Association between cerebrospinal fluid (CSF) cystatin C concentration and long-term neurological disability in 11 patients with acute myelitis who experienced at least 1 relapse during follow-up. A, The regression line (solid line) of CSF cystatin C densitometric values at the time of the initial episode of acute myelitis (x-axis) is shown according to the Expanded Disability Status Scale (EDSS) score at the last visit (y-axis). Dotted lines represent the 95% confidence interval for the regression line. B, Median EDSS score at the last visit with range (bars) according to CSF cystatin C densitometric values in the patients who relapsed. The median cystatin C densitometric value in the subgroup (ie, 6.4) was chosen as the cutoff value. Cerebrospinal fluid cystatin C densitometric values were calculated as described in the “Data” subsection of the “Methods” section.
Gajofatto A, Monaco S, Fiorini M, Zanusso G, Vedovello M, Rossi F, Turatti M, Benedetti MD. Assessment of Outcome Predictors in First-Episode Acute MyelitisA Retrospective Study of 53 Cases. Arch Neurol. 2010;67(6):724-730. doi:10.1001/archneurol.2010.107
To identify predictors of short- and long-term outcomes in acute myelitis (AM).
First episodes of AM were retrospectively identified in a single institution. Information regarding demographics, clinical status, laboratory workup, magnetic resonance imaging of the spine and brain, and electrophysiological assessment was collected. Tau, 14-3-3 protein, and cystatin C levels were assessed de novo in stored cerebrospinal fluid samples.
A neurological department database.
Fifty-three patients with a first episode of AM.
Main Outcome Measures
The prognostic value of all variables was analyzed for the following outcomes: recovery from the initial event, symptom recurrence, conversion to multiple sclerosis (MS), and long-term disability.
Median follow-up was 6.2 years. Six patients (11%) remained monophasic; 5 (9%) developed recurrent myelitis; and 42 (79%) underwent conversion to MS. Sensory level absence, no sphincter involvement, abnormal magnetic resonance imaging findings in the brain, spinal cord lesions shorter than 3 vertebral segments, and abnormal somatosensory evoked potentials predicted MS conversion. Fifteen of 32 patients with pyramidal dysfunction at onset (47%) and 17 of 43 with relapses during follow-up (40%) had significant disability at the last visit compared with 2 of 21 patients without pyramidal manifestations (10%) and none of the patients without exacerbations (P = .006 and P = .02, respectively). In 11 patients with exacerbations, we observed a significant correlation between cerebrospinal fluid levels of cystatin C and the degree of neurological disability at the last visit (Spearman ρ = 0.69; P = .03).
For patients with first-episode AM, the conversion rate to MS is high. Motor dysfunction at onset and relapse occurrence are associated with worse outcome. Cerebrospinal fluid levels of cystatin C may prove useful for predicting the prognosis of such patients.
Acute myelitis (AM) is an inflammatory condition characterized by rapid onset of spinal cord dysfunction. The etiology is extremely heterogeneous and includes infections, paraneoplastic syndromes, systemic inflammatory diseases, multiple sclerosis (MS), and neuromyelitis optica (NMO). Despite a thorough diagnostic workup, the cause of AM remains undefined in a considerable proportion of cases (idiopathic AM).1 Independent of etiology, AM prognosis is highly variable and unpredictable. It has been reported that patients with shocklike symptoms and/or back pain at the onset of acute transverse myelitis (ATM) fail to recover in most cases.2,3 Absent cortical responses at motor and somatosensory evoked potentials (EPs) and signs of denervation on electromyography have also been associated with poor prognosis in patients with ATM.4 In addition, the presence of 14-3-3 protein and increased levels of interleukin 6 in the cerebrospinal fluid (CSF) of patients with ATM have been proposed as potential predictors of permanent neurological disability.5,6 In a significant proportion of patients, AM is the initial manifestation of a chronic disease, particularly MS. In subjects who develop subsequent conversion to MS, the first episode of AM tends to be less severe, with prominent asymmetric sensory symptoms and a better recovery compared with AM that is unrelated to MS.7,8 Other factors associated with increased risk of developing MS are the presence of basal brain abnormalities on magnetic resonance imaging (MRI) suggestive of demyelination, positive findings for CSF oligoclonal bands (OBs), and delayed visual EPs.9,10 Patients with longitudinally extensive transverse myelitis who are seropositive for anti–aquaporin 4 antibody seem to have an increased risk of NMO.11 Although some prognostic factors for AM have been suggested, robust markers are still lacking. Patients at risk of significant disability, recurrence of symptoms, or conversion to a specific neurological disease may prompt therapeutic and monitoring choices radically different from subjects with more favorable prognostic profiles.
