Clinically isolated acute transverse myelitis: prognostic features and incidence

John Young (Department of Neurology, Christchurch Public Hospital, Christchurch, New Zealand); Stephen Quinn (Menzies Research Institute, University of Tasmania, Hobart, Australia); Mike Hurrell (Department of Radiology, Christchurch Public Hospital, Christchurch, New Zealand); and Bruce Taylor (Menzies Research Institute)

The final definitive version of “Clinically isolated acute transverse myelitis: prognostic feature and incidence” has been published in Multiple Sclerosis 15(11) 1295 – 1302 by Sage Publications Ltd, All rights reserved. © [John Young, Stephen Quinn, Mike Hurrell, and Bruce Taylor]

Abstract
Background  Demyelinating acute transverse myelitis (ATM) may be the first presentation of multiple sclerosis or remain a clinically isolated syndrome.  North Canterbury, New Zealand provides a well circumscribed population to study ATM.

Objective To identify prognostic features, clinical outcomes and incidence of ATM in North Canterbury, New Zealand. 

Methods All patients with ATM as a first neurological presentation diagnosed from January 2001 to December 2005 at a single institution providing all neurological care for North Canterbury were assessed for clinical data, MRI findings, CSF results and clinical outcomes.  CHAMPS, Barkhof/Tintore and Swanton criteria were applied to brain MRI. 

Results Sixty-one patients were identified with a mean duration of follow-up of 30 ± 17 months.  50% of patients with ATM with brain lesions by CHAMPS criteria converted to clinically definite multiple sclerosis (CDMS).  No patients with idiopathic ATM converted to CDMS.  There was a strong association with conversion to CDMS and abnormal brain MRI by CHAMPS criteria (hazard ratio, 5.63; 1.83-17.3), Barkhof/Tintore criteria (hazard ratio, 6.43; 2.31-17.9) and Swanton criteria (hazard ratio, 4.53; 1.67-12.3).  The age standardised annual incidence of ATM was 24.6 (18.2-31.1) per million, of definite and possible idiopathic ATM was 6.2 (2.9-9.6) per million, and of ATM with brain lesions was 4.7 (1.9-7.6) per million. 

ConclusionPatients with idiopathic ATM are at low risk for conversion to CDMS.  Abnormal brain MRI by CHAMPS criteria is a sensitive predictor of conversion to CDMS.  The annual incidence of ATM in North Canterbury, New Zealand is significantly higher than previously reported.

Introduction
Acute transverse myelitis (ATM) manifests as sensory, motor or autonomic dysfunction due to inflammation of the spinal cord.  The etiology of non-compressive acute transverse myelopathy may be classified as delayed radiation effect, spinal cord infarction, parainfectious, systemic autoimmune disease, multiple sclerosis (MS) or idiopathic [1].  Diagnostic criteria for idiopathic ATM were proposed by the Transverse Myelitis Consortium Working Group (TMCWG) [2].  These include bilateral signs or symptoms of spinal cord dysfunction, a clearly defined sensory level, inflammation within spinal cord demonstrated by CSF pleocytosis or elevated IgG index or gadolinium enhancement, and progression to nadir between four hours and twenty-one days.

Increasing use of brain MRI to demonstrate demyelinating lesions disseminated in time and space led to new diagnostic criteria for MS incorporating these findings [3,4].  Applying these criteria to brain MR imaging to define lesions disseminated in space is predictive of conversion to clinically definite MS (CDMS) [5]. 

Patients with idiopathic ATM and MS associated ATM are at risk of further demyelinating events.   Patients with ATM as a clinically isolated syndrome (CIS) who subsequently develop a second event disseminated in time and place meet criteria for CDMS [3]. Patients at higher risk of conversion to CDMS include those with unilateral signs or symptoms without a clear sensory level, those with spine MRI lesions measuring less than two segments and those with oligoclonal bands identified on CSF examination [6-8]. 

