Factors Associated With Relapse and Treatment of Myelin Oligodendrocyte Glycoprotein Antibody–Associated Disease in the United Kingdom

This cohort study investigates factors associated with long-term risk of relapse among patients with myelin oligodendrocyte glycoprotein antibody–associated disease in the United Kingdom.


Introduction
Myelin oligodendrocyte glycoprotein antibody-associated disease (MOGAD) is an inflammatory demyelinating disease of the central nervous system with antibodies against myelin oligodendrocyte glycoprotein (MOG-Ab) that are expressed on the cell surface of oligodendrocytes 1 and myelin sheath in the central nervous system. The disease is a recently characterized acquired demyelinating syndrome, which has been shown in several reports [2][3][4][5] to have clinical and pathological features distinct from multiple sclerosis (MS) and aquaporin-4 antibody-positive neuromyelitis spectrum disorder (AQP4-NMOSD). [6][7][8] There appears to be an age-related factor associated with the onset phenotype, with adult cohorts having an increased proportion of ON involvement [9][10][11][12] and pediatric cohorts having acute disseminated encephalomyelitis (ADEM). [13][14][15] MOGAD is likely to have at least 2-fold as the prevalence of NMOSD in some regions. 16 Because MOGAD can range from a monophasic condition to a treatment-resistant relapsing disorder, it is essential to understand the risk of long-term relapse and factors associated with prognosis when deciding immunosuppressive treatments. 9,11,[17][18][19][20] However, this risk is unclear owing to the use of nonincident cohorts or short follow-up times in previous studies. Additionally, most previous studies have reported results from adult-predominant centers or pediatric services; thus, the association of age with outcome may differ by study.
We collected data from the 5 main hospitals in the UK that care for patients with MOGAD, including 2 pediatric specialist centers in the UK, to study risk of relapse, factors associated with relapse, and treatment response among UK patients with MOGAD across a range of ages with longer followup times.

Methods
Patients included in this cohort study had been enrolled according to research and development review board with informed consent or service evaluation committee approval for each of the

JAMA Network Open | Neurology
Factors Associated With Relapse of Myelin Oligodendrocyte Glycoprotein Antibody-Associated Disease

Study Design and Clinical Data Collection
This UK study of 276 patients with MOGAD used databases of patients from 5 health care centers: 146 patients at Oxford and its outreach sites, 65 patients at Liverpool, 32 patients at a children's hospital in Birmingham, 22 patients at a children's hospital in London, and 11 patients at Cardiff, Wales. All patients had clinical events in keeping with MOGAD and were positive on the cell-based IgG1 assay 21,22 in the Oxford Autoimmune Neurology Diagnostic Laboratory. Shorter-term data (ie, with a last follow-up date of 2016) was previously reported from 75 patients in the Oxford group, which included a small incident group of 44 patients. 11 Clinical data included sex, self-identified race, age at onset of MOGAD, family history of autoimmune disease, clinical attack phenotype, number of attacks, disease duration, treatment of acute attacks, chronic immunosuppressant therapy, and presence of cerebrospinal fluid oligoclonal bands. Race was collected because of the observed association of race with outcomes in AQP4-NMOSD to investigate if there were similar associations among patients with MOGAD. 6 Clinical presentation at onset was categorized into 4 clinical subgroups: ON, TM, simultaneous ON with TM, and ADEM, brain, or brainstem. The latter categorization included individuals who presented with a combination of brain or brainstem attacks with other onset locations, such as ON and TM (eTable 1 in the Supplement).
To analyze risk of relapse, we identified from the total cohort an incident group (prospectively followed from onset) defined as individuals diagnosed with MOG antibodies before a second attack.
This was done in order to remove the bias of overestimating relapse risk by enriching data with individuals referred years after onset only because they presented with a relapse (ie, retrospectively recruited).
Onset age groups for subgroup analysis were divided into ages younger than 12 years, 12 to 18 years, older than 18 to 40 years, and older than 40 years, also referred to as young pediatric, teenage, young adult, and older adult groups, respectively. The age of 12 years was selected as being a reasonably pragmatic cutoff for puberty, and the age range of 12 to 18 years was chosen for teenage years; other groups were identified after reviewing the frequency plot ( Figure 1A) to ensure that an equitable number of patients were represented in each group. To compare the association of immunosuppression and immunomodulation with relapse outcomes, we analyzed only patients with relapsing disease in treatment analysis (a first clinical event in MOGAD is not a currently established indication for long-term immunosuppression or immunomodulation). We categorized patients into 5 maintenance therapy subgroups: (1) prednisolone (P), (2) steroid-sparing immunosuppressive treatment (IST; ie, azathioprine, methotrexate, or mycophenolate mofetil), various combinations of 1 and 2, (3) MS-disease-modifying therapy (MS-DMT), (4) intravenous immunoglobulin (IVIG; alone or in combination with rituximab, P, IST, or P and IST), and (5) rituximab (alone or in combination with P, IST, or P and IST). The no-treatment group refers to the off-treatment phase among patients in the relapsing group, which could occur between any attack or between the last attack and last follow-up.
First-line therapy was analyzed for treatment groups with sufficient numbers of patients. Patients who received treatment with cyclosporin, cyclophosphamide, tocilizumab, or mitoxantrone were excluded from the survival analysis because of the low number of patients. Annualized relapse rates (ARRs) were calculated as number of relapses per year (inclusive of incident event) and included patients with at least 12 months of follow-up to prevent overinflation of rates.

