SPINE Volume 32, Number 15, pp 1675–1678

©2007, Lippincott Williams & Wilkins, Inc.

MR T2 Image Classification in Cervical

Compression Myelopathy
Predictor of Surgical Outcomes

Yasutsugu Yukawa, MD, Fumihiko Kato, MD, Hisatake Yoshihara, MD, Makoto Yanase, MD,
and Keigo Ito, MD

Study Design. Prospective imaging study of patients erative recovery. Classification of ISI can be a predictor of
undergoing surgery for cervical compressive myelopathy. surgical outcome.
Objectives. To investigate whether the classification of Key words: cervical compression myelopathy, magnetic
increased signal intensity (ISI) on magnetic resonance resonance imaging, classification, increased signal inten-
imaging (MRI) in patients with cervical compressive my- sity, laminoplasty, predictor. Spine 2007;32:1675–1678
elopathy reflects the severity of symptoms and surgical
outcome.
Summary of Background Data. The association be-
Magnetic resonance imaging (MRI) is an invaluable ex-
tween ISI and surgical outcome in cervical myelopathy
remains controversial. The degree of ISI has not been well amination in patients with cervical compressive myelop-
discussed. athy. MRI can show not only the degree of spinal canal
Methods. A total of 104 patients with cervical com- stenosis but also the intramedullary state of the spinal
pressive myelopathy were prospectively enrolled. All cord in detail.1 Increased signal intensity (ISI) of the spi-
were treated with cervical expansive laminoplasty. MRI
was performed in all patients before surgery. ISI of
nal cord on T2-weighted MRI and decreased signal in-
spinal cord was classified into three groups based on tensity on T1-weighted MRI are well-known changes in
sagittal T2-weighted images as follows: Grade 0, none; spinal cord lesions. These signal changes are considered
Grade 1, light (obscure); and Grade 2, intense (bright). to reflect various intramedullary lesions. In particular,
The severity of myelopathy was evaluated according to ISI is often seen in patients with cervical compressive
the Japanese Orthopedic Association (JOA) score for
cervical myelopathy.
myelopathy. However, the significance of ISI for prog-
Results. Eighty-six patients (83%) showed ISI before nosis remains controversial. With advances in MRI tech-
surgery. Patients with ISI were significantly older, and niques and software, it has become possible to detect two
had a longer duration of disease, a lower postoperative different types of ISI: light (obscure) and intense (bright)
JOA score, and a worse postoperative recovery rate of signal changes. Unfortunately, to our knowledge, the as-
JOA score than those without ISI. Preoperative MRI
sociation between the degree of ISI in the spinal cord,
showed 18 patients in Grade 0, 49 patients in Grade 1, and
37 in Grade 2. Duration of disease was the shortest in clinical symptoms, and surgical outcome has not been
Grade 0 and longest in Grade 2. Although there was no well discussed. The purpose of the present study was to
significant difference in preoperative JOA scores among elucidate whether preoperative grades of ISI reflect the
the three groups, Grade 0 patients had a higher postop- severity of symptoms and surgical outcome in patients
erative JOA score and the best postoperative recovery,
with cervical compressive myelopathy.
and Grade 2 had a lower postoperative JOA score and the
worst postoperative recovery. Materials and Methods
Conclusion. Preoperative ISI on T2-weighted sagittal
MRI was correlated with patient age, duration of disease, A total of 142 patients with cervical compressive myelopathy
postoperative JOA score, and postoperative recovery were studied prospectively from April 1995 to December 2000.
rate. Patients with the greatest ISI had the worst postop- Among them, 104 patients who were followed up for more
than 12 months were enrolled in this study. There were 67 men
and 37 women, and the mean age was 61.0 years (range, 34 –79
years). The concomitant diagnoses causing cervical compres-
From the Department of Orthopedic Surgery, Chubu Rosai Hospital, sive myelopathy were cervical spondylotic myelopathy in 74
Nagoya, Japan. patients, ossification of the posterior longitudinal ligament in
Acknowledgment date: August 15, 2006. First revision date: September
6, 2006. Second revision date: October 31, 2006. Third revision date:
20 patients, cervical disc herniation in 6 patients, and calcifi-
December 19, 2006. Fourth revision date: December 21, 2006. Accep- cation of the yellow ligament in 4 patients. Patients with cere-
tance date: December 21, 2006. bral palsy, rheumatoid arthritis, or other spinal disease, and
Presented at the Cervical Spine Research Society, 30th Annual Meeting, those who needed spinal instrumentation during surgery due to
Phoenix, AZ, December 2003. kyphotic deformity or severe instability were excluded from
The manuscript submitted does not contain information about medical
device(s)/drug(s). this study. Patients with traumatic cervical cord injury without
No funds were received in support of this work. No benefits in any bony lesions, the so-called central cord injury, were also ex-
form have been or will be received from a commercial party related cluded. The mean duration of disease was 20 months (range,
directly or indirectly to the subject of this manuscript. 1–228 months) before surgery. Expansive laminoplasty from
Address correspondence and reprint requests to Yasutsugu Yukawa,
MD, Department of Orthopedic Surgery, Chubu Rosai Hospital,
C3–C7 was performed in all patients.
Nagoya, Aichi 455-0830, Japan; E-mail: [email protected]. All patients underwent high-resolution MRI with a 1.5-T
go.jp Signa (GE Medical Systems, WI) imager before surgery. T1-

