Journal of Clinical Neuroscience 16 (2009) 666–671

Contents lists available at ScienceDirect

Journal of Clinical Neuroscience

journal homepage: www.elsevier.com/locate/jocn

Clinical Study

Microsurgical treatment and functional outcomes of multi-segment

intramedullary spinal cord tumors
Jianjun Sun *, Zhenyu Wang, Zhendong Li, Bin Liu
Department of Neurosurgery, Peking University Third Hospital, No. 49 Huayuan North Road, Haidian District, Beijing 100083, China

a r t i c l e i n f o a b s t r a c t

Article history: We aimed to prospectively analyze correlations between clinical features and histological classification of
Received 1 June 2008 multi-segment intramedullary spinal cord tumors (MSICTs), and the extent of microsurgical resection
Accepted 5 August 2008 and functional outcomes. Fifty-six patients with MSICTs underwent microsurgery for tumor removal
using a posterior approach. The tumor was exposed through a dorsal myelotomy. Pre-operative and
post-operative nervous function was scored using the Improved Japanese Orthopaedic Association (IJOA)
Keywords: grading system. Correlation analyses were performed between functional outcome (IJOA score) and his-
Spinal cord
tological features, age, tumor location, and the longitudinal extent of spinal cord involvement. The most
Intramedullary tumor
Multi-segment
frequently involved levels were the medullo cervical and the cervicothoracic regions (51.8%, 29/56) fol-
Microsurgery lowed by the conus terminalis (26.8%, 15/56) and the thoracic region (14.3%, 8/56). Ependymoma was the
Outcome most frequent MSICT type, seen in 22 of 56 patients (39%), followed by low grade astrocytoma (17
patients, 30%) and glioblastoma multiforme (3 patients, 5%). Gross total tumor removal was achieved
in 33 cases (58%), subtotal resection in 4 (7%), and partial resection in 16 (28%). The histological classifi-
cation of the tumor was the most important factor influencing the extent of surgical removal (v2 = 22.17,
p = 0.00). The overall difference between pre-operative and post-operative neurological state was not sig-
nificant (v2 = 5.44, p = 0.61). Thus, MSICTs were most commonly seen in the medullo cervical and cervi-
cothoracic regions, with ependymoma and low grade astrocytoma the most common tumour types. We
stress the importance of early microsurgical treatment for MSICTs while the patients do not have severe
dysfunction.
Crown Copyright Ó 2008 Published by Elsevier Ltd. All rights reserved.

1. Introduction 2. Materials and methods

The incidence of spinal tumors is less than that of intracranial This was a prospective, descriptive study using clinical records.
tumors, with an overall incidence of about one spinal tumor for Ethics approval from the Research Ethics Board of Peking Univer-
every four intracranial lesions. The incidence of multi-segment sity Third Hospital was obtained to conduct this study.
intramedullary spinal cord tumors (MSICTs) is much less (about
1 MSICT for every 10 spinal tumors).1,2 Although usually not life
2.1. Patients
threatening, MSICTs cause permanent disability in patients, and
consequently also place large burdens on their families and on
A consecutive series of 56 patients diagnosed with MSICTs
society. In the past, total removal of MSICTs was rarely attempted
extending for 3 or more spinal segments were referred to our
because of the difficulty of the operative procedure, and the wors-
institution and underwent microsurgical treatment.2,3 Basic demo-
ening post-operative neurological status.2,3 With the advance of
graphic data, clinical and radiological presentation, and intraoper-
modern microsurgical techniques, it is now possible to completely
ative observations were evaluated. Pre-operative neuroimaging,
resect MSICTs without worsening the post-operative neurological
including MRI, was performed in all cases. Spinal angiography
status. For this reason, total removal has now become the gold
was performed if the MRI suggested a possible vascular lesion.
standard for the treatment of MSICTs where possible. In this article,
we prospectively analyzed 56 patients presenting with MSICTs
who received microsurgery at this hospital from January 2002 to 2.2. Surgical technique
October 2007.
All 56 patients underwent their first operation at our institu-
* Corresponding author. Tel.: +86 10 62017358; fax: +86 10 62017700. tion. The operation followed the usual procedures for intramedul-
E-mail address: sunjianjun_2008@yahoo.com.cn (J. Sun). lary spinal cord tumors. Laminectomy was performed over the

