Molecular Psychiatry (2021) 26:2590–2604 https://doi.org/10.1038/s41380-020-00899-w ARTICLE Clinical, cognitive and neuroanatomical associations of serum NMDAR autoantibodies in people at clinical high risk for psychosis Thomas A. Pollak 1,2,3 Matthew J. Kempton1,2 Conrad Iyegbe1 Angela Vincent 4 Sarosh R. Irani5 ● ● ● ● ● Ester Coutinho6 David A. Menassa4,7 Leslie Jacobson4 Lieuwe de Haan8 Stephan Ruhrmann9 Gabriele Sachs10 ● ● ● ● ● ● Anita Riecher-Rössler11 Marie-Odile Krebs12 Paul Amminger13,14 Birte Glenthøj15 Neus Barrantes-Vidal16 ● ● ● ● ● Jim van Os17,18 Bart P. F. Rutten 18 Rodrigo A. Bressan19 Mark van der Gaag 20 Robert Yolken21 ● ● ● ● ● Matthew Hotopf22 Lucia Valmaggia23 James Stone 1,2,3,24 Anthony S. David1,25 The EUGEI High-Risk Study ● ● ● ● Philip McGuire1,2,3 Received: 21 April 2019 / Revised: 28 July 2020 / Accepted: 21 September 2020 / Published online: 19 October 2020 © The Author(s) 2020. This article is published with open access Abstract Serum neuronal autoantibodies, such as those to the NMDA receptor (NMDAR), are detectable in a subgroup of patients 1234567890();,: 1234567890();,: with psychotic disorders. It is not known if they are present before the onset of psychosis or whether they are associated with particular clinical features or outcomes. In a case–control study, sera from 254 subjects at clinical high risk (CHR) for psychosis and 116 healthy volunteers were tested for antibodies against multiple neuronal antigens implicated in CNS autoimmune disorders, using fixed and live cell-based assays (CBAs). Within the CHR group, the relationship between NMDAR antibodies and symptoms, cognitive function and clinical outcomes over 24 month follow-up was examined. CHR subjects were not more frequently seropositive for neuronal autoantibodies than controls (8.3% vs. 5.2%; OR = 1.50; 95% CI: 0.58–3.90). The NMDAR was the most common target antigen and NMDAR IgGs were more sensitively detected with live versus fixed CBAs (p < 0.001). Preliminary phenotypic analyses revealed that within the CHR sample, the NMDAR antibody seropositive subjects had higher levels of current depression, performed worse on the Rey Auditory Verbal Learning Task (p < 0.05), and had a markedly lower IQ (p < 0.01). NMDAR IgGs were not more frequent in subjects who later became psychotic than those who did not. NMDAR antibody serostatus and titre was associated with poorer levels of functioning at follow-up (p < 0.05) and the presence of a neuronal autoantibody was associated with larger amygdala volumes (p < 0.05). Altogether, these findings demonstrate that NMDAR autoantibodies are detectable in a subgroup of CHR subjects at equal rates to controls. In the CHR group, they are associated with affective psychopathology, impairments in verbal memory, and overall cognitive function: these findings are qualitatively and individually similar to core features of autoimmune encephalitis and/or animal models of NMDAR antibody-mediated CNS disease. Overall the current work supports further evaluation of NMDAR autoantibodies as a possible prognostic biomarker and aetiological factor in a subset of people already meeting CHR criteria. Introduction Neuronal surface autoantibodies (NSAbs) have been iden- tified in blood samples from a minority of patients with Members of the EU-GEI High-Risk Study Group are listed below psychotic disorders and several other unrelated neurological Acknowledgements. and psychiatric conditions [1–3]. These autoantibodies tar- Supplementary information The online version of this article (https:// get cell-surface proteins, such as the NMDA receptor doi.org/10.1038/s41380-020-00899-w) contains supplementary (NMDAR), gamma-aminobutyric acidA (GABAA) or material, which is available to authorised users. GABAB receptor, α-amino-3-hydroxy-5-methyl-4- * Thomas A. Pollak isoxazolepropionic acid (AMPA) receptor, leucine-rich thomas.pollak@kcl.ac.uk glioma-inactivated 1 (LGI1) or contactin-associated Extended author information available on the last page of the article protein-like 2 (CASPR2) or intracellular proteins, such Clinical, cognitive and neuroanatomical associations of serum NMDAR autoantibodies in people at. . . 2591 as glutamic acid decarboxylase (GAD). When present in disorder. In all, 10–35% of patients with CHR may subse- cerebrospinal fluid (CSF), these IgG-subclass cell surface- quently develop psychosis, with the proportion varying targeting antibodies are considered pathogenic and associate between samples [32, 33], and there is an unmet clinical with severe autoimmune encephalopathy syndromes, which need for biomarkers that can predict which individuals will can present with varied psychiatric features, particularly in later become psychotic [34]. Therefore, we sought to the case of NMDAR antibody encephalitis, plus impaired address this by investigating various NSAbs in people who cognitive function and pathological changes centred around are at CHR for psychosis and examining the relationship limbic brain regions [4–8]. Estimates of the prevalence of between NMDAR antibodies and clinical outcomes, cog- NSAbs in patients with psychosis vary depending on the nitive function and psychopathology in CHR subjects. age, condition studied and the assay used. Live cell-based assays (CBAs) consistently detect higher rates than fixed CBAs, and fixed CBA positivity rates vary widely Materials and methods [1, 9–11]. Therefore, the clinical relevance of these serum- only antibodies remains unclear. Samples At present, there are no data on whether NSAbs are present prior to the onset of psychosis and, if so, whether Sera were taken from 254 subjects from the EU-GEI high- they predict clinical outcomes. People at clinical high risk risk study who met the PACE criteria for the CHR state (CHR) for psychosis experience subthreshold psychotic [35], and from 116 healthy controls (HC). CHR subjects symptoms, display memory impairments and show altera- were recruited from 11 centres (London, Amsterdam, The tions in the structure of limbic brain areas [12–14], findings Hague, Copenhagen, Vienna, Basel, Cologne, Melbourne, that show a partial resemblance to those seen in NMDAR Paris, Barcelona, and São Paulo) and were reassessed at a antibody-mediated encephalitis [4, 15–17]. timepoint as close to 12 and 24 months as possible. Outside of the autoimmune encephalitis context there is Exclusion criteria were presence of a current or past psy- increasing evidence that NMDAR antibodies may be asso- chotic disorder, symptoms relevant for inclusion being ciated with cognitive impairment (e.g., in cancer [18, 19], or explained by a medical disorder or drugs or alcohol following viral encephalitis [20]) but this association has dependency or IQ < 60. not been examined in the context of psychosis-spectrum One-hundred sixteen HC subjects, frequency matched by disorders. In patients who have developed psychosis, most age and sex, were recruited from the EU-GEI high-risk studies have not found clinically significant differences in study (n = 55), and the South East London Community symptom profile between seropositive and seronegative Health (SELCOH) study [36] (n = 61). HCs from the EU- patients [3, 9, 10, 21], raising potential doubts over their GEI study were excluded if they met criteria for the CHR pathogenic relevance. However, in vitro, serum NMDAR state, while HCs from SELCOH were excluded if they antibodies from patients with psychosis disrupt receptor screened positive on the Psychosis Screening Questionnaire dynamics [22, 23] and function [9, 24], suggesting that the [37]. Ethical approval for the study was obtained from the antibodies have pathogenic potential at a molecular level. local research ethics committees at each of the sites. All It is known that NR1-specific intrathecal B cells—which participants provided written informed consent. have not been reported in psychiatric disorders—are a All serum samples were stored at −80 degrees Celsius hallmark of NMDAR antibody encephalitis [25] and that until assayed and tested blind to case–control status. circulating B cells can produce these antibodies [26]. The clinical significance of serum autoantibodies may depend Phenotype data on the presence of additional factors, such as blood–brain barrier (BBB) disruption [9, 27], which could allow them All subjects completed a sociodemographic schedule. access to the CNS. There is widespread evidence for BBB Baseline psychopathology was assessed using the Com- disruption in psychotic disorders [28], although human prehensive Assessment of At-Risk Mental States studies are limited by the requirement for CSF samples or (CAARMS) [38], Brief Psychiatric Rating Scale (BPRS) the use of proxy clinical markers such as a history of brain [39], Scale for the Assessment of Negative Symptoms injury. One such marker is the astrocytic calcium-binding (SANS) [40], Young Mania Rating Scale, (YMRS) [41], protein S100B, which is elevated in schizophrenia [28–31], and the Montgomery-Åsberg Depression Rating Scale and can be measured in serum samples alongside (MADRS) [42]. The Rey Auditory Verbal Learning Task autoantibodies. [43] was used to assess verbal memory function and IQ was If autoantibodies to the NMDAR play an aetiological estimated using the shortened Wechsler Adult Intelligence role in some cases of psychosis, one would expect that they Scale III [44]. Overall social functioning was assessed using would be present before the full clinical expression of the the Global Assessment of Functioning (GAF) scale. 