Article Text
PDF
Abstract
Background Cerebrospinal fluid myelin oligodendrocyte glycoprotein IgG (CSF MOG-IgG) are found in a proportion of patients with MOG antibody-associated disorder (MOGAD) and have been associated with severe disease presentations. However, most studies did not systematically investigate the role of MOG-IgG intrathecal synthesis (ITS).
Methods We retrospectively studied 960 consecutive patients with paired serum and CSF samples screened for MOG-IgG using a live cell-based assays. MOG-IgG-specific antibody index (AIMOG) was systematically calculated using serum and CSF titres to assess MOG-IgG ITS, and clinical features were compared between MOG-IgG CSF+/CSF− and ITS+/ITS− patients.
Results MOG-IgG were found in 55/960 patients (5.7%; serum+/CSF−: 58.2%, serum+/CSF+: 34.5%; serum−/CSF+: 7.3%). Serum/CSF MOG-IgG titres showed a moderate correlation in patients without ITS (ρ=0.47 (CI 0.18 to 0.68), p<0.001), but not in those with ITS (ρ=0.14 (CI −0.46 to –0.65), p=0.65). There were no clinical–paraclinical differences between MOG-IgG CSF+ vs CSF− patients. Conversely, patients with MOG-IgG ITS showed pyramidal symptoms (73% vs 32%, p=0.03), spinal cord involvement (82% vs 39%, p=0.02) and severe outcome at follow-up (36% vs 5%, p=0.02) more frequently than those without MOG-IgG ITS. A multivariate logistic regression model indicated that MOG-IgG ITS was an independent predictor of a poor outcome (OR: 14.93 (CI 1.40 to 19.1); p=0.03). AIMOG correlated with Expanded Disability Status Scale (EDSS) scores at disease nadir and at last follow-up (p=0.02 and p=0.01).
Conclusions Consistently with physiopathology, MOG-IgG ITS is a promising prognostic factor in MOGAD, and its calculation could enhance the clinical relevance of CSF MOG-IgG testing, making a case for its introduction in clinical practice.
- NEUROIMMUNOLOGY
- IMMUNOLOGY
- MULTIPLE SCLEROSIS
Data availability statement
Data are available in a public, open access repository. Anonymised data not published within this article will be made available by request from any qualified investigator. Raw data are available at the Zenodo repository (doi:10.5281/zenodo.10405063).
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
MOG antibody-associated disorder (MOGAD) is diagnosed through serum MOG-IgG detection with cell-based assays. Recent diagnostic criteria have discussed a role for cerebrospinal fluid myelin oligodendrocyte glycoprotein IgG (CSF MOG-IgG) to identify an additional group of patients who are seronegative, although they need further examination and validation as they may lack specificity. CSF MOG-IgG have also been associated with worse disease outcome.
WHAT THIS STUDY ADDS
This study shows that, in MOGAD, MOG-IgG intrathecal synthesis might be a more accurate disease outcome predictor than CSF MOG-IgG detection. A correct interpretation of the presence of MOG-IgG in the CSF should take into account the inflammation-dependent permeability of the blood-CSF-barrier.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
The results of our study further support the role of MOG-IgG measurement in the CSF samples for MOGAD diagnosis and disease course prediction. Titration of MOG-IgG in serum and CSF samples wit ITS calculation could prove useful for clinicians to orient treatment strategies early in the disease. Future studies on CSF MOG-IgG should consider protocol standardisation among different centres.
Introduction
MOG antibody-associated disorder (MOGAD) is a demyelinating disease of the Central Nervous System (CNS) commonly associated with optic neuritis and transverse myelitis in adults or acute disseminated encephalomyelitis (ADEM) in children, within a clinical spectrum that includes heterogeneous and rarer phenotypes, and monophasic or relapsing courses.1–4 MOG-IgG play a pivotal diagnostic role, and serum is the recommended specimen to assess their presence, as these antibodies are believed to originate peripherally5 6.
