Comparison of Clinical Outcomes of Transverse Myelitis Among Adults With Myelin Oligodendrocyte Glycoprotein Antibody vs Aquaporin-4 Antibody Disease

IMPORTANCE Recognizing the differences between transverse myelitis (TM) associated with myelin oligodendrocyte glycoprotein (MOG) antibody (Ab) disease vs aquaporin-4 (AQP4)-Ab disease and prognosticating patients within each group may be an important factor for better clinical treatment for these respective patients. OBJECTIVES To compare the clinical and radiological findings of the first TM episode in patients with MOG-Ab disease vs patients with AQP4-Ab disease and to assess factors associated with worse outcomes and relapse risk. DESIGN, SETTING, AND PARTICIPANTS This retrospective cross-sectional study used data collected from the Oxford Neuromyelitis Optica Service database, a national service that serves the south of England, including detailed clinical data, and high-quality imaging from within 4 weeks of the first TM episode from patients with MOG-Ab disease or AQP4-Ab disease and a confirmed history of TM from April 2018 to January 2019. Data analyses were conducted from February 2019 to April 2019. MAIN OUTCOMES AND MEASURES Onset features of each condition measured using the Expanded Disability Status Scale (EDSS) score, time to an EDSS score of 6, time to relapse, and residual sphincter dysfunction at least 6 months after the first TM episode and at last follow-up. Abstract(continued) after TM occurred in 17 patients with MOG-Ab disease (37%) and 36 patients with AQP4-Ab disease (52%). Concomitant brainstem lesions in patients with MOG-Ab disease were associated with a higher mean (SD) EDSS score at recovery (3.5 [2.3] vs 1.4 [0.9]; P < .001). In patients with AQP4-Ab disease, those younger than 50 years were more likely to relapse (27 of 36 patients aged <50 years [75%] vs 9 of 33 patients aged (cid:2) 50 years [27%]; P < .001) and those 50 years and older were more likely to reach an EDSS score of 6 (19 of 33 patients aged (cid:2) 50 years [58%] vs 11 of 36 patients aged <50 years [31%]; P = .03). CONCLUSIONS AND RELEVANCE This study found that in patients who experienced a TM episode, short and multiple lesions at onset were more common in those with MOG-Ab disease than among those with AQP4-Ab disease. The presence of a brainstem lesion at the time of a TM episode in patients with MOG-Ab disease was associated with a worse recovery. In patients with AQP4-Ab disease, those 50 years and older at disease onset had more disability, and those younger than 50 years at disease onset had more relapses. to NMOSD. 2 In both conditions, TM is a hallmark feature, and disability appears to be associated with TM episodes. A 2017 study 3 suggested that MOG-Ab disease is milder than NMOSD associated with AQP4-Ab. In this study, we describe the demographic characteristics, presentation, clinical and paraclinical features, and outcomes in the largest cohort of adult patients with TM who have tested positive for serum MOG-Ab to date, to our knowledge, and compare them with those of adult patients with TM and AQP4-Ab disease. In particular, we assess the features of the TM episode for factors associated with worse outcomes and increased relapse risk. With the understanding that MOG-Ab disease has been defined as a separate entity from AQP4-Ab disease 4 and multiple sclerosis (MS), we aim to describe the characteristics associated with TM in this condition and highlight clinically relevant long-term outcomes. MOG-Ab disease and 41 patients (59%) with AQP4-Ab disease. These were conducted at least 6 months after an acute relapse. Complete resolution of abnormalities in MRI findings, without atrophy, occurred in 16 patients (62%) with MOG-Ab disease and 7 patients with AQP4-Ab disease ( A 2017 study 21 of follow-up brain MRIs outside of relapse suggested resolution of lesions was more common in MOG-Ab disease. Our study’s findings in the spinal cord were consistent with this. Up to 62% of patients with MOG-Ab disease showed complete resolution of lesions in MRI findings at least 6 months after the first TM episode. This occurred in only 17% of patients with AQP4-Ab disease, in whom residual lesions or atrophy were more common. Studies using nonconventional MRI methods to examine the integrity of the normal-appearing cord after an episode of TM are needed to establish if this recovery is full. The pathological basis for these differences in clinical and MRI features should also be further explored.


