A subset of patients with neuromyelitis optica spectrum disorders (NMOSD) has been shown to be seropositive for myelin oligodendrocyte glycoprotein antibodies (MOG-IgG). To describe the epidemiological, clinical, radiological, cerebrospinal fluid (CSF), and electrophysiological features of a large cohort of MOG-IgG-positive patients with optic neuritis (ON) and/or myelitis (n = 50) as well as attack and long-term treatment outcomes. Retrospective multicenter study. The sex ratio was 1:2.8 (m:f). Median age at onset was 31 years (range 6-70). The disease followed a multiphasic course in 80% (median time-to-first-relapse 5 months; annualized relapse rate 0.92) and resulted in significant disability in 40% (mean follow-up 75 ± 46.5 months), with severe visual impairment or functional blindness (36%) and markedly impaired ambulation due to paresis or ataxia (25%) as the most common long-term sequelae. Functional blindness in one or both eyes was noted during at least one ON attack in around 70%. Perioptic enhancement was present in several patients. Besides acute tetra-/paraparesis, dysesthesia and pain were common in acute myelitis (70%). Longitudinally extensive spinal cord lesions were frequent, but short lesions occurred at least once in 44%. Fourty-one percent had a history of simultaneous ON and myelitis. Clinical or radiological involvement of the brain, brainstem, or cerebellum was present in 50%; extra-opticospinal symptoms included intractable nausea and vomiting and respiratory insufficiency (fatal in one). CSF pleocytosis (partly neutrophilic) was present in 70%, oligoclonal bands in only 13%, and blood-CSF-barrier dysfunction in 32%. Intravenous methylprednisolone (IVMP) and long-term immunosuppression were often effective; however, treatment failure leading to rapid accumulation of disability was noted in many patients as well as flare-ups after steroid withdrawal. Full recovery was achieved by plasma exchange in some cases, including after IVMP failure. Breakthrough attacks under azathioprine were linked to the drug-specific latency period and a lack of cotreatment with oral steroids. Methotrexate was effective in 5/6 patients. Interferon-beta was associated with ongoing or increasing disease activity. Rituximab and ofatumumab were effective in some patients. However, treatment with rituximab was followed by early relapses in several cases; end-of-dose relapses occurred 9-12 months after the first infusion. Coexisting autoimmunity was rare (9%). Wingerchuk’s 2006 and 2015 criteria for NMO(SD) and Barkhof and McDonald criteria for multiple sclerosis (MS) were met by 28%, 32%, 15%, 33%, respectively; MS had been suspected in 36%. Disease onset or relapses were preceded by infection, vaccination, or pregnancy/delivery in several cases. Our findings from a predominantly Caucasian cohort strongly argue against the concept of MOG-IgG denoting a mild and usually monophasic variant of NMOSD. The predominantly relapsing and often severe disease course and the short median time to second attack support the use of prophylactic long-term treatments in patients with MOG-IgG-positive ON and/or myelitis.

Antibodies to myelin oligodendrocyte glycoprotein (MOG-IgG) have been suggested to play a role in a subset of patients with neuromyelitis optica and related disorders. To assess (i) the frequency of MOG-IgG in a large and predominantly Caucasian cohort of patients with optic neuritis (ON) and/or myelitis; (ii) the frequency of MOG-IgG among AQP4-IgG-positive patients and vice versa; (iii) the origin and frequency of MOG-IgG in the cerebrospinal fluid (CSF); (iv) the presence of MOG-IgG at disease onset; and (v) the influence of disease activity and treatment status on MOG-IgG titers. 614 serum samples from patients with ON and/or myelitis and from controls, including 92 follow-up samples from 55 subjects, and 18 CSF samples were tested for MOG-IgG using a live cell-based assay (CBA) employing full-length human MOG-transfected HEK293A cells. MOG-IgG was detected in 95 sera from 50 patients with ON and/or myelitis, including 22/54 (40.7%) patients with a history of both ON and myelitis, 22/103 (21.4%) with a history of ON but no myelitis and 6/45 (13.3%) with a history of longitudinally extensive transverse myelitis but no ON, and in 1 control patient with encephalitis and a connective tissue disorder, all of whom were negative for AQP4-IgG. MOG-IgG was absent in 221 further controls, including 83 patients with AQP4-IgG-seropositive neuromyelitis optica spectrum disorders and 85 with multiple sclerosis (MS). MOG-IgG was found in 12/18 (67%) CSF samples from MOG-IgG-seropositive patients; the MOG-IgG-specific antibody index was negative in all cases, indicating a predominantly peripheral origin of CSF MOG-IgG. Serum and CSF MOG-IgG belonged to the complement-activating IgG1 subclass. MOG-IgG was present already at disease onset. The antibodies remained detectable in 40/45 (89%) follow-up samples obtained over a median period of 16.5 months (range 0–123). Serum titers were higher during attacks than during remission (p < 0.0001), highest during attacks of simultaneous myelitis and ON, lowest during acute isolated ON, and declined following treatment. To date, this is the largest cohort studied for IgG to human full-length MOG by means of an up-to-date CBA. MOG-IgG is present in a substantial subset of patients with ON and/or myelitis, but not in classical MS. Co-existence of MOG-IgG and AQP4-IgG is highly uncommon. CSF MOG-IgG is of extrathecal origin. Serum MOG-IgG is present already at disease onset and remains detectable in the long-term course. Serum titers depend on disease activity and treatment status.

