- Email: [email protected]
Specific MRI findings help distinguish acute transverse myelitis of Neuromyelitis Optica from spinal cord infarction
Author’s Accepted Manuscript SPECIFIC MRI FINDINGS HELP DISTINGUISH ACUTE TRANSVERSE MYELITIS OF NEUROMYELITIS OPTICA FROM SPINAL CORD INFARCTION I Kister, E Johnson, E Raz, J Babb, J Loh, TM Shepherd www.elsevier.com/locate/msard
PII: DOI: Reference:
S2211-0348(16)30048-7 http://dx.doi.org/10.1016/j.msard.2016.04.005 MSARD389
To appear in: Multiple Sclerosis and Related Disorders Received date: 30 September 2015 Revised date: 26 February 2016 Accepted date: 11 April 2016 Cite this article as: I Kister, E Johnson, E Raz, J Babb, J Loh and TM Shepherd, SPECIFIC MRI FINDINGS HELP DISTINGUISH ACUTE TRANSVERSE MYELITIS OF NEUROMYELITIS OPTICA FROM SPINAL CORD I N FA R C T I O N , Multiple Sclerosis and Related Disorders, http://dx.doi.org/10.1016/j.msard.2016.04.005 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 galley proof before it is published in its final citable 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.
SPECIFIC MRI FINDINGS HELP DISTINGUISH ACUTE TRANSVERSE MYELITIS OF NEUROMYELITIS OPTICA FROM SPINAL CORD INFARCTION Kister I1, Johnson E2, Raz E2, Babb, J2, Loh J2, Shepherd TM2 1
NYU Multiple Sclerosis Comprehensive Care Center, Department of Neurology,
New York University School of Medicine, New York, NY, USA. 2 Neuroradiology
Section, Department of Radiology, New York University School of
Medicine, New York, NY, USA.
Corresponding author:
Timothy Shepherd, M.D./Ph.D. [email protected] 660 First Ave., 2nd Floor, New York, NY 10016 Phone – 212-263-8487 Fax – 212-263-3838
Key words: Spinal cord, Myelitis, Devic’s Disease, Ischemia, MRI, Multiple Sclerosis Word count: 2128 words Abbreviations: Magnetic Resonance Imaging (MRI), Neuromyelitis Optica (NMO), Spinal Cord Infarct (SCI).
ABSTRACT Background: There is substantial overlap between MRI of acute spinal cord lesions from neuromyelitis optica (NMO) and spinal cord infarct (SCI) in clinical practice. However, early differentiation is important since management approaches to minimize morbidity from NMO or SCI differ significantly. Objective: To identify MRI features at initial presentation that may help to differentiate NMO acute myelitis from SCI. Methods: 2 board-certified neuroradiologists, blinded to final diagnosis, retrospectively characterized MRI features at symptom onset for subjects with serologically-proven NMO (N=13) or SCI (N =11) from a single institution. Univariate and multivariate analyses were used to identify factors associated with NMO or SCI. Results: SCI was more common in men and Caucasians, while NMO was more common in non-Caucasian women (P < 0.05). MRI features associated with NMO acute myelitis (P<0.05) included location within 7-cm of cervicomedullary junction; lesion extending to pial surface; ‘bright spotty lesions’ on axial T2 MRI; and gadolinium enhancement. Patient’s age, lesion length and cross-sectional area, cord expansion, and the “owl’s eyes” sign did not differ between the two groups (P > 0.05). Conclusion: Along with patient demographic characteristics, lesion features on MRI, including lesion location, extension to pial border and presence of ‘bright spotty lesion’ can help differentiate acute myelitis of NMO from SCI in the acute setting.
1. INTRODUCTION Acute spinal cord injury, characterized by rapid evolution of bilateral sensori-motor deficits and autonomic dysfunction is a neurologic emergency that carries risk of considerable long-term disability. When MRI reveals a longitudinally extensive cord lesion, the main diagnostic considerations are acute transverse myelitis secondary to infectious or autoimmune etiologies, versus spinal cord ischemia (SCI)(Kitley et al. , 2012). It is essential to establish the correct diagnosis promptly in order to initiate treatment (and avoid inappropriate therapy). It would be advantageous to know whether there are any etiology-specific MRI features that could be used to improve diagnostic accuracy in the acute setting. To address this question, we carried out a retrospective spinal cord MRI review of patients evaluated at our institution with acute myelopathy due to Neuromyelitis Optica (NMO) and patients with spinal cord ischemia (SCI). We choose to focus on subset of acute myelopathy due to NMO because of a very high degree of diagnostic certainty in these cases, and because lesions of NMO and SCI both tend to be several vertebral bodies long, expansile and centered around gray matter. Further, during the presentation of either condition, the companion brain MRI is typically normal, making their differentiation especially challenging.
2. MATERIALS AND METHODS 2.1 – Subject Selection
This was a HIPAA-compliant retrospective case-control study. We performed a single-institution retrospective chart review over a 10-year period (2005-2014) to identify cases with final diagnosis of spinal cord infarct (SCI) or acute myelitis due to clinically definite NMO (Wingerchuk et al., 2006) The diagnosis of spinal cord infarct (SCI) was made in 10 patients based on a combination of clinical and MRI features and, where available, angiographic evidence. Direct angiographic confirmation was available in 4 cases (anterior spinal artery occlusion, hyperemia in the infarcted spinal cord territory and increased visualization of the collateral circulation). In 4 patients with typical features of SCI (hyperacute onset-to-nadir time of <4 hours and severe radicular pain at onset), diagnosis of SCI was supported by such MRI features that have never been described in myelitis: vertebral and/or muscle infarction in 3 cases and persistent root and ventral cord enhancement several months after event in 2 cases. In two patients, symptoms developed in the immediate following medical procedures - discectomy in one case and epidural injection in a patient with hemochromatosis in the other case. All 13 patients with NMO myelitis fulfilled diagnostic criteria for NMO and were AQP-4-Ab seropositive.
