Phenotypes and prognostic factors of syringomyelia in single-center patients with Chiari I malformation: moniliform type as a special configuration

Patient selection

From January 2017 and December 2020, a consecutive series of CM-Ⅰ patients with syringomyelia were retrospectively reviewed in a single institution. Clinical records and radiologic data in this study were retrospectively analyzed from a prospectively maintained syringomyelia database. According to screening, a total of 202 pediatric and adult CM-Ⅰ patients with syringomyelia who underwent PFD finally met the inclusion criteria of our study.

Inclusion criteria: (1) CM-Ⅰ with syringomyelia, with MRI showing cerebellar tonsils more than 5 mm below the foramen magnum; (2) patients underwent PFD with duraplasty in our center.

Exclusion criteria: (1) patients with atlantoaxial dislocation, basilar invagination, congenital vertebral anomalies, degenerative cervical spondylosis, tumor, trauma, myelomeningocele or tethered cord; (2) secondary CM-Ⅰ due to hydrocephalus or intracranial space-occupying lesions; (3) a history of craniovertebral junction surgery with cervical or occipital fusion and instrumentation; (4) incomplete clinical data or lost follow-up after surgery.

The local ethics committees in our institution approved this study, which was conducted by the ethical standards of the 1964 Declaration of Helsinki. We obtained consent from all patients participating in this study.

Clinical evaluation

Clinical symptoms were divided into general symptoms, spinal cord-related symptoms such as suspended sensory disorder, amyotrophy and weakness of limbs, and other symptoms such as cerebellar syndrome and cranial nerve dysfunction. Some patients had multiple clinical symptoms at the same time. The Chiari Severity Index (CSI),[22] a preoperative grading system proposed by Greenberg et al. was used to assess preoperative status for each patient. The specific data were summarized in Table 1.

Radiographic evaluation

Measurements for preoperative X-ray were made using the CAD Canvas X (Version 20.0; Canvas GFX, Boston, USA), a computer-aided drafting software. All relative radiologic measurements were repeated at least three times, independently by one surgeon and another radiologist. We could determine whether there was obvious scoliosis (Cobb angle of main curve ≥ 10˚ in the coronal X-ray). The value of cervical Cobb angle in the sagittal plane at the range of -5˚~5˚ was defined as the straight cervical spine, and the positive value larger than this range was defined as the lordosis; the negative value lower as the kyphosis.

RadiAnt DICOM Viewer software (Version 4.6.9, Medixant, Poznan, Poland) was used to view and measure MRI images. A routine MRI examination was taken within one week before surgery. Postoperative MRI was conducted regularly on a 1 month, 3 months (± 1 month), 6 months (± 3 month), 1 year, 1.5 years basis, and later every year after decompression surgery until evidence for effective syrinx resolution was found. The following indicators were measured at the T2-weighted median sagittal and transverse positions of MRI: (1) Maximal syrinx/cord (S/C) ratio: the ratio of syrinx diameter to the spinal cord diameter at the same level. (2) Syrinx length: the number of vertebral segments spanned by the syrinx. (3) A radiographic improvement rate in the maximal S/C ratio or syrinx length was calculated using the following formula: (preoperative value-postoperative value)/preoperative value. Furthermore, we assessed the following angles that related to the CVJ region and posterior cranial fossa (PCF): basal angle, the angle of the tentorium cerebelli to the Twining’s line, clivo-axial angle, and clivus gradient. Ventral and dorsal SAS at foramen magnum (FM) on median sagittal MRI. (Figure S1)

Operation strategy

Operation reports were reviewed including 202 patients who underwent PFD with duraplasty and C1 laminectomy. The posterior arch of the C1 (1.5 cm on each side) and the foramen magnum (1.5 cm on each side, about 3 cm upwards) were resected until the dura mater transition at the cerebellar hemisphere was seen. After incision of the dura mater, the arachnoid was opened for further exploration of the fourth ventricle outlet.

Prognostic evaluation

Immediate postoperative outcomes were evaluated at discharge. Long-term follow-up was defined as the last available follow-up that was more than 6 months after surgery.

The Chicago Chiari Outcome Scale (CCOS) was used to evaluate the surgical efficacy from four aspects: pain symptoms, non-pain symptoms, functionality, and complications, with a score from 1 to 4 for each item.[23] The better the prognosis, the higher the score, with the CCOS score groupings (4–8, 9–12, 13–16) presenting the prognosis as favorable between 13 and 16, no obvious improvement between 9 and 12, and worse between 4 and 8. The latter two groups (4–12) were defined as having no improvement.

