Results

There were no patients with an anomaly of the craniovertebral junction in our study. Degenerative changes on CT that was carried out in all 34 patients were either absent or minimal (20 patients) or moderate or severe (14 patients, table 2). Cervical spine MRI was performed on the same day as the CT and none showed evidence of a myelopathy.

We grouped patients according to the level and side of degenerative changes seen on CT. From a functional point of view the cervical spine can be divided into a more mobile upper segment (occipital-C1 and C1-C2 joints) and a lower less mobile segment (C2/C3 facet/uncovertebral joints and below). We grouped degenerative changes in patients accordingly. In all 14 patients with moderate to severe degenerative disease, changes occurred at C2/C3 and below. Five of these patients also had moderate or severe disease above the C2/C3 level, four of whom had significant changes in the C1 arch-peg articulation only (one patient with right laterocollis, two with left laterocollis, and one with left torticollis as their main component respectively), and one patient (with right laterocollis) had changes in the C1 arch-peg and right C1/C2 lateral mass articulation.

Unilateral degenerative changes at C2/C3 and below were significantly more likely to occur on the side of the main direction of spasmodic torticollis: patients with left torticollis or laterocollis were more likely to have left sided degenerative changes and vice versa (table3). The few patients with moderate or severe disease at the OC1C2 articulations precluded statistical analysis.

We compared the two groups of patients with none or minimal or moderate or severe degenerative changes (table 4). There was no significant difference in age, sex, duration of torticollis, overall severity, degree of disability, or amount of pain (before surgery) as determined by the TWSTRS. However, the duration without adequate treatment differed significantly between the groups. In other words, patients with moderate or severe changes either had a longer interval between onset of symptoms of spasmodic torticollis and start of effective treatment (usually BT injections) or an overall shorter lasting response to BT treatment. Among the moderate or severe change group there was a higher proportion of patients with primary failure of BT treatment (six out of 14) than in the none or minimal degenerative change group (3 out of 20), but the difference was not significant. However, the proportion of patients with primary BT treatment failure in the moderate or severe group was significantly greater than the expected proportion of primary BT failures in a non-selected group with spasmodic torticollis (less than 10%)35 (p=0.038, Fisher's exact test). This was not the case in the none or mild change group.

Clinically there were other distinguishing features in the moderate or severe group. Firstly, although there was no difference in overall severity of spasmodic torticollis before surgery between the two groups there was a striking difference in restriction of head movement. Patients in the moderate or severe group were significantly less able to turn or tilt the head to the side contralateral to the dystonic head position than those in the none or mild group (table 4). The combined rating (restriction of contralateral tilt+contralateral turn) was also significantly higher in the first group. Secondly, head tremor was more severe in the moderate or severe group (table 4).

Finally, in the none or mild group (n=20) there was a significant improvement of pain and severity, but not disability, 3 months after surgery. By contrast, in the moderate to severe group (n=14) none of the three outcome variables changed significantly after surgery (table5). Head mobility (contralateral turn or tilt) tended to improve after surgery in the none or mild group, but not in the moderate to severe group. Tremor did not change significantly after surgery in either group.

Discussion

The most striking finding of this study is the occurrence of moderate to severe degenerative changes in the C2/3 facets or below in 14 out of 34 patients with spasmodic torticollis. Osteoarthritis in the cervical spine presents with joint narrowing, obliteration of joint space, subchondral sclerosis, and osteophytosis on plain films, CT, and MRI36-39 and is strongly associated with age.40 41 The degree of degenerative changes encountered is more than would be expected in a patient group with a mean age of 50 years. Premature spondylosis has been documented in generalised dystonia11-26 but there is little information on patients with spasmodic torticollis, particularly with regard to primary or secondary failure of BT treatment.

