Accepted Manuscript MRI-based Morphological Evidence of Spinal Cord Tethering Predicts Curve Progression in Adolescent Idiopathic Scoliosis Min Deng, MMed, Steve CN. Hui, MSc, Fiona WP. Yu, MSc, Tsz-ping Lam, MBBS, Yong Qiu, MD, Bobby KW. Ng, MBBS, Jack CY. Cheng, MBBS, MD, Winnie CW. Chu, MBChB, MD PII:
S1529-9430(15)00199-0
DOI:
10.1016/j.spinee.2015.02.033
Reference:
SPINEE 56216
To appear in:
The Spine Journal
Received Date: 16 July 2014 Revised Date:
4 February 2015
Accepted Date: 18 February 2015
Please cite this article as: Deng M, Hui SC, Yu FW, Lam T-p, Qiu Y, Ng BK, Cheng JC, Chu WC, MRIbased Morphological Evidence of Spinal Cord Tethering Predicts Curve Progression in Adolescent Idiopathic Scoliosis, The Spine Journal (2015), doi: 10.1016/j.spinee.2015.02.033. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
MRI-based Morphological Evidence of Spinal Cord Tethering Predicts Curve Progression in Adolescent Idiopathic Scoliosis Min Deng, MMed1 , Steve CN Hui, MSc1 , Fiona WP Yu, MSc2 , Tsz-ping Lam, MBBS2,4 , MBChB, MD*1
RI PT
Yong Qiu, MD3,4 , Bobby KW Ng, MBBS2, Jack CY Cheng, MBBS, MD2,4 , Winnie CW Chu,
Department of Imaging and Interventional Radiology, The Chinese University of Hong Kong
2
Department of Orthopaedics and Traumatology, The Chinese University of Hong Kong
3
Spine Surgery, The Affiliated Drum Tower Hospital of Nanjing University Medical School,
SC
1
4
M AN U
Nanjing, China
Joint Scoliosis Research Center of the Chinese University of Hong Kong and Nanjing
University
Prof. Winnie CW Chu
TE D
*Corresponding Author:
Email address: [email protected]
EP
Mailing address: Department of Imaging and Interventional Radiology, The Chinese University
AC C
of Hong Kong, Shatin, New Territories, Hong Kong.
ACCEPTED MANUSCRIPT
MRI-based Morphological Evidence of Spinal Cord Tethering
2
Predicts Curve Progression in Adolescent Idiopathic Scoliosis
3
Abstract
4
BACKGROUND CONTEXT: Existing prognostic factors for adolescent idiopathic
5
scoliosis (AIS) patients have focused mainly on curve, maturity and bone related factors.
6
Previous studies have shown significant associations between curve severity and
7
morphological evidences of relative shorter spinal cord tethering in AIS. In conjunction with
8
increased prevalence of abnormal somatosensory cortical evoked potentials (SSEP) and
9
low-lying cerebellar tonsil in severe AIS, suggest that there might be neural morphological
M AN U
SC
RI PT
1
predictors for curve progression.
11
PURPOSE: We sought to identify any morphological predictors associated with cord
12
tethering, as measured by magnetic resonance imaging (MRI), for curve progression in AIS
13
patients.
14
STUDY DESIGN/SETTING: A prospective cohort study.
15
PATIENT SAMPLE: A total of 81 female AIS subjects between 10 and 14 years, without
16
surgical intervention during the follow up period.
17
OUTCOME MEASURES: MRI scans of hindbrain and whole spine, and areal bone mineral
18
density (BMD) at bilateral femoral necks were performed.
19
METHODS: All AIS patients were longitudinally followed up starting from initiation of
20
bracing beyond skeletal maturity in 6-monthly interval. Clinical and radiographic data were
21
recorded at each clinic visit. BMD and MRI measurements including ratio of spinal cord to
22
vertebral column length, ratio of anteroposterior (AP) and transverse (TS) diameter of cord,
23
lateral cord space (LCS) ratio, cerebellar tonsil level, and conus medullaris position were
24
obtained at baseline. Only compliant patients with a minimum 2-year follow-up were
25
analyzed. AIS girls were assigned into three groups according to bracing outcome: (A)
AC C
EP
TE D
10
1
ACCEPTED MANUSCRIPT
Non-Progression (curvature increase≤5°); (B) Progression (curvature increase≥6°); (C)
2
Progression with surgery indication (Cobb angle≥50° after skeletal maturity despite bracing).
3
The predictors for curve progression were evaluated using univariate analysis and
4
multivariate ordinal regression model.
5
RESULTS: The average duration of follow-up was 3.4 years (range: 2.0-5.6 years). There
6
were 46 girls (57%) in group A, 19 girls (23%) in group B and 16 girls (20%) in group C. No
7
significant intergroup differences were found in spinal cord length, tonsil level and conus
8
position. Group C had significantly longer vertebral column length, smaller cord-vertebral
9
length ratio, and higher AP/TS cord ratio as compared to group A, while LCS ratio in group C
10
was significantly increased when compared with both group A and group B. In regression
11
model, five significant independent predictors including cord-vertebral length ratio(Odds
12
Ratio (OR): 1.993 [95% CI: 1.053-3.771, P=0.034]), LCS ratio (OR: 2.639 [95% CI:
13
1.128-6.174, P=0.025]), initial Cobb angle (OR: 1.156 [95% CI: 1.043-1.281, P=0.006]),
14
menarche age (OR: 1.688 [95% CI: 1.010-2.823, P=0.046]), BMD (OR: 2.960 [95% CI:
15
1.301-6.731, P=0.010]), and a marginally significant predictor namely AP/TS cord ratio(OR:
16
1.463 [95% CI: 0.791-2.706, P=0.096]) were obtained.
17
CONCLUSIONS: On baseline MRI measurement, cord-vertebral length ratio and LCS ratio
18
are identified as new significant independent predictors for curve progression in AIS while
19
AP/TS cord ratio is suggested as a potential predictor requiring further validations. The above
20
MRI parameters can be taken into accounts for prognostication of bracing outcome.
21
Keywords:
22
Predictor; Spinal cord;
SC
M AN U
TE D
EP
AC C
23
RI PT
1
Adolescent idiopathic scoliosis; Magnetic resonance imaging; Bracing;
2
ACCEPTED MANUSCRIPT
1
Introduction Adolescent idiopathic scoliosis (AIS) is a three dimensional structural deformity of the
3
spine that occurs in otherwise healthy children, predominantly in adolescent girls that
4
typically progresses during the adolescent growth spurts, affecting 2%-3% of children
5
worldwide [1, 2]. Despite decades of dedicated research, the aetiopathogenesis of this classic
6
orthopaedic disorder remains uncertain. Scoliosis curves may remain static, progress slowly
7
or rapidly while the prognostic factors governing curve progression are not totally clear.