Statistical Analysis
We applied χ 2 and 1-way analysis of variance (or the Kruskal-Wallis test) to compare multiple groups of data between subgroups. Kaplan-Meier survival curves were used for the time-to-event analyses, and across-group comparisons used a univariate Cox proportional regression model. Univariate analyses were performed for nontreatment factors (ie, sex, race, health care center, onset age, and onset phenotype on relapse risk). These analyses were adjusted for other factors as indicated in the results. Analyses for maintenance therapies and risk of relapse were performed using multivariate models, adjusting for covariates. We selected covariates associated with a change in relapse risk  25 Mixed race and other race were combined owing to low numbers; mixed race was a specific term used in data collection. There was an increased proportion of Asian individuals in the young pediatric group (12 of 57 individuals with race data [21.1%]) compared with other age subgroups.     As expected, the incident group had a shorter median disease duration and an increased proportion of patients with monophasic disease. The age of onset had 2 peaks in childhood (below and above age 12 years), 1 peak between ages 19 and 40 years, and another peak between ages 50 and 60 years ( Figure 1A). There were similar sex ratios, rates of coexisting other autoantibodies, and annual relapse rates across age groups (Table 1). Although the 2 younger age of onset groups had the longest disease duration, 37 of 69 patients (53.6%) in the youngest age group had relapsing disease, which is similar to rate of the older age groups. Oligoclonal bands were more frequent in the young pediatric group and less frequent in the teenage group (Table 1).  However, among 38 patients aged younger than 12 years, ADEM followed by ON was the most common phenotype (7 patients [18.4%]). Relapsing TM was found only among adults (ie, those, aged >18 years), and relapsing ADEM was found only among patients aged younger than 12 years (eFigure in the Supplement).

Risk of Relapse After Onset Attack
Because the MOG antibody test has been available only in recent years, patients with onset many years ago will be unlikely to represent for diagnosis unless they relapse. Thus, relapse risk among individuals with onset long ago is likely to appear higher that its true value, and this will be associated with a bias toward an increased relapse rate in the nonincident group. Figure 2A  0.20-0.82; P = .01) (Figure 3C), and this association remained when a follow-on course of prednisolone was included as a covariate (HR, 0.47; 95% CI, 0.22-0.99; P = .049).
There was no linear association of onset age with time to relapse ( Figure 2D)    No overall significant difference was found in time to relapse across 5 main treatment groups, although numbers in some groups were small ( Figure 3A). Differences in risk of relapse among patients treated with prednisolone, IST, or prednisolone with IST compared with the no-treatment group were not statistically significant (HR, 0.65; 95% CI, 0.42-1.01; P = .06). There was no difference in risk of relapse after adjustment for previously identified covariates. The group treated with rituximab relapsed sooner, with an initial decline in the MS-DMTs survival curve ( Figure 3A). Figure 3B shows the separation of the most common treatment group to explore if there was any differences among prednisolone alone, IST alone, and a combination (IST with P). There was no significant  It is difficult to fully adjust for indication bias; thus, we compared first-line maintenance therapies with the no-treatment group using multivariate analysis, including age at onset, isolated ON at onset, and any TM at onset as covariates ( Figure 3C). Although across all groups there was no significant difference, in multivariate analysis at 24 months, there was a significantly decreased risk of relapse for patients in the prednisolone group compared with patients in the no-treatment group (HR, 0.33; 95% CI, 0.12-0.92; P = .03) ( Figure 3C).  Figure 3D).

