Diffusion Tensor Imaging in Idiopathic Acute Transverse Myelitis

Diffusion Tensor Imaging in Idiopathic Acute Transverse MyelitisW52 AJR:191, August 2008
Application of DTI to the brain has pro- vided characterization of microstructural changes in multiple sclerosis, schizophrenia, dyslexia, trauma, amyotrophic lateral sclero- sis, stroke, white matter injury of preterm infants, and aging [10–20]. Applications of DTI to the spinal cord have been reported in multiple sclerosis, spinal cord compression, cervical spondylosis, and astrocytoma [1–4, 6–8, 21–26]. It has been previously reported that DTI is more sensitive than T2-weighted imaging for detecting white matter involve- ment [1–4, 6–8, 10, 23–25, 27].
A study has shown the application of DTI in idiopathic acute transverse myelitis (ATM) [27]. It was the study about DTI in inflamma- tory diseases of the spinal cord that included multiple sclerosis, sarcoidosis, transverse my- elitis, and polyradiculoneuritis, but there were only two cases of transverse myelitis [27]. Except for that study, there have been no reports of the application of DTI in idiopathic
Diffusion Tensor Imaging in Idiopathic Acute Transverse Myelitis
Joon Woo Lee1 Kyung Seok Park2 Jae Hyoung Kim1 Ja-Young Choi3 Sung Hwan Hong3 Seong-Ho Park2 Heung Sik Kang1
Lee JW, Park KS, Kim JH, et al.
1Department of Radiology, Seoul National University Bundang Hospital, Gyeonggi-do, Korea.
2Department of Neurology, Seoul National University Bundang Hospital, 300 Gumi-dong, Bundang-gu, Seongnam-si, Gyeonggi-do 463-707, Korea. Address correspondence to K. S. Park ([email protected]).
3Department of Radiology, Seoul National University College of Medicine, Seoul, Korea.
Musculoskeleta l Imaging • Or ig ina l Research
WEB This is a Web exclusive article.
AJR 2008; 191:W52–W57
© American Roentgen Ray Society
D iffusion tensor imaging (DTI) is an MRI technique that evaluates the scalar properties of the diffu- sion of extracellular water mole-
cules within white matter fibers [1–3]. Frac- tional anisotropy (FA), a parameter derived from DTI computations, reflects the global anisotropy of the analyzed structure. FA val- ues depend on the water diffusion in the ex- tracellular space along the axon fibers [1–3]. Parameters such as myelination and axonal membrane thickness and changes in extra- or intracellular components can affect FA val- ues. The closer the FA value is to 1, the more anisotropic this structure is [1–5]. DTI pro- vides unique quantitative information per- taining to structural and orientational fea- tures of the CNS tissue. Although DTI is not in routine clinical use, it has proven to be an invaluable tool for detecting subtle damage to white matter that appears normal on con- ventional T2-weighted MR images [1–10].
Keywords: acute transverse myelitis, diffusion tensor imaging, spine
DOI:10.2214/AJR.07.2800
Received June 29, 2007; accepted after revision August 9, 2007.
Supported by grant No. 02-2006-040 from the Seoul National University Bundang Hospital Research Fund.
OBJECTIVE. Our study was based on our hypotheses that in idiopathic acute transverse myelitis (ATM), fractional anisotropy (FA) values would be abnormal not only in the T2- hyperintense lesion but also in the surrounding normal-appearing spinal cord and that the abnormal FA values in the spinal cord could be related to clinical outcome.
SUBJECTS AND METHODS. Sagittal diffusion tensor imaging (DTI) was performed in 10 patients with idiopathic ATM (four men, six women; mean age, 45 years; age range, 20–66 years) and 10 sex- and age-matched normal volunteers. FA measurements were made in the spinal cord at three levels: lesion, proximal normal-appearing spinal cord, and distal normal-appearing spinal cord. The grade of FA decrease (mild, less than 10% decrease [(FA normal − FA pt) × 100 / FA normal]; moderate, 10–20%; severe, more than 20%) was related to the clinical outcome, which was determined by a neurologist using Paine’s scale of normal, good, fair, or poor.