Objective To update the consensus recommendations for reporting of quantitative optical coherence tomography (OCT) study results, thus revising the previously published Advised Protocol for OCT Study Terminology and Elements (APOSTEL) recommendations. Methods To identify studies reporting quantitative OCT results, we performed a PubMed search for the terms “quantitative” and “optical coherence tomography” from 2015 to 2017. Corresponding authors of the identified publications were invited to provide feedback on the initial APOSTEL recommendations via online surveys following the principle of a modified Delphi method. The results were evaluated and discussed by a panel of experts and changes to the initial recommendations were proposed. A final survey was recirculated among the corresponding authors to obtain a majority vote on the proposed changes. Results A total of 116 authors participated in the surveys, resulting in 15 suggestions, of which 12 were finally accepted and incorporated into an updated 9-point checklist. We harmonized the nomenclature of the outer retinal layers, added the exact area of measurement to the description of volume scans, and suggested reporting device-specific features. We advised to address potential bias in manual segmentation or manual correction of segmentation errors. References to specific reporting guidelines and room light conditions were removed. The participants9 consensus with the recommendations increased from 80% for the previous APOSTEL version to greater than 90%. Conclusions The modified Delphi method resulted in an expert-led guideline (evidence Class III; Grading of Recommendations, Assessment, Development and Evaluations [GRADE] criteria) concerning study protocol, acquisition device, acquisition settings, scanning protocol, funduscopic imaging, postacquisition data selection, postacquisition analysis, nomenclature and abbreviations, and statistical approach. It will be essential to update these recommendations to new research and practices regularly.

Antibodies against myelin oligodendrocyte glycoprotein (MOG-IgG) have been reported in patients with aquaporin-4 antibody (AQP4-IgG)-negative neuromyelitis optica spectrum disorders (NMOSD). The objective of this study was to describe optic neuritis (ON)-induced neuro-axonal damage in the retina of MOG-IgG-positive patients in comparison with AQP4-IgG-positive NMOSD patients. Afferent visual system damage following ON was bilaterally assessed in 16 MOG-IgG-positive patients with a history of ON and compared with that in 16 AQP4-IgG-positive NMOSD patients. In addition, 16 healthy controls matched for age, sex, and disease duration were analyzed. Study data included ON history, retinal optical coherence tomography, visual acuity, and visual evoked potentials. Eight MOG-IgG-positive patients had a previous diagnosis of AQP4-IgG-negative NMOSD with ON and myelitis, and eight of (mainly recurrent) ON. Twenty-nine of the 32 eyes of the MOG-IgG-positive patients had been affected by at least one episode of ON. Peripapillary retinal nerve fiber layer thickness (pRNFL) and ganglion cell and inner plexiform layer volume (GCIP) were significantly reduced in ON eyes of MOG-IgG-positive patients (pRNFL = 59 ± 23 μm; GCIP = 1.50 ± 0.34 mm3) compared with healthy controls (pRNFL = 99 ± 6 μm, p < 0.001; GCIP = 1.97 ± 0.11 mm3, p < 0.001). Visual acuity was impaired in eyes after ON in MOG-IgG-positive patients (0.35 ± 0.88 logMAR). There were no significant differences in any structural or functional visual parameters between MOG-IgG-positive and AQP4-IgG-positive patients (pRNFL: 59 ± 21 μm; GCIP: 1.41 ± 0.27 mm3; Visual acuity = 0.72 ± 1.09 logMAR). Importantly, MOG-IgG-positive patients had a significantly higher annual ON relapse rate than AQP4-IgG-positive patients (median 0.69 vs. 0.29 attacks/year, p = 0.004), meaning that on average a single ON episode caused less damage in MOG-IgG-positive than in AQP4-IgG-positive patients. pRNFL and GCIP loss correlated with the number of ON episodes in MOG-IgG-positive patients (p < 0.001), but not in AQP4-IgG-positive patients. Retinal neuro-axonal damage and visual impairment after ON in MOG-IgG-positive patients are as severe as in AQP4-IgG-positive NMOSD patients. In MOG-IgG-positive patients, damage accrual may be driven by higher relapse rates, whereas AQP4-IgG-positive patients showed fewer but more severe episodes of ON. Given the marked damage in some of our MOG-IgG-positive patients, early diagnosis and timely initiation and close monitoring of immunosuppressive therapy are important.