The aim of this study was to investigate the prognostic significance of a wide range of variables, including a panel of CSF molecular markers, in a group of subjects with first-episode AM.
Potential study subjects were identified by querying our in-patient clinic database for a discharge diagnosis of myelopathy, (transverse) myelitis, encephalomyelitis, MS, or NMO. In addition, patients who initially reported sensory, motor, or autonomic disturbance were identified in the database of our outpatient center for demyelinating disorders. Patients identified from these sources were included in the study if they met the following criteria: (1) development of sensory, motor, or autonomic dysfunction attributable to the spinal cord; (2) progression to nadir within 21 days after the onset of symptoms; (3) inflammation within the spinal cord demonstrated by CSF pleocytosis, positive OB findings, or elevated IgG index or by gadolinium enhancement on MRI; and (4) at least 1 follow-up visit at our center 12 or more months after symptoms onset. Exclusion criteria were (1) compressive or neoplastic cause suspected because of MRI or myelography findings, (2) symptoms and signs attributable to simultaneous extraspinal involvement, and (3) previous episodes suggestive of central nervous system inflammation. Fulfillment of inclusion and exclusion criteria was assessed through a review of medical records by a trained neurologist (A.G.).
Demographic and clinical data were obtained from databases or from medical records if electronic information was missing or incomplete. Sex, ethnicity, age at symptom onset, personal or family history of autoimmune disease, functional system affected, presentation as acute partial myelitis (APM) as opposed to ATM,12 symmetry of manifestations, and maximum degree of neurological impairment were included. The maximum degree of neurological impairment was defined on the basis of Expanded Disability Status Scale (EDSS)13 functional systems scores as: (1) mild (≥1 functional system with a score of 1, with the others 0); (2) moderate (1 or 2 functional systems with a score of 2, with the others 0 or 1); or (3) severe (≥1 functional system with a score ≥3 or 3 functional systems with a score of 2).14
Laboratory, radiological, and electrophysiological data were considered for analysis only if obtained within 6 months after symptom onset or before a relapse occurred.
Cerebrospinal fluid cell count, protein concentration, IgG index, and OB status were collected from original reports or medical records. Presence of 14-3-3 protein, concentration of tau protein, and expression of cystatin C were assessed on CSF samples stored at −80°C in the neuropathology laboratory of our hospital. The presence of 14-3-3 protein was examined by means of immunoblot with anti–pan 14-3-3 polyclonal antibodies (Santa Cruz Biotechnology Inc, Santa Cruz, California), after sodium dodecyl sulfate–polyacrylamide gel electrophoresis. For this purpose, the equivalent of 25 μL of CSF was tested as previously described.15 Tau protein levels were measured in duplicate using the sandwich enzyme-linked immunosorbent assay kit (BioSource human total tau ELISA kit; Invitrogen Corporation, Carlsbad, California), according to the manufacturer's instructions. Tau concentrations were estimated from standard curves made for each assay. Detection of cystatin C was performed by Western blot of 2.5 mL of CSF after sodium dodecyl sulfate–polyacrylamide gel electrophoresis. Membranes were incubated with anti–cystatin C polyclonal antibody (Upstate Biotechnology, Lake Placid, New York) at 1:1000 dilution and revealed by using the Western blotting system (ECL; GE Healthcare, Milwaukee, Wisconsin). Cystatin C concentration was expressed as a densitometric value, that is, the ratio between the band optic density in each patient and mean optic density of bands in healthy controls (individuals without neurological diseases).16
Radiological data were obtained from original MRIs of the spinal cord and brain and reviewed by a trained neurologist (A.G.). When images were not available, data were collected from reports. Spinal cord and brain MRIs were considered abnormal if there was at least 1 T2-weighted hyperintense signal abnormality suggestive of an inflammatory lesion. Information regarding lesions number, location, longitudinal and axial extension, and presence of gadolinium enhancement on spinal cord MRI was also collected. Lesion number and location and gadolinium enhancement on brain MRI were assessed. We determined the presence of spatial dissemination of lesions suggestive of MS according to the recent McDonald criteria revision.17
Data regarding somatosensory, motor, visual, and brainstem auditory EPs were also included. Evoked responses were considered abnormal if the central conduction time was increased, showed a significant morphologic alteration, or was not recognizable. The normative values were those used in the laboratory in which the electrophysiological evaluation was performed.