MOG-IgG can also be detected in the cerebrospinal fluid (CSF) of patients with demyelinating syndromes, either in association with serum MOG-IgG or, in rare cases, exclusively in the CSF.7–9 This latter finding was suggested to increase diagnostic sensitivity and was included in the recently published diagnostic criteria for MOGAD.6 Even though these criteria emphasise the importance of serum MOG-IgG detection with cell-based assays (CBAs), and their quantitative measurement to improve specificity and positive predictive value (PPV),6 they also allow to make a diagnosis of MOGAD in CSF-only positive patients in the presence of supporting features. However, CSF MOG-IgG have also been reported in patients with multiple sclerosis (MS),10 and in other inflammatory neurological disorders, such as bacterial meningitis.11 For this reason, MOG-IgG CSF testing is only recommended in seronegative patients with a phenotype highly suggestive of MOGAD and no alternative diagnosis.6
The presence of MOG-IgG in the CSF, with or without coexisting seropositivity, has been associated with ADEM and cortical encephalitis MOGAD phenotypes,9 and with a more severe disease course.8 However, CSF MOG-IgG positivity has been often analysed independently of both serum titres and inflammation-related dysfunction of the blood-CSF-barrier (B-CSF-B). This does not allow a distinction between intrathecal production of antibodies by CNS infiltrating cells of the B lymphocyte lineage (intrathecal synthesis, ITS) and a serum-to-CSF passive transfer of antibodies, driven by a gradient of concentration. Interestingly, CSF-restricted MOG-IgG positivity has also been associated to poor prognosis in MOGAD patients.8
In this study, we aimed to evaluate the clinical relevance of MOG-IgG testing on paired serum and CSF samples, with a specific focus on ITS.
Methods
We included consecutive patients with suspected CNS demyelinating conditions, and paired serum and CSF samples sent to the Pavia Neuroimmunology Laboratory (n=960), from July 2019 to January 2023. Samples were collected during both remission and active phases, with the inclusion of one time point sample for each patient.
Samples were all tested on arrival using a live CBA, as previously described.12 Screening for total MOG-IgG was performed at 1:20 dilution in serum, and 1:5 in the CSF. When positive, the samples were also tested with an IgG1-specific secondary antibody. All samples positive at screening dilution underwent MOG total IgG endpoint titration with twofold dilutions.12 Samples with MOG-IgG titres ≥1:160 (serum) and ≥1:5 (CSF) were considered positive. MOGAD diagnosis was assessed according to published criteria.6 To this purpose, serum titres ≥1:640 were considered ‘clear positives’ and required at least one ‘core feature’ to define the diagnosis of MOGAD, whereas titres between 1:160 and 1:320 were considered ‘low positives’ and required additional supporting criteria.6 Patients were classified as being MOG-IgG positive in serum and CSF (serum+/CSF+), in serum only (serum+/CSF−), and in CSF-only (serum−/CSF+).
For MOG-IgG-positive patients, detailed clinical, radiological and laboratory information was gathered. Disability at onset and follow-up was measured with the Expanded Disability Status Scale (EDSS). A poor outcome was defined as an EDSS ≥6. We also considered an EDSS ≥3 as a further disability cut-off, to reproduce earlier studies on the subject.8 Detailed CSF analysis results were collected when available.
In patients with CSF+MOG IgG, we calculated the following parameters: albumin quotient (QAlb, CSF/serum albumin ratio), IgG quotient (QIgG-total, CSF/serum total IgG ratio), Reiber’s empiric hyperbolic function (Qlim), MOG-IgG-specific quotient (QMOG, CSF/serum MOG-IgG titres ratio), and finally MOG-IgG-specific antibody index (AIMOG). The latter was used to assess the ITS of MOG-IgG. AIMOG was calculated as the ratio between QMOG and QIgG, similarly to previous reports for AQP4-IgG13 and MOG-IgG.14 To further increase the detection of intrathecally synthesised antibodies, we applied corrected AI, which considers the functioning and integrity of the B-CSF-B. To this end, Reiber’s empiric hyperbolic function Qlim was used to prevent underestimation of ITS due to alterations of the B-CSF-B function.15 16 For antibody measurements expressed as concentrations or optical density values, reference range of the AI is 0.6−1.3, and an AI ≥1.5 indicates ITS. However, when considering the imprecision inherent to the titration procedure, the cut-off of AI >4 is recommended, and, as such, was used in this study as well as in previous studies on AQP4-IgG and MOG-IgG.9 15 16 Detailed formulae for the above-mentioned parameters are found in online supplemental methods. Individual patients were represented as dots on Reibergrams, graphical representations that allow the recognition of disease-related immunoglobulin patterns at a glance. Our Reibergrams were adapted to show the relation between QAlb and QMOG. To contextualise our results, we performed an aggregated analysis of the literature, last updated on 31 January 2024, including published studies on the topic. To this end, we searched the Pubmed library for reports describing case series of patients with CSF MOG-IgG through the following research string for titles: (“Myelin Oligodendrocyte Protein” OR “MOG” OR “MOGAD” OR “MOG-IgG”) AND (“CSF” OR “Cerebrospinal Fluid” OR “Cerebrospinal Fluids”). We retrieved 17 papers. After excluding commentaries/reviews, single case reports and cohort studies without sufficient clinical data, we eventually included six full papers.7–11 14
Supplemental material
Statistical analysis
Quantitative variables were reported as medians (range), and categorial variables as percentages. Subgroup comparisons were performed with parametric (χ2 test) or non-parametric (Mann-Whitney, Fisher’s exact) tests, as appropriate. Titre differences in paired serum-CSF samples were analysed with the Wilcoxon signed-rank test. Correlation analyses were performed with Spearman’s rank correlation coefficient. The relationship between AIMOG and EDSS was assessed through univariate linear regression. The role of MOG-IgG ITS as predictor of a severe outcome was analysed through a multivariate logistic regression model built by adding significant variables from the univariate analysis. All analyses employed Stata/IC 14.0 for Mac (64-bit, StataCorp, College Station, Texas) and GraphPad Prism (V.9.0, GraphPad Software, La Jolla, California). Values of p<0.05 were considered significant.
