Optical coherence tomography in myelin-oligodendrocyte-glycoprotein antibody-seropositive patients: a longitudinal study

Background Serum antibodies against myelin-oligodendrocyte-glycoprotein (MOG-IgG) are detectable in a proportion of patients with acute or relapsing neuroinflammation. It is unclear, if neuro-axonal damage occurs only in an attack-dependent manner or also progressively. Therefore, this study aimed to investigate longitudinally intra-retinal layer changes in eyes without new optic neuritis (ON) in MOG-IgG-seropositive patients. Methods We included 38 eyes of 24 patients without ON during follow-up (F/U) [median years (IQR)] 1.9 (1.0–2.2) and 56 eyes of 28 age- and sex-matched healthy controls (HC). The patient group’s eyes included 18 eyes without (EyeON-) and 20 eyes with history of ON (EyeON+). Using spectral domain optical coherence tomography (OCT), we acquired peripapillary retinal nerve fiber layer thickness (pRNFL) and volumes of combined ganglion cell and inner plexiform layer (GCIP), inner nuclear layer (INL), and macular volume (MV). High-contrast visual acuity (VA) was assessed at baseline. Results At baseline in EyeON-, pRNFL (94.3 ± 15.9 μm, p = 0.36), INL (0.26 ± 0.03 mm3, p = 0.11), and MV (2.34 ± 0.11 mm3, p = 0.29) were not reduced compared to HC; GCIP showed thinning (0.57 ± 0.07 mm3; p = 0.008), and VA was reduced (logMAR 0.05 ± 0.15 vs. − 0.09 ± 0.14, p = 0.008) in comparison to HC. Longitudinally, we observed pRNFL thinning in models including all patient eyes (annual reduction − 2.20 ± 4.29 μm vs. − 0.35 ± 1.17 μm, p = 0.009) in comparison to HC. Twelve EyeON- with other than ipsilateral ON attacks ≤ 6 months before baseline showed thicker pRNFL at baseline and more severe pRNFL thinning in comparison to 6 EyeON- without other clinical relapses. Conclusions We observed pRNFL thinning in patients with MOG-IgG during F/U, which was not accompanied by progressive GCIP reduction. This effect could be caused by a small number of EyeON- with other than ipsilateral ON attacks within 6 months before baseline. One possible interpretation could be a reduction of the swelling, which could mean that MOG-IgG patients show immune-related swelling in the CNS also outside of an attack’s target area. Electronic supplementary material The online version of this article (10.1186/s12974-019-1521-5) contains supplementary material, which is available to authorized users.

optic neuritis (ON) enhanced inflammation and induced demyelination upon transfer into experimental animals indicating the pathogenic potential of MOG-IgG detected in the blood of these patients [7]. It is discussed whether MOG-IgG define a separate disease entity tentatively called MOG-IgG-associated diseases, MOG-IgG autoimmunity or MOG-IgG seropositive encephalomyelitis rather than being part of several autoimmune disorders, especially neuromyelitis optica spectrum disorders (NMOSD) [1,3,8,9]. However, the bouquet of clinical phenotypes in MOG-IgG-associated diseases at clinical onset is not easy to differentiate and overlaps with aquaporin-4-IgG (AQP4-IgG)-seropositive NMOSD and in rare cases with MS [2,[10][11][12], although distinct clinical features such as seizures have been described [13][14][15]. ON is the most common manifestation and can lead to substantial neuro-axonal damage after multiple relapses, as shown in different cohorts [11,16]. The pattern of retinal degeneration after ON seems to be similar in all MOG-IgG-seropositive cohorts as shown by optical coherence tomography (OCT) studies [11,16]. OCT proved to be a precise and reproducible method for non-invasive visualization and quantification of retinal layers and plays a crucial role in analyzing retinal changes in various neuroinflammatory disorders [17][18][19][20]. In a cross-sectional study, MOG-IgG-related OCT features indicated subclinical pathology in eyes without a history of ON (Eye ON-) [16]. However, no longitudinal OCT data is reported in MOG-IgG-associated diseases so far and the pattern of longitudinal retinal damage still remains elusive. Using OCT, we assessed retinal layer thinning as a marker of neuro-axonal damage in a cohort of MOG-IgGseropositive patients without ON during follow-up (F/U). We aimed to investigate at baseline and longitudinally microstructural changes in MOG-IgG-seropositive patients, extending previous work in AQP4-IgG-seropositive NMOSD [21,22].