2.2 – Data Analysis Patient demographic information that would be readily available to radiologists (age, gender and race) were included in the analysis. MRI of the spine obtained within 2 days of clinical presentation was reviewed by two board-certified neuroradiologists (TS and JL) who were blinded to patient demographic
information and the final diagnosis.
For some patients, initial MRI study was
obtained at an outside institution prior to transfer to our medical center, however the images were loaded into the local radiology PACS. All studies included axial and sagittal T2-weighted sequences of the total spine. The sagittal extent of the T2 hyperintense acute lesion was characterized by the number of vertebral bodies spanned and craniocaudal length in cm, e.g. a lesion between C3 and C5 could be characterized as 3 vertebral bodies, or 3.4-cm long by measurement on the mid-sagittal T2 image. The position of the lesion along the cranio-caudal axis was recorded as the vertebral body at the lesion center. We measured the distance of the cranial and caudal lesion edges to the cervicomedullary junction and cord termination at the conus respectively. We recorded whether the lesion crossed the craniocervical, cervicothoracic or thoracolumbar junctions. The midpoint of the lesion was characterized using the axial T2 images. The cross-sectional area of the lesion and respective overall cord cross-sectional area were measured manually on the same T2-weighted axial image; then the ratio of lesion-to-cord cross-sectional area was calculated. We recorded presence of T2 'bright spotty’ lesions(Yonezu et al. , 2014), ‘owl (or snake) eyes’ sign Lebouteux et al. , 2014) and whether the lesion extended to pial surface of cord on any axial T2W image. The presence or absence of cord expansion was assessed on sagittal T2 images. We also evaluated for evidence of adjacent vertebral body infarct defined as geographic marrow hyperintensity in the sagittal T2-weighted images. Presence or absence of contrast enhancement was assessed both on axial and sagittal post-contrast images (available for 20 of 24 subjects). Diffusion-weighted
MRI and apparent diffusion coefficient maps were used for evaluation of reduced diffusion (available for 8 of 24 subjects).
2.3 – Statistical Analysis Univariate analyses were conducted to identify individual factors associated with NMO or SCI. Exact Mann-Whitney tests were used to compare patients with and without NMO in terms of each numeric factor, whereas Fisher exact tests were used to identify binary factors associated with NMO. Stepwise selection in the context of binary logistic regression was used to identify sets of two or more independent predictors of NMO from among the factors showing a significant association with NMO in the univariate analysis. All statistical tests were conducted at the two-sided 5% significance level using SAS 9.3 (SAS Institute, Cary, NC). Figure 4 demonstrates an axial T2-weighted MRI for a selection of lesions from subjects with NMO acute myelitis or SCI.
3. RESULTS Figures 1 and 2 demonstrate typical imaging data obtained from subjects presenting with myelitis due to NMO and spinal cord infarct, respectively. The distribution of lesions in the spinal cord for NMO and spinal cord infarct patients is graphically depicted in Figure 3. Table 1 presents a summary of differences between the 2 subject populations. Race and gender proved helpful in favoring one diagnosis over the other – 69% of patients with NMO myelitis were non-Caucasian females, while the majority
of patients with spinal cord infarct were Caucasian and/or male. The location of the lesion along the craniocaudal axis also helped distinguish the two populations: center of NMO lesion was more commonly observed within 7 cm of either the conus terminus or cervicomedullary junction. It was more common to see contrast enhancement in NMO subjects during the acute presentation or for the T2hyperintensity to extend to the pial surface on axial images. The presence of T2 ‘bright spotty lesions’, defined as “very hyperintense spotty lesions on axial T2WI that are visually more hyperintense than that of surrounding cerebrospinal fluid without flow void effects” (Yonezu, 2014), helped distinguish NMO from spinal cord infarct. In 2 subjects a consensus between the two neuroradiologists could not be reached whether T2 bright spots were present or absent so these data were excluded. Finally, ‘owl’s eye’ sign - bilateral hyperintensities of the anterior horns on axial T2-weighted spinal cord MRI - was present in similar proportion of NMO (4/13) and SCI (4/11) cases (p=1.0). Reduced diffusion was present in the only NMO subject and 6 of the 7 spinal cord infarct subjects where the sequence was included (insufficient data for statistical analysis). Despite the many significant differences for individual factors of the two subject populations (Table 1), multivariable analysis did not identify 2 or more independent predictors to better distinguish NMO from spinal cord infarct. This most likely reflects insufficient statistical power for multivariate analysis in this single-institution study.
4. DISCUSSION
Our study confirms that imaging features of SCI and NMO overlap (Asgari et al. , 2013, Cheshire et al. , 1996, Lennon et al. , 2004, Masson et al. , 2004, Novy et al. , 2006, Robertson et al. , 2012, Salvador de la Barrera et al. , 2001, Tackley et al. , 2014, Weidauer et al. , 2002, Wingerchuk et al. , 2007, Wingerchuk, Lennon, 2006, Wingerchuk and Weinshenker, 2014). Lesion length, cross-sectional area, and the presence of cord expansion were similar in SCI and NMO. Yet, we were able to identify a number of differences in their radiologic appearance. Many of the observed differences could be plausibly explained on an anatomic and pathophysiologic basis. Vascular ischemic lesions tended to occur in areas where collateral circulation is poor (mid-thoracic cord and conus) and were not seen in the upper cervical cord, with its dual supply from vertebral and spinal arteries. In contrast, NMO lesions tended to cluster around ‘junctions’ - particularly the cervicomedullary and cervico-thoracic junctions (Fig 2) – areas with abundant collateral circulation. SCI lesions rarely extended to pial surface, presumably because collateral blood supply from spinal cord surface arteries arteries protected peripheral tissue against ischemic insult. Restricted blood supply in the acute phase of ischemia also likely explains lack of enhancement in during acute phase of SCI (as is the case in cerebral stroke). In contrast, NMO lesions were typically enhancing, suggesting intact (or increased) blood flow with inflammation-induced breakdown of blood-cord barrier. Considering that gadolinium-enhanced images yield important diagnostic clues, it may be advisable to obtain contrast-enhanced sequences in all patients with acute cord dysfunction.