The variables were tested for their predictive value in regard to the following outcomes: (1) recovery from the initial AM episode; (2) occurrence, location, and time interval of first relapse (if any); (3) final diagnosis at the last follow-up visit of monophasic AM, recurrent AM, or clinically definite MS (CDMS)18; MS according to the Polman criteria17 or NMO19; (4) annual relapse rate; and (5) EDSS score of at least 2.5 at the last follow-up visit during the remission phase.
The degree of recovery from the initial AM episode was defined as the highest level of neurological recovery reached before a relapse or disability progression occurred and was scored as (1) full (EDSS score, 0), (2) mild residual disability (EDSS score, 1.0-2.0), or (3) significant residual disability (EDSS score, ≥2.5). A relapse was defined as an episode of new or recurring neurological symptoms suggestive of an inflammatory central nervous system lesion lasting for at least 24 hours in the absence of increased body temperature or infection after a remission of 30 days or more.20 A diagnosis of monophasic AM was made if no relapse occurred and no new spinal cord or brain lesions appeared on MRI that allowed a diagnosis of MS. Recurrent AM was diagnosed if at least 1 spinal cord relapse occurred in the absence of extraspinal relapses and brain MRI activity during follow-up.
Differences between groups were determined with analysis of variance, 2-tailed t test, Mann-Whitney test, or Kruskal-Wallis test for quantitative variables and with χ2 test or Fisher exact test for categorical variables, as appropriate. The Bonferroni correction was used for multiple comparisons. For time-dependent outcomes, the Kaplan-Meier survival analysis method was also used, setting the appropriate time interval for each outcome. The difference between survival curves was analyzed with the log-rank test. The correlation between quantitative variables and outcomes was analyzed by computing the Spearman coefficient. A 2-tailed significance level α = .05 and 95% confidence intervals were adopted. All the analyses were performed using SPSS statistical software, version 16.0 (SPSS Inc, Chicago, Illinois).
Of the 570 patients identified through the initial database screening, 53 were eligible for the study. Follow-up time shorter than 1 year, progressive course, previous episodes of neurological dysfunction compatible with demyelination, polyregional involvement, and failure to demonstrate inflammation within the spinal cord were the most frequent reasons for exclusion. Descriptive characteristics of the study patients are shown in Table 1. The CSF samples of 19 patients were available in our laboratory for molecular markers analysis (Table 2).
Of the 53 study patients, 15 recovered fully from the initial AM episode, whereas 32 recovered with mild and 6 with significant residual disability. Variables associated with recovery are shown in Table 3.
Forty-three patients (81%) experienced at least 1 relapse during follow-up. Factors associated with relapse occurrence are shown in Table 4. Median interval from initial AM to onset of the first relapse was 10.9 months (range, 1.0-151.9 months). Median survival time to first relapse was 9.0 months for patients with motor dysfunction during the initial episode of AM vs 17.9 months for patients without motor dysfunction (P = .01, log-rank test). First relapse location was the spinal cord in 34 cases (79%), optic nerve in 4 (9%), brainstem in 3 (7%), and polyregional in 2 (5%). In patients with isolated sensory disturbances during the initial AM, the first relapse occurred in the spinal cord in 9 of 15 cases (60%) compared with 25 of 28 patients with motor and/or sphincter involvement in addition to sensory manifestations (89%) (P = .046).
