Relationship of syrinx size and tonsillar descent to spinal deformity in Chiari malformation Type I with associated syringomyelia

Abstract

Object

Chiari malformation Type I (CM-I) is a developmental abnormality often associated with a spinal syrinx. Patients with syringomyelia are known to have an increased risk of scoliosis, yet the influence of specific radiographically demonstrated features on the prevalence of scoliosis remains unclear. The primary objective of the present study was to investigate the relationship of maximum syrinx diameter and tonsillar descent to the presence of scoliosis in patients with CM-I–associated syringomyelia [AQ? Edit okay? If not, please advise. JG: edit correct]. A secondary objective was to explore the role of craniovertebral junction (CVJ) characteristics for additional risk factors for scoliosis.

Methods

The authors conducted a retrospective review of pediatric patients evaluated for CM-I with syringomyelia at a single institution in the period from 2000 to 2012. Syrinx morphology and CVJ parameters were evaluated with MRI, whereas the presence of scoliosis was determined using standard radiographic criteria. Multiple logistic regression was used to analyze radiological features that were independently associated with scoliosis.

Results

Ninety-two patients with CM-I and syringomyelia were identified. The mean age was 10.5 ± 5 years. Thirty-five (38%) of 92 patients had spine deformity; 23 (66%) of these 35 were referred primarily for deformity, and 12 (34%) were diagnosed with deformity during workup for other symptoms. Multiple regression analysis revealed maximum syrinx diameter > 6 mm (OR 12.1, 95% CI 3.63–40.57, p < 0.001) and moderate (5–12 mm) rather than severe (> 12 mm) tonsillar herniation (OR 7.64, 95% CI 2.3–25.31, p = 0.001) as significant predictors of spine deformity when controlling for age, sex, and syrinx location.

Conclusions

The current study further elucidates the association between CM-I and spinal deformity by defining specific radiographic characteristics associated with the presence of scoliosis. Specifically, patients presenting with larger maximum syrinx diameters (> 6 mm) have an increased risk of scoliosis.

Chiari malformation Type I (CM-I) is a developmental abnormality of the craniovertebral junction (CVJ), often associated with spinal cord abnormalities such as syringomyelia and scoliosis.^10,33 The rate of scoliosis in pediatric patients with CM-I has been reported to be as high as 80% in those with concurrent syringomyelia.^{6,10,16,20,31} Some authors have speculated that the etiology of scoliosis in such patients is related to the effect of the expanding spinal cord syrinx on the function of medially located motor neurons; the resulting imbalance in the paraspinal musculature is thought to predispose the individual to spinal deformity.^11,13,18 Others have questioned the validity of this model and suggested tonsillar compression of the cervicomedullary junction as an alternate etiology of scoliosis in patients with CM-I.⁴

Despite the uncertain etiology, scoliosis in the context of syringomyelia and CM-I is frequently treated with decompression of the CVJ to stabilize the spinal deformity.^7,8,15 Prior studies have demonstrated the importance of early neurosurgical intervention in patients with scoliosis to avoid or delay curve progression.^6,7,16 While this approach is widely practiced, no guidelines are in place to identify the patients at risk for concurrent spinal deformity. Although authors have examined radiological features of CM-I–related scoliosis, none have reported the differences in the radiological features of syringomyelia and the CVJ between patients with and those without scoliosis.

We hypothesized that a larger maximum syrinx size portends a greater risk of spinal deformity. In the current study, we investigated the relationship between maximum syrinx diameter and the presence of scoliosis in patients with CM-I and syringomyelia. A secondary aim was to explore the role of CVJ characteristics as additional risk factors for scoliosis.

Methods

Patient Population

After obtaining institutional review board approval from Washington University School of Medicine [AQ? What do you mean by “our institution”? Please name it.] and St. Louis Shriner's Hospital, where patients were comanaged, we queried a pediatric neurosurgical database for patients evaluated for CM-I with syringomyelia in the period between 2000 [AQ? In the abstract you refer to 2001. Please advise which year is correct: 2000 or 2001.JG: changes made] and 2012. Criteria for inclusion in the search were an age of 18 years or younger, diagnosis of CM-I (tonsillar descent beyond the foramen magnum ≥ 5 mm), and maximum anteroposterior (AP) diameter of syringomyelia ≥ 2 mm. Exclusion criteria were previous decompression or spinal fusion; missing preoperative MRI or radiographic studies; idiopathic syringomyelia or syringomyelia related to tumor, infection, arachnoiditis, or previous trauma to the spinal cord; history of myelomeningocele, tethered cord, or other dysraphism; and connective tissue diseases or genetic syndromes. We reviewed demographic data, presenting symptoms, neurological deficits, and medical comorbidities, as well as pre- and post-operative radiological studies [AQ? Do you perhaps mean preoperative and postoperative radiological studies? If not, please clarify your intended meaning.].