Optical coherence tomography
All centers used SPECTRALIS spectral-domain OCT (Heidelberg Engineering, Heidelberg, Germany) with automatic real-time (ART) function for image averaging. We acquired peripapillary retinal nerve fiber layer thickness (pRNFL) and volumes of combined ganglion cell and inner plexiform layer (GCIP), inner nuclear layer (INL) and macular volume (MV) by OCT. GCIP, INL and MV were calculated as a 3 mm diameter cylinder around the fovea from a macular volume scan (25°× 30°, 61 vertical B-scans, 12 ≤ ART ≤ 18; 20°× 20°, 25 vertical B-scans, 27 ≤ ART ≤ 49). The peripapillary RNFL (pRNFL) was measured with activated eye tracker using ring scans around the optic nerve (12°, 1536 A-scans, 57 ≤ ART ≤ 100) or the most inner ring of a star-and-ring scan around the optic nerve (12°, 768 A-scans, 27 ≤ ART ≤ 33). For two patients (8.3%), the ring scan protocol changed during the acquisition period (ring scan to inner ring of a star-and-ring scan). Segmentation of all layers was performed semi-automatically using software provided by the OCT manufacturer (Eye Explorer 1.9.10.0 with viewing module 6.3.4.0, Heidelberg Engineering, Heidelberg, Germany). Experienced raters (BK for TU Munich data, JH for all other data) carefully checked all scans for sufficient quality and segmentation errors and corrected if necessary. OCT data in this study is reported and analyzed according to the APOSTEL and OSCAR-IB recommendations [24,25]. Macular microcysts were defined as the presence of cystic lesions on at least one scan detected by experienced raters (BK for TU Munich scans, JH for all other scans). Additionally, we collected habitually corrected monocular high-contrast visual acuity (VA) using ETDRS (Early Treatment Diabetic Retinopathy Study) charts at baseline in 20 ft distance for a subset of patients (N = 15).

Statistical methods
Group differences between MOG-IgG patients and HC were tested by chi-squared test for sex and Wilcoxon rank-sum test for age. Main outcomes were change of GCIP, pRNFL, INL and MV and VA over F/U. Crosssectional differences of OCT values and VA between all groups were analyzed pairwise by generalized estimating equation (GEE) models to account for inter-eye withinpatient correlations of monocular measurements. Longitudinal analyses of OCT and VA were performed with linear mixed effects models using time from baseline and group as fixed effects and patient-ID and eye-ID as random effects; results are reported for effect "Time from Baseline * Group", which reflects the groupspecific change over time. Annual loss was estimated for each individual as change to baseline at last visit divided by F/U time in years. All tests and graphical representations were performed with R version 3.3.1 [http://www. R-project.org]. Statistical significance was established at p < 0.05, and all results were interpreted in the context of an exploratory analysis and therefore not adjusted for multiple comparison.