RESULTS. Mean FA values in patients were significantly lower than those in normal vol- unteers in lesions (0.5328 vs 0.7125, p = 0.002) and distal normal-appearing spinal cord (0.6676 vs 0.7720, p = 0.0137). All three patients with a mild FA decrease or increase in distal normal-appearing spinal cord showed a normal or good outcome, but all three patients with a severe FA decrease in distal normal-appearing spinal cord showed a fair outcome, among the eight patients to whom steroid treatment was given.
CONCLUSION. FA values in lesions and in distal normal-appearing spinal cord sig- nificantly decreased in patients with idiopathic ATM, and FA decrease in distal normal-ap- pearing spinal cord might be related to clinical outcome.
Lee et al. Diffusion Tensor Imaging in Acute Transverse Myelitis
Musculoskeletal Imaging Original Research
Diffusion Tensor Imaging in Acute Transverse Myelitis
ATM. To our knowledge, there has also been no study relating FA values in the spinal cord with the clinical outcome of the patient after treatment.
Our hypotheses were that in idiopathic ATM, FA values would be abnormal, not only in the T2-hyperintense lesion but also in the surrounding normal-appearing spinal cord because DTI has been known to be more sensitive for detecting subtle white matter change than T2-weighted imaging and that the abnormal FA values in the spinal cord could be one of the severity indicators related to clinical outcome because DTI is known to reflect microstructural white mat- ter changes.
Subjects and Methods Subjects
This study was approved by the institutional review board, and informed consent was obtained. Patients with idiopathic ATM with a lesion in the cervical spinal cord who visited the department of neurology in our hospital during a 1-year period from January 2006 to December 2006 were prospectively enrolled in this study. Exclusion criteria were uncertain diagnosis for idiopathic ATM, a lesion in the thoracic spinal cord, and refusal to undergo the DTI scan. Clinical assessment of patients was performed by one neurologist. The diagnosis was based on the criteria of the Transverse Myelitis Consortium Working Group [28]. According to the criteria, nine patients had definite idiopathic ATM and one had possible idiopathic ATM. There were four
male and six female patients, whose age ranged from 20 to 66 years (mean age, 45 years). The interval between symptom onset and MRI ranged from 0.25 to 11 months (mean, 3.8 months). All but two patients were treated with IV methyl- prednisolone (1 g/d for 5 days). MRI was per- formed before methylprednisolone treatment in eight patients. Clinical outcome after IV methyl- prednisolone treatment was evaluated by one neurologist using Paine’s scale: 1, normal, full recovery; 2, good, gait essentially normal but mild urinary symptoms or minimal sensory and upper motor neuron sign; 3, fair, mild spasticity but independent ambulation, urgency of micturition, or constipation with some sensory signs; 4, poor, unable to walk or severe gait disturbances, absence of sphincter control, and sensory deficit [29]. Sex- and age-matched normal volunteers with no his- tory of neurologic disorders and with a normal neurologic examination underwent the same scan- ning procedure as the patients. Clinical features of the patients are outlined in Table 1.
DTI Technique A 1.5-T MR scanner (Gyroscan Intera, Philips
Healthcare) was used. Head and neck coils were applied to all subjects. The sensitivity encoding (SENSE) single-shot echo-planar imaging (EPI) with pulse sequence and SENSE factor 2 was used for the sagittal DTI in the cervical spinal cord with b value, 900 s/mm2; number of diffusion gradient directions, 15; number of excitations, 5; and slice thickness, 4 mm [30]. The diffusion gradient strength was 30 mT/m, the foldover direction was anteroposterior, and the fat shift
direction was posterior. The TR/TE was 7,000/ 100; matrix, 112 × 128; and field of view, 224 × 224 mm. The slice thickness was 4 mm.