8
Severe scoliosis is associated with significant morbidities and disfiguration of body image. It
9
is a major challenge to estimate the likelihood of curve deterioration so that proper treatment
10
planning and counseling to the patients and the parents can be provided. Bracing has been
11
widely applied in AIS patients aiming for preventing curve progression so that corrective
12
surgery can be avoided [3-6]. Despite the previous controversies of bracing effectiveness
13
[7-10], recent researches including a latest multicenter large-scale study have shown that
14
bracing is superior to observation in clinical outcome[5, 11, 12]. Predictors for curve
15
progression should be redefined for AIS patients with bracing treatment, rather than
16
following predictors deriving from natural progression of curve in patients with observation.
17
Existing known predictors for curve progression despite bracing treatment in AIS are
18
mainly focused on clinical factors such as chronologic age[13], Risser sign[14], menarchal
19
status[13, 15], and radiographic factors such as curve magnitude[13, 15], curve pattern[16,
20
17], BMD[16] [18]. Recently, calcaneal stiffness index detected by quantitative ultrasound
21
(QUS) has been determined as a new predictor [13].
SC
M AN U
TE D
EP
AC C
22
RI PT
2
MRI is an excellent imaging modality for morphological delineation of soft tissue and
23
neural structures. It is now adopted in the protocol of pre-operative planning in many centers.
24
Research papers reporting abnormalities in the neuroaxis have renewed interest in abnormal
25
neuroanatomy associated with AIS. Previous cross-sectional studies have revealed significant
26
correlations between curve severity and morphological evidences of subclinical tethering of
27
relative shorter spinal cord in AIS, such as reduced spinal cord to anterior vertebral column
28
length ratio and altered cross-sectional shape of spinal cord[19, 20]. In conjunction with
29
increased prevalence of abnormal SSEP and low-lying cerebellar tonsil in severe AIS as 3
ACCEPTED MANUSCRIPT
1
compared with mild to moderate AIS [19, 21], neuro-related factors might have predictive
2
value for curve progression. To the best of our knowledge, there is no longitudinal MR study evaluating the
4
relationship between morphological evidences of relative shorter cord tethering with the
5
progression of scoliosis curve. We sought to identify any new neural morphological
6
predictors, as measured by MRI, to determine curve progression in brace-treated AIS
7
patients.
8
Materials and methods
9
Subjects
M AN U
SC
RI PT
3
During the period from July 2007 to April 2011, AIS subjects were prospectively recruited
11
from our scoliosis clinics, which was one of the only two tertiary referral centers specialized
12
in scoliosis, serving a population of seven million citizens. All subjects should be female with
13
right-sided thoracic/thoracolumbar curves with apical vertebra ranged from T6-T12. All
14
recruited subjects have been offered bracing according to following Scoliosis Research
15
Society (SRS) modified inclusion criteria: (i) age 10-14 years old; (ii) skeletal immaturity
16
(Risser sign 0-2); (iii) primary curve Cobb angle of 20° to 40°; (iv) no prior treatment; (iv)
17
either premenarchal or less than 1 year postmenarchal. Exclusion criteria included conditions
18
and medications that would affect bone remodeling and calcium metabolism, neuromuscular
19
abnormalities, genetic diseases, chromosomal defects, autoimmune disorders, endocrine
20
disturbances. Ethical approval was obtained from the University and Hospital Research
21
Ethics Committee. All subjects provided written informed consent.
22
Clinical assessment
23
Before bracing started, menarche status was recorded. Body weight, standing height, sitting
24
height, and arm span were measured with standard stadiometry techniques [22]. Corrected
25
standing height was derived from Bjure formula (logy =0.011x-0.177), where y represents the
26
loss of trunk height (cm) and x represents major Cobb angle. The body mass index (BMI) for
27
AIS patients was calculated by specific formula (body weight [kg] / square of arm span
AC C
EP
TE D
10
4
ACCEPTED MANUSCRIPT
1
[m2])[23]. According to method described in our previous study [24], areal BMD at bilateral femoral
3
neck was measured by Dual-energy X-ray absorptiometry (XR-46; Norland Medical Systems,
4
Fort Atlinson,WI, USA), with the precision errors of 1.1% to 3.7%. Skeletal maturity was
5
assessed on pelvis anterior-posterior radiographs using Risser sign grading system [25].
6
Standing radiographs of the whole spine were obtained for Cobb angle measurement while
7
curve pattern was evaluated according to King’s classification [26].
8
Bracing treatment and follow-up
9
The type of bracing was determined by curve patterns. Tailor made underarm brace by
10
experienced orthotist was used for AIS patients with major thoracic/thoracolumbar curve (Fig
11
1). Bracing treatments mainly followed the protocols described in our previous study[27].
12
Daily bracing time of 22 hours per day was initiated, followed by modification in terms of
13
temporary bracing outcome between two consecutive visits described by Yrjönen et al[28]: (i)
14
if improve (Cobb angle decrease >6°) or with a Risser sign greater than 3 and/or over-1-year
15
after menarche, shortening 2 to 4 hours per day at each follow-up. (ii) if stable (Cobb angle
16
change6°), remaining 22 hours per day. All subjects
18
underwent longitudinal follow-up at 6-month intervals beyond skeletal maturity in a special
19
clinic. Radiographic and clinical data were collected at each follow-up visit. Compliance ratio
20
was defined as the ratio of the actual daily bracing time to the recommended daily time. Only
21
AIS patients who had a minimum 2-year follow-up post bracing and a compliance ratio
22
≥75 % were analyzed.
SC
M AN U
TE D
EP
AC C
23
RI PT
2
Final Cobb angle change was calculated by subtracting Cobb angle measured on latest
24
follow-up radiographs from the baseline angle measured before implementation of bracing.
25
The bracing outcome was evaluated referring to the standardized criteria of SRS committee
26
in principle [29]. AIS girls were assigned into three groups according to bracing outcome: (A)
27
Non-Progression (curvature increase ≤5°); (B) Progression without surgery indication
28
(curvature increase ≥6°); (C) Progression with surgery indication (Cobb angle ≥50° after
29
skeletal maturity despite bracing). 5
ACCEPTED MANUSCRIPT
MRI protocol
2
All AIS patients underwent baseline MRI scan before bracing initiation. MRI data acquisition
3
was performed on 3.0 T MR scanner (Achieva TX; Philips Healthcare, Best, The Netherlands)
4
using a spine array coil. Imaging protocol included sagittal scans of hindbrain and the whole
5
spine from foramen magnum to sacrum, and transverse (TS) scans of 5 vertebral levels
6
around the vertebral apex (including the apical vertebra, 2 vertebrae above and below).