Standard protocol approvals, registrations and patient consents
The project was approved by the Institutional Review Board of the IRCCS Policlinico San Matteo, Pavia (project codes: 0020308/23). All patients included in the study provided their informed consent.
Data availability
Anonymised data not published within this article will be made available by request from any qualified investigator. Raw data are available at the Zenodo repository (doi:10.5281/zenodo.10405063).17
Results
Serum and CSF MOG-IgG titres, intrathecal MOG-IgG synthesis and IgG1 subclass
In our cohort of 960 patients, 55 (5.7%) were MOG-IgG-positive in at least one specimen (serum ≥1:160, CSF ≥1:5 or both) (figure 1A).
Study design flowchart and detailed serostatus of the study population. (A) Study flowchart. (B) Pie charts describing serum/cerebrospinal fluid positivity and presence of intrathecal synthesis. CSF, cerebrospinal fluid; MOG, myelin oligodendrocyte glycoprotein; MOGAD, MOG antibody-associated disorder; ITS, intrathecal synthesis.
MOG-IgG were positive exclusively in the serum in 32/55 patients (serum+/CSF−, 58.2%), in serum and CSF in 19/55 (serum+/CSF+, 34.5%) and in CSF-only in 4/55 (serum−/CSF+, 7.4%). (figure 1B). Three of serum−/CSF+MOG IgG patients had detectable, but below the cut-off serum MOG-IgG (1:20, 1:40 and 1:80).
Median titres were 1:1280 (range 1:20–1:40940) in serum, and 1:20 (range 1:5–1:320) in the CSF. In patients with detectable MOG-IgG in both serum and CSF, serum titres were always higher than CSF titres (1:1280 vs 1:20, p<0.001), except for two patients (figure 2A–C). B-CSF-B dysfunction (QAlb above the reference value of 0.7%) was found in 16/55 patients (29.1%; median: 1.39%; range: 0.88%–5.00%), more frequently in patients with CSF MOG-IgG (11/23; 48%) than in those with serum-restricted MOG-IgG (5/32; 16%; p<0.001). Among the 11 CSF MOG-IgG patients with B-CSF-B dysfunction, 2 were serum−/CSF+ and 9 were serum+/CSF+. Patients with serum-restricted MOG-IgG and barrier dysfunction had low serum titre (median 1:640, range 1:160–1:640) and mild barrier dysfunction degree (mean QALB values, 1.27%, range 0.93–2.33).
Serum and CSF titres correlation and intrathecal synthesis. (A) Paired analysis of titres in serum and CSF for all patients, (B) for patients with and (C) for patients without MOG-IgG ITS. (D) Correlation of serum and CSF titres for all patients, (E) for patients with and (F) for patients without MOG-IgG ITS. (G) Reibergram showing CSF+patients as dots and their relationship with barrier damage and intrathecal synthesis (H) Scatterplot of serum and CSF titres in MOGAD patients on the Reibergram, showing antibody index values for each patient. Values in the top left corner indicate a CSF origin (ie, intrathecal production). Dots are patients, titres are expressed as log10 for graphical purposes, the red line indicates an AI of 4. Asterisk in (A)–(C) denotes significant differences among groups (p<0.05). In (D)–(F), blue-dotted lines represent negative samples (titre value=0), while red-dotted lines represent the cut-off for positivity (≥1:160 for serum and ≥1:5 for CSF). In G, blue dotted line represents Qalb reference cut-off for barrier dysfunction (0.7%). AI, antibody index; CSF, cerebrospinal fluid; ITS, intrathecal synthesis; MOG, myelin oligodendrocyte glycoprotein; MOGAD, MOG antibody-associated disorder.