0967-5868/$ - see front matter Crown Copyright Ó 2008 Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.jocn.2008.08.016
 J. Sun et al. / Journal of Clinical Neuroscience 16 (2009) 666–671
region of the tumor on the basis of pre-operative neuro-diagnostic
testing. Only medial facet joint exposure was required in most
cases, and an effort was made to preserve the facet joint capsule
in all cases. After midline dural incision, the operating microscope
was brought into the field and a midline myelotomy was performed
and pia traction obtained using sutures to the dura. The resection
was then modified based on the tumor dissection plane, and infil-
tration as determined by pre-operative MRI and the appearance un-
der intra-operative microscopy. Intra-operative neurophysiological
monitoring was regularly used. The pia mater was not normally
reattached at the end of the resection. For spinal ependymomas
(Fig. 2), midline myelotomy was carried out to expose the proximal
and distal ends of the tumor mass. Spinal ependymomas originate
from the central canal, grow concentrically, and demonstrate a
clear plane to normal cord tissue. Ultrasonic aspiration was used
to resect the tumor to its interface with the normal cord white mat-
ter. This dissection plane between the tumor mass and normal cord
tissue was relatively easy to preserve in most cases.
In spinal astrocytic tumors, in contrast, tumor extension is typ-
ically eccentric without any defined plane to normal cord tissue.
Due to this intramedullary eccentric mass, the posterior median
sulcus is often shifted to the opposite side. Midline myelotomy
was used cautiously to identify the posterior median sulcus. The
malignant astrocytic tumors in this series, including anaplastic
astrocytomas (Fig. 3) and glioblastomas, mostly invaded through
the pia mater or arachnoid. After the extramedullary tumor was
removed, the opening of the spinal cord was extended in the midline.
The tumor was dissected along the cleavage plane with the spinal
cord as much as possible. In areas of obvious infiltration, the tumor
was removed layer by layer starting from the innermost layers until
white matter was identified. In cases where malignant astrocytoma
was suspected at frozen biopsy, partial removal was performed and
complete resection was not attempted. Radiotherapy was then
performed.
Some well-defined MSICTs (including hemangioblastoma
(Fig. 1) and schwannoma) may partially invade the spinal cord
and extrude through the pia mater. In these cases, the draining
veins were carefully displaced from the field before the feeding
arteries were clipped and then obstructed with bipolar coagula-
tion. Tumors were partially dissected and not totally removed
where necessary. In patients with hemangioblastomas the tumor
cyst was not removed. In order to prevent the contents of terato-
mas from seeding the subarachnoid space, well-defined intramed-
ullary cystic teratomas were shielded with cotton slips before
being cut open. The cyst walls of tumors that adhered tightly
to the spinal cord were cauterized with micro-power bipolar
coagulation, rather than separated and removed by force. Lipomas
between the spinal cord and nerve root were debulked and decom-
pressed partially; total resection was not attempted.
The extent of tumor removal was graded using four classes
according to intraoperative observations and post-operative MRI.
Cases with no residual enhancement on post-operative MRI were
classified as class 1. Subtotal resection of about 80 to 90 percent
was defined as class 2. Partial resection of about 60% to 80% of
the tumor was defined as class 3. All other procedures, including
decompression and biopsy, were classified as class 4.
The Improved Japanese Orthopaedic Association (IJOA) grading
system was used to evaluate pre-operative and post-operative ner-