Imaging Data

Two different investigators, including a board-certified radiologist, assessed features of the posterior cranial fossa and the CVJ by using brain or cervical spine MRI. They measured 1) tonsillar herniation, as defined by the distance between the tip of the cerebellar tonsil perpendicular to McRae's line (basion-opisthion line); 2) position of the obex relative to the foramen magnum (quantified by the perpendicular distance between the obex and a line drawn from the basion perpendicular to the central canal);^35,36 3) CM 1.5, if present (defined as descent of the obex below the foramen magnum);^34,36 4) pb-C2, perpendicular distance between the ventral dura to the line that joins the basion to the posterior portion of the axis body inferior endplate;⁹ 5) odontoid inclination (angle between the dens and the posterior wall of the C-2 vertebral body);³⁷ 6) clivus canal angle (CXA; determined by the angle between the Wackenheim's clivus line and the posterior wall of the C-2 veretbral body),^28,31 7) basilar invagination, if present;²⁷ and 8) medullary kinking, if present [AQ? Edits okay? If not, please clarify your intended meaning.].²⁷

The location and morphology of the syrinx were evaluated using both T1- and T2-weighted MRI. Sagittal images of the spine and brain were used to determine the rostral and caudal extent of the syrinx, recorded as the number of vertebral segments traversed (as described by Ono et al.²⁰). Axial images were used to determine maximal AP diameter of the syrinx (in mm). Results of CSF flow studies were obtained when available.

The location and severity of spinal deformity were recorded using the standard Cobb technique with available imaging.³¹ Scoliosis was defined as a Cobb angle > 10°.¹⁴ Formal spine radiographs (standing AP and lateral long cassette radiographs) were available in 35 (100%) of 35 patients [AQ? Does this refer to preoperative or postoperative radiographs? Will move this information to Results JG: this refers to pre-operative radiographs.]. The patients without available imaging records were excluded from analysis, despite prior description of clinically significant scoliosis (> 10°) in the medical records.

Statistical Analysis

Univariate statistical analysis was performed to identify the unadjusted association of basic demographic and radiological covariates with the presence of spinal deformity. Proportions were analyzed using Fisher's exact test, and continuous data were analyzed using univariate logistic regression [AQ? Edit okay? If not, please clarify your intended meaning.JG: edits correct]. Tonsillar herniation and CXA were categorized using a cutoff of 1 SD from the mean. Collinearity between radiological covariates was examined using Pearson's correlation coefficients. Those covariates satisfying the statistical entry criteria (p < 0.2) on univariate analysis were included in the logistic regression model. Multiple logistic regression analysis was performed to assess independent risk factors for spinal deformity while adjusting for age and sex. The Hosmer-Lemeshow test was used to check goodness-of-fit of the model, and the predictive value of the multivariate model was expressed using the c-statistic. Statistical significance was established using a cutoff of p < 0.05. Data were analyzed using SPSS version 19.0 software (IBM).

Results

Patient Characteristics

Basic demographic data and chief complaints of the 92 patients, 51 females and 41 males, with CM-I and syringomyelia are presented in Table 1. The mean age at syrinx diagnosis was 10 ± 5 years [AQ? Overall, correct?].

TABLE 1.

Basic demographic and clinical characteristics of 92 patients with CM-I and syringomyelia^*

Parameter	w/o Scoliosis	w/ Scoliosis	p Value
no. of patients	57	35
mean age at syrinx diagnosis in yrs [AQ? Correct addition in yellow?]	10 ± 5.3	9.6 ± 3.8	0.696
no. of males (%)	27 (47)	14 (40)	0.490
mean BMI	18 ± 6.5	19.2 ± 6.5	0.498
chief complaint (%)
scoliosis	0 (0)	23 (66)	—
neurological symptoms	33 (58)	17 (48)	—
headache	38 (67)	9 (26)	—
combination of symptoms	19 (33)	14 (40)	—

^*Values represent the means ± standard deviation unless indicated otherwise. BMI = body mass index; — = not applicable [AQ? Correct for this and all other tables? If not, please provide correct definition.].