Spondylotic changes in the cervical spine usually occur between C3 and C742 43 but not above C3. In normal aging the C5/6 uncovertebral joints are most often involved, followed by those at C6/7. Changes in the facet (zygapophyseal) joints usually occur at the C3/4, C4/5 and C5/6 levels. Overall, the C2/3 level is the least often affected, by contrast with our findings in patients with spasmodic torticollis. The pattern of degenerative changes we found in patients with spasmodic torticollis has previously been described by others11 26-28 and has also been noted in patients with generalised dystonia.11-18 20-22 24 26

Functional anatomy might help to explain the occurrence of degenerative changes predominantly in the upper cervical segments in patients with spasmodic torticollis. The C2/3 intervertebral joint is a transitional area situated between the upper cervical spine, where there is little flexion or extension and most rotation, and the lower cervical spine, where sagittal plane motion predominates (C5/6 having the greatest range) as well as some rotation.44 Nearly 50% of axial rotation occurs at the C1/C2 articulations.45 There is no or minimal axial rotation at the OC1 articulations and negligible lateral flexion at the C1/C2 articulations.46 47 It has been shown that rotation of the C1/C2 segment does not decrease with age but increases slightly to compensate for the overall decreased motion that occurs in the lower segments.48 It is therefore likely that the continuous abnormal and complex head movements that occur in spasmodic torticollis strain the upper cervical spine articulations more than the lower cervical spine, as there is less mobility at lower levels. Moreover, coupling of motion in the spine seems very important in spasmodic torticollis. For example, lateral flexion of the neck to the left will result in rotation of the cervical vertebrae to the left with the spinous process moving to the right.45 49 Hence, if patients with spasmodic torticollis have an abnormal posture for prolonged periods, this may result in more complex and asymmetric joint strain than would happen if the head spent more time in the neutral position.

Degenerative changes in the upper cervical spine tended to be unilateral. In patients with moderate to severe degenerative changes, the side of arthritic changes was significantly associated with the direction of spasmodic torticollis (for example, patients with left torticollis or left laterocollis had predominantly left sided degenerative disease and vice versa). Abnormal motion results in increased stress, particularly at joint margins, which is likely to be greater on the side towards which the head is tilted or rotated, accelerating degenerative changes on this side. Levineet al 14 studied plain radiographs of the cervical spine of patients with spasmodic torticollis and also found that spondylosis was greater on the side to which the head rotated.

Overall, it seems plausible that head position and sustained non-physiological stress on joints facilitates degenerative disease.50 However, degenerative changes were either absent or mild in 20 out of 34 patients in this selected series of patients with spasmodic torticollis. Surprisingly, neither age nor overall duration of spasmodic torticollis differed between the two groups. However, a significant difference was found with respect to duration of inadequate or insufficient treatment and occurrence of degenerative changes. Thus, in the group with moderate or severe changes the interval between onset of symptoms and the start of effective treatment was significantly longer than in the none or minimal group. Similarly, six out of the 14 patients in the moderate or severe group were primary BT injection failures and consequently never had adequate treatment. This proportion of primary BT injection failures was significantly higher than expected in a non-selected group of patients with spasmodic torticollis in which primary failure to BT injections is usually less than 10%.35 Effective treatment with BT may, probably due to improvement of head posture, protect patients from developing premature cervical spondylosis.

Although overall severity on the TWSTR scale did not differ between the two groups, patients in the moderate to severe group had some distinguishing clinical features. Firstly, using a validated rating scale34 their head tremor was significantly more severe. Tremulous head movements in spasmodic torticollis are often jerky and complex and could strain cervical joints considerably, particularly in the mobile transition zone of C2/3 between the upper and the lower cervical spine, which may promote the formation of osteophytes. Dystonic head tremor often responds less well to BT treatment51 and this also may put patients with head tremor at higher risk of developing degenerative changes. Secondly, head mobility was significantly more restricted in patients in the moderate to severe group. This finding is probably a consequence of underlying osteoarthritis that in advanced stages will inevitably restrict movement. On the other hand, it might also be argued that a fixed abnormal position may be a risk factor for the development of degenerative cervical spine disease. Sometimes patients develop a relatively fixed head posture shortly after onset of the disease—for example, in acute onset post-traumatic torticollis after minor trauma that usually starts several days and sometimes weeks after the trauma,52 in post-traumatic torticollis after cervical spine injury,53 54 or occasionally in tardive dystonia (personal observation). These patients may have an increased risk of developing premature osteoarthritis. Interestingly, three patients with severe degenerative changes had a history of acute onset post-traumatic torticollis, severe cervical spine injury, and tardive dystonia respectively, but only one patient with mild degenerative changes had a history of acute onset after trauma (table 4). However, these numbers are too small to draw firm conclusions.

