Introduction

Surgical procedures for syringomyelia associated with Chiari type I malformation (CMI) are mainly directed toward decompressing cerebellar tonsils at the foramen magnum and restoring CSF flow dynamics to normal. Despite extensive clinical investigations, considerable controversy about the technical details of the operation is still present [25, 37]. The most common technical modalities described in the literature include the following: (1) foramen magnum decompression (FMD) with incision of the external layer of the dura [16]; (2) FMD, dural opening, arachnoid preservation with or without duraplasty [27]; (3) FMD, dural opening, arachnoid dissection with or without coagulation or resection of cerebellar tonsils and duraplasty [17]. However, most studies lack enough patients to allow sound conclusions on how patients with syringomyelia-CMI complex should be operatively treated. In particular, the question about whether the arachnoid should be dissected after the osteo-dural decompression and the efficacy of arachnoid dissection on the post-operative shrinkage of syringomyelia has not been fully elucidated. Improved understanding of the treatment-related radiological outcome after foramen magnum decompression can help practitioners in the selection of the surgical procedure for syringomyelia associated with CMI.

The aim of our meta-analysis was to examine and compare the radiological outcome of syringomyelia in patients with CMI after different surgical strategies.

Methods

Literature search

A comprehensive literature search of PubMed, Ovid MEDLINE, and Ovid EMBASE was conducted for studies published from January, 1990 to May, 2018. PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analysis [23]) guidelines were followed. The key words and the detailed search strategy are reported in Table 1.

Table 1 Search syntax
Full size table

The inclusion criteria were the following: (1) studies reporting surgical series of adult patients (> 18 years old) with syringomyelia associated with CMI; (2) studies reporting series of patients treated with FMD with or without dural opening and with or without arachnoid dissection; (3) studies detailing post-operative complications and reporting radiological outcome of syringomyelia. Exclusion criteria were the following: (1) studies reporting series with less than 10 patients; (2) studies reporting patients with craniovertebral junction (CVJ) malformations or hydrocephalus; (3) review articles; (4) studies published in languages other than English. In cases of overlapping patient populations, only the series with the largest number of patients or most detailed data were included.

Two independent readers screened articles (F.C. and Y.A.) in their entirety to determine eligibility for inclusion. A third author solved discrepancies (D.D.C.).

Data collection

From each study, we extracted the following: (1) demographic data; (2) number of patients with CMI associated with syringomyelia; (3) preoperative clinical and radiological data; (4) post-operative complications (pseudomeningocele, CSF leakage, aseptic meningitis, and infection). Preoperative clinical symptoms were classified as follows: pain-related (tussive sub-occipital headache, non-tussive headache, neck pain), spinal cord-related (motor signs including paresis/weakness of at least one limb, spasticity, muscle atrophy, sensory symptoms including hypoesthesias, dysesthesias, paresthesias, peripheral nerve neuralgias), brainstem-related (including dizziness, diplopia, nystagmus), hiccup cerebellar-related (ataxia, gait impairment, vertigo, dysarthria), and lower cranial nerve-related (LCN). For each series, we collected site and extent of syrinx.

FMD was sub-grouped as follows: (1) FMD with splitting of the dura (opening of the outer layer only) (FMDds); (2) FMD with dural opening, arachnoid preservation with or without duraplasty (FMDdo); (3) FMD with dural opening, arachnoid exploration, and duraplasty (FMDae) (Fig. 1a–c). In the last sub-group, different surgical maneuvers were reported after arachnoid dissection (such as lysis of arachnoid bands/veils, cerebellar tonsil shrinkage, and plugging of the obex). A sub-analysis of this population was beyond the scope of our meta-analysis, and it was not reported.

Fig. 1
figure 1

Illustrations depicting the three modifications of foramen magnum decompression (FMD) investigated in this study. a FMD with splitting of the dura (opening of the outer layer). b FMD with durotomy (dural opening and arachnoid preservation) with or without duraplasty. c FMD with durotomy, arachnoid exploration/dissection, and duraplasty

Full size image

Radiological outcome was assessed at the last MRI follow-up and dichotomized as follows: better/resolved and unchanged/worse. Post-operative complications were dichotomized as follows: CSF-related (pseudomeningocele, CSF leak) and inflammatory/infection-related (aseptic meningitis, wound infections, meningitis).