Our analysis serves as a cautionary note against assigning too much significance to any single MRI feature. Such ‘classic’ radiographic feature of SCI as “owl eyes” (or “snake eye”) appearance (Figure 4), was as common in NMO as SCI. ‘Bright spotty lesions’, considered to be nearly pathognomonic for NMO(Yonezu, Ito, 2014), were observed in SCI as well, albeit less frequently than in NMO. Diffusion MRI was only available in minority of cases, and were unable to assess whether it will prove useful in differentiation of myelitis from stroke. Notably, the one NMO subject with the sequence had reduced diffusion in the spinal cord lesion; similarly, acute inflammatory multiple sclerosis lesions and infarct both show reduced diffusion in the brain. Pre-MRI probability of SCI vs NMO depended on gender and ethnicity of the subject. This is not surprising as NMO is over-represented among women (7:1 or even 10:1 ration in most studies(Collongues et al. , 2010)) and non-Caucasians. In keeping with these epidemiologic observations, all NMO subjects in our study were women, and 69% were non-Caucasian. SCI patients were as likely to be men as women and were overwhelmingly Caucasian. Demographics of the catchment area will, to some extent, dictate pre-test probability of NMO or SCI, but our results suggests that basic demographic facts need to be taken into consideration by radiologists in weighing probabilities of specific diagnoses for longitudinallyextensive transverse myelopathy lesions. Our single-center, retrospective study comparing NMO with SCI has some limitations. We focused exclusively on acute lesions in the cord, but, clearly, radiographic evidence of past disease activity in brain, cord or optic nerve would be
very helpful for diagnosis of NMO, and must be taken into account. We have only included in our analysis basic demographic/clinical data as would be available to radiologist at time of review. Additional clinical features, such as onset-to-nadir time of under 4 hours and presence of radicular pain, both of which favor SCI(Rigney et al., 2015), could further help discriminate between the two conditions. We compared NMO with SCI in view of their radiographically similar appearance and ‘grey-matter centeredness’. This decision improved our diagnostic certainty, but limited applicability of our finding to the ‘real life’ setting, where other causes of acute longitudinally extensive myelopathy must be considered(Kitley, 2012). Whether the observed differences between NMO and SCI carry over to other inflammatory causes of cord injury must await a larger, preferably, multi-center study that would include non-NMO myelopathies as a separate comparator. It must also be acknowledged that MRI protocols and quality differed between cases over the 10-year sampling period. Some sequences, such as diffusion-weighted imaging of the spine, are technically challenging and only recently have become available in clinical practice. Finally, our sample size was small, a limitation that is difficult to overcome in a single-institution study given the relative rarity of both NMO and SCI. Small sample size is likely responsible for failure of binary logistic regression to identify a set of two or more independent predictors that would differentiate NMO from SCI. It is likely that a combination of statistical significant individual factors in Table 1 can be used to more rapidly diagnose SCI or NMO as the underlying cause of acute myelopathy. However, a larger, multi-institution study is required to validate
and expand our findings leading to practical algorithm for successful differentiation of inflammatory from vascular causes of acute cord injury.
REFERENCES Asgari N, Skejoe HP, Lillevang ST, Steenstrup T, Stenager E, Kyvik KO. Modifications of longitudinally extensive transverse myelitis and brainstem lesions in the course of neuromyelitis optica (NMO): a population-based, descriptive study. BMC neurology. 2013;13:33. Cheshire WP, Santos CC, Massey EW, Howard JF, Jr. Spinal cord infarction: etiology and outcome. Neurology. 1996;47:321-30. Collongues N, Marignier R, Zephir H, Papeix C, Blanc F, Ritleng C, et al. Neuromyelitis optica in France: a multicenter study of 125 patients. Neurology. 2010;74:736-42. Kitley JL, Leite MI, George JS, Palace JA. The differential diagnosis of longitudinally extensive transverse myelitis. Multiple sclerosis. 2012;18:271-85. Lebouteux MV, Franques J, Guillevin R, Delmont E, Lenglet T, Bede P, et al. Revisiting the spectrum of lower motor neuron diseases with snake eyes appearance on magnetic resonance imaging. European journal of neurology 2014;21:1233-41. Lennon VA, Wingerchuk DM, Kryzer TJ, Pittock SJ, Lucchinetti CF, Fujihara K, et al. A serum autoantibody marker of neuromyelitis optica: distinction from multiple sclerosis. Lancet. 2004;364:2106-12. Masson C, Pruvo JP, Meder JF, Cordonnier C, Touze E, De La Sayette V, et al. Spinal cord infarction: clinical and magnetic resonance imaging findings and short term outcome. Journal of neurology, neurosurgery, and psychiatry. 2004;75:1431-5. Novy J, Carruzzo A, Maeder P, Bogousslavsky J. Spinal cord ischemia: clinical and imaging patterns, pathogenesis, and outcomes in 27 patients. Archives of neurology. 2006;63:1113-20. Rigney L, Cappelen-Smith C, Sebire D, Beran RG, Cordato D. Nontraumatic spinal cord ischaemic syndrome. Journal of clinical neuroscience : official journal of the Neurosurgical Society of Australasia. 2015. Robertson CE, Brown RD, Jr., Wijdicks EF, Rabinstein AA. Recovery after spinal cord infarcts: long-term outcome in 115 patients. Neurology. 2012;78:114-21. Salvador de la Barrera S, Barca-Buyo A, Montoto-Marques A, Ferreiro-Velasco ME, Cidoncha-Dans M, Rodriguez-Sotillo A. Spinal cord infarction: prognosis and recovery in a series of 36 patients. Spinal cord. 2001;39:520-5. Tackley G, Kuker W, Palace J. Magnetic resonance imaging in neuromyelitis optica. Multiple sclerosis. 2014.