Measurements After sending all source images of the DTI to a
PC, one radiologist who was blinded to clinical outcomes and history measured FA in the cervical spinal cord by PRIDE software (Philips Health care). In patients, FA measurements were made in the spinal cord at three different levels (lesion, proximal normal-appearing spinal cord, and distal normal- appearing spinal cord) by using regions of interest (ROIs) and the most accurate B0 axial image. Lesion level was determined as the center of a T2- hyperintense lesion in the spinal cord. Proximal normal-appearing spinal cord level was determined as a level with the T2-isointense spinal cord and one spine segment proximal to the cranial end of the T2-hyperintense spinal cord lesion. Distal normal-appearing spinal cord level was determined as a level with the T2-isointense spinal cord and one spine segment distal to the caudal end of the T2-hyperintense spinal cord lesion (Fig. 1). Special attention was paid in ROI selection to avoid partial volume effect, magnetic susceptibility effects, and motion artifacts. To maintain the same ROI for the same spinal cord, we carefully placed a circular ROI in side the spinal cord as large as possible and without containing CSF outside the spinal cord. After FA measure ments, the three levels of measurements were recorded in each patient, and FA measure ments were also made at the same levels on the match ed normal volunteers. The ROIs in patients and normal volunteers were
TABLE 1: Clinical Features of 10 Patients with Idiopathic Acute Transverse Myelitis (ATM)
Patient No. Age (y) Sex Symptoms
Intervala (mo)
Clinical Outcomee
1 57 M Weakness, sensory change, sphincter dysfunction 2 Yes Yes Definite Fair
2 48 F Weakness, sensory change, sphincter dysfunction 2 No Yes Definite Fair
3 29 F Weakness, sensory change 0.25 No Yes Definite Good
4 51 M Sensory change 2 No Yes Definite Normal
5 47 M Weakness, sensory change 11 NA Yes Definite Fair
6 48 F Weakness, sensory change, sphincter dysfunction 4 No Yes Definite NA
7 66 F Weakness, sensory change 5 No Yes Definite Fair
8 48 F Sensory change 1 No Yes Definite Normal
9 20 M Weakness, sensory change, sphincter dysfunction 9 NA Yes Definite NA
10 32 F Sensory change 1.3 No No Possible Normal
Note—Patient no. indicates case number of patients with idiopathic ATM. NA indicates not applicable. aMonths between onset of symptoms and diffusion tensor imaging. bCSF findings of pleocytosis or IgG index elevation suggestive of inflammatory evidence. cPresence of intramedullary enhancement on MRI. dDiagnosis of idiopathic ATM on the basis of the criteria of the Transverse Myelitis Consortium Working Group [28]. eClinical outcome after steroid treatment assessed by Paine’s scale as normal, good, fair, or poor.
D ow
nl oa
de d
fr om
w w
w .a
jr on
lin e.
or g
by 2
7. 70
.1 29
.2 0
on 0
3/ 23
/2 3
fr om
I P
ad dr
es s
27 .7
0. 12
9. 20
. C op
yr ig
ht A
R R
S. F
or p
er so
Lee et al.
matched at each level. Levels and FA values are summarized in Table 2.
With FA values on matched levels of patients (FA pt) and normal volunteers (FA normal), the percentage of decrease of FA values was calculated as follows:
Percentage of decrease = (FA normal − FA pt) × 100 / FA normal.
On the basis of the percentage of decrease, the FA decrease was graded as mild (less than 10% by percentage of decrease), moderate (10–20% by percentage of decrease), severe (more than 20% by percentage of decrease). One neurologist and one radiologist compared the grade of FA decrease with the clinical outcome.
Statistical Analysis Wilcoxon’s matched-pairs signed rank test was
used for evaluating differences of FA values in the spinal cord at each matched level (lesion, proximal normal-appearing spinal cord, and distal normal- appearing spinal cord) between patients and normal volunteers. Graphpad Instat software (Graphpad Software) was used for statistical analysis.
Results Levels and FA values in 10 patients with
idiopathic ATM are summarized in Table 2. FA values in the spinal cord at three matched levels (lesion, proximal normal-appearing spinal cord, and distal normal-appearing spi- nal cord) in 10 patients with idiopathic ATM and in 10 age- and sex-matched normal vol- unteers are summarized in Table 3. Mean FA values for the lesion level in patients were significantly lower than those in normal vol- unteers (0.5328 vs 0.7125, p = 0.002). Mean FA values for distal normal-appearing spinal
cord in patients were also significantly lower than those in normal volunteers (0.6676 vs 0.7720, p = 0.0137). However, mean FA val- ues of spinal cord in proximal normal-ap- pearing spinal cord were not significantly different between patients and normal vol- unteers (0.6443 vs 0.6307, p = 0.693). FA values in the spinal cord tended to be lower in the upper cervical level in the normal volunteers. At the C1 level, FA values in the spinal cord were lower than 0.6 in all normal volunteers and patients.