Introduction
Neuromyelitis optica spectrum disorder (NMOSD) is an autoimmune inflammatory disorder of the central nervous system characterized by episodes of optic neuritis (ON) or transverse myelitis (TM).
Aquaporin-4 (AQP4) antibodies (Abs) are a highly specific pathologic marker of NMOSD and associated with a chronic relapsing condition with high morbidity and mortality. 1 The discovery of myelin oligodendrocyte glycoprotein (MOG)-Ab disease has suggested that MOG-Ab disease is clinically and radiologically similar to NMOSD. 2 In both conditions, TM is a hallmark feature, and disability appears to be associated with TM episodes. A 2017 study 3 suggested that MOG-Ab disease is milder than NMOSD associated with AQP4-Ab. In this study, we describe the demographic characteristics, presentation, clinical and paraclinical features, and outcomes in the largest cohort of adult patients with TM who have tested positive for serum MOG-Ab to date, to our knowledge, and compare them with those of adult patients with TM and AQP4-Ab disease. In particular, we assess the features of the TM episode for factors associated with worse outcomes and increased relapse risk. With the understanding that MOG-Ab disease has been defined as a separate entity from AQP4-Ab disease 4 and multiple sclerosis (MS), we aim to describe the characteristics associated with TM in this condition and highlight clinically relevant long-term outcomes.

Methods
Ethics patients with MOG-Ab disease were found not to have AQP4-Ab disease and vice versa. In particular, the MOG-Ab assay was to full-length MOG with an IgG1-specific secondary antibody, which increases specificity. 5

Cohort
Every patient referred to our clinic is entered into a database using Access database software version 12.0 (Microsoft Corp) to facilitate patient care. At their first visit, each patient is offered the opportunity to sign written informed consent for their clinical data to be used for research. We screened the database for adult patients who had experienced a TM episode and had either MOG-Ab disease or AQP4-Ab disease from April 2018 to January 2019. Patients who had confounding neurological comorbidities, who had not signed consent, or for whom acute imaging or clinical details from the acute TM episode were unavailable were excluded. We collected demographic characteristics (ie, age at onset of disease, sex, and self-reported race/ethnicity), date of onset of the disease and date of last follow-up (used to calculate disease duration), features of the onset TM (ie, signs and symptoms), Expanded Disability Status Scale (EDSS) score at nadir, time to acute treatment and type of acute treatment, magnetic resonance image (MRI) features during acute episode, longterm disability outcomes (ie, motor and sphincter dysfunction), relapse history, long-term treatment, and features of any follow-up imaging performed outside of acute relapse treatment. Recovery from the onset episode of TM was described using a subgroup of patients who had been relapse free for at least 1 year after the TM episode to allow for full recovery to take place before the residual disability was measured.
Our service receives referrals from the entire southern United Kingdom, so MRI scans were performed locally during the acute TM episode and then reviewed at our center. All MRI scans were performed on 1.5-T or 3-T MRI machines and included sagittal cervical and thoracic T1 (slice thickness, 3-4 mm; echo time, 6.8-12.2 milliseconds; repetition time, 469-540 milliseconds) and T2 spinal cord imaging (slice thickness, 3-4 mm; echo time, 100-116 milliseconds; repetition time, 3000-4840 milliseconds). Not all participants were given contrast. Axial images acquired among these patients were either T2 or T2* weighted images in the affected area (slice thickness, 4-5 mm). These scans were then reviewed by one of us who specializes in inflammatory neuroradiology (W.K.) and 4 of us This study aimed to identify the features of TM in patients with MOG-Ab disease or AQP4-Ab disease during the acute episode and during follow up. We described and compared the clinical MRI features in each patient group during the TM episode. We then examined long-term outcomes of interest: disability, as measured by the EDSS, and the presence of long-term sphincter dysfunction.
Finally, we aimed to assess the TM episode features that might be associated with a worse prognosis in each group. Owing to the differences in demographic characteristics that have been described in these conditions, 7 we hypothesized that age and sex would be different between the groups and so were considered confounders. In all analyses, age, sex, and disease duration were included as variables.
factors were selected and each assessed with a univariate analysis. The significant factors were then entered into the multiple regression model with the factors that differed between MOG-Ab disease and AQP4-Ab disease groups (ie, sex, age, and disease duration). P values were 2-tailed, and statistical significance was set at .05. Data analysis was conducted from February 2019 to April 2019.