The incidence of ATM of all causes is between 1.34 to 4.6 per million per year [9,10].  The incidence of idiopathic and MS associated ATM are both approximately 1.0 per million per year in Albuquerque, New Mexico [10].  The incidence of demyelinating disease is known to vary geographically but there are no other estimates of incidence of idiopathic ATM [11].  The aim of our study was to identify predictors and risk of conversion to CDMS in a community based series of clinically isolated ATM patients, and to estimate the incidence of idiopathic and MS associated ATM in a well circumscribed New Zealand population.
 
Patients and methods 

Setting 
The Department of Neurology at Christchurch Public Hospital is the only neurology referral centre in the North Canterbury area of New Zealand and provides neurology care to a population of 430,000.  The diagnoses of all patients seen by the Department of Neurology are maintained in the prospective Neurology Department Database (NDD).  All neurologists practicing in the Canterbury region of New Zealand during the time of this study practiced at Christchurch Public Hospital and they registered all outpatients and inpatients on the NDD. 

Patients 
All patients diagnosed with acute transverse myelitis by a consultant neurologist between January 1, 2001 and December 31, 2005 were retrospectively identified from the NDD.  All patients must have undergone spinal MR imaging.  Exclusion criteria were an identifiable compressive, radiation, ischaemic, infectious or systemic autoimmune cause for myelopathy, previous symptoms consistent with demyelination at any site or a progression to nadir less than four hours or greater than 21 days as per the TMCWG criteria [2].  Demographic data, symptoms, signs, CSF examination results, follow-up and clinical outcomes were obtained from the case records.  Oligoclonal banding (OCB) was tested by electrophoresis with immunofixation.  Further tests including CSF virology and autoimmune screening were performed at the discretion of the treating neurologist.  Conversion to CDMS was defined as a second neurological event consistent with demyelination at a different site or spinal level.  Further follow-up and clinical information were obtained from primary care physicians and direct patient review. 

Classification 
Patients meeting the TMCWG criteria, using CSF white cell count greater than five as evidence of pleocytosis, were categorised as definite idiopathic ATM [2].  Patients who had no CSF or MRI evidence of inflammation but otherwise met these criteria were categorised as possible idiopathic ATM.  Patients with bilateral signs or symptoms and clearly defined sensory level with abnormal brain MRI by CHAMPS criteria [12] (two or more brain lesions at least three mm in diameter with at least one lesion being periventricular or ovoid) were categorised as ATM with brain lesions.  Patients presenting with unilateral signs or symptoms or without a clear sensory level, with normal brain MRI by CHAMPS criteria were categorised as partial ATM without brain lesions.  Patients fulfilling criteria for partial ATM with abnormal brain MRI by CHAMPS criteria were categorised as partial ATM with brain lesions.  Patients with incomplete records of progression to nadir were categorised as ATM with incomplete data. 

MRI acquisition and analysis 
MR images were acquired on 1.5 Tesla scanners.  Spine and brain MR images were retrospectively reviewed by a neuroradiologist (MH) blinded to the clinical details.  Brain MRI findings were classified according to the Barkhof/Tintore criteria [13,14]  as used in the McDonald criteria [3] (three of the following four features: one or more gadolinium enhancing lesions or nine T2 hyperintense lesions; one or more infratentorial lesions; one or more juxtacortical lesions; or three or more periventricular lesions), the Swanton criteria [15] (at least one T2 hyperintense lesion in two or more of the following regions; periventricular, juxtacortical, infratentorial), and CHAMPS criteria [12] (as above).  Intra-rater reliability (IRR) was assessed by blinded review of 10 randomly selected brain and spine scans with an IRR of 0.97 for reported MRI measures.

Incidence 
Identified patients residing outside the North Canterbury region were excluded from the incidence calculation but not from other analyses.  Population figures for the North Canterbury region were taken from the 2001 New Zealand census.  All incidence results were age standardised to the world population [16].