Myelin oligodendrocyte glycoprotein antibodies (MOG-IgG) are present in a subset of aquaporin-4 (AQP4)-IgG-negative patients with optic neuritis (ON) and/or myelitis. Little is known so far about brainstem involvement in MOG-IgG-positive patients. To investigate the frequency, clinical and paraclinical features, course, outcome, and prognostic implications of brainstem involvement in MOG-IgG-positive ON and/or myelitis. Retrospective case study. Among 50 patients with MOG-IgG-positive ON and/or myelitis, 15 (30 %) with a history of brainstem encephalitis were identified. All were negative for AQP4-IgG. Symptoms included respiratory insufficiency, intractable nausea and vomiting (INV), dysarthria, dysphagia, impaired cough reflex, oculomotor nerve palsy and diplopia, nystagmus, internuclear ophthalmoplegia (INO), facial nerve paresis, trigeminal hypesthesia/dysesthesia, vertigo, hearing loss, balance difficulties, and gait and limb ataxia; brainstem involvement was asymptomatic in three cases. Brainstem inflammation was already present at or very shortly after disease onset in 7/15 (47 %) patients. 16/21 (76.2 %) brainstem attacks were accompanied by acute myelitis and/or ON. Lesions were located in the pons (11/13), medulla oblongata (8/14), mesencephalon (cerebral peduncles; 2/14), and cerebellar peduncles (5/14), were adjacent to the fourth ventricle in 2/12, and periaqueductal in 1/12; some had concomitant diencephalic (2/13) or cerebellar lesions (1/14). MRI or laboratory signs of blood-brain barrier damage were present in 5/12. Cerebrospinal fluid pleocytosis was found in 11/14 cases, with neutrophils in 7/11 (3-34 % of all CSF white blood cells), and oligoclonal bands in 4/14. Attacks were preceded by acute infection or vaccination in 5/15 (33.3 %). A history of teratoma was noted in one case. The disease followed a relapsing course in 13/15 (87 %); the brainstem was involved more than once in 6. Immunosuppression was not always effective in preventing relapses. Interferon-beta was followed by new attacks in two patients. While one patient died from central hypoventilation, partial or complete recovery was achieved in the remainder following treatment with high-dose steroids and/or plasma exchange. Brainstem involvement was associated with a more aggressive general disease course (higher relapse rate, more myelitis attacks, more frequently supratentorial brain lesions, worse EDSS at last follow-up). Brainstem involvement is present in around one third of MOG-IgG-positive patients with ON and/or myelitis. Clinical manifestations are diverse and may include symptoms typically seen in AQP4-IgG-positive neuromyelitis optica, such as INV and respiratory insufficiency, or in multiple sclerosis, such as INO. As MOG-IgG-positive brainstem encephalitis may take a serious or even fatal course, particular attention should be paid to signs or symptoms of additional brainstem involvement in patients presenting with MOG-IgG-positive ON and/or myelitis.

Machine learning-based imaging diagnostics has recently reached or even surpassed the level of clinical experts in several clinical domains. However, classification decisions of a trained machine learning system are typically non-transparent, a major hindrance for clinical integration, error tracking or knowledge discovery. In this study, we present a transparent deep learning framework relying on 3D convolutional neural networks (CNNs) and layer-wise relevance propagation (LRP) for diagnosing multiple sclerosis (MS), the most widespread autoimmune neuroinflammatory disease. MS is commonly diagnosed utilizing a combination of clinical presentation and conventional magnetic resonance imaging (MRI), specifically the occurrence and presentation of white matter lesions in T2-weighted images. We hypothesized that using LRP in a naive predictive model would enable us to uncover relevant image features that a trained CNN uses for decision-making. Since imaging markers in MS are well-established this would enable us to validate the respective CNN model. First, we pre-trained a CNN on MRI data from the Alzheimer's Disease Neuroimaging Initiative (n = 921), afterwards specializing the CNN to discriminate between MS patients (n = 76) and healthy controls (n = 71). Using LRP, we then produced a heatmap for each subject in the holdout set depicting the voxel-wise relevance for a particular classification decision. The resulting CNN model resulted in a balanced accuracy of 87.04% and an area under the curve of 96.08% in a receiver operating characteristic curve. The subsequent LRP visualization revealed that the CNN model focuses indeed on individual lesions, but also incorporates additional information such as lesion location, non-lesional white matter or gray matter areas such as the thalamus, which are established conventional and advanced MRI markers in MS. We conclude that LRP and the proposed framework have the capability to make diagnostic decisions of CNN models transparent, which could serve to justify classification decisions for clinical review, verify diagnosis-relevant features and potentially gather new disease knowledge.