Specific pattern of gadolinium enhancement in spondylotic myelopathy Eoin P. Flanagan, MB.BCh,1 Karl N. Krecke, MD,2 Richard W. Marsh, MD,3 Caterina Giannini, MD, PhD,4 B Mark Keegan, MD,1 and Brian G. Weinshenker, MD,1
Running Head: Enhancement in spondylosis Department of Neurology,1 Radiology,2 Neurosurgery,3 and Anatomical Pathology,4 Mayo Clinic, Rochester, MN, USA
Correspondence to: Dr. Brian G. Weinshenker, Mayo Clinic, Department of Neurology, 200 First Street SW, Rochester, MN, 55905, USA. E-mail: [email protected]
Word Count: Title: characters with spaces = 69 (maximum 80) Running Title: characters with spaces = 26 (maximum 50) Abstract: 249 (max 250) Text = Introduction:
309 (Limit 500 words)
Methods/Results:
2996 (No limit)
Discussion:
1430 (Limit 1500 words)
References = 23 (maximum 40) Figures/Tables: 2 Tables; 6 figures Supplemental data: none
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process which may lead to differences between this version and the Version of Record. Please cite this article as an ‘Accepted Article’, doi: 10.1002/ana.24184
Annals of Neurology
Key words:
Cervical spondylosis; gadolinium enhancement; magnetic resonance
imaging; persistent; spinal cord tumor; transverse myelitis;
Financial Disclosure Statements Drs. Flanagan, Marsh, Gianinni and Krecke have no disclosures. Dr Weinshenker has received a research grant from the Guthy Jackson Foundation. He receives royalties from RSR for a technology license related to a test for aquaporin-4 autoantibodies for diagnosis of neuromyelitis optica. He serves on data safety monitoring committees for Novartis, Biogen-Idec and Mitsubishi pharmaceutical companies, and serves on an adjudication panel for Medimmune Pharmaceuticals. He served as a consultant for GlaxoSmithKline, Elan, Ono, Chugai and Alexion pharmaceutical companies and for Asahi Kasei Medical Company. He serves on editorial boards for Neurology, the Canadian Journal of Neurological Sciences, and Turkish Journal of Neurology. Keegan is compensated as a Chief Editor of eMedicine, has consulted for Bristol Meyers Squibb, Novartis and Bionest and receive research support from Terumo BCT. Conflict of Interest Statement The authors declare that they have no conflicts of interest. Author contributions Dr Flanagan was involved in abstracting the clinical information from the electronic medical record, drafting and revising the manuscript for content, including medical writing for content, analysis and interpretation of data, and acquisition of data. Dr Krecke was involved in revising the manuscript for content and analysis and interpretation of data. Dr Marsh was involved in revising the manuscript for content and analysis and interpretation of data.
John Wiley & Sons
Page 2 of 35
Page 3 of 35
Annals of Neurology
Dr Giannini was involved in revising the manuscript for content and analysis and interpretation of data. Dr Keegan was involved in revising the manuscript for content and analysis and interpretation of data. Dr Weinshenker was involved in drafting and revising the manuscript for content, including medical writing for content, analysis and interpretation of data, acquisition of data and study supervision.
John Wiley & Sons
Annals of Neurology
Abstract Objective: To highlight a specific under-recognized radiological feature of spondylotic myelopathy often resulting in misdiagnosis. Methods: Patients evaluated between 1/1/1996-12/31/2012 who met the following criteria were included: 1) suspected spondylotic myelopathy; 2) gadolinium enhancement detected; 3) spinal surgery performed. Results: Fifty six patients (70% men) whose median age was 53.5 years (range, 24-80) were included. Spinal cord MRI (cervical, 52; thoracic, 4) revealed longitudinal-spindleshaped-T2-signal hyperintensity (100%) and cord enlargement (79%) accompanied by a characteristic pancake-like transverse band of gadolinium enhancement in 41 (73%) typically immediately caudal to the site of maximal spinal stenosis. Forty (71%) patients were initially diagnosed with neoplastic or inflammatory myelopathies and decompressive surgery was delayed by a median of 11 months (range 1-64). Spinal cord biopsy in 6 did not reveal any alternative diagnosis. Ninety five percent were stable or improved. Gadolinium enhancement persisted in 75% at 12 months, raising concern about accuracy of initial diagnosis. Twenty patients required a gait aid (36%) at last follow up (median 60 months, range 10-172). The need for a gait aid pre-operatively (p=0.005), but not delay to surgery, predicted need for gait aid at final follow-up. Interpretation: Transverse pancake-like gadolinium enhancement associated with and just caudal to the site of maximal stenosis and at the rostrocaudal midpoint of a spindleshaped T2-hyperintensity suggests that spondylosis is the cause of the myelopathy. Persistent enhancement for months to years following decompressive surgery is common.
John Wiley & Sons
Page 4 of 35
Page 5 of 35
Annals of Neurology
Recognition is important to prevent inappropriate interventions or delay in consideration of a potentially beneficial decompressive surgery.
John Wiley & Sons
Annals of Neurology
Cervical spondylosis is the commonest cause of myelopathy accounting for 23.6% of all non-traumatic myelopathies in one study.1 It is essential to recognize the spectrum of radiological features of cervical spondylotic myelopathy (SM), including those that are atypical, given its frequency and the potential for successful treatment with surgical decompression, although class I evidence supporting the benefits of surgery is lacking. Magnetic resonance imaging (MRI) of cervical SM typically reveals degenerative disease of the spine, canal stenosis and cord compression.2 In cervical SM spinal cord T2-signal abnormalities occur in approximately 15%3 and gadolinium enhancement in 7.3%.4 We have recognized a distinctive pattern of long fusiform T2-signal abnormality with flat, transversely oriented pancake-like enhancement at the site of maximal compression in cervical SM.5, 6 This radiological feature of SM often leads to misdiagnosis as inflammatory6 or neoplastic conditions,7 delay of potentially effective surgical treatment and occasionally inappropriate biopsy. Prompt, accurate and confident diagnosis of SM is important for a number of reasons. Although a recommendation of surgical decompression is based on a thorough differential diagnosis of myelopathy, a rigorous assessment of the severity of stenosis and an assessment of risks and benefits, surgical decompression may improve neurological prognosis.8 Investigations to detect alternative etiologies are costly and often imperfectly sensitive (e.g. angiotensin converting enzyme to diagnose neurosarcoidosis). Hence, exclusion of alternative causes of myelopathy associated with gadolinium enhancement with complete certainty is impossible. Spinal cord biopsy carries significant risk of neurological deficits and may be non-diagnostic. Empiric treatments for suspected alternative etiologies (e.g. long-term immunosuppressant medications) may be harmful. We observed persisting gadolinium
John Wiley & Sons
Page 6 of 35
Page 7 of 35
Annals of Neurology
enhancement for months-years after surgery for SM, leading physicians to question the accuracy of the diagnosis and re-investigate for other causes despite the lack of clinical worsening. Herein we describe the clinical features and radiological characteristics including their evolution in 56 patients with SM associated with gadolinium enhancement.
Subjects and Methods Identifying Patients. Patients were identified retrospectively from 1/1/1996-12/31/2012. Inclusion criteria were: 1) suspected SM; 2) pre-operative MRI available; 3) gadolinium enhancement on MRI; 4) spinal surgery undertaken (primary decompression or secondary decompression during spinal cord biopsy). Patients were excluded if they did not have myelopathy, if they did not undergo MRI or if they did not have surgery. To identify patients we: 1) prospectively collected patients evaluated by one of the authors either at Mayo Clinic [n=12] or by electronic-consultation [n=1]; 2) searched the surgical records for additional patients who underwent cervical spine decompression surgery for cervical spondylosis from 2003-2012 [n=34]; and 3) reviewed patients who underwent conjunctival biopsy to assess for sarcoidosis, a commonly suspected alternative diagnosis in this context, in patients with a progressive myelopathy of unknown origin [n=9]. In total 56 patients were included. Two patients with pancake-like gadolinium enhancement meeting other criteria were excluded as surgery was not performed during the follow-up period. One patient presented with a subacute myelopathy with weakness, stiffness and sensory loss. At last clinical follow-up 7 months after onset the patient was stable but had continuing symptoms of myelopathy and the
John Wiley & Sons
Annals of Neurology
enhancing lesion persisted. A second patient presented with a chronically progressive myelopathy and had stable symptoms and persistent gadolinium enhancement 13 months after onset. Both patients had moderate spondylotic changes and the final diagnosis in both was myelopathy of uncertain etiology. Surgery was not recommended as the significance of this radiologic finding was not recognized and because the stenosis was moderate rather than severe. None of the 20 patients with cervical or thoracic radiculopathy identified by our search of the surgical records and excluded as they did not have SM had gadolinium enhancement within the spinal cord. To determine if the radiological pattern of gadolinium enhancement with SM that we describe in this study is seen with other etiologies of spinal cord pathology we reviewed the MRIs of 136 patients with alternative diagnoses that were encountered during our search of the surgical records (n=43), patients selected from our database of longitudinally extensive myelopathies (n=56), patients with transverse myelitis from Olmsted County (n=27) and patients from a series of primary intramedullary spinal cord lymphoma (n=10)9. Six patients in our study have been reported previously.5, 6 Follow up Information. We contacted patients who had not been evaluated within 6 months of the study ascertainment date by telephone. We recorded their current clinical symptomatology including ambulatory status, additional surgeries undertaken and alternative diagnoses made. We also requested that their most recent MRI of the spine (cervical and/or thoracic) be sent to us for review.
John Wiley & Sons
Page 8 of 35
Page 9 of 35
Annals of Neurology
Neuroimaging. MRI’s from each patient were reviewed by a licensed neuroradiologist (K.N.K). We recorded the length of T2-signal abnormality, the pattern of gadolinium enhancement (sagittal and axial) and the duration of gadolinium enhancement prior to and after surgery. A total of 257 MRI’s at different time points (median of 4 MRI’s per patient; range, 1-13) were reviewed. Pre-operative MRI’s were performed with gadolinium in 54 patients and without gadolinium in two patients in whom post-operative pancake-like enhancement was detected. Pathology. All pathology samples were evaluated at Mayo Clinic by an experienced licensed neuropathologist. The photomicrograph was acquired in JPEG format and built in photoshop. A slight adjustment in brightness was made for optimal quality. Statistical Analysis. Descriptive summary statistics were reported as median (range, minimum-maximum) for continuous variables and frequencies and percentages for categorical variables. Comparisons were performed using Wilcoxon rank sum test or Fisher’s exact tests as appropriate using JMP 8.0 software (SAS®). Standard protocol approvals, registrations, and patient consents All patients consented to the use of their medical record for research purposes.