Statistical analysis 
Survivorship functions were obtained using Kaplan-Meier analysis and differences in survival were measured by the log-rank test.  All hazards ratios were calculated using Cox proportional hazards models.  As all males were OCB -ve we adjusted for age only in these instances.  When the outcome-exposure contingency tables contained a zero cell in the multivariable analysis the logistic models did not converge and we reported Fisher exact p‑values.  Otherwise, all outcomes were adjusted for age and sex and presented with 95% confidence intervals.  There was no evidence that the proportional hazards assumption was violated and model adequacy was verified using the May‑Hosmer goodness-of-fit test [17].  This test identifies potential problems in fit after using a Cox proportional hazards model by partitioning the data into deciles of risk and comparing observed and expected predicted numbers of events.  All analyses were undertaken using Stata version 10.0.
 
Results 
A total of 61 patients were identified with a first diagnosis of ATM made by a consultant neurologist.   The clinical characteristics of this group and the subgroups are shown in table 1.  Ten patients had incomplete data regarding progression to nadir.  Had their progression to nadir met the TMCWG criteria, four would have been categorised as definite idiopathic ATM, one as partial ATM without brain lesions and five as partial ATM with brain lesions.  There was no statistical difference in the demographic characteristics between these 10 patients and the remaining group. 

Investigations 
The investigation characteristics are shown in table 1.

Conversion to clinically definite MS 
The conversion to CDMS and mean follow up are shown in table 2.  Four patients categorised as ATM with incomplete data converted to CDMS.  Had their progression to nadir met TMCWG criteria three would have been categorised as partial ATM with brain lesions and the fourth as definite idiopathic ATM.  Overall conversion and conversion by subgroup utilising KM survival analysis are shown in figure 1.  The conversion rate of definite and possible idiopathic ATM was different from ATM with brain lesions (p = 0.003), partial ATM with brain lesions (p < 0.001) and partial ATM without brain lesions (p = 0.007).  The conversion rates of ATM with brain lesions, partial ATM with brain lesions and partial ATM without brain lesions did not statistically differ.

Incidence 
Three patients resided outside the North Canterbury region at the time of diagnosis.  For the remaining patients age standardised annual incidence is shown in table 2. 
 
Application of different MRI brain criteria 
Univariable analysis of the different criteria are shown in table 3.  

Others predictors of conversion 
Univariable analysis of MRI spine and CSF characteristics are shown in table 3.
 
Multivariable outcome analysis 
Bivariable and multivariable analysis of the three factors associated with relapse (OCB +ve, abnormal brain MRI, ≥ two spinal lesions) and the three protective factors (normal brain, single spinal lesion, OCB  –ve) are shown in table 4.

Discussion 
 
Our annual incidence of idiopathic and MS associated ATM combined, of 10.8 per million, is approximately five times higher than that of Albuquerque, New Mexico and two times higher than previous estimates of ATM of all causes [9,10].  These earlier studies did not include partial ATM with or without brain lesions and so we exclude these patients in this comparison.  The incidence of demyelinating disease is known to vary according to latitude and North Canterbury, latitude 43°S, is further from the equator than New Mexico, latitude 35°N, and Israel, latitude 31°N.  Our finding of an overall annual incidence of clinically isolated ATM of 24.6 per million is consistent with the known high incidence of demyelinating disease in Canterbury [18].  A latitudinal gradient of CIS, including ATM, has also been reported from Australia [11].

Idiopathic ATM is identified as a distinct subgroup of transverse myelitis.  Earlier studies demonstrate clinical and radiological characteristics different to those of MS associated ATM.  The spine MRI appearances of the lesions of idiopathic ATM are longer than those of MS associated ATM [7].  Idiopathic ATM has a conversion rate to CDMS of close to 0% compared with the up to 83% conversion rate of patients with CIS, including ATM, and an abnormal brain MRI [19,20].  However, these early studies of idiopathic ATM used varying definitions.  This heterogeneity and the need for further research prompted the TMCWG to propose uniform diagnostic criteria for idiopathic ATM [2].  The criteria emphasise a clinical history and examination consistent with spinal cord inflammation, confirmation of inflammation by CSF or MRI examination and exclusion of other demonstrable etiology.  MS associated ATM is differentiated from idiopathic ATM by the criterion “brain MRI abnormalities suggestive of MS”.  Partial ATM is differentiated by the necessity to demonstrate bilateral signs or symptoms and a clearly defined sensory level. 

