et al. (1991). Examination of relatives is likely to be much more rewarding than reliance on history, and Waddy et al. (1991) found 35.7% of cases of idiopathic dystonic torticollis had relatives who showed signs of dystonia. It led the authors to propose that a single autosomal dominant gene might be responsible for the majority of inherited dystonia in Britain, whether focal or generalized. Similar findings were also observed in two other studies where first degree relatives of patients with any form of adult-onset focal dystonia were examined by the investigators (Defazio et al. 1993, Stojanovic et al. 1995). These studies also suggested autosomal dominant transmission of focal dystonia with reduced penetrance, although they did not analyze the different types of dystonia separately. It was unclear from the latter studies whether affected family members tended to develop the same type of dystonia as the index cases. In addition to the issue of apparently isolated cervical dystonia being a manifestation of generalized primary dystonia, spasmodic torticollis itself appears to be familial in a small number of cases (Thompson 1896, van Bogaert 1941, Gilbert 1977). Patterson and Little (1943) found a family history in 4% of cases, which accords with the 4.6% noted by Rondot et al. (1991). There is evidence to suggest this may be more frequent if the disorder is severe (Nuutila and Wickstrom 1971). The exact status of such families, however, is uncertain and it has not been conclusively shown that they are not just variants of generalized primary dystonia.

However, some large families with adult-onset cervical dystonia have been described (see Table 38.2). In some but not all, known loci were excluded. One family had five individuals definitely affected with torticollis (including a pair of monozygotic twins) among two generations (Uitti and Marganore 1993). Another five individuals were said to be possibly affected. The mean age of onset was 35.2 years and dystonia was limited to the neck. Arm tremor was present in one family member with definite dystonia. Another family of Mennonite origin with torticollis has been described with seven affected family members in two generations (Bressman et al. 1996). In three individuals there had been spread to involve the arm and face but none had generalized dystonia. The DYT1 locus was excluded in these two families (Bressman et al. 1996) and the family was later found to carry mutations in the THAP1 gene causing DYT6 dystonia. However, larger studies screening cohorts of adult-onset torticollis have shown that THAP1 mutations are not a major cause of this condition (see Chapter 35).

Another family from Germany with adult onset dystonia mainly affecting the cervical muscles has been reported (Leube et al. 1996). There were seven affected and six possibly affected individuals across four generations. The dystonia was mainly cervical, but there was one individual with laryngeal involvement. Two other individuals with torticollis had segmental dystonia. Linkage to chromosome 18p was identified (Leube et al. 1996) and this gene locus has been designated as DYT7. The DYT7 gene appears to be transmitted with an autosomal dominant pattern with reduced penetrance. Investigation of 15 sporadic patients with focal cervical dystonia from the same region indicated that these patients shared the same haplotype with family members linked to DYT7, thus suggesting a founder mutation in northwest Germany (Leube et al. 1997). However, there are other families with predominant cervical dystonia not linked to DYT6 or 7 loci, suggesting genetic heterogeneity (Leube et al. 1997, Munchau et al. 2000).

Trauma has been found to precede the onset of spasmodic torticollis in a number of studies and overall the incidence is in the order of 10%. In addition, cervical pain in the absence of trauma may precede the onset of neck rotation (Table 38.3). The need for patients to explain the cause of their symptoms may lead to unrelated trauma being reported and the significance of these observations is uncertain. Onset of neck pain immediately after trauma, commencement of torticollis within days of this, marked limitation of range of motion, persistence during and lack of improvement after sleep, absence of a ‘geste antagoniste’, lack of improvement with support, marked spasm of paracervical muscles, and absence of deterioration with action have been said to be features distinguishing this from spontaneously occurring spasmodic torticollis (Truong et al. 1991) and this raises the question as to whether some of the cases may not just be examples of non-specific musculo-skeletal injury. Patients who have had the onset of pain with generalized spasm of neck muscles and restriction of all cervical movements without torticollis immediately after local trauma have been labelled as having ‘post-traumatic cervical dystonia’ (Goldman and Ahlskog 1993). Such individuals seem quite different from the majority of cases of spasmodic torticollis and it is doubtful they are true examples of dystonia. In not all traumatically induced cases is there direct cervical injury and occasionally basal ganglia damage may be responsible (Isaac and Cohen 1989, Krauss et al. 1992).

Emotional stress aggravates the clinical features of many extrapyramidal disorders and may occasionally appear to precipitate the disease. Although severe psychological stress may immediately antedate the onset of head turning (Meares 1973) and, as mentioned above, a high incidence of preceding psychopathology has been noted in some series, a non-specific triggering effect cannot be excluded (Sheehy and Marsden 1980). In a small number of cases, the disorder may be provoked by hyperthyroidism and is relieved when the patient becomes euthyroid (Stern 1902, Cantonnet 1904, Gilbert 1972). Other dystonias have also been noted in hyperthyroidism (see Chapter 43). In some patients onset of torticollis has occurred after hyperthyroidism has been treated and a history of other forms of thyroid disease has been noted in some studies. Six out of 11 cases reported by Duane (1988) had some form of thyroid abnormality and 24% of 347 cases had a history of thyroid problems. However, no predominance of thyroid disease has been found in other large series (Chan et al. 1991, Rondot et al. 1991).

Abstinence in alcoholics has been occasionally associated with the appearance of permanent spasmodic torticollis, but the significance is uncertain (De Keyser 1991).

