Introduction

Over the past decades, due to the more frequent routine use of magnetic resonance imaging (MRI) in the diagnostic process and follow-up for back pain and spine injuries, even distinctive features are increasingly detected. The attribution of a T2-hyperintense medullary signal as a prominent central canal, hydro- and syringomyelia has been classified by Milhorat [40, 41]. However, diagnostic criteria and terminology are used inconsistently. Batnitzky [6] differentiated primary congenital and secondary acquired forms of syringomyelia due to trauma, infection, tumor, or vascular disturbances. The pathophysiological mechanism of congenital syringomyelia is explained by the absence of a perforation of the rhombic roof and formation of the foramen Magendie during fetal weeks 6 to 8 [21], resulting in a persistent patent central canal. Conversely, acquired syringomyelia has been linked to intermittent sharp increases in cerebrospinal fluid (CSF) pressure associated with venous pressure fluctuations as the underlying distending force [61]. An experimental modeling of a phase difference between the pressure pulse in the spinal subarachnoid space and the perivascular spaces suggests that mechanical perturbations caused by arachnoiditis exacerbate the phase-lag effect [15]. As soon as the intrinsic fluid storage capacity of the spinal cord is overloaded, medullary edema may develop, presenting as a hyperintense T2-weighted signal and referred to as the “pre-syrinx” state [20].

Following spinal cord injury (SCI), local ischemia, liquefaction of hematoma, and/or autolytic processes, as well as subarachnoid scars limit the CSF flow [17], thereby rendering the development of possible posttraumatic syringomyelia (PTS). In PTS, delayed progressive myelopathy develops often corresponding to spinal segments distant from the level of the original lesion. Besides CSF flow disturbances, posttraumatic kyphosis and the resulting spinal canal stenosis may promote the progression of PTS [1, 44].

Concerning preferred therapeutic strategies, in 2010, a consensus panel gave a strong recommendation for surgical intervention in the setting of motor neurological deterioration and a weak recommendation for spinal cord untethering with expansive duraplasty as the preferred first-line surgical technique [7]. Furthermore, they recommended no decompression at the time of initial injury to limit the future risk of syringomyelia, or for patients developing pain, sensory loss or for asymptomatic but expanding syrinx [7]. While for cervical spondylotic myelopathy, two randomized controlled trials (RCT) compare different surgical techniques [19, 22], no RCT, or, at least, a prospective observational study compares a non-operative to surgical treatment—neither in cervical myelopathy nor in PTS. Here, we present the results of a systematic literature search on PTS for treatment strategy, outcome, and complications.

Methods

Eligibility

All studies describing the treatment or reporting the effects of treatment of PTS from 1980 to 2020 were included in this review. Case reports including less than three individuals and animal studies were not included. Apart from this, no restrictions were placed on the study type (experimental or observational studies), or sample size.

Literature search and data extraction

The retrieval of studies was performed in PubMed using the combined filter and Medical Subject Headings (MeSH) term: (“spinal cord injuries” [Mesh] AND “syringomyelia” [Mesh]). Additionally, the Cochrane Central Register of Controlled Trials (CENTRAL), the Cochrane Database of Systematic Reviews, Web of Science, Scopus, and Google Scholar were searched for eligible studies. All records were screened based on title and abstract independently by two authors (FDS and AK) separately. In cases of discordance, the records were included in the full-text screening. Finally, the remaining records were evaluated by reading the full-text papers. All relevant characteristics (study type, sample size, level and severity of the injury, interval injury to symptoms, surgical technique, follow-up period, main findings) reported in the manuscripts were extracted into evidence tables. Due to a high level of heterogeneity of the studies and many insufficient study designs, no pooled effect estimates were calculated. Instead, a descriptive summary was carried out.

Risk of bias assessment

Since the Newcastle-Ottawa Scale (NOS) is only suitable for assessing the quality of case-control and prospective cohort studies, we used a risk-of-bias measurement instrument based on the NOS but also suitable for studies which cannot be subsumed under these gold standard observational study designs. A detailed description of this tool has been published by Dodoo-Schittko et al. [12]. All included studies were evaluated by two authors (AK and FDS). Subsequently, disparate ratings were discussed until consensus was reached.