Demographic Characteristics
Among 153 patients admitted to our clinic with MOG-Ab disease or AQP4-Ab disease who had experienced a TM episode, 5 were excluded for having confounding neurological comorbidities, 13 patients were excluded because they had not signed informed consent, and 10 were excluded because we were unable to track any acute imaging or clinical details from the acute TM episode.

Clinical Presentation and Baseline Features of First TM
The frequency of motor and sensory symptoms and sphincter involvement at nadir did not

Spinal Cord MRI Features Within 4 Weeks of Symptom Onset
Short lesions (<3 vertebral levels) with no concomitant long lesion occurred in 11 patients (24%) with MOG-Ab disease and 8 patients (12%) with AQP4-Ab disease (Table). The presence of short lesions was particularly prevalent in patients with MOG-Ab disease whose first TM episode occurred as a relapse rather than at onset; 50% of these patients had only short lesions (7 of 14 patients) compared with 13% of those whose TM episode occurred at the onset of the disease (4 of 32 patients). This pattern did not occur in the AQP4-Ab group; the distribution of short lesions was 13% among patients who experienced the first TM episode at disease onset (7 of

Brainstem Imaging
The presence of associated brainstem lesions was not significantly different between groups.
However, when an associated brainstem lesion occurred at the time of a TM, it was more commonly contiguous with the cord lesion in patients with AQP4-Ab disease than patients with MOG-Ab disease   In patients with AQP4-Ab disease, age was associated with relapse risk, with younger patients more at risk (odds ratio, 0.94 [95% CI, 0.88-0.98]; P < .001). Disease duration was also associated with risk of relapse (odds ratio, 1.02 [95% CI, 1.01-1.04]; P = .02). There were no associations of treatment, sex, or onset severity with risk of relapse. To further explore the association of age with relapse, a survival analysis was performed. Using the median cutoff of age 50 years, patients in the younger group (n = 36) were more likely to relapse than those in the older group (n = 33) (log-rank: P = .003) ( Figure 1A). The mean disease duration did not differ between the older and younger groups.

Relapses
Among patients with MOG-Ab disease, there was no specific characteristic associated with relapses, and the only statistically significant factor was disease duration (odds ratio, 1.002 [95% CI, 1.001-1.004]; P = .004). The association of age with relapse observed in patients with AQP4-Ab disease was not observed in those with MOG-Ab disease.

Disability
In patients with AQP4-Ab disease, the factors associated with EDSS score in the univariate analysis were age and severity of the onset episode, and they retained statistical significance in a multivariate analysis (R 2 = 0.67; P = .003). In a survival analysis using the 2 age groups, the younger group, although more likely to relapse, was less likely to reach an EDSS score of 6 (log-rank: P < .001) ( Figure 1B).  (Figure 2A). Because of the association of age with disability in patients with AQP4-Ab disease, and acknowledging that the mean age of this group was older, we used the nearest neighbor method in propensity score-matching to obtain 2 more closely matched subgroups. Kaplan-Meier analysis on these groups, and patients with AQP4-Ab disease were still more likely to reach an EDSS score of 6 ( Figure 2B).