Weidauer S, Nichtweiss M, Lanfermann H, Zanella FE. Spinal cord infarction: MR imaging and clinical features in 16 cases. Neuroradiology. 2002;44:851-7. Weinshenker BG, Wingerchuk DM, Pittock SJ, Lucchinetti CF, Lennon VA. NMO-IgG: a specific biomarker for neuromyelitis optica. Disease markers. 2006;22:197-206. Wingerchuk DM, Lennon VA, Lucchinetti CF, Pittock SJ, Weinshenker BG. The spectrum of neuromyelitis optica. Lancet neurology. 2007;6:805-15. Wingerchuk DM, Lennon VA, Pittock SJ, Lucchinetti CF, Weinshenker BG. Revised diagnostic criteria for neuromyelitis optica. Neurology. 2006;66:1485-9. Wingerchuk DM, Weinshenker BG. Neuromyelitis optica (Devic's syndrome). Handbook of clinical neurology. 2014;122:581-99. Yonezu T, Ito S, Mori M, Ogawa Y, Makino T, Uzawa A, et al. "Bright spotty lesions" on spinal magnetic resonance imaging differentiate neuromyelitis optica from multiple sclerosis. Multiple sclerosis. 2014;20:331-7.
FIGURES
Figure 1 – Typical imaging features of acute myelitis from neuromyelitis optica. This patient presented to the emergency department of an outside facility with abdominal numbness, hand tremors, urinary hesitancy, and forearm myalgias. MRI demonstrated a longitudinally-extensive central area of cord T2 hyperintensity and expansion. Axial T2-weighted image demonstrated T2 bright spotty lesions and T2 hyperintensity extending to the right ventral spinal cord periphery. Companion MRI of the brain and orbits was unremarkable. Subsequent aquaporin 4 antibody testing was positive, confirming the diagnosis of neuromyelitis optica.
Figure 2 – Typical imaging features of acute myelopathy from spinal cord ischemia progressing to infarct. This patient rapidly progressed to complete paraplegia and MRI (A-B) revealed spinal cord swelling and signal abnormality thought to be consistent with transverse myelitis versus vascular event. Emergency spinal angiogram revealed the artery of Adamkiewicz originating from the left L1 segmental artery (white dashed arrows), giving rise to a radiculomedullary artery (white arrows in C and D) contributing to the anterior spinal artery (black arrow). At the superior endplate level of the T12 vertebral body, the anterior spinal axis was disrupted in a pattern suggestive of a thrombotic or embolic occlusion (asterisk, C). After intra-arterial injection of tPA via the microcatheter, the anterior spinal artery axis is recanalized (black arrows in D).
Figure 3 – Bar graphs demonstrating the lesion distribution along the spinal axis for subjects with confirmed NMO (A) and spinal cord infarct (B) (bold outlines indicate a lesion that also touches the cord periphery). The majority of NMO acute myelitis lesions were near the cervicomedullary junction and extended to the periphery, whereas most spinal cord infarcts were located in the distal cord and lesions did not extend to the periphery.
Figure 4 – Axial T2-weighted MRI of the spinal cord lesion in several subjects with serologically proven NMO (A-E) and SCI (F-J). Many subjects presented with classic spinal cord MRI changes attributed to their final diagnoses (A, D-G). For example, note the concurrent geographic, T2 bright vertebral body infarct associated with spinal cord infarct in panel I. “Owl’s eye” sign, typically associated with SCI, was seen in a equal proportion of NMO cases (B. Several SCI cases demonstrate atypical features (H-J), including extension to the periphery.
Lesion
Location
46% (5)
0.32 ± 0.08
Relative cross-sectional area
Touches cord periphery
0.17 ± 0.06
Area (cm2)
73% (8)
Within 7 cm of conus
92% (12)
0.36 ± 0.12
0.22 ± 0.11
8% (1)
62% (8)
0% (0)
5.4 ± 3.5 54% (7)
3.4 ± 2.0
Vertebral body span
8.7 ± 5.5
9% (1)
8.9 ± 6.2
Lesion craniocaudal length (cm)
7.6 ± 4.7
31% (4)
100% (13)
43.9 ± 13.6
13
Neuromyelitis Optica
Involves cervicomedullary or cervicothoracic junctions Within 7 cm of foramen magnum
15.4 ± 4.2
91% (10)
Race (% Caucasian)
Center (1 to 24, where C3 = 3)
54% (6)
Sex (% female)
53.8 ± 16.5
Age (years)
Demographics
Spinal Cord Infarct 11
Factor
Total Subjects
Category
0.023
0.259
0.327
0.002
0.002
0.033
0.178
0.862
<0.001
0.005
0.011
0.142
NA
P-Value*
Table 1 – Comparison of clinical and MRI features for subjects with spinal cord infarct or acute myelitis from neuromyelitis optica.
92% (12)
36% (4/11) 85% (11)
92% (11/12)
0.142
1.000 0.022
0.006
Not all subjects included for these factors (the total included is denoted in the denominator, where different)
This work was supported by the Center for Advanced Imaging Innovation and Research, a National Institute for Biomedical Imaging and Bioengineering Biomedical Technology Resource Center [P41EB017183]. However, this funding source had no role influencing study design; in the collection, analysis and interpretation of data; in the writing of the report; or in the decision to submit the article for publication.
Role of Funding Source:
All authors report no actual or potential conflicts of interest including any financial, personal or other relationships with other people or organizations within three years of beginning the submitted work that could inappropriately influence, or be perceived to influence, their work.
Conflicts of Interest:
+
Statistical significance was determined a priori as P < 0.05.
* Numerical or ordinal data were analyzed with Mann-Whitney tests, while binary data were analyzed with Fisher exact test.
64% (7)
31% (4/13) 29% (2/7)
“Owl’s eyes’ sign Contrast enhancement present+
Cord expansion present
30% (3/10)
Bright spotty lesion present+
JOURNAL REQUIREMENTS = 3-5 bulleted points, none longer than 85 characters counting spaces.
“Owl eyes” gray matter T2 hyperintensities were not pathognomonic for cord infarct.
Patient sex and race can help separate spinal cord infarct from NMO acute myelitis.