Variable Definition

The preoperative symptom duration was defined as the time from the onset of symptoms to the initial diagnosis of syringomyelia at admission to our institution. Observation time represented the time from the indication for surgery determined (with syringomyelia confirmed by MRI, but the patient refused surgery) to decompression surgery. The postoperative syrinx was reported to be either reduced (syrinx decreased in size ≥ 20%) or unchanged (syrinx decreased in size less than 20% or remained the same size). The event of significant improvement of syrinx was defined as a more than 20% decrease in maximal S/C ratio on follow-up MRI.

As previously proposed by Ono et al. [24], the configuration of syrinx can be divided into four types: A, Distended type; B, Moniliform type; C, Slender type; D, Circumscribed type. (Fig. 1) Syrinx with continuous septations (≥ 3) was referred to as moniliform type. The preoperative descent grades of cerebellar tonsillar herniation (TH) were divided into three stages: Grade 1, the cerebellar tonsil descends beyond the foramen magnum but does not reach the C1 arch; grade 2, reaches the C1 arch; grade 3, descends beyond the C1 arch.

Statistical analysis

SPSS software (version 22.0; IBM, NY, USA) was used for statistical analysis. For continuous variables, the mean values are presented with standard deviations, and the differences between two groups were analyzed using the student’s T-test and the Kruskal-Wallis test was used among multiple groups (≥ 3). A χ² or Fisher exact test was used to assess associations between dichotomous categorical variables. Wilcoxon rank-sum test was used for multiple categorical variables. P-value < 0.05 was considered statistically significant. Intraclass correlation coefficient (ICC) was used to evaluate the intra-repeatability from different observers (interobserver reliability). One independent researcher blinded to the group allocation completed the evaluations.

Patient Characteristics

A total of 202 CM-Ⅰ patients with syringomyelia were admitted to a single center between January 2017 and December 2020. There were 28 (13.8%) cases younger than 18 years old (yr), including 16 (57.1%) females and 12 (42.9%) males with a mean age of 7.8 ± 4.1 year (range, 1.0–17.0 year) in the pediatric group. There were 71.8% females and 28.2% males with a mean age of 47.4 ± 10.4 year (range, 23.0–73.0 year) in the adult group. Compared with the adult group, the pediatric group showed a much lower incidence of paresthesia of limbs and a proportion of patients who suffered weakness of limbs. There was quite a high incidence of scoliosis in children than that in adults. A longer clinical presentation was seen in the adult group. (Table 1)

At admission, the mean maximal S/C of the syrinx was 0.61 ± 0.18 (ICC = 0.911) and the mean syrinx length (vertebral segments) was 8.55 ± 3.62 (ICC = 0.924). The difference wasn’t seen in syrinx location, syrinx deviation, and cervical physio-curve. There was no significant difference in CSI and TH between the two groups. Moniliform type was more common in children than adults. (Table 2)

Surgical outcome

The average CCOS score was 14.56 ± 1.78 points. The overall postoperative prognosis was good, among all the patients, 152 cases (75.25%) with a favorable prognosis, and 50 cases (24.75%) were without improvement. There was no significant difference in the prognosis evaluated by CCOS and effective resolution of syrinx between the two groups. Syrinx was resolved effectively in 172 cases (85.15%), and 30 cases (14.85%) stayed unchangeable or worsened in terms of syrinx resolution.

No serious neurologic complications occurred in the two groups of patients. There were seven cases (3.5%) with postoperative transient headache, nine (4.5%) cases with short-term cerebrospinal fluid leakage, four cases (1.5%) incisional infection.

Factors predicting the surgical outcome in the adult group

The preoperative clinical parameters and radiologic measurements from the 174 adult patients were assessed using univariate and multivariate analyses to identify factors for predicting clinical improvement (composite CCOS) following surgery.

The primary univariate rank-sum test showed that longer preoperative syndrome duration (P = .000) and longer observation time (P = .000) were associated with a statistically significant increase in the likelihood of no improvement. In addition, moniliform type (P = .008) and syrinx length (P = .002) were also significant predictors of clinical improvement (composite CCOS). However, CSI, TH grade, CVJ parameters including BA, CAA, and CG, or any other factors were not associated with prognosis.