MOG-IgG ITS was found in 13/23 patients (56%; figure 1B; online supplemental table 1). This count also included the only patient with MOG-IgG undetectable in serum, but detectable in the CSF at 1:5 dilution. Median AIMOG was 4.6 (range: 0–141.6). Serum and CSF MOG-IgG titres showed a significant correlation in patients without MOG-IgG ITS (Spearman’s ρ: 0.47 (CI 0.18 to 0.68), p<0.001) but not in patients with MOG-IgG ITS (ρ=0.14 (CI −0.46 to 0.65), p=0.65) (figure 2D–F). Patients were represented as individual dots on a scatterplot and a Reibergram, showing a visual relation between QMOG, B-CSF-B damage expressed by the QAlb, and AIMOG values (figure 2G,H).
All positive samples were tested for the IgG1 subclass, and MOG-IgG1 were positive in at least one specimen in 47/55 patients (85%). Serum MOG-IgG1 were positive in serum+ MOG IgG samples (45/51; 88.2%) more frequently than in CSF+ (10/23; 43.5%; p<0.001). CSF MOG-IgG1 were positive in two out of the four with CSF-only MOG-IgG (online supplemental figure 1A). Among IgG1-negative patients, subclass testing revealed one IgG2 and one IgG3-positive patient.
When considering all samples with detectable serum MOG-IgG, median serum titres in MOG-IgG1-positive patients were higher (1:1280; range, 1:160-1:40960) than in MOG-IgG1 seronegative (1:160; range, 1:20–1:5120; p<0.001). Similarly, median CSF MOG-IgG titres in CSF-IgG1 positives were higher than in negatives (1:40; range, 1:10-1:320 vs 1:10; range, 1:2−1:20; p<0.001; online supplemental figure 1B).
Diagnostic relevance of CSF MOG-IgG
Among the 55 MOG-IgG positive patients, 3 did not have sufficient clinical information to assess MOGAD diagnosis, and were excluded from further analysis (figure 1A). MOGAD diagnosis was not confirmed in 2/29 patients with serum-only MOG-IgG that were diagnosed as MS (one with 1:5120 titre, IgG1 positive; one with 1:160 titre, IgG1 negative). Both patients had multiple white matter lesions in brain and spinal cord suggestive of MS (coexistence of T2 hyperintense lesions and T1 hypointense lesions) with no features suggestive of MOGAD such as fluffy margins or medullary H-sign, and had CSF-only oligoclonal bands. Finally, one of the four patients with CSF-only MOG-IgG was diagnosed with non-arteritic anterior ischaemic optic neuropathy (NAION). This sample was negative when tested for MOG-IgG1. All patients positive for both serum and CSF MOG-IgG had confirmed MOGAD.
Clinical features of patients with MOGAD according to CSF MOG-IgG versus intrathecal MOG-IgG synthesis
Among the 49 patients with sufficient clinical information and confirmed MOGAD, 29 (59%) were women, and 21 (43%) were paediatric. The most common clinical presentations were optic neuritis and myelitis, both found in 35% of patients, followed by ADEM (18%) and brainstem syndrome (14%). Detailed clinical information is reported in table 1.
Detailed clinical and paraclinical features in patients with MOG-IgG according to CSF positivity and intrathecal synthesis.
Samples were collected at disease onset in 42 patients (86%), during a relapse in 2/49 (4.1%) or during remission phases in 5/49 (10.2%). Median time from disease onset to sampling was 21 days (IQR 6–49 days). Samples at onset were always collected before the administration of acute phase treatments (such as intravenous steroids, plasma exchange (PLEX) or intravenous immunoglobulins (IvIg)). All patients whose samples were collected at relapse or remission did not receive immunosuppressive treatments in the 3 months prior to sample collection. Median follow-up time was 12 months (IQR 4–18.5).
