Introduction

Posttraumatic syringomyelia can cause neurological deterioration after spinal cord injury (SCI) and thus represents a potentially serious complication. Sensory disturbance, gradual loss of motor function, pain at or above the level of injury and increased spasticity are the characteristic and most common clinical signs.1, 2, 3, 4, 5, 6, 7 The reported incidences of posttraumatic syringomyelia vary greatly. The incidence of posttraumatic syringomyelia with clinical manifestation ranges from <1% to 7%.2, 3, 4, 5, 8, 9 However, the incidence in radiological investigations is considerably higher and ranges >50%.6, 10, 11 Clinically relevant posttraumatic syringomyelia does not occur frequently; nevertheless, progressive neurological deterioration in the wake of posttraumatic syringomyelia represents a serious complication with devastating consequences for the affected patients.

Posttraumatic syringomyelia has been observed more frequently in individuals with motor and sensory complete SCI compared with those with incomplete SCI.1, 2, 8 Furthermore, it has been reported that the onset of posttraumatic syringomyelia is earlier with older age at the time of SCI and with cervical and thoracic SCI compared with lumbar SCI.8 However, there is a lack of data concerning the relevant characteristics and risk factors of posttraumatic syringomyelia. On the basis of risk factors, screening procedures for the timely diagnosis of posttraumatic syringomyelia could be defined.

We have therefore investigated the characteristics of symptomatic posttraumatic syringomyelia in individuals with SCI and the effects of patient and syrinx characteristics on unfavorable syringomyelia presentation (early formation, cranial progression and elongation). The following hypotheses were tested: age at the time of SCI, injury level (that is, tetraplegic or paraplegic), injury severity (that is, motor and sensory complete or incomplete SCI), and syrinx location (that is, cervical or non-cervical) have a significant effect on early syrinx formation (that is, <5 years after SCI), syrinx length and syrinx extending cranial to the lesion.

Materials and methods

Setting

This investigation was carried out in a SCI rehabilitation center with 140 beds. All applicable institutional and governmental regulations concerning the ethical use of health-related data were followed during the course of this research.

Collected data

A 10-year time period (2002–2012) was evaluated. Individuals to be investigated were identified by searching our patient database with the terms ‘syringomyelia’, ‘syrinx’ and ‘hydromyelia’. Individuals with an etiology other than traumatic SCI and no magnetic resonance imaging of the affected area were excluded. Personal characteristics were collected from electronic patient records. The severity of SCI was classified using the American Spinal Injury Association Impairment Scale (AIS). The syrinx location and length were determined on T2-weighted magnetic resonance images of the spine in neutral position using the Phoenix PACS software (version 3.20.34233, Phoenix-PACS GmbH, Freiburg, Germany). Furthermore, the date of the diagnosis was collected.

Statistical analysis

The data were calculated as medians and 95% confidence intervals (CIs) if appropriate. The Chi-square or Fisher’s exact test was used to investigate the differences in observed frequencies between groups. Comparisons between independent samples were performed using Mann–Whitney U-test. The effects of age at the time of SCI, injury level (that is, tetraplegic or paraplegic), injury severity (that is, AIS A or AIS B–D) and syrinx location (that is, cervical or non-cervical) on early syrinx formation (that is, <5 years after SCI), syrinx length and syrinx extending cranial to the lesion were investigated using binary logistic regression analysis. The age was grouped using 15-year increments.12 The statistical analyses were performed using the SPSS software (version 18.0.3, IBM, Somers, NY, USA). A P-value of ⩽0.05 was considered significant.

Results

A total of 222 patients with symptomatic syringomyelia were identified. The data of 32 patients with non-traumatic SCI and 52 patients with myelomeningocele were excluded from analysis. Thus, the data of 138 patients were analyzed.

Patients’ characteristics, including gender, injury location and injury severity, are presented in Table 1. The majority of the patients (that is, 110/138, 79.7%) were male, and most patients (that is, 108/138, 78.3%) suffered from motor and sensory complete SCI (AIS A). The median age at the time of SCI was 25 years (95% CI 23–27 years, range 12–63 years). The symptoms associated with syringomyelia are presented in Table 2. Deterioration of motor and sensory function, pain and increased spasticity were the most commonly observed symptoms.