Through further multivariate analysis, it was found that preoperative syndrome duration, observation time, and with/without moniliform type were independent factors affecting the prognosis. The AOC curve suggested that the preoperative syndrome duration predicted the clinical improvement (composite CCOS) with high accuracy (AUC 87.1%, P = .000). (Fig. 2)

A Kaplan-Meier analysis demonstrated the cumulative incidence rate of effective syrinx resolution. The effectiveness of syrinx resolution in patients with moniliform type was significantly higher than that with the non-moniliform type (p = 0.000 by log-rank test) in the adult group. It was seen that pediatric patients with moniliform type could also achieve better syrinx resolution (p = 0.042 by log-rank test). (Fig. 2)

Differences between moniliform type and non-moniliform type

The most obvious difference between moniliform type and non-moniliform type lies in the patients with straightened cervical physio-curve or not, preoperative symptom duration, and ventral SAS.

Differences between pediatric and adult patients with CM-Ⅰ and syringomyelia

It is considered that children are still in the stage of growth and development. In a subset of pediatric patients with Chiari malformation without surgical treatment, the syringomyelia could even resolve spontaneously at follow-up.[26] According to comparisons of the preoperative clinical and radiologic characteristics in our continuous series of patients between adults and children, it illustrated that the main difference lies in the duration of preoperative symptoms. What’s more, children often found CM-Ⅰ with syringomyelia on further workup due to scoliosis. In adults, syringomyelia is mostly diagnosed when weakness or paresthesia of limbs occurs. All patients in the two groups completed the operation successfully, and there was no perioperative death or serious fatal complications.

Improvements in syringomyelia-related symptoms, however, do not always correspond to reductions in the size of the syrinx,[27, 28] so we analyzed both syrinx resolution and CCOS score respectively to comprehensively assess the surgical outcomes. The prognosis of both groups was generally favorable, no significant differences existed in the syrinx resolution and clinical outcomes between the two groups. Through further comparisons of radiologic parameters, it was found that syrinx in moniliform type was more common in children, which may be related to the relatively shorter natural history of CM-I in the pediatric group. There were no statistically significant differences in other baseline factors. It is unknown whether the more frequent occurrence of the moniliform syrinx in children is due to the high compensation of syringomyelia in pediatric CM-I. If so, it is necessary to further include phenotypic factors of syringomyelia in adults to analyze its association with prognosis through multivariate analysis.

Univariate and multivariate logistic regression models in adult patients

By reviewing the natural history of CM-I in children and adults, combined with the comparison between the two groups of cases in this series (children vs adults), it is suggested to further explore the relevant factors affecting the prognosis in adult patients, and whether it is related to the specificity of typical moniliform type.

Firstly, through univariate analysis in adults, we concluded that preoperative syndrome duration (P = .000), observation time (P = .000), the length of the syrinx (P = .002), and with/without moniliform type (P = .008) were related factors affecting the prognosis. Next, the multivariate analysis illustrated that the duration of preoperative symptoms, observation time before surgery, and with/without moniliform type were independent prognostic factors. Moreover, the effectiveness of syrinx resolution in patients with moniliform type was significantly higher than that with non-moniliform type.

Appropriate time of intervention for syringomyelia was important.[19, 29] The shorter the preoperative symptom duration and the shorter the observation time for decompression surgery, the better the prognosis. Timely and effective decompression surgery could achieve a better outcome, which was reflected in a higher CCOS score and syrinx resolution rate. The AOC curve also illustrated the duration of preoperative symptoms to guide prognosis with high accuracy. Therefore, surgical decompression should be taken as soon as possible once the indication for surgery is determined, which is beneficial to improve the prognosis. According to the natural history of the occurrence and development of syringomyelia, timely surgical intervention for the moniliform syringomyelia may allow patients to obtain better surgical efficacy, which would be discussed further in the following text.

There was a paucity of detailed studies on the relationship between prognosis after decompression surgery and syringomyelia morphology, which contained syrinx length, width, configuration, and deviation.[30–32] Syrinx configuration has rarely been regarded as predicting factors in previous studies. Our results confirmed that most factors above were not involved in predicting surgical outcomes. The syrinx length was only a related factor, rather than an independent factor affecting the prognosis. However, we interestingly found that the CCOS score improvement in syrinx with moniliform type was relatively better than that with non-moniliform type.