Results Patient demographics and clinical features Fifty-six patients (92% Caucasian; 8% African-American) were included (Table 1). Onset was subacute (≤8 weeks) in 21 (37%) and insidious (>8weeks) in 35 (63%). The most common symptoms were: numbness, paresthesias and/or pain (localized or shooting), 51 (91%); weakness, 48 (86%); bowel/bladder disturbance, 29 (52%); classic or “reverse” (with neck extension) Lhermitte’s symptom, 9 (16%). Examination revealed:
John Wiley & Sons
Annals of Neurology
spastic quadriparesis, 14; spastic paraparesis, 11; Brown-Sequard syndrome, 6; upper extremity weakness alone, 6; hemiparesis, 4; sensory ataxia, 3; spastic gait without weakness, 3; spastic-ataxic gait, 2; and mild myelopathy without weakness, 7. Ambulatory status immediately pre-operatively was: no gait aid, 41 (73%); cane, 5 (9%); walker, 4 (7%); and wheelchair, 6 (11%). Spine MRI (Table 1) Spondylotic myelopathy affected cervical cord in 52 (93%) and thoracic cord in 4 (7%). All patients had T2-weighted signal abnormalities (45% extending ≥3 vertebral segments) typically spindle shaped on sagittal images (median length, 33 mm; range, 6140) and central on axial images (median width of cervical T2-lesions 11.2 mm, range 2.5-18.5). T2-signal heterogeneity at the gadolinium enhancement site was noted in 13 (23%, Figure 1[B1 and C1]). Thoracic epidural lipomatosis contributed to stenosis in one. No cystic changes or syrinxes were noted. Fifty five patients (98%) had a single enhancing lesion (all extending ≤1 segment) and one had 3 focal regions of enhancement at three sites of severe narrowing. Enhancement was typically located at the middle of the T2-signal hyperintensity on sagittal images. Gadolinium enhancement frequently had a characteristic transverse band or flat pancake-like appearance in the sagittal imaging plane (41 patients [73%]). The transverse band of enhancement was complete in 67%, while in 6% there was central sparing noted on sagittal images (Figure 1C). The band of enhancement was located just inferior to the site of maximum stenosis, most commonly at C5 or C6 (48%). In the other 15 patients (27%) a focal region of enhancement that did not form a transverse band was seen in a similar rostrocaudal position within the area of T2 signal abnormality (Figure
John Wiley & Sons
Page 10 of 35
Page 11 of 35
Annals of Neurology
1A2). The rostrocaudal length of the enhancement was slightly greater in the dorsal than in the ventral cord in 14 patients (25%, Figure 1C2, Figure 2A2). On axial postgadolinium images, the enhancement pattern was circumferential located in the white matter, sparing gray matter in 29 of 51 with available images (57%: Figures 2A3 and 3[A-D]), focal in 14 (27%; peripheral, 12: Figure 3F; central, 2); multifocal/patchy 7 (14%) and diffuse in 1 (2%). Overall, enhancement affected white matter and spared gray matter in 45 of 51 (88%). The circumferential pattern was sometimes incomplete (Figure 3D), and occasionally involved the hemicord (Figure 3E). Pial enhancement was never observed. Other Investigations Head MRI was essentially normal in 43 patients investigated. Cerebrospinal fluid (CSF) commonly showed elevated protein (Table 1). Chest computed tomography revealed hilar adenopathy in 2 of 33 evaluated (6% [see potentially confounding illnesses below]). Whole body 18-Flourodeoxyglucose positron emission tomography (FDG-PET) was normal in one and revealed mild hypermetabolism at the three sites of maximum stenosis in one patient with multilevel stenosis. Initial diagnoses and treatments In forty patients (71%), an etiology other than SM and sometimes multiple alternative diagnoses were proposed including: spinal cord tumor, 18; transverse myelitis, 11; multiple sclerosis (MS), 9; sarcoidosis, 9; seronegative neuromyelitis optica (NMO), 4; inflammatory/autoimmune myelopathy not otherwise specified, 2; vascular malformation, 1. Thirty patients (54%) received one or more immunotherapies for suspected inflammatory/autoimmune myelopathies including: corticosteroids, 28;
John Wiley & Sons
Annals of Neurology
interferon-β1a, 3; azathioprine, 3; plasmapheresis, 2; intravenous immune-globulin, 1; mycophenolate, 1; glatiramer acetate, 1; and mitoxantrone, 1. Thirteen of 28 patients (46%) treated with corticosteroids reported transient clinical improvement (possibly related to decreased edema) but clinical and radiological progression often followed prior to surgical treatment. One patient who received immunosuppression developed septic shock. Surgical Treatments The median time from symptom onset to surgery (at Mayo Clinic, 70%; elsewhere, 30%) was 11 months (range, 1-64). Spinal surgery details are outlined in Table 1 and 2. Thirtynine (70%) patients reported some improvement after surgery and 14 reported stability while 3 continued to deteriorate despite surgery. Eleven patients reported delayed worsening after surgery (20%). Seven underwent further decompressive surgery at the same site or an adjacent level a median of 9 months (range, 1-26) after the initial surgery and all reported benefit. Pathology findings Spinal cord biopsy in six patients revealed: non-specific perivascular inflammation, 1 (Figure 4); non-specific gliosis, 1; non-specific inflammatory changes not otherwise specified, 1; mild parenchymal and perivascular histiocytic infiltrate with evidence of perivascular chronic inflammation and gliosis, 1; minute fragments of cord parenchyma with mild hypercellularity and atypia, occasional Rosenthal fibers and mild gliosis without features of demyelination, 1; and reactive lymphoplasmacytic infiltrate with preservation of myelin and axons, 1. Potentially confounding illnesses
John Wiley & Sons
Page 12 of 35
Page 13 of 35
Annals of Neurology
One patient was subsequently diagnosed with relapsing remitting multiple sclerosis, but persistent enhancement for 37 months and symptom resolution after decompressive surgery suggested SM as the cause of their initial progressive myelopathy. Three patients had pathological evidence of systemic sarcoidosis (lung biopsy, 2; conjunctival biopsy, 1). Marked clinical and radiological improvement after surgery without immunosuppressant treatments excluded neurosarcoidosis as the cause of the myelopathy in one of the three. In the other two patients sarcoidosis may have contributed to the myelopathy but spondylosis was felt the more likely due to the pancake-like enhancement and improvement after surgery. The CSF findings in these two patients revealed a normal white cell count in one and a mild pleocytosis (12/µL [normal, ≤5/ µL]) but markedly elevated total protein (178 mg/dL [normal ≤45 mg/dL]) in the other. MRI follow-up (Table 1) In 51patients with (>1 MRI) the median duration of MRI follow up was 23 months (range, 1-131). In two patients T2-signal abnormalities worsened immediately after surgery, but subsequently regressed. The duration of gadolinium enhancement observed represents the minimum duration and was reported as being at least the duration at which it was last observed. Persistent, although often decreasing, enhancement after surgery was present at least ≥3 months in 34 of 34 (100%) and at least ≥ 12 months in 12 of 16 (75%) with close MRI follow up (Table 1). The median duration of enhancement after surgery was 7.5 months (range, 1-101 months), but enhancement was present in 35 (62%) at last follow up. In twenty-six patients (46%) the treating clinician documented uncertainty about the SM diagnosis due to persistent enhancement after surgery, frequently resulting in treatments for alternative etiologies despite clinical stability. No patient developed
John Wiley & Sons
Annals of Neurology
progressive spinal cord enlargement over time on MRI and resolution of swelling sometimes accompanied by atrophy was noted in those with long-term MRI follow up. During the period of resolution continuous peripheral circumferential enhancement typically became incomplete or multifocal but remained confined to white matter (Figure 2[A5 and A7]). Myelomalacia demonstrated by residual T2-signal change was evident in all patients at last follow up. Clinical Follow-up The median duration from symptom onset to last follow up was 60 months (range, 10172). Seven of 15 patients (47%) improved ambulatory status after surgery (e.g. walker to cane). At last follow up 36% needed a gait aid (Table 1). The only statistically significant predictor of the need for gait aid at last follow up was the need for a gait aid at our initial evaluation (p=0.005: Table 2). All patients in this series had surgery precluding a rigorous analysis of the benefits of surgery. To investigate the role of surgery, we analyzed a subgroup of 28 patients in whom surgery was delayed by one year or greater (median delay 21 months; range, 12-66]); progressive worsening was noted over the observation period prior to surgery in all patients. A small peripheral focus of gadolinium enhancement noted in one patient resolved prior to surgery despite continued clinical worsening accompanied by increasing T2-signal abnormality and worsening cord compression. Sixteen of the 28(57%) patients improved after surgery and 10 stabilized (36%). Two of these patients deteriorated despite surgery, one of whom was judged to have inadequate decompression based on imaging and improved after a second surgery. Potential radiological mimics
John Wiley & Sons
Page 14 of 35
Page 15 of 35
Annals of Neurology
None of the 136 patients with established alternative diagnoses (Figure 5) had pancakelike enhancement. Diagnoses included: pathologically confirmed intramedullary spinal cord tumors 44 (ependymomas, 16; primary intramedullary spinal cord lymphoma, 10; astrocytomas, 8; hemangioblastomas, 6; cord primitive neuroectodermal tumor, 1; dermoid, 1; ganglioglioma, 1; and subependymoma, 1); neurosarcoid 28; neuromyelitis optica 28; transverse myelitis 27 (12 confirmed on follow-up to have relapsing remitting MS) and vascular lesions 9 (angiographically confirmed dural-arteriovenous fistula, 6; cavernous angiomas, 3).