NMO cord lesions usually are cervical, extend to pial surface, were enhancing on presentation and more commonly have T2 bright spotty lesions.
Highlights:
Acknowledgements: The authors wish to thank Ms. Tamar Bacon for her help in preparation of Figure 3.
PII: DOI: Reference:
S2211-0348(16)30048-7 http://dx.doi.org/10.1016/j.msard.2016.04.005 MSARD389
To appear in: Multiple Sclerosis and Related Disorders Received date: 30 September 2015 Revised date: 26 February 2016 Accepted date: 11 April 2016 Cite this article as: I Kister, E Johnson, E Raz, J Babb, J Loh and TM Shepherd, SPECIFIC MRI FINDINGS HELP DISTINGUISH ACUTE TRANSVERSE MYELITIS OF NEUROMYELITIS OPTICA FROM SPINAL CORD I N FA R C T I O N , Multiple Sclerosis and Related Disorders, http://dx.doi.org/10.1016/j.msard.2016.04.005 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 galley proof before it is published in its final citable 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.
SPECIFIC MRI FINDINGS HELP DISTINGUISH ACUTE TRANSVERSE MYELITIS OF NEUROMYELITIS OPTICA FROM SPINAL CORD INFARCTION Kister I1, Johnson E2, Raz E2, Babb, J2, Loh J2, Shepherd TM2 1
NYU Multiple Sclerosis Comprehensive Care Center, Department of Neurology,
New York University School of Medicine, New York, NY, USA. 2 Neuroradiology
Section, Department of Radiology, New York University School of
Medicine, New York, NY, USA.
Corresponding author:
Timothy Shepherd, M.D./Ph.D. [email protected] 660 First Ave., 2nd Floor, New York, NY 10016 Phone – 212-263-8487 Fax – 212-263-3838
Key words: Spinal cord, Myelitis, Devic’s Disease, Ischemia, MRI, Multiple Sclerosis Word count: 2128 words Abbreviations: Magnetic Resonance Imaging (MRI), Neuromyelitis Optica (NMO), Spinal Cord Infarct (SCI).
ABSTRACT Background: There is substantial overlap between MRI of acute spinal cord lesions from neuromyelitis optica (NMO) and spinal cord infarct (SCI) in clinical practice. However, early differentiation is important since management approaches to minimize morbidity from NMO or SCI differ significantly. Objective: To identify MRI features at initial presentation that may help to differentiate NMO acute myelitis from SCI. Methods: 2 board-certified neuroradiologists, blinded to final diagnosis, retrospectively characterized MRI features at symptom onset for subjects with serologically-proven NMO (N=13) or SCI (N =11) from a single institution. Univariate and multivariate analyses were used to identify factors associated with NMO or SCI. Results: SCI was more common in men and Caucasians, while NMO was more common in non-Caucasian women (P < 0.05). MRI features associated with NMO acute myelitis (P<0.05) included location within 7-cm of cervicomedullary junction; lesion extending to pial surface; ‘bright spotty lesions’ on axial T2 MRI; and gadolinium enhancement. Patient’s age, lesion length and cross-sectional area, cord expansion, and the “owl’s eyes” sign did not differ between the two groups (P > 0.05). Conclusion: Along with patient demographic characteristics, lesion features on MRI, including lesion location, extension to pial border and presence of ‘bright spotty lesion’ can help differentiate acute myelitis of NMO from SCI in the acute setting.
1. INTRODUCTION Acute spinal cord injury, characterized by rapid evolution of bilateral sensori-motor deficits and autonomic dysfunction is a neurologic emergency that carries risk of considerable long-term disability. When MRI reveals a longitudinally extensive cord lesion, the main diagnostic considerations are acute transverse myelitis secondary to infectious or autoimmune etiologies, versus spinal cord ischemia (SCI)(Kitley et al. , 2012). It is essential to establish the correct diagnosis promptly in order to initiate treatment (and avoid inappropriate therapy). It would be advantageous to know whether there are any etiology-specific MRI features that could be used to improve diagnostic accuracy in the acute setting. To address this question, we carried out a retrospective spinal cord MRI review of patients evaluated at our institution with acute myelopathy due to Neuromyelitis Optica (NMO) and patients with spinal cord ischemia (SCI). We choose to focus on subset of acute myelopathy due to NMO because of a very high degree of diagnostic certainty in these cases, and because lesions of NMO and SCI both tend to be several vertebral bodies long, expansile and centered around gray matter. Further, during the presentation of either condition, the companion brain MRI is typically normal, making their differentiation especially challenging.
2. MATERIALS AND METHODS 2.1 – Subject Selection
This was a HIPAA-compliant retrospective case-control study. We performed a single-institution retrospective chart review over a 10-year period (2005-2014) to identify cases with final diagnosis of spinal cord infarct (SCI) or acute myelitis due to clinically definite NMO (Wingerchuk et al., 2006) The diagnosis of spinal cord infarct (SCI) was made in 10 patients based on a combination of clinical and MRI features and, where available, angiographic evidence. Direct angiographic confirmation was available in 4 cases (anterior spinal artery occlusion, hyperemia in the infarcted spinal cord territory and increased visualization of the collateral circulation). In 4 patients with typical features of SCI (hyperacute onset-to-nadir time of <4 hours and severe radicular pain at onset), diagnosis of SCI was supported by such MRI features that have never been described in myelitis: vertebral and/or muscle infarction in 3 cases and persistent root and ventral cord enhancement several months after event in 2 cases. In two patients, symptoms developed in the immediate following medical procedures - discectomy in one case and epidural injection in a patient with hemochromatosis in the other case. All 13 patients with NMO myelitis fulfilled diagnostic criteria for NMO and were AQP-4-Ab seropositive.
2.2 – Data Analysis Patient demographic information that would be readily available to radiologists (age, gender and race) were included in the analysis. MRI of the spine obtained within 2 days of clinical presentation was reviewed by two board-certified neuroradiologists (TS and JL) who were blinded to patient demographic
information and the final diagnosis.