2592 T. A. Pollak et al. Participants were invited for face-to-face follow-up meet- preprocessed using Freesurfer version 6.0.0, using a stan- ings at 1 year and 2 years after baseline. Where face-to-face dard pipeline for quantitation of subcortical volumes meetings were not possible, participants were followed up (described in full on the FreeSurferWiki page (https://surfer. for 2 years using available clinical records, and this follow- nmr.mgh.harvard.edu/fswiki). First, intensity correction and up was extended when additional clinical data was skull stripping were performed, followed by segmentation available. of grey and white matter along with segmentation of sub- cortical structures in which subject scans are warped to a Immunoassays training set atlas and individual voxels labelled using a Gaussian classifier analysis, yielding maps of deep grey Sera were tested for antibodies of the IgG, IgA and IgM matter, white matter, and CSF structures. subtypes to 32 different substrates (see Supplementary Analyses were performed on Freesurfer’s subcortical Appendix 1 for a full list), by indirect immunofluorescence segmentation output volumes and estimates of total intra- on cell-based assays using fixed recombinant HEK293 cells cranial volumes. Analysis was a priori restricted to the and on frozen sections of rodent and monkey brain tissue hippocampus and the amygdala (both bilaterally), as these (hippocampus and cerebellum) at Euroimmun AG, Lübeck, are the areas that are predominantly affected in autoimmune Germany, as described previously [45]). encephalopathies associated with NSAbs, and these areas Live CBA autoantibody testing for the detection of IgG have been independently linked to the later onset of psy- antibodies to NMDAR was performed using HEK293 cells chosis in neuroimaging studies of clinical high-risk popu- (European Collection of Authenticated Cell Cultures lations [49, 50]. (ECACC), Public Health England) at the Nuffield Depart- ment of Clinical Neurosciences, University of Oxford as Statistical analysis described previously [46] with some modifications. Speci- fically, intensity of fluorescence of each sample was rated Statistical analyses were performed using SPSS 23. on a semi-quantitative visual scoring system of 0–4. Scores Demographic variables were compared between cases and of 0 and 1 were rated as negative, 1.5 as low positive and controls and between seropositive and seronegative CHR 2–4 as positive. All positive samples were confirmed at subjects with independent samples t-tests and dilutions of 1:20, 1:100 and 1:500 using an IgG-Fc specific Mann–Whitney U-tests for continuous variables, and chi secondary antibody (goat antihuman Fc IgG; Thermo Sci- square and Fisher’s exact tests for categorical data. Based entific 31125), then tested for non-specific binding using on previous work on neuronal autoantibody prevalence in HEK293 cells that had been transfected with another anti- first episode psychosis populations [10], sample sizes were gen (dopamine D2 receptor). The titre of the antibody was selected to have statistical power to detect differences. given as the dilution of serum giving a score of 1. IgG- For symptom and cognitive scales analyses of variance positive samples were tested in the Oxford laboratory for was performed within the General Linear Model with binding to live, cultured rat hippocampal neurons as NMDAR antibody status (positive or negative), gender and described previously [47]. All cell lines are regularly tested ethnicity as factors and with age as covariate, after homo- for Mycoplasma contamination using immunofluorescence. geneity of variance was confirmed. Given the polymorphic Levels of S100B, a putative marker of blood–brain bar- psychiatric presentation in NMDAR antibody encephalitis rier disruption, were assessed using a chemiluminescence [7, 51] we aimed to look at associations with a broad range assay (Diasorin). High-sensitivity C-reactive protein of psychopathology rather than restricting analyses to any (hsCRP) was measured as a general marker of systemic one symptom domain or dimension. In addition to total inflammation, using high-sensitivity enzyme immunoassays scores for the CAARMS (whole-instrument score), YMRS, as previously described [48]. MADRS and SANS, a positive symptoms factor was computed based on the BPRS components analysis of Neuroimaging analysis Dingemans et al. [52]. For dichotomous outcome measures (transition, non- MR Images were acquired using the ADNI-2 T1 weighted remission from CHR state), logistic regression was per- volumetric sequence (http://adni.loni.usc.edu/methods/ formed with NMDAR antibody status (positive or documents/mri-protocols/) from 159 CHR subjects on 3T negative), age, gender and ethnicity as covariates. scanners. The ADNI-2 sequence has been specifically To compare subcortical ‘limbic’ volumes between ser- developed for multicentre studies and has been optimised to opositive and seronegative CHR subjects multivariate ana- give similar contrast across MRI scanners. Nevertheless, the lysis of variance with NSAb status (positive or negative), scanner was modelled as a factor in the statistical analysis to MRI site, gender and ethnicity as factors, and total intra- control for this. T1 weighted volumes were and cranial volume (TICV) and age as covariates, was Clinical, cognitive and neuroanatomical associations of serum NMDAR autoantibodies in people at. . . 2593 performed in the General Linear Model. All main effects inflammation as indexed by hsCRP (p = 0.048). Smoking were included in our models and the analysis of interactions and antipsychotic usage were more frequent in cases than in was a priori limited to the interaction between NSAb status controls. and S100B status. Significance threshold was set to p < 0.05; mean ± sd Prevalence of NSAbs detected using commercial, presented. Owing to the exploratory nature of the analyses, fixed CBA and live CBA significance values are given for two-tailed tests where applicable and uncorrected for multiple comparisons. Using commercial, fixed assays, NSAbs were detected in 21 (8.3%) CHR subjects and 6 (5.2%) of matched, healthy controls (OR = 1.50; 95% CI: 0.58–3.90; p = 0.401). Pre- Results valence did not differ by site (p = 0.800). IgG was the most commonly detected antibody isotype (48.1%) followed by Participants IgA (33.3%) and IgM (22.2%). Isotype distribution did not differ between CHRs and HCs (IgG: p = 1.000; IgA: p = Demographic and clinical information is presented in 0.283; IgM: p = 0.670). See Fig. 1c for distribution of all Table 1. Cases and controls differed in ethnicity, with fewer NSAb targets. NMDAR was the most commonly detected black participants in the CHR group (8.7%) than in the antigen in the CHR group (n = 11 [4.3%]; comprising healthy control group (20.7%). The groups did not differ on IgG = 1 [0.4%], IgA = 6 [2.4%] and IgM = 4 [1.6%]) and any other demographic variable. There was no difference in in the HC group (n = 2 [1.7%]; comprising IgG = 1 [0.9%], mean serum S100B level between CHR subjects and con- IgA = 0 and IgM = 1 [0.9%]) (Fig. 1d). There was no dif- trols (p = 0.833); CHR subjects had slightly higher baseline ference in overall NMDAR antibody seroprevalence Table 1 Demographic/basic clinical information and NSAb serostatus by group. Total CHR cohort Total HC cohort p (n = 254) (n = 116) Age 22.70 ± 5.00 (range 14 to 45) 23.54 ± 3.38 (range 17 to 34) 0.057 Sex [M: n (%)] 136 (53.5) 60 (51.7) 0.822 Ethnicity [n (%)] White 178 (70.4), black 22 (8.7), White 76 (65.5), black 24 0.003 other 53 (20.9) (20.7), other 16 (13.8) Smokera [n (%)] 133 (53.4) 31 (26.7) <0.001 BMI 24.31 ± 5.30 23.70 ± 4.44 0.288 Current antipsychotic use 24 (9.4) 0 (0) <0.001 [n (%)] S100B (ug/L)b 0.040 ± 0.049 0.050 ± 0.071 0.833 CRP (mg/L)b 1.23 ± 1.08 0.95 ± 0.95 0.048 Any antibody (fixed CBA) 21 (8.3) 6 (5.2) 0.390 [n (%)] IgG [n (%)] 9 (3.5) 4 (3.4) 1.000 IgA [n (%)] 8 (3.1) 1 (0.9) 0.283 IgM [n (%)] 5 (2.0) 1 (0.9) 0.670 NMDAR antibody (fixed 11 (4.3) 2 (1.7) 0.360 CBA) [n (%)] NMDAR IgG [n (%)] 1 (0.4) 1 (0.9) 0.529 NMDAR IgA [n (%)] 6 (2.4) 0 (0.0) 0.183 NMDAR IgM [n (%)] 4 (1.6) 1 (0.9) 1.000 NMDAR IgG (live CBA) 13 (5.1) 6 (5.2) 1.000 [n (%)] Bold values indicate statistical significance p < 0.05. CHR clinical high risk, HC healthy control, CBA cell-based assay, S100B S100 calcium-binding protein B, CRP C-reactive protein, NMDAR N- methyl-d-aspartate receptor. a Data available for 365 of 370 subjects. b Adjusted for BMI, gender, age, ethnicity and smoking status. 