Outcome measures

The primary objective of this meta-analysis was to assess the rate of post-operative radiological improvement of syringomyelia after FMD and to compare the radiological outcome according to the surgical approach used. The secondary objective was to evaluate the impact of potential confounders on the analyzed outcome. The analysis was performed for the following variables: rate of post-operative CSF-related complications, rate of post-operative infections, median follow-up (radiological or clinical), rate of patients with radiological evidence of tonsils below C1, rate of patients with syringomyelia limited to the cervical spine, mean duration of symptoms before surgery. In addition, the analysis of post-operative complications for each sub-group was conducted.

Quality scoring

A modified version of the Newcastle-Ottawa Scale (NOS) [38] was used for the quality assessment of the included studies. The details are reported in Table 2. Two authors performed the quality assessment independently (F.C. and Y.A.) and a third author solved discrepancies (D.D.C.). Studies rated with a NOS ≥ 6 were considered as “high quality.”

Table 2 Summary of the studies included in the systematic review, accounted for FMD surgical technique
Full size table

Statistical analysis

The Wald method was used to calculate confidence intervals for event rates. A two-tailed t test was used for continuous data and Fisher’s exact test for categorical variables. In order to assess the heterogeneity of the data, the Higgins index (I2) [15] was used, in which I2 > 50% suggests substantial heterogeneity. DerSimonian and Laird random effects model was subsequently applied [8]. The graphical representation of the meta-analysis was performed by forest plot. To verify the consistency of meta-analysis, the influence of each individual study was assessed by the sensitivity analysis (“leave-one-out” approach) [42], and subsequently funnel plot followed by Egger’s linear regression test were analyzed [10, 11]. Random effects meta-regression [2] was used to assess significant association between the effect size and the following moderators: surgical technique (FMDds, FMDdo, FMDae), patient’s mean age, CSF-related complications, post-operative infections, median follow-up, and study publication year. Differences were considered significant at p < 0.05. Statistical analyses were performed with SPSS version 23 (SPSS Inc. SPSS® Chicago, IL, USA), with ProMeta version3 (Internovi, Cesena, Italy) and OpenMeta[Analyst] (http://www.cebm. brown.edu/openmeta/).

Results

Literature review

A total of 659 potentially eligible articles were identified (Fig. 2). Thirty-one more articles were added by manual search or author’s knowledge. Among them, a total of 652 articles were excluded. Thus, a total of 38 full-text articles were further assessed for eligibility. Of these, 25 more articles were excluded because they did not meet our inclusion criteria. Therefore, a total of 13 articles finally met our inclusion criteria for further study [7, 12, 18, 26,27,28,29,30,31,32,33, 35, 43].

Fig. 2
figure 2

PRISMA diagram detailing the specifics of the systematic literature review

Full size image

Quality of studies

Overall, 10 studies (76.9%) were rated “high quality.” Six articles were based on a prospective design, and seven more studies were retrospective single center (details in online resource 1).

Demographic data

A total of 316 adult patients with CMI without history of previous FMD were included in our systematic review. Of these, 127 were females (M:F ratio 1:1.41), and the mean age was 40.72 years (range 18–70). A total of 276 patients with CMI associated with syringomyelia were identified and evaluated. The mean follow-up was 33.91 months (range 6–78). Studies included in our systematic review are summarized in Table 2.

Clinical and radiological data

Spinal cord-related symptoms were the most common clinical presentation (254 patients). A total of 125 patients (49.2%, 95% CI 43.1–55.3%) complained of sensory disturbances, and 98 patients (38.6%, 95% CI 32.8–44.7%) presented with motor dysfunction (10 studies). Pain-related symptoms were reported in 139 patients. Tussive sub-occipital headache (Valsalva-related) and non-tussive headache were reported by 55 patients (39.6%, 95% CI 31.8–47.9%), and neck pain was described in 66 patients (47.5%, 95% CI 39.4–55.7%). Brainstem compression, cerebellar, and LCN-related symptoms were reported in 21, 36, and 14 patients respectively. Preoperative radiological data were available in four articles (122 patients). Syrinx location was cervical in 30 patients (24.6% 95% CI 17.8–33.0%), cervicothoracic in 73 patients (59.8% 95% CI 50.0–68.1%), thoracic in 4 patients (3.3%, 95% CI 1.0–8.4%), and holocord in 15 patients (12.3%, 95% CI 7.5–19.4%). Its mean extension was 8.8 vertebral levels (range 5–9). Detailed data are presented in Table 3.