The onset is usually gradual, although occasionally it may be abrupt. Sudden onset was reported in 11% of Rondot et al.'s (1991) series, although 36% were said to progress rapidly. An insidious onset was noted in 72% of Patterson and Little's (1943) patients. In 68% the first symptom was a pulling or drawing sensation and in 29% it was a spasmodic jerking. This is similar to other series (Jankovic et al. 1991). When fully developed the involuntary movement may be tonic or consist of a series of clonic jerks, causing the head to deviate to one side (Table 38.3). It may be associated with lateral flexion, antecollis, or retrocollis (Fig. 38.5). Movements may become increasingly severe and persistant. They have been recorded as intermittent in two thirds and continuous in one third of patients presenting for consideration of surgical treatment (Sorensen and Hamby 1966). Such ‘spasmodic’ activity was divided into ‘head jerks’ and ‘neck spasms’ by Chan et al. (1991) and were present in 35% and 37% respectively. Although most series have had equal numbers of patients turn to left and right sides (Matthews et al. 1978, Setjskal and Tomanek 1981), in Patterson and Little's (1943) large series, 52% rotated to the left, while only 36% rotated to the right. Four percent of these patients had retrocollis, 3% antecollis, and in 5% the direction of deviation was uncertain. In Jankovic et al.'s(1991) 300 patients, 82% had horizontal rotation with equal proportions turning to each side, 29% had retrocollis, 25% anterocollis, 42% laterocollis, and two thirds had a mixture of positions. Chan et al. (1991) found only 19% had pure rotational torticollis, while this was present in 37% of Rondot et al.'s (1991) cases. Of the 220 patients in the latter series, only 3.8% and 1.1% had isolated laterocollis and retrocollis respectively, with the remainder showing a mixture of head tilts, especially lateral and/or backward movement combined with torticollis, which was present in over half of the whole series.

Although the direction of rotation is usually constant in any patient, in a few it may be variable. Thus, temporary remission may be followed by head turning to the opposite side (see under ‘Prognosis’) and occasionally it may even vary from spasm to spasm (Meares 1973, Matthews et al. 1978). Although the sternomastoid on the side opposite the direction of torticollis contracts, a number of other neck muscles invariably are involved. Thus, trapezius, splenius capitis, semispinalis capitus, and a number of other of the deep cervical muscles usually go into spasm (Patterson and Little 1943, Herz and Hoefer 1949, Matthews et al. 1978). As with other forms of dystonia, it may be aggravated by emotional stress and it disappears in sleep (Patterson and Little 1943, Meares 1973).

The antagonistic gesture or ‘geste antagoniste’ described in early French writings refers to the placement of a hand on the head or neck in an attempt to correct the deviation (Fig. 38.5). It has been claimed that mere light touch, or even making a gesture ‘as if to touch’, is sufficient to alleviate torsion (Wilson 1940). Some have therefore suggested that spasmodic torticollis (and perhaps other dystonias) might result from a disturbance of central processing of the afferent input conveying head position information at least in those who are sensitive to sensory stimuli (Hallett 1995, Karnath et al. 2000). In this context prolonged vibration was shown to improve torticollis in one study (Karnath et al. 2000). Other workers have interpreted the improvement with the ‘geste’ as being due to triggering of proprioceptive brainstem reflexes (Patterson and Little 1943) analogous to the relaxation of spasm in cervical muscles of decerebrate animals produced by pressing on the face (Barre 1929). Touching the face, chin, neck, temporal or parieto-occipital areas has been claimed to be effective (Patterson and Little 1943, Sorensen and Hamby 1966, Setjskal 1980). Pressure on the face or temporal area was effective in 83% of cases reported by Sorensen and Hamby (1966) and in a similar proportion of those investigation by Patterson and Little (1943). The latter, however, found that relief occurred more frequently when the hand was placed on the face and chin than on the neck, but that there was no difference in touching the side to which the chin had turned, or the opposite one. They reported, however, that pressure with the hand contralateral to the direction of torticollis was effective in 81% of cases, but in only 65% if the ipsilateral hand was used. Using electromyography Matthews et al. (1978) were unable to demonstrate consistent changes in cervical muscle activity with such manoeuvres. In a similar but more extensive study of 65 patients, Stejskel (1980) found that the hand used was always ipsilateral to the direction of the torticollis in 55% and always contralateral in 25%. Contralateral pressure was particularly used if there was a preponderance of tilting. Mere touch was ineffective and substantial pressure proportional to the severity of the torticollis had to be applied. In some cases scalp hair was pulled to keep the head straight. He interpreted these findings as attempts at mechanical correction without evidence of involvement of tonic neck reflexes. Other events have been reported to alter the intensity of torticollis including visual stimulation, caloric activation of the vestibular apparatus, and activity of the upper limbs (Podivinsky 1968). In particular, lifting a weight in the hand ipsilateral to the direction of the torticollis has been reported to aggravate spasm, whereas using the contralateral hand has been said to relieve it. Matthews et al. (1978), however, were unable to demonstrate any consistent effect using such manoeuvres.

Eventually the neck may be persistently held in an abnormal position. Even at this stage, however, it can usually be straightened. Sometimes contracture of sternomastoid and other muscles makes correction impossible. Hypertrophy of the sternomastoid contralateral to the direction of torticollis is frequent (Patterson and Little 1943). Cervical pain is common and becomes persistent in approximately half to three quarters of cases (Patterson and Little 1943, Chan et al. 1991, Jankovic et al. 1991)). Cervical spondylosis may be aggravated and possibly precipitated with resulting local or radicular pain. Occasionally spondylitic myelopathy with paraparesis or tetraparesis can occur (Walsh 1976, Waterston et al. 1989).