Radiological Features of Study Population

Syringomyelia was present in all patients in the study, with a mean maximal AP syrinx diameter of 5.9 ± 3.9 mm and syrinx length of 10 ± 5 vertebral levels. The syrinx crossed the cervicothoracic junction in 52 (56%) of 92 patients [AQ? Above you refer to 92 patients. Okay to change 97 to 92. JG: Thank you for catching this, have changed the numbers], was isolated to the thoracic or lumbar spine in 12 (13%) of 92, and spanned the length of the spinal cord in 6 (7%) 92. On preoperative T1-weighted MRI, the mean cerebellar tonsillar herniation was 14.4 ± 7.2 mm, mean obex descent was 10.2 ± 5.0 mm, mean pb-C2 was 3.9 ± 1.8 mm, mean odontoid inclination angle was 16° ± 8°, and mean CXA angle was 155° ± 11°. Basilar invagination was identified in 3 patients and medullary kinking in 18 (20%) of 92. There was no correlation between tonsillar descent (Spearman's = −0.152, p = 0.149) or CXA (Spearman's = −0.159, p = 0.131) and maximum syrinx diameter [AQ? Did you mean between tonsillar descent and max syrinx diameter AND between CXA and max syrinx diameter? If not, please clarify your intended meaning. JG: This is correct, tonsillar descent AND syrinx diameter; CXA AND syring diameter]. Syrinx size between the 2 groups of tonsillar descent (> 12 mm, 5–12 mm) was comparable (5.7 vs 6.1, p = 0.818).

Cerebrospinal fluid flow studies were performed in 33 patients (36%); 7 (21%) of these patients demonstrated normal flow, 12 (36%) demonstrated posterior flow restriction only, and 14 (42%) demonstrated both anterior and posterior flow restriction. Tonsillar pulsation was observed in 14 patients (42%). Flow restriction was more often observed in patients with severe tonsillar herniation than in those with moderate herniation (100% vs 50%, p = 0.001); however, no difference existed in flow restriction between those with and those without scoliosis (100% vs 76%, p = 0.55). Syrinx size was unrelated to extent of flow restriction (p = 0.26) and presence of tonsillar pulsation (p = 0.572). These data were excluded from multivariate analysis because of the limited number of patients with available CSF flow studies.

Radiographic Features of Scoliosis

Thirty-five (38%) of 92 patients had spine deformity; 14 were male [AQ? In Table 2, you refer to 14 males. Please advise. JG: thank you for your correction] and 21 were female (p = 0.472). Spinal deformity had been previously diagnosed in 23 (66%) of the 35 and was diagnosed during workup for presenting symptoms in 12 (34%). Preoperative radiographic records were available for review in all patients with scoliosis (Table 2). The mean major coronal Cobb angle was 30.4° ± 15°; 18 curves (51%) were apex left. The mean thoracic kyphosis was 47° ± 15°, with 32 (91%) of 35 patients demonstrating a sagittal curve > 30°. Patients diagnosed with scoliosis prior to referral to our institution demonstrated a larger Cobb angle (35° vs 22°, p = 0.017).

TABLE 2.

Radiographic features of scoliosis in 35 patients with CM-I and syringomyelia

Parameter	Value
mean age in yrs	10 ± 4
no. of males (%)	14 (40)
mean BMI	21 ± 8
coronal
mean major Cobb angle	30.4 ± 15°
no. convex lt curves	18 (51)
apex: IQR	T8-11
mean proximal thoracic Cobb angle	14 ± 11°
mean middle thoracic Cobb angle	27 ± 17°
mean thoracolumbar Cobb angle	17 ± 11°
mean coronal balance in mm	10 ± 9
sagittal
mean major Cobb angle†	47 ± 15°
T2–5 Cobb angle	15 ± 7°
T5–12 Cobb angle	31 ± 19°
T12–S1 Cobb angle	55 ± 17°
apex: IQR	T6-8

*IQR = interquartile range.

^†Ninety-one percent of patients had sagittal Cobb angle >30°.