vous function. IJOA is based on the JOA grading system, with the
addition of scoring stool function as normal, slight dysfunction, se-
vere dysfunction, or incontinent. Functional outcome was defined
as post-operative IJOA score minus pre-operative IJOA score. The
clinical outcomes were independently analyzed by an observer
who was blind to other features.
The first post-operative follow-up MRI was done at two weeks
post-operative and the second at 6 months. MRI was then repeated
667
annually in order to detect disease progression in cases without
symptomatic change. If patients experienced symptomatic change,
then MRI was performed immediately. Disease progression was
defined as recurrence in cases of complete removal, and regrowth
in cases of incomplete removal.
2.3. Post-operative management
Post-operatively all patients received routine administration of
methylprednisolone with a flushing dose (about 500–1000 mg/d
according to body weight) for three days, and vitamin B1 and
B12 for at least two months. A neck collar or waistline brace was
used to fix position and prevent vertebral deformation. All patients
received rehabilitation training at local physical therapy centers.
The condition of patients was evaluated according to the func-
tional status and graded as improved, unchanged, or worsened.
2.4. Statistical analysis
To compare the two groups, the data was analyzed by cross
table studies with the p-values generated using either the
chi-squared test or Wilcoxon Signed Rank test. Data analysis was
performed using a computer-based statistical program, Statistical
Package for the Social Sciences v. 11.0 for Windows (SPSS, Chicago,
IL, USA). Data were expressed as the mean ± standard error (s.e.).
P values of less than 0.05 were considered significant.
3. Results
3.1. Clinical factors
Of the 56 patients, 34 were male (61%) and 22 were female
(39%). Children less than age 10 years represented 6 cases, ages
10–19 years 6 cases, ages 20–29 years 13 cases, ages 30–39 years
13 cases, ages 40–49 years 8 cases, and ages 50 years or older 10
cases (mean age, 32.6 ± 15.6 years, Table 1).
The most common initial symptom was pain (52%, 29/56), fol-
lowed by sensory disturbance (23%, 13/56), motor weakness
(14%, 8/56), gait deterioration (9%, 5/56) and sphincter dysfunction
(2%, 1/56) (Table 1). Age and initial presentation have prognostic
significance (v2 = 15.4, p = 0.02). Adolescent patients tended to
present with subtle changes in gait or more frequent falls. Older
patients mostly presented with sensory disturbance and weakness.
The most commonly involved location was the medullo cervical
and the cervicothoracic segments (51.8%, 29/56), followed by the
conus terminalis (26.8%, 15/56), the thoracic region (14.3%, 8/56),
and the lumbosacral region (7.1%, 4/56). Three spinal segments
were involved in 20 (35.7%) cases, four in 16 (28.6%) cases, five
in 10 (17.8%) cases, and more than five in 10 (17.9%) cases.
The most frequent MSICTs were ependymomas (39.3%, 22/56),
followed by astrocytomas (30.3%, 17/56, grade I-II in 14 cases,
grade III in 3 cases), cystic teratomas (10.7%, 6/56), glioblastomas
(5.4%, 3/56), lipomas (5.4%, 3/56), hemangioblastomas (5.4%, 3/
56), and schwannomas (3.5%, 2/56). The ependymomas and astro-
cytomas mostly involved the cervical spinal cord and the terato-
mas mostly involved the medullary conus. The location of
MSICTs did not predict histology (v2 = 7.51, p = 0.28).
3.2. Surgical outcomes
Gross total tumor removal was achieved in 33 cases (58.9%),
subtotal resection in 4 (7.1%), and partial resection in 16 (28.6%).
Owing to the absence of a clear dissection plane in three patients
(5.4%), only a decompression or biopsy was performed (one glio-
blastoma, one ependymoma, and one teratoma). The histological
668 J. Sun et al. / Journal of Clinical Neuroscience 16 (2009) 666–671