Dystonia may develop at other sites and occasionally torticollis may be the first manifestation of generalized primary dystonia (dystonia musculorum deformans and then genetic forms like DYT6 dystonia should be considered). In most patients, however, spread to other muscles is restricted to the head or upper limbs. Occasionally the trunk may be involved. Crouch (1976[a]) reported that 80% of patients with spasmodic torticollis had extranuchal dystonia. In 33% it was limited to the shoulder and in 47% it involved the trunk, arms, and legs. In 50% spasms involved the face. In Jankovic et al.'s (1991) study, there was elevation of the shoulder in 54%, scoliosis in 39%, and dystonia of the face, mouth, and jaw in 12%, 16%, and 12% respectively. Other workers, however, have not shown such a high incidence. Jahanshahi et al.(1990) found that 31.9% of cases of adult onset idiopathic spasmodic torticollis later developed dystonia elsewhere. Patterson and Little(1943) reported involvement of the upper limbs in 25%, back muscles in 9%, face in 7%, platsma in 5%, masseters in 2%, palate in 1%, and legs in 2%. It is uncertain, however, what proportion of their patients showed dystonia at multiple sites and what proportion were free of all additional movements. Approximately 13% of Matthews et al.'s (1978) patients showed additional dystonic movements. Only 3% of Marsden's (1976) cases with adult onset torticollis progressed to involve the trunk or upper limbs and none spread to the legs, although extension to cranial dystonia was somewhat more common. The reason for this gross discrepancy in the reported incidence of extranuchal dystonia is uncertain but probably different populations were studied with different criteria for diagnosis of spasmodic torticollis. Inclusion of patients with obvious generalized or symptomatic dystonia could have to some extent contaminated the populations described and there is lack of strict uniformity. For example, in the series of Rondot et al. (1991) 38% had other antecedent extrapyramidal features and the study of Jankovic et al. (1991) contained 6% of patients with tardive dystonia. Clinical experience suggests that those presenting with isolated spasmodic torticollis in adult life may develop involvement of adjacent segments, but dystonia at distal sites is unusual (Dauer et al. 1998).

Other extrapyramidal features may be present and as with extranuchal dystonia there has been great variation in reported incidence. In addition to tremor of the neck, tremor at other sites is common and was present in 32% of Jankovic et al.'s (1991) patients. Chan et al. (1991) noted hand tremor in 23% and Rondot et al. (1991) found postural tremor in 21%. Reduced arm swing may, however, be present and has been noted in more than half of 100 patients with adult onset primary cervical dystonia (Kagi et al. 2008). Crouch (1976[a]) found tremor in 87% of patients and parkinsonian features in 33%. Essential tremor, mask-like facies, and frank parkinsonism were each present in approximately 5% of Patterson and Little's (1943) cases. Matthews et al. (1978) noted tremor and Parkinson's disease in single patients in a small series of 30 cases. Increased tone in the upper limbs was reported in 3% of Patterson and Little's (1943) patients but was able to be elicited using sykinesis in 50% of cases investigated by Meares (1971[c]). Assessment of this latter abnormality is subjective and as controls were not used it seems likely to be an overestimate. As with extranuchal dystonia, usual clinical experience does not support a high incidence of other major extrapyramidal signs and many of the above features may be subtle.

Patterson and Little (1943) reported neurological abnormalities other than the spasmodic torticollis in 48% of patients, including cranial nerve signs in 35%. Pupillary abnormalities were present in 9%, nystagmus in 6%, seventh nerve paresis in 5%, and dysarthria in 3%. In addition 27% of patients were said to have an abnormality of deep or superficial reflexes, most commonly hyper-reflexia on the side to which the chin was rotated. These same authors also found cutaneous sensory abnormalities in a number of patients. Although there have been occasional reports of neurological abnormalities outside of the extrapyramidal system (Meares 1971[c], Matthews et al. 1978) these are infrequent and appear to be incidental findings. It thus seems unlikely that this high incidence of neurological abnormalities noted by Patterson and Little(1943) is a valid finding and they have not been noted in other large series (Jankovic et al. 1991).

Swallowing difficulties have not been a feature of most studies. In a group of 43 patients referred for ventral rhizotomy, and hence perhaps consisting of patients with severe symptoms, 34.9% had dysphagia and 51.2% of the whole group showed abnormalities of swallowing on videofluoroscopic evaluation (Riski et al. 1990). Ertekin et al. (2002) assessed 25 patients with cervical dystonia for features of dysphagia by electrophysiological methods. In his cohort, dysphagia was suspected in 36% of patients with cervical dystonia on the basis of clinical assessment. The incidence of dysphagia increased to 72% on electrophysiological evaluation of pharyngeal swallowing. Submental muscle electromyography and laryngeal relocation times were significantly prolonged and the triggering time to swallowing reflex was significantly delayed. Some abnormalities seen in cricopharyngeal sphincter muscle EMG indicated that the striated sphincter muscle may be hyperreflexive in some patients. The authors concluded that neurogenic dysphagia is more prominent and longer lasting than mechanical dysphagia, which was transient and varied from patient to patient. Although these electrophysiological methods were not suitable for detecting anatomical changes during swallowing, as in videofluoroscopic studies, observations supported the neurogenic cause of dysphagia in patients with any kind of cervical dystonia. Finally, swallowing problems may also be related to treatment in those patients receiving botulinum toxin injections. In addition to the physical signs, patients may develop psychological symptoms as the result of the disorder. Thus, Matthews et al.(1978) found that patients were frightened of being ridiculed and were reluctant to leave their homes. Depression is significantly more frequent than in controls with cervical spondylosis (see earlier in section on ‘Pathophysiological mechanisms’).

As well as these additional features, which patients with spasmodic torticollis may develop, and the incidence of dystonia in their pedigrees, there may be a family history of other neurological complaints. Thus, Jahanshahi et al. (1990) found 6.9% of cases had relatives with tremor and Rondot et al. (1991) noted this in 14.6%. The latter authors also found a 3.3% prevalence of Parkinson's disease and a 2% prevalence of tics in the families of patients, but they did not have any controls. The significance of such observations is uncertain and remains a matter of debate.