Results

Systematic literature search

The electronic search revealed 599 scientific reports, and the PRISMA flow diagram of the screening process is presented in Fig. 1. After screening the title and abstract, 61 articles were included in the full-text screening process. One study was excluded because of overlapping data [54], and one because the full text was not available in English [10]. Finally, a total of 43 studies met the eligibility criteria. The extracted information of these studies is shown in Table 1. All cohorts/samples were independent, and we could not identify any overlap of included individuals. Overall, the sample sizes of the clinical studies detected by literature search ranged from case series of three up to studies including 600 individuals [2, 4, 5, 9, 11, 13, 14, 16, 18, 23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40, 42,43,44,45,46,47,48,49,50,51,52,53, 55, 56, 59, 60, 62]. Observational designs were applied in all studies. One study used an experimental design investigating the effect of shunt insertion on syrinx length and size in an animal model [8] and an observational study in dogs [3]. Both were excluded because they were not including human subjects.

Fig. 1
figure 1

PRISMA flow diagram of the screening process

Full size image
Table 1 Evidence table of the literature search
Full size table

Patients characteristics

The studies collect data from 1803 patients. The time interval from the spinal injury to the onset of symptoms varies between 42 and 264 months. Pain, motor, and sensory function compromise are the most frequent symptoms, while autonomic dysfunction is uncommon. Complete SCI preponderates (mean 63%) over incomplete (mean 37%). The most frequent locations of spinal injuries were the thoracic region (mean 48%) and the cervical region (mean 41%). The lumbar region (mean 10%) and medulla oblongata (mean < 1%) were hardly affected.

Clinical outcome

The mean follow-up period was 56 months in the observational studies. While 89% of patients were treated surgically, of the remaining 11% treated conservatively, only those of Ronen et al. [45] received a rehabilitation (n = 5). The summarized results of the surgical and conservative treatment are presented in Table 2. Following surgery for PTS, 43% of patients improved clinically, and the MRI findings in 75% of patients; 50% remained stable, 16% deteriorated, and 0.8% died peri-procedural. Sufficient information about complications of the performed surgical procedures is not available in all studies. Six studies do not describe complications, and in another six studies, no complications were observed. Out of the 403 complications following surgery (26%), a drain or valve dysfunction was most often reported (21%). A CSF leak was reported in 2.9%, a transient neurological deficit in 2.8%, a permanent neurological deficit in 2.5%, and wound complications were reported in 2% of patients. Other complications included venous thromboembolic event (N = 7), meningitis (N = 6), pneumocephalus (N = 3), air embolism (N = 2), cerebellar tonsillar descent (N = 1), and cardiac arrest (N = 1).

Table 2 Results of surgical and conservative treatment in posttraumatic syringomyelia
Full size table

A detailed analysis of surgical results concerning pain, sensory, motor, and autonomic function was performed by Vernon et al. in 27 patients [59], Lee et al. in 87 patients [37, 38], and Schaan et al. in 30 patients [47]. The results are presented in Table 3. Pain is improved in 43% of patients, sensory function in 49%, and motor function in 55%. On the other hand, pain is deteriorated in 15%, sensory function in 27%, motor function in 25%, and autonomic function in almost all cases.

Table 3 Detailed analysis of surgical results concerning pain, sensory, motor, and autonomic function
Full size table

The type of surgical procedure on PTS was specified in 866 patients and is presented in Table 4. Arachnoid lysis was the procedure that was performed most often (N = 418; 48%), followed by various techniques of drain placement (N = 267; 31%). Procedures that were performed less frequently were cord transection (N = 51; 5.9%), syringostomy (N = 49; 5.7%), duraplasty (N = 41; 4.7%), a combination of arachnoid lysis and drain (N = 30; 3.5%), decompression alone (N = 5; 0.6%), or shunt alone (N = 5; 0.6%).