Of the 35 patients with spinal deformity, 15 (43%) ultimately required definitive spinal fusion within 24 months of posterior fossa decompression. Surgeries were offered in cases in which the coronal deformity had progressed to more than 50°, as measured using the traditional Cobb technique. Before reaching this radiographic measure, all patients were nonoperatively treated with various techniques, according to physician and patient preferences. Patients requiring spinal reconstruction demonstrated greater initial coronal deformity (40° vs 25°, p = 0.003) and greater sagittal deformity (55° vs 42°, p = 0.027). An initial coronal Cobb angle > 30° (p = 0.014) and kyphosis > 45° (p = 003) corresponded to a higher likelihood of scoliosis surgery. Age at decompression (p = 0.786), sex (p = 0.176), body mass index (p = 0.779), preoperative maximal syrinx diameter (p = 0.152), and tonsillar herniation (p = 0.426) were not associated with scoliosis surgery.

Radiological Features of Scoliosis

Univariate analysis of radiological risk factors associated with scoliosis are presented in Table 3. The presence of scoliosis was significantly associated with a larger maximum AP syrinx diameter (OR 1.5, 95% CI 1.3–1.8, p < 0.001); a syrinx diameter > 6 mm corresponded to a higher rate of scoliosis (OR 6.80, 95% CI 2.64–17.52, p < 0.001 [AQ? The p value appears as <0.001 in Table 3. Please advise which value is correct. JG: Value in Table is correct]). The relationship between syrinx size and scoliosis is demonstrated in Fig. 1. A cervicothoracic syrinx was more common in patients with spinal deformity (83% vs 40%, p = 0.002). The mean tonsillar herniation was 11.7 ± 5.7 mm in patients with scoliosis, as compared with 16.2 ± 7.6 mm in those without scoliosis (OR 0.898, 95% CI 0.83–0.97, p = 0.004). Patients with tonsillar herniation of 5–12 mm were more likely to demonstrate scoliosis (OR 3.38, 95% CI 1.40–8.16, p = 0.007 [AQ? Table 3 reads 0.007. Please advise which value is correct. JG: value in table is correct]), despite no difference in syrinx size (p = 0.818). No difference existed in the CXA (p = 0.912), pb-C2 (p = 0.598), odontoid inclination (p = 0.547), or presence of CM 1.5 (p = 0.92) between patients with and without scoliosis.

TABLE 3.

Univariate analysis of radiological features associated with scoliosis in patients with CM-I and syringomyelia

Parameter	w/o Scoliosis (%)	w/ Scoliosis (%)	p Value (2-sided)	OR (95% CI)
no. of patients	57	35
CVJ
mean tonsillar herniation in mm	16.2 ± 7.6	11.7 ± 5.7	0.004	0.898 (0.832–0.969)
no. w/ tonsillar herniation of 5–12 mm (%)	19 (33)	22 (63)	0.007	3.385 (1.405–8.155)
no. w/ CM 1.5	17 (30)	11 (31)	0.920	1.07 (0.433–2.68)
mean obex descent in mm	11 ± 5	9.1 ± 4.7	0.084	1.084 (0.989–1.187)
no. w/ obex <11 mm	28 (49)	12 (34)	0.144	0.552 (0.218–1.249)
mean odontoid pb-C2 in mm	3.8 ± 1.9	4 ± 1.6	0.598	1.067 (0.842–1.351)
mean dens inclination	15.9 ± 7.8°	14.9 ± 7.7°	0.547	0.983 (0.930–1.039)
mean CXA	155.5 ± 10.8°	155.8 ± 11.4°	0.912	1.002 (0.964–1.042)
no. w/ CXA >150°	37 (65)	22 (63)	0.755	1.151 (0.447–2.777)
no. w/ basilar invagination	2 (4)	1 (3)	0.865	1.236 (0.108–14.16)
no. w/ medullary kinking	13 (23)	5 (14)	0.321	1.773 (0.572–5.493)
syrinx morphology
mean syrinx diameter in mm	4.1 ± 3	8.9 ± 3.4	<0.001	1.537 (1.285–1.839)
no. w/ syrinx diameter >6 mm	17 (30)	26 (74)	<0.001	6.797 (2.637–17.521)
no. w/ syrinx diameter >7 mm	10 (18)	24 (68)	<0.001	10.255 (3.82–27.53)
mean syrinx length (no. of levels)	8.1 ± 5	13.8 ± 4	<0.001	1.264 (1.136–1.406)
syrinx location
no. cervical	21 (37)	1 (3)	—	—
no. cervicothoracic	23 (40)	29 (83)	0.002	26.5 (3.31–211.8)
no. thoracolumbar	9 (16)	3 (9)	0.111	7 (0.639–76.708)
no. holocord	4 (7)	2 (6)	0.079	10.5 (0.758–145.35)
CSF flow studies
no. w/ flow study performed	29 (51)	4 (11)	<0.001	—
no. w/ flow restriction	22 (76)	4 (100)	0.555	—
no. w/ tonsillar pulsation	11 (38)	3 (75)	0.279	—

Fig. 1.