Table 1 formed in two patients following complete resection of giant

The relationship between age at presentation and initial symptoms ependymomas (the length of enhancement of the tumors was
Age Initial symptoms (no. patients/ n = 56) Total more than 20 cm) involving the thoracolumbar segment (Fig. 4).
(years)
Gait Pain Sphincter Sensory Weakness
deterioration dysfunction disturbance 3.3. Follow-up results
<10 3 3 0 0 0 6
10–19 0 5 1 0 0 6 The mean follow-up period was 28.6 ± 20.82 months (range 3–
20–29 2 6 0 2 3 13 72 months). Compared to the pre-operative neurological status,
30–39 0 8 0 3 2 13 the overall post-operative status at last follow-up was improved
40–49 0 3 0 4 1 8
in 83.9% patients (n = 47), unchanged in 10.7% patients (n = 6)
50–59 0 2 0 7 0 9
>60 0 0 0 0 1 1 and worse in 5.4% patients (n = 3, all diagnosed with malignant
Total 5(8.9%) 29(51.8%) 1(1.8%) 13(23.2%) 8(14.3%) 56 astrocytoma). In 37 patients with total or subtotal tumor resection,
v2 15.4 36 (97.2%) enjoyed a complete recovery. Half a year after the oper-
p value 0.02 ation, the patient in Fig. 1 and 3 had returned to school. A year after
the operation, the patient in Fig. 2 was commuting to work on a
motorcycle. Two years after the operation, the patient in Fig. 4
characteristics of the tumors were the most important factors had returned to work and married. In 16 patients with partial tu-
influencing the extent of surgical removal (v2 = 22.17, p < 0.01). mor resection, 10 (62.5%) retained minimal dysfunction.
The gross total tumor removal rate of ependymomas was 90.9%. Histological grade was significant in the prognosis of patients
The partial resection and biopsy rate of high-grade astrocytomas with multi-segment intramedullary malignant astrocytoma (v2 =
was 83.3%. 9.09, p = 0.01). In this study, a 14-year-old male with a glioblas-
The overall difference between pre-operative and post-opera- toma in the conus terminalis exhibited iliac and femoral bone
tive neurological states (as shown by IJOA score differences) was metastases five months after surgery as well as lateral cerebral
not significant (v2 = 5.44, p = 0.61). Post-operatively, the clinical ventricle seeding, and died of cerebral herniation 16 months later.
function of the patients with ependymomas and low-grade astro- A 30-year-old male patient without total removal of a grade III
cytomas recovered more quickly and completely. The extent of astrocytoma exhibited in situ recurrence eight months after sur-
surgery was the most important factor influencing the IJOA score gery. At the time of the second laminectomy, tumor was detected
differences (Z = 5.11, p < 0.001). in the subarachnoid space. Five months after the biopsy operation,
AO (Arbeitsgemeinschaft für Osteosynthesefragen) Universal a 35-year-old female patient with a cervical spinal cord malignant
Spine System (AO-USS) spinal vertebral internal fixation was per- astrocytoma died of respiratory failure.

Fig. 1. A 20-year-old female presented with progressive left extremity sensory disturbance and weakness for half a year. MRI showed an intramedullary enhancing solid mass
in the left posterior portion of the spinal cord at level C5–7. (A) Post-gadolinium-enhanced sagittal T1-weighted MRI; (B) spinal angiograms showed a 35 mm hypervascular
tumor, fed by a radiculopial artery arising from the anterior spinal artery; (E) intra-operative photograph showing a laminotomy over C5–C7 with an intramedullary vascular
lesion; (F) the vascular tumor was removed totally; (C) an MRI showing spinal cord swelling two weeks after the operation; (D) an MRI eight months later showing no
recurrent lesion. The histological result was a hemangioblastoma. (This figure is available in colour at www.sciencedirect.com.)
 J. Sun et al. / Journal of Clinical Neuroscience 16 (2009) 666–671 669

Fig. 2. A 35-year-old male presented with right lower extremity pain and weakness for a year, with the weakness progressing and spreading to all four limbs. MRI showed an
intramedullary partially enhancing mass, extending from the medulla oblongata to C6 with a syrinx at both ends. (A) Sagittal T2-weighted MRI; (B) enhanced sagittal T1-
weighted MRI; (E) intra-operative photograph showing an intramedullary gray tumor under the microscope with a clear dissection plane; (F) a cavity emerged after tumor
total resection; (C–D) two weeks after the operation, an MRI showed spinal cord swelling; (G–J) one year later, MRI showed no recurrent lesion. The histological result was an
ependymoma. (This figure is available in colour at www.sciencedirect.com.)