Table 3 Systematic review main data
Full size table

Surgical approaches

FMDds was reported in 46 patients (14.6%, 95% CI 11.0–18.9%). FMDdo was reported in six studies. In total, 160 patients (50.6%, 95% CI 45.1–56.1%) underwent FMDdo. Finally, 110 patients (34.8%, 95% CI 30.0–40.2%) underwent FMDae.

Radiological outcome

Overall, the rate of post-operative radiological improvement at last follow-up was 81.1% (95% CI 73.3–88.9%; p < 0.001; I2 = 71.4%). Furthermore, a lower rate of post-operative radiological syrinx shrinkage was observed after FMDds (21/39 = 55.6% 95% CI 40.5–70.8%, I2 = 0%), compared with FMDdo and FMDae (125/142 = 90%, 95% CI 85.1–94.8%, I2 = 0% and 69/95 = 79.8%, 95% CI 61.7–98%, I2 = 85.5%, respectively) (p = 0.003). Forest plots are shown in Fig. 3. On the other hand, the difference between extra-arachnoidal technique and arachnoid dissection was not significant in our analysis (p = 0.307).

Fig. 3
figure 3

Forest plot detailing sub-group analysis of pooled rate and 95% confidence intervals for the rate of syrinx radiological improvement after FMD. Yellow diamond sub-group pooled rate, blue diamond overall pooled rate. ds dural splitting, do dural opening, ae arachnoid exploration

Full size image

No significant association arose between the rate of post-operative radiological outcome CSF-related complications, post-operative infections, tonsillar prolapse below C1, and the rate of cervical syringomyelia. Similarly, median follow-up did not affect the effect size, whereas longer preoperative duration of symptoms was associated with a lower rate of radiological improvement of syringomyelia (Table 4). Scatter plots are available in online resource 2.

Table 4 Impact of potential confounders on the rate of radiological syrinx improvement ensuing FMD in random effects meta-regression
Full size table

Complications

Overall, CSF-related complications were reported in 19/297 patients (6.4% 95% CI 4.1–10%) (11 studies), and no significant relationship arose when considering surgical technique (FMDds, 1/45 = 2% 95% CI 0–12.6%; FMDdo, 14/146 = 9.6% 95% CI 5.7–15.6%; FMDae, 4/106 = 3.8% 95% CI 1.2–9.6%) (p = 0.11). The overall rate of post-operative infections was 10/270 = 3.7% (95% CI 1.9–6.8%) (10 studies) and was similar among sub-groups (FMDds, 2/44 = 4.5% 95% CI 0.4–16%; FMDdo, 6/118 = 5.1% 95% CI 2.12–10.9%; FMDae, 2/108 = 1.9% 95% CI 0.1–6.9%) (p = 0.44).

Study heterogeneity and publication bias

I2 was greater than 50% for the rate of radiological outcome (71.4%). When considering surgical technique, heterogeneity was substantial for the FMDae sub-group (roughly 85%), and it was null for FMDds and FMDdo populations. No individual study significantly influenced outcomes, and the funnel plot followed by Egger’s linear regression excluded the risk of publication bias (p = 0.13) (Fig. 4).

Fig. 4
figure 4

a Leave-one-out sensitivity analysis for radiological outcome analysis ensuing FMD. b Funnel plot detailing publication bias among the included studies. Asterisk indicates statistical significance of Egger’s linear regression test that excluded publication bias (p = 0.13)

Full size image

Discussion

FMD is widely accepted as the treatment of syringomyelia-CMI complex. Several details of the surgical procedure are under discussion and consensus on the ideal surgical treatment for this condition has not been reached. Pooling the results of 13 studies, our analysis provides representative data on the clinical and radiological outcome after FMD in patients with syringomyelia associated with CMI.