A scatterplot of scoliosis frequency against maximal AP syrinx diameter. Patients with a larger maximal syrinx diameter have scoliosis at a higher frequency. A maximum AP syrinx diameter > 6 mm indicates a higher prevalence of scoliosis.

Following multivariate logistic regression analysis, maximum syrinx diameter > 6 mm [AQ? Edit correct? If not, please clarify your intended meaning.] (OR 12.1, 95% CI 3.63–40.57, p < <0.001 [AQ? In the abstract and in Table 4 you refer to a p value <0.001. Please advise which value is correct. JG: Value in table is correct, thank you]) and moderate tonsillar herniation (5–12 mm; OR 7.64 [AQ? Abstract and Table 4 have 7.64 as the OR. Please advise. JG: 7.64 is correct], 95% CI 2.3–25.31, p = 0.001) remained independently associated with the presence of spine deformity when adjusted for age, sex, and syrinx location (Table 4). No significant interactions were found. The model was stable and robust with a Hosmer-Lemeshow chi-square of 0.28 (p = 0.869) and c-statistic of 0.807 (p < 0.001).

TABLE 4.

Independent risk factors for scoliosis in patients with CM-I and syringomyelia

Parameter	OR	95% CI	p Value
max syrinx diameter >6 mm [AQ? Correct to add the yellow-highlighted words?]	12.1	3.63–40.57	<0.001
tonsillar herniation of 5–12 mm	7.64	2.3–25.31	0.001

Hosmer-Lemeshow chi-square = 0.28, p = 0.869. [AQ? Where in the table should an asterisk be placed; that is, to what does this statement apply?]

Discussion

We have shown a strong association of the radiological features of syringomyelia and the CVJ with scoliosis in 92 children with CM-I and syringomyelia. Specifically, a larger maximum syrinx diameter [AQ? When you refer to a “larger maximum syrinx diameter,” do you mean >6 mm or >7 mm? JG: referring to “larger” syrinx size, as it increases from 1 to 12 etc. Not referring here to >6 or >7. Initially, reviewer comments had not agreed with using term “increasing” therefore have substituted “larger”] and moderate (5–12 mm) rather than severe (> 12 mm) tonsillar ectopia were independently associated with a higher frequency of scoliosis. No association was observed between syrinx location and the presence of scoliosis when adjusted for syrinx size. While the severity of the spinal deformity did not correlate with the diameter of the syrinx, patients with more severe initial deformity (scoliosis > 30° or kyphosis > 45°) were more likely to require surgery for deformity correction despite decompression.

Previous authors have recognized the association among CM-I, scoliosis, and syringomyelia.^6,20,36 The atypical patterns of CM-I–related deformities, such as left thoracic apices and hyperkyphosis, rapid progression, and early onset, suggests an etiology unique from that of adolescent idiopathic scoliosis.^24,38 Both syringomyelia and hindbrain compression have been hypothesized in the pathogenesis of CM-I–related scoliosis; however, the role of each as a risk factor remains poorly defined.^3,21,25,26 Although previous studies have recognized that CM-I patients with syringomyelia are more likely to demonstrate scoliosis,^16,20,36 risk factors for clinically significant spinal deformity and the need for corrective spinal surgery remain controversial [AQ? Do you mean that they are still debated? JG: Correct].^20,21,24