7
Above MR images were obtained using 2D Turbo spin-echo T2 weighted sequence with
8
following parameters: (i) sagittal: turbo factor=33, TR=4048 milliseconds(ms), TE=120 ms,
9
matrix=312×248, slice thickness=3.5 mm, slice gap=0 mm, field of view= 690 mm, NSA=2
10
(ii) axial: turbo factor=40, TR=645 ms, TE= 120 ms, matrix= 284×186, slice thickness=2 mm,
11
slice gap=0 mm, field of view=200 mm, NSA=1). The MRI images were reformatted on a
12
workstation (Workspace, Philips Medical System, Best, The Netherlands).
13
MRI morphologic measurement
14
The MRI morphological measurements have been reported in a number of previous
15
cross-sectional studies [19, 20]. In this study, all measurements were made by a single
16
observer (radiologist with > 10 years of experience with MRI imaging). In brief, the
17
parameters are as follows:
18
(a) Length of the anterior vertebral column and spinal cord
TE D
M AN U
SC
RI PT
1
On the reformatted straightened best mid sagittal image of the anterior vertebral column or
20
spinal cord, the total length of the anterior vertebral column was measured from the tip of
21
odontoid process of C2 down to the inferior endplate of L5 along the mid sagittal line of the
22
anterior vertebral column while length of cord was measured from the level of the tip of
23
odontoid process of C2 down to the conus medullaris (Fig 2).
24
(b) Conus position
25
The tip of conus medullaris was identified and its position was defined in relation to adjacent
26
vertebrae/disc unit which was divided into 4 segments according to the method described by
27
Saifuddin et al[30], including 3 equal portions of vertebral body (upper, middle, and lower
28
thirds) as well as the separate distal disc. From the upper third of T12 down to middle third of
AC C
EP
19
6
ACCEPTED MANUSCRIPT
L3, total 14 segments were assigned with ordinal C-values ranging from 0 to 13. For example,
2
the conus position at lower third of L1 level was represented with a C-value “6” (Fig 3).
3
(c) AP/TS cord ratio and Lateral cord space (LCS) ratio
4
The longest axis of the cord was taken for measuring maximum TS diameter at the
5
cross-sectional plane of apical vertebra in major curve, while the largest anteroposterior (AP)
6
diameter of the cord was measured along a perpendicular axis to the TS diameter (Figure 4a).
7
The AP/TS ratio was then calculated.
RI PT
1
A line was drawn bisecting the chosen vertebra by 3 standardized datum points as
9
illustrated in our previous study [20] . This line represented the vertebral axis. Parallel to the
10
vertebral axis, 4 lines were drawn from right to left (Figure 4b): (i) tangential to the inner
11
surface of the pedicle on the right/convex side, (ii) tangential to the lateral cord surface on the
12
right/convex side, (iii) tangential to the lateral cord surface on the left/concave side, and (iv)
13
tangential to the inner surface of the pedicle on the left/concave side. The distance of LCS to
14
the right side of the cord between line (i) and (ii) was defined as “A,” whereas the distance of
15
LCS to the left side of the cord between line (iii) and (iv) was defined as “B” and the cord
16
diameter “C” was defined as the distance between line (ii) and (iii). A ratio was calculated by
17
the following formula: (A+C/2) / (B+C/2) (Figure 4b), which was known as the LCS ratio-
18
convex over concave, reflecting the position of the cord.
19
(d) Cerebellar tonsil level
20
The position of cerebellar tonsil was obtained by the same method as described in our
21
previous studies [21, 31]. The level of cerebellar tonsil relative to a reference line connecting
22
the basion and opsithion (BO line) was measured in terms of the method described by
23
Aboulezz et al[32] (Fig 5).
24
Statistical Analysis
M AN U
TE D
EP
AC C
25
SC
8
Linear regression was used to test the effect of age on various MRI parameters. For
26
continuous variables with normally distribution, data were presented as mean ± standard
27
deviation, and one-way analysis of variance (ANOVA) with least significant difference (LSD)
28
test was used for group comparison (one-way ANOVA was replaced by univariate ANOVA if
29
parameter associated with age). For continuous variables with skewed or unknown 7
ACCEPTED MANUSCRIPT
1
distribution, data were presented as median with interquartile range, and Kruskal-Wallis test
2
along with Mann-Whitney U test were used for group comparison. The chi-square test was
3
used to compare the percentages among groups for categorical variables. Multivariate ordinal logistic regression analysis was performed to identify independent
5
risk factors for curve progression of AIS patients during bracing. Only variables with a
6
univariate P value of 0.05). After adjustment for age, group C
29
showed significantly longer anterior vertebral column length than group A. Group C also had 8
ACCEPTED MANUSCRIPT
significantly smaller cord-vertebral length ratio than group A. No significant intergroup
2
differences were found in spinal cord length and tonsil level. Group C showed significantly
3
higher AP/TS cord ratio and LCS ratio at apical vertebra as compared to other 2 groups. LCS
4
ratio in group A was significantly lower than group B. There was no significant difference in
5
distribution of conus position among subgroups ( χ2 test, P=0.641).
6
Predictors for AIS curve progression despite bracing
7
The average duration of follow-up was 3.4 years (range: 2.0-5.6 years). Using univariate
8
analysis, following variables with P-value less than 0.1 were eligible to enter regression
9
model: initial Cobb angle (P=0.005), age at menarche (P< 0.001), BMI (P=0.002), BMD
10
(P=0.005), duration of bracing (P=0.078), duration of follow-up (P=0.014),vertebral column
11
length (P=0.071), cord-vertebral length ratio(P=0.04), AP/TS cord ratio (P=0.038) and LCS
12
ratio(P=0.009) (Table1).
M AN U
SC
RI PT
1
In final multivariate regression model, five variables were identified as the significant
14
independent predictors for curve progression during bracing including cord-vertebral length
15
ratio(Odds Ratio (OR): 1.993 [95% CI: 1.053-3.771, P=0.034]), LCS ratio (OR: 2.639 [95%
16
CI: 1.128-6.174, P=0.025]), initial Cobb angle (OR: 1.156 [95% CI: 1.043-1.281, P=0.006]),
17
menarche age (OR: 1.688 [95% CI: 1.010-2.823, P=0.046]), BMD (OR: 2.960 [95% CI:
18
1.301-6.731, P=0.010]) , and one marginally significant predictor namely AP/TS cord ratio
19
(OR: 1.463 [95% CI: 0.791-2.706, P=0.096]) were obtained (Table 2). The P value of
20
Hosmer and Lemeshow test for prediction model was 0.886. A ROC curve for prediction
21
model is shown in Figure 6. The area under the curve was 0.799 (95% CI: 0.675, 0.923). At a
22
probability cutoff of 0.661, the sensitivity and specificity of prediction model was 76.5% and
23
76.5%, respectively.
