Background: Primary late-onset focal dystonias may spread over time to adjacent body regions, but differences in the risk of spread over time among the various focal forms and the influence of age at dystonia onset on the risk of spread are not well established.
Methods: Patients presenting with primary late-onset focal blepharospasm (BSP, n = 124), cervical dystonia (CD, n = 73) and focal hand dystonia (FHD, n = 24) with 10 years or more of disease duration (mean ± SD, 15.3 (SD 4.9) years) were included in the study. The relationship between demographic/clinical variables and spread of dystonia was assessed by Kaplan–Meier survival curves and Cox proportional hazard regression models.
Results: Patients starting with BSP, CD and FHD had similar age, sex and disease duration. Age at dystonia onset, age at initial spread and the risk of initial spread were significantly higher, whereas time elapsing from onset to initial spread was significantly lower in the BSP group than in those with onset in the neck or in the upper extremities. Conversely, these parameters were similar in the CD and FHD groups. The greater risk of spread in the BSP group was mainly evident in the first 5 years of history; thereafter, it declined and became similar to that of patients with CD/FHD. The difference in the risk of initial spread by site of onset was partly confounded by age at dystonia onset. Site of and age at dystonia onset, and age at first spread, were not significant predictors of the risk of a second spread.
Conclusion: This study adds new insights into the phenomenon of spread of primary late-onset focal dystonia and provides the framework for future studies aimed at an indepth investigation of the mechanism(s) of spread.
- focal dystonia
- risk factors
Statistics from Altmetric.com
Primary late-onset dystonia is often focal at onset but may spread over time to adjacent body regions.1 The few published studies on this topic have suggested that different sites of onset confer different risks of spread.2–5 These reports have, nevertheless, left some questions unanswered. First, the clinical observation for spreading referred to patients with relatively short follow-up, which was no more than 5 years in most cases.2–5 Whether the type and extent of spread differ in patients with different onset even after 5 years remains unknown. Second, previous studies failed to control the risk of spread in relation to age at onset of dystonia. Given that the age at onset differs among the various focal types of primary late-onset dystonia,1 age at onset may be a confounding factor for the risk of spread among the various focal dystonias. Finally, none of the previous studies specifically investigated factors potentially affecting the spread of primary late-onset dystonia to more than one body site.
In this multicentre clinical study, we assessed the spread of primary late-onset dystonia in a sample of patients presenting with blepharospasm (BSP), cervical dystonia (CD) and focal hand dystonia (FHD), all of whom had their disease for 10 years or more. This long disease duration allowed us to check for differences in the extent and body distribution of spread among the various focal forms over time and to evaluate the influence of age at dystonia onset on the risk of spread.
PATIENTS AND METHODS
Patients were selected among consecutive outpatients with late-onset primary dystonia who were periodically followed up at the movement disorders clinics of the neurological departments at the Universities of Genoa, Rome “La Sapienza”, Messina and Bari, Italy.
Inclusion criteria for this study were: a diagnosis of focal, segmental or multifocal dystonia with an identifiable specific site of dystonia onset, made by senior neurologists who specialised in movement disorders according to the published standard criteria;6 age at dystonia onset (defined as time of first symptom) >20 years; and duration of disease ⩾10 years. Exclusion criteria were neurological abnormalities in addition to dystonia, except tremors associated with dystonia;6 features suggesting dopa-responsive dystonia, paroxysmal dystonia and alcohol-responsive myoclonus-dystonia;1 a history of exposure to dopamine receptor-blocking agents within 6 months before the onset of dystonia, as well as other causes of secondary dystonia;7 features suggesting non-neurological (ophthalmological, orthopaedic, psychogenic) causes of dystonia;1 and lack of a specific site of dystonia onset identified by history.
Data were obtained from a common electronic medical records system that contained all patient information collected from 1999 to 2006. Collected information included history data obtained at the initial clinical evaluation and post-evaluation data. All patients were under botulinum toxin treatment (which did not seem to affect spread)8 and none of the patients reported neuroleptic drug intake after the onset of dystonia. For each body site affected by dystonia, the date of initial involvement and the date of spread to other body parts (approximated to 1 year) was obtained from the patient’s historical report, records from other physicians, when available, and our own observations. Affected body parts were categorised as eyelids, oromandibular region (mouth, jaw and tongue), larynx, neck, and upper and lower extremities.
We assessed the accuracy of the self-report of age at dystonia onset with a test–retest repeatability study, which involved a sample of case patients randomly retested 1 year (SD 2 months) after the initial physician office visit.