Comparison of FA decrease in the spinal cord of patients and outcome results are sum- marized in Table 4. Among eight patients to whom steroid treatment was given, there were three patients with normal, one with good, and four with fair results. Among four patients with a fair clinical outcome, severe FA decrease was shown at the lesion level in all (100%) and in distal normal-appearing spinal cord in three (75%). Among four pa- tients with normal or good clinical outcome, severe FA decrease was shown in two (50%)
C D
A B
Fig. 1—29-year-old woman with idiopathic acute transverse myelitis (patient no. 3). A, T2-weighted sagittal image shows hyperintense lesion in spinal cord from C2 to C6 level. Lesion level was determined to be center of T2-hyperintense lesion in spinal cord. Proximal normal-appearing spinal cord level (PNLSC) was determined as level with T2-isointense spinal cord and one spine segment proximal to cranial end of T2-hyperintense spinal cord lesion. Distal normal-appearing spinal cord level (DNLSC) was determined as level with T2-isointense spinal cord and one spine segment distal to caudal end of T2-hyperintense spinal cord lesion. B–D, Diffusion tensor image of fractional anisotropy (FA) map (B) and color tensor map image (C) show decreased FA in lesion. FA measurements were made in spinal cord at three different levels (lesion, proximal normal-appearing spinal cord, and distal normal-appearing spinal cord) in most accurate B0 axial image (D).
D ow
nl oa
de d
fr om
w w
w .a
jr on
lin e.
or g
by 2
7. 70
.1 29
.2 0
on 0
3/ 23
/2 3
fr om
I P
ad dr
es s
27 .7
0. 12
9. 20
. C op
yr ig
ht A
R R
S. F
or p
er so
Diffusion Tensor Imaging in Acute Transverse Myelitis
at the lesion level and none (0%) in distal normal-appearing spinal cord. In terms of FA decrease at the lesion level, FA decrease was mild in one, moderate in one, and severe in eight. Two cases with mild or moderate FA decrease at lesion level had a normal clinical outcome. Among eight cases with severe FA decrease at the lesion level, only two cases showed normal (patient no. 8) or good out- come (patient no. 3, Fig. 1), which showed moderately decreased FA or increased FA in distal normal-appearing spinal cord. In terms of FA decrease in distal normal-appearing spinal cord and the clinical outcome of eight patients to whom steroid treatment was giv- en, all three patients with mild FA decrease or FA increase in distal normal-appearing spinal cord showed a normal or good clinical outcome, two patients with moderate FA de- crease in distal normal-appearing spinal cord showed a normal or fair outcome, but all
three patients with severe FA decrease in dis- tal normal-appearing spinal cord showed a fair clinical outcome. There was no relation of FA decrease in proximal normal-appear- ing spinal cord to clinical outcome.
Discussion Acute noncompressive myelopathies were
first recognized in the 19th century, and early pathologic studies identified both inflamma- tory and vascular causes [31–33]. ATM is a syndrome characterized by anterior and pos- terior (hence transverse) spinal cord impair- ment resulting in weakness, a lowered sen- sory level, and autonomic dysfunction [28, 34–36]. Possible causes include vascular, in- fectious, neoplastic, collagen vascular, iatro- genic, and autoimmune abnormalities. In ad- dition, the syndrome can occur as part of a CNS demyelinating disease such as neuro- myelitis optica or, uncommonly, typical mul-
tiple sclerosis. After such recognizable causes are excluded, idiopathic ATM can be diagnosed [28, 29, 34–36]. On the basis of expert opinion, the Transverse Myelitis Con- sortium Working Group proposed diagnostic criteria for idiopathic ATM [28]. The clini- cal requirements include bilateral sensory, motor, or autonomic dysfunction referable to the spinal cord, with a clearly defined senso- ry level that progresses to the nadir over 4–21 days from onset. Neuroimaging, preferably MRI, is used to eliminate structural causes. Evidence to support an inflammatory cause is also required; this may be shown through MRI evidence of gadolinium enhancement within the cord or by CSF findings of pleocy- tosis or IgG index elevation. Patients meeting all diagnostic criteria are considered to have definite idiopathic ATM, whereas those who do not meet the MRI or CSF criteria for in- flammation have possible idiopathic ATM.