Both Ozerdemoglu et al.²¹ and Qiu [AQ? You refer to Qui et al., which is reference 24, but you cite reference 38, which is Wu et al. Please advise which reference you meant to cite. Intended Reference was Qiu et al, reference 24] et al. ²⁴ have explored the relationship between syrinx morphology and spinal deformity characteristics. In a series of 112 pediatric patients with scoliosis, Ozerdemoglu and colleagues reported a range of syrinx diameters (2–18 mm) with no correlation to curve magnitude.²¹ Recently, Qiu et al.²⁴ found no association between syrinx length or the ratio of syrinx/cord diameter and curve severity in 83 patients [AQ? Do you mean an association between syrinx length or ratio of syrinx/cord diameter AND curve severity? (Note that “between...and...” are used together to express a relationship.)JG: yes that is correct]. Similar to the study by Yeom et al.,³⁹ the current study revealed no association between syrinx characteristics and coronal Cobb angle. Only Ono et al.²⁰ noted a correlation between scoliosis severity (12.1° vs 37.7°) and syrinx length (13.4 vs 15.9 levels) in an adult population with CM-I. Yet, in a study by Krieger et al.,¹⁶ all 79 patients with CM-I and scoliosis demonstrated a syrinx width > 6 mm. Recognizing the strong association between syrinx parameters and scoliosis, these authors implicated syrinx size in spinal growth. The preliminary association between greater maximum syrinx diameter and scoliosis in the current study builds on and significantly strengthens this theory. As we have shown, patients with a syrinx diameter > 6 mm demonstrate a significantly higher rate of scoliosis (OR 8.08). This finding also suggests that the mechanisms involved in scoliosis development may be distinct from those involved in curve progression, as syrinx size does not appear to be associated with the magnitude of coronal deformity.

Some authors describe a spatial component in the relationship between syringomyelia and scoliosis in CM-I.²¹ Ozerdemoglu et al.²¹ found that patients with a caudally located maximal syrinx diameter demonstrated distal major curve apices. The cervicothoracic junction was most often affected by a syrinx, with the largest syrinx area in the C5–7 region. In Attenello et al.,¹ the majority of patients with scoliosis demonstrated cervical (90%) and thoracic (95%) syringes. Tokunaga et al.³² described a syrinx at the cervicothoracic junction in 96% (22 of 23) of patients with scoliosis. The mean syrinx/spinal cord ratio in their study was 49%, with maximum dilation occurring most often at the C-6 level (17 of 23 cases). The current study showed a higher rate of scoliosis among cervicothoracic and holocord syringes on univariate analysis; however, when adjusted for syrinx length and maximum diameter, syrinx location did not confer an additional risk for scoliosis. This may suggest the susceptibility of the cervicothoracic junction to syrinx expansion and highlights the strength of the association between syrinx size and scoliosis.²¹

The prevalence of scoliosis in CM-I without syringomyelia has been reported in 8%–30% of patients.^19,31 Some authors suggest that this prevalence implicates tonsillar herniation as the underlying etiology of CM-I–related scoliosis, while others argue that this rate may reflect a basal level of scoliosis unrelated to CM-I [AQ? Correct to add the yellow-highlighted word? If not, please clarify your intended meaning. JG: Yes this is correct].¹⁶ Although this unique patient population was not included in the current study design, we intend to investigate it in a future study.

Few studies have shown a direct association between CVJ parameters and scoliosis in CM-I patients with or without syringomyelia. Only Ono et al.²⁰ [AQ? You refer to Ono et al., which is reference 20, but you cite reference 1, which is Attenello et al. Please advise which reference was truly meant.JG: Ono was meant, thank you.] reported a relation between levels of tonsillar descent and spinal deformity characteristics; they documented 70% concordance between the laterality of cerebellar tonsil asymmetry and the direction of major curve convexity. To our knowledge, no previous study has found moderate tonsillar ectopia (5–12 mm) to be associated with scoliosis in patients with CM-I and syringomyelia. Attenello et al.¹ found that 90% of patients with CM-I and scoliosis demonstrated “moderate ectopia,” which was defined as tonsillar descent between C-1 and C-2. Similar to the value noted in the current study, Tokunaga et al.³² reported a mean tonsillar herniation of 9.7 mm in 23 patients with CM-I–related scoliosis. Other studies have also found a high proportion of patients with moderate descent. Qiu et al.²⁴ showed that 44% of tonsils descended to the region between the foramen magnum and the C-1 level in 87 patients with scoliosis. They found no correlation between tonsillar herniation and curve severity. In a study of CM-I–related scoliosis, Krieger et al.¹⁶ found that 84% of patients had tonsils above the C-2 level.

