J Neurosurg Pediatrics 1:456–460, 2008

Suboccipital decompression for Chiari

malformation–associated scoliosis: risk factors and time
course of deformity progression

FRANK J. ATTENELLO, M.S., MATTHEW J. MCGIRT, M.D., APRIL ATIBA, B.S.,

MURAYA GATHINJI, M.S., GHAZALA DATOO, B.S., JON WEINGART, M.D.,
BENJAMIN CARSON, M.D., AND GEORGE I. JALLO, M.D.
Department of Neurosurgery, The Johns Hopkins School of Medicine, Baltimore, Maryland

Object. Chiari malformation Type I (CM-I) is often associated with scoliosis. It remains unclear which subgroups
of patients are most likely to experience progression of spinal deformity after cervicomedullary decompression. The
authors’ goal was to determine the time frame of curvature progression and assess which patient subgroups are at great-
est risk for progression of spinal deformity after surgery.
Methods. The authors retrospectively reviewed the records of all pediatric patients with significant scoliosis in
whom suboccipital decompression was performed to treat for CM-I during a 10-year period at a single academic insti-
tution. Clinical, radiological, and operative variables were assessed as independent factors for failure (worsening of
scoliosis) by using a univariate regression analysis.
Results. Twenty-one children (mean age 9 6 3 years; 4 male) underwent hindbrain decompression for CM-I–asso-
ciated scoliosis and were followed for a mean of 39 months. All patients harbored a syrinx. Eight patients (38%) expe-
rienced improvement in scoliosis curvature, whereas 10 (48%) suffered a progression. Thoracolumbar junction sco-
liosis (p = 0.04) and failure of the syrinx to improve (p = 0.05) were associated with 5- and 4-fold respective increases
in the likelihood of deformity progression. Each increasing degree of preoperative Cobb angle was associated with an
11% increase in the likelihood of scoliotic curve progression (p , 0.05).
Conclusions. Over one third of patients with CM-I–associated scoliosis will improve after cervicomedullary decom-
pression alone. Cervicomedullary decompression is a good first-line option, particularly in children with concordant
posterior fossa symptoms. Patients presenting with more severe scoliosis (increasing Cobb angle) or scoliosis that cross-
es the thoracolumbar junction may benefit from earlier orthopedic involvement and should be monitored regularly for
curvature progression after cervicomedullary decompression. In cases in which there is a failure of the syrinx to show
improvement after suboccipital decompression, the patients are also more likely to develop curvature progression.
(DOI: 10.3171/PED/2008/1/6/456)

KEY WORDS • Chiari malformation • outcome • predictors • scoliosis

malformation Type I, defined as caudal dis-
C
HIARI
placement of the cerebellar tonsils into the cervical
canal, was first documented by Hans Chiari4 in
1891. Over the last century, multiple symptoms of cere-
bellar, brainstem, and spinal cord pathology have been at-
tributed to this complex disease.11,13,14 Standard surgical
management for CM-I remains posterior fossa decompres-
sion.2,5,10 Up to 30% of all patients with CM-I and 60% of
patients with Chiari malformation–associated syringomy-
elia will present with scoliosis.1,3,6–9,12,13
The authors of previous studies have reported a wide
range of outcomes after hindbrain decompression for Chiari
malformation–associated scoliosis with 20–70% of patients
experiencing curve progression.1,3,6–9,12 Reports of improve-
Abbreviations used in this paper: CI = confidence interval; CM-I =
Chiari malformation Type 1; HR = Hazard ratio.
ment in scoliosis after hindbrain decompression also range
widely from 20 to 60%.1,3,6–9,12 Hence, there is no consensus
on the absolute risk and time frame of scoliosis progression
after hindbrain decompression. Furthermore, it remains un-
clear which subgroups of patients are most likely to suffer
progression of their spinal deformity after cervicomedullary
decompression. We set out to determine the time frame of
curvature progression and assess which patient subgroups
are at greatest risk for progression of spinal deformity post-
operatively.
Clinical Materials and Methods
In 21 consecutive patients undergoing posterior fossa de-
compression for CM-I–associated scoliosis at The Johns
Hopkins Hospital between 1995 and 2005, we reviewed pre-
senting symptoms, neurological deficits, demographic data,