For some patients, initial MRI study was
obtained at an outside institution prior to transfer to our medical center, however the images were loaded into the local radiology PACS. All studies included axial and sagittal T2-weighted sequences of the total spine. The sagittal extent of the T2 hyperintense acute lesion was characterized by the number of vertebral bodies spanned and craniocaudal length in cm, e.g. a lesion between C3 and C5 could be characterized as 3 vertebral bodies, or 3.4-cm long by measurement on the mid-sagittal T2 image. The position of the lesion along the cranio-caudal axis was recorded as the vertebral body at the lesion center. We measured the distance of the cranial and caudal lesion edges to the cervicomedullary junction and cord termination at the conus respectively. We recorded whether the lesion crossed the craniocervical, cervicothoracic or thoracolumbar junctions. The midpoint of the lesion was characterized using the axial T2 images. The cross-sectional area of the lesion and respective overall cord cross-sectional area were measured manually on the same T2-weighted axial image; then the ratio of lesion-to-cord cross-sectional area was calculated. We recorded presence of T2 'bright spotty’ lesions(Yonezu et al. , 2014), ‘owl (or snake) eyes’ sign Lebouteux et al. , 2014) and whether the lesion extended to pial surface of cord on any axial T2W image. The presence or absence of cord expansion was assessed on sagittal T2 images. We also evaluated for evidence of adjacent vertebral body infarct defined as geographic marrow hyperintensity in the sagittal T2-weighted images. Presence or absence of contrast enhancement was assessed both on axial and sagittal post-contrast images (available for 20 of 24 subjects). Diffusion-weighted
MRI and apparent diffusion coefficient maps were used for evaluation of reduced diffusion (available for 8 of 24 subjects).
2.3 – Statistical Analysis Univariate analyses were conducted to identify individual factors associated with NMO or SCI. Exact Mann-Whitney tests were used to compare patients with and without NMO in terms of each numeric factor, whereas Fisher exact tests were used to identify binary factors associated with NMO. Stepwise selection in the context of binary logistic regression was used to identify sets of two or more independent predictors of NMO from among the factors showing a significant association with NMO in the univariate analysis. All statistical tests were conducted at the two-sided 5% significance level using SAS 9.3 (SAS Institute, Cary, NC). Figure 4 demonstrates an axial T2-weighted MRI for a selection of lesions from subjects with NMO acute myelitis or SCI.
3. RESULTS Figures 1 and 2 demonstrate typical imaging data obtained from subjects presenting with myelitis due to NMO and spinal cord infarct, respectively. The distribution of lesions in the spinal cord for NMO and spinal cord infarct patients is graphically depicted in Figure 3. Table 1 presents a summary of differences between the 2 subject populations. Race and gender proved helpful in favoring one diagnosis over the other – 69% of patients with NMO myelitis were non-Caucasian females, while the majority
of patients with spinal cord infarct were Caucasian and/or male. The location of the lesion along the craniocaudal axis also helped distinguish the two populations: center of NMO lesion was more commonly observed within 7 cm of either the conus terminus or cervicomedullary junction. It was more common to see contrast enhancement in NMO subjects during the acute presentation or for the T2hyperintensity to extend to the pial surface on axial images. The presence of T2 ‘bright spotty lesions’, defined as “very hyperintense spotty lesions on axial T2WI that are visually more hyperintense than that of surrounding cerebrospinal fluid without flow void effects” (Yonezu, 2014), helped distinguish NMO from spinal cord infarct. In 2 subjects a consensus between the two neuroradiologists could not be reached whether T2 bright spots were present or absent so these data were excluded. Finally, ‘owl’s eye’ sign - bilateral hyperintensities of the anterior horns on axial T2-weighted spinal cord MRI - was present in similar proportion of NMO (4/13) and SCI (4/11) cases (p=1.0). Reduced diffusion was present in the only NMO subject and 6 of the 7 spinal cord infarct subjects where the sequence was included (insufficient data for statistical analysis). Despite the many significant differences for individual factors of the two subject populations (Table 1), multivariable analysis did not identify 2 or more independent predictors to better distinguish NMO from spinal cord infarct. This most likely reflects insufficient statistical power for multivariate analysis in this single-institution study.
4. DISCUSSION
Our study confirms that imaging features of SCI and NMO overlap (Asgari et al. , 2013, Cheshire et al. , 1996, Lennon et al. , 2004, Masson et al. , 2004, Novy et al. , 2006, Robertson et al. , 2012, Salvador de la Barrera et al. , 2001, Tackley et al. , 2014, Weidauer et al. , 2002, Wingerchuk et al. , 2007, Wingerchuk, Lennon, 2006, Wingerchuk and Weinshenker, 2014). Lesion length, cross-sectional area, and the presence of cord expansion were similar in SCI and NMO. Yet, we were able to identify a number of differences in their radiologic appearance. Many of the observed differences could be plausibly explained on an anatomic and pathophysiologic basis. Vascular ischemic lesions tended to occur in areas where collateral circulation is poor (mid-thoracic cord and conus) and were not seen in the upper cervical cord, with its dual supply from vertebral and spinal arteries. In contrast, NMO lesions tended to cluster around ‘junctions’ - particularly the cervicomedullary and cervico-thoracic junctions (Fig 2) – areas with abundant collateral circulation. SCI lesions rarely extended to pial surface, presumably because collateral blood supply from spinal cord surface arteries arteries protected peripheral tissue against ischemic insult. Restricted blood supply in the acute phase of ischemia also likely explains lack of enhancement in during acute phase of SCI (as is the case in cerebral stroke). In contrast, NMO lesions were typically enhancing, suggesting intact (or increased) blood flow with inflammation-induced breakdown of blood-cord barrier. Considering that gadolinium-enhanced images yield important diagnostic clues, it may be advisable to obtain contrast-enhanced sequences in all patients with acute cord dysfunction.