In 1893 Gowers stated ‘the prognosis must be grave in every developed case’. In reviewing the outcome of patients almost a century later, Matthews et al. (1978) concluded there was little reason to change this view. Meares (1971[c]) claimed that onset associated with pain or jerking of the head, rather than tonic deviation, was associated with a better prognosis, as was associated anxiety, neuroticism, and marked marital or sexual disturbance. Others have not been able to confirm these claims (Matthews et al. 1978) and it seems unlikely they are correct. The outcome has also been claimed to be better in both conservatively and surgically treated groups if there is pure head rotation without tilting (Stejskal 1975). Spontaneous remission has been reported in approximately 25% of cases (Meares 1971[d], Tibbetts 1971, Matthews et al. 1978). This may be partial or complete and usually commences early in the course of the disorder, particularly during the first year (Friedman and Fahn 1986). Such remissions commonly last weeks or months but sometimes persist for years. Only 2–3% of patients (Patterson and Little 1943, Matthews et al. 1978) have spontaneous ‘cures’, but as relapses occur after remissions of between 10 and 30 years it is difficult to regard patients as ever being completely cured. A proportion of patients, however, eventually improve and some may become free of movements, except under stress. Jayne et al. (1984) suggested prognosis was better than is often thought, with 38% of patients experiencing remission, and in 23% this was sustained with a median duration of 8 years. The median duration of torticollis before remission was 3 years. Lowenstein and Aminoff (1988) found 13% underwent complete remission and a further 33% had partial remission. They noted that patients not experiencing improvement tended to be older at onset of illness. Friedman and Fahn (1986), however, found that remission lasting longer than 1 year occurred in only 12% and seemed more frequent in younger patients. Chan et al. (1991) noted remission in only 9.8% of 266 patients, and again this was more likely to occur in younger subjects. Of such patients, however, 85% relapsed, 27% reporting a change in direction of head rotation.

It has been suggested that there are three phases to the disorder. During the first 5 years there is usually gradual deterioration, although this is the stage during which spontaneous remission may occur. During the second 5 years the disorder is usually static and following this there may be a little sustained improvement (Meares 1971[d], Matthews et al. (1978). Patterson and Little (1943) found torticollis prevented 10% of their patients from working and a further 50% were partially incapacitated for work. In 23% of Jankovic et al.'s (1991) patients, the disorder resulted in unemployment.

Routine haematology, plasma biochemistry, and CSF analysis show no abnormality apart from in those cases associated with thyroid disorder. Cervical spine X-rays may show cervical spondylosis (Patterson and Little 1943, West 1977). The findings on pneumoencephalography have been mentioned above under ‘Pathophysiological mechanisms’. CT and MRI brain scans do not usually show any structural change, but T2 values for the putamen and pallidum have been reported to be higher bilaterally in patients than controls (Schneider et al. 1994). The significance of this observation is uncertain. For other imaging findings also see earlier under ‘Anatomical pathology’. Rarely, basal ganglia lesions have been reported in apparently idiopathic adult onset spasmodic torticollis, but such cases are really examples of secondary dystonia (Molho and Factor 1993).

Functional imaging studies with a D2 receptor ligand found a non-significant trend towards increased uptake in the striatum contralateral to the torticollis (Leenders et al. 1993). Hierholzer et al. (1994) also reported significantly greater striatal binding contralateral to the direction of head turn in torticollis patients studied by single photon emission computed tomography (SPECT) using a D2 receptor ligand. The significance of these findings is unclear.

Regional cerebral glucose metabolism, as measured by position emission tomography utilizing 18-fluorodeoyglucose, in normal subjects shows significant correlation between the caudate, lentiform nucleus, and thalamus ipsilaterally. In addition, metabolic activity in each of these structures is correlated with the same region contralaterally. It has been reported that in spasmodic torticollis, relationships between the sides is maintained but that those between the thalamus and both the ipsilateral caudate and lentiform nuclei are reduced. It has been suggested this may result from disruption of pallidothalamic projections and possibly imply a disturbance in gamma aminobuteric acids (Stoessl et al. 1986).

Using transcranial sonography, structural changes with increased echogenicity have been detected in the lentiform nuclei, predominantly in the contralateral pallidum in patients with adult onset primary focal dystonia (Becker et al. 2001).

The electroencephalogram has been found to show minor non-specific abnormalities in approximately half the patients (Kaste et al. 1981). It has also been reported that clonic rhythmic jerking of the head may be associated with suppressed, asynchronous background activity, whereas tonically sustained torticollis does not have this effect (Podivinsky 1968). Patients with tonic deviation have also been said to have increased amplitude of late components of somatosensory cortical evoked responses which is greater over the hemisphere contralateral to the direction of torticollis (Podivinsky 1968). These neurophysiological findings are of uncertain significance.

A slow vertex negative potential, known as the contingent negative variation, can be recorded after pairs of stimuli which warn subjects to prepare for, and then execute, goal-directed movements. This contingent negative variation is abnormal in patients with spasmodic torticollis when performing head rotation, in that the late components are markedly attenuated. It is normal for finger movements (Kaji et al. 1995). It has been suggested this reflects a failure of neural activities programming head rotation, resulting in co-contraction of agonist and antagonist muscles.

EMG shows that activity involves many neck muscles bilaterally, even when the movement is a simple rotatory one. Both tonic and clonic patterns of activity can be distinguished, the latter sometimes being semirhythmic. Both may be seen in the same patients at different times during the course of the illness or they may even be mingled in the same contraction (Podivinsky 1968). Tijssen and collegues analyzed EMG activity from sternocleidomastoid and splenius capitus muscles in the frequency and time domains and found that the EMG activity in these two muscles was in phase in patients with torticollis but not in matched controls imitating the torticollis. This short-term synchronization was suggestive of an abnormal corticoreticular and corticospinal drive in patients with dystonic torticollis and this also provided a method to distinguish organic from psychogenic torticollis (Tijssen et al. 2000).

Electronystagmography has shown increased duration, amplitude, and velocity of nystagmus, with the slow phase in the direction of the torticollis (see Fig. 38.6) (Stejksal 1981, Bronstein and Rudge 1988, Diamond et al. 1988). Other workers have reported minor degrees of directional preponderance, but consistency has been less dramatic (Matthews et al. 1978). Overall, however, there seems little doubt that such asymmetries are genuine. Other evidence of disturbed vestibular function consists of spontaneous nystagmus in the dark and the presence of counter-rolling abnormalities (Diamond et al. 1988). The latter are seen as lack of sustained eye torsion at the extreme position, when subjects are rotated back and forth through 180 degrees in both the naso-occipital and earth horizontal axes. The significance of such findings has been disputed, with some workers suggesting they are secondary to abnormal head position (Huygen et al. 1989), while others argue against this (Stell et al. 1991). Persistence of such findings after correction of torticollis has been cited as evidence to support the latter position (Stell et al. 1991). Minor abnormalities in auditory functions also have been claimed with an increase in interpeak latencies between waves I and III and waves I and V of the brainstem auditory evoked responses ipsilateral to the head deviation (Drake 1988).