456 J. Neurosurg.: Pediatrics / Volume 1 / June 2008

Suboccipital decompression for CM-I–associated scoliosis
comorbidities, pre- and postoperative radiological studies,
operative records, and follow-up clinical records. An elec-
tronic database was created by inputting patients’ demo-
graphic information and presenting symptoms. Additionally,
we also identified and recorded the presence and location of
syringomyelia and scoliosis on magnetic resonance images
or plain radiographs. All patients in this group were offered
surgical decompression if we documented the following: 1)
cerebellar tonsillar herniation $ 5 mm below the foramen
magnum and either 2) hindbrain symptoms consistent with
CM-I (tussive headache, cervical pain, central apnea, dys-
phagia, aspiration, vertigo, vocal cord paralysis, motor/sen-
sory deficits, nystagmus, ataxia, uncoordination, and syrin-
gomyelia) or 3) syringomyelia manifesting with a scoliosis
curvature or Cobb angle . 10° on standing radiographs. Pa-
tients with minimal ectopia (. 5 mm below the foramen
magnum but rostral to C-1) underwent suboccipital decom-
pression only if they experienced tussive, reproducible head-
ache in the presence of a syrinx and scoliosis. All patients
underwent suboccipital decompression of the foramen mag-
num and C-1 laminectomy. The majority of patients also
underwent duraplasty. Duraplasty was not performed in pa-
tients with minimal tonsillar ectopia and evidence of phys-
iological hindbrain cerebrospinal fluid flow and tonsillar
pulsations on intraoperative ultrasonography after decom-
pression. Coagulation of the tonsil was only performed when
tonsillar herniation was observed intraoperatively to extend
below C-2. Patients who underwent spinal fusion prior to or
during the CM-I decompression were excluded from this
analysis. As a result, no patients in our study had spinal fu-
sion either before or at the same time as CM-I decompres-
sion. Hence, this series represents patients presenting with
Chiari malformation–associated syringomyelia whose treat-
ment consisted solely of hindbrain decompression and ob-
servation.
The location of the deformity was defined as the follow-
ing: with the apex at T2–10 for thoracic scoliosis, at T10–L1
for thoracolumbar scoliosis, and at L1–4 for lumbar scolio-
sis. Improvement or worsening in spinal curvature was de-
termined by assessing postoperative radiographs of standing
plain radiographs, with a change in the Cobb angle of . 10°
regarded as a significant change in spinal curvature. Im-
provement or worsening in syringomyelia was also evaluat-
ed on postoperative MR images, with . 20% change in syr-
inx size regarded as significant.
This CM-I database was retrospectively analyzed to as-
sess the independent association of presenting symptoms,
physical examination findings, radiological variables, and
operative details with the postoperative progression of sco-
liosis; this was done using Kaplan–Meier curves and log-
rank analysis for stratified covariates as well as Cox propor-
tional hazards analysis for continuous covariates. Mean data
are presented 6 the standard deviation.