Statistical analysis was performed with a standard statistical package (STATA 8). Unless otherwise expressed, all data are means ± SD. For the test–retest study dealing with the continuous variable age at dystonia onset, the intraclass correlation coefficient (ICC) was calculated.9 The ICC value that was considered to indicate high repeatability ranged between 0.80 and 1.00.10 Differences between groups were examined using the chi-square test and a one-way ANOVA with Newman–Keuls post-hoc test. The relationship between the site of onset and spread was estimated by Kaplan–Meier survival curves and Cox proportional hazard regression models.11 For both methods, the endpoint was time to spread. Patients without spread were included in the survival functions for as long as the length of observation, and their data were censored beyond that time. The comparison between Kaplan–Meier survival curves was computed by log-rank test. The Cox analysis evaluated whether age, age at dystonia onset, sex and duration of disease had a confounding effect on the relationship between site of onset and risk of spread. For the purpose of the Cox analysis, Hazard ratios (HRs), two-sided 95% confidence intervals (CIs), and p values were calculated. Age, age at dystonia onset, and duration of disease were analysed as continuous variables, and gender was analysed as a categorical variable (1 for a woman, 0 for a man). A factor was considered to be an important confounding variable if its inclusion in the model changed the estimate of an already included variable by >15%.12 The accuracy of the Cox model was evaluated for the observed data, and the proportional hazards assumption relative to the variable site of onset was graphically assessed by plotting the log-logistic model. The statistical power was assessed by the equation to control type II error, as reported by Parmar and Machin13. For all statistical methods, p values <0.05 were considered to be significant.
The test–retest sample included 33 case patients (10 men and 23 women) aged 65.8 (SD 8.8) years, with 15.5 (SD 5.9) years of disease duration at the time of last examination. Gender distribution, age, disease duration and type of dystonia were similar to those of the overall sample (see below). These subjects showed high repeatability in recalling age at dystonia onset (ICC = 0.97; 95% CI, 0.93 to 1.0; p<0.001).
Demographic and clinical features of the study sample
A total of 229 patients presenting with BSP (n = 124), OMD (n = 6), CD (n = 73) and FHD (n = 26) met the eligibility criteria. Owing to the small number of patients starting with OMD, we analysed only the 223 patients whose dystonia started as BSP, CD and FHD. There were 155 women and 68 men aged 65.3 (SD 12.3) years. The mean age at dystonia onset was 49.3 (SD 14.1) years, and mean duration of disease was 15.3 (SD 4.9) years; this resulted from historical report and from the length of time patients were followed up in the clinic (mean prospective follow-up, 8.7 years; range, 3–14 years). During the overall follow-up period, dystonia spread to other sites in 85 patients (38%): in 65 of such patients (29%) it spread to one body part, and in 20 patients (9%) to two body parts. The chronological order of spread events was based on patients’ self-report in 31/85 patients, records from other doctors in 8/85, and our own findings in 46/85.
Stratifying by site of onset, age, gender and duration of disease produced similar results in the three groups presenting with BSP, CD and FHD, whereas age at onset and the frequency of spread were significantly higher in the BSP group (table 1). The duration of patients’ follow-up did not significantly differ between groups with different site of dystonia onset (data not shown). The groups with different site of onset differed significantly in age at first spread and time elapsing between onset and first spread (table 1). Post-hoc comparisons showed an older age at the time of spread and a shorter time to spread in the BSP group than in the other groups. Age at first spread and time elapsing from onset to first spread were comparable in the CD and FHD groups. The differences among the three groups in age at and time of spread to a second body part (second spread) failed to reach statistical significance, presumably owing to the small size of the samples (table 1).
In patients presenting with BSP, dystonia spread more frequently to the oromandibular region, whereas in patients presenting with CD it spread more frequently to the upper limb and in those with FHD to the neck (table 2). In the BSP group, dystonia tended to spread more rapidly to adjacent body sites (table 2).
Risk of initial spread
Kaplan–Meier survival curves suggested that patients presenting with BSP were significantly more likely to have spread than those with initial onset in the neck or in the upper limb (fig 1). The Cox analysis confirmed a significantly higher risk of spread in patients with BSP than in those with onset in the neck (HR = 2.6; 95% CI, 1.5 to 4.6; p<0.001) or in the upper extremities (HR = 2.4; 95% CI, 1.1 to 5.5; p = 0.045). No significant difference was found in the risk of spread between patients with onset in the neck and those with onset in the upper extremities (HR = 1.3; 95% CI, 0.53 to 3.1; p = 0.6).