Using the same techniques described in Chapter 37 on ‘Idiopathic (Primary) Cranial Dystonias’, the R2 component of the blink reflex to a second stimulus can be shown to recover more rapidly than in normals (Fig. 38.7), suggesting disinhibition of brainstem interneurons (see earlier in ‘Pathophysiological mechanisms’). This can also be inferred from the fact that electrical stimulation of the supraorbital nerve, which results in two phases of reflex facilitation in the sternomastoid with an interpolated phase of inhibition, shows a decrease in the degree of inhibition (Nakashima et al. 1989). As also briefly discussed in Chapter 37, studies using transcranial magnetic stimulation suggest over-excitability and a reduction in motor inhibitory circuit activity/function evident at many levels of the nervous system, but most likely with its origins in the basal ganglia; and abnormal plasticity in dystonia (see Table 38.4).

Early attempts to control torticollis involved a variety of methods of restraint, including casts, collars, and metal bands (Fig. 38.8). These are generally ineffective and have been suggested to cause deterioration (Meares 1973), although this seems doubtful.

Many different types of drugs have been used to treat spasmodic torticollis (Table 38.5). As for blepharospasm botulinum toxin may be drug of choice; however with regard to oral medication similar to other forms of dystonia, it has been reported to respond to anticholinergic therapy. Tanner et al. (1979) noted improvement in six patients given parenteral scopolamine, whose effect was antagonized by the centrally acting cholinergic agent physostigmine. Lang et al. (1983) noticed only slight improvement in eight patients given parenteral atropine, benztropine, and chlorpheniramine. Only two patients responded by more than 20% and this was to chlorpheniramine. Improvement was usually associated with drowsiness. Using high dose anticholinergic therapy (trihexyphenidyl 6–59 mg/day or ethopropazine 100–800 mg/day) Fahn (1983) and Marsden et al. (1984) noted improvement in 17 out of 40 (42%) patients. This treatment is less effective in adults and in focal or segmental dystonia than in children or in generalized dystonia. Side effects, including drowsiness, confusion, blurred vision, and dry mouth, are more prominent in adults, limiting the dose able to be tolerated. At the present time anticholinergics are probably the most effective oral drugs available for the treatment of this disorder. However, botulinum toxin is the first choice. Nutt et al. (1984) did not find differences between centrally acting anticholinergics, peripheral anticholinergics, and placebo in patients with cranial dystonia (class III evidence). A retrospective class IV study of adult onset dystonia was carried out by Lang et al. (1982) which found no consistent benefit from anticholinergics in patients with adult onset focal dystonia and concluded that only a minority of patients with cranial dystonia respond to anticholinergics. For a Cochrane review see Albanese and colleagues (2006). The authors concluded that there is no proof of efficacy of anticholinergics in adults; therefore no recommendations could be made to guide prescribing.

A number of antidopaminergic agents have been tried. Shaw et al. (1972) found only one out six patients improved with haloperidol treatment. More enthusiastic results were reported by Crouch (1976[b]) with 14 out of 16 patients improving in a double blind cross over trial. Gilbert (1972) noted improvement in all seven patients treated with a combination of haloperidol and amantidine. In a double blind cross over trial West (1977) was unable to demonstrate any effect by this combination or by amantadine alone. It seems likely that only occasional patients respond to this combination or to haloperidol alone. The dopamine store depleting drug tetrabenazine may also produce relief, but the number of patients responding is disappointingly small (Shaw et al. 1972, Swash et al. 1972). The atypical neuroleptic clozapine is ineffective (Thiel et al. 1994). Albanese et al. (2006) in their Cochrane review concluded that there is lack of evidence to give recommendations for this type of drugs.

Dopaminergic drugs have also been used to treat spasmodic torticollis. l-dopa produced only very occasional improvement (Barrett et al. 1970, Shaw et al. 1972). Similarly the dopamine receptor agonists bromocriptine, lisuride, and apomorphine have only infrequently alleviated torticollis (Lees et al. 1976, Tolosa 1978, Juntunen et al. 1979, Quinn et al. 1985). The Cochrane recommendations (Albanese et al. 2006) state that ‘following a positive diagnostic trial with levodopa, chronic treatment with levodopa should be initiated and adjusted according to the clinical response’. However, this may raise the suspicion of dopa-responsive dystonia and further investigation should be considered (Chapter 36).

Modulation of GABA receptor activity has also been reported to be helpful. Benzodiazepines may increase the efficacy of GABA at its receptor and frequently produces slight alleviation of torticollis. Occasional patients show a more dramatic response (Ahmed and Meeran 1979, Francis 1983). Increased brain GABA levels may result from administration of l-glutamine, which is converted to GABA, and isoniazid which inhibits GABA breakdown. Administration of these two drugs, along with diazepam (and pyridoxine), improved spasmodic torticollis in seven out of 14 patients, but the degree of response and side effects were such that only two were maintained on long-term treatment (Korein et al. 1981). Euphoria and elevation of plasma creatinine, urea, and hepatic enzymes were common. This therapy should be regarded as experimental.

Lithium carbonate has occasionally been reported to improve spasmodic torticollis (Couper-Smartt 1973), as have l-tryptophan (Lal et al. 1981) and l-5 hydroxytryptophan (Mori et al. 1975) and recently mexelitene, particularly for painful torticollis and generalized dystonia (Lucetti et al. 2000). Other drugs that have been tested include nabilone (a cannabinoid receptor agonist) which did not show efficacy, lidocaine, diphenhydramine, tizanidine, and oestrogens (see Albanese et al. 2006).