Discussion This case series demonstrates that prominent pancake-like gadolinium enhancement on sagittal images of the spinal cord immediately below a stenosis and circumferential enhancement on axial images sparing gray matter in the setting of progressive myelopathy associated with intramedullary T2-weighted lesions is strongly indicative of SM. Patients with this radiological finding are frequently diagnosed with, or treated for, alternative etiologies of myelopathy other than SM. Potentially beneficial surgical decompression is delayed or deferred. Persistent enhancement despite clinical stability months-to-years following successful surgical decompression is common and should not lead to unnecessary investigations and treatments. The characteristic pancake-like enhancement pattern described in this study generally follows certain rules that may be helpful for the practicing clinician when encountering a suspected case: 1) transverse band appearance, greater in transverse than vertical extent on sagittal images; 2) location just below the site of maximum stenosis at
John Wiley & Sons
Annals of Neurology
the center of a spindle shaped T2 hyperintensity; and 3) circumferential enhancement sparing gray matter on axial images. The enhancement varied slightly, sometimes appearing “signet ring” like with longer dorsal than ventral enhancement (Figure 1C2 and Figure 2A2) or appearing discontinuous sparing gray matter on the sagittal images (Figure 1[C2 and F2]; Figure 2A2). Nonetheless the three characteristics above were generally respected. The accompanying spinal stenosis may be moderate rather than severe (Figure 1[B, C, D and F] and Figure 2). The associated spinal cord swelling and enhancement led to suspicion of a primary intramedullary lesion such as tumor or inflammation in 71% in this study. The enhancement pattern may be a key diagnostic clue to SM. Enhancement was most often at C5 or C6, the commonest level of involvement in SM2 and extended over one vertebral segment or less in rostrocaudal extent. A quarter of patients in this study with SM and enhancement did not have this characteristic pattern. Technical issues may have contributed to a less than typical appearance in these cases as suggested by relatively weak intensity of enhancement. In neoplastic or inflammatory lesions of the cord, gadolinium enhancement is longest in rostrocaudal dimension. Sagittal and axial images from prior publications of SM mimicking tumors or inflammation show a similar pattern7, 10 although one patient had a cylindrical pattern.11 A prospective study in which all patients with cervical SM undergoing MRI received gadolinium reported the prevalence of enhancement to be 7%.4 The retrospective design and methods of patient identification precluded an assessment of the frequency of gadolinium enhancement in SM in this study. Many patients appropriately undergo MRI without gadolinium when evaluating spondylosis. However, many patients receive gadolinium only when after a prominent T2-signal abnormality is
John Wiley & Sons
Page 16 of 35
Page 17 of 35
Annals of Neurology
detected in the setting of moderate spondylosis (as was the case for many patients in this study) to address the differential diagnosis of intramedullary lesions. The radiological finding we report may be helpful in this context. Persistent enhancement for one year or longer after decompressive surgery is common (75% in this study and 40% in a prior report4) and should be considered compatible with SM in the absence of clinical deterioration. In our study persistent enhancement often led physicians to question SM diagnosis and treat with immunotherapies for suspected inflammation or consider spinal cord biopsy, particularly when stenosis was no longer present (e.g. Figure 2A6). However, the enhancement invariably decreased after surgery, albeit slowly, as previously described.7, 10 Late clinical worsening or increasing enhancement in the months-years after surgery should raise concern for an alternative diagnosis or incomplete decompression; seven patients in our study required an additional decompressive surgery. The pathophysiology of gadolinium enhancement in SM is poorly understood but probably relates to focal blood-spinal-cord-barrier breakdown.4 Formation of new blood vessels in response to injury may cause a leaky blood-spinal-cord-barrier4 although increased vascular permeability from venous hypertension has also been implicated.10 Subarachnoid scarring may alter CSF/extracellular fluid flow dynamics and contribute to edema.7 Similarly, cauda equina enhancement in lumbar spondylosis may relate to bloodnerve-barrier damage.12 The pathological features of perivascular lymphocytic infiltration (Figure 4) suggest inflammatory cells had access to the spinal cord through the damaged blood-spinal-cord-barrier and this is similar to a previously reported biopsied case11 and may explain the mild CSF pleocytosis seen in a few patients in our study. Persistent
John Wiley & Sons
Annals of Neurology
enhancement for months-to-years suggests that healing of blood-spinal-cord-barrier is a lengthy process. MRI worsening during extension has been reported in SM mimicking tumors7 and transient repetitive compression may explain why the signal change and enhancement often appeared out of proportion to the severity of stenosis (Figure 1B and Figure 2). Flexion/extension MRIs (not performed routinely in our study) may be helpful in such cases. As the spinal cord may move considerably during flexion and extension13 prevention of such movement and the related repeated transient compression may account for perceived benefit from a cervical collar by some patients. The clinical and radiological features of SM in our study facilitate the differential diagnosis of myelopathy (Figure 6). In contrast to progressive SM, the clinical nadir in transverse myelitis cases occurs within 21 days. Persistent enhancement beyond 2 months occurs in ≤4%14 of patients with MS and oligoclonal bands, which were not seen in our study, are detected in >85%. Although a predilection for MS lesions to occur at sites of trauma15 including adjacent to spondylotic ridges (Figure 5F) has been reported, the claims remain controversial.16 T2-signal hyperintensities in NMO are often accompanied by long patchy enhancement (Figure 5D).17 Markedly elevated CSF protein, likely due to spinal block, without other features of inflammation was typical in our study and argued against primary inflammatory etiologies. Spinal cord sarcoidosis lesions demonstrate persistent enhancement but normal CSF white cell count, absence of pial enhancement (Figure 5E) and normal chest CT rule out this diagnosis in most patients in our series. One of six previously described patients with cord edema and enhancement suspicious for cervical spondylosis worsened after decompression and subsequent cord biopsy confirmed neurosarcoidosis, highlighting the potential for misdiagnosis and importance
John Wiley & Sons
Page 18 of 35
Page 19 of 35
Annals of Neurology
of clinical and radiological follow-up, particularly of patients with less typical imaging findings.18 Lack of conus involvement and absence of venous flow voids made dural arterio-venous fistula unlikely.19 While non-diagnostic spinal cord biopsies may occur20 6 negative spinal cord biopsies make it unlikely that patients in our series had tumors. Furthermore, we did not detect cysts or syrinxes that occur in 60% of spinal cord tumors nor did we observe progressive enlargement of the spinal cord on median follow up of 60 months.21 FDG-PET spinal cord hypermetabolism is more suggestive of a neoplastic than of inflammatory myelopathy.22 Focal FDG-PET may occur at the site(s) of maximal spinal stenosis,23 and was noted in one patient in our study. The retrospective nature of the analysis and highly selected criteria prevent us from concluding that the radiologic finding we report is pathognomonic of SM. Caution is advised particularly in less typical cases (Figure 3F) or in those with abnormalities that raise the possibility of alternative diagnoses (CSF pleocytosis, abnormal MRI head or hilar adenopathy). Our study is also limited by the lack of SM patients without enhancement to compare outcomes and determine its prognostic significance, but prior studies have found enhancement to be a poor prognostic factor.4 The small numbers (n=2) of patients with pancake-like enhancement who did not undergo surgery, our search strategy that identified mostly patients who underwent surgery, and our general approach of advising surgery in this situation, taken together, prevent us from reaching firm conclusions about benefits of surgery compared to nonsurgical management. Recommendations for surgical decompression should be based on a variety of factors: progressive symptoms and signs, lack of other identifiable cause after thorough evaluation, assessment of severity of cord compression based on axial T2-weighted
John Wiley & Sons
Annals of Neurology
images, risk:benefit assessment for an individual patient. The presence of pancake-like gadolinium enhancement alone should not be the basis of a surgical recommendation. Most patients in our series had progressive neurological worsening prior to surgery and moderate to severe myelopathy. In this situation, surgical decompression is often recommended, despite lack of support from randomized trials.2 While the majority improved or stabilized after surgery in this retrospective study, our search strategy identified mainly patients who underwent surgery and therefore may have been biased towards patients with continued deterioration for whom surgery is more likely to be recommended than for those with stable myelopathy. Prospective studies addressing patients who do and do not undergo surgery are warranted in SM with gadolinium enhancement to assess the benefit of surgery. Despite these limitations, this hallmark radiological finding is clearly under-recognized and greater appreciation of this characteristic radiological feature of SM may result in earlier potentially beneficial decompressive surgery, reduce inappropriate interventions and possibly improve outcomes for patients.
John Wiley & Sons
Page 20 of 35
Page 21 of 35
Annals of Neurology
References 1.
Moore AP, Blumhardt LD. A prospective survey of the causes of non-traumatic
spastic paraparesis and tetraparesis in 585 patients. Spinal Cord. 1997 Jun;35(6):361-7. 2.
Toledano M, Bartleson JD. Cervical spondylotic myelopathy. Neurol Clin. 2013
Feb;31(1):287-305. 3.
Takahashi M, Yamashita Y, Sakamoto Y, Kojima R. Chronic cervical cord
compression: clinical significance of increased signal intensity on MR images. Radiology. 1989 Oct;173(1):219-24. 4.
Ozawa H, Sato T, Hyodo H, et al. Clinical significance of intramedullary Gd-
DTPA enhancement in cervical myelopathy. Spinal Cord. 2010 May;48(5):415-22. 5.
Flanagan EP, Marsh RW, Weinshenker BG. Teaching neuroimages: "pancake-
like" gadolinium enhancement suggests compressive myelopathy due to spondylosis. Neurology. 2013 May 21;80(21):e229. 6.
Kelley BJ, Erickson BJ, Weinshenker BG. Compressive myelopathy mimicking
transverse myelitis. Neurologist. 2010 Mar;16(2):120-2. 7.
Sasamori T, Hida K, Yano S, Takeshi A, Iwasaki Y. Spinal cord swelling with
abnormal gadolinium-enhancement mimicking intramedullary tumors in cervical spondylosis patients: Three case reports and review of the literature. Asian J Neurosurg. 2010 Jul;5(2):1-9. 8.
McCormick WE, Steinmetz MP, Benzel EC. Cervical spondylotic myelopathy:
make the difficult diagnosis, then refer for surgery. Cleve Clin J Med. 2003 Oct;70(10):899-904.
John Wiley & Sons
Annals of Neurology
9.
Flanagan EP, O'Neill BP, Porter AB, Lanzino G, Haberman TM, Keegan BM.
Primary intramedullary spinal cord lymphoma. Neurology. 2011 Aug 23;77(8):784-91. 10.
Lee J, Koyanagi I, Hida K, Seki T, Iwasaki Y, Mitsumori K. Spinal cord edema:
unusual magnetic resonance imaging findings in cervical spondylosis. J Neurosurg. 2003 Jul;99(1 Suppl):8-13. 11.
Cabraja M, Abbushi A, Costa-Blechschmidt C, et al. Atypical cervical
spondylotic myelopathy mimicking intramedullary tumor. Spine (Phila Pa 1976). 2008 Mar 15;33(6):E183-7. 12.
Jinkins JR. Gd-DTPA enhanced MR of the lumbar spinal canal in patients with
claudication. J Comput Assist Tomogr. 1993 Jul-Aug;17(4):555-62. 13.
Adams CB, Logue V. Studies in cervical spondylotic myelopathy. I. Movement of
the cervical roots, dura and cord, and their relation to the course of the extrathecal roots. Brain. 1971;94(3):557-68. 14.
Cotton F, Weiner HL, Jolesz FA, Guttmann CR. MRI contrast uptake in new
lesions in relapsing-remitting MS followed at weekly intervals. Neurology. 2003 Feb 25;60(4):640-6. 15.
Poser CM. Trauma to the central nervous system may result in formation or
enlargement of multiple sclerosis plaques. Arch Neurol. 2000 Jul;57(7):1074-7, discussion 8. 16.
Ronthal M. On the coincidence of cervical spondylosis and multiple sclerosis.
Clin Neurol Neurosurg. 2006 Mar;108(3):275-7.
John Wiley & Sons
Page 22 of 35
Page 23 of 35
Annals of Neurology
17.
Wingerchuk DM, Weinshenker BG. Acute disseminated encephalomyelitis,
transverse myelitis, and neuromyelitis optica. Continuum (Minneap Minn). 2013 Aug;19(4 Multiple Sclerosis):944-67. 18.
Nurboja B, Chaudhuri A, David KM, Casey AT, Choi D. Swelling and
enhancement of the cervical spinal cord: when is a tumour not a tumour? Br J Neurosurg. 2012 Aug;26(4):450-5. 19.
Rubin MN, Rabinstein AA. Vascular diseases of the spinal cord. Neurol Clin.
2013 Feb;31(1):153-81. 20.
Cohen-Gadol AA, Zikel OM, Miller GM, Aksamit AJ, Scheithauer BW, Krauss
WE. Spinal cord biopsy: a review of 38 cases. Neurosurgery. 2003 Apr;52(4):806-15; discussion 15-6. 21.
Koeller KK, Rosenblum RS, Morrison AL. Neoplasms of the spinal cord and
filum terminale: radiologic-pathologic correlation. Radiographics. 2000 NovDec;20(6):1721-49. 22.
Flanagan EP, Hunt CH, Lowe V, et al. [(18)F]-fluorodeoxyglucose-positron
emission tomography in patients with active myelopathy. Mayo Clin Proc. 2013 Nov;88(11):1204-12. 23.
Floeth FW, Stoffels G, Herdmann J, et al. Prognostic value of 18F-FDG PET in
monosegmental stenosis and myelopathy of the cervical spinal cord. J Nucl Med. 2011 Sep;52(9):1385-91.