2594 T. A. Pollak et al. Fig. 1 NSAb seroprevalence and associations of seropositivity in binding to cultured hippocampal neurons. Cell nuclei are DAPI- CHR and HCs—antibodies detected using fixed assay. a Examples labelled; human IgG is labelled green using a fluorescent secondary of seropositive immunoassays. Top: fixed CBA seropositive for antibody (AlexaFluor 488 mouse anti-human IgG1; Invitrogen) and NMDAR antibody IgG (CHR subject); human IgG is labelled green hippocampal neurons are labelled red using mouse MAP2 antibodies using a fluorescent secondary antibody. Bottom: immunohistochem- (Monoclonal-Anti-MAP2, clone HM-2, Sigma Aldrich). Scalebar = istry showing IgG binding to fixed and permeabilised rat hippocampus 25 µm. c Distribution of antigen target by group. d Distribution of (CHR subject seronegative for other specific antigenic targets); human antibody isotype by group. e Distribution of antibody titre by group IgG is labelled green using a fluorescent secondary antibody. b Con- and assay type (any antibody or NMDAR antibody only); IgG = focal microscope image of IgG from CHR subject showing strong green, IgA = red, IgM = blue. between the CHR and HC groups (p = 0.36). Although differences [10, 11, 22], indicating greater analytical sen- NMDAR antibody isotype distribution was not different sitivity, we retested all samples for NMDAR IgG using a between the two groups (IgG: p = 0.529; IgA: p = 0.183; live CBA (Fig. 2a). Live CBA detected a significantly IgM: p = 1.000), IgA NMDAR antibody seropositivity was higher prevalence of NMDAR IgG antibodies in the total seen in six CHR subjects but not in any HC. cohort than did fixed CBA (19 (5.1%) IgG seropositive On fixed CBAs, titres to specific antigens ranged from subjects on live CBA vs. 2 (0.5%)) on fixed CBA: p = 1:10 to 1:320 in CHR subjects and from 1:32 to 1:320 in 0.0001), but were found similarly in 5.1% of CHR cases HCs (Fig. 1e), but none of these bound brain tissue slices on and 5.2% of HCs (p = 1.000). Prevalence did not differ by immunohistochemistry conducted in the same laboratory. site (p = 0.145). Titres ranged from 1:30 to 1:750 in CHR However, 11 of 12 (91.7%) NSAb IgG seropositive samples subjects and from 1:40 to 1:500 in HCs (Fig. 2b). The two bound cultured rat hippocampal neurons, with three samples subjects who were seropositive for NMDAR IgG on fixed showing strong binding. Serum from seven subjects (five CBA were also seropositive on live CBA. CHR subjects, two HCs) had IgG, IgA or IgM antibodies Seventeen of 19 samples that were NMDAR IgG ser- which bound cerebellum or hippocampus sections (titre opositive on live assays were tested for binding to cultured 1:32 to 1:320) without antigen-specific antibodies detected. rat hippocampal neurons: 10 of 17 (58.8%) samples bound No subjects were seropositive for more than one antigen on neurons, with no samples showing strong binding. All four fixed CBA. samples with a CBA titre of 1:250 or over bound neurons; of Since studies that have used live CBA to detect NMDAR the 13 samples with lower titres, six (46.2%) bound neurons. IgG in psychotic patients and controls have tended to yield IgG titre on CBA and hippocampal binding intensity score greater seropositivity rates [23] and find case–control were positively correlated (r = 0.491; p = 0.048). Clinical, cognitive and neuroanatomical associations of serum NMDAR autoantibodies in people at. . . 2595 Fig. 2 NMDAR IgG seroprevalence and associations of seropositivity— antibodies detected using live assay. a Representative example of NMDAR IgG-positive live CBA. Human IgG is labelled red using a fluorescent secondary antibody (AlexaFluor 568 goat anti-mouse IgG (H + L); Invitrogen) and shown binding to cells that co-express eGFP (green). Scalebar = 25 µm. b Distribution of NMDAR IgG titre by group and assay type (fixed vs. live assay). There were no age, sex, ethnicity or smoking status Scores of functioning were assessed in CHR subjects at differences between seropositive and seronegative subjects the assessment timepoints closest to two years after base- (Table 2). S100B and hsCRP levels did not differ between line. Mean change in GAF scores was negative (indicating seropositive and seronegative CHR subjects, for any NSAb worsening function) for seropositive subjects (live assay), or for NMDAR antibodies measured using either assay and positive (indicating improvement) in seronegative (Table 2). subjects; in multiple regression analyses NMDAR antibody There were no associations between antibody serostatus seropositivity was significantly associated with change in and self-reported medical history including neurological GAF disability scores (Supplementary Table 8). There was disorder, autoimmune disorder, head trauma, cancer, history a significant negative correlation between NMDAR IgG of surgical procedures, CNS infection or serious infection titre (live assay) and change in GAF disability scores (Supplementary Table 1). Notably the only subject that had (Spearman’s rho = −0.542; p = 0.024), with higher titre had a CNS infection (meningitis) was seropositive for subjects showing a more negative change in GAF scores, NMDAR IgG on live assay. indicating deteriorating function. Clinical outcomes of CHR subjects Cognitive associations of NMDAR antibody seropositivity in CHR subjects Of the 254 CHR subjects from whom serum had been collected and tested for NSAbs, 49 (19.3%) developed a Total RAVLT immediate recall score was lower in CHR psychotic illness in the follow-up period. The subjects subjects seropositive for NMDAR antibodies on fixed CBA who transitioned to psychosis did not differ from those compared to seronegative CHR subjects (p = 0.030). These who did not transition in age or gender (Supplementary differences became more substantial after controlling for Table 2). total CAARMS score (p = 0.007), suggesting that the Of the subjects who transitioned to psychosis, 1 (2.0%) relationship with auditory verbal memory was not simply was seropositive for NMDAR antibodies on fixed CBA and secondary to an effect of the severity of psychopathology 3 (6.1%) were seropositive for NMDAR IgG on live assay (Table 3). at baseline; of 205 CHR subjects who did not transition to In order to explore whether the association with cogni- psychosis, 10 (4.9%) were seropositive on fixed assay and tive impairment was specific to verbal memory function, we 10 (4.9%) were seropositive on live assay (Supplementary then compared IQ (estimated using the WAIS-III) in ser- Table 3). In logistic regression analyses, NMDAR antibody opositive and seronegative subjects. Mean WAIS-III IQ in serostatus at baseline did not predict later transition to NMDAR antibody seropositive CHR subjects was almost psychosis or nonremission from the CHR state, regardless 15 points lower than that in seronegative CHR subjects of the assay used (fixed or live CBA) for antibody detection (p = 0.010). This group difference also remained significant (Supplementary Tables 4–7). after controlling for total CAARMS score (p = 0.008) 2596 T. A. Pollak et al. (Table 3). Both effects appeared to be driven by IgA/M 0.418 0.330 0.563 0.826 0.815 0.463 0.506 0.783 p antibodies: after restricting the analysis to IgA/M ser- positive (live CBA) opositive subjects only, both AVLT immediate recall (p = 0.011) and WAIS-III estimated IQ (p = 0.018) remained NMDAR antibody White 11 (57.9), significantly impaired in seropositive subjects. black 3 (15.8), other 5 (26.3) 0.037 ± 0.031 23.79 ± 5.73 23.82 ± 5.10 In fact, within the total CHR sample (n = 254), only 1.17 ± 0.94 11 (57.9) (n = 19) 9 (47.4) 2 (10.5) 9 subjects had an IQ below 70 (the threshold for a diagnosis of intellectual disability) but 3 of these CHR subjects (33%) were seropositive for NMDAR antibodies. Subjects who were NMDAR IgG seropositive on live CBA did not differ negative (live CBA) NMDAR antibody 0.797 White 243 (69.4), from seronegative subjects on measures of cognition black 43 (12.3), other 64 (18.3) including AVLT immediate recall (p = 0.337) and IQ CBA cell-based assay, S100B S100 calcium-binding protein B, CRP C-reactive protein, NMDAR N-methyl-D-aspartate receptor, BMI body mass index. 0.994 0.043 ± 0.058 0.829 22.92 ± 4.50 0.591 24.12 ± 5.02 (p = 0.796). 0.925 1.14 ± 1.05 0.270 188 (53.6) 0.398 155 (44.8) (n = 351) 0.203 22 (6.3) Symptomatic associations of NMDAR antibody seropositivity in CHR subjects p Nine of 13 (69.2%) subjects who were seropositive for positive (fixed CBA) NMDAR antibodies on live assay met DSM-IV criteria for NMDAR antibody current depressive episode, compared to 31.2% of ser- White 10 (76.9), onegative subjects (p = 0.012). Rates of other DSM-IV other 2 (15.4) black 1 (7.7), 22.69 ± 4.57 0.04 ± 0.043 1.28 ± 1.41 diagnoses did not differ according to serostatus. There was (n = 13) 9 (69.2) 4 (30.8) 2 (15.4) no difference in the rates of specific DSM-IV diagnoses 24.94 between NMDAR antibody fixed CBA seropositive and seronegative subjects (Supplementary Table 9). negative (fixed CBA) While symptom severity scores were not significantly NMDAR antibody different between seropositive and seronegative subjects for 0.148 White 44 (68.5), black 45 (12.6), other 67 (18.8) most symptom domains when looking at antibodies detec- 0.613 0.043 ± 0.057 0.458 22.97 ± 4.57 0.991 1.14 ± 1.04 0.163 187 (52.4) 0.233 160 (45.5) ted using either assay, MADRS scores were borderline (n = 357) 0.403 22 (6.2) significantly higher in CHR subjects seropositive on live 0.370 24.07 CBA (p = 0.051), indicating greater depression symptom severity. Across the whole CHR cohort, MADRS scores were weakly correlated with NMDAR IgG (live CBA) titre p (Spearman’s rho = 0.167; p = 0.009). Statistic adjusted for BMI, gender, age, ethnicity and smoking status. Any antibody negative Any antibody positive Table 2 Demographic/basic clinical information by NSAb serostatus. There was no association between NMDAR antibody White 23 (85.2), serostatus and CHR subtype (vulnerability group, atte- other 3 (11.1) 0.038 ± 0.033 black 1 (3.7), nuated psychosis symptoms [APS] or brief limited inter- (fixed CBA) 22.33 ± 4.13 23.19 ± 5.55 1.12 ± 1.21 mittent psychotic symptoms [BLIPS]) (fixed CBA: p = 18 (66.7) (n = 27) 9 (33.3) 3 (11.1) 0.316; live CBA: p = 0.865), although the numbers of subjects in the vulnerability (n = 41) and BLIPS (n = 16) subgroups were small. White 231 (67.5), Data available for 365 of 370 subjects. black 45 (13.2), Association between blood–brain barrier disruption other 66 (19.3) 0.043 ± 0.058 23.01 ± 4.60 24.17 ± 4.98 and symptoms 1.14 ± 1.04 178 (51.9) 155 (45.9) (n = 343) Current antipsychotic 21 (6.1) In order to assess the possible role of blood–brain barrier disruption as a moderating variable for the associations of NMDAR serostatus with symptom measures, we performed Ethnicity [n (%)] a preliminary (due to small n in the seropositive groups) Smokera [n (%)] Sex [M: n (%)] b S100B (ug/L) CRP (mg/L)b analysis of the association between serum S100B levels and use [n (%)] symptom scores in NMDAR antibody seropositive and seronegative groups. In subjects for whom NMDAR IgG BMI Age were detected using live CBA, positive associations b a Clinical, cognitive and neuroanatomical associations of serum NMDAR autoantibodies in people at. . . 