Thus, the drug treatment of spasmodic torticollis is unsatisfactory. Only a small proportion of patients respond to most agents and improvement is unpredictable. A reasonable plan of action would be to begin with diazepam, but if improvement is insufficient an anticholinergic agent should be substituted, with only very gradual increments in dose. This probably offers the best chance of success, but side effects may limit the dose. If this is ineffective other agents should be tried. It is largely a matter of trial and error and the percentage of patients showing a satisfactory response to any agent will be small.

The use of botulinum A toxin in focal dystonia has been discussed in Chapter 37 on ‘Idiopathic (Primary) Cranial Dystonias’. It has also proved of major benefit in the treatment of spasmodic torticollis and is currently accepted as the most effective form of symptomatic treatment for this condition. In a randomized prospective double blind controlled study, botulinum toxin was found to be significantly more effective than trihexiphenidyl with less adverse effects (Brans et al. 1996). It is usual to determine clinically the muscles that are especially involved and to inject the two most active sites. Some authors have used electromyography to identify and aid injection into affected muscles (Dubinsky et al. 1991), whereas others (Anderson et al. 1992) have not. Sternomastoid and the contralateral splenius capitus are usually injected in purely rotational torticollis, but the trapezius may also need treating, especially if there is elevation of the shoulder (Fig. 38.9). More complex movements may necessitate administration of toxin at other sites. The beneficial effects usually last 2–4 months.

Beneficial effects have been reported in the majority of patients (Figure 38.10, Table 38.6). In an early double blind placebo controlled study Tsui et al. (1986) reported a favourable response in 18 out of 19 patients, while Stell et al. (1988) noted all of 10 patients had reduced severity scores on both clinical and ‘blind video’ ratings at 6 weeks after treatment. Mild dysphagia and dysphonia were found by some authors to be common after injections into the sternomastoid. Mild swallowing difficulty, mainly with dry solids, was noted in seven out of 10 of Stell et al.'s (1988) patients, while two severe cases occurred after giving the same total dose in half the volume of dilutent. Five of these 10 patients also had mild deepening and reduction in volume of voice. These complications were not noted by Tsui et al. (1986), who found side effects were more frequent in the saline injected placebo group and they consisted mainly of local discomfort.

Subsequently there have been a large number of reports of the effect of botulinum toxin injections in the treatment of torticollis (Table 38.7). About three quarters of the patients experience improvement in their movements and relief from pain. The treatment is also very effective in lessening the degree of muscular hypertrophy. In spite of this, the patients’ overall assessment of benefit is somewhat less with only about two thirds feeling they have been helped. Like all forms of practical procedures, however, effects are dependent on the exact technique used and better results have been reported by some authors (e.g. Anderson et al. 1992). With repeated injections, side effects are common but usually minor (Table 38.7). Anderson et al. (1992) found 84% of 107 patients, given a total of 483 treatments, experienced an adverse effect at some stage, with dysphagia, dry mouth, dysphonia, and troublesome neck weakness being the most common. Dysphagia, which was the most troublesome, was regarded as mild after 30%, moderate after 5%, and severe after 2% of the treatments. It occurred only after sternomastoid injections and was related to the dose of botulinum toxin used. The incidence of side effects, particularly dysphagia, was greater in this series than in most others and probably relates to a relatively higher dose being used. A difference in the relative potency of the toxins used may underlie this (Anderson et al. 1992). In many studies, especially those from North America, the incidence of dysphagia has been 0–11% (Tsui et al. 1986, Brin et al. 1987, Gelb et al. 1989, Greene et al. 1990, Jankovic and Schwartz 1990). The majority (94%) of Anderson et al.'s (1992) patients continued to derive benefit from repeated injections, but 4.6% exhibited diminishing or loss of effect. Of these latter patients 60% were shown to have developed neutralizing antibodies, whereas only 1% of those with continuing long-term response had these (in low titre). Failure to detect antibodies in many unresponsive patients is probably due to the rather insensitive (but highly specific) mouse lethality assay which is the method used for detection of antibodies (Borodic et al. 1996). Short intervals between treatments and high doses are suspected to be associated with development of antibodies (Greene et al. 1994, Jankovic and Schwartz 1995). New serotypes F and B which are being developed may be useful in patients with antibodies to botulinum toxin A. Brachial plexopathy may be a rare adverse effect and may be on the basis of an immune response (Glanzman et al. 1990, Sampio et al. 1993). Distant effects of botulinum toxin on neuromuscular transmission detected by single fibre EMG have been reported (Girlanda et al. 1992); however, only rarely are there clinical signs of generalized weakness following local injections (Bhatia et al. 1998).

The average latency between injection and onset of improvement is about 1 week. Botulinum toxin injections usually have to be repeated every 3–6 months and the duration of improvement is usually 6–14 weeks (Jankovic and Schwartz 1990, Anderson et al. 1992), but there is considerable individual variation.

The history of surgery for spasmodic torticollis dates back to at least 1641 when a German army surgeon, Isaac Minnius, resected the sternomastoid muscles. In 1812 Dupuytren performed the first closed tenotomy of this muscle (Finney and Hughson 1925). A variety of operations followed, including excision of part of the spinal accessory nerve (de Morgan 1866) and cervical nerve roots. Initially the anterior roots were divided unilaterally (Keen 1891) but subsequently anterior and posterior roots were sectioned bilaterally, along with the peripheral part of the spinal accessory nerve. In 1930 Dandy combined this with intraspinal section of the spinal accessory nerve. A variety of similar operations are performed but perhaps the most common involves bilateral anterior rhizotomy of the upper three or four cervical segments associated with partial or complete spinal accessory nerve section. The simplest procedure, peripheral spinal accessory nerve section alone, produces only limited benefit in a small number of patients (Walsh 1976). The reported results of other forms of surgery have shown considerable variation. In a comparison of operated and unoperated cases, Meares (1971[d]) reported that operated patients ‘fared considerably worse’ and sometimes showed severe disability which was not present in the other groups. Thus, weakness of neck muscles necessitated cervical support in some, and pain, stiffness, or persistent involuntary movements troubled others. More encouraging results were reported by Walsh (1976) and 26 out of 33 patients were said to have had a ‘satisfactory’ result, although there was one postoperative death and a significant incidence of residual neck weakness, dysphagia, and difficulty in elevating the arms above the shoulders. Other studies have shown approximately 90% of patients deriving some benefit, with approximately 55–65% having ‘excellent’ results (Sorensen and Hamby 1966, Fabinyi and Dutton 1980, Gauthier et al. 1988).