Our analysis serves as a cautionary note against assigning too much significance to any single MRI feature. Such ‘classic’ radiographic feature of SCI as “owl eyes” (or “snake eye”) appearance (Figure 4), was as common in NMO as SCI. ‘Bright spotty lesions’, considered to be nearly pathognomonic for NMO(Yonezu, Ito, 2014), were observed in SCI as well, albeit less frequently than in NMO. Diffusion MRI was only available in minority of cases, and were unable to assess whether it will prove useful in differentiation of myelitis from stroke. Notably, the one NMO subject with the sequence had reduced diffusion in the spinal cord lesion; similarly, acute inflammatory multiple sclerosis lesions and infarct both show reduced diffusion in the brain. Pre-MRI probability of SCI vs NMO depended on gender and ethnicity of the subject. This is not surprising as NMO is over-represented among women (7:1 or even 10:1 ration in most studies(Collongues et al. , 2010)) and non-Caucasians. In keeping with these epidemiologic observations, all NMO subjects in our study were women, and 69% were non-Caucasian. SCI patients were as likely to be men as women and were overwhelmingly Caucasian. Demographics of the catchment area will, to some extent, dictate pre-test probability of NMO or SCI, but our results suggests that basic demographic facts need to be taken into consideration by radiologists in weighing probabilities of specific diagnoses for longitudinallyextensive transverse myelopathy lesions. Our single-center, retrospective study comparing NMO with SCI has some limitations. We focused exclusively on acute lesions in the cord, but, clearly, radiographic evidence of past disease activity in brain, cord or optic nerve would be
very helpful for diagnosis of NMO, and must be taken into account. We have only included in our analysis basic demographic/clinical data as would be available to radiologist at time of review. Additional clinical features, such as onset-to-nadir time of under 4 hours and presence of radicular pain, both of which favor SCI(Rigney et al., 2015), could further help discriminate between the two conditions. We compared NMO with SCI in view of their radiographically similar appearance and ‘grey-matter centeredness’. This decision improved our diagnostic certainty, but limited applicability of our finding to the ‘real life’ setting, where other causes of acute longitudinally extensive myelopathy must be considered(Kitley, 2012). Whether the observed differences between NMO and SCI carry over to other inflammatory causes of cord injury must await a larger, preferably, multi-center study that would include non-NMO myelopathies as a separate comparator. It must also be acknowledged that MRI protocols and quality differed between cases over the 10-year sampling period. Some sequences, such as diffusion-weighted imaging of the spine, are technically challenging and only recently have become available in clinical practice. Finally, our sample size was small, a limitation that is difficult to overcome in a single-institution study given the relative rarity of both NMO and SCI. Small sample size is likely responsible for failure of binary logistic regression to identify a set of two or more independent predictors that would differentiate NMO from SCI. It is likely that a combination of statistical significant individual factors in Table 1 can be used to more rapidly diagnose SCI or NMO as the underlying cause of acute myelopathy. However, a larger, multi-institution study is required to validate
and expand our findings leading to practical algorithm for successful differentiation of inflammatory from vascular causes of acute cord injury.
REFERENCES Asgari N, Skejoe HP, Lillevang ST, Steenstrup T, Stenager E, Kyvik KO. Modifications of longitudinally extensive transverse myelitis and brainstem lesions in the course of neuromyelitis optica (NMO): a population-based, descriptive study. BMC neurology. 2013;13:33. Cheshire WP, Santos CC, Massey EW, Howard JF, Jr. Spinal cord infarction: etiology and outcome. Neurology. 1996;47:321-30. Collongues N, Marignier R, Zephir H, Papeix C, Blanc F, Ritleng C, et al. Neuromyelitis optica in France: a multicenter study of 125 patients. Neurology. 2010;74:736-42. Kitley JL, Leite MI, George JS, Palace JA. The differential diagnosis of longitudinally extensive transverse myelitis. Multiple sclerosis. 2012;18:271-85. Lebouteux MV, Franques J, Guillevin R, Delmont E, Lenglet T, Bede P, et al. Revisiting the spectrum of lower motor neuron diseases with snake eyes appearance on magnetic resonance imaging. European journal of neurology 2014;21:1233-41. Lennon VA, Wingerchuk DM, Kryzer TJ, Pittock SJ, Lucchinetti CF, Fujihara K, et al. A serum autoantibody marker of neuromyelitis optica: distinction from multiple sclerosis. Lancet. 2004;364:2106-12. Masson C, Pruvo JP, Meder JF, Cordonnier C, Touze E, De La Sayette V, et al. Spinal cord infarction: clinical and magnetic resonance imaging findings and short term outcome. Journal of neurology, neurosurgery, and psychiatry. 2004;75:1431-5. Novy J, Carruzzo A, Maeder P, Bogousslavsky J. Spinal cord ischemia: clinical and imaging patterns, pathogenesis, and outcomes in 27 patients. Archives of neurology. 2006;63:1113-20. Rigney L, Cappelen-Smith C, Sebire D, Beran RG, Cordato D. Nontraumatic spinal cord ischaemic syndrome. Journal of clinical neuroscience : official journal of the Neurosurgical Society of Australasia. 2015. Robertson CE, Brown RD, Jr., Wijdicks EF, Rabinstein AA. Recovery after spinal cord infarcts: long-term outcome in 115 patients. Neurology. 2012;78:114-21. Salvador de la Barrera S, Barca-Buyo A, Montoto-Marques A, Ferreiro-Velasco ME, Cidoncha-Dans M, Rodriguez-Sotillo A. Spinal cord infarction: prognosis and recovery in a series of 36 patients. Spinal cord. 2001;39:520-5. Tackley G, Kuker W, Palace J. Magnetic resonance imaging in neuromyelitis optica. Multiple sclerosis. 2014.