Finally, two adult patients with cervical ependymomas involv-
ing more than five spinal segments developed cervical flexion
deformity one year after laminectomy at both the axial cervical
spine (C1 or C2) and the cervicothoracic junction. This was proba-
bly related to the physical requirements of their occupations. Rec-
tification of the unhealthy posture and enhancement of the cervical
hypokinesis using exercise may prevent the cervical vertebra from
deformation in these patients.
4. Discussion
MSICTs require more complex surgical manipulation and in-
volve higher operative risk than single- or two-segment intramed-
ullary spinal cord tumors. Microsurgery techniques improve the
complete tumor removal rate without worsening outcomes, under-
scoring the importance of early diagnosis and microsurgical treat-
ments for MSICTs. Sandalcioglu et al. retrospectively analyzed 44
cases with spinal low-grade astrocytomas and concluded that
tumors extending more than three segments showed a good func-
tional outcome in 69% of patients (16/23) and those involving three
or fewer segments showed a good outcome in only 52% of patients
(11/21).3 Our data showed a good functional outcome in 61% of
patients (34/56). This observation can be explained by the fact that
tumors extending by more than three segments were mostly
ependymomas and low-grade astrocytomas, and as mentioned
above, characterized by a well-defined plane of dissection
compared to high-grade infiltrative astrocytomas. Functionally,
patients with these types of tumors were able to recover more
quickly and completely after surgery.
In our series, the shift and malformation of the spinal poster-
ior median sulcus was minor in intramedullary ependymomas
and low-grade astrocytomas and was identified easily under a
microscope. However, the shift and malformation of the spinal
posterior median sulcus was serious for malignant gliomas and
670 J. Sun et al. / Journal of Clinical Neuroscience 16 (2009) 666–671

Fig. 3. A 13-year-old male presented with right lower extremity numbness and left lower extremity weakness for 20 days. An MRI showed an intramedullary unevenly
enhancing mass, at level C5–T5 without a syrinx. (A) Sagittal T1-weighted MRI; (B) enhanced sagittal T1-weighted MRI; (E) intra-operative photograph showing an
intramedullary yellow -brown tumor under the microscope without a dissection plane; (F) a cavity emerged after subtotal resection of the tumor; (C–D) MRI two weeks after
the operation showing spinal cord swelling. The histological result was anaplastic astrocytoma. (This figure is available in colour at www.sciencedirect.com.)