Methods of assessment, however, have often been poorly specified and although control of movements may be achieved, the extent of residual disability is sometimes not clear. Perhaps one of the most explicit reports is that of Hambey and Schiffer (1970). In 80 operated patients they had two postoperative deaths; 80% of patients were improved, but a half still had some head rotation and half still experienced neck pain. Cervical instability and swallowing difficulties were persistent in a third, but in most they were not regarded as particularly severe. More than a third required further surgery.

Using prior EMG to identify which muscles are responsible for the abnormal movement, combined with a trial of preoperative temporary nerve blocks, avulsion of a limited number of nerves is carried out. Typically this involves the spinal accessory nerve and the contralateral posterior roots of C1 and C2 and the posterior primary divisions (rami) of C3, C4, C5, and usually C6. In a series of 111 such operations, 14 patients had spinal accessory denervation, 11 had unilateral ramisectomy, 76 had a combination of these two procedures, and the remainder had multiple denervations preformed because of widespread abnormal muscle activation. The result was said to have been very good in 63 and excellent in 34 patients, with less wasting and better cervical movement and stability than with anterior rhizotomy (Bertrand and Molina-Negro 1988). In an expanded series of 260 cases, Bertrand (1993) reported total or marked relief of symptoms with preservation of normal or nearly normal movements in 88%, where such surgery was followed by early physiotherapy. Sequelae of the procedure include sensory loss in the distribution of the greater occipital nerve, parasthesiae, and tic-like pain. Other series have reported similar good results (Braun and Richter 1994). However, there is still incomplete information about the length of follow-up, lack of use standard rating scales, and no blinded evaluation of outcome. Although it is difficult to predict which patients may be best for this sort of surgery, there is an impression that those with simple rotational torticollis or a marked (past) beneficial response to botulinum toxin injections do best (Braun et al. 1995).

Munchau et al. (2001) undertook a prospective study of selective peripheral denervation in cervical dystonia patients with primary or secondary botulinum toxin treatment failure using independent standardized assessment. Of the 62 patients who were assessed, 22 (35.5%) were not offered surgery, most commonly because of widespread dystonia. Of the remaining 40 patients, 37 have so far had surgery, 31 of whom have been followed up for at least 1 year and 15 for 18 months after surgery (mean follow-up duration 16.7 months). Using the Toronto Western Spasmodic Torticollis Rating Scale (TWSTRS) global outcome score, 68% of patients derived functionally relevant improvement at 12 months after surgery. In the entire operated group, total TWSTRS scores were reduced by 30% at 6 and 12 months after surgery. The subscores for severity, disability, and pain were reduced by 20, 30, and 40%, respectively, at 6 months and 20, 40, and 30%, respectively, at 12 months. Pain, however, increased over time, which appeared to result from muscle reinnervation. TWSTRS scores were not significantly improved in the six patients with primary botulinum toxin treatment failure. Head tremor did not change. There was a significant improvement of body concept, perceived disfigurement, stigma, and quality of life in the 12 patients whose psychosocial function was assessed. Preoperative disability and restriction of head movement were negatively correlated and the initial response to botulinum toxin treatment positively correlated with global outcome score. Spread or deterioration of dystonia elsewhere in the body occurred in three patients, with unpleasant sensory symptoms in denervated posterior cervical segments occurring in 14. Ten patients developed mild to moderate dysphagia and two developed severe dysphagia. The authors concluded that this form of surgery may be an effective treatment for patients with secondary but probably not for those with primary dystonia, unresponsive to treatment. However, reinnervation is not infrequent and can compromise outcome. Although postoperative morbidity is low, there may by a risk of dysphagia.

Because results with nerve root and peripheral nerve surgery were unsatisfactory, stereotaxic thalamotomy was introduced. Cooper (1977) reported the results of bilateral ventrolateral thalamotomy in 160 patients. He found it necessary to make additional lesions in the ventro-postero-medial nucleus and the centromedian nuclei; 60% of patients improved, but pseudobulbar palsy with dysphonia developed in a fifth of patients. Others have also found that although bilateral thalamotomy may lessen or abolish movements in approximately half of cases, it is associated with an unacceptable complication rate (Meares 1971[d], Walsh 1976) (also see Chapter 35 under ‘Generalized Primary Dystonia’). Unilateral thalamotomy avoids the risk of disastrous pseudobulbar palsy but may be less effective. The lesion is made contralateral to the direction of chin rotation. In 1970 Hassler and Dieckman reported that unilateral thalamotomy could be made more effective by lesioning the internal segment of the oral part of the ventrolateral nucleus (VLo) and the adjacent field of Forel in pure rotatory torticollis or the internal part of VLo plus adjacent interstitio-thalamic fibres if there was associated lateral head tilt. Using these methods Dieckman (1976) found that 47% of 70 patients had complete relief of torticollis and a further 29% had a good result. The beneficial effects often took several months to appear and a tendency to neglect the contralateral limbs was present in 16%. Subsequent reports have been less enthusiastic and Laitinen and Vilkki (1977) found only two out of seven patients had long-lasting improvement and in none was the result regarded as good. After left-sided thalamotomy, receptive and expressive verbal efficiencies were impaired. Other workers have had more success with unilateral thalamotomy with the majority of patients showing some improvement (Bertrand et al. 1978, von Essen et al. 1980). In some cases, however, persisting torticollis has necessitated additional peripheral denervation (Bertrand et al. 1978). Some authors have reported a delay of between several months to several years before apparent benefit from unilateral thalamotomy (Walsh 1976, von Essen 1980). As many patients with long-standing spasmodic torticollis may show spontaneous improvement, the significance of these later recoveries is uncertain.