Given the similar age at dystonia onset, age of spread and frequency of spread between patients with CD and FHD, these two groups were collapsed into a single group to gain power in the subsequent analysis versus the BSP group. The significantly greater risk of spread in the BSP group (table 3) remained statistically unchanged after adjusting for age, sex, referral centre and disease duration; adjusting for age at dystonia onset reduced the HR by >15% but the reduction nevertheless remained significant (table 3). Age at dystonia onset correlated with an increased risk of spread (HR = 1.04; 95% CI, 1.01 to 1.06; p = 0.001) but, after adjusting for site of onset, the association lacked significance (adjusted HR = 1.02; 95% CI, 0.99 to 1.04; p = 0.15). No significantly different risk of spread was associated with age, sex, duration of disease or referral centre (data not shown).
The rate of spread by site of onset varied over time (fig 1). In the BSP group, most spread events (45/58, 78%) developed within about 5 years after the initial onset of BSP; in patients starting with CD/FHD, 33% of spread events (9/27) occurred within about 5 years after the initial onset of dystonia and 67% of spread events (18/27) developed 5–15 years after dystonia onset. The difference between BSP and CD/FHD groups was statistically significant (Chi square: 15.6, p<0.001).
The Cox analysis restricted to the first 5 years of history disclosed an almost 4 times greater risk of spread in the BSP group than in the CD/FHD group (HR = 4 within about 5 years after the initial onset of BSP 95% CI, 2.1 to 9; p<0.001). The adjustment for age of dystonia onset reduced the HR by >15% (3.3; 95% CI, 1.5 to 7.4; p = 0.003) and the reduction remained significant. After the first 5 years of disease duration, the risk of spread was comparable in the BSP and CD/FHD groups (HR = 0.87; 95% CI, 0.42 to 1.8; p = 0.71). The study had an estimated >90% chance of detecting a two-fold change in the risk of spread with α = 0.05 (two-sided) for BSP versus CD/FHD combined, assuming 5 years of history as time zero.
In the BSP group, 36/45 spread events occurring during the first 5 years of history were to the oromandibular region, compared with 6/13 spread events occurring in the subsequent follow-up (Chi square test, p = 0.02). Conversely, in the CD/FHD group, a comparable number of spread events to cranial sites was observed in the first 5 years of history and thereafter (1/8 vs. 8/18, p = 0.2).
Risk of second spread
Of the 85 patients whose dystonia spread to at least one site, 20 (23%) had also spread to a second body site. A mean 4.3 (SD 4.1) years elapsed between initial spread and second spread. Patients who had spread to a second body site and those who did not had a similar age (68.4 (SD 9.4) vs. 67.1 (SD 10.8) years, p = 0.61), female preponderance (15/20 vs. 43/63, p = 0.7), age at dystonia onset (51.7 (SD 12.2) vs. 50.5 (SD 14.1) years, p = 0.56), age at first spread (57 (SD 9.4) vs. 56.8 (SD 12) years, p = 0.94), time elapsing from dystonia onset to first spread (6.1 (SD 6.9) vs. 6.7 (SD 8.3) years, p = 0.77) and duration of follow-up after initial spread (11 (SD 4.2) vs 10 (SD 4.8) years, p = 0.42).
The Cox analysis showed that patients whose dystonia started as BSP and those with onset in the neck or upper limb had a similar tendency for another event of spread to develop (HR = 1.01; 95% CI, 0.95 to 1.08; p = 0.7), assuming that the first spread was at baseline. This estimate remained unaffected by age, sex, age at dystonia onset, age at first spread and duration of follow-up after the first spread (data not shown). The study had an estimated >85% chance of detecting a two-fold change in the risk of second spread with α = 0.05 (two-sided) for BSP versus CD/FHD combined.
In this multicentre clinical series enrolling patients with primary late-onset dystonia who had dystonia for 10 years or more, patients whose dystonia started as BSP had a higher risk of spread than patients whose disease started as CD or FHD. This is partly in agreement with previous observations, particularly those by Weiss et al,4 who compared a large number of dystonia patients with different sites of onset in a retrospective study. In contrast to our study, the case series by Weiss et al4 was characterised by a shorter follow-up period, mainly restricting the validity of their findings to the first 5 years of disease duration. By studying patients with a longer disease duration (⩾10 years), we confirmed that most patients whose dystonia started as BSP had spread within 5 years from the onset. We also observed that in most patients whose disease began with CD or FHD, focal dystonia could spread later, within 15 years from onset. After the first 5 years of disease duration, the rate of spread slowed in patients with BSP and became similar to that of patients with CD/FHD.