24
The probability for progression without surgery indication (P1) and for progression with
25
surgery indication (P2) could be calculated respectively by following equations:
26
AC C
EP
TE D
13
(1) Logit P1=Loge (P1/1- P1)
27
= -22.99 (Cord-vertebral length ratio) +1.941(LCS ratio)
28
+ 0.145(Cobb angle) +0.52 (menarche age)-10.852 (Femoral neck
29
BMD)-7.062 9
ACCEPTED MANUSCRIPT
1
(2) Logit P2= Loge (P2/1- P2) = -22.99 (Cord-vertebral length ratio) +1.941 (LCS ratio)
3
+ 0.145 (Cobb angle) + 0.52 (menarche age)-10.852 (Femoral neck
4 5
RI PT
2
BMD)-4.994
6
After dividing subjects into non-progression group and progression group, i.e. combining
8
Group B and C, ROC analysis showed the optimal cut-off value, corresponding sensitivity
9
and specificity for cord-vertebral length ratio was 0.73, 62.5%, 91.3%; for LCS ratio was 2.57, 37.5%, 82.6%
11
Discussion
(Table3).
M AN U
10
SC
7
This is the first prospective study to introduce MRI measurements into prognostication
13
of AIS. Both cord-vertebral length ratio and LCS ratio are found to be novel predictors for
14
curve progression, while AP/TS cord ratio is indicated as a potential predictor which
15
requires further validations. Taking above neuro-related factors into consideration may be
16
conducive to prediction of bracing effect so as to formulate treatment strategies
17
appropriately.
TE D
12
MRI is radiation free and non-invasive. It is the optimal imaging modality to quantify
19
morphology of deformed spinal cord and canal in AIS, with excellent interobserver reliability
20
(intraclass correlation > 0.9) in assessing different parameters [19, 20]. Though not a main
21
indication, MRI is able to provide similar information as plain film/ CT regarding the curve
22
classification and Cobb angle measurement with an acceptable range of error [33, 34] in AIS
23
patients who are at the age with particular concern about radiation hazard from ionizing
24
imaging.
AC C
EP
18
25
This is the first study taking into account the predictive value of spinal cord morphology
26
for curve progression in AIS. There is no generally accepted “top theory” for etiology and
27
etiopathogenesis of AIS. A variety of concepts are involved in this field, for example, relative
28
anterior spinal overgrowth(RASO)[35], uncoupled spinal neuro-osseous growth [36-39], 10
ACCEPTED MANUSCRIPT
biomechanical
spinal
growth
modulation[40],
thoraco-spinal
concept[41],
2
neurodevelopmental mechanism[42], melatonin-signaling pathway dysfunction[43, 44],
3
platelet calmodulin dysfunction[45]. MRI studies have renewed the interest of abnormal
4
neuroanatomy associated with AIS. Findings from previous cross-sectional studies advocated
5
the concept of “subclinical cord tethering”, which was an extended concept of asynchronous
6
neuro-osseous growth by Roth-Porter [20], as a possible mechanism as curve development
7
and progression in AIS. Roth postulated that when longitudinal growth of the spinal cord and
8
nerve roots failed to keep pace with the growth of the vertebral column, adaptive scoliotic
9
curvature might develop to compensate for above disproportional neuro-osseous growth[36] .
10
Porter attempted to verify Roth’s opinion with findings that overall length of spinal canal was
11
shorter in relation to summated vertebral bodies in anatomic specimens of AIS and the
12
rotation of vertebral column was around the axis of the cord [37-39]. Previous MRI studies
13
confirmed relative shortening of cord in AIS, as reflected by reduced cord-vertebral length
14
ratio, was attributed to relative anterior spinal overgrowth, whereas spinal cord in AIS did not
15
show corresponding lengthening but maintained a relatively fixed conus position regardless
16
of curve severity when compared with gender- and age-matched controls [19, 20].
17
Furthermore, increased AP/TS cord ratio was observed in AIS, which was attributed to
18
relative shorter cord undergoing a stretching-tethering force along longitudinal axis from two
19
extremities cranially and caudally, resulting in a more roundish cross-sectional shape of the
20
cord instead of the typical oval shape. The cord was also deviated towards the concavity side
21
of the scoliotic curve as an adaptation to the altered structure of the spinal canal, which
22
contributed to increased LCS ratio on an axial view [20]. All the above morphological
23
changes observed in previous cross-sectional studies were substantiated by the current
24
longitudinal study, in which cord-vertebral length ratio and LCS ratio were found to be
25
significant independent predictors for curve progression.
AC C
EP
TE D
M AN U
SC
RI PT
1
26
Cerebellar tonsil level did not turn out to be an independent prognostic factor in this study,
27
although increased incidence of tonsillar ectopia in severe AIS and positive correlation
28
between tonsil level and cord-vertebral length ratio were found[20, 21]. The possible
29
explanation is that tonsillar ectopia is itself a secondary sign of tethered cord and does not 11
ACCEPTED MANUSCRIPT
have direct impact on curve development. Furthermore, a larger cerebellar volume and a
2
smaller posterior cranial fossa in AIS might also contribute to the descent of cerebellar tonsils
3
in response to a shorter cord and relatively fixed conus position [46, 47]. Besides, more
4
significant decent and excursion of cerebellar tonsil can be detected in the upright position
5
whereas supine MRI imaging was done in current study.
RI PT
1
Several studies have reported prognostic factors for bracing outcome in AIS. The focus
7
of prognostic factors could be classified as "curve-related", "maturity-related" and
8
"bone-related" factors. Curve pattern and magnitude, age at presentation, Risser sign and
9
menarchal status were found to be prognostic factors in 1020 patients treated with
10
Milwaukee brace [15]. Sun et al [16] found that AIS patients with osteopenia were prone to
11
curve progression despite Milwaukee or Boston brace treatment, while Vijvermans et al[14]
12
supplemented a few prognostic factors including apical rotation and interval between curve
13
discovery and bracing initiation. In present study, additional neuro-related factors were
14
added into pool of predictors for bracing outcome.
M AN U
SC
6
The final independent prognostic factors revealed by our regression model comprised
16
cord-vertebral length ratio, LCS ratio, menarchal status, curve magnitude, and BMD. Our
17
prediction model was well-fitting, indicated by goodness-of-fit test (P=0.886). Meanwhile,
18
the accuracy of this prediction model was favorable, represented by an area under the ROC
19
curve of 0.799. Interestingly, age and Risser sign, which used to be considered as common
20
prognostic factors in AIS, did not emerged in our predictive model, being in agreement with
21
another study enrolling 68 AIS patients [16]. Additionally, curve pattern, which appeared to
22
be a disputable risk factor for bracing in earlier studies [16, 17], was not defined as a
23
predictor in our analysis either. Above discrepancies might be related to small number of
24
subjects in our cohort. Other possible confounding factors include non-uniform inclusion
25
criteria for bracing, inconsistent definition of curve progression and surgery indication,
26
differences in compliance level and statistic approaches.