John Wiley & Sons
Annals of Neurology
Page 24 of 35
Table 1. Clinical, laboratory, neuroimaging, surgical and pathological characteristics of spondylotic myelopathy with gadolinium enhancement (n=56) Number of patients
Median (range)a
(%) Demographics Gender -
Male
-
Age at onset in years
39 (70%) 53.5 (24-80)
CSF findings (n=34) Elevated white cell count (>5/µl)
4 of 32 (12.5%)
Elevated protein (>45 mg/dL)
23 of 33 (70%)
Oligoclonal bands (>3) or IgG index (>0.85)
13.5 (10-19/µl) 82.5 (49-178 mg/dL)
0 of 32 (0%)
MRI spine T2 signal hyperintensity length in vertebral segments
56 (100%)
Spinal cord enlargementb
44 (79%)
Sagittal enhancement thickness rostro-caudally in mmc
2 (1-8)
7.3 (2.6-15)
Flat,transverse band pancake-like enhancement
41 (73%) d
Axial circumferential pattern sparing gray matter
29 of 51 (57%)
Persistent enhancement at 3 months
34 of 34 (100%)
Persistent enhancement at 12 months
12 of 16 (75%)
Surgical treatments Decompressive laminectomies without fusion
21 (37.5%)
Decompression with fusion
21 (37.5%)
Spinal cord biopsy with associated decompressione
6 (11%)
Otherf
8 (14%)
Pathology of cord biopsy: non specific inflammation
6 of 6 (100%)
Ambulatory status at last follow up No gait aid
36 (64%)
Cane
6 (11%)
Walker
9 (16%)
Wheelchair
5 (9%)
Abbreviations: CSF, cerebrospinal fluid; mm, millimeters; MRI, magnetic resonance imaging.
John Wiley & Sons
Page 25 of 35
Annals of Neurology
a
for CSF results the median and range are of the abnormal values
b
Median width of cervical spinal cord axially 0.5 vertebral segemnts below the site of maximal stenosis was
15.2 mm (range, 11.8-20.1). For thoracic spine median width of spinal cord axially 0.5 vertebral segments below the site of maximal stenosis was 11.95 mm (range, 11.6-12.1) and the median width of T2-signal change in the thoracic cord was 8.8 mm (range, 6.6-8.8). c
maximum length of enhancement rostrocaudally used for measurements
d
76% of those with a circumferential pattern sparing gray matter were symmetric
e
f
No complications from the cord biopsies were noted
Corpectomy with fusion, 3; diskectomy, 3; laminoplasty, 1. Decompression not otherwise specified, 1.
\
John Wiley & Sons
Annals of Neurology
Page 26 of 35
Table 2. Statistical comparison of predictors of the need for a gait aid at last followup in cervical spondylosis associated with gadolinium enhancement No Gait aid at
Gait aid at last
last follow up
follow-up
(n=36)
(n=20)
p-value
Demographics/clinical details -
Male sex
28 (78%)
11 (55%)
0.08
-
Median (range) age at onset, years
51 (32-72)
58.5 (24-80)
0.10
-
Median (range) time to surgery,
10 (1.6-66)
13 (0.1-60.6)
0.61
5 (14%)
10 (50%)
0.005
2 (1-8)
2 (1-8)
0.78
6 (17%)
6 (30%)
0.25
months -
Need for gait aid prior to surgery
MRI Features -
Median (range) length of T2 signal (vertebral segments)
-
Enhancement >12 months
CSF Results -
Elevated white cell count (>5/µL)
3 of 23 (13%)
1 of 9 (11%)
1.0
-
Elevated CSF protein (normal, ≤45
17 of 24 (71%)
6 of 9 (67%)
1.0
mg/dL) Surgery performed -
Anterior decompression
12 of 28a (43%)
2 of 15a (13%)
0.09
-
Posterior decompression
16 of 28a (57%)
13 of 15a (87%)
0.09
-
Spinal fusion
12 (33%)
9 (45%)
0.39
-
Laminectomy
13 (36%)
8 (40%)
0.77
-
Spinal cord biopsy
5 (14%)
1 (5%)
0.40
Abbreviations: CSF, cerebrospinal fluid; MRI, magnetic resonance imaging; SD, standard deviation. a
those in whom details of the surgical approach was available
John Wiley & Sons
Page 27 of 35
Annals of Neurology
Figure Legends: Figure 1: Neuroimaging of spondylosis with gadolinium enhancement Sagittal MRI reveals a three vertebral segment spindle shaped T2-signal hyperintensity in the spinal cord (A1) with transverse gadolinium enhancement (A2) just below the site of maximum stenosis at the C5/C6 interspace. Sagittal MRI reveals three segment spindle shaped T2-signal hyperintensity with heterogenous T2-signal (B1, white arrows) at the site of flat pancake-like gadolinium enhancement (B2). Postoperative sagittal MRI of a separate patient demonstrates decompression (C1, arrowhead) and a two segment spindle shaped T2 hyperintensity with signal heterogeneity (C1, white arrow) at the site of flat pancake-like enhancement that is thicker dorsally than ventrally (C2, white arrow). A longitudinally-extensive T2-weighted signal abnormality with moderate cervical spondylosis is most prominent at C4-5 on sagittal T2-weighted images (D1), below which a flat pancake-like band of gadolinium enhancement is evident (D2). A spindle shaped T2-hyperintensity is seen on sagittal MRI (E1), at the middle of which focal pancake-like enhancement is present (E2). A longitudinally extensive T2-hyperintensity is present on sagittal T2-weighted MRI (F1) with pancake-like enhancement at the site of maximal stenosis; central sparing of gray matter is suggested (F2), which is better appreciated on axial images (not shown). Figure 2: MRI evolution after cervical spondylosis surgery Sagittal MRI demonstrates T2-signal hyperintensity and cord swelling which appears out of proportion to degree of canal stenosis (A1) with transverse gadolinium enhancement (A2) of the white matter tracts and sparing of gray matter on axial T1-weighted images
John Wiley & Sons
Annals of Neurology
(A3). Subsequent gadolinium-enhanced sagittal and axial T1-weighted images show gradual asymmetric resolution of enhancement over time (A4-A9). Figure 3: Axial gadolinium enhancement patterns in cervical spondylosis A typical pattern of peripheral enhancement on T1-weighted axial images was noted in 57% of patients (A-C). The pattern is circumferential (illustrated by white arrows) and spares gray matter in most (A-C). The peripheral enhancement pattern may be incomplete or partial (D) or involve only the hemi-cord (E). In 24% of patients, a single focus of enhancement was seen (F) which is not specific for spondylotic myelopathy but may represent a fragment of the general pattern of peripheral enhancement. Figure 4: Pathology of spinal cord biopsy in a case of SM with gadolinium enhancement Biopsy demonstrating gliotic white matter (A) with focal perivascular inflammation (B), making one suspect an inflammatory process. The axons, possibly due to prior freezing, are brightly eosinophilic nearly mimicking Rosenthal fibers (A and B). The scale bar represents 30 micron (µm). Figure 5: Sagittal MRIs of potential mimics of SM with enhancement and coexisting spondylosis Many causes of myelopathy can have coexisting spondylosis (arrowheads) but the enhancement pattern differs (white arrows) when the myelopathy etiology is not secondary to spondylosis. Examples of alternative myelopathy etiologies with co-existing spondylosis shown here include: ependymoma with cylindrical enhancement (A); multiple hemangioblastomas with circular enhancing lesions (B); astrocytoma with multifocal rostro-caudal enhancement (C); seropositive neuromyelitis optica with patchy
John Wiley & Sons
Page 28 of 35
Page 29 of 35
Annals of Neurology
rostro-caudal enhancement (E); spinal cord sarcoidosis with pial enhancement (E); and multiple sclerosis with focal dorsal cord enhancement (F). Figure 6: Spondylosis with Enhancement -- Diagnostic Algorithm Abbreviations: AQP4-IgG, aquaporin-4-IgG; AV, arteriovenous; CSF, cerebrospinal fluid; CT, computed tomography; MRA, magnetic resonance angiography; MRI, magnetic resonance imaging; MS, multiple sclerosis; NMO, neuromyelitis optica.
John Wiley & Sons
Annals of Neurology
Figure 1: Neuroimaging of spondylosis with gadolinium enhancement Sagittal MRI reveals a three vertebral segment spindle shaped T2-signal hyperintensity in the spinal cord (A1) with transverse gadolinium enhancement (A2) just below the site of maximum stenosis at the C5/C6 interspace. Sagittal MRI reveals three segment spindle shaped T2-signal hyperintensity with heterogenous T2-signal (B1, white arrows) at the site of flat pancake-like gadolinium enhancement (B2). Postoperative sagittal MRI of a separate patient demonstrates decompression (C1, arrowhead) and a two segment spindle shaped T2 hyperintensity with signal heterogeneity (C1, white arrow) at the site of flat pancake-like enhancement that is thicker dorsally than ventrally (C2, white arrow). A longitudinally-extensive T2weighted signal abnormality with moderate cervical spondylosis is most prominent at C4-5 on sagittal T2weighted images (D1), below which a flat pancake-like band of gadolinium enhancement is evident (D2). A spindle shaped T2-hyperintensity is seen on sagittal MRI (E1), at the middle of which focal pancake-like enhancement is present (E2). A longitudinally extensive T2-hyperintensity is present on sagittal T2-weighted MRI (F1) with pancake-like enhancement at the site of maximal stenosis; central sparing of gray matter is suggested (F2), which is better appreciated on axial images (not shown). 254x190mm (96 x 96 DPI)
John Wiley & Sons
Page 30 of 35
Page 31 of 35
Annals of Neurology
Figure 2: MRI evolution after cervical spondylosis surgery Sagittal MRI demonstrates T2-signal hyperintensity and cord swelling which appears out of proportion to degree of canal stenosis (A1) with transverse gadolinium enhancement (A2) of the white matter tracts and sparing of gray matter on axial T1-weighted images (A3). Subsequent gadolinium-enhanced sagittal and axial T1-weighted images show gradual asymmetric resolution of enhancement over time (A4-A9). 254x190mm (96 x 96 DPI)
John Wiley & Sons
Annals of Neurology
Figure 3: Axial gadolinium enhancement patterns in cervical spondylosis A typical pattern of peripheral enhancement on T1-weighted axial images was noted in 57% of patients (AC). The pattern is circumferential (illustrated by white arrows) and spares gray matter in most (A-C). The peripheral enhancement pattern may be incomplete or partial (D) or involve only the hemi-cord (E). In 24% of patients, a single focus of enhancement was seen (F) which is not specific for spondylotic myelopathy but may represent a fragment of the general pattern of peripheral enhancement. 254x169mm (96 x 96 DPI)
John Wiley & Sons
Page 32 of 35
Page 33 of 35
Annals of Neurology
Figure 4: Pathology of spinal cord biopsy in a case of SM with gadolinium enhancement Biopsy demonstrating gliotic white matter (A) with focal perivascular inflammation (B), making one suspect an inflammatory process. The axons, possibly due to prior freezing, are brightly eosinophilic nearly mimicking Rosenthal fibers (A and B). The scale bar represents 30 micron (µm). 177x127mm (300 x 300 DPI)
John Wiley & Sons
Annals of Neurology
Figure 5: Sagittal MRIs of potential mimics of SM with enhancement and coexisting spondylosis Many causes of myelopathy can have coexisting spondylosis (arrowheads) but the enhancement pattern differs (white arrows) when the myelopathy etiology is not secondary to spondylosis. Examples of alternative myelopathy etiologies with co-existing spondylosis shown here include: ependymoma with cylindrical enhancement (A); multiple hemangioblastomas with circular enhancing lesions (B); astrocytoma with multifocal rostro-caudal enhancement (C); seropositive neuromyelitis optica with patchy rostro-caudal enhancement (E); spinal cord sarcoidosis with pial enhancement (E); and multiple sclerosis with focal dorsal cord enhancement (F). 254x190mm (96 x 96 DPI)
John Wiley & Sons
Page 34 of 35
Page 35 of 35
Annals of Neurology
Figure 6: Spondylosis with Enhancement -- Diagnostic Algorithm Abbreviations: AQP4-IgG, aquaporin-4-IgG; AV, arteriovenous; CSF, cerebrospinal fluid; CT, computed tomography; MRA, magnetic resonance angiography; MRI, magnetic resonance imaging; MS, multiple sclerosis; NMO, neuromyelitis optica. 211x178mm (96 x 96 DPI)
John Wiley & Sons
Running Head: Enhancement in spondylosis Department of Neurology,1 Radiology,2 Neurosurgery,3 and Anatomical Pathology,4 Mayo Clinic, Rochester, MN, USA
Correspondence to: Dr. Brian G. Weinshenker, Mayo Clinic, Department of Neurology, 200 First Street SW, Rochester, MN, 55905, USA. E-mail: [email protected]
Word Count: Title: characters with spaces = 69 (maximum 80) Running Title: characters with spaces = 26 (maximum 50) Abstract: 249 (max 250) Text = Introduction:
309 (Limit 500 words)
Methods/Results:
2996 (No limit)
Discussion:
1430 (Limit 1500 words)
References = 23 (maximum 40) Figures/Tables: 2 Tables; 6 figures Supplemental data: none
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process which may lead to differences between this version and the Version of Record. Please cite this article as an ‘Accepted Article’, doi: 10.1002/ana.24184
Annals of Neurology
Key words:
Cervical spondylosis; gadolinium enhancement; magnetic resonance
imaging; persistent; spinal cord tumor; transverse myelitis;
Financial Disclosure Statements Drs. Flanagan, Marsh, Gianinni and Krecke have no disclosures. Dr Weinshenker has received a research grant from the Guthy Jackson Foundation. He receives royalties from RSR for a technology license related to a test for aquaporin-4 autoantibodies for diagnosis of neuromyelitis optica. He serves on data safety monitoring committees for Novartis, Biogen-Idec and Mitsubishi pharmaceutical companies, and serves on an adjudication panel for Medimmune Pharmaceuticals. He served as a consultant for GlaxoSmithKline, Elan, Ono, Chugai and Alexion pharmaceutical companies and for Asahi Kasei Medical Company. He serves on editorial boards for Neurology, the Canadian Journal of Neurological Sciences, and Turkish Journal of Neurology. Keegan is compensated as a Chief Editor of eMedicine, has consulted for Bristol Meyers Squibb, Novartis and Bionest and receive research support from Terumo BCT. Conflict of Interest Statement The authors declare that they have no conflicts of interest. Author contributions Dr Flanagan was involved in abstracting the clinical information from the electronic medical record, drafting and revising the manuscript for content, including medical writing for content, analysis and interpretation of data, and acquisition of data. Dr Krecke was involved in revising the manuscript for content and analysis and interpretation of data. Dr Marsh was involved in revising the manuscript for content and analysis and interpretation of data.