Results
Patient Population
Among 258 pediatric patients undergoing first-time pos-
terior fossa decompression for CM-I, 33 patients (13%) pre-
sented with CM-I–associated scoliosis during the study pe-
riod. Two patients (6%) underwent fusion planned at the
time of CM-I decompression, 2 (6%) underwent planned fu-
J. Neurosurg.: Pediatrics / Volume 1 / June 2008
sion after posterior fossa decompression, and 7 (21%) un-
derwent follow-up at an outside institution, with 1 case (3%)
removed due to the presence of CM-I and scoliosis without
a syrinx. Patients in whom fusion was planned during or af-
ter posterior fossa decompression had severe (. 40° Cobb
angle) and progressive scoliosis in the previous year. The 21
remaining patients continued to undergo follow-up at our in-
stitution and were included in the present study. Mean age at
time of surgery was 9 6 3 years, and 4 patients (19%) were
male. Presenting symptoms included hindbrain headache in
5 patients (24%; mean duration 12 months) and symptoms
of syrinx or scoliosis alone (no headache) in 16 (76%; mean
duration 20 months). Presenting symptoms are summarized
in Table 1. Tonsillar herniation was . 5 mm below the fora-
men magnum but rostral to C-1 in 1 patient (5%; mild ec-
topia), between C-1 and C-2 in 19 patients (90%; moderate
ectopia), and below C-2 in 1 patient (5%; severe ectopia).
Syringomyelia was present in all patients, involving the
cervical spine in 19 (90%), the thoracic spine in 20 (95%),
and the conus medullaris in 4 (19%) patients. When a syrinx
was present, it spanned a mean 9 6 4 spinal levels. One
patient (5%) had cervical scoliosis, 10 (48%) had thoracic
scoliosis, 7 (33%) had thoracolumbar scoliosis, and 3 (14%)
had lumbar scoliosis. The mean Cobb angle of the scoliotic
curves was 28 6 8°.
Only 1 patient did not undergo duraplasty, and in this case
scoliosis progression was observed by 9 months postopera-
tively and there was lack of syrinx improvement.
Imaging Follow-Up
All patients underwent preoperative radiography at an
average of 2 6 1 months before surgery. Patients routinely
underwent standing radiography 3 months after surgery,
with repeated radiography every 6 months unless symptom
change prompted earlier imaging. Fourteen patients (67%)
underwent their first postoperative radiographic studies
within 4 months of surgery, with the remaining patients un-
dergoing their radiological studies between 4 and 6 months
after surgery. Twelve patients (57%) underwent a second
postoperative radiographic examination within 1 year of sur-
TABLE 1
Summary of presenting symptoms and signs in 21
consecutive pediatric patients with scoliosis and
syringomyelia undergoing surgical decompression for CM-I*
Variable No. of Patients (%)
male sex 4 (19)
headache 5 (24)
cranial nerve or scoliosis symptoms alone 16 (76)
sensory deficiency 3 (14)
motor deficiency 3 (14)
scoliosis
cervical 1 (5)
thoracic 10 (48)
lumbar 3 (14)
curve crosses CTJ 0 (0)
curve crosses TLJ 7 (33)
syrinx
cervical 19 (90)
thoracic 20 (95)
conus medullaris 4 (19)
* CTJ = cervicothoracic junction; TLJ = thoracolumbar junction.
457
 F. J. Attenello et al.