Cohort description and follow-up
We included 38 eyes of 24 patients without ON during F/U. 70% of the patients from Berlin [7/10] and 64% of the patients from LMU Munich [7/11] have been included in previous cross-sectional studies [7,10,16]. MOG-IgG-seropositive patients had the following diag-  9)). From the 18 Eye ON-, we identified 12 eyes with other than ipsilateral ON attacks within 6 months before baseline (five eyes of three patients with a myelitis, four eyes of two patients with myelitis and brainstem attacks, one eye of one patient with myelitis and contralateral ON and 2 eyes of 2 patients with contralateral ON; age 40 ± 9, male/female 5/ 3, EDSS 2.5, median follow-up 14 ± 5.9 months)) and six eyes without other attacks (age 39.0 ± 21.0, male/female 1/4, EDSS 3.5, median follow-up 26 ± 4.5 months). Retrospectively, one patient (2 eyes) could not be included in the study analysis because he had ONs on both sides during F/U and another patient (2 eyes) could not be included because he had insufficient follow-up less than 8 months. Data of further 8 eyes had to be excluded (five eyes with ON during F/U, one eye with ON less than 5 months before study inclusion, one eye with missing data, one eye with OCT-confounding disease). Clinical characteristics of all included patients are shown in Table 1.

Group differences at baseline
First, we analyzed group differences at baseline between MOG-IgG-seropositive patient eyes with a history of ON (Eye ON+ ), patient eyes without previous ON (Eye ON-) and eyes from HC. At baseline, in Eye ON-, pRNFL, INL and MV were not significantly different, but GCIP was significantly thinner in comparison to HC (p = 0.008) ( Table 2, Fig. 1). VA was lower in Eye ONin comparison to HC (p = 0.013).
One Eye ONshowed a massive thinness of the pRNFL at baseline despite a missing history of ON. We found macular microcysts within the INL in 6/20 (30%) Eye ON+ .
OCT changes during F/U Longitudinally, we observed pRNFL thinning, which was not accompanied by progressive GCIP reduction, in eyes without ON during F/U (annual loss: − 2.20 ± 4.29 μm vs. HC -0.35 ± 1.17 μm, p = 0.009) ( Fig. 2; individual changes in Additional file 1). There were no longitudinal group differences between Eye ON+ and Eye ONfor GCIP, pRNFL, INL and MV as well as between MOG-IgGseropositive NMOSD and other MOG-IgG-seropositive patients (Table 3). In a previous study investigating spinal cord changes in MOG-IgG patients, we suspected edematous changes in patients close to a clinical attack [27]. We therefore investigated patients with a nonipsilateral ON attack within 6 months of the baseline visit in a subgroup analysis. At baseline, the pRNFL in 12 Eye ONwith a non-ipsilateral ON attack within the 6 months before baseline was thicker in comparison to 6 Eye ONwithout a non-ipsilateral ON attack within the 6 months before baseline (pRNFL 100.2 ± 12.7 μm vs. 82.7 ± 16.2 μm (p = 0.019)) ( Fig. 3A). Reduction of pRNFL thickness was seen mainly in 3 eyes of the subgroup analysis. Two of the 3 eyes had no clinical evidence of unilateral ON attacks of the contralateral eye within the 6 months prior to inclusion in the study. One of the 3 eyes had a relapse complex with myelitis, brain attack and contralateral ON within 6 months prior to baseline. An ON-affection of these 12 Eye ONwith a non-ipsilateral ON attack was further ruled out by a stable high-contrast visual acuity (HCVA) without a change during F/U (HCVA as decimal, median (range): at baseline 1.0 (0.6-1.1); at last visit 1.0 (0.6-1.6)). A longitudinal graphical display of Eye ONshowed the pRNFL thinning to be predominantly present in Eye ONwith an attack before baseline (Fig. 3B). However, due to the small sample size, no statistical analysis could be performed.

