key: cord-0327624-zsoyupfm authors: Sperber, P. S.; Gebert, P.; Broersen, L. H. A.; Huo, S.; Sophie, P. K.; Pruess, H.; Heuschmann, P. U.; Endres, M.; Liman, T. G.; Siegerink, B. title: Serum anti-NMDA-receptor antibodies and cognitive function after ischemic stroke (PROSCIS-B) date: 2020-04-18 journal: nan DOI: 10.1101/2020.04.15.20066324 sha: 9f5f873c7dd4ca4255e00db53dd43d1bd96b339c doc_id: 327624 cord_uid: zsoyupfm Objective: We aimed to investigate whether serum anti-N-Methyl-D-Aspartate-receptor GluN1 antibodies (NMDAR1-abs) are associated with worse cognitive function over three years after first ischemic stroke. Methods: Data were used from the PROSpective Cohort with Incident Stroke-Berlin (PROSCIS-B;NCT01363856). NMDAR1-abs (IgM/IgA/IgG) were measured with cell-based assays from serum obtained within seven days after first-ever stroke. Seropositivity was defined as titers >1:10, low titers as <1:100 and high titers as >1:100. We assessed cognitive status annually up to three years after stroke with the Telephone Interview for Cognitive Status - modified (TICS-m) and used crude and adjusted linear mixed models to estimate the impact of NMDAR1-abs exposure on TICS-m over time. Results: Data on NMDAR1-abs (median day of sampling=4[IQR=2-5]) were available in 583 of 621 PROSCIS-B patients (39% female; median NIHSS=2[IQR=1-4]; median MMSE=28[IQR:26-30]; median mRS=2[IQR=1-3]) of whom 76(13%) were seropositive (IgM:n=48/IgA:n=43/IgG:n=2). TICS-m over time was not different in NMDAR1-abs seropositive compared to seronegative patients (Crude beta=0.38[95%CI=-1.00 to 1.76]; adjusted beta Model3=0.30[95%CI=-1.14 to 1.73]). In subgroups, TICS-m over time was not different in patients with low titers (Crude beta=1.53[95%CI=-0.06 to 3.11]; adjusted beta Model3=1.42[95%CI= 0.23 to 3.08]), however, in patients with high titers TICS-m was lower in the crude model (Crude beta= 2.54[95%CI=-4.99 to -0.08]), with a similar effect size after confounder adjustment (adjusted beta Model3=-2.30[95%CI=-4.82 to 0.21]). All groups were compared to seronegative patients, respectively. Conclusion: Overall NMDAR1-abs seropositivity was not associated with cognitive function over time after first-ever mild-to-moderate ischemic stroke. Our data suggest that high titers associate with impaired cognitive function after stroke, warranting larger studies. Overall NMDAR1-abs seropositivity was not associated with cognitive function over time after first-ever mild-to-moderate ischemic stroke. Our data suggest that high titers associate with impaired cognitive function after stroke, warranting larger studies. Cognitive impairment is frequent after stroke and up to one third of all stroke patients develop incident post stroke dementia.(1,2) Serum anti-NMDA (N-Methyl-D-Aspartate)-receptor antibodies (NMDAR1-abs), primarily of the IgA and IgM isotypes have been observed in about 10% of the apparently healthy population, however, other studies also showed associations between seropositivity and cognitive impairment. (3) (4) (5) We previously observed that NMDAR1-abs seropositivity with high titers (>1:100) was associated with unfavorable functional outcome at one year after stroke and overall seropositivity was associated with increased vascular recurrent risk or death within three years after ischemic stroke, (6) contrasting previous findings that indicated beneficial effects of serum NMDAR1-abs on infarct volumes in ischemic stroke. (7, 8) The functionality of IgA and IgM antibodies has been studied previously and remains controversial, especially the clinical implication of these isotypes. (9, 10) Physiological NMDAR function is crucial for mechanisms of synaptic plasticity and normal neuronal function. (11) We consider that serum NMDAR1-abs may enter brain parenchyma following blood brain barrier disruption as a consequence of stroke where they may lead to a downregulation of NMDA-receptors and hamper functional recovery of the damaged tissue.(10) Therefore, this study was motivated by the hypothesis that NMDAR1-abs have detrimental effects on cognitive outcome after stroke. We herewith aimed to quantify the impact of NMDAR1-abs on cognitive functioning up to three years in a large cohort of first-ever ischemic stroke patients. The PROSCIS -B study (ClinicalTrials.gov identifier: NCT01363856) is a prospective observational hospital-based cohort study, which recruited patients at three tertiary university hospital stroke units of the Charité-Universitätsmedizin Berlin with first-ever stroke according to WHO criteria, to study stroke secondary risks. (12) Patients presenting with brain tumor, brain metastasis of a tumor of other origin, or patients participating