The particularity of syringomyelia with moniliform type

Why does syrinx with moniliform type show better prognostic improvement? Does the formation of moniliform syrinx indicate some kind of protective mechanism, or bring about some inspiration in the clinical diagnosis and treatment of the syringomyelia-related disease? The clinical prognosis of syringomyelia in moniliform type is relatively better, simultaneously with higher effectiveness of syrinx resolution shown by the survival curve in both pediatric and adult group. (Fig. 2) On the one hand, this may be related to the relatively short duration of its natural history, on the other hand, such syrinx may have strong compensatory effects for decompression surgery. During the postoperative follow-up, we found that the moniliform syringomyelia tended to move towards a smaller S/C, but the intrinsic separation of the syringomyelia appeared to persist in the short period after surgery. (Figs. 3 and 4) In terms of imaging characteristics, its continuous separation located inside the syrinx may be the source of the highly compensatory pathophysiological mechanism of such syringomyelia.[33]

From the point of the association among cerebellar tonsillar herniation, CSF circulation obstruction, and syringomyelia, we agreed that syringomyelia generally underwent a dynamic process: pre-syringomyelia stage, progressive stage, stable stage.[34] We further analyzed the differences of various factors at baseline between moniliform type and non-moniliform type syringomyelia and tried to trace the particularity from the origin and development of the moniliform syringomyelia. Such configuration tends to have a shorter duration of preoperative symptoms, and ventral SAS ≥ 2 mm are more concentrated in this type, suggesting that a moniliform type may represent the morphological feature in the early course of the disease. Because the cerebellar tonsillar herniation obstructs the circulation of CSF at the foramen magnum (especially in dorsal SAS), ventral SAS ≥ 2 mm was more common in the moniliform type group, which could be derived from some “buffer space” in the ventral SAS in the early stage.

What's more, straightened cervical physio-curve appears to be more common in patients with moniliform syrinx than non-moniliform syrinx. Some scholars have previously reported that compared with normal people, patients with syringomyelia will have a certain loss in cervical lordosis, but they have not mentioned the intrinsic relationship between the specific syringomyelia type and the corresponding changes in cervical sagittal alignment.[35] We speculated that the compensatory decrease in cervical lordosis during syringomyelia formation may act as a compensatory physiological response to get better CSF circulation in the SAS. This mechanism may be more pronounced in the moniliform syrinx. With the syrinx resolution after surgery, the lordosis may have a certain tendency to recover (Fig. 4), but that still needs to be confirmed by more RCTs.

Clinical significance of this study

The results in this study did not show that other factors such as CSI, TH grade, CVJ measurements, the presence or absence of scoliosis, and cervical physio-curve were directly related to clinical prognosis, nor did it show a difference in these factors between the moniliform and non-moniliform syringomyelia. Although they may be relevant factors that need to be comprehensively considered before surgery, they had no obvious predictive significance for guiding prognosis. However, CM-Ⅰ often accompanies other CVJ malformations, and it is difficult to identify the prognosis through only one single index. It must be analyzed on a case-by-case basis combined with other comprehensive factors.

This paper proposes a special morphological feature of syringomyelia, the moniliform type syringomyelia, associated with the surgical outcome from the perspective of clinical symptoms, imaging features (phenotypes of syringomyelia and biomechanical structures of the CVJ and cervical spine), and multivariate prognostic analysis. Based on the previous studies,[24] we give a new definition for this type of syringomyelia, that is moniliform syrinx with continuous obvious separation on MRI, which presents as a wide ventral SAS at FM, and tends to have a relatively shorter natural history and more common straightened cervical curve, and most importantly it may suggest a better prognosis. In the future, more clinical research is needed to dynamically observe the change of syringomyelia from the perspective of the pathophysiological mechanism of the occurrence and development of syringomyelia, finally to make it clear the structure and function of syringomyelia separation and the biological characteristics of moniliform syringomyelia.

Strengths and limitations

Moniliform type was elicited based on clinical and radiographic comparisons between different age groups. It was confirmed that such type may have an impact on the prognosis in adults, which was assessed using both syrinx resolution and CCOS. Finally, the particularity of moniliform syringomyelia was analyzed.

However, it was undeniable that there were some shortcomings in our study. The survival curve was analyzed based on the event of the effective resolution of the syrinx by postoperative MRI, so a short time interval was selected as much as possible, but there was still interval error, for example, the occurrence of the event happened to fall within the empty period between the checkpoints. Besides, there was a lack of longer follow-up after the effective resolution of the syringomyelia, so more research and evaluations are needed in the future to achieve the maximal ablation effect of the syringomyelia.