2597 Table 3 Cognitive function and psychopathology scores by NMDAR antibody serostatus (descriptive means shown with F and p values from ANOVA). Antibody Antibody F (serostatus) p (serostatus) positive negative NMDAR antibody (IgG, IgA, IgM; fixed assay) AVLT IR total (7:160) 52.00 ± 13.79 63.06 ± 11.90 4.817 0.030 AVLT IR total (CAARMS adjusted) 45.80 ± 10.69 62.50 ± 11.90 7.651 0.007 (5:127) WAIS estimated total IQ (9:222) 84.22 ± 20.35 98.71 ± 16.90 6.687 0.010 WAIS estimated total IQ (CAARMS 83.29 ± 21.62 100.32 ± 16.86 7.134 0.008 adjusted) (7:182) CAARMS total (9:198) 49.00 ± 16.17 49.49 ± 15.00 0.001 0.981 BPRS-positive sx (11:224) 14.91 ± 5.43 13.55 ± 4.54 0.731 0.393 SANS (9:210) 20.44 ± 15.30 15.53 ± 11.74 1.074 0.301 YMRS (11:225) 4.27 ± 5.35 4.00 ± 4.79 0.03 0.957 MADRS (11:230) 20.91 ± 12.98 18.69 ± 9.16 0.650 0.421 NMDAR antibody (IgG; LIVE ASSAY) AVLT IR total (9:158) 58.00 ± 13.90 62.86 ± 12.03 0.929 0.337 AVLT IR total (CAARMS adjusted) 58.00 ± 13.90 62.15 ± 12.13 0.826 0.365 (9:123) WAIS estimated total IQ (12:219) 95.00 ± 19.98 98.32 ± 17.10 0.067 0.796 WAIS estimated total IQ (CAARMS 95.00 ± 19.98 100.01 ± 17.11 0.353 0.553 adjusted) (12:177) CAARMS total (13:194) 51.00 ± 11.90 49.37 ± 15.21 0.197 0.657 BPRS-positive sx (13:222) 14.62 ± 4.82 13.56 ± 4.57 0.683 0.409 SANS (13:206) 15.31 ± 11.39 15.76 ± 11.96 0.058 0.810 YMRS (12:224) 6.50 ± 5.16 3.89 ± 4.76 3.179 0.076 MADRS (13:228) 23.69 ± 9.05 18.51 ± 9.30 3.847 0.051 Bold values indicate statistical significance p < 0.05. Numbers in brackets under variable names are the number of seropositive subjects and seronegative subjects for whom data were available for that variable. BPRS total score excluded from analysis due to heteroscedasticity. NMDAR N-methyl-D-aspartate receptor, AVLT Rey Auditory Verbal Learning Task, IR immediate recall, CAARMS Comprehensive Assessment of At-Risk Mental States, WAIS Wechsler Adult Intelligence Scale III, BPRS Brief Psychiatric Rating Scale, SANS Scale for the Assessment of Negative Symptoms, YMRS Young Mania Rating Scale, MADRS Montgomery-Asberg Depression Rating Scale. between S100B and symptom severity at baseline were visits (i.e., had new neuronal antibodies detectable). Of observed for total BPRS score and for negative symptoms. 11 subjects who were seropositive at baseline, 4 (36.4%) In the seronegative group, modest negative associations were still seropositive on at least one subsequent visit (p = between S100B and symptom severity were observed for 0.018). In 3 of these 4 subjects (75%), NSAbs detectable at multiple symptom scales (Supplementary Table 10). follow-up were to the same antigen and of the same isotype. None of the four subjects with persistent antibodies (i.e., Longitudinal NSAb serostatus NSAbs detectable at baseline and at a subsequent timepoint) became psychotic during follow-up. Only one of 9 subjects One-hundred seventy-nine follow-up serum samples were who developed de novo NSAbs at follow-up became psy- available on 120 CHR subjects (modal interval between chotic: this subject developed NMDAR IgM antibodies baseline and repeat blood sample: 12 months). Baseline (1:10) a year after initial assessment, having been ser- serostatus was predictive of follow-up serostatus, i.e., sub- onegative at baseline; he went on to transition to psychosis a jects who were seropositive at baseline were more likely year later, but no blood samples were available at the time than subjects who were seronegative at baseline to be ser- he became psychotic or afterwards. opositive at follow-up. In total, of 109 CHR subjects who Of five subjects who had NMDAR antibodies detectable were seronegative at baseline and who had blood taken at at baseline and on whom follow-up samples were available, subsequent visits, 9 (8.3%) were seropositive at subsequent two (40%) had NSAbs at one or more follow-up point. One 2598 T. A. Pollak et al. case with NMDAR IgA antibodies (1:32) at baseline had resemblance to (but is less severe than) the pattern of IgG antibodies at visit 2 (12 months) that bound hippo- findings seen in the acute stages of many autoimmune campus and cerebellum (1:100) but not NMDAR-specific encephalopathies. This suggests that NMDAR antibodies, antibodies detectable, and at visit 4 (24 months) had no and possibly other NSAbs, could play an analogous, but not antibodies at all detectable. One case with NMDAR IgM necessarily identical, role in the pathophysiology of some antibodies at baseline (1:100) had NMDAR IgM antibodies patients who present with psychiatric symptoms [59]. This (1:100) detectable at visit 3. Neither of these cases devel- would be consistent with recent evidence that NMDAR oped psychosis. antibodies isolated from psychiatric patients have patho- Of 11 CHR subjects on whom serum was taken at the genicity in vitro and in vivo [9, 22, 24]. time of transition to psychosis, none were seropositive. The reductions in verbal memory performance in NMDAR antibody-positive subjects are reminiscent of Limbic volumes in NSAb seropositive CHR subjects those seen in NMDAR encephalitis, and were of a com- parable magnitude [19]. It is notable, however, that ser- As only a small number of the CHR subjects who were opositive CHR subjects did not show other features of NMDAR antibody seropositive had also had an MRI scan autoimmune encephalopathies, such as seizures or frank (n = 5), we investigated the relationship between auto- movement disorders. Neurological soft signs were not antibodies and MRI measures in subjects who were ser- assessed in the present study, but subtle movement signs opositive for any NSAb (n = 11). CHR subjects who were have been observed in a proportion of CHR subjects seropositive for NSAbs had larger amgydala volumes (p = [60–62] and have been associated with NMDAR antibody 0.048) than seronegative CHR subjects (n = 148). There seropositivity in patients with psychosis [9]. were no group differences in hippocampal volume (Sup- Furthermore, NMDAR antibody encephalitis is classi- plementary Table 11). cally associated with a polymorphic affective psychosis with catatonic features [7], which is broader than the phe- notypic associations of seropositivity that we observed. Discussion However a rodent model involving the intraventricular administration of NMDAR antibodies generated a cogni- Summary of results tively impaired, anxious and depressive phenotype [63], which is consistent with our findings of an association This is the first study of CNS autoantibodies in people at between NMDAR antibodies and MADRS scores and CHR for psychosis. We found an overall seroprevalence of current depressive episodes in CHR subjects. It has been 8.3%, indicating that NSAbs are detectable in a minority of suggested that the multifaceted symptomatology of these individuals. The NMDAR was the most common NMDAR encephalitis represents more than the effects of target antigen and NMDAR IgGs were more sensitively the NMDAR antibody at the synapse, and that additional detected with live versus fixed CBAs. The prevalence was immune and non-immune factors help shape clinical pre- not significantly greater than in HCs, mirroring findings sentation. Most passive immunisation animal models do not from recent studies that have found similar case–control recapitulate a full encephalopathic phenotype, instead seroprevalence rates in patients with psychosis and other showing convergence around a more restricted set of psychiatric and non-psychiatric disorders [1, 21, 53]. symptoms, which could be conceived as a ‘ketamine-like’ NMDAR antibody seropositivity was not associated with synaptopathy [64], without features of frank CNS inflam- the development of a psychotic disorder, which was our mation. If NMDAR antibodies do have a causal role in main outcome of interest, but was associated with a dete- some psychiatric disorders, it is therefore possible that the rioration in disability-associated functional status. Within effects represent more of a synaptopathic than a frank the CHR sample, NMDAR antibody seropositivity was encephalitic process. associated with impaired verbal memory performance, Additionally, the possibility that NMDAR antibodies lower IQ and with depressive symptoms. Exploratory MRI could play a causal role in some CHR cases does not analysis indicated that NSAb seropositivity was associated imply that the overall underlying immunology in these with increased amygdala volume. cases would be the same as in NMDAR encephalitis, a disorder which is characterised by the presence of NR1- Relevance of NSAbs in CHR subjects specific intrathecal B cells. Indeed the importance of BBB disruption in determining the relevance of peripheral The combination of impaired cognitive function, severe NMDAR antibodies in many human and animal studies of affective psychopathology [46, 51, 54] and increased NSAb-mediated psychiatric disorders [9, 27, 64]) sug- volume of limbic brain regions [55–58] bears some gests that intrathecal synthesis may be minimal or absent Clinical, cognitive and neuroanatomical associations of serum NMDAR autoantibodies in people at. . . 