1675
1676 Spine • Volume 32 • Number 15 • 2007

Figure 1. Classification of in-

creased signal intensity (ISI) of
the spinal cord in T2-weighted
MRI.

and T2-weighted images of sagittal views of the cervical cord Results

were obtained using a spin echo sequence system for T1-
weighted images and a fast spin echo sequence system for The mean JOA score was 9.8 before surgery, 13.4 at 1
T2-weighted images. A surface coil was used. The slice width year after surgery, and 13.2 at the final follow-up. The
was 4 mm and the acquisition matrix was 512 ⫻ 256. The mean recovery rate of the JOA score was 48.2% at the
sequence parameters were repetition time (TR) 400 millisec- final follow-up. The mean follow-up period was 40
onds/echo time (TE) 11 milliseconds for T1-weighted im- months (range, 12–90 months) after surgery. No patients
ages, and TR 4000 milliseconds/TE 126 milliseconds for needed additional cervical decompressive surgery due to
T2-weighted images. In sagittal T2 images, ISI of the spinal restenosis or residual stenosis.
cord at the narrowest level was classified from Grade 0 to In the study group, 86 patients (83%) had ISI and 18
Grade 2 (Grade 0, none; Grade 1, light [obscure]; Grade 2, patients (17%) did not. Patients with ISI were older, and
intense [bright]) (Figure 1). Intense ISI was defined as similar
to the signal of cerebrospinal fluid. These classifications Table 1. Evaluation of Cervical Myelopathy Using a
were made by two independent radiologists experienced in Scoring System Proposed by Japanese Orthopaedic
spinal imaging. The concordance between the two observers Association (JOA Score) and Recovery Rate of JOA Score
in evaluating the signal change on T2-weighted images was
0.88 (kappa ⫽ 0.80; P ⬍ 0.001). The two observers estab- JOA score
I. Motor function of the upper extremity
lished the final classification by consensus.
0. Impossible to eat with chopsticks or spoon
The severity of myelopathy was evaluated according to the 1. Possible to eat with spoon, but not with chopsticks
Japanese Orthopedic Association (JOA) score for cervical my- 2. Possible to eat with chopsticks, but inadequate
elopathy (Table 1). Improvement of postoperative symptoms 3. Possible to eat with chopsticks, but awkward
4. Normal
was represented using the recovery rate of the JOA score ([post- II. Motor function of the lower extremity
operative JOA score ⫺ preoperative JOA score]/[17 ⫺ preop- 0. Impossible to walk
erative JOA score] ⫻ 100%). A recovery rate of 100% was the 1. Needs cane or aid on flat ground
best possible postoperative improvement. 2. Needs cane or aid only on stairs
3. Possible to walk without cane or aid but slowly
The multifactorial effects of variables such as age, duration 4. Normal
of disease, preoperative JOA score, postoperative JOA score, III. Sensory function
recovery rate of JOA score, and preoperative ISI on MRI were A. Upper extremity
studied. 0. Apparent sensory loss
1. Minimal sensory loss
2. Normal
Statistical Analysis. A standard StatView (SAS Institute, B. Lower extremity (same as A)
Cary, NC) software package was used for the statistical anal- C. Trunk (same as A)
IV. Bladder function
ysis. Nonparametric analysis using Mann-Whitney U test was 0. Complete retension
used for analyzing the differences between the two groups and 1. Severe disturbance (sense of retention, dribbling, incomplete
the Kruskal-Wallis test followed by the Mann-Whitney U test continence)
for analyzing the differences among three groups. Repeated- 2. Mild disturbance (urinary frequency, urinary hesitancy)
3. Normal
measures analyses of variance in the same group were per-
formed using the Wilcoxon test. P ⬍ 0.05 was considered sta- Recovery rate of JOA score (Hirabayashi method). Recovery rate (%) ⫽ 关Postopera-
tive score ⫺ Preoperative score兴 / 关Normal score (17) ⫺ Preoperative score兴 ⫻ 100.
tistically significant.
 Cervical Compression Myelopathy • Yukawa et al 1677