It is surprising that patients with only moderate tonsillar ectopia (> 5 mm, < 12 mm) have a greater risk of clinically significant scoliosis, whereas patients with more severe tonsillar descent (> 12 mm) demonstrate a lower prevalence of scoliosis. While no difference in syrinx size exists between the 2 groups, several studies have reported a similarly perplexing association between moderate ectopia and the presence of syringomyelia in CM-I.^23,30,37 In light of this observation, our findings may reflect a complex process at the CVJ.¹⁷ Several authors have suggested that the degree of obex descent at the CVJ could influence the spinal axis and the development of syringomyelia.²⁹ Yet, we observed no association between obex descent and the presence of CM 1.5 with scoliosis. Although dynamic imaging of CSF flow in patients with CM has helped shape our understanding of the pathophysiology of syringomyelia,⁵ we were unable to include these studies in the multivariate analysis because of the small number of patients with complete testing. Likely, [AQ? Please provide a referent for “This”; that is, to what does “This” refer? JG: Thank you for your comment, I have changed to sentence to reflect the need of dynamic imaging in the future as our data are limited in addressing this aspect of the relationship ] there is a dynamic component to the development of scoliosis and warrants further investigation using dynamic imaging in the future.

Ultimately, the goal of decompression in CM-I patients with scoliosis is to stabilize curve progression and thereby prevent the need for complex spinal reconstruction. Although the natural history of scoliosis in CM-I remains unclear, a recent meta-analysis by Hwang et al.¹² suggests the benefit of decompression and syrinx reduction in improving scoliotic curve progression. But despite adequate posterior fossa decompression, the rate of scoliosis surgery ranges between 15% and 50%, comparable with the 43% observed in our patient population[AQ? By “current study” do you mean the one by Hwang et al. or the one by you? If you, where is this rate of scoliosis surgery discussed? What am I missing? JG: this refers to the 43% of patients that required scoliosis surgery, see Results, Page 4/11].^2,6,16,25 Past authors have described an older age, level of spinal deformity, extent of syrinx resolution, and degree of initial scoliosis as risk factors for curve progression and spinal fusion.^{1,2,6,12,22,25} In the largest study to date, Krieger et al.¹⁶ noted a higher rate of surgery in patients with an initial Cobb angle > 25°, with no association with preoperative syrinx size or extent of syrinx resolution. Similarly, we observed that patients with scoliosis > 30° as well as those with kyphosis > 50° were significantly more likely to require spinal reconstruction within the follow-up period. However, the decision to perform spinal reconstruction involves multiple variables, among which are patient and physician preferences [AQ? Edit correct? If not, please clarify your intended meaning. JG: these changes are correct]. While it is likely that early intervention plays a role in avoiding surgery, additional studies are needed to adequately determine the optimal management of spinal health in these complex cases.

A central limitation of our study is its retrospective, cross-sectional design. The data and their interpretation are limited by completeness of the medical records and consistency between preoperative evaluations, particularly the availability of standing plain films of the spine. A number of patients did not seek orthopedic care at our institution, thus limiting our ability to capture longitudinal data and examine the temporal relationship among syringomyelia, scoliosis development, and scoliosis surgery. Importantly, our study population included only those patients with syringomyelia ≥ 2 mm on MRI evaluation; we hope to investigate patients with CM-I–related scoliosis without syringomyelia in the future. Note also that this study is subject to the bias of a tertiary care referral center, as the composition of our study population represents only those patients referred to our clinic. To capture the true incidence of spinal deformity in patients with CM-I and better understand risk factors of progression and functional outcome, future research in this field must come in the form of a prospective, multicenter effort such as that of the Park-Reeves Syringomyelia Research Consortium.

No study has examined radiological features of the CVJ and syringomyelia as potential risk factors for the presence of scoliosis in CM-I with syringomyelia. We have shown that coexistent scoliosis is more likely with a larger maximum syrinx diameter and moderate tonsillar herniation (5–12 mm). While the results of this study have limited clinical implications, the findings presented herein suggest that dilation of the spinal cord by syringomyelia, as well as compression at the cervicomedullary junction, appears to influence spinal growth in CM-I. Prospective studies conducted through the Park-Reeves Syringomyelia Research Consortium are directed toward further elucidating these complex relationships.

Conclusions

Results from this retrospective study demonstrate a strong association between maximum syrinx size and the presence of spinal deformity in patients with CM-I and syringomyelia. This finding represents an early but important component in the investigation of potential common mechanisms involved in the pathogenesis of these complex conditions.

Abbreviations used in this paper

AP

anteroposterior

CM-I

Chiari malformation Type I

CVJ

craniovertebral junction

CXA

clivus canal angle