2599 in seropositive subjects with isolated psychiatric In total 72.4% of IgG seropositive samples bound presentations. hippocampal neurons; this was more true of IgG ser- Given the associations we report between NSAbs and opositive samples on fixed CBA (91.7%) than on live both amygdala enlargement and affective symptomatology, CBA (58.8%). This is consistent with live CBA poten- it is interesting to note that whereas amygdala volumes are tially detecting antibodies with higher sensitivity than the reduced in schizophrenia and first episode schizophrenic fixed CBA. The fact that in our laboratory not all live psychosis, they are increased in the first episode of affective CBA-positive samples bound rat hippocampal neurons psychoses and other nonschizophrenic psychoses (including (see also ref. [21]) does not undermine the potential brief psychotic disorder) [65]. The increase in amygdala disease-relevance of NMDAR IgG seropositivity, parti- volumes in our NSAb seropositive subjects could represent cularly as these antibodies have pathogenic potential at acute inflammation, as is the case in acute limbic ence- the level of the NMDAR [22, 23]. Immunocytochemistry phalitis. Importantly, in the post-acute phase of autoimmune with cultured neurons, as with all assays, has a detection encephalitis there is evidence of volume loss in these same threshold, and this should be understood in the context regions, which is in turn linked to the severity of persisting that CBAs overexpress the antigen at the surface but in deficits [5, 15, 17]. If the neuroanatomical associations of immunocytochemistry expression levels are broadly NSAb seropositivity truly represent a similar mechanism to physiological. Therefore, it might be expected that some that of autoimmune encephalitis, one testable hypothesis is lower titre samples that are positive on live assay will not that these brain regions will show accelerated atrophy or demonstrate binding to neurons. The observed correlation volume loss at follow-up scans in seropositive CHR between NMDAR IgG live CBA titre and hippocampal subjects. binding intensity is suggestive that live CBA is measuring Prüss, Finke and colleagues have highlighted associa- genuine IgG binding, and indeed the live CBA threshold tions between IgA NMDAR antibodies and cognitive above which NMDAR IgG-positive samples will reliably impairments in multiple clinical contexts [2, 18, 66]. In the bind to neurons appears to be 1:250. present study, 6 CHR subjects had IgA NMDAR anti- bodies, compared with no HCs. It is thus possible that IgA The role of the blood–brain barrier NMDAR antibodies drive the cognitive impairment observed in our seropositive subjects. IgM, and many IgAs, Ehrenreich and colleagues have stressed that for peripheral are not frequently reported in the CSF, although this may NSAbs to be clinically relevant there may also need to be potentially reflect a bias towards the measurement of IgG disruption of the BBB [64]. This has been demonstrated in only. While IgA can exist as monomers in blood, IgMs are mice with an ApoE-/- haplotype [68] and in humans with a multimerised antibodies, potentially making their crossing ‘history of neurotrauma’ or obstetric complications as of the blood–brain barrier less likely than IgGs. medical proxies [9]. A parsimonious interpretation is that We observed that in CHR individuals NMDAR antibody when the integrity of the BBB is compromised, NSAbs serostatus fluctuates over time. Pan and colleagues recently have greater access to the CNS and are more likely to reported similar findings in animals and human subjects disrupt brain function [28]. However, it is also possible that (following stroke): a majority of individuals remained ser- serum NSAbs are peripheral biomarkers of a central dys- onegative, and a minority either remained seropositive or function that is accompanied by BBB disruption, such as acquired de novo antibodies at 1–3-year follow-up [67]. neuroinflammation. Consistent with the role of the BBB as What causes these fluctuations, and whether they have any proposed by Ehrenreich and colleagues, in a supplemental direct clinical consequences, has not been established analysis we observed positive associations between S100B (although there is some emerging evidence for chronic levels (a widely used proxy for BBB disruption [28, 31]) stress as a potential inducer of NMDAR antibodies). and symptom scores in subjects who were seropositive for NMDAR antibodies on live CBA; these associations were Comparison of cell-based assays in the opposite direction in subjects without detectable serum antibodies. Given the small numbers of seropositive In the largest direct assay comparison reported to date, we subjects, however, this finding mandates replication in a found that live CBA detects a nine times higher prevalence larger sample, potentially with an additional marker of BBB of IgG NMDAR antibodies in serum than fixed CBA, disruption. The ‘gold standard’ of BBB quantification in confirming a previous report in a first episode psychosis humans is the CSF-serum albumin quotient, which has cohort [23]. In this earlier report, even in the absence of consistently been found to be abnormal in psychosis [28]. seropositivity on fixed CBA, these IgG both targeted and However, this requires a lumbar puncture, which is not were capable of altering the surface dynamics of the usually part of the clinical assessment of patients with NMDAR, suggesting pathogenic potential. psychosis, let alone individuals at CHR [69]. A less 2600 T. A. Pollak et al. invasive method for BBB quantification in these individuals CHR state. As Castillo-Gomez and colleagues have demon- is the use of dynamic contrast-enhanced MRI [70]. strated, in conditions in which antibody production is puta- tively lower than in autoimmune encephalitis the brain may Antibodies as biomarkers in the CHR state act as ‘immunoprecipitator’ of CSF NSAbs, and NSAbs may not be detectable in CSF, despite having functional con- Separately, or in addition to, the question of the patho- sequences in the brain [68]. Although we studied a large genicity of NMDAR antibodies in the CHR state, the pre- sample of CHR subjects, because only a minority were sent study indicates a potential role for these antibodies as antibody-positive, the numbers of subjects in some of the biomarkers of poor cognition, depressive symptomatology comparisons was small and hence we did not perform cor- and/or outcome in the CHR state. rections for multiple comparisons. We also cannot exclude In CHR subjects, severe psychopathology and cognitive the possibility that we failed to detect some differences impairments, and alterations in limbic brain volumes at because of limited statistical power. This issue may be presentation are associated with an increased risk of later addressed by pooling data from different CHR cohorts, as in transition to psychosis [12–14, 71, 72]. Impaired verbal the Harmony collaboration (https://www.pronia.eu/news- memory performance in CHR individuals is one of the pronia/news-detail/tx_news/initiation-of-the-multi-center- strongest cognitive predictors of subsequent transition to collaboration-harmony-in-october-copy-1/). psychosis [73, 74]. Low IQ has also been identified as a predictor of conversion to psychosis in CHR subjects [75], Acknowledgements We also acknowledge the SELCOH study team and the participation of all South London and the Maudsley NIHR and is an established risk factor for psychosis in the general BioResource volunteers, and thank the South London and the population [76]. Our study suggests, complementing similar Maudsley BioResource staff for their help with volunteer recruitment. findings in other patient groups [2, 18, 20, 66], that NMDAR antibodies may have a role as a biomarker of The EUGEI High-Risk Study Group Philip McGuire1,2,3, Lucia cognitive impairment in CHR subjects. Since we observed Valmaggia23, Matthew J. Kempton1,2, Maria Calem1, Stefania Tognin1, Gemma Modinos1, Lieuwe de Haan8, Mark van der Gaag20, that antibody serostatus fluctuates over time, an important Eva Velthorst26,27, Tamar C. Kraan26, Daniella S. van Dam26, Nadine question is whether cognitive status fluctuates in association Burger28, Barnaby Nelson13, Patrick McGorry13, Paul Amminger13,14, with serostatus. This latter might be expected if NMDAR Christos Pantelis13, Athena Politis13, Joanne Goodall13, Anita Riecher- antibodies represent state (rather than trait) biomarkers for Rössler11, Stefan Borgwardt29,30, Sarah Ittig31, Erich Studerus32, Renata Smieskova31, Rodrigo A. Bressan19, Ary Gadelha19, Elisa poor cognition: indeed most passive transfer models suggest Brietzke33, Graccielle Asevedo19, Elson Asevedo19, Andre Zugman19, that the antibody effects are reversible. However a different Neus Barrantes-Vidal16, Araceli Rosa34, Anna Racioppi35, Manel relationship might be expected if serostatus is in fact a Monsonet35, Lídia Hinojosa-Marqués35, Thomas R. Kwapil36, marker of poor cognitive performance