These criteria identified patients who meet the clinical criteria of idiopathic ATM but without confirmed inflammation [21].  After discussion with the TMCWG these patients were categorised as possible idiopathic ATM and included with definite idiopathic ATM in their analysis.  Studies using these new criteria still define a patient population with a low conversion rate to CDMS [21,22].  Our study confirms that patients with idiopathic ATM are at low risk of conversion to CDMS with none of our patients progressing to a second demyelinating event within the timeframe of this study. 

Early case series of ATM included only patients with bilateral signs and clearly defined sensory levels [9,23].  Patients with unilateral signs or without sensory levels were classified as partial ATM and excluded.  Recent series specifically address partial ATM and demonstrate a strong association between this presentation and brain MRI suggestive of MS [24].  Our study identifies 24 patients with unilateral signs or without sensory levels, of whom 19 had brain MRI performed and seven had abnormalities consistent with MS by CHAMPS criteria.   Previous studies of patients meeting our definition of partial ATM without brain lesions demonstrate a rate of conversion to CDMS between 15-44% [25-27].  Our conversion rate of 41% is similar to these studies.  In studies of patients meeting our definition of partial ATM with brain lesions the conversion rate was 44-93% [6,27,28].  Our conversion rate of 71% is similar and confirms that this group has the highest risk of early conversion to CDMS.

The subgroup of MS associated ATM is not well defined.  The TMCWG criteria exclude any patient with brain MRI abnormalities suggestive of MS but do not define what these are.  The McDonald criteria include the Barkhof/Tintore criteria for brain MRI to demonstrate dissemination in space or, alternatively, the presence of two or more MRI lesions consistent with MS when associated with a positive CSF.  Newer MRI criteria were proposed by Swanton et al. in 2006 [15].   We used the simpler CHAMPS definition of two or more white matter lesions to define abnormal brain MRI and demonstrate that these criteria identify more patients at risk of conversion than Swanton or Barkhof/Tintore criteria, with similar hazard ratios.  Conversely, excluding patients from the idiopathic ATM group by using the CHAMPS criteria identifies a group at low risk of conversion.  The CHAMPS criteria are simple to apply and we demonstrate that their use in brain MRI, along with accurate categorisation using clinical features of ATM, provides strong prognostic information regarding conversion to CDMS. 

Presence of oligoclonal banding on CSF examination varies according to category of ATM.  Oligoclonal banding was previously demonstrated in 17% of patients with definite and possible idiopathic ATM compared with 62% of patients with partial ATM without brain lesions and 62.5% of partial ATM with and without brain lesions [21,25,26].  Our results confirm these findings, with the lowest percentage in the idiopathic ATM group and higher percentages in the partial ATM without brain lesions and ATM with brain lesions groups.  The small number of CSF examinations performed in the partial MS associated group reflects local clinical practice.  Oligoclonal banding is associated with increased risk of conversion to CDMS after clinically isolated ATM and our results confirm this association [8,28].

Our findings do not confirm the greater risk of conversion to CDMS associated with spinal lesions less than two vertebral bodies long.  Longitudinally extensive cord lesions did not provide a protective effect, but, with only five cases with this finding, the interpretation is limited.  None of the patients were thought to have neuromyelitis optica according to Wingerchuk criteria [29].  NMO antibody testing was not performed as the test was not available to us at the time.