Discussion
In this study, we investigated longitudinally, MOG-IgG-seropositive patients for potential progressive or covert damage in the retina in the absence of new clinical ON. We could not detect progressive GCIP thinning during F/U in MOG-IgG-seropositive patients, which is in contrast to progressive GCIP  reduction in AQP4-IgG-seropositive NMOSD and MS [22,27]. Instead, we observed a longitudinal pRNFL reduction, which in a consequent subgroup analysis appeared to primarily occur in patients with nonipsilateral ON attacks within 6 months before baseline. A hypothetical explanation of this finding could be a remission of pRNFL edema. Cross-sectional retinal imaging studies have shown conflicting results as to whether MOG-IgGassociated diseases have a more favorable outcome compared to patients with ON in other disease contexts [28][29][30][31][32][33][34]. The presumed higher relapse rates in MOG-IgG-seropositive patients could be associated with a severe retinal neuroaxonal loss and an unfavorable visual outcome [11]. Although OCT data regarding MOG-IgG-associated retinal damage are inconsistent [11,16,30,35], neuroaxonal retinal damage may occur as a consequence of clinical episode(s) of ON or of subclinical involvement [11,16]. ON was associated with macular microcysts, a biomarker suggestive of severe optic neuropathy [16,36,37]. A previous study investigating a smaller cohort of MOG-IgG-positive patients showed a significant reduction of the pRNFL and the ganglion cell layer in Eye ONcompared to HC cross-sectionally [16]. By contrast, in our current study, we could only confirm a significant GCIP reduction in Eye ONat baseline but no significant reduction of the pRNFL as a hint towards subclinical retinal pathology. However, pRNFL edema as a marker of immune-related swelling in the CNS after relapses and also outside of relapses could have contributed to this finding. The GCIP reduction at baseline could be discussed as progressive neurodegenerative retinal involvement, subclinical optic nerve pathology, chiasmal crossover of ON in contralateral eyes, or as an expression of subclinical ON in the previous patient's history. However, according to Ramanathan et al., only 5% of ONs in MOG-IgG-seropositive patients shows chiasmal involvement [38].
Longitudinally, we observed pRNFL but not GCIP thinning. We hypothesize that this can be explained not only by subclinical retinal or optic nerve involvement or druginduced retinal damage related to immunosuppressive treatment, but also by a remission of non-ipsilateral ON attacks that has occurred in Eye ONwithin 6 months before baseline since patients without clinical attacks ≤ 6 months before baseline did not present significant pRNFL or GCIP loss during F/U. This is clearly in contrast to our recently published data about longitudinal GCIP thinning in AQP4-IgG-seropositive NMOSD [22] or earlier studies reporting GCIP loss in MS [27] and might be an important hint towards the differentiation of MOG-IgGassociated diseases from AQP4-IgG-seropositive NMOSD. AQP4-IgG-seropositive NMOSD is an astrocytopathy, and a primary retinopathy caused by antibody-mediated damage is supported by animal studies and recently also clinical studies [22,39]. In contrast, the retina does not harbor myelin-producing oligodendrocytes and an expression of MOG has not been shown, making a primary retinopathy unlikely.
Further, data showing clear differences between AQP4-IgG-seropositive NMOSD and MOG-IgG-associated diseases were presented recently by Chien et al. [40]. Spinal cord imaging data showed differences in spinal cord affection patterns and disability accumulation. A higher prevalence of myelitis with clinical attacks and chronic spinal cord lesions was detected for AQP4-IgG-seropositive NMOSD patients in comparison to MOG-IgG-associated diseases [40]. Interestingly, MOG-IgG-seropositive   patients showed a swelling of the upper cervical cord area during other non-myelitis attacks, also pointing towards a systemic inflammatory affection in MOG-IgG-associated diseases as potentially shown here in the pRNFL during different attacks [40]. Our data is in line with the conclusion that AQP4-IgG-seropositive NMOSD and MOG-IgG-associated diseases are distinct immunological disorders, but share common clinical patterns [22,[40][41][42].
Limitations of our study are the heterogeneity of MOG-IgG-seropositive patients with different clinical phenotypes in our cohort, the heterogeneity of immunosuppressive treatments of our patients, and due to the rarity of MOG-IgG-seropositive patients in Europe, the small sample size, which leads to outliers possibly having a larger effect on the results, short and variable F/U, and the evaluation of MOG-IgG by different labs using different assays. Additionally, our study lacks magnetic resonance imaging data on optic nerve lesion lengths and lesion volumes of the afferent visual system as well as whole-brain lesion volume to further evaluate subclinical retinal atrophy in MOG-IgG-associated diseases.