Current diagnosis at last follow-up visit was monophasic AM in 6 patients (11%), recurrent AM in 5 (9%), and MS (according to the Polman criteria) in 42 (79%) (35 with relapsing-remitting MS, 3 with secondary progressive MS, and 4 with clinically isolated syndromes plus MRI criteria for dissemination in space and time). Patients' characteristics according to diagnosis are reported in Table 1. Of the patients who developed MS, 38 fulfilled criteria for CDMS. One of the 6 patients with monophasic AM also had positive serology findings (IgM) for Epstein-Barr virus in the CSF. One of the 5 patients with recurrent AM was diagnosed as having systemic lupus erythematosus and another was diagnosed as having Sjögren syndrome. Twelve of the 53 study patients presented with ATM (the others had APM), but only 2 of them fulfilled current diagnostic criteria for idiopathic ATM12 at onset. Both of these patients had subsequent spinal cord relapses and received a diagnosis of recurrent AM. In patients with recurrent AM, the median time to first relapse was 7.9 (range, 2.9-29.0) months. Median time to MS conversion according to the Polman criteria was 12.9 (range, 1.9-151.9) months, whereas the median time to CDMS conversion was 16.9 (range, 1.9-174.0) months.
In patients with recurrent AM or MS, the annual relapse rate was 0.4 (range, 0.1-1.1), without significant differences between the 2 groups. None of the variables analyzed was significantly associated with the annual relapse rate.
In the whole study group, median EDSS score at the last follow-up visit was 2.0 (range, 0-8.0). Variables associated with an EDSS of at least 2.5 at the last follow-up visit are shown in Table 5. In the subgroup of 11 patients with at least 1 relapse (mostly patients with MS), the CSF cystatin C densitometric value was significantly correlated with the last follow-up EDSS score (Figure).
This study retrospectively examined a large series of patients with a first episode of AM, which was defined according to diagnostic criteria for ATM, APM, and longitudinally extensive transverse myelitis.
We found that recovery from AM was complete or with mild disability in nearly 95% of patients, which is in contrast to earlier observations1,3 reporting a higher proportion of cases with severe residual disability. This discrepancy likely depends on differences in patient selection, such as exclusion of ischemic acute myelopathy and inclusion of APM cases in the present study. Severity of AM did not influence the degree of recovery, whereas symmetric motor manifestations and negative CSF OB findings predicted more pronounced residual disability.
We observed that an increased risk of relapse was associated with CSF cell count of 10/μL or less and with bilateral manifestations at onset.
Conversion to MS occurred in nearly 80% of AM cases. Our high prevalence of MS compared with other similar studies21,22 is probably explained by the application of the Polman criteria, longer follow-up in the group of patients with MS, and selection bias of an MS-specialized center. In most patients with MS, recurrence was found at the spinal cord level, an observation suggesting that the onset location of MS may predict the location of subsequent relapses.23 We confirm that spinal cord lesions spanning less than 3 vertebral segments on MRI,24 abnormal brain MRI findings at baseline (particularly the presence of ≥3 periventricular lesions), and dissemination of lesions in space according to the Polman criteria are predictive of subsequent conversion to MS. We did not replicate the findings of a recent study in which family history of MS and severe impairment at onset were associated with an increased risk of conversion to MS after a first episode of acute partial transverse myelitis.25 We found that a medical history negative for connective tissue disease, absence of sphincters involvement, absence of sensory level, and abnormal somatosensory EPs at onset were indicative of MS rather than monophasic or recurrent AM. Asymmetric/partial manifestations, positive CSF OB findings, and abnormal visual EPs were not predictive of MS, in contrast to data previously reported.8,10,26 However, a number of patients with monophasic or recurrent AM and positive OB findings and/or abnormal visual EPs may eventually develop MS during follow-up. Although in our study most patients who later developed MS had a partial spinal cord syndrome as the initial manifestation, the rate of MS conversion did not differ significantly between patients with APM and those with ATM (33 of 41 and 9 of 12, respectively). This finding may reflect that most of the patients in both groups had an abnormal baseline finding on the brain MRI (33 of 41in the APM group and 10 of 12 in the ATM group), which similarly increased their risk of developing MS. Previous studies differentiated the risk of MS conversion between APM and ATM, mostly in patients with normal findings on the brain MRI. However, when we looked at patients fulfilling diagnostic criteria for idiopathic ATM (which included a brain MRI that did not suggest MS), none developed MS, in accordance with previous reports.27- 29 The high prevalence of MS-like brain MRI abnormalities among patients with ATM in our series is likely due to the selection bias toward MS cases.
The 5 patients with longitudinal AM (lesion of ≥3 vertebral segments on spinal MRI) and patients with brain MRI findings not suggestive of MS and a history of optic neuritis after AM are currently undergoing assessment for serum anti–aquaporin 4 antibody (NMO-IgG) to identify potential cases of NMO spectrum disorders.11,19
Our study showed that symmetric motor dysfunction at the initial AM episode and the occurrence of at least 1 relapse were the only clinical variables associated with disability on the EDSS score at the last follow-up visit.