due to the impact of Mathilde Kazes37, Claire Daban37, Julie Bourgin37, Olivier Gay37, Célia Mam-Lam-Fook37, Marie-Odile Krebs12, Dorte Nordholm38, an as-yet undetermined confounder. Lasse Randers38, Kristine Krakauer38, Louise Glenthøj38, Birte In our study, NMDAR antibodies did not predict tran- Glenthøj15, Merete Nordentoft38, Stephan Ruhrmann9, Dominika sition to psychosis, as defined by operationalised criteria, or Gebhard9, Julia Arnhold39, Joachim Klosterkötter9, Gabriele Sachs10, nonremission from the CHR state, despite all but one Iris Lasser10, Bernadette Winklbaur10, Philippe A. Delespaul18,40, Bart P.F. Rutten18, Jim van Os17,18 NMDAR IgG seropositive subjects still meeting CHR cri- teria at follow-up. NMDAR antibody seropositivity was 26 AMC, Academic Psychiatric Centre, Department Early Psychosis, however associated with a deterioration in disability- Meibergdreef 5, 1105 AZ Amsterdam, The Netherlands; 27Icahn associated functioning, short of transition, and indeed School of Medicine at Mount Sinai, Department of Psychiatry, 1425 greater antibody titre was associated with greater dete- Madison Avenue, New York, NY 10029, USA; 28Parnassia Psychia- tric Institute, Department of Psychosis Research, Zoutkeetsingel 40, rioration in function, overall suggesting that NMDAR 2512 HN The Hague, The Netherlands; 29Department of Psychiatry, antibody serostatus should be further evaluated as a pre- University of Basel, Basel, Switzerland; 30Department of Psychiatry dictive marker of functional outcome. and Psychotherapy, University of Lübeck, Lübeck, Germany; 31Uni- versity Psychiatric Hospital, Wilhelm Klein-Strasse 27, 4002 Basel, Switzerland; 32University of Basel, Department of Psychology, Limitations Division of Personality and Developmental Psychology, Basel, Switzerland; 33Depto Psiquiatria, Escola Paulista de Medicina, We measured NSAbs in serum. Assessing NSAbs in CSF Universidade Federal de São Paulo—UNIFESP, Sao Paulo, Brazil; 34 may provide a better way of investigating their role in some Departament de Biologia Evolutiva, Ecologia i Ciències Ambientals (Universitat de Barcelona), Barcelona, Spain; 35Departament de Psi- autoimmune CNS disorders, but lumbar puncture is more cologia Clínica i de la Salut (Universitat Autònoma de Barcelona), invasive and is not usually carried out in CHR subjects. Barcelona, Spain; 36Department of Psychology, University of Illinois However, while CSF NSAbs are important for the diagnosis at Urbana-Champaign, Champaign, IL, USA; 37University Paris Des- of autoimmune encephalitis, this does not imply that serum cartes, Hôpital Sainte-Anne, C’JAAD, Service Hospitalo-Uni- versitaire, Inserm U894, Institut de Psychiatrie (CNRS 3557), NSAbs are irrelevant in psychiatric conditions such as the Clinical, cognitive and neuroanatomical associations of serum NMDAR autoantibodies in people at. . . 2601 Paris, France; 38Mental Health Center Copenhagen and Center for study received financial support by French Government’s Agence Clinical Intervention and Neuropsychiatric Schizophrenia Research, Nationale pour la Recherche (ANR, 08-MNP-007) and by the French CINS, Mental Health Center Glostrup, Mental Health Services in the Health Ministry’s Programme Hospitalier de Recherche Clinique Capital Region of Copenhagen, University of Copenhagen, (PHRC, AOM-07-118, ‘Influence of cannabis psychopathological Copenhagen, Denmark; 39Psyberlin, Berlin, Germany; 40Mondriaan outcome in At risk mental state’ (ICAAR study)). The Sainte-Anne Mental Health Trust, Heerlen, The Netherlands hospital center promoted the study. Additional financial support was obtained from the Institut National de la Santé et de la Recherche Author contributions TP and PM conceived the study. TP performed Médicale (INSERM, recurrent funding and fellowships) and by Fon- analyses and wrote the manuscript. MK coordinated data collection dation Pierre Deniker. AV and SRI are coapplicants and receive roy- across EU-GEI sites, analysed neuroimaging data and contributed to alties on patent application WO/2010/046716 (U.K. patent no., PCT/ the manuscript. CI, DAM, EC, LJ, RY performed analyses and edited GB2009/051441) entitled ‘Neurological Autoimmune Disorders’. The the manuscript. AV and SI supervised analyses, assisted with inter- patent has been licensed to Euroimmun AG for the development of pretation and edited the manuscript. LdH, SR, GS, AR-R, M-OK, BN, assays for LGI1 and other VGKC-complex antibodies. MOK partici- BG, NVB, BR, JvO, RAB, MvdG, LV, MH, SH led participant pated on boards (Roche, Janssen) and received financial support from recruitment at their sites and edited the manuscript. JS, AD and PM Janssen, Otsuka Lundbeck alliance, EIsai for conference or dis- oversaw data collection, assisted with study design and data inter- semination initiatives. pretation and edited the manuscript. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Compliance with ethical standards Open Access This article is licensed under a Creative Commons Conflict of interest Dr. Pollak was supported by a clinical research Attribution 4.0 International License, which permits use, sharing, training fellowship grant from the Wellcome Trust (no 105758/Z/14/ adaptation, distribution and reproduction in any medium or format, as Z). The European Network of National Schizophrenia Networks long as you give appropriate credit to the original author(s) and the Studying Gene- Environment Interactions (EU-GEI) Project is funded source, provide a link to the Creative Commons license, and indicate if by grant agreement HEALTH-F2–2010–241909 (Project EU-GEI) changes were made. The images or other third party material in this from the European Community’s Seventh Framework Programme. article are included in the article’s Creative Commons license, unless Additional support was provided by a Medical Research Council indicated otherwise in a credit line to the material. If material is not Fellowship to M Kempton (grant MR/J008915/1). Dr. Irani is sup- included in the article’s Creative Commons license and your intended ported by a Wellcome Trust Intermediate Fellowship (104079/Z/14/Z), use is not permitted by statutory regulation or exceeds the permitted the British Medical Association Research Grants: Vera Down (2013) use, you will need to obtain permission directly from the copyright and Margaret Temple (2017). Dr. Nelson was supported by a NHMRC holder. To view a copy of this license, visit http://creativecommons. Senior Research Fellowship. Dr. Glenthøj is the leader of a Lundbeck org/licenses/by/4.0/. Foundation Centre of Excellence for Clinical Intervention and Neu- ropsychiatric Schizophrenia Research (CINS), which is partially financed by an independent grant from the Lundbeck Foundation References based on international review and partially financed by the Mental Health Services in the Capital Region of Denmark, the University of 1. Dahm L, Ott C, Steiner J, Stepniak B, Teegen B, Saschenbrecker Copenhagen, and other foundations. Her group has also received a S, et al. Seroprevalence of autoantibodies against brain antigens in research grant from Lundbeck A/S for another independent health and disease. Ann Neurol. 2014;76:82–94. investigator-initiated study. All grants are the property of the Mental 2. Doss S, Wandinger KP, Hyman BT, Panzer JA, Synofzik M, Health Services in the Capital Region of Denmark and administrated Dickerson B, et al. High prevalence of NMDA receptor IgA/IgM by them. She has no other conflicts to disclose. Dr. De Haan has no antibodies in different dementia types. Ann Clin Transl Neurol. conflicts to disclose. Dr. Ruhrmann is a project consultant of Boeh- 2014;1:822–32. ringer Ingelheim. Bart PF Rutten is funded by a VIDI award number 3. Zandi MS, Irani SR, Lang B, Waters P, Jones PB, McKenna P, 91718336 from the Netherlands Scientific Organisation. Barnaby et al. Disease-relevant autoantibodies in first episode schizo- Nelson was supported by an NHMRC Senior Research Fellowship phrenia. J Neurol. 2011;258:686–8. (1137687). ASD is supported by the NIHR University College London 4. Pollak TA, Beck K, Irani SR, Howes OD, David AS, McGuire Hospital (UCLH) Biomedical Research Centre. Dr. Barrantes-Vidal PK. Autoantibodies to central nervous system neuronal surface was supported by the Ministerio de Ciencia, Innovación e Uni- antigens: psychiatric symptoms and psychopharmacological versidades (PSI2017-87512-C2-1-R), Generalitat de Catalunya implications. Psychopharmacology. 2016;233:1605–21. (2017SGR1612 and ICREA Academia Award). Dr Bressan has 5. Heine J, Pruss H, Bartsch T, Ploner CJ, Paul F, Finke C. Imaging received research grants from Janssen and the governmental funding of autoimmune encephalitis-Relevance for clinical practice and research agencies: CAPES, CNPq and FAPESP (2016/022465 and hippocampal function. Neuroscience. 2015;309:68–83. 2011/50740-5) and has participated in speaker bureaus for Janssen and 6. Bien CG, Vincent A, Barnett MH, Becker AJ, Blumcke I, Graus Sanofi-Aventis, all outside the submitted work. The authors F, et al. Immunopathology of autoantibody-associated encephali- acknowledge financial support from the National Institute for Health tides: clues for pathogenesis. Brain: a J Neurol. Research (NIHR) Biomedical Research Centres at the South London 2012;135:1622–38. and Maudsley NHS Foundation Trust, University College London 7. Al-Diwani A, Handel A, Townsend L, Pollak T, Leite MI, Hospital and the John Radcliffe Hospital NHS Foundation Trust, Harrison PJ, et al. The psychopathology of NMDAR-antibody Oxford. The views expressed are those of the author(s) and not encephalitis in adults: a systematic review and phenotypic analysis necessarily those of the NHS, the NIHR or the Department of Health of individual patient data. Lancet Psychiatry. 