John Wiley & Sons
Page 2 of 35
Page 3 of 35
Annals of Neurology
Dr Giannini was involved in revising the manuscript for content and analysis and interpretation of data. Dr Keegan was involved in revising the manuscript for content and analysis and interpretation of data. Dr Weinshenker was involved in drafting and revising the manuscript for content, including medical writing for content, analysis and interpretation of data, acquisition of data and study supervision.
John Wiley & Sons
Annals of Neurology
Abstract Objective: To highlight a specific under-recognized radiological feature of spondylotic myelopathy often resulting in misdiagnosis. Methods: Patients evaluated between 1/1/1996-12/31/2012 who met the following criteria were included: 1) suspected spondylotic myelopathy; 2) gadolinium enhancement detected; 3) spinal surgery performed. Results: Fifty six patients (70% men) whose median age was 53.5 years (range, 24-80) were included. Spinal cord MRI (cervical, 52; thoracic, 4) revealed longitudinal-spindleshaped-T2-signal hyperintensity (100%) and cord enlargement (79%) accompanied by a characteristic pancake-like transverse band of gadolinium enhancement in 41 (73%) typically immediately caudal to the site of maximal spinal stenosis. Forty (71%) patients were initially diagnosed with neoplastic or inflammatory myelopathies and decompressive surgery was delayed by a median of 11 months (range 1-64). Spinal cord biopsy in 6 did not reveal any alternative diagnosis. Ninety five percent were stable or improved. Gadolinium enhancement persisted in 75% at 12 months, raising concern about accuracy of initial diagnosis. Twenty patients required a gait aid (36%) at last follow up (median 60 months, range 10-172). The need for a gait aid pre-operatively (p=0.005), but not delay to surgery, predicted need for gait aid at final follow-up. Interpretation: Transverse pancake-like gadolinium enhancement associated with and just caudal to the site of maximal stenosis and at the rostrocaudal midpoint of a spindleshaped T2-hyperintensity suggests that spondylosis is the cause of the myelopathy. Persistent enhancement for months to years following decompressive surgery is common.
John Wiley & Sons
Page 4 of 35
Page 5 of 35
Annals of Neurology
Recognition is important to prevent inappropriate interventions or delay in consideration of a potentially beneficial decompressive surgery.
John Wiley & Sons
Annals of Neurology
Cervical spondylosis is the commonest cause of myelopathy accounting for 23.6% of all non-traumatic myelopathies in one study.1 It is essential to recognize the spectrum of radiological features of cervical spondylotic myelopathy (SM), including those that are atypical, given its frequency and the potential for successful treatment with surgical decompression, although class I evidence supporting the benefits of surgery is lacking. Magnetic resonance imaging (MRI) of cervical SM typically reveals degenerative disease of the spine, canal stenosis and cord compression.2 In cervical SM spinal cord T2-signal abnormalities occur in approximately 15%3 and gadolinium enhancement in 7.3%.4 We have recognized a distinctive pattern of long fusiform T2-signal abnormality with flat, transversely oriented pancake-like enhancement at the site of maximal compression in cervical SM.5, 6 This radiological feature of SM often leads to misdiagnosis as inflammatory6 or neoplastic conditions,7 delay of potentially effective surgical treatment and occasionally inappropriate biopsy. Prompt, accurate and confident diagnosis of SM is important for a number of reasons. Although a recommendation of surgical decompression is based on a thorough differential diagnosis of myelopathy, a rigorous assessment of the severity of stenosis and an assessment of risks and benefits, surgical decompression may improve neurological prognosis.8 Investigations to detect alternative etiologies are costly and often imperfectly sensitive (e.g. angiotensin converting enzyme to diagnose neurosarcoidosis). Hence, exclusion of alternative causes of myelopathy associated with gadolinium enhancement with complete certainty is impossible. Spinal cord biopsy carries significant risk of neurological deficits and may be non-diagnostic. Empiric treatments for suspected alternative etiologies (e.g. long-term immunosuppressant medications) may be harmful. We observed persisting gadolinium
John Wiley & Sons
Page 6 of 35
Page 7 of 35
Annals of Neurology
enhancement for months-years after surgery for SM, leading physicians to question the accuracy of the diagnosis and re-investigate for other causes despite the lack of clinical worsening. Herein we describe the clinical features and radiological characteristics including their evolution in 56 patients with SM associated with gadolinium enhancement.
Subjects and Methods Identifying Patients. Patients were identified retrospectively from 1/1/1996-12/31/2012. Inclusion criteria were: 1) suspected SM; 2) pre-operative MRI available; 3) gadolinium enhancement on MRI; 4) spinal surgery undertaken (primary decompression or secondary decompression during spinal cord biopsy). Patients were excluded if they did not have myelopathy, if they did not undergo MRI or if they did not have surgery. To identify patients we: 1) prospectively collected patients evaluated by one of the authors either at Mayo Clinic [n=12] or by electronic-consultation [n=1]; 2) searched the surgical records for additional patients who underwent cervical spine decompression surgery for cervical spondylosis from 2003-2012 [n=34]; and 3) reviewed patients who underwent conjunctival biopsy to assess for sarcoidosis, a commonly suspected alternative diagnosis in this context, in patients with a progressive myelopathy of unknown origin [n=9]. In total 56 patients were included. Two patients with pancake-like gadolinium enhancement meeting other criteria were excluded as surgery was not performed during the follow-up period. One patient presented with a subacute myelopathy with weakness, stiffness and sensory loss. At last clinical follow-up 7 months after onset the patient was stable but had continuing symptoms of myelopathy and the
John Wiley & Sons
Annals of Neurology
enhancing lesion persisted. A second patient presented with a chronically progressive myelopathy and had stable symptoms and persistent gadolinium enhancement 13 months after onset. Both patients had moderate spondylotic changes and the final diagnosis in both was myelopathy of uncertain etiology. Surgery was not recommended as the significance of this radiologic finding was not recognized and because the stenosis was moderate rather than severe. None of the 20 patients with cervical or thoracic radiculopathy identified by our search of the surgical records and excluded as they did not have SM had gadolinium enhancement within the spinal cord. To determine if the radiological pattern of gadolinium enhancement with SM that we describe in this study is seen with other etiologies of spinal cord pathology we reviewed the MRIs of 136 patients with alternative diagnoses that were encountered during our search of the surgical records (n=43), patients selected from our database of longitudinally extensive myelopathies (n=56), patients with transverse myelitis from Olmsted County (n=27) and patients from a series of primary intramedullary spinal cord lymphoma (n=10)9. Six patients in our study have been reported previously.5, 6 Follow up Information. We contacted patients who had not been evaluated within 6 months of the study ascertainment date by telephone. We recorded their current clinical symptomatology including ambulatory status, additional surgeries undertaken and alternative diagnoses made. We also requested that their most recent MRI of the spine (cervical and/or thoracic) be sent to us for review.
John Wiley & Sons
Page 8 of 35
Page 9 of 35
Annals of Neurology
Neuroimaging. MRI’s from each patient were reviewed by a licensed neuroradiologist (K.N.K). We recorded the length of T2-signal abnormality, the pattern of gadolinium enhancement (sagittal and axial) and the duration of gadolinium enhancement prior to and after surgery. A total of 257 MRI’s at different time points (median of 4 MRI’s per patient; range, 1-13) were reviewed. Pre-operative MRI’s were performed with gadolinium in 54 patients and without gadolinium in two patients in whom post-operative pancake-like enhancement was detected. Pathology. All pathology samples were evaluated at Mayo Clinic by an experienced licensed neuropathologist. The photomicrograph was acquired in JPEG format and built in photoshop. A slight adjustment in brightness was made for optimal quality. Statistical Analysis. Descriptive summary statistics were reported as median (range, minimum-maximum) for continuous variables and frequencies and percentages for categorical variables. Comparisons were performed using Wilcoxon rank sum test or Fisher’s exact tests as appropriate using JMP 8.0 software (SAS®). Standard protocol approvals, registrations, and patient consents All patients consented to the use of their medical record for research purposes.