Table 4 Detailed analysis of results of different surgical techniques in the treatment of posttraumatic syringomyelia
Full size table

Four studies directly compared the results of different treatment regimes in separate cohorts. The allocation process was either based on clinical findings [37, 38] or not described [2, 47], and the time point of outcome assessment is not always specified. In two separate studies, Lee et al. performed arachnoid lysis (N = 33), drain placement (N = 33), and a combination of both (N = 13). While the failure of drain placement (N = 5; 15%) is higher than of arachnoid lysis (N = 2; 6%), the incidence of procedure-related neurological deficits (9%: arachnoid lysis N = 3; drain N = 3) and the overall improvement is comparable (pain 30%: each N = 11; sensory 21%: each N = 7; motor 30–33% arachnoid lysis N = 10, drain N = 11). The results reported by Aghakhani et al. postoperatively were worse for drain placement (N = 15; improvement 0%; deterioration 53%) than for arachnolysis (N = 19: improvement 16%; deterioration 5%) but improved over time [2]. A combination of both techniques was performed in a minority of patients and was associated with a considerable morbidity (53–75%) [2, 37, 38]. Schaan et al. compared drain placement (N = 18: improvement 33–38%; deterioration 13–31%) and duraplasty (N = 12; improvement 45–100%; deterioration 0–18%) [47].

The results of the less-often performed procedures are better, although based on small sample sizes: cord incision or transection resulted in an improvement of 50 to 100% (N = 8) [58], syringostomy resulted in an improvement of 47% to 71% (N = 19) [58], and duraplasty resulted in an improvement of 33% to 100% (N = 12) [47].

Without a surgical treatment—i.e., following a “conservative” treatment—2% of patients improved and 88% remained stable, while 10% of patients deteriorated (total N = 198; Table 2) [4, 5, 11, 14, 25, 32,33,34, 40, 45, 46, 49].

One experimental study including six patients (male, 30–50 years) into a phase 2 trial injecting autologous mesenchymal stromal cells into the syrinx of PTS patients was not included in the above-mentioned analysis because it is a novel therapeutic approach [57]. The time interval between SCI and treatment varied from 5.8 to 27.7 years, and irrespective of syrinx size, 300 × 106 autologous expanded mesenchymal stromal cells, supported in autologous plasma were administered into the syrinx. The patients were followed up for 6 months, and the authors report in all patients variable improvement in clinical scales, mainly in the scales related to sphincter dysfunction and neuropathic pain [57].

Discussion

Over the past four decades, PTS is diagnosed more frequently and surgical techniques became more elaborate. Consequently, in a review paper (2010), Bonfield et al. presented recommendations of a consensus panel for surgical intervention in the setting of motor neurological deterioration and for spinal cord untethering with duraplasty as the preferred surgical technique [7]. Interestingly, they recommended against the direct decompression at the time of initial injury as well as against surgical interventions for patients developing pain, sensory loss, or for asymptomatic but radiologically expanding syrinx [7]. By 2020, still, no prospective study is available comparing non-operative and surgical treatment—neither in cervical myelopathy nor in PTS.

Here, we present the results of a systematic bibliographic literature search on PTS for treatment strategy, outcome, and complications. The risk of bias assessment revealed a high or unclear risk of selection bias in all studies. In addition, the risk of information bias was present in many studies, notably in the assessment of patient-reported outcome measurements. This is because of the use of non-validated measurement instruments. The present literature review reveals that 89% of the included 1803 patients were treated surgically. This fact can be probably attributed to a publication bias. Nevertheless, 12 studies including 198 PTS patients treated conservatively have been published [4, 5, 11, 14, 25, 32,33,34, 40, 45, 46, 49]. In contrast with the expert recommendations of 2010 [7], we also appreciate the results of these conservatively treated PTS patients.

Effect of intervention on specific symptoms

The 2010 recommendations for surgical intervention in PTS advocate spinal cord untethering with duraplasty in the setting of motor neurological deterioration but against surgery for pain and sensory deterioration [7]. When we analyzed the outcome of surgery for PTS concerning specific neurological functions (Table 3), motor and sensory dysfunction responded better than pain to an intervention (improved motor 55%, sensory 49%, pain 43%) but were also at a higher risk for deterioration (motor 25%, sensory 27%, pain 15%) while the autonomic function was at a high risk for deterioration [37, 38, 47, 59]. Hence, although it is common practice in neurosurgery to prioritize motor symptoms in decision making for surgery and in guidelines, the recommendation by Bonfield et al. do not necessarily reflect the common practice in the existing literature on PTS.