Discussion
This cohort study was a multicenter collaboration analyzing UK patients with MOGAD with good representation from all age groups, which included a large incident group with longer follow-up than seen in previous studies. Overall, we found a 31.7% risk of relapse within 4 years and a 36.3% risk of relapse within 8 years; risk was increased among young adults (ie, those aged older than 18 to 40 years), with a 45.8% risk of relapse within 4 years and a 59.4% risk within 8 years. Having ON alone at onset was associated with increased risk of relapse before adjusting for a course of oral prednisolone, and any TM was associated with decreased risk. Proportions of the population by race varied by age group, with an increased proportion of Asian individuals in the young pediatric group.
There were also more ADEM presentations in the young pediatric group. First-line therapy with prednisolone, nonsteroidal immunosuppressants (ie, azathioprine, mycophenolate, or methotrexate), or a combination was associated with decreased risk of relapse by approximately 50%.
This study supports baseline features across ages noted in previous large MOGAD cohort reports. [9][10][11][12]19,20,23,24 A 2021 large MOGAD cohort publication 10,15 compared baseline features across age ranges that are similar to those investigated in our study and noted more ADEM at onset and less ON at onset in the young pediatric group but not in the older pediatric or adolescent group, which is in line with our findings. We did not find a decreased percentage of TM in the adolescent group, although our age categorization was slightly older: ages 12 to 18 years compared with 10 to 17 years in the Cobo-Calvo et al 10 cohort. In accordance with their study, we also noted an increased relapse risk in the young adult group. However, unlike our study, Cobo-Calvo et al 10 did not find decreased risk with any TM, but the cohorts differed in that our analysis was performed on an incident group (ie, focused on individuals diagnosed at onset). The findings of a previous study 11 suggested that a course of oral prednisolone after the onset attack for more than 3 months were associated with decreased risk of relapse, and we noted that patients with ON at onset were less likely to receive a tapering off of oral prednisolone. This may be associated with such patients being seen first by ophthalmologists, who in the UK may be influenced by the Beck study. 26 That study's findings suggested that oral prednisolone may increase the risk of relapse after an attack of ON.
Adjusting for a tapering off of prednisolone (categorized as any prednisolone or a longer course) removed the ON alone vs TM alone effect but not the decreased risk of relapse among individuals with any TM at onset.

JAMA Network Open | Neurology
Factors Associated With Relapse of Myelin Oligodendrocyte Glycoprotein Antibody-Associated Disease In a mixed cohort, patients experiencing relapse with transverse myelitis may be more likely to be referred than patients experiencing relapse with optic neuritis, which could obscure differences in the results. Additionally, we noted variation in racial proportions across age groups, which was not previously studied, to our knowledge. Moreover, a lack of short TM and relapsing TM has not been seen in pediatric groups and relapsing ADEM has not been seen in adults, to our knowledge. It is interesting to note that, in our study, bilateral ON was predominant except among young adults.
Only an incident study can give an unbiased cohort risk of relapse over time. There have been 2 previous incident studies reported, one with 44 patients included in our study with a shorter follow-up (median 15.5 months) 11 and one from the Cobo-Calvo et al study 23  Reports suggesting the effectiveness of corticosteroids and immunosuppression in relapse prevention in smaller groups 11,19,20 were more convincing in a larger study using propensity score matching. The study found that the long-term immunosuppression or immunomodulation treatment group (40 patients treated with azathioprine, mycophenolate, or rituximab) had a decreased risk of relapse compared with 59 patients receiving no treatment, with an HR of 0.41; however, corticosteroid treatment was not included owing to the low number of patients. 9 We found that using prednisolone, IST, or a combination as first-line treatment was associated with a similar decrease in relapse risk, with an HR of 0.51 vs no treatment.

Limitations and Strengths
Our study has some limitations. It included a small number of patients in some of subgroups, and multiple treatments were combined for some patients, making it difficult to separate associations of individual treatments and adjust for indication bias. However, including covariates that may bias relapse risk and using only first-line treatments should help adjust for these biases. Only time to next relapse was analyzed in the study, which reflects the clinical labeling of patients with monophasic or relapsing phenotypes and also the primary outcome in clinical trials. However, future, larger studies should include more complex recurrent event analyses to assess the association of treatment with subsequent relapses. It is important to consider caution in the interpretation of findings when there are multiple interactions and confounders and small numbers of patients, which is why we tested covariates based on factors previously reported to be associated with relapse risk or based on results in this study.
The strengths of our study are that it includes the largest and longest follow-up incident group of patients MOGAD, to our knowledge. We were also able to study differences across a range of age categories separating prepubertal children from adolescents and young adults from older adults in order to investigate relapse risk by age. Additionally, we were able to study the association of prednisolone with relapse prevention, which is a common treatment used in relapsing MOGAD.

Conclusions
We found that MOGAD was associated with distinct clinical presentations, demographics, and relapse risks across age categories with a decreased risk of relapse overall than previously reported.
Our study found that young adults had an increased risk of relapse and that standard first-line immunosuppression or prednisolone were associated with a decrease in the risk of relapse in relapsing patients by nearly one-half. Understanding longer-term relapse risk may help clinicians make a more informed decisions about when to start immunosuppression treatment considering its association with milder disability from relapses compared with AQP4 antibody NMOSD. Currently, the evidence for maintenance therapies for relapse prevention in MOGAD is poor and relies on observational studies, and therefore future prospective randomized clinical trials are crucial to fill this gap.