Shortcomings of this retrospective study include the variation in investigations, with brain MRI and CSF examinations not performed on all patients.  Follow-up MR imaging was not routinely performed and so we have not assessed dissemination of lesions in time to fulfill criteria for MS.  The clinical follow-up was not uniform, of moderate duration, and reliant on contact with primary care physicians to identify further neurological events in some cases. However, demyelinating disease is a common disorder in our community and there is a high level of awareness of the association of spinal cord syndromes with MS among neurologists, neurosurgeons and general practitioners.  It is very unlikely that a person with a spinal cord syndrome would not be referred to the specialist neurology services.  Due to the lack of standardized follow up we are unable to comment on the degree of recovery from ATM.

This study provides an overview of the clinical outcomes of clinically isolated ATM in a community based neurological service in North Canterbury, New Zealand.  It shows the utility of applying CHAMPS MRI criteria to this patient population to predict conversion to CDMS and emphasises the good prognosis in the short term of those with definite or possible idiopathic ATM in relation to conversion to MS.  It also demonstrates that ATM incidence may vary geographically as does the incidence of MS.

Acknowledgements

Statistical analysis was conducted by SQ, who is a member of the International Biometrical Society. The authors have no conflicts of interest to report.

References

1. de Seze J, Stojkovic T, Breteau G, Lucas C, Michon-Pasturel U, Gauvrit JY et al.  Acute myelopathies: Clinical, laboratory and outcome profiles in 79 cases.  Brain 2001; 124: 1509-21.
2. Transverse Myelitis Consortium Working Group.  Proposed diagnostic criteria and nosology of acute transverse myelitis.  Neurology 2002; 59: 499-505.
3. McDonald WI, Compston A, Edan G, Goodkin D, Hartung HP, Lublin FD et al.  Recommended diagnostic criteria for multiple sclerosis: guidelines from the International Panel on the diagnosis of multiple sclerosis.  Ann Neurol 2001; 50: 121-27.
4. Polman CH, Reingold SC, Edan G, Filippi M, Hartung HP, Kappos L et al.  Diagnostic criteria for multiple sclerosis: 2005 revisions to the “McDonald Criteria”.  Ann Neurol 2005; 58: 840-46.
5. Tintore M, Rovira A, Rio J, Nos C, Grive E, Sastre-Garriga J et al.  New diagnostic criteria for multiple sclerosis: application in first demyelinating episode.  Neurology 2003; 60: 27-30.
6. Ford B, Tampieri D, Francis G.  Long-term follow-up of acute partial transverse myelopathy.  Neurology 1992; 42: 250-52.
7. Bakshi R, Kinkel PR, Mechtler LL, Bates VE, Lindsay BD, Esposito SE et al.  Magnetic resonance imaging findings in 22 cases of myelitis: comparison between patients with and without multiple sclerosis.  Eur J Neurol 1998; 5: 35-48.
8. Sharief MK, Thompson EJ.  The predictive value of intrathecal immunoglobulin synthesis and magnetic resonance imaging in acute isolated syndromes for subsequent development of multiple sclerosis.  Ann Neurol 1991; 29: 147-51.
9. Berman M, Feldman S, Alter M, Zilber N, Kahana E.  Acute transverse myelitis: incidence and etiologic considerations.  Neurology 1981; 31: 966-71.
10. Jeffery DR, Mandler RN, Davis LE.  Transverse myelitis.  Retrospective analysis of 33 cases, with differentiation of cases associated with multiple sclerosis and parainfectious events.  Arch Neurol 1993; 50: 532-35.