Conclusions
We report in this small explorative study of MOG-IgGassociated diseases no evidence of GCIP thinning during F/U. Additionally, we found pRNFL reduction without GCIP loss during F/U predominantly in Eye ONwith other than ipsilateral ON attacks ≤ 6 months before baseline. We will investigate in a planned longitudinal study involving more centers, whether this reduction is actually due to a remission of edema or reflects retinal neurodegenerative processes or drug-induced retinal damage related to aggressive immunosuppressive treatment.

Additional file
Additional file 1: Figure S1. Spaghetti plots of longitudinal OCT data.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Ethics approval and consent to participate
Twenty-four patients were seen at four university tertiary care centers specialized in neuroimmunological diseases (Institute of clinical Neuroimmunology, Ludwig-Maximilians University (LMU), Munich, NeuroCure Clinical Research Center, Charité -Universitätsmedizin Berlin, Germany; Department of Neurology, University of Lille Hospital, Lille, France; Department of Neurology, Klinikum Rechts der Isar, Technische Universität München (TUM), Munich, Germany). Written informed consent was obtained from all patients participating in the study. The local ethics committees approved the study protocol in accordance with the Declaration of Helsinki (1964) in its currently applicable version.

Consent for publication Not applicable
Competing interests JH reports a grant for OCT research from the Friedrich-Baur-Stiftung; personal fees and non-financial support from Merck, Novartis, Roche, Bayer Healthcare, Santhera, Biogen, Sanofi Genzyme; and non-financial support of the Guthy-Jackson Charitable Foundation, all outside the submitted work. FCO was an employee of Nocturne UG, unrelated to this work. OO reports grant for research from Novartis-Pharma and grants and personal fees from Biogen, Genzyme, Merck, Novartis-Pharma, and Teva-Pharma, outside the submitted work. BK reports grants from the Bundesministerium für Bildung und Forschung (Kompetenznetz Multiple Sklerose KKNMS), intramural funding from the Technical University of Munich (KKF program) and a grant from Novartis unrelated to this study. HZ reports a research grant from Novartis related to this study and speaking fees from Teva unrelated to this study. LB has nothing to disclose. JBS has received travel grants and speaking fees from Bayer Healthcare, Biogen Idec, Merck Serono, Sanofi-Aventis/ Genzyme, Teva Pharmaceuticals unrelated to this study. SJ has nothing to disclose. The Neurological Research Laboratory (Medical University of Innsbruck and Tirol Kliniken, MR) receives payments for antibody assays (AQP4-and anti-neuronal antibodies) and for AQP4-and MOG-antibody validation experiments organized by Euroimmun (Germany). KR was supported by the German Ministry of Education and Research (BMBF/KKNMS, Competence Network Multiple Sclerosis) and has received research support from Novartis and Merck Serono as well as speaking fees and travel grants from Guthy Jackson Charitable Foundation, Bayer Healthcare, Biogen Idec, Merck Serono, Sanofi-Aventis/Genzyme, Teva Pharmaceuticals, Roche and Novartis. EM received an honorarium from Roche, Novartis and Genzyme, and grant support from Novartis and Genzyme. RH received research grants and/or speaker honoraria from Actelion, Genzyme-Sanofi, Novartis, Immunic, Roche. FP reports research grants and speaker honoraria from Bayer, Teva, Genzyme, Merck, Novartis, MedImmune and is a member of the steering committee of the OCTIMS study (Novartis), all unrelated to this work. AUB is the founder and holds shares of Motognosis and Nocturne. He is named as an inventor on several patent applications describing serum biomarkers for MS, perceptive visual computing for tracking of motor dysfunction and OCT image analysis. TK received travel expenses and personal compensations from Bayer Healthcare, Teva Pharma, Merck, Novartis Pharma, Sanofi-Aventis/Genzyme, Roche, and Biogen as well as grant support from Bayer-Schering AG, Novartis and Chugai Pharma, unrelated to this work.