When comparing clinical and paraclinical features of 28/49 patients (57%) with serum+/CSF− MOG-IgG versus the 21/49 (43%) with CSF+MOG IgG (either with, or without serum MOG-IgG), we found no significant differences (figure 3, table 1). Among the three patients with CSF-restricted MOG-IgG, one presented with a longitudinally extensive transverse myelitis (LETM) followed by a bilateral optic neuritis, one with an encephalo-myeloradiculitis, and one with a myeloradiculitis with leptomenigeal enhancement on spinal cord MRI. In both patients with myeloradiculitis, the Peripheral Nervous System (PNS) involvement coexisted with the myelitis and was supported by MRI evidence of spinal root contrast enhancement and by nerve conduction studies findings of increased F wave latencies. The phenotype was consistent with that of other reported cases.18
Clinical-paraclinical features in patients with MOG-IgG according to CSF positivity and intrathecal synthesis. Left: patients with or without CSF MOG-IgG; right, patients with or without MOG-IgG intrathecal synthesis. Asterisk denotes significant differences among groups (p<0.05). CSF, cerebrospinal fluid; ADEM, acute disseminated encephalomyelitis; MOG, myelin oligodendrocyte glycoprotein.
We then compared clinic-radiological features in 11/49 patients (22%) with versus 38/49 (78%) without MOG-IgG ITS. We found that patients with MOG-IgG ITS presented more often with pyramidal symptoms (73% vs 32%; p=0.03) and had more frequent spinal cord involvement (82% vs 39%; p=0.02), particularly in the cervical region (73% vs 21%; p=0.003) on MRI scans (figure 3). Of note, only 3% of patients without MOG-IgG ITS had a severe outcome at follow-up (EDSS ≥6), compared with 36% of patients with MOG-IgG ITS (p=0.007; figure 3). All patients received intravenous steroids during the acute attack, and patients with MOG-IgG ITS were more often treated with additional IvIg or PLEX cycles (4/11 vs 3/38, p=0.036), whereas there was no difference among MOG-IgG CSF+ versus CSF− patients (5/21 vs 2/28, p=0.12). The presence of MOG-IgG ITS predicted a poor outcome (EDSS ≥6) in the univariate logistic regression model (OR: 10.28 (CI 1.57 to 67.44); p=0.015); another predictor of a severe outcome was the presence of an LETM at onset (OR: 13.33 (CI 1.83 to 96.98); p=0.01). Both MOG-IgG ITS (OR: 14.93 (CI 1.40 to 159.19); p=0.025) and LETM at onset (OR 20.43 (CI 1.60 to 261.23); p=0.02) were still significant in the multivariate model (figure 4A; table 2), even when adjusting the multivariate regression for age and sex (p=0.035 and p=0.03, respectively).
Univariate and multivariate logistic regression results for predictors of a severe outcome
Intrathecal MOG-IgG synthesis, MOG-IgG Antibody Index and disability. (A) The forest plot shows that MOG-IgG ITS, as well as LETM at onset, was an independent predictor of long-term poor outcome (defined as EDSS≥6 at last follow-up) in a multivariate logistic analysis. (B, C) In patients with MOG-IgG ITS, AIMOG values significantly correlated with EDSS scores at disease nadir (B) and (C) at last follow-up. In (B) and (C), the red dot represents the only patient with undetectable serum MOG-IgG; we attributed to this patient the arbitrary serum titre of 1:10, which corresponds to an AI value of 186.8. AIMOG, MOG-IgG Antibody Index; EDSS, Expanded Disability Status Scale; ITS, intrathecal synthesis; LETM, longitudinally extensive transverse myelitis; MOG, myelin oligodendrocyte glycoprotein.
We built an additional logistic regression model to test a different measure of poor outcome (EDSS ≥3); here, MOG-IgG ITS predicted a poor outcome in the univariate (OR: 4.44 (CI 1.02 to 19.38); p=0.04), but not in the multivariate model (OR: 3.28 (CI 0.64 to 16.86); p=0.15). In addition, in patients with MOG-IgG ITS, AIMOG values correlated both with EDSS at the nadir of the disease (ρ=0.70, p=0.02) and at last follow-up (ρ=0.75, p=0.01) (figure 4B,C). In linear regression analysis, increases in AIMOG values were associated with higher EDSS values at the end of follow-up (R2=0.37, p=0.04). There was no significant follow-up length difference among patients with and without severe disease outcome (9.8 vs 12.9 months; p=0.54). None of the patients with severe disease outcome had a relapsing disease course.
