We retrospectively screened between 2017 and 2020 a cohort of 167 patients (Figure 1) from the Department of Psychiatry and Psychotherapy, University Medical Center Göttingen. All patients presented different psychiatric diagnoses and were screened for specific autoantibodies (see section autoantibody determination) because we suspected that their psychiatric symptoms had an organic origin. Organic origin has been suspected due to various features associated with the presentation of clinical features with a fulminant course, the acuity of symptoms, the rapid progression of symptoms, the atypical severity of cognitive dysfunction associated with other symptoms or alone, or the presence of autoimmune indicators recently presented in a review (4). Severe cognitive dysfunction including orientation and attention functions, which is not explained otherwise, is a strong indicator than the red flag (5) of an autoimmune base of symptoms and thus another reason for screening neuronal autoantibodies in patients. Our cohort was divided into two subgroups: (1) consisted of 35 serum antibody-positive and (2) contained only 132 serum antibody-negative patients (Figure 1). Our inclusion criterion was a psychiatric disorder for which we had determined autoantibodies for differential diagnostic reasons. We did not exclude patients with comorbidities or additional neurologic conditions. All patients recruited in 2020 had been screened for a CoV-SARS-2 infection. Patients recruited in 2016–2019 did not undergo testing for CoV-SARS-2. Their psychopathology was retrospectively assessed via two psychopathological classification systems by studying patient files. Patient data were classified by one rater (IMG) according to the AMPD system (9) and the HiTOP classification system (10, 11). An ultimate total of 154 patients who had undergone complete psychopathological investigation according to AMDP were included in our study. Among the total 154 patients, our groups consisted of 35 serum antibody-positive and 119 serum antibody-negative patients (Figure 1). We used the STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) guidelines to organize and describe our observational study better. Our investigations were approved by our local ethics committee and are in accordance with the Declaration of Helsinki.

Psychopathology was retrospectively assessed from patient files by one rater (IMG). Psychopathological features according to AMDP were assessed as single items from the following domains: disturbances of consciousness, orientation, memory and attention, formal thought disorder, worries and compulsions, delusions, disorders of perception, hallucinations, ego disturbances, disturbances of affect, disorders of drive and psychomotor activity, circadian disturbances, and other disorders. Furthermore, we applied spectra from the HiTOP classification system entailing somatoform, internalizing, thought disorder, disinhibited externalizing, antagonistic externalizing, and detachment spectra. Sexual problems, eating pathologies, fear, distress, mania, substance abuse, and antisocial behavior functioned as HiTOP factors. The two HiTOP superspectra emotional dysfunction (12) and psychosis (13) were also evaluated. The superspectra emotional dysfunction consists of the single spectra somatoform and internalizing, whereas the superspectra psychosis comprises the single spectra detachment and thought disorder. For AMDP domains, we established a total score consisting of the AMDP items from one AMDP domain that were present in the patients. Thus, the total AMDP domain's scoring range is different [disturbances of consciousness (0–4), disturbances of orientation (0–4), disturbances of memory and attention (0–6), formal thought disorder (0–12), worries and compulsions (0–6), delusions (0–6), disorders of perception (0–6), hallucinations (0–6), ego disturbances (0–6), disturbances of affect (0–21), disorders of drive and psychomotor activity (0–9), circadian disturbances, and other disturbances (0–10)]. Moreover, we set up an HiTOP sum score for the superspectra emotional dysfunction and psychosis consisting of the presence of individual spectra present in the patients [psychosis (0–2), emotional dysfunction (0–2)]. Psychopathology was retrospectively assessed from patient files by one rater (IMG).

Our cerebrospinal fluid samples resulted from lumbar puncture. CSF and blood samples were taken at the same time and were processed according to the standard protocol in the neurochemistry laboratory in the University Medical Center Göttingen's Neurology Department. Freshly extracted cerebrospinal fluid is used, as in vitro cytolysis is so rapid. All biomaterial samples were processed within 1–2 h after they were received to ensure accurate cell count samples. The CSF samples were centrifuged to separate cells from CSF at these settings (700 revolutions correspond to 105 × g, 10 min, room temperature). The degeneration marker (ptau181, t-tau) and Reiber scheme were determined on the same day. However, the serum was centrifugated different settings (5,000 revolutions corresponding to 4,528 × g for 5 min at 20°). All serum samples were stored until they were sent as aliquots (500 μl serum) to the Euroimmun laboratory for the detection of autoantibodies. About 300 μl CSF aliquots were also sent to the Euroimmun laboratory. All patients' neural autoantibodies were assessed by the Clinical Immunological Laboratory Prof. Stöcker. We sought diverse neural autoantibodies in peripheral blood (PB) and CSF entailing autoantibodies against membrane cell surface targets [α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors 1/2 (AMPAR1/2), Aquaporin 4, contactin-associated protein 2 (CASPR2), dipeptidyl-peptidase-like 6 protein (DPPX), gamma aminobutyric acid B1/2 receptor (GABAB1/2R), glutamic acid decarboxylase (GAD65), leucine-rich glioma inactivated protein 1 (LGI1), and N-methyl-D-aspartate receptor (NMDAR)] and autoantibodies against intracellular often paraneoplastic antigen targets [autoantibodies against intracellular autoantibodies: amphiphysin, CV2, glutamic acid decarboxylase (GAD65), HuD, Ma1/ Ma2, neurochondrin (NC), Ri, TR, Yo, and Zic4]. Euroline immunoblot and cell-based assays were utilized to determine autoantibodies. Neural autoantibodies were assessed in a semiquantitative scale possessing three intensity levels: 1 (mild intensity), 2 (moderate intensity), and 3 (strong intensity). The groups were divided into patients with neural autoantibodies and those without them in serum and or cerebrospinal fluid (CSF). The CSF was analyzed in the Neurochemistry Laboratory, University Medical Center Göttingen.