TABLE 3: Fractional Anisotropy (FA) Values in Spinal Cord at Matched Three Levels (Lesion, Proximal Normal- Appearing Spinal Cord, and Distal Normal-Appearing Spinal Cord) in 10 Patients with Idiopathic Acute Transverse Myelitis (ATM) and 10 Age- and Sex-Matched Normal Volunteers
Group
Mean SD Mean SD Mean SD
Normal volunteers 0.7125 0.0794 0.6307 0.1017 0.7720 0.0527
Patients 0.5328 0.0658 0.6443 0.0906 0.6676 0.0821
p 0.0020 0.6953 0.0137
TABLE 2: Fractional Anisotropy (FA) Measurement in 10 Patients with Idiopathic Acute Transverse Myelitis (ATM) and 10 Age- and Sex-Matched Normal Volunteers
Patient No. Lesiona Lesion Levelb
Lesion FAc Proximal Normal-
Distal FAg
Patient Group
Normal Group
Patient Group
Normal Group
Patient Group
Normal Group
1 C2–C4 C3 0.5117 0.7950 C1 0.5840 0.5657 C5 0.5389 0.7904
2 C5 C5 0.5517 0.7311 C4 0.6135 0.8044 C6 0.5929 0.7618
3 C2–C6 C4 0.4197 0.6672 C1 0.4595 0.5221 C7 0.7835 0.7783
4 C7 C7 0.6297 0.6446 C6 0.7045 0.6203 T1 0.7368 0.7888
5 C5–C6 C5 0.4846 0.6527 C4 0.7263 0.7124 C7 0.7476 0.8618
6 C7 C7 0.4460 0.5657 C6 0.6465 0.6120 T1 0.6493 0.6748
7 C5–C6 C5 0.5751 0.7940 C4 0.7387 0.7741 C7 0.5961 0.7635
8 C5–C6 C6 0.5737 0.7849 C4 0.6059 0.6271 C7 0.6592 0.8129
9 C2–C4 C3 0.5582 0.7886 C1 0.6061 0.5162 C5 0.6243 0.7854
10 C2 C2 0.5776 0.7013 C1 0.7581 0.5532 C3 0.7479 0.7029
Note—Patient no. indicates case number of 10 patients with idiopathic ATM and 10 age- and sex-matched normal volunteers. aLevel of T2-hyperintensity in the spinal cord. bLevel of center of T2-hyperintensity in the spinal cord. cFA value of lesion level measured in regions of interest in the spinal cord. dOne segment proximal to the cranial end of the T2-hyperintense lesion. eFA value measured in regions of interest in the spinal cord of proximal normal-appearing spinal cord level. fOne segment distal to the caudal end of the T2-hyperintense lesion. gFA value measured in regions of interest in the spinal cord of distal normal-appearing spinal cord level.
D ow
nl oa
de d
fr om
w w
w .a
jr on
lin e.
or g
by 2
7. 70
.1 29
.2 0
on 0
3/ 23
/2 3
fr om
I P
ad dr
es s
27 .7
0. 12
9. 20
. C op
yr ig
ht A
R R
S. F
or p
er so
Lee et al.
The intent of these criteria is to identify a relatively homogeneous patient cohort for re- search. In this study, we used these criteria to identify cases with idiopathic ATM [28].
According to our study, FA values in the lesion level and distal normal-appearing spi- nal cord significantly decreased in the pa- tients compared with normal volunteers. Pa- tients with mild FA decrease or FA increase in distal normal-appearing spinal cord showed a better clinical outcome than pa- tients with severe FA decrease in the distal level. Patients with…