in an intervention study, were excluded. Details on the study design have been described previously. (13, 14) For a detailed baseline characterization, an extensive clinical and technical examination was performed within seven days after the acute event including blood sampling for laboratory measures. Patients were followed-up annually by telephone interviews or postal mail contact assessing i.a. cognitive function and functional outcome up to three years after the index event. For this investigation, only patients with mild-to-moderate ischemic stroke events (National Institutes of Health Stroke Scale [NIHSS] < 16) were included, as we counted very few cases with severe strokes (NIHSS > 15, n=6). Serum blood samples were obtained from patients within 7 days after stroke and stored at -80°C until they were first-ever thawed for antibody measurements. NMDAR1-abs were measured with cell-based assays by the Euroimmun laboratory in Luebeck, Germany. In summary, HEK293 cells were transfected with GluN1 subunits of NMDA-receptors to bind antibodies of the IgM, IgA and IgG isotype from patient serum. Fluorescein isothiocyanate anti-human IgG was secondary administered to manually obtain staining with fluorescence microscopy. The assessors had no insight to patient data. Details on the procedure have been described elsewhere. (15, 16) Titer levels started from a dilution of 1:10, which defined All rights reserved. No reuse allowed without permission. the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is . https://doi.org/10.1101/2020.04. 15.20066324 doi: medRxiv preprint seropositivity in our study. For sub-groups, we defined titers of 1:10 to 1:100 as low titers and titers > 1:100 as high titers a priori, in line with previous analyses. (6) Outcome definitions: Methods II (online only). We calculated 95% confidence intervals in addition to our effect estimate as a measure of precision from our sample. All rights reserved. No reuse allowed without permission. the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is . https://doi.org/10.1101/2020.04. 15.20066324 doi: medRxiv preprint Potential confounders were selected a priori according to their presumable impact on both, NMDAR1-abs serostatus and post-stroke cognitive functioning. We adjusted our first model (Model1) for age, sex and years of school education in 2 categories (≤ 10 years of school;>10 years of school), the second Model (Model2) was adjusted for a propensity score, which was built using the variables age, sex, years of school education, current smoking (yes/no), and a disease history assessment requesting history of 1) arterial hypertension (yes/no), 2) peripheral artery disease (yes/no), 3) atrial fibrillation (yes/no), 4) diabetes mellitus (yes/no), the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is . https://doi.org/10.1101/2020.04. 15.20066324 doi: medRxiv preprint All patients or their legal guardian gave written informed consent for study participation. Berlin and the study was conducted in concordance to ethical principles framed by the declaration of Helsinki. The data that support the findings of this study are available from the qualified principal investigator of PROSCIS-B (T.G. Liman, thomas.liman@charite.de) upon reasonable request. Fifty-five patients died during the three-year follow up period and were therefore missing single or multiple TICS-m assessments. PROSCIS-B patients had a mean age of 67 (Standard deviation [SD]=13), 39% were female, median baseline NIHSS was 2 (IQR=1 to 4) and baseline mRS was 2 (IQR=1 to 3). We All rights reserved. No reuse allowed without permission. the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is . https://doi.org/10.1101/2020.04. 15.20066324 doi: medRxiv preprint measured NMDAR1-abs seropositivity in 76 patients (13%) in whom IgM NMDAR1-abs were present in serum of 49 patients (8%), followed by IgA (43 patients (7%)), and IgG isotype antibodies were detected in 2 patients (0.3%), only. Seventeen patients (3%) were presenting IgM and IgA antibodies, simultaneously. We did not observe major differences in baseline characteristics between NMDAR1-abs seropositive and seronegative patients and likewise not between subgroups as shown in Table 1 . Cognitive function at baseline was assessed with the MMSE and was 28 points (IQR=26 to 30) on a median in the total cohort and in seronegative patients. In overall seropositive patients the median MMSE was 29 points (IQR=27 to 30) in seropositive patients with low titers the median MMSE was 29 points (IQR=27.5 to 30) and for those patients with high titers it was 27 points (IQR=24 to 29). Twenty-eight percent of PROSCIS-B patients with MMSE assessment (n=605) were cognitive impaired with test results of equal or less than 26 points. Among seronegative patients with MMSE and NMDAR1-abs measurement (n=497) this proportion was 29% and among seropositive patients with both assessments (n=72) 22% were cognitively impaired. Out of 52 patients with low titers and both assessments 15% of the patients were cognitively impaired at baseline and in patients with high titers this proportion was 40% (n=20). All rights reserved. No reuse allowed without permission. the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is (4) anti-NMDAR GluN1 IgM antibodies n (%) 49 (8) -49 (8) 34 (6) 15 (3) anti-NMDAR GluN1 IgA antibodies n (%) 43 (7) -43 (7) 31 (5) 12 (2) anti-NMDAR GluN1 IgG antibodies n (%) (13) 67 (13) 66 (14) 65 (14) TICS-m scores for all observations over time after stroke are visualized in Figure 2 and Table 2 . Effect estimates for all potential confounders including a quantification of the explained and residual variance within these models are presented in Supplemental Table I . All rights reserved. No reuse allowed without permission. the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is . https://doi.org/10.1101/2020.04.15.20066324 doi: medRxiv preprint Table 2 Serostatus, anti-NMDAR antibody seroprevalence. β, effect size (points on the Telephone Interview for Cognitive Status-modified [TICS-m]) in relation to reference group. 95% CI, 95% confidence interval. ref., reference category. *Crude analyses; †Model 1, adjusted for age and sex and years of school education. ‡Model 2, adjusted for propensity score (indicating the propensity for antibody serum presentation according to age, sex, years of school education, current smoking, habitual alcohol consumption, history of hypertension, history of myocardial infarction, history of diabetes mellitus, history of peripheral artery disease, history of atrial fibrillation; for sub-analyses a propensity score for titers ≤ 1:100 and titers ≥ 1:320 was build using the same variables, respectively). §Model 3, adjusted for propensity score, National Institutes of Health Stroke Scale (NIHSS), Trial of Org 10172 in Acute Stoke Treatment (TOAST) categories. Crude We provide full data in Supplemental Table V . In our study of mild-to-moderately affected first-ever ischemic stroke patients, NMDAR1-abs seropositivity in the acute phase, primarily IgM and IgA, was not associated with cognitive function over three years after stroke. In subgroups, we also observed no differences between seronegative patients and patients with low titers. However, our data indicate decreased cognitive function over time after ischemic stroke in patients with high titers of NMDAR1abs compared to seronegative patients, which is independent from several other risk factors. All rights reserved. No reuse allowed without permission. the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is . https://doi.org/10.1101/2020.04. 15.20066324 doi: medRxiv preprint NMDAR1-abs serum prevalence has been observed in about ~10% of the healthy population, and in those with various diseases, including stroke. (9, 10, 20) Similar to our study results, mainly IgM and IgA antibodies were observed. (10) The swift blood sampling in our study and the similar proportion of seropositive patients in healthy individuals suggests that antibodies were preexistent before the stroke and thus not a consequence of the acute event. (20) This is further supported by findings from the study of Zerche et al. who measured NMDAR1-abs in a more standardized time-window swiftly after the stroke (<36h) and measured a similar serum prevalence, using the same determination technique like we used for our study. (7) The high serum prevalence of IgM and IgA NMDAR1-abs in healthy elderly individuals questions a pathological significance of these isotypes on their own. Cumulative evidence from disease cohorts other than stroke patients link serum presentation particularly with cognitive outcomes: for example was NMDAR1-abs IgA and IgM seropositivity associated with cognitive impairment in melanoma patients, and others found serum IgA to be associated with different types of slowly progressive cognitive impairment. (3) (4) (5) In stroke, however, NMDAR1-abs seropositivity was foremost linked with beneficial effects on infarct volumes, which was explained by presumed effects of NMDAR1-abs on NMDA-receptor mediated excitotoxicity. (7, 8, 21) In line with other studies mentioned earlier, (3) (4) (5) in our study of firstever stroke patients high titers were associated with decreased cognitive function, challenging previous hypotheses of beneficial effects of these antibodies in stroke pathology. (7, 8) This current