Table 1 Patients’ characteristics
Full size table
Table 2 Clinical symptoms associated with syringomyelia
Full size table

Syringomyelia was diagnosed a median 15.0 years (95% CI 10.6–16.7 years, range 0–40 years) after SCI had occurred at a median age of 42 years (95% CI 40–44 years, range 17–79 years). Syringomyelia was most commonly observed in the cervical–thoracic (that is, 58/138, 42.0%) or the thoracic spine (that is, 50/138, 36.2%). A total of 19 (13.8%) and 5 (3.6%) syrinx occurred in the cervical and thoraco-lumbar spine, respectively. In two patients, the syrinx stretched from the cervical to the lumbar spine. The median syrinx length was seven vertebral levels (95% CI 6–8 levels, range 1–19 levels). The direction of extension was cranial in 18 (14.4%) patients, caudal in 36 (28.8%) and both in 71 (56.8%) patients. The syrinxes were extending a median three (95% CI 3–4 levels, range 1–19 levels) and eight vertebral levels (95% CI 6–9 levels, range 2–18 levels) in the cranial and caudal direction, respectively.

Injury severity (P=0.035) and age (P=0.001) were significant predictors for the early (that is, <5 years after SCI) formation of syringomyelia (Table 3). The odds of early syrinx formation were increased more than nine times in patients with motor and sensory complete SCI (AIS A) compared with those with incomplete SCI (AIS B–D). Furthermore, the odds of early syrinx formation increased with age (Table 3). In patients aged >45 years, 53% experienced syrinx formation in <5 years after SCI compared with 15% in patients who were aged ⩽30 years (Table 4). Syringomyelia occurred significantly (P⩽0.002) earlier in older (>30 years) compared with younger patients (⩽30 years) (Table 5). In patients with motor and sensory complete SCI (AIS A), syringomyelia occurred significantly sooner (P=0.027) (Table 5) and more often earlier than 5 years after SCI (P=0.003) compared with patients with incomplete SCI (AIS B–D) (Table 4).

Table 3 Binary logistic regression results concerning the effects of age, injury level, injury severity and syrinx location on early syrinx formation, syrinx length and cranial syrinx progression
Full size table
Table 4 Occurrence of early syrinx formation for different age and injury severity groups
Full size table
Table 5 Time from spinal cord injury to syrinx formation for different age, injury severity, injury level and syrinx location groups
Full size table

Age, injury level, injury severity and syrinx location did not have any significant (P>0.9) effect on cranial syrinx progression or syrinx length (Table 3).

Discussion

To our knowledge, this is the first investigation of possible predictors of early syrinx formation, cranial syrinx progression and syrinx length. Injury severity was a significant predictor of early syrinx formation. The odds of early syrinx formation were increased more than nine times in patients with motor and sensory complete SCI compared with those with incomplete SCI. A trauma resulting in complete SCI may cause a greater impairment of the cerebrospinal fluid circulation compared with one resulting in incomplete SCI and thus facilitate the development of syringomyelia.13 Other investigators have also observed earlier syrinx formation in patients with complete SCI.8 However, statistical significance was not reached, most likely because of insufficient sample size and statistical power. There is a lack of data concerning the effect of injury severity on the time course of syringomyelia after traumatic SCI. Previous reports contain data concerning the incidence and occurrence of syringomyelia in patients with complete and incomplete SCI. The majority of reports support the hypothesis that syringomyelia is more common in patients with complete SCI.1, 3, 4, 8, 14 In the present study, approximately 75% of the patients with syringomyelia suffered from complete SCI compared with 52–82% in these previous reports. Few authors have reported similar or lower (that is, 50–33%) proportions of patients with complete compared with those with incomplete SCI among patients suffering from syringomyelia.2, 9, 10 In contrast to present and other results, Ko et al.2 have reported that only 49% of patients with early syringomyelia suffered from complete SCI.