The decision as to if, when, and what type of treatment to offer to a patient with PTS has changed over the past four decades. During the 1980s and early 1990s, surgical intervention including syringostomy and syrinx drainage was the preferred treatment option [58, 60]. Drainage complications resulted in the preference for the reconstruction of the subarachnoid space [50] or even a conservative treatment [5, 14, 33, 45, 49]. Furthermore, one has to keep in mind that both, improvement as well as an arrest of deterioration, may constitute the goal of treatment.

Effect of type of surgical procedure on the outcome

The type of surgical procedure was specified in almost 900 patients (Table 4). Forty-eight percent of authors performed untethering, another 31% various techniques of drain placement, and 4% combined both procedures. Cord transection, syringostomy, or duraplasty were performed by around 5%, and less than 1% applied decompression or shunt alone. While no surgical technique for PTS provides substantially superior results, any type of drain placement is associated with a failure rate of up to 20% [37, 38, 47]. The reason for the weak recommendation of the 2010 consensus panel for spinal cord untethering with expansive duraplasty as the preferred first-line surgical technique is not obvious [7]. From a pathophysiological point of view, an individualized therapeutic approach would be desirable based on the patient’s history, symptoms, and radiological findings. After the detection of local alterations in CSF flow, the restoration of physiological flow patterns should be the first goal followed by draining CSF trapped in cysts. Duraplasty may aid in the creation of extra CSF space and avoid new scar formation. Cell transplantation therapies in PTS are under investigation, although their relevance has not been confirmed yet [57, 63].

Time interval to surgical intervention

Experimental evidence indicates that mechanical perturbations of arachnoiditis form the basis of syrinx development [15], and medullary edema in MRI can be interpreted as a “pre-syrinx “[20] as suggested by Lyons et al. in the pre-MRI area [40]. These pathophysiological considerations may stress the relevance of an early intervention to divert CSF if the spinal canal stenosis is not absolute [13, 29, 40, 53]. Irrespectively, it is self-evident that the spinal stabilization and re-alignment of fractures preventing medullar compression and kyphosis is mandatory [2, 44, 49]. Interestingly, the 2010 consensus committee recommended against the direct decompression at the time of initial injury as well as against surgical interventions for patients developing sensory loss, a pain syndrome, or for asymptomatic but expanding syrinx [7].

Some authors do not support the use of a surgical intervention in PTS, even in patients with progressive neurological deterioration [45]. Our analysis suggests that a conservative treatment as reported by some authors [4, 5, 11, 14, 25, 33, 34, 40, 45, 49] may be an alternative to surgical procedures flawed by complications. However, because of the observational design of all studies, we cannot exclude that our analysis is biased by a crossover of patients who were initially treated conservatively and were referred to a subsequent surgical treatment because of clinical deterioration. Furthermore, it is important to keep in mind that scar formation itself without PTS may result in neurological deterioration.

Conclusion

Here, we present the analysis of a systematic literature search on therapeutic options for PTS over four decades. The outcome of conservative and surgical treatment is not directly comparable because of the exclusively observational study design with the subsequent selection bias and cross-over. While a satisfying outcome defined as either an improved or stable situation is identical (conservative 85%; surgery 88%), the reduction of deterioration from 15.5% without surgery to 9.1% with surgery is accompanied by a 0.33% surgery-related mortality and 23% complications. The evidence of the efficacy of the different treatment modalities is very low mainly resulting from the application of observational study designs with a consistently high risk of selection bias. This points to the necessity of additional research using appropriate study designs to reveal the causal relationship between treatment and outcome. However, concerning the existing literature, there is no satisfactory standard treatment for syringomyelia even diagnosing PTS early in its evolution. Hence, PTS remains a neurosurgical challenge even diagnosed early in its course.