2019;6:235–46. and Social Care. Additional support was provided by the Maudsley 8. Dalmau J, Gleichman AJ, Hughes EG, Rossi JE, Peng X, Lai M, Charity (Grant Ref. 980) and Guy’s and St. Thomas’s Charity (Grant et al. Anti-NMDA-receptor encephalitis: case series and analysis Ref. STR130505). These funders had no involvement in study design, of the effects of antibodies. Lancet Neurol. 2008;7:1091–8. data collection, analysis or the decision to submit for publication. The 2602 T. A. Pollak et al. 9. Hammer C, Stepniak B, Schneider A, Papiol S, Tantra M, 26. Makuch M, Wilson R, Al-Diwani A, Varley J, Kienzler AK, Begemann M, et al. Neuropsychiatric disease relevance of circu- Taylor J, et al. N-methyl-D-aspartate receptor antibody production lating anti-NMDA receptor autoantibodies depends on blood- from germinal center reactions: therapeutic implications. Ann brain barrier integrity. Mol Psychiatry. 2014;19:1143–9. Neurol. 2018;83:553–61. 10. Lennox BR, Palmer-Cooper EC, Pollak T, Hainsworth J, Marks J, 27. Hammer C, Zerche M, Schneider A, Begemann M, Nave KA, Jacobson L, et al. Prevalence and clinical characteristics of serum Ehrenreich H. Apolipoprotein E4 carrier status plus circulating neuronal cell surface antibodies in first-episode psychosis: a case- anti-NMDAR1 autoantibodies: association with schizoaffective control study. Lancet Psychiatry. 2017;4:42–8. disorder. Mol Psychiatry. 2014;19:1054–6. 11. Pathmanandavel K, Starling J, Merheb V, Ramanathan S, Sinmaz 28. Pollak TA, Drndarski S, Stone JM, David AS, McGuire P, Abbott N, Dale RC, et al. Antibodies to surface dopamine-2 receptor and NJ. The blood-brain barrier in psychosis. Lancet Psychiatry. N-methyl-D-aspartate receptor in the first episode of acute psy- 2017;5:79–92. chosis in children. Biol Psychiatry. 2015;77:537–47. 29. Schumberg K, Polyakova M, Steiner J, Schroeter ML. Serum 12. Fusar-Poli P, Deste G, Smieskova R, Barlati S, Yung AR, Howes S100B is related to illness duration and clinical symptoms in O, et al. Cognitive functioning in prodromal psychosis: a meta- schizophrenia-A meta-regression analysis. Front Cell Neurosci. analysis. Arch Gen Psychiatry. 2012;69:562–71. 2016;10:46. 13. Perkins DO, Jeffries CD, Cornblatt BA, Woods SW, Addington J, 30. Aleksovska K, Leoncini E, Bonassi S, Cesario A, Boccia S, Bearden CE, et al. Severity of thought disorder predicts psychosis Frustaci A. Systematic review and meta-analysis of circulating in persons at clinical high-risk. Schizophrenia Res. S100B blood levels in schizophrenia. PLoS ONE. 2014;9: 2015;169:169–77. e106342. 14. Mechelli A, Riecher-Rossler A, Meisenzahl EM, Tognin S, Wood 31. Strathmann FG, Schulte S, Goerl K, Petron DJ. Blood-based SJ, Borgwardt SJ, et al. Neuroanatomical abnormalities that pre- biomarkers for traumatic brain injury: evaluation of research date the onset of psychosis: a multicenter study. Arch Gen Psy- approaches, available methods and potential utility from the chiatry. 2011;68:489–95. clinician and clinical laboratory perspectives. Clin Biochem. 15. Finke C, Kopp UA, Pajkert A, Behrens JR, Leypoldt F, Wuerfel 2014;47:876–88. JT, et al. Structural hippocampal damage following anti-N- 32. Fusar-Poli P, Cappucciati M, Borgwardt S, Woods SW, methyl-D-aspartate receptor encephalitis. Biol Psychiatry. Addington J, Nelson B, et al. Heterogeneity of psychosis risk 2016;79:727–34. within individuals at clinical high risk: a meta-analytical stratifi- 16. Finke C, Kopp UA, Pruss H, Dalmau J, Wandinger KP, Ploner CJ. cation. JAMA Psychiatry. 2016;73:113–20. Cognitive deficits following anti-NMDA receptor encephalitis. J 33. Fusar-Poli P, Bonoldi I, Yung AR, Borgwardt S, Kempton MJ, Neurol Neurosurg Psychiatry. 2012;83:195–8. Valmaggia L, et al. Predicting psychosis: meta-analysis of tran- 17. Finke C, Kopp UA, Scheel M, Pech LM, Soemmer C, Schlichting sition outcomes in individuals at high clinical risk. Arch Gen J, et al. Functional and structural brain changes in anti-N-methyl- Psychiatry. 2012;69:220–9. D-aspartate receptor encephalitis. Ann Neurol. 2013;74:284–96. 34. McGuire P, Sato JR, Mechelli A, Jackowski A, Bressan RA, 18. Bartels F, Stronisch T, Farmer K, Rentzsch K, Kiecker F, Finke C. Zugman A. Can neuroimaging be used to predict the onset of Neuronal autoantibodies associated with cognitive impairment in psychosis? Lancet Psychiatry. 2015;2:1117–22. melanoma patients. Ann Oncol: Off J Eur Soc Med Oncol/ESMO. 35. Kraan TC, Velthorst E, Themmen M, Valmaggia L, Kempton MJ, 2019;30:823–9. McGuire P, et al. Child maltreatment and clinical outcome in 19. Finke C, Bartels F, Lutt A, Pruss H, Harms L. High prevalence of individuals at ultra-high risk for psychosis in the EU-GEI high risk neuronal surface autoantibodies associated with cognitive deficits study. Schizophrenia Bull. 2018;44:584–92. in cancer patients. J Neurol. 2017;264:1968–77. 36. Hatch SL, Frissa S, Verdecchia M, Stewart R, Fear NT, Reich- 20. Westman G, Studahl M, Ahlm C, Eriksson BM, Persson B, enberg A, et al. Identifying socio-demographic and socioeconomic Ronnelid J, et al. N-methyl-d-aspartate receptor autoimmunity determinants of health inequalities in a diverse London commu- affects cognitive performance in herpes simplex encephalitis. Clin nity: the South East London Community Health (SELCoH) study. Microbiol Infect. 2016;22:934–40. BMC Public Health. 2011;11:861. 21. Gaughran F, Lally J, Beck K, McCormack R, Gardner-Sood P, 37. Bebbington P, Nayani T. The psychosis screening questionnaire. Coutinho E, et al. Brain-relevant antibodies in first-episode psy- Int J Methods Psychiatr Res. 1995;5:11–9. chosis: a matched case-control study. Psychol Med. 38. Yung AR, Yuen HP, McGorry PD, Phillips LJ, Kelly D, Dell’Olio 2017;48:1257–63. M, et al. Mapping the onset of psychosis: the comprehensive 22. Jezequel J, Johansson EM, Dupuis JP, Rogemond V, Grea H, assessment of at-risk mental states. Aust NZ J Psychiatry. Kellermayer B, et al. Dynamic disorganization of synaptic NMDA 2005;39:964–71. receptors triggered by autoantibodies from psychotic patients. Nat 39. Lukoff D, Liberman RP, Nuechterlein KH. Symptom monitoring Commun. 2017;8:1791. in the rehabilitation of schizophrenic patients. Schizophrenia Bull. 23. Jezequel J, Rogemond V, Pollak T, Lepleux M, Jacobson L, Grea 1986;12:578–602. H, et al. Cell- and single molecule-based methods to detect anti-N- 40. Andreasen NC. The scale for the assessment of negative symp- methyl-D-aspartate receptor autoantibodies in patients with first- toms (SANS): conceptual and theoretical foundations. Br J Psy- episode psychosis from the OPTiMiSE project. Biol Psychiatry. chiatry Suppl. 1989;7:49–58. 2017;82:766–72. 41. Young RC, Biggs JT, Ziegler VE, Meyer DA. A rating scale for 24. Castillo-Gomez E, Oliveira B, Tapken D, Bertrand S, Klein- mania: reliability, validity and sensitivity. Br J Psychiatry: J Ment Schmidt C, Pan H, et al. All naturally occurring autoantibodies Sci. 1978;133:429–35. against the NMDA receptor subunit NR1 have pathogenic 42. Montgomery SA, Asberg M. A new depression scale designed to potential irrespective of epitope and immunoglobulin class. Mol be sensitive to change. Br J Psychiatry: J Ment Sci. Psychiatry. 2016;22:1776–84. 1979;134:382–9. 25. Kreye J, Wenke NK, Chayka M, Leubner J, Murugan R, Maier N, 43. Schmidt M. Rey auditory and verbal learning test: a handbook. et al. Human cerebrospinal fluid monoclonal N-methyl-D- Los Angeles, CA: Western Psychological Services; 1996. aspartate receptor autoantibodies are sufficient for encephalitis 44. Wechsler D. Manual for the Wechsler Adult Intelligence Scale— pathogenesis. Brain: J Neurol. 2016;139:2641–52. Revised. New York: Psychological Corporation; 1981. Clinical, cognitive and neuroanatomical associations of serum NMDAR autoantibodies in people at. . . 2603 45. Wandinger KP, Saschenbrecker S, Stoecker W, Dalmau J. Anti- progression in youth at ultrahigh risk for psychosis. Schizophrenia NMDA-receptor encephalitis: a severe, multistage, treatable dis- Res. 2015;162:86–9. order presenting with psychosis. J Neuroimmunol. 62. Tamagni C, Studerus E, Gschwandtner U, Aston J, Borgwardt S, 2011;231:86–91. Riecher-Rossler A. Are neurological soft signs pre-existing mar- 46. Irani SR, Bera K, Waters P, Zuliani L, Maxwell S, Zandi MS, kers in individuals with an at-risk mental state for psychosis? et al. N-methyl-D-aspartate antibody encephalitis: temporal pro- Psychiatry Res. 2013;210:427–31. gression of clinical and paraclinical observations in a pre- 63. Planaguma J, Leypoldt F, Mannara F, Gutierrez-Cuesta J, Martin- dominantly non-paraneoplastic disorder of both sexes. Brain: J Garcia E, Aguilar E, et al. Human N-methyl D-aspartate receptor Neurol. 2010;133:1655–67. antibodies alter memory and behaviour in mice. Brain: J Neurol. 47. Irani SR, Alexander S, Waters P, Kleopa KA, Pettingill P, Zuliani 2015;138:94–109. L, et al. Antibodies to Kv1 potassium channel-complex proteins 64. Ehrenreich H. Autoantibodies against the N-methyl-d-aspartate leucine-rich, glioma inactivated 1 protein and contactin-associated receptor subunit NR1: untangling apparent inconsistencies for protein-2 in limbic encephalitis, Morvan’s syndrome and acquired clinical practice. Front Immunol. 