Age at the time of SCI was the second significant predictor of early syrinx formation. Syringomyelia occurred a median 3.5 and 5 years after SCI in patients aged >45 and 31–45 years, respectively, compared with 17 years in patients aged ⩽30 years. Other authors have also reported a significantly earlier onset of syringomyelia with increasing age at the time of SCI.8 The mean time after SCI was approximately 6, 10 and 13 years in patients aged >40 years, 31–40 years and up to 30 years, respectively, in this previous report.8 In individuals aged >30 years, age-related degenerative changes of the spinal canal and cord tissue affecting cerebrospinal fluid circulation seem to accelerate the development of posttraumatic syringomyelia.8

In the present study, the level of SCI (that is, tetraplegia vs paraplegia) did not have any significant effect on early syrinx formation, syrinx length or cranial syrinx progression. This is in accordance with other investigations, in which there was no association between the injury level and the time to syrinx formation.2, 4, 8 In contrast, some authors have observed a higher percentage of paraplegics among patients affected by syringomyelia.9, 10 In the present study, the time from SCI to syrinx formation was approximately 16 years in tetraplegic patients, compared with 7–9 years in previous reports.4, 8 The elapsed time after SCI in paraplegic patients was approximately 14 and 5–11 years in the present and previous investigations,4, 8 respectively.

In the present study, none of the investigated factors was a significant predictor of cranial syrinx progression or syrinx length. An earlier report corroborates that the direction of syrinx progression does not depend on the severity of SCI.7 Factors associated with the mechanism of syringomyelia formation in the spinal cord may affect syrinx length and cranial progression.13

In our patients, syringomyelia had been diagnosed a median 15 years after SCI at a median age of 42 years. Interestingly, there is a broad agreement concerning the occurrence of syringomyelia in patients in their forties1, 2, 7, 10 and thus between one and one-and-a-half decades after SCI.1, 8, 9, 15 Syringomyelia was most commonly observed in the cervical–thoracic or the thoracic spine (that is, 78%) and stretched over a median seven vertebral levels. The reported average extent of syringomyelia ranges from 7 to 12 vertebral levels.1, 10 In the present investigation, syringomyelia had progressed cranially, caudally and in both directions in 14, 29 and 57% of the patients, respectively. The main difference to previous data7, 10 is the greater percentage of progression in both direction (57% vs 33%) and the smaller percentage of cranial progression (14% vs 47%) in our study cohort. The extent of progression was a median 3 and 8 vertebral levels in the cranial and caudal direction, respectively, compared with 4.5 and 2 levels reported previously.10

The retrospective and cross-sectional study design pertains to the limitations of the present study. The odds ratio values should be considered with caution because of the wide CIs. Possible predictors of syringomyelia such as spinal canal stenosis,1, 2, 10 posttraumatic kyphosis of the vertebral column1 or pathophysiological changes in the spinal cord13, 16 have not been investigated. However, the aim of the present study was to investigate readily available and easily assessable predictors of syringomyelia. Further prospective longitudinal studies are needed to confirm the present results and to investigate other possible predictors of syringomyelia.

Progressive neurological deterioration in the wake of posttraumatic syringomyelia represents a serious complication with devastating consequences for the affected patients. Health professionals entrusted with the care of SCI individuals should be aware of the possibility of syringomyelia in motor and sensory complete SCI patients in their fortie s, approximately 10 years after injury, presenting with symptoms such as sensory and motor function loss, pain or increased spasticity. Individuals with motor and sensory complete SCI aged >30 years are at risk of early syrinx formation and should be screened for syringomyelia (annual magnetic resonance imaging). The extent and frequency of screening procedures should be investigated in future studies.

In conclusion, syringomyelia mainly occurs in patients with motor and sensory complete SCI and involves the cervical spine in three of five patients. Patients with motor and sensory complete SCI and those aged >30 years have an increased risk of syrinx formation within 5 years after injury.

Data archiving

There were no data to deposit.