AC C
EP
TE D
15
27
Increasing evidences have demonstrated that bracing treatment has more favorable
28
clinical outcome as compared with observation. Weinstein et al. [5] has newly investigated 12
ACCEPTED MANUSCRIPT
242 AIS patients in which the rate of treatment success (curve progression
S1529-9430(15)00199-0
DOI:
10.1016/j.spinee.2015.02.033
Reference:
SPINEE 56216
To appear in:
The Spine Journal
Received Date: 16 July 2014 Revised Date:
4 February 2015
Accepted Date: 18 February 2015
Please cite this article as: Deng M, Hui SC, Yu FW, Lam T-p, Qiu Y, Ng BK, Cheng JC, Chu WC, MRIbased Morphological Evidence of Spinal Cord Tethering Predicts Curve Progression in Adolescent Idiopathic Scoliosis, The Spine Journal (2015), doi: 10.1016/j.spinee.2015.02.033. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
MRI-based Morphological Evidence of Spinal Cord Tethering Predicts Curve Progression in Adolescent Idiopathic Scoliosis Min Deng, MMed1 , Steve CN Hui, MSc1 , Fiona WP Yu, MSc2 , Tsz-ping Lam, MBBS2,4 , MBChB, MD*1
RI PT
Yong Qiu, MD3,4 , Bobby KW Ng, MBBS2, Jack CY Cheng, MBBS, MD2,4 , Winnie CW Chu,
Department of Imaging and Interventional Radiology, The Chinese University of Hong Kong
2
Department of Orthopaedics and Traumatology, The Chinese University of Hong Kong
3
Spine Surgery, The Affiliated Drum Tower Hospital of Nanjing University Medical School,
SC
1
4
M AN U
Nanjing, China
Joint Scoliosis Research Center of the Chinese University of Hong Kong and Nanjing
University
Prof. Winnie CW Chu
TE D
*Corresponding Author:
Email address: [email protected]
EP
Mailing address: Department of Imaging and Interventional Radiology, The Chinese University
AC C
of Hong Kong, Shatin, New Territories, Hong Kong.
ACCEPTED MANUSCRIPT
MRI-based Morphological Evidence of Spinal Cord Tethering
2
Predicts Curve Progression in Adolescent Idiopathic Scoliosis
3
Abstract
4
BACKGROUND CONTEXT: Existing prognostic factors for adolescent idiopathic
5
scoliosis (AIS) patients have focused mainly on curve, maturity and bone related factors.
6
Previous studies have shown significant associations between curve severity and
7
morphological evidences of relative shorter spinal cord tethering in AIS. In conjunction with
8
increased prevalence of abnormal somatosensory cortical evoked potentials (SSEP) and
9
low-lying cerebellar tonsil in severe AIS, suggest that there might be neural morphological
M AN U
SC
RI PT
1
predictors for curve progression.
11
PURPOSE: We sought to identify any morphological predictors associated with cord
12
tethering, as measured by magnetic resonance imaging (MRI), for curve progression in AIS
13
patients.
14
STUDY DESIGN/SETTING: A prospective cohort study.
15
PATIENT SAMPLE: A total of 81 female AIS subjects between 10 and 14 years, without
16
surgical intervention during the follow up period.
17
OUTCOME MEASURES: MRI scans of hindbrain and whole spine, and areal bone mineral
18
density (BMD) at bilateral femoral necks were performed.
19
METHODS: All AIS patients were longitudinally followed up starting from initiation of
20
bracing beyond skeletal maturity in 6-monthly interval. Clinical and radiographic data were
21
recorded at each clinic visit. BMD and MRI measurements including ratio of spinal cord to
22
vertebral column length, ratio of anteroposterior (AP) and transverse (TS) diameter of cord,
23
lateral cord space (LCS) ratio, cerebellar tonsil level, and conus medullaris position were
24
obtained at baseline. Only compliant patients with a minimum 2-year follow-up were
25
analyzed. AIS girls were assigned into three groups according to bracing outcome: (A)
AC C
EP
TE D
10
1
ACCEPTED MANUSCRIPT
Non-Progression (curvature increase≤5°); (B) Progression (curvature increase≥6°); (C)
2
Progression with surgery indication (Cobb angle≥50° after skeletal maturity despite bracing).
3
The predictors for curve progression were evaluated using univariate analysis and
4
multivariate ordinal regression model.
5
RESULTS: The average duration of follow-up was 3.4 years (range: 2.0-5.6 years). There
6
were 46 girls (57%) in group A, 19 girls (23%) in group B and 16 girls (20%) in group C. No
7
significant intergroup differences were found in spinal cord length, tonsil level and conus
8
position. Group C had significantly longer vertebral column length, smaller cord-vertebral
9
length ratio, and higher AP/TS cord ratio as compared to group A, while LCS ratio in group C
10
was significantly increased when compared with both group A and group B. In regression
11
model, five significant independent predictors including cord-vertebral length ratio(Odds
12
Ratio (OR): 1.993 [95% CI: 1.053-3.771, P=0.034]), LCS ratio (OR: 2.639 [95% CI:
13
1.128-6.174, P=0.025]), initial Cobb angle (OR: 1.156 [95% CI: 1.043-1.281, P=0.006]),
14
menarche age (OR: 1.688 [95% CI: 1.010-2.823, P=0.046]), BMD (OR: 2.960 [95% CI:
15
1.301-6.731, P=0.010]), and a marginally significant predictor namely AP/TS cord ratio(OR:
16
1.463 [95% CI: 0.791-2.706, P=0.096]) were obtained.
17
CONCLUSIONS: On baseline MRI measurement, cord-vertebral length ratio and LCS ratio
18
are identified as new significant independent predictors for curve progression in AIS while
19
AP/TS cord ratio is suggested as a potential predictor requiring further validations. The above
20
MRI parameters can be taken into accounts for prognostication of bracing outcome.
21
Keywords:
22
Predictor; Spinal cord;
SC
M AN U
TE D
EP
AC C
23
RI PT
1
Adolescent idiopathic scoliosis; Magnetic resonance imaging; Bracing;
2
ACCEPTED MANUSCRIPT
1
Introduction Adolescent idiopathic scoliosis (AIS) is a three dimensional structural deformity of the
3
spine that occurs in otherwise healthy children, predominantly in adolescent girls that
4
typically progresses during the adolescent growth spurts, affecting 2%-3% of children
5
worldwide [1, 2]. Despite decades of dedicated research, the aetiopathogenesis of this classic
6
orthopaedic disorder remains uncertain. Scoliosis curves may remain static, progress slowly
7
or rapidly while the prognostic factors governing curve progression are not totally clear.