TABLE 2
Summary of pre- and postoperative characteristics
Case Scoliosis
No. Age at Surgery (yrs) Preop Cobb Angle (°) Curvature Outcome

1 4 35 improvement
2 8 24 improvement
3 8 19 improvement
4 8 22 improvement
5 10 25 improvement
6 10 32 improvement
7 11 33 improvement
8 12 20 improvement
9 8 27 no change
10 10 23 no change
11 13 48 no change
12 5 25 progression
13 6 20 progression
14 7 20 progression
15 9 21 progression
16 10 32 progression
17 10 26 progression
18 11 38 progression
19 11 37 progression
20 13 46 progression
21 14 35 progression

elia at an average of 8 months. Failure of the syrinx to de-

FIG. 1. Kaplan–Meier plots of time of onset of curve improve-
ment (upper) and time of onset of curve worsening (lower), indi-
cated by Cobb angle on standing radiographs, as a function of time
after hindbrain decompression in children with CM-I. Eight patients
(38%) experienced improvement in their scoliosis curvature, and 10
(48%) experienced scoliosis progression postoperatively. Of those
patients with improvement, 50% exhibited improvement by 6
months, 88% by 2 years, and all patients by 3 years. Of those pa-
tients in whom the deformity worsened, 60% exhibited worsening
at 1 year, 80% by 2 years, and all patients by 4 years.
gery, and in the remaining patients the studies were per-
formed between 12 and 18 months postoperatively. In 6 pa-
tients (29%) standing radiography was conducted . 2 years
after surgery.
The mean follow-up duration was 39 6 36 months (range
4–117 months). Only 1 patient underwent follow-up for ,
6 months (4 months). All other patients were followed up to
2 years. Fifteen patients (71%) were followed for . 2 years.
Nine patients (43%) were followed for at least 3 years.
Outcome and Predictors of Treatment Failure
In 8 patients (38%) the scoliotic curvature improved after
surgery, whereas in 10 (48%) the deformity progressed post-
operatively (Fig. 1, Table 2). In 3 patients (14%) there was
no change in scoliotic curvature at last follow-up examina-
tion. Three (30%) of 10 patients with deformity progression
did eventually require spinal fusion during the study period.
Fourteen patients (67%) exhibited improved syringomy-
crease in size was associated with a 4-fold likelihood of sco-
liosis progression (HR 3.80 [95% CI 1.0–14.5], p = 0.05)
(Fig. 2, Table 3). Syrinx improvement preceded curvature
improvement by an average of 7 months.
Scoliosis with apex at the thoracolumbar junction was
associated with a 5-fold increase in the likelihood of curve
progression (HR 4.75 [95% CI 1.06–21.3], p = 0.04) (Fig.
2, Table 3). In addition, each increasing degree of Cobb
angle was associated with an 11% increase in the likelihood
of curve progression (p , 0.05). Duraplasty was not per-
formed in 1 patient and was not assessed as a predictor of
outcome based on sample size.
Discussion
In our experience of hindbrain decompression as the ini-
tial treatment for CM-I–associated scoliosis, we observed
that scoliosis improved radiographically in more than one
third of patients after hindbrain decompression alone. Pa-
tients presenting with an increasing Cobb angle or scoliosis
crossing the thoracolumbar junction were at significantly
greater risk of curve progression. Failure of the syrinx to im-
prove was also associated with curve progression. The rel-
ative risk of scoliosis progression was greatest between 6
and 12 months after surgery with 29% of the deformities
progressing by 12 months. After the 1st year, the estimated
annual risk of progression was decreased, with 10% of the
cases developing curve progression over each of the next 2
years, reaching 49% by 36 months. Improvement in sco-
liosis was noted in one third of patients. The curvature im-
proved within the first 36 months of hindbrain decompres-
sion in all patients. Curvature improvement occurred at
a similar incidence within the 1st and 2nd year postopera-
tively.
Reports on progression of scoliosis after hindbrain de-
compression for CM-I remain limited. Farley et al.7 report-
ed on 9 patients who underwent decompression, 8 of whom