observation further coincides with our observation regarding NMDAR1-abs and other outcomes in the PROSCIS study, i.e. increased vascular risk for overall seropositive patients and unfavorable functional outcome in patients with high titers, after stroke respectively. (6) Considering that NMDARs are highly expressed in the central nervous system and that the blood brain barrier disruption in stroke allows serum NMDAR1-abs to access vulnerable neuronal tissue, it appears reasonable to assume antigen binding of NMDAR1-abs to their targeted receptor and that they subsequently exert functional effects. If binding results in All rights reserved. No reuse allowed without permission. the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is . https://doi.org/10.1101/2020.04. 15.20066324 doi: medRxiv preprint NMDAR downregulation, as proposed previously, (10, 22) this could hamper functional recovery and lead to impaired tissue reorganization long-term after stroke.(23) This is because NMDARs are crucial for mechanisms of synaptic plasticity and normal neuronal functioning. (11, 21) Although different groups reported functional effects of IgA and IgM on NMDARs in terms of receptor internalization, the question whether these effects are sufficient to have clinical implications remains a controversy. (4, 9, 10) NMDAR1-abs may also be markers of an underlying immunology which associates with vascular risk, functional and neuropsychiatric impairment and may therefore be predictive biomarkers for outcomes after stroke, irrespective of an underlying causal relationship. Either way our effect sizes combined with the precision of estimation is not indicative for the quality of an underlying relationship. For more clarification, our study would have benefitted from a deep immune-characterization of study subjects. We could not observe any associations of NMDAR1-abs serum presentation with low titers and cognitive outcome. If we take into account that only high titers associate with unfavorable functional outcome after stroke, a higher cut-off for seropositivity should be considered. (6) We consider the effects observed for high NMDAR1-abs titers as clinically significant, to -1.05) in the PROSCIS-B study and thus essentially the same effect than the one we observed for NMDAR1-abs seropositivity with high titers on TICS-m, the importance may become clearer. Overall, our Model2 may be the most reliable estimate because we comprehensively adjusted for potential confounders while incorporating only a single parameter, i.e. the propensity score, which prevented over-parametrization of the model. In a sensitivity analysis, we excluded TICS-m observations of depressed patients and effects All rights reserved. No reuse allowed without permission. the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is . https://doi.org/10.1101/2020.04. 15.20066324 doi: medRxiv preprint observed for seropositivity with high titers were extremely attenuated. This is explained by the exclusion of a high proportion of seropositive participants with high titers (see Supplemental Table IV ) and raises the question whether those excluded patients suffered both, i.e. cognitive impairment and depression, or whether pseudodementia due to depression was present in these patients and cognitive decline played a minor role. Our cognitive outcome scores were increasing over time indicated by a precise positive effect size for time in our models (see Supplemental Table I ), which we interpret as a learning effect for TICS-m due to repeated measures. (25) Otherwise, our models seem to accurately pick up on the effects, as demonstrated by single estimates for our prechosen explanatory variables: for example, higher school education (>10 years) caused an increase of 3.57 TICS-m points compared to lower school education (<= 10 years) (Supplemental Table I ), which supports that despite this learning effect our study remains representative. Our study results may be biased towards the null due to a learning effect in the TICS-m instrument. This consideration was previously reported for TICS-m in other studies. (25) Furthermore, we were not able to conduct a detailed neuropsychological testing, which could discriminate even minor disabilities in patients with only mild-to-moderate stroke events. To reduce internal heterogeneity we excluded those six patients with a baseline NIHSS of ≥16. However, differences in cognitive function may be more pronounced in severely affected stroke patients. Furthermore, our study results cannot be generalized to patients with severe stroke events. Selection bias may be present, as we recorded a total of ~ 40% of the observations were missing TICS-m assessments. We addressed this short-coming in two sensitivity analyses, one of which was a multiple imputation procedure for missing values. However, as we cannot exclude that patients were missing an assessments not at random bias may still be present or even worsened. (26) Our results need further validation, especially All rights reserved. No reuse allowed without permission. the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is . https://doi.org/10.1101/2020.04. 