8
Severe scoliosis is associated with significant morbidities and disfiguration of body image. It
9
is a major challenge to estimate the likelihood of curve deterioration so that proper treatment
10
planning and counseling to the patients and the parents can be provided. Bracing has been
11
widely applied in AIS patients aiming for preventing curve progression so that corrective
12
surgery can be avoided [3-6]. Despite the previous controversies of bracing effectiveness
13
[7-10], recent researches including a latest multicenter large-scale study have shown that
14
bracing is superior to observation in clinical outcome[5, 11, 12]. Predictors for curve
15
progression should be redefined for AIS patients with bracing treatment, rather than
16
following predictors deriving from natural progression of curve in patients with observation.
17
Existing known predictors for curve progression despite bracing treatment in AIS are
18
mainly focused on clinical factors such as chronologic age[13], Risser sign[14], menarchal
19
status[13, 15], and radiographic factors such as curve magnitude[13, 15], curve pattern[16,
20
17], BMD[16] [18]. Recently, calcaneal stiffness index detected by quantitative ultrasound
21
(QUS) has been determined as a new predictor [13].
SC
M AN U
TE D
EP
AC C
22
RI PT
2
MRI is an excellent imaging modality for morphological delineation of soft tissue and
23
neural structures. It is now adopted in the protocol of pre-operative planning in many centers.
24
Research papers reporting abnormalities in the neuroaxis have renewed interest in abnormal
25
neuroanatomy associated with AIS. Previous cross-sectional studies have revealed significant
26
correlations between curve severity and morphological evidences of subclinical tethering of
27
relative shorter spinal cord in AIS, such as reduced spinal cord to anterior vertebral column
28
length ratio and altered cross-sectional shape of spinal cord[19, 20]. In conjunction with
29
increased prevalence of abnormal SSEP and low-lying cerebellar tonsil in severe AIS as 3
ACCEPTED MANUSCRIPT
1
compared with mild to moderate AIS [19, 21], neuro-related factors might have predictive
2
value for curve progression. To the best of our knowledge, there is no longitudinal MR study evaluating the
4
relationship between morphological evidences of relative shorter cord tethering with the
5
progression of scoliosis curve. We sought to identify any new neural morphological
6
predictors, as measured by MRI, to determine curve progression in brace-treated AIS
7
patients.
8
Materials and methods
9
Subjects
M AN U
SC
RI PT
3
During the period from July 2007 to April 2011, AIS subjects were prospectively recruited
11
from our scoliosis clinics, which was one of the only two tertiary referral centers specialized
12
in scoliosis, serving a population of seven million citizens. All subjects should be female with
13
right-sided thoracic/thoracolumbar curves with apical vertebra ranged from T6-T12. All
14
recruited subjects have been offered bracing according to following Scoliosis Research
15
Society (SRS) modified inclusion criteria: (i) age 10-14 years old; (ii) skeletal immaturity
16
(Risser sign 0-2); (iii) primary curve Cobb angle of 20° to 40°; (iv) no prior treatment; (iv)
17
either premenarchal or less than 1 year postmenarchal. Exclusion criteria included conditions
18
and medications that would affect bone remodeling and calcium metabolism, neuromuscular
19
abnormalities, genetic diseases, chromosomal defects, autoimmune disorders, endocrine
20
disturbances. Ethical approval was obtained from the University and Hospital Research
21
Ethics Committee. All subjects provided written informed consent.
22
Clinical assessment
23
Before bracing started, menarche status was recorded. Body weight, standing height, sitting
24
height, and arm span were measured with standard stadiometry techniques [22]. Corrected
25
standing height was derived from Bjure formula (logy =0.011x-0.177), where y represents the
26
loss of trunk height (cm) and x represents major Cobb angle. The body mass index (BMI) for
27
AIS patients was calculated by specific formula (body weight [kg] / square of arm span
AC C
EP
TE D
10
4
ACCEPTED MANUSCRIPT
1
[m2])[23]. According to method described in our previous study [24], areal BMD at bilateral femoral
3
neck was measured by Dual-energy X-ray absorptiometry (XR-46; Norland Medical Systems,
4
Fort Atlinson,WI, USA), with the precision errors of 1.1% to 3.7%. Skeletal maturity was
5
assessed on pelvis anterior-posterior radiographs using Risser sign grading system [25].
6
Standing radiographs of the whole spine were obtained for Cobb angle measurement while
7
curve pattern was evaluated according to King’s classification [26].
8
Bracing treatment and follow-up
9
The type of bracing was determined by curve patterns. Tailor made underarm brace by
10
experienced orthotist was used for AIS patients with major thoracic/thoracolumbar curve (Fig
11
1). Bracing treatments mainly followed the protocols described in our previous study[27].
12
Daily bracing time of 22 hours per day was initiated, followed by modification in terms of
13
temporary bracing outcome between two consecutive visits described by Yrjönen et al[28]: (i)
14
if improve (Cobb angle decrease >6°) or with a Risser sign greater than 3 and/or over-1-year
15
after menarche, shortening 2 to 4 hours per day at each follow-up. (ii) if stable (Cobb angle
16
change6°), remaining 22 hours per day. All subjects
18
underwent longitudinal follow-up at 6-month intervals beyond skeletal maturity in a special
19
clinic. Radiographic and clinical data were collected at each follow-up visit. Compliance ratio
20
was defined as the ratio of the actual daily bracing time to the recommended daily time. Only
21
AIS patients who had a minimum 2-year follow-up post bracing and a compliance ratio
22
≥75 % were analyzed.
SC
M AN U
TE D
EP
AC C
23
RI PT
2
Final Cobb angle change was calculated by subtracting Cobb angle measured on latest
24
follow-up radiographs from the baseline angle measured before implementation of bracing.
25
The bracing outcome was evaluated referring to the standardized criteria of SRS committee
26
in principle [29]. AIS girls were assigned into three groups according to bracing outcome: (A)
27
Non-Progression (curvature increase ≤5°); (B) Progression without surgery indication
28
(curvature increase ≥6°); (C) Progression with surgery indication (Cobb angle ≥50° after
29
skeletal maturity despite bracing). 5
ACCEPTED MANUSCRIPT
MRI protocol
2
All AIS patients underwent baseline MRI scan before bracing initiation. MRI data acquisition
3
was performed on 3.0 T MR scanner (Achieva TX; Philips Healthcare, Best, The Netherlands)
4
using a spine array coil. Imaging protocol included sagittal scans of hindbrain and the whole
5
spine from foramen magnum to sacrum, and transverse (TS) scans of 5 vertebral levels
6
around the vertebral apex (including the apical vertebra, 2 vertebrae above and below).
7
Above MR images were obtained using 2D Turbo spin-echo T2 weighted sequence with
8
following parameters: (i) sagittal: turbo factor=33, TR=4048 milliseconds(ms), TE=120 ms,
9
matrix=312×248, slice thickness=3.5 mm, slice gap=0 mm, field of view= 690 mm, NSA=2
10
(ii) axial: turbo factor=40, TR=645 ms, TE= 120 ms, matrix= 284×186, slice thickness=2 mm,
11
slice gap=0 mm, field of view=200 mm, NSA=1). The MRI images were reformatted on a
12
workstation (Workspace, Philips Medical System, Best, The Netherlands).