Fig. 4. A 26-year-old male presented with lumbodorsal pain and lower extremity numbness and weakness for more than one year. An MRI showed a diffuse lightly enhancing
mass, at levels T12-L5. (A–C) Post-enhanced sagittal T1-weighted MRI; (E) intra-operative photograph showing an intramedullary tumor under the microscope occupying the
space between the spinal cord and nerve root; (F) AO Universal Spine System internal fixation was performed; (D) an anteroposterior and lateral X-ray of the lumbosacral
spine two weeks after the operation showing excellent fixation position. The histological result was ependymoma. (This figure is available in colour at www.
sciencedirect.com.)
 J. Sun et al. / Journal of Clinical Neuroscience 16 (2009) 666–671
was identified with difficulty under the microscope. This perhaps
relates to the worse post-operative neurological dysfunction in
these patients. Once the normal plasticity of the spinal cord was
impaired, surgery in our series had few benefits in terms of func-
tional improvement.
In this study, the overall difference between pre-operative and
post-operative neurological state quantified by IJOA scores was
not statistically significant. Pre-operative neurological status has
been reported to be an important factor in long-term prognosis.4,5
Brotchi et al. recommended that tumors within the spinal cord
should be operated on while the patient is still able to walk.6
McGirt et al. retrospectively analyzed 58 patients undergoing
surgical resection of cervical intramedullary spinal cord tumors
and found that only 11 of 58 patients (19%) required fusion.7 They
found that decompression spanning both the axial cervical spine
(C1–C2) and the cervicothoracic junction (C7–T1) increased the
risk for progressive spinal deformity requiring fusion. Surgery at
these high stress areas predisposed the patients to post-operative
deformity by impairing the spine’s ability to withstand the stress
associated with cervical motion. However, involvement of C1–C2
or C7–T1 alone in the decompression was not associated with
post-operative deformity. In this study, if laminectomy strictly pre-
served the facet joint capsule and surgical decompression did not
span both the axial cervical spine and the cervicothoracic junction,
then fusion at the time of tumor resection should not be required.
The patients in this series whose tumors involved more than five
spinal segments required close monitoring for the development
of progressive deformity, and the use of cervical hypokinesis exer-
cise to avoid long-term head drop.
Tumor histological classification was found to be the only sig-
nificant factor influencing the extent of surgical removal in this
study. In this series, surgery was limited by the absence of a clear
plane of cleavage to partial resection, decompression, or biopsy
only. High-grade astrocytomas especially were characterized by
an infiltrative growth, often leading to incomplete tumor resection.
Ependymomas were generally recognized as resectable lesions due
to a clearly defined plane of dissection.8 Vascular tumors were
characterized by a clear plane of dissection facilitating removal.3
Kim et al. summarized 28 adult patients diagnosed with intramed-
ullary spinal cord astrocytomas and found that complete tumor re-
moval for low-grade astrocytomas was achieved in 38.9% of
patients (7/18) and for high-grade astrocytomas in 20% (2/10).5
The extent of surgery was closely associated with recurrence and
poor prognosis.
The difference between age and initial presentation was statis-
tically significant (v2 = 15.4, p = 0.02). Adolescent patients pre-
sented with subtle changes in gait or more frequent falls, while
671
older patients tended to present with sensory disturbance and
weakness. Goethem et al. observed that motor deficits and dyses-
thesia of the four extremities and urinary dysfunction were the
most common initial presentation in adult patients with MSICTs.9
McGirt et al. considered that parents might notice subtle gait
changes or more frequent falls in their children.10 Weakness may
be asymmetrical as the tumor mass is usually eccentric within
the spinal cord center.11,12
In summary, MSICTs were most commonly seen in the medullo
cervical and cervicothoracic region, with ependymona and astrocy-
toma as the most common types. We stress the importance of early
microsurgical treatment for MSICTs while patients do not have se-
vere dysfunction.
Acknowledgements
This work was supported in part by a Peking University Third
Hospital Research Grant. The authors thank Dr. Simon Glynn (from
the Department of Neurology, University of Pennsylvania School of
Medicine) for the revision and comments of the manuscript and Ms
Isabelle Liu (from the Washington University in St Louis School of
Medicine) for the final English language editing.
References
1. Peng L, Qi ST, Chen Z, et al. Radical microsurgical treatment of intramedullary
spinal cord tumors. Chin Med J 2006;119:1343–7.
2. Chen Z, Jian FZ, Wang YL, et al. Micro neurosurgery of multisegmental
intramedullary spinal cord ependymomas. Chin J Neurosurg 2006;22:14–7.
3. Sandalcioglu IE, Gasser T, Asgari S, et al. Functional outcome after surgical
treatment of intramedullary spinal cord tumors: experience with 78 patients.
Spinal cord 2005;43:34–41.
4. Ferrante L, Mastronardi P, Celli P, et al. Intramedullary spinal cord
ependymomas–a study of 45 cases with long-term follow-up. Acta Neurochir
(Wien) 1992;119:74–9.
5. Kim MS, Chung CK, Choe GY, et al. Intramedullary spinal cord astrocytoma in
adults: postoperative outcome. J Neuro-oncol 2001;52:85–94.
6. Brotchi J, Dewitte O, Levivier M, et al. A survey of 65 tumors within the spinal
cord: surgical results and the importance of preoperative magnetic resonance
imaging. Neurosurgery 1991;29:651–7.
7. McGirt MJ, Chaichana KL, Attenello F, et al. Spinal deformity after resection of
cervical intramedullary spinal cord tumors in children. Childs Nerv Syst
2008;24:93–7.
8. Chang UK, Choe WJ, Chung SK, et al. Surgical outcome and prognostic factors of
spinal intramedullary ependymomas in adults. J Neuro-oncol 2002;57:133–9.
9. Goethem VJ, Hauwe VDL, Özsarlak Ö, et al. Spinal tumors. Eur J Radiol
2004;50:159–76.
10. McGirt MJ, Chaichana KL, Atiba A, et al. Neurological outcome after resection of
intramedullary spinal cord tumors in children. Childs Nerv Syst 2007;24:93–7.
11. Epstein FJ, Farmer JP, Freed D. Adult intramedullary spinal cord ependymomas:
The result of surgery in 38 patients. J Neurosurg 1993;79:204–9.
12. Houten JK, Weiner HI. Pediatric intramedullary spinal cord tumors: special
considerations. J Neuro-oncol 2000;47:225–30.