Our analysis provided new information on the relationship between the site of dystonia onset, age of dystonia onset and risk of spread. When we adjusted the data for age at onset of dystonia, the difference in the risk of initial spread between the BSP and CD/FHD groups diminished, yet was not abolished. This effect was specific as the estimated survival function remained appreciably unchanged after adjusting for age, sex or duration of disease. Demographic and clinical features (including site of dystonia onset, age at dystonia onset and age at first spread) did not influence the risk of second spread. We specifically chose not to assess the effect of family history of dystonia on the risk of spread because family history data collection limited to patients’ report has proved inaccurate,14 and we could not directly examine patients’ relatives.
One possible explanation for the different spread between BSP and CD/FHD is that the mechanisms responsible for the spread of primary late-onset dystonia (that are currently unknown) vary by site of onset. Neurophysiological and imaging studies show that the various focal dystonias differ in some pathophysiological features and that others are shared.15 In our study sample, we detected the differential risk of spread according to the site of dystonia onset only in the first years of disease duration, not throughout follow-up. In the early years after onset, the faster rate of spread in patients with BSP was accounted for by spread to the oromandibular region. BSP and oromandibular dystonia (OMD) classically present at similar ages (5th to 6th decade), unlike the age at onset of extracranial dystonias (3rd and 4th decades),1 6 15 develop at the same time in a large proportion of cases (Meige’s syndrome, which was excluded from our study),1 6 and share neurophysiological abnormalities in the cranial district (such as a shortened cortical silent period in facial muscles,16 and abnormal blink reflex R2 component and masseter inhibitory reflex SP2 phase17 that are usually less evident in CD and FHD). In addition, physiological supranuclear control of the oromandibular muscles resembles that of eyelid muscles more than cervical and limb muscles.18 Overall, the close anatomical and pathophysiological features of BSP and OMD might account for the more frequent and rapid spread to the oromandibular region in patients with BSP. Another possibility is that patients whose dystonia starts with BSP, CD and FHD may be differently exposed to risk factors for OMD. Information on risk factors for OMD is limited. Anecdotal observations suggested a role for injury or surgical intervention at the facio-buccal area, or for dental abnormalities (including edentulism).19 20 Some of these putative risk factors (eg, edentulism) are age-related. If age-related risk factors alone are responsible for the differential risk of spread according to the site of onset, however, this would have been entirely confounded by age at dystonia onset, yet it was not.
Our study has limitations. First, this study assessed self-reported age at dystonia onset and timing of spread of dystonia to other body parts in a retrospective fashion. Although we showed that the age at dystonia onset is reliably described in patients with primary late-onset dystonia, the validity of the self-reported age at onset may be difficult to determine only on the basis of retrospective reports. The similar difficulty in ascertaining the validity of the self-reported date of spread notwithstanding, we directly assessed the date of spread in more than half of the patients. Of note, the duration of patients’ follow-up was similar between groups of patients with different site of dystonia onset. Second, our cases were ascertained from tertiary referral movement disorders centres and strictly selected on the basis of disease duration of 10 years or more. This criterion may have introduced a selection bias. Nevertheless, the demographic and clinical features of this series were similar to those of the general population of cases.1 4 15 The second part of our analysis entailed comparing patients starting with BSP with patients with onset in the neck or upper limb combined. The appropriateness of collapsing the CD and FHD groups into a single group was supported by the strict similarities of age at dystonia onset, age at and time to spread, and frequency of spread between the CD and FHD groups found in present and previous studies.4 Overall, we feel confident that our methods and procedures reduced the major bias inherent to this partly retrospective investigation. A lack of precision could also be excluded because the Cox analysis of both initial and second spread by site of dystonia onset achieved satisfactory study power.
In conclusion, this study showed a higher frequency of spread during the first 5 years after onset in patients whose dystonia starts as BSP than in those starting with CD or FHD. In addition to prior observations, our study showed that after 5 years of disease duration, the site of dystonia onset left the risk of initial spread and the risk of subsequent spread unchanged. The site of dystonia onset influenced the timing of spread since most patients starting with CD or FHD experienced spread 5–15 years after dystonia onset, whereas most patients who started with BSP tended to experience spread earlier in the course of the disease. The difference in the risk of initial spread by site of onset was partly confounded by age at dystonia onset. The findings from this long-term study may be useful for planning studies in the future that are aimed at indepth investigations of the mechanism responsible for spread of dystonia, and designing ad hoc clinical trials to evaluate possible protective therapies.
Competing interests: None declared.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.