Table 2. Clinical Features and Surgical Outcomes With in signal from the cord on T2-weighted images in detail:
or Without ISI from 0 (none) to 4 (very intense). These authors also men-
tioned that patients who had an intense T2 signal before
Without ISI With ISI P
surgery seemed to do better than those with mild preoper-
Age (yr) 55.3 ⫾ 9.4 62.3 ⫾ 9.4 0.008 ative abnormality. However, their grading scale is too me-
Duration of disease (mo) 12.1 ⫾ 27.9 22.1 ⫾ 33.6 0.008 ticulous to classify the signal change. We classified the sig-
Preoperative JOA score 10.6 ⫾ 2.2 9.6 ⫾ 2.8 NS
Postoperative JOA score 14.6 ⫾ 1.9 13.0 ⫾ 2.6 0.012
nal changes into three categories from Grade 0 to Grade 2
Recovery rate (%) 62.3 ⫾ 25.9 45.3 ⫾ 31.1 0.033 (Grade 0, none; Grade 1, light [obscure]; and Grade 2,
NS indicates not significant.
intense [bright]). This grading system seems to be subjective
for each observer. However, the concordance between the
two observers in evaluating signal changes was relatively
high, and the reproducibility of grading was considered to
had a longer duration of disease, a worse postoperative
be good.
JOA score, and a worse postoperative recovery rate than
ISI on preoperative MRI was seen in 86 patients
those without ISI (Table 2).
(83%): 49 patients with Grade 1 and 37 with Grade 2.
Preoperative MRI showed 18 patients in Grade 0, 49
In the present study, preoperative MRI classification
in Grade 1, and 37 in Grade 2. The mean age was the
was consistent with clinical symptoms and surgical
youngest in Grade 0. Duration of disease increased with
outcome. Patients with ISI were significantly older,
grade, being 12.1 in Grade 0, 14.7 in Grade 1, and 32.0
and had a longer duration of disease, and showed less
in Grade 2. The respective preoperative and postopera-
improvement after surgery than patients without ISI.
tive JOA scores in Grades 0, 1, and 2 were 10.6, 9.4, 9.9
We also found a significant association among patient age,
and 14.6, 13.3, 12.7, respectively. There was no differ-
duration of disease, postoperative JOA score, recovery rate
ence in preoperative JOA scores, but the postoperative
of JOA score, and preoperative MRI classification. Patients
JOA score of Grade 2 was significantly the lowest. Re-
with the greatest ISI had a longer duration of disease and
covery rates in Grades 0, 1, and 2 were 62.3%, 52%, and
the worst surgical results. To our knowledge, this is the first
37%, respectively. Recovery rates decreased significantly
report to demonstrate a significant association among ob-
with grades (Table 3).
jective signs, surgical outcome, and preoperative MRI clas-
Discussion sification.
In a neuropathologic study of patients with cervical
ISI of the spinal cord on T2-weighted MRI is often seen in
spondylotic myelopathy, there appeared to be a com-
patients with cervical compressive myelopathy. With recent
mon pattern of lesion progression, from mild alter-
advances in MRI techniques and software, we can detect
ation of the spinal cord to severe.10 In studies of his-
various degrees of ISI in these patients. However, opinions
topathologic examination and MRI, ISI without signal
vary on the association between the presence of ISI, clinical
change on T1-weighted images appeared nonspecifi-
features, and surgical outcome. The significance of ISI for
cally in mildly altered lesions such as loss of nerve cells,
prognosis remains controversial. The purpose of the
gliosis, edema in gray matter, Wallerian degeneration, de-
present study was to classify ISI before surgery and prospec-
myelination, and edema in white matter.1,11 ISI of the
tively, and to verify whether the classification could reflect
spinal cord on T2-weighted images has been considered
clinical features and surgical outcome.
to include a wide spectrum of compressive myelomalacic
Many authors have investigated the association between
pathology and reflects a wide range of spinal cord recuper-
ISI and surgical outcomes. Some authors report that pa-
ative potential.7 Based on our findings and previous stud-
tients with ISI have poor prognosis after surgery,2– 6 but
ies,1,10,11 it can be concluded that light ISI reflects mild
others could not find any association.7–9 Most of them
neuropathologic alteration in the spinal cord and reflects
mentioned the presence of ISI on T2-weighted images and
greater recuperative potential, and intense ISI reflects severe
decreased signal intensity on T1-weighted images but did
alteration and reflects less recuperative potential. In cervical
not refer to the degree of ISI. Only Mehalic et al2 have
compressive myelopathy, the signal intensity of spinal cord
described a grading scale for classifying the relative increase
changes from none to light ISI and then to intense ISI with
the progress of the disease, and the recovery rate after
Table 3. Clinical Features and Surgical Outcomes in surgery decreases.
Each Grade of ISI
Grade 0 Grade 1 Grade 2 P Key Points
Age (yr) 55.3 ⫾ 9.4 62.1 ⫾ 10.1 62.5 ⫾ 8.4 0.031 ● Increased signal intensity (ISI) on magnetic reso-
Duration of disease (mo) 12.1 ⫾ 27.9 14.7 ⫾ 21.3 32.0 ⫾ 43.3 0.001 nance images (MRIs) in patients with cervical com-
Preoperative JOA score 10.6 ⫾ 2.2 9.4 ⫾ 2.8 9.8 ⫾ 2.8 NS
Postoperative JOA score 14.6 ⫾ 1.9 13.3 ⫾ 2.6 12.6 ⫾ 2.6 0.020 pressive myelopathy were prospectively classified
Recovery rate (%) 62.3 ⫾ 25.9 51.6 ⫾ 25.0 36.9 ⫾ 36.4 0.018 to three grades.
NS indicates not significant.
1678 Spine • Volume 32 • Number 15 • 2007