Post-operative changes of syringomyelia

The rate of post-operative reduction of syrinx size is quite variable in the literature ranging from 39 to 100% [25, 37]. The different surgical approaches and the heterogeneous populations reported in the clinical series may explain this high variability. In order to minimize the heterogeneity of our population, we did not select studies including patients with craniovertebral junction malformations and/or hydrocephalus. Our meta-analysis of 276 adult patients with CMI and syringomyelia found a post-operative reduction of syringomyelia in roughly 80% of cases with a mean radiological follow-up of 34 months (range 9–78). Among the analyzed confounders potentially affecting the rate of radiological shrinkage of syringomyelia, only the mean preoperative duration of symptoms was significant (p = 0.03). This result is in agreement with previous clinical studies and supports the concept that surgery should be offered early after diagnosis of syringomyelia [1, 5, 24]. One of the most relevant findings of this study is that patients undergoing FMD with splitting of the dura had significantly lower rates of syrinx reduction than patients treated with dural opening with or without arachnoid dissection (55.6% versus 79.8% and 90% [p = 0.003]). Dura splitting decompression was originally described by Isu et al. [16] in a small series of adult patients with syringomyelia associated with CMI. They reported a decrease in size of the syrinx in all patients and no complications. Subsequently, clinical series on decompression with incision of the outer dural layer in adults published rates ranging from 50 to 80% of syrinx reduction [7, 32]. In our analysis, a post-operative reduction of the syrinx after splitting of the dura was observed in 55.6% of patients (95% CI 40.5–70.8%) with low heterogeneity (I2 = 0%). The lower number of post-operative syrinx shrinkage suggests that splitting of the dura may not be sufficient to completely relieve the occlusion of the free pulsatile CSF pathways across the craniovertebral junction [14]. Series recommending dural opening with or without arachnoid dissection reported higher rates of syrinx reductions ranging from 53 to 100% [28, 39]. Our study highlighted that rates of syrinx reduction were high among both groups of FMD with the extra-arachnoidal technique (90%, 95% CI 85.1–94.8%, I2 = 0%) and arachnoid dissection (79.8%, 95% CI 61.7–98%, I2 = 85.5%) without significant difference (p = 0.307). This finding neatly questions the issue of the pathophysiological role of intradural pathology in the occurrence of syringomyelia in the setting of CMI. Advocates for dissecting the arachnoid argue that intradural pathology, such as arachnoid veils or web and medialized tonsils, can partially or completely occlude the CSF flow channels through the foramen of Magendie, potentially leading to syringomyelia formation [9]. Tubbs et al. [36] found arachnoid veils in 9 (11.2%) out of 80 patients with CMI with an associated syrinx and recommended exploration of the fourth ventricle to ensure free CSF flow. However, these authors recognized that the intradural pathology cannot fully explain syringomyelia formation because arachnoid veils were not encountered on exploration of the fourth ventricle in most of the patients with syringes. In a recent clinical study, Dlouhy et al. [9] found obstructing arachnoid veils in 59.5% of patients with CMI associated with syringomyelia and suggested that this condition obstructs CSF flow out of the fourth ventricle with resultant downward herniation and piston-like effect of the tonsils. Similarly, Klekamp [17] in a clinical study of 371 FMDs found that the grade of arachnoid pathology correlated with the rate of associated syringomyelia and recommended arachnoid opening in every decompression with the aim to restore the CSF pathways. However, Wetjen et al. [39], in a prospective study on 29 consecutive patients with syringomyelia-CMI complex operated on with the extra-arachnoidal technique, found that syringomyelia decreased in diameter and length during the radiological follow-up in all patients suggesting that the underlying pathophysiology can be reversed without addressing the intradural pathology. In fact, in a seminal anatomical study on the foramen of Magendie, Barr [4] found that cases of imperforate foramina of Luschka and cases of shelf-like midline membranous projections blocking the median aperture are common in otherwise healthy humans. According to these findings, the pathophysiological role of intradural pathology in syringomyelia formation remains elusive and the role of intradural exploration in the treatment of syringomyelia-CMI complex continues to be under discussion.

Pathogenesis of syringomyelia associated with Chiari type I malformation and surgical implications