13
MRI morphologic measurement
14
The MRI morphological measurements have been reported in a number of previous
15
cross-sectional studies [19, 20]. In this study, all measurements were made by a single
16
observer (radiologist with > 10 years of experience with MRI imaging). In brief, the
17
parameters are as follows:
18
(a) Length of the anterior vertebral column and spinal cord
TE D
M AN U
SC
RI PT
1
On the reformatted straightened best mid sagittal image of the anterior vertebral column or
20
spinal cord, the total length of the anterior vertebral column was measured from the tip of
21
odontoid process of C2 down to the inferior endplate of L5 along the mid sagittal line of the
22
anterior vertebral column while length of cord was measured from the level of the tip of
23
odontoid process of C2 down to the conus medullaris (Fig 2).
24
(b) Conus position
25
The tip of conus medullaris was identified and its position was defined in relation to adjacent
26
vertebrae/disc unit which was divided into 4 segments according to the method described by
27
Saifuddin et al[30], including 3 equal portions of vertebral body (upper, middle, and lower
28
thirds) as well as the separate distal disc. From the upper third of T12 down to middle third of
AC C
EP
19
6
ACCEPTED MANUSCRIPT
L3, total 14 segments were assigned with ordinal C-values ranging from 0 to 13. For example,
2
the conus position at lower third of L1 level was represented with a C-value “6” (Fig 3).
3
(c) AP/TS cord ratio and Lateral cord space (LCS) ratio
4
The longest axis of the cord was taken for measuring maximum TS diameter at the
5
cross-sectional plane of apical vertebra in major curve, while the largest anteroposterior (AP)
6
diameter of the cord was measured along a perpendicular axis to the TS diameter (Figure 4a).
7
The AP/TS ratio was then calculated.
RI PT
1
A line was drawn bisecting the chosen vertebra by 3 standardized datum points as
9
illustrated in our previous study [20] . This line represented the vertebral axis. Parallel to the
10
vertebral axis, 4 lines were drawn from right to left (Figure 4b): (i) tangential to the inner
11
surface of the pedicle on the right/convex side, (ii) tangential to the lateral cord surface on the
12
right/convex side, (iii) tangential to the lateral cord surface on the left/concave side, and (iv)
13
tangential to the inner surface of the pedicle on the left/concave side. The distance of LCS to
14
the right side of the cord between line (i) and (ii) was defined as “A,” whereas the distance of
15
LCS to the left side of the cord between line (iii) and (iv) was defined as “B” and the cord
16
diameter “C” was defined as the distance between line (ii) and (iii). A ratio was calculated by
17
the following formula: (A+C/2) / (B+C/2) (Figure 4b), which was known as the LCS ratio-
18
convex over concave, reflecting the position of the cord.
19
(d) Cerebellar tonsil level
20
The position of cerebellar tonsil was obtained by the same method as described in our
21
previous studies [21, 31]. The level of cerebellar tonsil relative to a reference line connecting
22
the basion and opsithion (BO line) was measured in terms of the method described by
23
Aboulezz et al[32] (Fig 5).
24
Statistical Analysis
M AN U
TE D
EP
AC C
25
SC
8
Linear regression was used to test the effect of age on various MRI parameters. For
26
continuous variables with normally distribution, data were presented as mean ± standard
27
deviation, and one-way analysis of variance (ANOVA) with least significant difference (LSD)
28
test was used for group comparison (one-way ANOVA was replaced by univariate ANOVA if
29
parameter associated with age). For continuous variables with skewed or unknown 7
ACCEPTED MANUSCRIPT
1
distribution, data were presented as median with interquartile range, and Kruskal-Wallis test
2
along with Mann-Whitney U test were used for group comparison. The chi-square test was
3
used to compare the percentages among groups for categorical variables. Multivariate ordinal logistic regression analysis was performed to identify independent
5
risk factors for curve progression of AIS patients during bracing. Only variables with a
6
univariate P value of 0.05). After adjustment for age, group C
29
showed significantly longer anterior vertebral column length than group A. Group C also had 8
ACCEPTED MANUSCRIPT
significantly smaller cord-vertebral length ratio than group A. No significant intergroup
2
differences were found in spinal cord length and tonsil level. Group C showed significantly
3
higher AP/TS cord ratio and LCS ratio at apical vertebra as compared to other 2 groups. LCS
4
ratio in group A was significantly lower than group B. There was no significant difference in
5
distribution of conus position among subgroups ( χ2 test, P=0.641).
6
Predictors for AIS curve progression despite bracing
7
The average duration of follow-up was 3.4 years (range: 2.0-5.6 years). Using univariate
8
analysis, following variables with P-value less than 0.1 were eligible to enter regression
9
model: initial Cobb angle (P=0.005), age at menarche (P< 0.001), BMI (P=0.002), BMD
10
(P=0.005), duration of bracing (P=0.078), duration of follow-up (P=0.014),vertebral column
11
length (P=0.071), cord-vertebral length ratio(P=0.04), AP/TS cord ratio (P=0.038) and LCS
12
ratio(P=0.009) (Table1).
M AN U
SC
RI PT
1
In final multivariate regression model, five variables were identified as the significant
14
independent predictors for curve progression during bracing including cord-vertebral length
15
ratio(Odds Ratio (OR): 1.993 [95% CI: 1.053-3.771, P=0.034]), LCS ratio (OR: 2.639 [95%
16
CI: 1.128-6.174, P=0.025]), initial Cobb angle (OR: 1.156 [95% CI: 1.043-1.281, P=0.006]),
17
menarche age (OR: 1.688 [95% CI: 1.010-2.823, P=0.046]), BMD (OR: 2.960 [95% CI:
18
1.301-6.731, P=0.010]) , and one marginally significant predictor namely AP/TS cord ratio
19
(OR: 1.463 [95% CI: 0.791-2.706, P=0.096]) were obtained (Table 2). The P value of
20
Hosmer and Lemeshow test for prediction model was 0.886. A ROC curve for prediction
21
model is shown in Figure 6. The area under the curve was 0.799 (95% CI: 0.675, 0.923). At a
22
probability cutoff of 0.661, the sensitivity and specificity of prediction model was 76.5% and
23
76.5%, respectively.