Results Patient demographics and clinical features Fifty-six patients (92% Caucasian; 8% African-American) were included (Table 1). Onset was subacute (≤8 weeks) in 21 (37%) and insidious (>8weeks) in 35 (63%). The most common symptoms were: numbness, paresthesias and/or pain (localized or shooting), 51 (91%); weakness, 48 (86%); bowel/bladder disturbance, 29 (52%); classic or “reverse” (with neck extension) Lhermitte’s symptom, 9 (16%). Examination revealed:
John Wiley & Sons
Annals of Neurology
spastic quadriparesis, 14; spastic paraparesis, 11; Brown-Sequard syndrome, 6; upper extremity weakness alone, 6; hemiparesis, 4; sensory ataxia, 3; spastic gait without weakness, 3; spastic-ataxic gait, 2; and mild myelopathy without weakness, 7. Ambulatory status immediately pre-operatively was: no gait aid, 41 (73%); cane, 5 (9%); walker, 4 (7%); and wheelchair, 6 (11%). Spine MRI (Table 1) Spondylotic myelopathy affected cervical cord in 52 (93%) and thoracic cord in 4 (7%). All patients had T2-weighted signal abnormalities (45% extending ≥3 vertebral segments) typically spindle shaped on sagittal images (median length, 33 mm; range, 6140) and central on axial images (median width of cervical T2-lesions 11.2 mm, range 2.5-18.5). T2-signal heterogeneity at the gadolinium enhancement site was noted in 13 (23%, Figure 1[B1 and C1]). Thoracic epidural lipomatosis contributed to stenosis in one. No cystic changes or syrinxes were noted. Fifty five patients (98%) had a single enhancing lesion (all extending ≤1 segment) and one had 3 focal regions of enhancement at three sites of severe narrowing. Enhancement was typically located at the middle of the T2-signal hyperintensity on sagittal images. Gadolinium enhancement frequently had a characteristic transverse band or flat pancake-like appearance in the sagittal imaging plane (41 patients [73%]). The transverse band of enhancement was complete in 67%, while in 6% there was central sparing noted on sagittal images (Figure 1C). The band of enhancement was located just inferior to the site of maximum stenosis, most commonly at C5 or C6 (48%). In the other 15 patients (27%) a focal region of enhancement that did not form a transverse band was seen in a similar rostrocaudal position within the area of T2 signal abnormality (Figure
John Wiley & Sons
Page 10 of 35
Page 11 of 35
Annals of Neurology
1A2). The rostrocaudal length of the enhancement was slightly greater in the dorsal than in the ventral cord in 14 patients (25%, Figure 1C2, Figure 2A2). On axial postgadolinium images, the enhancement pattern was circumferential located in the white matter, sparing gray matter in 29 of 51 with available images (57%: Figures 2A3 and 3[A-D]), focal in 14 (27%; peripheral, 12: Figure 3F; central, 2); multifocal/patchy 7 (14%) and diffuse in 1 (2%). Overall, enhancement affected white matter and spared gray matter in 45 of 51 (88%). The circumferential pattern was sometimes incomplete (Figure 3D), and occasionally involved the hemicord (Figure 3E). Pial enhancement was never observed. Other Investigations Head MRI was essentially normal in 43 patients investigated. Cerebrospinal fluid (CSF) commonly showed elevated protein (Table 1). Chest computed tomography revealed hilar adenopathy in 2 of 33 evaluated (6% [see potentially confounding illnesses below]). Whole body 18-Flourodeoxyglucose positron emission tomography (FDG-PET) was normal in one and revealed mild hypermetabolism at the three sites of maximum stenosis in one patient with multilevel stenosis. Initial diagnoses and treatments In forty patients (71%), an etiology other than SM and sometimes multiple alternative diagnoses were proposed including: spinal cord tumor, 18; transverse myelitis, 11; multiple sclerosis (MS), 9; sarcoidosis, 9; seronegative neuromyelitis optica (NMO), 4; inflammatory/autoimmune myelopathy not otherwise specified, 2; vascular malformation, 1. Thirty patients (54%) received one or more immunotherapies for suspected inflammatory/autoimmune myelopathies including: corticosteroids, 28;
John Wiley & Sons
Annals of Neurology
interferon-β1a, 3; azathioprine, 3; plasmapheresis, 2; intravenous immune-globulin, 1; mycophenolate, 1; glatiramer acetate, 1; and mitoxantrone, 1. Thirteen of 28 patients (46%) treated with corticosteroids reported transient clinical improvement (possibly related to decreased edema) but clinical and radiological progression often followed prior to surgical treatment. One patient who received immunosuppression developed septic shock. Surgical Treatments The median time from symptom onset to surgery (at Mayo Clinic, 70%; elsewhere, 30%) was 11 months (range, 1-64). Spinal surgery details are outlined in Table 1 and 2. Thirtynine (70%) patients reported some improvement after surgery and 14 reported stability while 3 continued to deteriorate despite surgery. Eleven patients reported delayed worsening after surgery (20%). Seven underwent further decompressive surgery at the same site or an adjacent level a median of 9 months (range, 1-26) after the initial surgery and all reported benefit. Pathology findings Spinal cord biopsy in six patients revealed: non-specific perivascular inflammation, 1 (Figure 4); non-specific gliosis, 1; non-specific inflammatory changes not otherwise specified, 1; mild parenchymal and perivascular histiocytic infiltrate with evidence of perivascular chronic inflammation and gliosis, 1; minute fragments of cord parenchyma with mild hypercellularity and atypia, occasional Rosenthal fibers and mild gliosis without features of demyelination, 1; and reactive lymphoplasmacytic infiltrate with preservation of myelin and axons, 1. Potentially confounding illnesses
John Wiley & Sons
Page 12 of 35
Page 13 of 35
Annals of Neurology
One patient was subsequently diagnosed with relapsing remitting multiple sclerosis, but persistent enhancement for 37 months and symptom resolution after decompressive surgery suggested SM as the cause of their initial progressive myelopathy. Three patients had pathological evidence of systemic sarcoidosis (lung biopsy, 2; conjunctival biopsy, 1). Marked clinical and radiological improvement after surgery without immunosuppressant treatments excluded neurosarcoidosis as the cause of the myelopathy in one of the three. In the other two patients sarcoidosis may have contributed to the myelopathy but spondylosis was felt the more likely due to the pancake-like enhancement and improvement after surgery. The CSF findings in these two patients revealed a normal white cell count in one and a mild pleocytosis (12/µL [normal, ≤5/ µL]) but markedly elevated total protein (178 mg/dL [normal ≤45 mg/dL]) in the other. MRI follow-up (Table 1) In 51patients with (>1 MRI) the median duration of MRI follow up was 23 months (range, 1-131). In two patients T2-signal abnormalities worsened immediately after surgery, but subsequently regressed. The duration of gadolinium enhancement observed represents the minimum duration and was reported as being at least the duration at which it was last observed. Persistent, although often decreasing, enhancement after surgery was present at least ≥3 months in 34 of 34 (100%) and at least ≥ 12 months in 12 of 16 (75%) with close MRI follow up (Table 1). The median duration of enhancement after surgery was 7.5 months (range, 1-101 months), but enhancement was present in 35 (62%) at last follow up. In twenty-six patients (46%) the treating clinician documented uncertainty about the SM diagnosis due to persistent enhancement after surgery, frequently resulting in treatments for alternative etiologies despite clinical stability. No patient developed
John Wiley & Sons
Annals of Neurology
progressive spinal cord enlargement over time on MRI and resolution of swelling sometimes accompanied by atrophy was noted in those with long-term MRI follow up. During the period of resolution continuous peripheral circumferential enhancement typically became incomplete or multifocal but remained confined to white matter (Figure 2[A5 and A7]). Myelomalacia demonstrated by residual T2-signal change was evident in all patients at last follow up. Clinical Follow-up The median duration from symptom onset to last follow up was 60 months (range, 10172). Seven of 15 patients (47%) improved ambulatory status after surgery (e.g. walker to cane). At last follow up 36% needed a gait aid (Table 1). The only statistically significant predictor of the need for gait aid at last follow up was the need for a gait aid at our initial evaluation (p=0.005: Table 2). All patients in this series had surgery precluding a rigorous analysis of the benefits of surgery. To investigate the role of surgery, we analyzed a subgroup of 28 patients in whom surgery was delayed by one year or greater (median delay 21 months; range, 12-66]); progressive worsening was noted over the observation period prior to surgery in all patients. A small peripheral focus of gadolinium enhancement noted in one patient resolved prior to surgery despite continued clinical worsening accompanied by increasing T2-signal abnormality and worsening cord compression. Sixteen of the 28(57%) patients improved after surgery and 10 stabilized (36%). Two of these patients deteriorated despite surgery, one of whom was judged to have inadequate decompression based on imaging and improved after a second surgery. Potential radiological mimics
John Wiley & Sons
Page 14 of 35
Page 15 of 35
Annals of Neurology
None of the 136 patients with established alternative diagnoses (Figure 5) had pancakelike enhancement. Diagnoses included: pathologically confirmed intramedullary spinal cord tumors 44 (ependymomas, 16; primary intramedullary spinal cord lymphoma, 10; astrocytomas, 8; hemangioblastomas, 6; cord primitive neuroectodermal tumor, 1; dermoid, 1; ganglioglioma, 1; and subependymoma, 1); neurosarcoid 28; neuromyelitis optica 28; transverse myelitis 27 (12 confirmed on follow-up to have relapsing remitting MS) and vascular lesions 9 (angiographically confirmed dural-arteriovenous fistula, 6; cavernous angiomas, 3).