The majority of the hypotheses dealing with the pathogenesis of syringomyelia associated with CMI supports the existence of an altered cranio-spinal CSF dynamics. With the “hydrodynamic theory” of syringomyelia, Gardner postulated that incomplete opening of the outlets of the fourth ventricle leads the CSF being forced into the central canal of the spinal cord with resultant syringomyelia formation [13]. Accordingly, in order to treat the intradural pathology, he proposed plugging the obex as the most physiological surgical option for syringomyelia in patients with CMI. Subsequently, Williams proposed that the existence of cranio-spinal CSF pressure dissociation in patients with CMI promotes the CSF flow from the fourth ventricle to the central canal [40]. Interestingly, Williams suggested to avoid plugging of the obex although he maintained a general agreement upon the necessity of intradural manipulation including arachnoid dissection and amputation of cerebellar tonsils [41]. Subsequently, many clinical and experimental studies suggested that the syrinx fluid could originate from the subarachnoid CSF entering into the spinal cord through the perivascular or interstitial spaces [3, 22, 25]. Oldfield et al. proposed that the mechanism for the development of syringomyelia lies in a piston-like effect of the herniated cerebellar tonsils on the partially enclosed spinal CSF with resultant increased subarachnoid pulse pressure [25]. According to this pathophysiological hypothesis, an extra-arachnoidal cranio-cervical decompression can effectively reverse the mechanism of syringomyelia formation and progression dismissing the role of intradural pathology. The concept that syringomyelia arises from accumulation of extracellular fluid coming from the spinal cord microcirculation originally suggested by Tannennerg in 1924 [34] and Liber and Lisa in 1937 [20] has been recently reproposed. Koyanagi and Houkin [19] hypothesized that in patients with CMI and syringomyelia, the reduced compliance of the posterior spinal veins due to the decreased compliance of the spinal subarachnoid space interferes with the absorption mechanism of the extracellular fluids through the intramedullary venous channels. The resultant enlargement of the central canal and interstitial edema promotes syringomyelia formation. Similarly to other hypotheses, also in the latter, the obstruction of the foramen magnum remains the initiating factor promoting syringomyelia formation and progression, whereas the role of intradural pathology remains poorly investigated.

Treatment-related complications

One of the main concerns of surgical treatment of syringomyelia associated with CMI is the risk of CSF-related complications, including CSF leakage, pseudomeningocele, aseptic meningitis, and hydrocephalus. Scanty information is available in the literature on surgical morbidity after FMD for CMI. In a recent retrospective study, Klekamp [17] reported a complication rate of 21.8% and the most common complication was CSF fistula. Previous meta-analyses reported a higher likelihood of CSF-related complications after FMD with duraplasty compared with FMD without duraplasty [6, 21]. Lin et al. [21] found a significant lower relative risk of overall complications in the group of FMD without duraplasty (RR, 0.78, 95% CI, 0.66–0.93; p < 0.05) compared with FMD with duraplasty, although no significant difference was found in the rate of wound infection between the two groups. The main limitation of these meta-analyses is the heterogeneity of the included population. In fact, duraplasty is usually performed in both FMD with extra-arachnoidal technique and arachnoid dissection and series exist in which extra-arachnoidal technique is accomplished without duraplasty. In our meta-analysis, which identified three different sub-groups of FMD, CSF-related complications occurred in 6.4% (95% CI 4.1–10%) of patients without significant differences among the surgical techniques. The occurrence of CSF leakage after FMD with incision of the external layer of the dura is the result of accidental opening of both the inner dural layer and the arachnoid [18]. Similarly, CSF-related complications after extra-arachnoidal technique are related to inadvertent arachnoidal lacerations and tears, as well as incompetent duraplasty [27].

Strength and limitations

Our study has several limitations. Series are often retrospective studies and small single-institution experiences. Because of the small number of cases, the comparison between sub-groups may not provide a comprehensive representation of outcomes difference among different surgical approaches. In addition, the series included in our study range was over a wide temporal period, and the surgical management of CMI changed in this time period. Furthermore, imaging technology has become more sensitive in detecting post-operative changes of syringomyelia. Due to the relatively short follow-up, our analysis is unable to comment on the long-term results of different techniques of FMD. Particularly, it is important to point out that it was impossible to assess the rate of repeat surgery for syringomyelia recurrence, due to insufficient data. In addition, we were not able to analyze the effect of arachnoid pathology on the radiological outcome due to insufficient data. Furthermore, three confounders (tonsillar prolapse below C1, rate of cervical syringomyelia, and median duration of symptoms) were available only in five studies, increasing the risk of selection bias. However, although retrospective data are low in quality, our meta-analysis is the best available evidence to assess the efficacy of different FMD techniques in syringomyelia resolution of patients with CMI.

Conclusions

Our meta-analysis found a post-operative reduction of syringomyelia in roughly 80% of adult patients after FMD for CMI. That rates of syrinx reduction did not differ significantly among the groups of decompression with the extra-arachnoidal technique and arachnoid dissection, whereas patients undergoing decompression with splitting of the dura had significantly lower rates of syrinx reduction. A large prospective, randomized study is required to conclusively answer the question on how the operation should be correctly tailored in adult patients with syringomyelia associated with CMI.