11. Lucas R, Taylor BV, Ponsonby A, Chapman C, Coulthard A, Dear K et al.  Latitudinal variations in incidence of first demyelinating events; descriptive analyses of case participants in the Ausimmune Study.  Mult Scler 2008; 14(supp 1): S190 –S191.  Abstract.
12. Jacobs LD, Beck RW, Simon JH, Kinkel RP, Brownscheidle CM, Murray TJ et al.  Intramuscular interferon beta-1a therapy initiated during a first demyelinating event in multiple sclerosis.  CHAMPS Study Group.  N Engl J Med 2000; 343: 898-904.
13. Barkhof F, Filippi M, Miller DH, Scheltens P, Campi A, Polman CH et al.  Comparison of MR imaging criteria at first presentation to predict conversion to clinically definite multiple sclerosis.  Brain 1997; 120: 2059-69.
14. Tintore M, Rovira A, Martinez MJ, Rio J, Díaz-Villoslada P, Brieva L et al.  Isolated demyelinating syndromes: comparison of different MR imaging criteria to predict conversion to clinically definite multiple sclerosis.  Am J Neuroradiol 2000; 21: 702-06.
15. Swanton JK, Fernando K, Dalton CM, Miszkiel KA, Thompson AJ, Plant GT et al.  Modification of MRI criteria for multiple sclerosis in patients with clinically isolated syndromes.  J Neurol Neurosurg Psychiatry 2006; 77: 830-33.
16. Doll R, Cook P.  Summarizing indices for comparison of cancer incidence data. Int J Cancer 1967; 2: 269-79.
17. May S, Hosmer DW.  A simplified method of calculating an overall goodness-of-fit test for the Cox proportional hazards model.  Lifetime Data Anal 1998; 4: 109-20.
18. Taylor BV, Richardson A, Mason DF, Willoughby E, Abernethy D, Sabel C.  Prevalence of multiple sclerosis in New Zealand.  Mult Scler 2008; 14(suppl 1): S202.  Abstract.
19. Al Deeb SM, Yaqub BA, Bruyn GW, Biary NM.  Acute transverse myelitis.  A localized form of postinfectious encephalomyelitis.  Brain 1997; 120: 1115-22.
20. O’Riordan JI, Thompson AJ, Kingsley DP, MacManus DG, Kendall BE, Rudge P et al.  The prognostic value of brain MRI in clinically isolated syndromes of the CNS.  A 10-year follow-up.  Brain 1998; 121: 495-503.
21. de Seze J, Lanctin C, Lebrun C, Malikova I, Papeix C, Wiertlewski S et al.  Idiopathic acute transverse myelitis: application of the recent diagnostic criteria.  Neurology 2005; 65: 1950-53.
22. Bruna J, Martinez-Yelamos S, Martinez-Yelamos A, Rubio F, Arbizu T.  Idiopathic acute transverse myelitis: a clinical study and prognostic markers in 45 cases.  Mult Scler 2006; 12: 169-73.
23. Lipton HL, Teasdall RD.  Acute transverse myelopathy in adults.  A follow-up study.  Arch Neurol 1973; 28: 252-57.
24. Scott TF, Bhagavatula K, Snyder PJ, Chieffe C.  Transverse myelitis.  Comparison with spinal cord presentations of multiple sclerosis.  Neurology 1998; 50: 429-33.
25. Cordonnier C, de Seze J, Breteau G, Ferriby D, Michelin E, Stojkovic T et al.  Prospective study of patients presenting with acute partial transverse myelopathy.  J Neurol 2003; 250: 1447-52.
26. Scott TF, Kassab SL, Singh S.  Acute partial transverse myelitis with normal cerebral magnetic resonance imaging: transition rate to clinically definite multiple sclerosis.  Mult Scler 2005; 11: 373-77.
27. Sellner J, Luthi N, Buhler R, Gebhardt A, Findling O, Greeve I et al.  Acute partial transverse myelitis: risk factors for conversion to multiple sclerosis.  Eur J Neurol 2008; 15: 398-405.
28. Miller DH, Ormerod IE, Rudge P, Kendall BE, Moseley IF, McDonald WI.  The early risk of multiple sclerosis following isolated acute syndromes of the brainstem and spinal cord.  Ann Neurol 1989; 26: 635-39.
29. Wingerchuk DM, Lennon VA, Pittock SJ, Lucchinetti CF, Weinshenker BG.  Revised diagnostic criteria for neuromyelitis optica.  Neurology 2006; 66: 1485-89.