24
The probability for progression without surgery indication (P1) and for progression with
25
surgery indication (P2) could be calculated respectively by following equations:
26
AC C
EP
TE D
13
(1) Logit P1=Loge (P1/1- P1)
27
= -22.99 (Cord-vertebral length ratio) +1.941(LCS ratio)
28
+ 0.145(Cobb angle) +0.52 (menarche age)-10.852 (Femoral neck
29
BMD)-7.062 9
ACCEPTED MANUSCRIPT
1
(2) Logit P2= Loge (P2/1- P2) = -22.99 (Cord-vertebral length ratio) +1.941 (LCS ratio)
3
+ 0.145 (Cobb angle) + 0.52 (menarche age)-10.852 (Femoral neck
4 5
RI PT
2
BMD)-4.994
6
After dividing subjects into non-progression group and progression group, i.e. combining
8
Group B and C, ROC analysis showed the optimal cut-off value, corresponding sensitivity
9
and specificity for cord-vertebral length ratio was 0.73, 62.5%, 91.3%; for LCS ratio was 2.57, 37.5%, 82.6%
11
Discussion
(Table3).
M AN U
10
SC
7
This is the first prospective study to introduce MRI measurements into prognostication
13
of AIS. Both cord-vertebral length ratio and LCS ratio are found to be novel predictors for
14
curve progression, while AP/TS cord ratio is indicated as a potential predictor which
15
requires further validations. Taking above neuro-related factors into consideration may be
16
conducive to prediction of bracing effect so as to formulate treatment strategies
17
appropriately.
TE D
12
MRI is radiation free and non-invasive. It is the optimal imaging modality to quantify
19
morphology of deformed spinal cord and canal in AIS, with excellent interobserver reliability
20
(intraclass correlation > 0.9) in assessing different parameters [19, 20]. Though not a main
21
indication, MRI is able to provide similar information as plain film/ CT regarding the curve
22
classification and Cobb angle measurement with an acceptable range of error [33, 34] in AIS
23
patients who are at the age with particular concern about radiation hazard from ionizing
24
imaging.
AC C
EP
18
25
This is the first study taking into account the predictive value of spinal cord morphology
26
for curve progression in AIS. There is no generally accepted “top theory” for etiology and
27
etiopathogenesis of AIS. A variety of concepts are involved in this field, for example, relative
28
anterior spinal overgrowth(RASO)[35], uncoupled spinal neuro-osseous growth [36-39], 10
ACCEPTED MANUSCRIPT
biomechanical
spinal
growth
modulation[40],
thoraco-spinal
concept[41],
2
neurodevelopmental mechanism[42], melatonin-signaling pathway dysfunction[43, 44],
3
platelet calmodulin dysfunction[45]. MRI studies have renewed the interest of abnormal
4
neuroanatomy associated with AIS. Findings from previous cross-sectional studies advocated
5
the concept of “subclinical cord tethering”, which was an extended concept of asynchronous
6
neuro-osseous growth by Roth-Porter [20], as a possible mechanism as curve development
7
and progression in AIS. Roth postulated that when longitudinal growth of the spinal cord and
8
nerve roots failed to keep pace with the growth of the vertebral column, adaptive scoliotic
9
curvature might develop to compensate for above disproportional neuro-osseous growth[36] .
10
Porter attempted to verify Roth’s opinion with findings that overall length of spinal canal was
11
shorter in relation to summated vertebral bodies in anatomic specimens of AIS and the
12
rotation of vertebral column was around the axis of the cord [37-39]. Previous MRI studies
13
confirmed relative shortening of cord in AIS, as reflected by reduced cord-vertebral length
14
ratio, was attributed to relative anterior spinal overgrowth, whereas spinal cord in AIS did not
15
show corresponding lengthening but maintained a relatively fixed conus position regardless
16
of curve severity when compared with gender- and age-matched controls [19, 20].
17
Furthermore, increased AP/TS cord ratio was observed in AIS, which was attributed to
18
relative shorter cord undergoing a stretching-tethering force along longitudinal axis from two
19
extremities cranially and caudally, resulting in a more roundish cross-sectional shape of the
20
cord instead of the typical oval shape. The cord was also deviated towards the concavity side
21
of the scoliotic curve as an adaptation to the altered structure of the spinal canal, which
22
contributed to increased LCS ratio on an axial view [20]. All the above morphological
23
changes observed in previous cross-sectional studies were substantiated by the current
24
longitudinal study, in which cord-vertebral length ratio and LCS ratio were found to be
25
significant independent predictors for curve progression.
AC C
EP
TE D
M AN U
SC
RI PT
1
26
Cerebellar tonsil level did not turn out to be an independent prognostic factor in this study,
27
although increased incidence of tonsillar ectopia in severe AIS and positive correlation
28
between tonsil level and cord-vertebral length ratio were found[20, 21]. The possible
29
explanation is that tonsillar ectopia is itself a secondary sign of tethered cord and does not 11
ACCEPTED MANUSCRIPT
have direct impact on curve development. Furthermore, a larger cerebellar volume and a
2
smaller posterior cranial fossa in AIS might also contribute to the descent of cerebellar tonsils
3
in response to a shorter cord and relatively fixed conus position [46, 47]. Besides, more
4
significant decent and excursion of cerebellar tonsil can be detected in the upright position
5
whereas supine MRI imaging was done in current study.
RI PT
1
Several studies have reported prognostic factors for bracing outcome in AIS. The focus
7
of prognostic factors could be classified as "curve-related", "maturity-related" and
8
"bone-related" factors. Curve pattern and magnitude, age at presentation, Risser sign and
9
menarchal status were found to be prognostic factors in 1020 patients treated with
10
Milwaukee brace [15]. Sun et al [16] found that AIS patients with osteopenia were prone to
11
curve progression despite Milwaukee or Boston brace treatment, while Vijvermans et al[14]
12
supplemented a few prognostic factors including apical rotation and interval between curve
13
discovery and bracing initiation. In present study, additional neuro-related factors were
14
added into pool of predictors for bracing outcome.
M AN U
SC
6
The final independent prognostic factors revealed by our regression model comprised
16
cord-vertebral length ratio, LCS ratio, menarchal status, curve magnitude, and BMD. Our
17
prediction model was well-fitting, indicated by goodness-of-fit test (P=0.886). Meanwhile,
18
the accuracy of this prediction model was favorable, represented by an area under the ROC
19
curve of 0.799. Interestingly, age and Risser sign, which used to be considered as common
20
prognostic factors in AIS, did not emerged in our predictive model, being in agreement with
21
another study enrolling 68 AIS patients [16]. Additionally, curve pattern, which appeared to
22
be a disputable risk factor for bracing in earlier studies [16, 17], was not defined as a
23
predictor in our analysis either. Above discrepancies might be related to small number of
24
subjects in our cohort. Other possible confounding factors include non-uniform inclusion
25
criteria for bracing, inconsistent definition of curve progression and surgery indication,
26
differences in compliance level and statistic approaches.
AC C
EP
TE D
15
27
Increasing evidences have demonstrated that bracing treatment has more favorable
28
clinical outcome as compared with observation. Weinstein et al. [5] has newly investigated 12
ACCEPTED MANUSCRIPT
242 AIS patients in which the rate of treatment success (curve progression