Three out of 20 and two out of 14 patients in the none or minimal and moderate or severe group respectively gave a history of mild head or neck trauma several months before the onset of dystonic symptoms (table4), a proportion that is in keeping with reports from the literature.55 This has been referred to as delayed onset post-traumatic torticollis,52 but clinically these patients are indistinguishable from patients with spasmodic torticollis who do not report a history of trauma.52 A history of minor trauma itself does not seem to be a risk factor for developing cervical spondylosis.

Do the degenerative changes we have found in the upper cervical spine matter? Severe myelopathy or radiculopathy caused by degenerative changes in the upper cervical spine can occur in generalised dystonia.11-26 However, no patient in our series had myelopathy and clinically there was no detectable weakness in cervical myotomes.

On the other hand compression of the C1/C2 roots could cause sensory symptoms in the C2 dermatome. We did not find sensory loss in the C2 dermatome before surgery in any of our patients but occipital and nuchal pain was a common complaint. Pain is often encountered in spasmodic torticollis and often involves overactive dystonic muscles in the neck. Apart from one patient with severe degenerative changes who reported constant nagging occipital pain starting shortly after a traction manoeuvre which could be interpreted as C2 radiculopathy, we were unable to recognise a difference in the quality, location, or behaviour of pain in the two groups of patients. Also, their ratings on the TWSTR pain scale were similar before operation. However, our measures may not have been sensitive enough to discern subtle differences in the character of the pain between the groups. The fact that pain significantly improved after peripheral denervation in the none or minimal group but not in the moderate or severe degenerative change group could indicate that pain in the second group is indeed partly caused by osteoarthritis rather than purely by dystonic muscular spasms. However, as the number of patients operated on with severe degenerative disease is small, these results must be interpreted with caution. Recently Kutvonen et al,56 assessing patients with spasmodic torticollis for the presence, quality, and location of pain using different established pain measurements, found no association with degenerative changes of the cervical spine.

It seems that osteoarthritis compromises treatment outcome as it impairs neck mobility. The proportion of primary BT failures was high in patients with moderate or severe degenerative changes and surgical outcome was unsatisfactory. Three months after surgery, when possible reinnervation that could limit benefit would not be expected to have occurred, patients with no or mild degenerative changes had a significant improvement of severity of torticollis and associated pain. However, in patients with moderate or severe changes none of the outcome variables had improved. Whereas postoperative neck mobility tended to improve in patients with mild osteoarthritis, this was not the case in patients with moderate or severe changes.

In a study of 242 patients with spasmodic torticollis, Jankovic and Schwartz57 found that poor treatment response after BT injections was associated with a long duration of disease before starting BT injections, and pointed out that these non-responders often had restricted head mobility. These features match the characteristics of our patients with marked degenerative changes in the cervical spine, which implies that osteoarthritis may limit response to treatment with BT, and also with SPD. Detailed imaging assessment of degenerative changes in the cervical spine may assist in selecting patients likely to benefit from BT and SPD treatment. Although plain radiographs with lower associated radiation dosage, might prove equally useful in separating patients with no or mild disease from those with moderate or severe degenerative disease, CT is necessary to confidently exclude some craniovertebral anomalies and is our investigation of choice in patients with spasmodic torticollis considered for surgery, or with a history of fixed, post-traumatic, or sudden onset torticollis.

Conclusions

Patients with spasmodic torticollis are at risk of premature degenerative changes in the upper cervical spine that tend to be on the side to which patients turn or tilt their heads. One risk factor for developing degenerative changes was long duration of the disease without adequate treatment. In other words, effective treatment, particularly BT injections, seemed to be protective. Head tremor and severe restriction of head movements were common in patients with moderate or severe degenerative changes and the second factor is likely to be a consequence of degenerative changes. However, both factors may also accelerate the degenerative process. Overall, treatment response after SPD in the group of patients with marked degenerative changes was unsatisfactory. These patients do not seem to be good candidates for surgery.

Imaging of the cervical spine should be considered in patients with spasmodic torticollis who present with significantly restricted head mobility and do not adequately respond to treatment. To demonstrate degenerative changes in the spine plain films or CT can be used. Spinal MRI should only be considered when clinical signs of myelopathy are present. As part of our assessment for surgery we now routinely scan patients from the occiput to C4 using CT.