Literature review and aggregate analysis on CSF MOG-IgG
Since one out of four of patients with CSF-only MOG-IgG in our cohort was ultimately diagnosed with a non-MOGAD condition, we also performed an aggregate analysis of our data and those extrapolated from six published studies addressing CSF-only MOG-IgG diagnostic relevance and laboratory methods used for their detection.7–11 14 Results are summarised in table 3. Of note, none of the papers was published after the 2023 MOGAD diagnostic criteria, and, therefore, MOGAD diagnosis was performed according to each study established criteria.
Aggregate data from the main case series reporting findings on CSF MOG antibodies
The proportion of CSF MOG-IgG positivity, either serum co-occurring, or CSF-restricted was highly variable among cohorts, ranging from 34.5% to 69.6%, and from 3.4% to 29.5%, respectively. Among the 90 patients with CSF-restricted MOG-IgG, data were available for 89: 12/89 (13.4%) had a non-MOGAD diagnosis (MS, 8; pneumococcal meningitis, 1; Susac syndrome, 1; acute polyradiculoneuritis, 1; NAION, 1). When considering only cohorts including consecutive patients sent for routine diagnostics (2879 patients), serum MOG-IgG positivity, according to each laboratory cut-off, had a PPV of 99.40% (CI 97.66% to 99.85%), while CSF-restricted MOG-IgG positivity had a PPV of 83.64% (CI 71.61% to 91.19%).
All studies tested CSF MOG-IgG using a live CBA, four of them using a total IgG secondary antibody (anti-Fc or anti-heavy and light chain; H-L), two an IgG1 secondary antibody, and one both anti-Fc and anti-IgG1.
Discussion
In this study, we explored the relevance of MOG-IgG testing in CSF in a large cohort of consecutive paired samples sent for MOG-IgG detection. We focused our investigation on the relationship between serum and CSF titres and the determination of MOG-IgG ITS. We found that MOG-IgG ITS, but not MOG-IgG presence in the CSF, is associated with a more aggressive disease phenotype and, importantly, can predict a worse outcome.
Several reports identified a potential role for CSF MOG-IgG testing, suggesting that their presence is often associated with encephalitic phenotypes and with severe forms of MOGAD with predominant spinal cord involvement.7–10 19 However, most of these studies only considered the presence or absence of MOG-IgG in the CSF.7 8 10 Under physiological conditions, IgG are not produced within the CNS. Therefore, IgG can be found in the CSF in three main and often overlapping scenarios: (a) diffusion of serum IgG through the B-CSF-B, at a gradient that is estimated between 1:200 and 1:40015 (b) enhanced transfer of serum IgG through a leaky B-CSF-B, which can show increased permeability in inflammatory conditions and (c) ITS of specific IgG within the CNS (figure 5). The latter phenomenon requires pathological immune changes that ultimately compartmentalise the immune reaction within the CNS, as it has been described in many autoimmune conditions such as MS.20 Hence, to investigate this relevant biological phenomenon, specific formulae that consider the relationship between QAlb and QMOG were used. These relationships were appropriately depicted by a Reibergram.16 This is particularly relevant for MOG-IgG, as previous studies showed that acute phase serum titres do not seem to be clinically or prognostically relevant.5 21–23 In our study, we confirmed these theoretical assumptions for MOG-IgG, as correlation between serum and CSF titres was only identified in patients without MOG-IgG ITS. In addition, patients with CSF MOG-IgG more frequently had an altered B-CSF-B compared with serum only MOG-IgG patients. It is, therefore, likely that MOG-IgG detection in CSF can be attributed in a proportion of cases to an increased diffusion through a leaky B-CSF-B. On the other hand, not all patients with an altered B-CSF-B necessarily present CSF MOG-IgG. Our knowledge of MOG-IgG CSF spillover is currently limited to the degree of B-CSF-B dysfunction and serum titres. For these reasons, we believe that future studies on the topic should, therefore, focus on MOG-IgG ITS rather than CSF positivity.