Recently, a set of 3 CSF polypeptides—tau protein, 14-3-3 protein, and cystatin C—has been proposed as biomarkers in patients with clinically isolated syndromes and CDMS.16 Our study explored the utility of the these markers as diagnostic and prognostic indicators in patients with first-episode AM. Results show that, similar to patients with clinically isolated syndromes and CDMS, a positive 14-3-3 protein assay coupled with a normal tau protein level (14-3-3/tau protein dissociation) was found in 12 of 15 patients with AM who later developed MS and was not significantly associated with irreversible neurological disability. Moreover, the presence of 14-3-3 protein in the CSF was not significantly associated with positive OB findings or increased IgG index in our study in contrast to the hypothesis that humoral immunity activation in the central nervous system may induce 14-3-3 protein leakage in the CSF.30 These findings further suggest that 14-3-3 protein may be detectable in the CSF in the absence of neurodegeneration.31 Hyperplastic reactive astrocytes seen in demyelinating lesions might be the source of 14-3-3 protein positivity in patients with MS.32
The third marker, cystatin C, is a protease inhibitor localized in glial and neuronal cells and actively synthesized by choroidal and leptomeningeal cells. Cystatin C counteracts the action of cathepsins, a family of lysosomal proteins released by activated microglia/macrophages, and it is thought to have a major role in modulating immune cell activation and cell death processes during inflammation.33,34 Cystatin C has been found to be downregulated in the CSF of a number of inflammatory conditions affecting the central nervous system and the peripheral nervous system.35,36 In animal models of MS, an increased expression of cystatin C has been found in white matter astrocytes, a finding paralleled by recent observations of increased CSF concentration of cystatin C in patients with MS at active and inactive stages of the disease.16,37 In our series, CSF cystatin C concentration was higher in study patients than in healthy controls. The discrepancy between this finding and the results of other studies that reported decreased levels of cystatin C in the CSF of patients with MS33 is likely explained by different patient selection criteria. All the patients with MS studied by Nagai and coworkers33 were in acute relapse at the time of lumbar puncture, whereas most patients in our series underwent lumbar puncture during AM remission (median interval from AM onset to lumbar puncture, 1 month [range, 0-6 months]). Although a decrease in CSF cystatin C levels in the acute phase may be related to high levels of cathepsin activity, this process might be reversed during the inactive stages of MS and related diseases. We observed a significant correlation between cystatin C levels in the CSF and the degree of confirmed neurological disability at the end of follow-up in patients who experienced at least 1 relapse. Although the small number of examined cases dictates caution, it can be speculated that cystatin C may mirror the extent of inflammation and demyelination at early disease stages, a factor that is known to influence the extent of later neurodegeneration and axonal loss and also triggers epitope spreading. In addition, recent studies suggest that, although cystatin C has a protective effect against oxidative stress–induced cell death, it may induce neuronal cell death in a dose-dependent manner.38,39 Based on our findings, it can be suggested that upregulation of CSF cystatin C at the time of first-episode AM may represent an unfavorable prognostic factor, to be confirmed in further studies.
Correspondence: Alberto Gajofatto, MD, Section of Clinical Neurology, Department of Neurological and Vision Sciences, University of Verona, Piazzale LA Scuro 10, 37134 Verona, Italy (firstname.lastname@example.org).
Accepted for Publication: October 22, 2009.
Author Contributions:Study concept and design: Gajofatto and Benedetti. Acquisition of data: Gajofatto, Fiorini, Vedovello, Rossi, Turatti, and Benedetti. Analysis and interpretation of data: Gajofatto, Monaco, Zanusso, and Benedetti. Drafting of the manuscript: Gajofatto and Monaco. Critical revision of the manuscript for important intellectual content: Monaco, Fiorini, Zanusso, Vedovello, Rossi, Turatti, and Benedetti. Statistical analysis: Gajofatto and Benedetti. Obtained funding: Gajofatto and Benedetti. Administrative, technical, and material support: Fiorini and Rossi. Study supervision: Benedetti.
Financial Disclosure: None reported.
Funding/Support: This study was supported by an unrestricted research grant from Merck-Serono.