Weidauer S, Nichtweiss M, Lanfermann H, Zanella FE. Spinal cord infarction: MR imaging and clinical features in 16 cases. Neuroradiology. 2002;44:851-7. Weinshenker BG, Wingerchuk DM, Pittock SJ, Lucchinetti CF, Lennon VA. NMO-IgG: a specific biomarker for neuromyelitis optica. Disease markers. 2006;22:197-206. Wingerchuk DM, Lennon VA, Lucchinetti CF, Pittock SJ, Weinshenker BG. The spectrum of neuromyelitis optica. Lancet neurology. 2007;6:805-15. Wingerchuk DM, Lennon VA, Pittock SJ, Lucchinetti CF, Weinshenker BG. Revised diagnostic criteria for neuromyelitis optica. Neurology. 2006;66:1485-9. Wingerchuk DM, Weinshenker BG. Neuromyelitis optica (Devic's syndrome). Handbook of clinical neurology. 2014;122:581-99. Yonezu T, Ito S, Mori M, Ogawa Y, Makino T, Uzawa A, et al. "Bright spotty lesions" on spinal magnetic resonance imaging differentiate neuromyelitis optica from multiple sclerosis. Multiple sclerosis. 2014;20:331-7.
FIGURES
Figure 1 – Typical imaging features of acute myelitis from neuromyelitis optica. This patient presented to the emergency department of an outside facility with abdominal numbness, hand tremors, urinary hesitancy, and forearm myalgias. MRI demonstrated a longitudinally-extensive central area of cord T2 hyperintensity and expansion. Axial T2-weighted image demonstrated T2 bright spotty lesions and T2 hyperintensity extending to the right ventral spinal cord periphery. Companion MRI of the brain and orbits was unremarkable. Subsequent aquaporin 4 antibody testing was positive, confirming the diagnosis of neuromyelitis optica.
Figure 2 – Typical imaging features of acute myelopathy from spinal cord ischemia progressing to infarct. This patient rapidly progressed to complete paraplegia and MRI (A-B) revealed spinal cord swelling and signal abnormality thought to be consistent with transverse myelitis versus vascular event. Emergency spinal angiogram revealed the artery of Adamkiewicz originating from the left L1 segmental artery (white dashed arrows), giving rise to a radiculomedullary artery (white arrows in C and D) contributing to the anterior spinal artery (black arrow). At the superior endplate level of the T12 vertebral body, the anterior spinal axis was disrupted in a pattern suggestive of a thrombotic or embolic occlusion (asterisk, C). After intra-arterial injection of tPA via the microcatheter, the anterior spinal artery axis is recanalized (black arrows in D).
Figure 3 – Bar graphs demonstrating the lesion distribution along the spinal axis for subjects with confirmed NMO (A) and spinal cord infarct (B) (bold outlines indicate a lesion that also touches the cord periphery). The majority of NMO acute myelitis lesions were near the cervicomedullary junction and extended to the periphery, whereas most spinal cord infarcts were located in the distal cord and lesions did not extend to the periphery.
Figure 4 – Axial T2-weighted MRI of the spinal cord lesion in several subjects with serologically proven NMO (A-E) and SCI (F-J). Many subjects presented with classic spinal cord MRI changes attributed to their final diagnoses (A, D-G). For example, note the concurrent geographic, T2 bright vertebral body infarct associated with spinal cord infarct in panel I. “Owl’s eye” sign, typically associated with SCI, was seen in a equal proportion of NMO cases (B. Several SCI cases demonstrate atypical features (H-J), including extension to the periphery.
Lesion
Location
46% (5)
0.32 ± 0.08
Relative cross-sectional area
Touches cord periphery
0.17 ± 0.06
Area (cm2)
73% (8)
Within 7 cm of conus
92% (12)
0.36 ± 0.12
0.22 ± 0.11
8% (1)
62% (8)
0% (0)
5.4 ± 3.5 54% (7)
3.4 ± 2.0
Vertebral body span
8.7 ± 5.5
9% (1)
8.9 ± 6.2
Lesion craniocaudal length (cm)
7.6 ± 4.7
31% (4)
100% (13)
43.9 ± 13.6
13
Neuromyelitis Optica
Involves cervicomedullary or cervicothoracic junctions Within 7 cm of foramen magnum
15.4 ± 4.2
91% (10)
Race (% Caucasian)
Center (1 to 24, where C3 = 3)
54% (6)
Sex (% female)
53.8 ± 16.5
Age (years)
Demographics
Spinal Cord Infarct 11
Factor
Total Subjects
Category
0.023
0.259
0.327
0.002
0.002
0.033
0.178
0.862
<0.001
0.005
0.011
0.142
NA
P-Value*
Table 1 – Comparison of clinical and MRI features for subjects with spinal cord infarct or acute myelitis from neuromyelitis optica.
92% (12)
36% (4/11) 85% (11)
92% (11/12)
0.142
1.000 0.022
0.006
Not all subjects included for these factors (the total included is denoted in the denominator, where different)
This work was supported by the Center for Advanced Imaging Innovation and Research, a National Institute for Biomedical Imaging and Bioengineering Biomedical Technology Resource Center [P41EB017183]. However, this funding source had no role influencing study design; in the collection, analysis and interpretation of data; in the writing of the report; or in the decision to submit the article for publication.
Role of Funding Source:
All authors report no actual or potential conflicts of interest including any financial, personal or other relationships with other people or organizations within three years of beginning the submitted work that could inappropriately influence, or be perceived to influence, their work.
Conflicts of Interest:
+
Statistical significance was determined a priori as P < 0.05.
* Numerical or ordinal data were analyzed with Mann-Whitney tests, while binary data were analyzed with Fisher exact test.
64% (7)
31% (4/13) 29% (2/7)
“Owl’s eyes’ sign Contrast enhancement present+
Cord expansion present
30% (3/10)
Bright spotty lesion present+
JOURNAL REQUIREMENTS = 3-5 bulleted points, none longer than 85 characters counting spaces.
“Owl eyes” gray matter T2 hyperintensities were not pathognomonic for cord infarct.
Patient sex and race can help separate spinal cord infarct from NMO acute myelitis.
NMO cord lesions usually are cervical, extend to pial surface, were enhancing on presentation and more commonly have T2 bright spotty lesions.
Highlights:
Acknowledgements: The authors wish to thank Ms. Tamar Bacon for her help in preparation of Figure 3.