A small number of patients with retrocollis have reported to be improved by stereotaxic lesioning of the midbrain prestitial nucleus (Hassler et al. 1981). In general, the results of stereotactic surgery in torticollis so far have been disappointing. However, recently there has been considerable interest in the internal segment of the globus pallidus as a target for lesioning or deep brain stimulation dystonia for both general and focal dystonia. Deep brain stimulation of the internal globus pallidus (GPi) in a small number of patients with cervical dystonia has been shown to produce favourable benefit, helping the posture and giving pain relief (Islekel et al. 1999, Krauss et al. 1999) (see later).

Stimulation of the dorsal cervical spinal cord has been reported to improve spasmodic torticollis (Gildenberg 1981). Waltz (1981), using percutaneously inserted epidural electrodes in 26 patients, found 58% markedly or moderately improved and 8% mildly improved. These assessments, however, were not performed blind and when such evaluations have been carried out this procedure has been found to be ineffective (Goetz et al. 1988).

More recently deep brain stimulation has become available and this is an option for patients with cervical dystonia who do not benefit from conservative treatment including local botulinum toxin injections [see Krauss et al. (2007) for review]. In one study reported by Kiss et al. (2007) 10 patients with severe, chronic, medication-resistant cervical dystonia were enrolled in prospective, single blind, multicentre study assessing the efficacy and safety of bilateral globus pallidus internus deep brain stimulation. The disease severity as rated by the TWSTRS score improved from a mean of 14.7 before surgery to 8.4 at 12 months postoperatively. Disability and pain scores improved from 14.9 and 26.6 before surgery to 5.4 and 9.2 at 12 months, respectively. General health and physical functioning as well as depression scores also improved significantly. Complications were mild and reversible in four patients. Some changes were observed in neuropsychological tests, although these did not impact daily life or employment. The authors concluded that this form of surgery is efficacious and safe for patients with severe and prolonged cervical dystonia who have failed medical management. In the same year, Hung et al. (2007) reported follow-up data of 10 patients with severe cervical dystonia unresponsive to medical treatment who had undergone the same form of deep brain stimulation and who were followed for 32 ± 21 months. At last follow-up, the TWSTRS severity score improved by 54.8%, the TWSTRS disability score improved by 59.1%, and the TWSTRS pain score improved by 50.4%.

In his review, Krauss (2007) summarizes that ‘given the fact that other surgical treatment options such as selective peripheral denervation are available, it may be considered third-line treatment in most instances. Chronic bilateral pallidal stimulation improves dystonic posture and movements, pain caused by dystonia and disability related to dystonia.’ The preliminary data on long-term follow-up appear to confirm the beneficial effect in the majority of patients. Krauss concludes that pallidal deep brain stimulation may play a major role in the future.

However, overall, the position of surgery in the management of patients with spasmodic torticollis remains controversial. Although it is undoubtedly effective in diminishing involuntary movements, there is significant morbidity and even occasional mortality which should be considered when dealing with an essentially benign condition. It should be reserved for severe cases with distressing symptoms unrelieved by other forms of therapy. Spasmodic torticollis should have been present for at least several years, making spontaneous remission unlikely. Selective posterior ramisectomy and spinal accessory division remain an option, but require expert neurophysiological guidance and meticulous attention to detail.

There is a long history of psychiatric therapy for spasmodic torticollis and aversion treatment has been used for 150 years (Gilby 1825, Brierly 1967, Cleeland 1973). Usually the patient receives an electric shock if the head turns. Although isolated successes are reported, it is doubtful if this therapy has any effect on the disorder. Other forms of psychotherapy, including suggestion, relaxation, hypnosis, or systematic desensitization have been reported effective in small numbers of cases (Meares 1973). Lack of scientific evaluation makes assessment of these results difficult. One of the more successful series is that of Patterson (1943), in which 16 out of 41 patients were said to recover or be improved. The duration of follow-up, however, was not clear. On the whole, psychiatric therapy appears to have little effect on the movements, although it may help alleviate mental anguish resulting from the condition.

Biofeedback using auditory and visual display of the EMG activity of neck muscles resulted in improvement persisting at 3 months in 54% of 69 patients, but at least 10% subsequently regressed (Korein and Brundy et al. 1976). Seven developed nerve root syndromes during treatment, necessitating permanent discontinuation of therapy in four. Utilizing combined biofeedback and aversion therapy Cleeland (1973) claimed sustained improvement over an average follow-up period of 20 months in six out of 10 patients. In a retrospective study of 72 patients Jahanshahi and Marsden (1989) found the majority reported some improvement with relaxation therapies, some of which included biofeedback. In addition, a sizable minority had been helped by physical therapies, such as wearing a collar, physiotherapy, acupuncture, osteopathy, and chiropractice. Other forms of physical therapy, including transcutaneous electrical stimulation of neck muscles (Gildenberg 1981) and horizontal stretching of the skin of the face and scalp (Christensen 1991), have been reported to be beneficial. Critical evaluation of such treatments is lacking and overall they seem of limited value, although they may provide some patients with temporary relief. In practice, patients often end up having a trial of these modalities in conjunction with medication, botulinum injections, or surgery.

On the basis that spasmodic torticollis might involve an abnormality of the utricle or its central connection Svien and Cody (1969) performed iontophoresis of the middle ear. Initial results were encouraging, but subsequent studies have not confirmed this work and it seems unlikely that it is effective (Duane 1988).

Like other forms of dystonia, spasmodic torticollis has to be causing significant symptoms to justify treatment. If this is the case, a suggested plan of management is to try botulinum toxin injection treatment as the first choice. If this is not possible, for example due to unavailability of an experienced injecter or being economically unviable (as in some developing countries), or the patient is immune to the injections, a trial of drug therapy should be given, starting with an anticholinergic and gradually building up to a high dose if tolerated and required. If drug trials are unsatisfactory and symptoms are distressing, selective denervation, usually involving section of the spinal accessory nerves and the posterior cervical rami, may be justified or deep brain stimulation of the internal segment of the globus pallidus may be considered.