JAMA Network Open | Neurology
Among patients with MOG-Ab disease, higher EDSS score was associated with the presence of a brainstem lesion, whether symptomatic or asymptomatic (R 2 = 0.34; P < .001), but not with age, sex, the presence of brain lesions, disease duration, or EDSS score at nadir. The mean (SD) long-term EDSS score in patients with brainstem lesions in MRI results was significantly higher than in patients without brainstem lesions (   at onset or as a relapse (Figure 3). Median (range) disease duration between the 2 groups was not significantly different (patients with brainstem lesions, 38

Long-term Sphincter Dysfunction
There was no significant difference between patients with MOG-Ab disease and patients with Among patients with MOG-Ab disease, the factor associated with need for long-term catheterization was the presence of a conus lesion (R 2 = 0.33; P = .002). The need for long-term catherization was not associated with the severity of the episode or EDSS score at last follow-up.
Among patients with AQP4-Ab disease, the univariate analysis suggested that the need for long-term catheterization was associated with EDSS score at last follow up (R 2 = 0.19; P = .02) and the presence of a conus lesion (R 2 = 0.19; P = .048). Figure 4 presents the locations of lesions in patients who required long-term catheterization, patients who had sphincter dysfunction without requiring catheterization, and unaffected patients in each group. Sphincter dysfunction existed as a spectrum in both conditions, from mild to requiring catheterization.
In the recovery analysis, erectile dysfunction occurred more commonly in patients with MOG-Ab disease than in those with AQP4-Ab disease ( with AQP4-Ab disease (P < .001).