Discussion This case series demonstrates that prominent pancake-like gadolinium enhancement on sagittal images of the spinal cord immediately below a stenosis and circumferential enhancement on axial images sparing gray matter in the setting of progressive myelopathy associated with intramedullary T2-weighted lesions is strongly indicative of SM. Patients with this radiological finding are frequently diagnosed with, or treated for, alternative etiologies of myelopathy other than SM. Potentially beneficial surgical decompression is delayed or deferred. Persistent enhancement despite clinical stability months-to-years following successful surgical decompression is common and should not lead to unnecessary investigations and treatments. The characteristic pancake-like enhancement pattern described in this study generally follows certain rules that may be helpful for the practicing clinician when encountering a suspected case: 1) transverse band appearance, greater in transverse than vertical extent on sagittal images; 2) location just below the site of maximum stenosis at
John Wiley & Sons
Annals of Neurology
the center of a spindle shaped T2 hyperintensity; and 3) circumferential enhancement sparing gray matter on axial images. The enhancement varied slightly, sometimes appearing “signet ring” like with longer dorsal than ventral enhancement (Figure 1C2 and Figure 2A2) or appearing discontinuous sparing gray matter on the sagittal images (Figure 1[C2 and F2]; Figure 2A2). Nonetheless the three characteristics above were generally respected. The accompanying spinal stenosis may be moderate rather than severe (Figure 1[B, C, D and F] and Figure 2). The associated spinal cord swelling and enhancement led to suspicion of a primary intramedullary lesion such as tumor or inflammation in 71% in this study. The enhancement pattern may be a key diagnostic clue to SM. Enhancement was most often at C5 or C6, the commonest level of involvement in SM2 and extended over one vertebral segment or less in rostrocaudal extent. A quarter of patients in this study with SM and enhancement did not have this characteristic pattern. Technical issues may have contributed to a less than typical appearance in these cases as suggested by relatively weak intensity of enhancement. In neoplastic or inflammatory lesions of the cord, gadolinium enhancement is longest in rostrocaudal dimension. Sagittal and axial images from prior publications of SM mimicking tumors or inflammation show a similar pattern7, 10 although one patient had a cylindrical pattern.11 A prospective study in which all patients with cervical SM undergoing MRI received gadolinium reported the prevalence of enhancement to be 7%.4 The retrospective design and methods of patient identification precluded an assessment of the frequency of gadolinium enhancement in SM in this study. Many patients appropriately undergo MRI without gadolinium when evaluating spondylosis. However, many patients receive gadolinium only when after a prominent T2-signal abnormality is
John Wiley & Sons
Page 16 of 35
Page 17 of 35
Annals of Neurology
detected in the setting of moderate spondylosis (as was the case for many patients in this study) to address the differential diagnosis of intramedullary lesions. The radiological finding we report may be helpful in this context. Persistent enhancement for one year or longer after decompressive surgery is common (75% in this study and 40% in a prior report4) and should be considered compatible with SM in the absence of clinical deterioration. In our study persistent enhancement often led physicians to question SM diagnosis and treat with immunotherapies for suspected inflammation or consider spinal cord biopsy, particularly when stenosis was no longer present (e.g. Figure 2A6). However, the enhancement invariably decreased after surgery, albeit slowly, as previously described.7, 10 Late clinical worsening or increasing enhancement in the months-years after surgery should raise concern for an alternative diagnosis or incomplete decompression; seven patients in our study required an additional decompressive surgery. The pathophysiology of gadolinium enhancement in SM is poorly understood but probably relates to focal blood-spinal-cord-barrier breakdown.4 Formation of new blood vessels in response to injury may cause a leaky blood-spinal-cord-barrier4 although increased vascular permeability from venous hypertension has also been implicated.10 Subarachnoid scarring may alter CSF/extracellular fluid flow dynamics and contribute to edema.7 Similarly, cauda equina enhancement in lumbar spondylosis may relate to bloodnerve-barrier damage.12 The pathological features of perivascular lymphocytic infiltration (Figure 4) suggest inflammatory cells had access to the spinal cord through the damaged blood-spinal-cord-barrier and this is similar to a previously reported biopsied case11 and may explain the mild CSF pleocytosis seen in a few patients in our study. Persistent
John Wiley & Sons
Annals of Neurology
enhancement for months-to-years suggests that healing of blood-spinal-cord-barrier is a lengthy process. MRI worsening during extension has been reported in SM mimicking tumors7 and transient repetitive compression may explain why the signal change and enhancement often appeared out of proportion to the severity of stenosis (Figure 1B and Figure 2). Flexion/extension MRIs (not performed routinely in our study) may be helpful in such cases. As the spinal cord may move considerably during flexion and extension13 prevention of such movement and the related repeated transient compression may account for perceived benefit from a cervical collar by some patients. The clinical and radiological features of SM in our study facilitate the differential diagnosis of myelopathy (Figure 6). In contrast to progressive SM, the clinical nadir in transverse myelitis cases occurs within 21 days. Persistent enhancement beyond 2 months occurs in ≤4%14 of patients with MS and oligoclonal bands, which were not seen in our study, are detected in >85%. Although a predilection for MS lesions to occur at sites of trauma15 including adjacent to spondylotic ridges (Figure 5F) has been reported, the claims remain controversial.16 T2-signal hyperintensities in NMO are often accompanied by long patchy enhancement (Figure 5D).17 Markedly elevated CSF protein, likely due to spinal block, without other features of inflammation was typical in our study and argued against primary inflammatory etiologies. Spinal cord sarcoidosis lesions demonstrate persistent enhancement but normal CSF white cell count, absence of pial enhancement (Figure 5E) and normal chest CT rule out this diagnosis in most patients in our series. One of six previously described patients with cord edema and enhancement suspicious for cervical spondylosis worsened after decompression and subsequent cord biopsy confirmed neurosarcoidosis, highlighting the potential for misdiagnosis and importance
John Wiley & Sons
Page 18 of 35
Page 19 of 35
Annals of Neurology
of clinical and radiological follow-up, particularly of patients with less typical imaging findings.18 Lack of conus involvement and absence of venous flow voids made dural arterio-venous fistula unlikely.19 While non-diagnostic spinal cord biopsies may occur20 6 negative spinal cord biopsies make it unlikely that patients in our series had tumors. Furthermore, we did not detect cysts or syrinxes that occur in 60% of spinal cord tumors nor did we observe progressive enlargement of the spinal cord on median follow up of 60 months.21 FDG-PET spinal cord hypermetabolism is more suggestive of a neoplastic than of inflammatory myelopathy.22 Focal FDG-PET may occur at the site(s) of maximal spinal stenosis,23 and was noted in one patient in our study. The retrospective nature of the analysis and highly selected criteria prevent us from concluding that the radiologic finding we report is pathognomonic of SM. Caution is advised particularly in less typical cases (Figure 3F) or in those with abnormalities that raise the possibility of alternative diagnoses (CSF pleocytosis, abnormal MRI head or hilar adenopathy). Our study is also limited by the lack of SM patients without enhancement to compare outcomes and determine its prognostic significance, but prior studies have found enhancement to be a poor prognostic factor.4 The small numbers (n=2) of patients with pancake-like enhancement who did not undergo surgery, our search strategy that identified mostly patients who underwent surgery, and our general approach of advising surgery in this situation, taken together, prevent us from reaching firm conclusions about benefits of surgery compared to nonsurgical management. Recommendations for surgical decompression should be based on a variety of factors: progressive symptoms and signs, lack of other identifiable cause after thorough evaluation, assessment of severity of cord compression based on axial T2-weighted
John Wiley & Sons
Annals of Neurology
images, risk:benefit assessment for an individual patient. The presence of pancake-like gadolinium enhancement alone should not be the basis of a surgical recommendation. Most patients in our series had progressive neurological worsening prior to surgery and moderate to severe myelopathy. In this situation, surgical decompression is often recommended, despite lack of support from randomized trials.2 While the majority improved or stabilized after surgery in this retrospective study, our search strategy identified mainly patients who underwent surgery and therefore may have been biased towards patients with continued deterioration for whom surgery is more likely to be recommended than for those with stable myelopathy. Prospective studies addressing patients who do and do not undergo surgery are warranted in SM with gadolinium enhancement to assess the benefit of surgery. Despite these limitations, this hallmark radiological finding is clearly under-recognized and greater appreciation of this characteristic radiological feature of SM may result in earlier potentially beneficial decompressive surgery, reduce inappropriate interventions and possibly improve outcomes for patients.
John Wiley & Sons
Page 20 of 35
Page 21 of 35
Annals of Neurology
References 1.
Moore AP, Blumhardt LD. A prospective survey of the causes of non-traumatic
spastic paraparesis and tetraparesis in 585 patients. Spinal Cord. 1997 Jun;35(6):361-7. 2.
Toledano M, Bartleson JD. Cervical spondylotic myelopathy. Neurol Clin. 2013
Feb;31(1):287-305. 3.
Takahashi M, Yamashita Y, Sakamoto Y, Kojima R. Chronic cervical cord
compression: clinical significance of increased signal intensity on MR images. Radiology. 1989 Oct;173(1):219-24. 4.
Ozawa H, Sato T, Hyodo H, et al. Clinical significance of intramedullary Gd-
DTPA enhancement in cervical myelopathy. Spinal Cord. 2010 May;48(5):415-22. 5.
Flanagan EP, Marsh RW, Weinshenker BG. Teaching neuroimages: "pancake-
like" gadolinium enhancement suggests compressive myelopathy due to spondylosis. Neurology. 2013 May 21;80(21):e229. 6.
Kelley BJ, Erickson BJ, Weinshenker BG. Compressive myelopathy mimicking
transverse myelitis. Neurologist. 2010 Mar;16(2):120-2. 7.
Sasamori T, Hida K, Yano S, Takeshi A, Iwasaki Y. Spinal cord swelling with
abnormal gadolinium-enhancement mimicking intramedullary tumors in cervical spondylosis patients: Three case reports and review of the literature. Asian J Neurosurg. 2010 Jul;5(2):1-9. 8.
McCormick WE, Steinmetz MP, Benzel EC. Cervical spondylotic myelopathy:
make the difficult diagnosis, then refer for surgery. Cleve Clin J Med. 2003 Oct;70(10):899-904.
John Wiley & Sons
Annals of Neurology
9.
Flanagan EP, O'Neill BP, Porter AB, Lanzino G, Haberman TM, Keegan BM.
Primary intramedullary spinal cord lymphoma. Neurology. 2011 Aug 23;77(8):784-91. 10.
Lee J, Koyanagi I, Hida K, Seki T, Iwasaki Y, Mitsumori K. Spinal cord edema:
unusual magnetic resonance imaging findings in cervical spondylosis. J Neurosurg. 2003 Jul;99(1 Suppl):8-13. 11.
Cabraja M, Abbushi A, Costa-Blechschmidt C, et al. Atypical cervical
spondylotic myelopathy mimicking intramedullary tumor. Spine (Phila Pa 1976). 2008 Mar 15;33(6):E183-7. 12.
Jinkins JR. Gd-DTPA enhanced MR of the lumbar spinal canal in patients with
claudication. J Comput Assist Tomogr. 1993 Jul-Aug;17(4):555-62. 13.
Adams CB, Logue V. Studies in cervical spondylotic myelopathy. I. Movement of
the cervical roots, dura and cord, and their relation to the course of the extrathecal roots. Brain. 1971;94(3):557-68. 14.
Cotton F, Weiner HL, Jolesz FA, Guttmann CR. MRI contrast uptake in new
lesions in relapsing-remitting MS followed at weekly intervals. Neurology. 2003 Feb 25;60(4):640-6. 15.
Poser CM. Trauma to the central nervous system may result in formation or
enlargement of multiple sclerosis plaques. Arch Neurol. 2000 Jul;57(7):1074-7, discussion 8. 16.
Ronthal M. On the coincidence of cervical spondylosis and multiple sclerosis.
Clin Neurol Neurosurg. 2006 Mar;108(3):275-7.
John Wiley & Sons
Page 22 of 35
Page 23 of 35
Annals of Neurology
17.
Wingerchuk DM, Weinshenker BG. Acute disseminated encephalomyelitis,
transverse myelitis, and neuromyelitis optica. Continuum (Minneap Minn). 2013 Aug;19(4 Multiple Sclerosis):944-67. 18.
Nurboja B, Chaudhuri A, David KM, Casey AT, Choi D. Swelling and
enhancement of the cervical spinal cord: when is a tumour not a tumour? Br J Neurosurg. 2012 Aug;26(4):450-5. 19.
Rubin MN, Rabinstein AA. Vascular diseases of the spinal cord. Neurol Clin.
2013 Feb;31(1):153-81. 20.
Cohen-Gadol AA, Zikel OM, Miller GM, Aksamit AJ, Scheithauer BW, Krauss
WE. Spinal cord biopsy: a review of 38 cases. Neurosurgery. 2003 Apr;52(4):806-15; discussion 15-6. 21.
Koeller KK, Rosenblum RS, Morrison AL. Neoplasms of the spinal cord and
filum terminale: radiologic-pathologic correlation. Radiographics. 2000 NovDec;20(6):1721-49. 22.
Flanagan EP, Hunt CH, Lowe V, et al. [(18)F]-fluorodeoxyglucose-positron
emission tomography in patients with active myelopathy. Mayo Clin Proc. 2013 Nov;88(11):1204-12. 23.
Floeth FW, Stoffels G, Herdmann J, et al. Prognostic value of 18F-FDG PET in
monosegmental stenosis and myelopathy of the cervical spinal cord. J Nucl Med. 2011 Sep;52(9):1385-91.
John Wiley & Sons
Annals of Neurology
Page 24 of 35
Table 1. Clinical, laboratory, neuroimaging, surgical and pathological characteristics of spondylotic myelopathy with gadolinium enhancement (n=56) Number of patients
Median (range)a
(%) Demographics Gender -
Male
-
Age at onset in years
39 (70%) 53.5 (24-80)
CSF findings (n=34) Elevated white cell count (>5/µl)
4 of 32 (12.5%)
Elevated protein (>45 mg/dL)
23 of 33 (70%)
Oligoclonal bands (>3) or IgG index (>0.85)
13.5 (10-19/µl) 82.5 (49-178 mg/dL)
0 of 32 (0%)
MRI spine T2 signal hyperintensity length in vertebral segments
56 (100%)
Spinal cord enlargementb
44 (79%)
Sagittal enhancement thickness rostro-caudally in mmc
2 (1-8)
7.3 (2.6-15)
Flat,transverse band pancake-like enhancement
41 (73%) d
Axial circumferential pattern sparing gray matter
29 of 51 (57%)
Persistent enhancement at 3 months
34 of 34 (100%)
Persistent enhancement at 12 months
12 of 16 (75%)
Surgical treatments Decompressive laminectomies without fusion
21 (37.5%)
Decompression with fusion
21 (37.5%)
Spinal cord biopsy with associated decompressione
6 (11%)
Otherf
8 (14%)
Pathology of cord biopsy: non specific inflammation
6 of 6 (100%)
Ambulatory status at last follow up No gait aid
36 (64%)
Cane
6 (11%)
Walker
9 (16%)
Wheelchair
5 (9%)
Abbreviations: CSF, cerebrospinal fluid; mm, millimeters; MRI, magnetic resonance imaging.