458 J. Neurosurg.: Pediatrics / Volume 1 / June 2008

Suboccipital decompression for CM-I–associated scoliosis
FIG. 2. Kaplan–Meier plots showing the incidence of curve progression onset as a function of time after hindbrain de-
compression for CM-I in children. Lack of syrinx improvement (p = 0.05) (upper) and presence of scoliosis at the thora-
columbar junction (p = 0.04) (lower) were associated with an increased risk of scoliosis curvature progression. T-L Jxn =
thoracolumbar junction.
were candidates for fusion by the end of the follow-up peri-
od. The study had a small sample size with a high mean
preoperative thoracic deformity (46 6 18°). In their study
involving hindbrain decompression, Hida et al.9 reported
improvement of the curvature in 38%, stabilization in 38%,
and progression in 23%. Brockmeyer et al.3 found that pa-
tients who were male, under the age of 10 years, and with
preoperative curves , 40° were more likely to improve af-
ter decompressive surgery. Eule et al.6 described a study in
which the vast majority of patients , 8 years of age under-
going decompression alone experienced improvement or
stabilization of their curves. Bhangoo and Sgouros1 report-
ed that age , 10 and preoperative curve , 30° were asso-
ciated with decreased need for additional scoliosis surgery.
The mean age and the mean Cobb angle of patients not re-
quiring further surgery were 10 years and 29°, respectively,
compared with 13 years and 76° in those in whom addition-
al surgery was required for deformity progression. Ghanem
and associates8 found that preoperative curves . 40° were
indicative of a worse outcome, but they found no correla-
tion with age. Although the authors of the aforementioned
studies described age and greater preoperative Cobb angles
as predictors of subsequent curve progression, there is not a
strong consensus as to the multiple variables that can iden-
tify high-risk patients outside of age and initial curve angle.
We found that an increased degree of scoliosis preopera-
J. Neurosurg.: Pediatrics / Volume 1 / June 2008
tively was a predictor of scoliosis progression after hind-
brain decompression. Our observations are similar to those
previously reported that suggest that either a Cobb angle .
30°1,3 or . 40°8 is a predictor of curvature progression. Fur-
thermore, we found that the spinal level involved in scolio-
sis was also indicative of postsurgical outcomes. Based on
Kaplan–Meier estimates, 60% of patients with thoraco-
lumbar scoliosis had curve progression within the first 12
months, and this rose to 80% by 24 months.
We also found a significant correlation between failure of
the syrinx to resolve and progressive scoliosis curvature
postoperatively. Of 7 patients with a decrease in syrinx size
and scoliotic curve, 57% exhibited a decrease in syrinx size
prior to scoliotic change and 43% exhibited this syrinx de-
crease concurrently with improvement in scoliosis. It is not
surprising that the decrease in syrinx size correlates with
scoliosis improvement, as the syrinx is likely to be the cause
of the scoliotic curve, and previous studies have commonly
noted the occurrence of scoliosis in up to 60% of patients
with syrinx.1,3,6–9,12
In our scoliosis cohort there was a predominance of fe-
males, constituting 80% of our series. This was a natural
variant of our population, as we did not include or exclude
patients on the basis of sex. In addition, our series demon-
strates that, among patients undergoing primary decompres-
sion for CM-I, 13% present with concurrent scoliosis. Re-
459
 F. J. Attenello et al.