Possible scenarios underlying MOG-IgG presence in serum and CSF according to blood-CSF barrier permeability. (A–C) are conditions where there is no MOG IgG ITS, while in (D–F) CNS resident plasma cells are responsible for MOG-IgG ITS. In the presence of an intact B-CSF-B (A–B), MOG-IgG cross the B-CSF-B according to the IgG gradient. When serum titres are low (A), MOG-IgG cannot be detected in the CSF, while higher serum titres (B) lead to a detectable amount in CSF. In the presence of a leaky B-CSF-B (C, depicted in red), MOG-IgG derived from passive diffusion can be detectable in the CSF even when serum titre is low. On the other hand, MOG-IgG can be synthetised intrathecally and be found in the CSF, either with (D) or without (E) concomitant serum MOG-IgG. This can be observed also with a leaky B-CSF-B (F). B-CSF-B, blood-cerebrospinal fluid-barrier; CNS, central nervous system; MOG, myelin oligodendrocyte glycoprotein. Created with BioRender.com.
From a clinical point of view, consistently with laboratory findings, we showed that only MOG-IgG ITS, and not CSF positivity in itself, is associated with specific clinical features in MOGAD patients, such as a more frequent spinal cord involvement and a severe presentation.
More importantly, MOG-IgG ITS was an independent predictor of poor outcome in our multivariate model, and AIMOG correlated with EDSS both at disease nadir and at last follow-up. This is consistent with a previous study where patients with CSF-restricted MOG-IgG (indicative of ITS) showed a poor outcome, but no systematic ITS calculation was performed.8 In another study where AIMOG was calculated, this was correlated with EDSS at presentation.9 In other CNS autoimmune disorders such as autoimmune encephalitis with Leucine-rich Glioma Inactivated 1 (LGI1) and N-methyl-D-aspartate (NMDAR) antibodies, ITS has been similarly associated with greater disease severity.24 25
MOGAD is widely considered a disease with more favourable outcome compared, for example, to Neuromyelitis Optica Spectrum Disorder (NMOSD), and patients can show a dramatic clinical improvement with treatment even after severe attacks.3 26–28 However, approximately 50%–60% patients develop neurological disability, which is mostly acquired during the first attack.27 29 30 In this scenario, most studies failed to identify clinically relevant prognostic biomarkers.3 28 31 Our results strengthen the prognostic relevance of CSF MOG-IgG and suggest that routine MOG-IgG titration in serum and CSF with ITS calculation could be a useful tool for clinicians to orient treatment strategies early in the disease course.8 Our findings are preliminary and need further validation in larger cohorts, but could support CSF testing of MOG-IgG on a larger scale, as the ceiling for MOG-IgG ITS as a predictor of disease severity and outcome is high.
Notably, our results show that LETM at disease presentation was an independent predictor of poor disease outcome, in analogy with other reports, such as one on a large cohort from the Oxford group, which associated transverse myelitis at onset with long-term disability.3
Other studies reported an association between CSF-restricted MOG-IgG and encephalitic (either ADEM or cortical encephalitis, CE) or spinal cord involvement, highlighting that no patient with this finding had isolated optic neuritis (ON).9 10 This was confirmed in our study, as all patients with CSF restricted MOG-IgG had either brain or spinal cord involvement. This is significant, as ON is the most common presenting phenotype in adults with MOGAD, further supporting the clinical relevance of CSF MOG-IgG. However, we did not identify different proportions of ADEM or CE when comparing patients with or without ITS, although this could be due to the small number of patients with these diagnoses in our study (11/49, 22%).
As for the use of CSF-restricted MOG-IgG for diagnostic purposes, the aggregate literature analysis, which includes the results of this study, shows that this finding can also occur in non-MOGAD patients, in a highly variable proportion among studies. Therefore, the overall PPV of CSF-restricted MOG-IgG seems to be lower than those of serum MOG-IgG and suggests caution in requesting CSF testing in patients with low pre-test probability.6 As a limitation, it must be considered that our aggregate analysis might have included duplicate records due to multiple articles published on the same cohort, possibly leading to a bias in our interpretation.
Furthermore, our aggregate literature analysis showed that the methods used to detect CSF MOG-IgG are highly heterogeneous, and variable results in terms of PPV are a likely reflection of this issue. Previous studies suggested that higher CSF MOG-IgG titres improved PPV, similarly to serum titres,10 32 and using an IgG1 secondary antibody for serum MOG-IgG detection further improves PPV.12 Conversely, no study so far compared the performances of different secondary antibodies (total MOG IgG vs IgG1), in the CSF. Besides, cumulative data from published studies showed that the use of IgG1 secondary antibodies was associated with a higher PPV. In our study, we observed that up to 43.5% of CSF MOG-IgG was IgG1 negative, and that this mostly occurred in patients with very low titres. However, we were not able to assess the diagnostic relevance of CSF titres or subclasses due to the low number of patients included. Interestingly, our only CSF MOG-IgG false-positive patients had a low titre and was IgG1 negative. Moreover, despite less than 40% of patients in our cohort showing CSF MOG-IgG1 positivity, two out of four CSF-restricted patients did show IgG1 positivity. This data could also be related to technical issues in CSF MOG-IgG testing, such as the sensitivity of the secondary antibodies used, further highlighting the need for standardisation studies to evaluate the best laboratory strategy for CSF MOG-IgG detection.