John Wiley & Sons
Page 25 of 35
Annals of Neurology
a
for CSF results the median and range are of the abnormal values
b
Median width of cervical spinal cord axially 0.5 vertebral segemnts below the site of maximal stenosis was
15.2 mm (range, 11.8-20.1). For thoracic spine median width of spinal cord axially 0.5 vertebral segments below the site of maximal stenosis was 11.95 mm (range, 11.6-12.1) and the median width of T2-signal change in the thoracic cord was 8.8 mm (range, 6.6-8.8). c
maximum length of enhancement rostrocaudally used for measurements
d
76% of those with a circumferential pattern sparing gray matter were symmetric
e
f
No complications from the cord biopsies were noted
Corpectomy with fusion, 3; diskectomy, 3; laminoplasty, 1. Decompression not otherwise specified, 1.
\
John Wiley & Sons
Annals of Neurology
Page 26 of 35
Table 2. Statistical comparison of predictors of the need for a gait aid at last followup in cervical spondylosis associated with gadolinium enhancement No Gait aid at
Gait aid at last
last follow up
follow-up
(n=36)
(n=20)
p-value
Demographics/clinical details -
Male sex
28 (78%)
11 (55%)
0.08
-
Median (range) age at onset, years
51 (32-72)
58.5 (24-80)
0.10
-
Median (range) time to surgery,
10 (1.6-66)
13 (0.1-60.6)
0.61
5 (14%)
10 (50%)
0.005
2 (1-8)
2 (1-8)
0.78
6 (17%)
6 (30%)
0.25
months -
Need for gait aid prior to surgery
MRI Features -
Median (range) length of T2 signal (vertebral segments)
-
Enhancement >12 months
CSF Results -
Elevated white cell count (>5/µL)
3 of 23 (13%)
1 of 9 (11%)
1.0
-
Elevated CSF protein (normal, ≤45
17 of 24 (71%)
6 of 9 (67%)
1.0
mg/dL) Surgery performed -
Anterior decompression
12 of 28a (43%)
2 of 15a (13%)
0.09
-
Posterior decompression
16 of 28a (57%)
13 of 15a (87%)
0.09
-
Spinal fusion
12 (33%)
9 (45%)
0.39
-
Laminectomy
13 (36%)
8 (40%)
0.77
-
Spinal cord biopsy
5 (14%)
1 (5%)
0.40
Abbreviations: CSF, cerebrospinal fluid; MRI, magnetic resonance imaging; SD, standard deviation. a
those in whom details of the surgical approach was available
John Wiley & Sons
Page 27 of 35
Annals of Neurology
Figure Legends: Figure 1: Neuroimaging of spondylosis with gadolinium enhancement Sagittal MRI reveals a three vertebral segment spindle shaped T2-signal hyperintensity in the spinal cord (A1) with transverse gadolinium enhancement (A2) just below the site of maximum stenosis at the C5/C6 interspace. Sagittal MRI reveals three segment spindle shaped T2-signal hyperintensity with heterogenous T2-signal (B1, white arrows) at the site of flat pancake-like gadolinium enhancement (B2). Postoperative sagittal MRI of a separate patient demonstrates decompression (C1, arrowhead) and a two segment spindle shaped T2 hyperintensity with signal heterogeneity (C1, white arrow) at the site of flat pancake-like enhancement that is thicker dorsally than ventrally (C2, white arrow). A longitudinally-extensive T2-weighted signal abnormality with moderate cervical spondylosis is most prominent at C4-5 on sagittal T2-weighted images (D1), below which a flat pancake-like band of gadolinium enhancement is evident (D2). A spindle shaped T2-hyperintensity is seen on sagittal MRI (E1), at the middle of which focal pancake-like enhancement is present (E2). A longitudinally extensive T2-hyperintensity is present on sagittal T2-weighted MRI (F1) with pancake-like enhancement at the site of maximal stenosis; central sparing of gray matter is suggested (F2), which is better appreciated on axial images (not shown). Figure 2: MRI evolution after cervical spondylosis surgery Sagittal MRI demonstrates T2-signal hyperintensity and cord swelling which appears out of proportion to degree of canal stenosis (A1) with transverse gadolinium enhancement (A2) of the white matter tracts and sparing of gray matter on axial T1-weighted images
John Wiley & Sons
Annals of Neurology
(A3). Subsequent gadolinium-enhanced sagittal and axial T1-weighted images show gradual asymmetric resolution of enhancement over time (A4-A9). Figure 3: Axial gadolinium enhancement patterns in cervical spondylosis A typical pattern of peripheral enhancement on T1-weighted axial images was noted in 57% of patients (A-C). The pattern is circumferential (illustrated by white arrows) and spares gray matter in most (A-C). The peripheral enhancement pattern may be incomplete or partial (D) or involve only the hemi-cord (E). In 24% of patients, a single focus of enhancement was seen (F) which is not specific for spondylotic myelopathy but may represent a fragment of the general pattern of peripheral enhancement. Figure 4: Pathology of spinal cord biopsy in a case of SM with gadolinium enhancement Biopsy demonstrating gliotic white matter (A) with focal perivascular inflammation (B), making one suspect an inflammatory process. The axons, possibly due to prior freezing, are brightly eosinophilic nearly mimicking Rosenthal fibers (A and B). The scale bar represents 30 micron (µm). Figure 5: Sagittal MRIs of potential mimics of SM with enhancement and coexisting spondylosis Many causes of myelopathy can have coexisting spondylosis (arrowheads) but the enhancement pattern differs (white arrows) when the myelopathy etiology is not secondary to spondylosis. Examples of alternative myelopathy etiologies with co-existing spondylosis shown here include: ependymoma with cylindrical enhancement (A); multiple hemangioblastomas with circular enhancing lesions (B); astrocytoma with multifocal rostro-caudal enhancement (C); seropositive neuromyelitis optica with patchy
John Wiley & Sons
Page 28 of 35
Page 29 of 35
Annals of Neurology
rostro-caudal enhancement (E); spinal cord sarcoidosis with pial enhancement (E); and multiple sclerosis with focal dorsal cord enhancement (F). Figure 6: Spondylosis with Enhancement -- Diagnostic Algorithm Abbreviations: AQP4-IgG, aquaporin-4-IgG; AV, arteriovenous; CSF, cerebrospinal fluid; CT, computed tomography; MRA, magnetic resonance angiography; MRI, magnetic resonance imaging; MS, multiple sclerosis; NMO, neuromyelitis optica.
John Wiley & Sons
Annals of Neurology
Figure 1: Neuroimaging of spondylosis with gadolinium enhancement Sagittal MRI reveals a three vertebral segment spindle shaped T2-signal hyperintensity in the spinal cord (A1) with transverse gadolinium enhancement (A2) just below the site of maximum stenosis at the C5/C6 interspace. Sagittal MRI reveals three segment spindle shaped T2-signal hyperintensity with heterogenous T2-signal (B1, white arrows) at the site of flat pancake-like gadolinium enhancement (B2). Postoperative sagittal MRI of a separate patient demonstrates decompression (C1, arrowhead) and a two segment spindle shaped T2 hyperintensity with signal heterogeneity (C1, white arrow) at the site of flat pancake-like enhancement that is thicker dorsally than ventrally (C2, white arrow). A longitudinally-extensive T2weighted signal abnormality with moderate cervical spondylosis is most prominent at C4-5 on sagittal T2weighted images (D1), below which a flat pancake-like band of gadolinium enhancement is evident (D2). A spindle shaped T2-hyperintensity is seen on sagittal MRI (E1), at the middle of which focal pancake-like enhancement is present (E2). A longitudinally extensive T2-hyperintensity is present on sagittal T2-weighted MRI (F1) with pancake-like enhancement at the site of maximal stenosis; central sparing of gray matter is suggested (F2), which is better appreciated on axial images (not shown). 254x190mm (96 x 96 DPI)
John Wiley & Sons
Page 30 of 35
Page 31 of 35
Annals of Neurology
Figure 2: MRI evolution after cervical spondylosis surgery Sagittal MRI demonstrates T2-signal hyperintensity and cord swelling which appears out of proportion to degree of canal stenosis (A1) with transverse gadolinium enhancement (A2) of the white matter tracts and sparing of gray matter on axial T1-weighted images (A3). Subsequent gadolinium-enhanced sagittal and axial T1-weighted images show gradual asymmetric resolution of enhancement over time (A4-A9). 254x190mm (96 x 96 DPI)
John Wiley & Sons
Annals of Neurology
Figure 3: Axial gadolinium enhancement patterns in cervical spondylosis A typical pattern of peripheral enhancement on T1-weighted axial images was noted in 57% of patients (AC). The pattern is circumferential (illustrated by white arrows) and spares gray matter in most (A-C). The peripheral enhancement pattern may be incomplete or partial (D) or involve only the hemi-cord (E). In 24% of patients, a single focus of enhancement was seen (F) which is not specific for spondylotic myelopathy but may represent a fragment of the general pattern of peripheral enhancement. 254x169mm (96 x 96 DPI)
John Wiley & Sons
Page 32 of 35
Page 33 of 35
Annals of Neurology
Figure 4: Pathology of spinal cord biopsy in a case of SM with gadolinium enhancement Biopsy demonstrating gliotic white matter (A) with focal perivascular inflammation (B), making one suspect an inflammatory process. The axons, possibly due to prior freezing, are brightly eosinophilic nearly mimicking Rosenthal fibers (A and B). The scale bar represents 30 micron (µm). 177x127mm (300 x 300 DPI)
John Wiley & Sons
Annals of Neurology
Figure 5: Sagittal MRIs of potential mimics of SM with enhancement and coexisting spondylosis Many causes of myelopathy can have coexisting spondylosis (arrowheads) but the enhancement pattern differs (white arrows) when the myelopathy etiology is not secondary to spondylosis. Examples of alternative myelopathy etiologies with co-existing spondylosis shown here include: ependymoma with cylindrical enhancement (A); multiple hemangioblastomas with circular enhancing lesions (B); astrocytoma with multifocal rostro-caudal enhancement (C); seropositive neuromyelitis optica with patchy rostro-caudal enhancement (E); spinal cord sarcoidosis with pial enhancement (E); and multiple sclerosis with focal dorsal cord enhancement (F). 254x190mm (96 x 96 DPI)
John Wiley & Sons
Page 34 of 35
Page 35 of 35
Annals of Neurology
Figure 6: Spondylosis with Enhancement -- Diagnostic Algorithm Abbreviations: AQP4-IgG, aquaporin-4-IgG; AV, arteriovenous; CSF, cerebrospinal fluid; CT, computed tomography; MRA, magnetic resonance angiography; MRI, magnetic resonance imaging; MS, multiple sclerosis; NMO, neuromyelitis optica. 211x178mm (96 x 96 DPI)
John Wiley & Sons