TABLE 3 Disclosure
Univariate association of clinical, radiographic, and
This study was supported by a grant from the Congress of Neuro-
surgical variables with postoperative worsening of logical Surgeons and the Syringomyelia Alliance Project.
scoliosis curvature following decompression for CM-I*
Variable† HR (95% CI) p Value
References
female sex 1.22 (0.27–5.53) 0.794
age 1.12 (0.82–1.52) 0.488 1. Bhangoo R, Sgouros S: Scoliosis in children with Chiari I-related
nausea/vomiting 3.16 (0.37–27.2) 0.294 syringomyelia. Childs Nerv Syst 22:1154–1157, 2006
incontinence 0.61 (0.24–8.48) 0.702 2. Bindal AK, Dunsker SB, Tew JM Jr: Chiari I malformation: clas-
sensory deficiency 1.60 (0.18–14.1) 0.670 sification and management. Neurosurgery 37:1069–1074, 1995
motor deficiency 3.07 (0.35–27.1) 0.312 3. Brockmeyer D, Gollogly S, Smith JT: Scoliosis associated with
tonsillar coagulation 0.89 (0.10–8.12) 0.915 Chiari 1 malformations: the effect of suboccipital decompression
no. of syrinx levels 1.02 (0.91–1.14) 0.746
on scoliosis curve progression: a preliminary study. Spine 28:
ectopia below C-2 3.16 (0.37–27.2) 0.294
no. of scoliosis levels 1.12 (0.90–1.39) 0.304
2505–2509, 2003
curve across TLJ 4.75 (1.06–21.3) 0.042 4. Chiari H: Über Veranderungen des Kleinhirns infolge von Hy-
Cobb angle 1.11 (1.00–1.24) 0.049 drocephalie des Grosshirns. Dtsch Med Wochenshr 17:1172–
lack of syrinx improvement 3.80 (1.00–14.5) 0.050 1175, 1891
5. Elster AD, Chen MY: Chiari I malformations: clinical and radio-
* Scoliosis crossing the thoracolumbar junction and failure of syrinx to logic reappraisal. Radiology 183:347–353, 1992
decrease . 20% at follow-up were associated with a 5- and 4-fold increase
in the likelihood of curve progression. Each increasing degree of Cobb angle 6. Eule JM, Erickson MA, O’Brien MF, Handler M: Chiari I mal-
curvature was associated with an 11% increase in the likelihood of curve formation associated with syringomyelia and scoliosis: a twenty-
progression. year review of surgical and nonsurgical treatment in a pediatric
† Variables associated with 0 or 1 case are not cited in the table. population. Spine 27:1451–1455, 2002
7. Farley FA, Puryear A, Hall JM, Muraszko K: Curve progression in
scoliosis associated with Chiari I malformation following suboc-
ported rates of scoliosis vary in the literature. In a series of cipital decompression. J Spinal Disord Tech 15:410–414, 2002
130 patients, Tubbs et al.13 demonstrated that 18% of those 8. Ghanem IB, Londono C, Delalande O, Dubousset JF: Chiari I
with CM-I presented with concurrent scoliosis. It is not clear malformation associated with syringomyelia and scoliosis. Spine
why our series shows a lower percentage of patients with 22:1313–1318, 1997
scoliosis, although it may be due to regional differences or 9. Hida K, Iwasaki Y, Koyanagi I, Abe H: Pediatric syringomyelia
our referral pattern. These differences may warrant further with chiari malformation: its clinical characteristics and surgical
investigation. outcomes. Surg Neurol 51:383–391, 1999
Several limitations of this study should be recognized. It 10. Meadows J, Kraut M, Guarnieri M, Haroun RI, Carson BS:
is retrospect in design and with nonstandardized follow-up Asymptomatic Chiari Type I malformations identified on mag-
netic resonance imaging. J Neurosurg 92:920–926, 2000
intervals. It also is composed of a small sample size (21 pa- 11. Milhorat TH, Chou MW, Trinidad EM, Kula RW, Mandell M,
tients). Hence, the study is underpowered to draw definitive Wolpert C, et al: Chiari I malformation redefined: clinical and ra-
negative conclusions. However, despite this small sample diographic findings for 364 symptomatic patients. Neurosurgery
size, the statistical significance noted for increasing Cobb an- 44:1005–1017, 1999
gle, scoliosis crossing the thoracolumbar junction, and lack 12. Sengupta DK, Dorgan J, Findlay GF: Can hindbrain decompres-
of syrinx resolution highlights the strong correlation of these sion for syringomyelia lead to regression of scoliosis? Eur Spine
variables with outcome. J 9:198–201, 2000
13. Tubbs RS, McGirt MJ, Oakes WJ: Surgical experience in 130 pe-
diatric patients with Chiari I malformations. J Neurosurg 99:291–
Conclusions 296, 2003
Over one third of patients with Chiari malformation–asso- 14. Weinberg JS, Freed DL, Sadock J, Handler M, Wisoff JH, Epstein
FJ: Headache and Chiari I malformation in the pediatric popula-
ciated scoliosis will improve after cervicomedullary decom- tion. Pediatr Neurosurg 29:14–18, 1998
pression alone. Cervicomedullary decompression is thus a
good first-line option particularly in children with concor-
dant posterior fossa symptoms. Patients presenting with
more severe scoliosis (increasing Cobb angle), lack of syrinx
improvement during follow-up, or scoliosis crossing the tho- Manuscript submitted August 14, 2007.
racolumbar junction may benefit from earlier orthopedic in- Accepted February 29, 2008.
volvement and should be monitored regularly for curve pro- Address correspondence to: Matthew J. McGirt, M.D., 3553 New-
gression after cervicomedullary decompression. land Road, Baltimore, Maryland 21218. email: mmcgirt1@jhmi.edu.

460 J. Neurosurg.: Pediatrics / Volume 1 / June 2008