Discussion
In this study, we describe the largest single-center longitudinal cohort, to our knowledge, of adult patients with MOG-Ab disease who experienced at least 1 TM episode and compared their outcomes with those of patients with AQP4-Ab disease who experienced at least 1 TM episode. In particular, we highlighted the outcomes and prognostic features in each group. We found that in patients with MOG-Ab disease, short spinal cord lesions and multiple lesions at disease onset were more common, which suggests that these patients may be at higher risk of receiving a misdiagnosis of MS. This is of particular importance in patients with MOG-Ab disease who have their first TM as a relapse, among whom approximately half presented with only short lesions on spinal cord MRI results. Brain lesions also occurred more commonly in patients with MOG-Ab disease at the time of their TM episode and were often asymptomatic. In particular, we found that the presence of a brainstem lesion at the time of a TM episode in patients with MOG-Ab disease may be a risk factor for a worse recovery. In patients with AQP4-Ab disease, the overall disability at last follow-up was greater with older age at disease onset and was associated with a worse recovery, and a younger age at disease onset was associated with a higher relapse risk.
There were noticeable differences in age and sex between the 2 groups. We used propensity score matching to analyze a more closely matched subgroup; however, owing to our AQP4-Ab group having significantly fewer men, it was difficult to fully match the groups without removing all men from the analysis. While all results should be considered in light of the age and sex differences, AQP4-Ab disease appeared to be associated with worse overall disability.
In both groups, TM was more likely to occur at disease onset. A 2013 study 9 found that up to  Previous smaller studies in AQP4-Ab disease have highlighted the significance in the length of the lesion at onset. 11,12 In this study, the length of the lesion was associated with the severity of the disease onset presentation in both groups. In patients with AQP4-Ab disease, older age at onset was associated with the long-term disability outcome of reaching an EDSS score of 6. A 2016 study 13 found an association of age with relapses in patients with AQP4-Ab disease, and our study also found that in patients with AQP4-Ab disease who experienced a TM episode, the younger age group was more likely to relapse; older patients were less likely to relapse but accumulated disability from the onset episode. In patients with MOG-Ab disease, residual disability was accumulated from the onset episode regardless of age at onset.
Important information was obtained from MRI findings in these conditions. We found that up to 48% of patients with MOG-Ab disease who experienced a TM episode presented with short lesions at onset, and 24% presented with only short lesions on their initial MRI results and no associated LETM. We also found that patients with MOG-Ab disease more commonly presented with multiple spinal cord lesions than patients with AQP4-Ab disease did, which agrees with results presented in a 2019 study by Dubey et al. 14 Short cord lesions may be misdiagnosed as MS, particularly in pediatric populations, in which patients with MOG-Ab disease have received an MS misdiagnosis in the past. 15 Therefore, axial imaging of the spinal cord is crucial in these cases, as is brain imaging. In our adultonly cohort, brain lesions occurred in more than half of patients with MOG-Ab disease at the time of the first TM episode, and most lesions were asymptomatic. Of note, 41% of patients with MOG-Ab disease had ADEM-like brain lesions, whereas none of the patients with AQP4-Ab disease had ADEMlike brain lesions. Therefore, ADEM-like lesions occurring together with an LETM or concomitant ON (unilateral or bilateral) may be an indicator for MOG-Ab disease.
This study found that involvement of the brainstem at the time of the TM, whether symptomatic or asymptomatic, was associated with worse outcomes in patients MOG-Ab disease. The mean EDSS score at recovery in patients with MOG-Ab disease with brainstem lesions was significantly higher than in those without brainstem lesions. To our knowledge, this is the first study that found this potentially important association. This association was not observed in patients with AQP4-Ab disease.
The predilection for MOG-Ab disease to involve the conus has been suggested as an important differentiator between MOG-Ab disease and AQP4-Ab disease. 3,7,14,16,17 The prevalence of conus involvement in our study in patients with MOG-Ab disease was 39%, and although prior reports have varied from 4% 18 to 75%, 19 a 2019 study by Dubey et al 14 aligned with our findings. We also found that involvement of the conus was associated with long-term sphincter dysfunction, particularly the requirement for catheterization. While sphincter dysfunction is not limited to patients with MOG-Ab disease, the proportion of patients with long-term sphincter dysfunction was not proportional to the recovery of mobility compared with patients with AQP4-Ab disease. Therefore, as an outcome measure, the EDSS score, which is primarily a measure of mobility, did not fully address the outcomes in MOG-Ab disease, in which mobility may be recovered but sphincter dysfunction remains an important outcome that needs to be quantified and treated effectively, as it significantly affects quality of life. 20 When examining the lesion locations in patients requiring a catheter, we hypothesize that sphincter dysfunction occurs by 1 of 2 mechanisms: (1) severe involvement of the spinal cord above the conus, such as that seen in AQP4-Ab disease, is generally more destructive and more likely to affect mobility, or (2) damage to the conus itself, which may be more vulnerable even with a less severe underlying disease, occurs more commonly in patients with MOG-Ab disease.
A 2017 study 21 of follow-up brain MRIs outside of relapse suggested resolution of lesions was more common in MOG-Ab disease. Our study's findings in the spinal cord were consistent with this.
Up to 62% of patients with MOG-Ab disease showed complete resolution of lesions in MRI findings at least 6 months after the first TM episode. This occurred in only 17% of patients with AQP4-Ab disease, in whom residual lesions or atrophy were more common. Studies using nonconventional MRI methods to examine the integrity of the normal-appearing cord after an episode of TM are needed to establish if this recovery is full. The pathological basis for these differences in clinical and MRI features should also be further explored.

Limitations and Strengths
Our study had limitations. One limitation was that we used retrospective data collection.
Additionally, MRI scans were from different centers, so not all patients were given contrast or had axial views taken. However, a strength of this study was that these data were more relevant to clinical practice with the observations having been made in a realistic clinical setting and patients having been scanned at different centers. Another limitation was that this study had small sample sizes, so variable selection for multivariate models may be misleading. 22 However, we selected clinically relevant confounders and included potential variables, namely age, sex, and disease duration.

Conclusions
In conclusion, the differences this study found between patients with MOG-Ab disease and patients with AQP4-Ab disease may assist clinicians in better risk stratification. Patients with MOG-Ab disease who experienced a TM episode were likely to fulfill criteria for Ab-negative NMOSD, so MOG-Ab disease should always be screened for. This study also found that short cord lesions were common in patients with MOG-Ab disease, particularly at the time of a TM relapse. Additionally, the differences between MOG-Ab disease and AQP4-Ab disease suggest that long-term disability outcomes are different in each group, and treatment response should be defined by disease-specific outcome measures.