the results of surgery for cervical compression myelopathy. Spine 1993;

18:2024 –9.
● Preoperative ISIs in MR T2-weighted sagittal im- 4. Ramanauskas WL, Wilner HI, Metes JJ, et al. MR imaging of compressive
myelomalacia. J Comput Assist Tomogr 1989;13:399 – 404.
ages were correlated with patient age, duration of 5. Suri A, Chabbra RP, Mehta VS, et al. Effect of intramedullary signal changes
disease and postoperative JOA score and recovery on the surgical outcome of patients with cervical spondylotic myelopathy.
rates. Spine J 2003;3:33– 45.
● ISI indicated a poor prognosis, and especially in- 6. Takahashi M, Yamashita Y, Sakamoto Y, et al. Chronic cervical cord com-
pression: clinical significance of increased signal intensity on MR images.
tense ISI indicated worse surgical outcomes than Radiology 1989;173:219 –24.
light ISI. 7. Morio Y, Teshima R, Nagashima H, et al. Correlation between operative
● Signal intensity change patterns in T2-weighted outcomes of cervical compression myelopathy and MRI of the spinal cord.
Spine 2001;26:1238 – 45.
sagittal images can be a predictor of surgical 8. Yone K, Sakou T, Yanase M, et al. Preoperative and postoperative magnetic
outcomes. resonance image evaluations of the spinal cord in cervical myelopathy. Spine
1992;17(suppl 10):388 –92.
9. Wada E, Yonenobu K, Suzuki S, et al. Can intramedullary signal change on
References magnetic resonance imaging predict surgical outcome in cervical spondylotic
myelopathy? Spine 1999;24:455– 61; discussion 462.
1. Ohshio I, Hatayama A, Kaneda K, et al. Correlation between histopathologic 10. Ito T, Oyanagi K, Takahashi H, et al. Cervical spondylotic myelopathy:
features and magnetic resonance images of spinal cord lesions. Spine 1993; clinicopathologic study on the progression pattern and thin myelinated fibers
18:1140 –9. of the lesions of seven patients examined during complete autopsy. Spine
2. Mehalic TF, Pezzuti RT, Applebaum BI. Magnetic resonance imaging and 1996;21:827–33.
cervical spondylotic myelopathy. Neurosurgery 1990;26:217–26; discussion 11. Mizuno J, Nakagawa H, Inoue T, et al. Clinicopathological study of ‘snake-
226 –7. eye appearance’ in compressive myelopathy of the cervical spinal cord.
3. Okada Y, Ikata T, Yamada H, et al. Magnetic resonance imaging study on J Neurosurg 2003;99(suppl 2):162– 8.