Our data suggest another relevant finding, as we found that among MOG-IgG false-positive patients, none had both serum and CSF MOG-IgG positivity combined. This suggests that the risk of false MOG-IgG positivity might be lower in patients with combined serum and CSF positivity.
Notably, as a limitation, all studies considered in our analysis were precedent to the publication of MOGAD diagnostic criteria, and the rate of ‘true’ and ‘false’ positives was assessed according to the criteria established by each study.
Our study has limitations. First, our cohort was retrospective and included only patients for which both serum and CSF were available, which represented 44.5% of those referred to our laboratory during the study period, and this might have introduced a selection bias. In addition, for practical reasons, we used a 1:5 dilution for CSF MOG-IgG testing, as this allowed to perform the assay with small amounts of sample. This might have prevented us to identify patients with lower CSF MOG-IgG titres and might partially explain differences in the proportion of CSF MOG-IgG positives compared with other studies. Unlike serum, there is no accepted cut-off value for CSF MOG-IgG positivity in MOGAD, and future studies should address this issue. Finally, the follow-up for some of our patients was short, with 46% of patients having a follow-up <1 year. This might have prevented us to detect relevant differences in long-term prognosis and future relapse risk. Moreover, the EDSS scale is less sensitive towards visual-related disability, that is, serious loss of vision due to optic neuritis. However, none of our patients with optic neuritis experienced severe, persistent visual loss after acute events resolved. Although most samples were collected during active phases of the disease, a small percentage was collected during remission. It is, therefore, impossible for us to determine whether in those patients MOG-IgG ITS might have been transiently present in earlier disease stages. Finally, our sample size is smaller than other studies investigating the role of CSF MOG-IgG; however, as the laboratory analysis was performed in a single centre, this allowed a high degree of consistency and standardization in the assays and their results.
In conclusion, we have shown that MOG-IgG ITS could be a promising clinical and prognostic marker in patients with MOGAD. Our findings could support MOG-IgG CSF testing and titration in all patients with suspected MOGAD. The routine calculation of AIMOG could represent a useful tool to orient early treatment strategies. A small proportion of MOGAD patients can be identified only after CSF testing, but CSF-restricted MOG-IgG positivity can also occur in non-MOGAD patients. Future studies should investigate further the relevance of MOG-IgG ITS as a prognostic marker and assess the best protocols for CSF MOG-IgG detection.
Data availability statement
Data are available in a public, open access repository. Anonymised data not published within this article will be made available by request from any qualified investigator. Raw data are available at the Zenodo repository (doi:10.5281/zenodo.10405063).
Ethics statements
Patient consent for publication
Ethics approval
The project was approved by the Institutional Review Board of the IRCCS Policlinico San Matteo, Pavia (project code: 0020308/23). All patients included in the study provided their informed consent. Participants gave informed consent to participate in the study before taking part.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Collaborators The PNI (Neuroimmunology Platform Group): Roberto Bergamaschi, Enrico Marchioni, Elisa Vegezzi, Paola Bini, Eleonora Tavazzi, Lara Ahmad, Fortuna Ricciardiello, Laura Giordano, Antonella Toriello, Cristian Sorrentino, Alice Passarini, Pierluigi Smilari, Monica Anna Maria Lodi, Antonio Lucia Spiezia.
Contributors GG and MG had full access to all of the data of the study and MG is responsible for the overall content as guarantor. GG, MG and DF contributed to the conception and design of the study, and to statistical analyses. All authors contributed to acquisition, analysis and interpretation of the data. Authors of the PNI Study Group have contributed to acquisition, analysis and interpretation of the data. GG, DF and MG contributed to drafting the text and preparing the figures. All authors contributed to the critical review of the manuscript for important intellectual content. MG obtained funding and supervised GG.
Funding This project was funded by the Ministero della Salute - Ricerca Corrente 2022-2024 (RRC2024-23684441) grant to the IRCCS Mondino Foundation
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer-reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.