15.20066324 doi: medRxiv preprint because our effect estimate did not appear robust in different sensitivity analyses. Besides these limitations, this study followed an a priori analysis plan with prespecified confounder adjustment and we benefit from highly standardized state-of-the-art antibody measurements. (27) We revisited our study results with several sensitivity analyses, which elucidated that our study results demand further validation. To conclude, NMDAR1-abs overall seropositivity and NMDAR1-abs was not associated with cognitive function over time after first-ever ischemic mild-to-moderate stroke. Our data point the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is . https://doi.org/10.1101/2020.04. 15.20066324 doi: medRxiv preprint the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is . https://doi.org/10.1101/2020.04. 15.20066324 doi: medRxiv preprint Post-stroke dementia -a comprehensive review Incidence and prevalence of dementia associated with transient ischaemic attack and stroke: analysis of the population-based Oxford Vascular Study High prevalence of NMDA receptor IgA/IgM antibodies in different dementia types IgA NMDA receptor antibodies are markers of synaptic immunity in slow cognitive impairment Neuronal autoantibodies associated with cognitive impairment in melanoma patients Receptor Antibodies and Long-Term Clinical Outcome After Stroke (PROSCIS-B) Preexisting Serum Autoantibodies Against the NMDAR Subunit NR1 Modulate Evolution of Lesion Size in Acute Ischemic Stroke An Oral Vaccine Against NMDAR1 with Efficacy in Experimental Stroke and Epilepsie No reuse allowed without permission. the author/funder, who has granted medRxiv a license to display the preprint in perpetuity Clinical and pathogenic significance of IgG, IgA, and IgM antibodies against the NMDA receptor All naturally occurring autoantibodies against the NMDA receptor subunit NR1 have pathogenic potential irrespective of epitope and immunoglobulin class NMDA receptor subunit diversity: impact on receptor properties, synaptic plasticity and disease Experience from a multicentre stroke register: a preliminary report Prediction of vascular risk after strokeprotocol and pilot data of the prospective Cohort with incident stroke (PROSCIS) Outcome after stroke attributable to baseline factors -The PROSpective Cohort with Incident Stroke ( PROSCIS ) Comparison of diagnostic accuracy of microscopy and flow cytometry in evaluating N-methyl-D-aspartate receptor antibodies in serum using a live cell-based assay Paraneoplastic anti-N-methyl-D-aspartate receptor encephalitis associated with ovarian teratoma No reuse allowed without permission. the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is Validation of the telephone interview for cognitive status-modified in subjects with normal cognition, mild cognitive impairment, or dementia Propensity Scoring. International Encyclopedia of Social & Behavioral Sciences. 2015.140-146p Seroprevalence of autoantibodies against brain antigens in health and disease Synaptic versus extrasynaptic NMDA receptor signalling: implications for neurodegenerative disorders Human cerebrospinal fluid monoclonal Nmethyl-D-aspartate receptor autoantibodies are sufficient for encephalitis pathogenesis Personal view Why did NMDA receptor antagonists fail clinical trials for stroke and traumatic brain injury ? Effect size -large, medium, and small Infectious burden and cognitive performance: The northern Manhattan study Multiple imputation for missing data in epidemiological and clinical research: Potential and pitfalls Multicentre comparison of a diagnostic assay: Aquaporin-4 antibodies in neuromyelitis optica No reuse allowed without permission. the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is The authors want to thank the EUROIMMUN laboratory for NMDAR1-abs measurements and Jane Thümmler for data management. PSS is a MD candidate at the Charité -Universitätsmedizin Berlin, and this work is submitted in partial fulfillment of her thesis. The PROSCIS-B study received funding from the Federal Ministry of Education and the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.The copyright holder for this preprint (which was not peer-reviewed) is .