2017;8:181. neuromyotonia. Brain. 2010;133:2734–48. 65. Velakoulis D, Wood SJ, Wong MT, McGorry PD, Yung A, 48. Dickerson F, Stallings C, Origoni A, Vaughan C, Khushalani S, Phillips L, et al. Hippocampal and amygdala volumes according to Yang S, et al. C-reactive protein is elevated in schizophrenia. psychosis stage and diagnosis: a magnetic resonance imaging Schizophrenia Res. 2013;143:198–202. study of chronic schizophrenia, first-episode psychosis, and ultra- 49. Allen P, Chaddock CA, Egerton A, Howes OD, Bonoldi I, Zelaya high-risk individuals. Arch Gen Psychiatry. 2006;63:139–49. F, et al. Resting hyperperfusion of the hippocampus, midbrain, 66. Pruss H, Holtje M, Maier N, Gomez A, Buchert R, Harms L, et al. and basal ganglia in people at high risk for psychosis. Am J IgA NMDA receptor antibodies are markers of synaptic immunity Psychiatry. 2016;173:392–9. in slow cognitive impairment. Neurology. 2012;78:1743–53. 50. Witthaus H, Mendes U, Brune M, Ozgurdal S, Bohner G, 67. Pan H, Steixner-Kumar AA, Seelbach A, Deutsch N, Ronnenberg Gudlowski Y, et al. Hippocampal subdivision and amygdalar A, Tapken D, et al. Multiple inducers and novel roles of auto- volumes in patients in an at-risk mental state for schizophrenia. J antibodies against the obligatory NMDAR subunit NR1: a trans- Psychiatry Neurosci. 2010;35:33–40. lational study from chronic life stress to brain injury. Mol 51. Gibson LL, Pollak TA, Blackman G, Thornton M, Moran N, Psychiatry. 2020. [Epub ahead of print]. David AS. The psychiatric phenotype of anti-NMDA receptor 68. Castillo-Gomez E, Kastner A, Steiner J, Schneider A, Hettling B, encephalitis. J Neuropsychiatry Clin Neurosci. 2019;31:70–9. Poggi G, et al. The brain as immunoprecipitator of serum auto- 52. Dingemans PM, Linszen DH, Lenior ME, Smeets RM. Compo- antibodies against N-Methyl-D-aspartate receptor subunit NR1. nent structure of the expanded Brief Psychiatric Rating Scale Ann Neurol. 2016;79:144–51. (BPRS-E). Psychopharmacology. 1995;122:263–7. 69. Pollak TA, Lennox BR. Time for a change of practice: the real- 53. Steiner J, Teegen B, Schiltz K, Bernstein HG, Stoecker W, world value of testing for neuronal autoantibodies in acute first- Bogerts B. Prevalence of N-methyl-D-aspartate receptor auto- episode psychosis. BJPsych Open. 2018;4:262–4. antibodies in the peripheral blood: healthy control samples 70. Heye AK, Culling RD, Valdes Hernandez Mdel C, Thrippleton revisited. JAMA Psychiatry. 2014;71:838–9. MJ, Wardlaw JM. Assessment of blood-brain barrier disruption 54. Kayser MS, Titulaer MJ, Gresa-Arribas N, Dalmau J. Frequency using dynamic contrast-enhanced MRI. A systematic review. and characteristics of isolated psychiatric episodes in anti-N- Neuroimage Clin. 2014;6:262–74. methyl-d-aspartate receptor encephalitis. JAMA Neurol. 71. Lieberman JA, Girgis RR, Brucato G, Moore H, Provenzano F, 2013;70:1133–9. Kegeles L, et al. Hippocampal dysfunction in the pathophysiology 55. Wagner J, Weber B, Elger CE. Early and chronic gray matter of schizophrenia: a selective review and hypothesis for early volume changes in limbic encephalitis revealed by voxel-based detection and intervention. Mol Psychiatry. 2018;23:1764–72. morphometry. Epilepsia. 2015;56:754–61. 72. Phillips LJ, Velakoulis D, Pantelis C, Wood S, Yuen HP, Yung 56. Wagner J, Witt JA, Helmstaedter C, Malter MP, Weber B, Elger AR, et al. Non-reduction in hippocampal volume is associated CE. Automated volumetry of the mesiotemporal structures in with higher risk of psychosis. Schizophrenia Res. 2002; antibody-associated limbic encephalitis. J Neurol Neurosurg 58:145–58. Psychiatry. 2015;86:735–42. 73. Simon AE, Gradel M, Cattapan-Ludewig K, Gruber K, Ballinari 57. Urbach H, Soeder BM, Jeub M, Klockgether T, Meyer B, Bien P, Roth B, et al. Cognitive functioning in at-risk mental states for CG. Serial MRI of limbic encephalitis. Neuroradiology psychosis and 2-year clinical outcome. Schizophrenia Res. 2006;48:380–6. 2012;142:108–15. 58. Taniguchi G, Fuse H, Okamura Y, Mori H, Kondo S, Kasai K, 74. Hauser M, Zhang JP, Sheridan EM, Burdick KE, Mogil R, et al. Improvement in anti-N-methyl-d-aspartate receptor Kane JM, et al. Neuropsychological test performance to enhance antibody-mediated temporal lobe epilepsy with amygdala enlar- identification of subjects at clinical high risk for psychosis gement without immunotherapy. Epilepsy Behav Case Rep. and to be most promising for predictive algorithms for 2018;10:96–8. conversion to psychosis: a meta-analysis. J Clin Psychiatry. 59. Al-Diwani A, Pollak T, Langford A, Lennox B. Synaptic and 2017;78:e28–40. neuronal autoantibody-associated psychiatric syndromes 75. Ziermans T, de Wit S, Schothorst P, Sprong M, van Engeland H, (SNAps): Controversies and Hypotheses. Front Psychiatry. Kahn R, et al. Neurocognitive and clinical predictors of long-term 2017;8:13. outcome in adolescents at ultra-high risk for psychosis: a 6-year 60. Dean DJ, Mittal VA. Spontaneous parkinsonisms and striatal follow-up. PLoS ONE. 2014;9:e93994. impairment in neuroleptic free youth at ultrahigh risk for psy- 76. Doody GA, Johnstone EC, Sanderson TL, Owens DG, Muir WJ. chosis. NPJ Schizophr. 2015;1:14006. ‘Pfropfschizophrenie’ revisited. Schizophrenia in people with 61. Dean DJ, Kent JS, Bernard JA, Orr JM, Gupta T, Pelletier-Baldelli mild learning disability. Br J Psychiatry: J Ment Sci. 1998; A, et al. Increased postural sway predicts negative symptom 173:145–53. 2604 T. A. Pollak et al. Affiliations Thomas A. Pollak 1,2,3 Matthew J. Kempton1,2 Conrad Iyegbe1 Angela Vincent 4 Sarosh R. Irani5 ● ● ● ● ● Ester Coutinho6 David A. Menassa4,7 Leslie Jacobson4 Lieuwe de Haan8 Stephan Ruhrmann9 Gabriele Sachs10 ● ● ● ● ● ● Anita Riecher-Rössler11 Marie-Odile Krebs12 Paul Amminger13,14 Birte Glenthøj15 Neus Barrantes-Vidal16 ● ● ● ● ● Jim van Os17,18 Bart P. F. Rutten 18 Rodrigo A. Bressan19 Mark van der Gaag 20 Robert Yolken21 ● ● ● ● ● Matthew Hotopf22 Lucia Valmaggia23 James Stone 1,2,3,24 Anthony S. David1,25 The EUGEI High-Risk Study ● ● ● ● Philip McGuire1,2,3 ● 1 15 Department of Psychosis Studies, Institute of Psychiatry, Centre for Neuropsychiatric Schizophrenia Research (CNSR) & Psychology and Neuroscience, King’s Health Partners, King’s Centre for Clinical Intervention and Neuropsychiatric College London, London, UK Schizophrenia Research (CINS), Mental Health Centre Glostrup, 2 University of Copenhagen, Glostrup, Denmark Precision Psychiatry Cluster, National Institute for Health 16 Research Biomedical Research Centre at South London and Departament de Psicologia Clínica I de la Salut (Universitat Maudsley NHS Foundation Trust and King’s College London, Autònoma de Barcelona), Fundació Sanitària Sant Pere Claver London, UK (Spain), Spanish Mental Health Research Network 3 (CIBERSAM), Barcelona, Spain South London and Maudsley NHS Foundation Trust, London, UK 17 4 Department of Psychiatry, UMC Utrecht Brain Centre, Nuffield Department of Clinical Neurosciences, John Radcliffe Utrecht, The Netherlands Hospital, Oxford, UK 18 5 Department of Psychiatry and Neuropsychology, School for Autoimmune Neurology Group, Nuffield Department of Clinical Mental Health and Neuroscience, Mental Health Research and Neurosciences, John Radcliffe Hospital, Oxford, UK Teaching Network, Maastricht University Medical Centre, 6 P.O. Box 616, 6200 MD 464, Maastricht, The Netherlands Division of Neuroscience, Institute of Psychiatry, Psychology and Neuroscience, King’s Health Partners, King’s College London, 19 PRISMA Clinic, Y-MIND—Institute for Prevention of Mental London, UK Disorders (PRISMA); LiNC—Lab Interdisciplinar Neurociências 7 Clínicas, Escola Paulista de Medicina, Universidade Federal de Biological Sciences, Faculty of Environmental and Life Sciences, São Paulo—UNIFESP, Sao Paulo, Brazil University of Southampton, Southampton, UK 20 8 VU University, Faculty of Behavioural and Movement Sciences, Amsterdam UMC, University of Amsterdam, Psychiatry, Department of Clinical Psychology and Amsterdam Public Mental Department Early Psychosis, Meibergdreef 9, Amsterdam, The Health research institute, van der Boechorststraat 7, 1081 BT Netherlands Amsterdam, The Netherlands 9 Department of Psychiatry and Psychotherapy, Faculty of Medicine 21 Stanley Neurovirology Division. Department of Pediatrics, Johns and University Hospital, University of Cologne, Hopkins School of Medicine, Baltimore, MD, USA Cologne, Germany 22 10 Department of Psychological Medicine, Institute of Psychiatry, Medical University of Vienna, Department of Psychiatry and Psychology and Neuroscience, King’s College London, Psychotherapy, Vienna, Austria London, UK 11 Medical Faculty, University of Basel, Basel, Switzerland 23 Department of Psychology, Institute of Psychiatry, Psychology & 12 University of Paris, GHU-Paris, Sainte-Anne, C’JAAD, Neuroscience, King’s College London, London, UK Hospitalo-Universitaire department SHU, Inserm U1266, Institut 24 Department of Neuroimaging, Centre for Neuroimaging Sciences, de Psychiatrie (CNRS 3557), Paris, France Institute of Psychiatry, Psychology and Neuroscience, King’s 13 Centre for Youth Mental Health, University of Melbourne, 35 College London, London, UK Poplar Road (Locked Bag 10), Parkville, VIC 4853052, Australia 25 Institute of Mental Health, University College London, Maple 14 Orygen, The National Centre of Excellence in Youth Mental House, 149 Tottenham Court Road, London W1T 7NF, UK Health, The University of Melbourne, Parkville, VIC, Australia