Data were analyzed with SPSS (IBM, SPSS, version 28, Ehningen, Germany). Graphs were made with Sigma Plot (Version 11, Palo Alto, United States). Mean and standard error of the mean were calculated for gender and age. In addition, for the investigation of age differences between men and women, the Mann–Whitney U-test was used. The psychopathological items and domains from AMDP as well as factors, superspectra, and spectra from HiTOP classification were compared between autoantibody-positive and autoantibody-negative groups via Fisher's exact test with no correction for multiple testing because of the data exploratory nature and our small sample. We conducted no power analysis beforehand, as our retrospective cohort analysis has exploratory character. We used Spearman's rho correlation including correcting for multiple testing via Bonferroni to investigate the relationship between the semiquantitative autoantibody-intensity score and psychopathological items sum score of the AMDP system and HiTOP classification. To estimate effect sizes and our results' robustness, we also conducted Bayesian testing and calculated the effect sizes via Cramer's V. A p-level of < 0.05 was considered as significant.

We enrolled 154 patients in this study with the following diagnostic spectrum shown in Table 1. In 35 of 154 (22.7%) psychiatric patients, we detected a neural antibody in the peripheral blood (Ab+ PB), and 13 of 154 (8.5%) had antibodies detected in CSF (Ab+ CSF). We divided the patients in a group of 35 psychiatric patients with peripheral blood autoantibodies (PsychAb + PB) and 119 psychiatric patients with no PB autoantibodies (PsychAb – PB). A patient classification found that 27 out of 35 (78%) of serum autoantibody-positive patients had possible autoimmune encephalitis and six out of 35 (17%) had definitive autoimmune encephalitis according to the criteria of Graus et al. (1), while seven out of 35 (20%) showed autoimmune psychosis according to the criteria of Pollak et al. (14) (Table 2). A lumbar puncture could not be performed in four patients with serum autoantibody detection and in three patients without serum autoantibody detection. Our patient groups consisted of various psychiatric diagnoses whose main diagnoses did not differ in their frequency between groups (Table 1). PsychAb + PB patients were diagnosed with disorders belonging to organic, including symptomatic, mental disorders (60%, n = 24), schizophrenia, schizotypical and delusional disorders (3%, n = 1), mood disorders (5%, n = 14), anxiety, dissociative, stress-related, somatoform, and other nonpsychotic mental disorders (4%, n = 2). On the contrary, PsychAb – PB patients were diagnosed with organic, including symptomatic, mental disorders (45%, n = 53), schizophrenia, schizotypical and delusional disorders (18%, n = 15), mood disorders (35%, n = 29), anxiety, dissociative, stress-related, somatoform and other nonpsychotic mental disorders (4%, n = 2), and adult personality and behavior disorders (2%, n = 2) (Table 1). The psychiatric diagnoses were made after the differential diagnostic approach was carried out, including the determination of neuronal autoantibodies. We formed a group of 13 of 154 (8.5%) patients who presented CSF neural autoantibodies (PsychAb + CSF) (Table 2) and a corresponding group of 141 patients with no CSF autoantibodies (PsychAb – CSF). See for autoantibodies of the PsychAb + PB group, serum and CSF group as well as CSF-only group autoantibodies (Table 2). In total, our cohort consisted of 80 women and 74 men. The men had a mean age of 54 years (54.3 ± 2.1 years), and the women were 61 years old (61.2 ± 1.7 years). However, the age of men and women did not differ throughout our entire cohort, the PsychAb + PB and PsychAb – PB patients. Age and gender did not differ between groups (PsychAb + PB vs. PsychAb – PB, and PsychAb + CSF vs. PsychAb – CSF) (Table 3). Several psychiatric syndromes did not differ among the (PsychAb + PB vs. PsychAb – PB, and PsychAb + CSF vs. PsychAb – CSF) groups (Table 3). However, many more psychiatric patients with a manic syndrome according to the AMDP classification presented neural autoantibodies (PsychAb+ or PsychAb + CSF) (8.6%) compared to neural autoantibody-negative psychiatric patients (PsychAb – PB or PsychAb – CSF) (0.8%) (p < 0.05) (Table 3).

We did not find relevant correlations between neural autoantibody intensity in serum (n = 27), neural autoantibody intensity in CSF (n = 5), combined neural autoantibody intensity in serum and CSF (n = 9), and the respective AMDP domain sum score, HiTOP spectrum score, HiTOP subfactor score, and HiTOP superspectrum scores.

To look for differences between autoantibody subtypes, we compared psychiatric patients with autoantibodies against membrane surface targets (n = 15) with autoantibodies against intracellular antigens (n = 17). We deleted these patients with the detection of not one, but various neural autoantibodies (n = 3). Surprisingly, we found more attention and memory impairment in patients with intracellular autoantibodies than in patients with antibodies to membrane surface antigens (Table 6A, Fisher's exact test: p < 0.005, Bayesian BF Poisson = 55.30). In affective psychopathology assessed by AMDP, it was surprisingly noticeable that patients with membrane surface autoantibodies did not have blunted affect in contrast to patients with autoantibodies against intracellular antigens (Table 6B, Fisher's exact test: p < 0.005, Bayesian BF Poisson = 3.26). However, no differences between the autoantibody subgroups with regard to the HiTOP spectra and subfactors could be demonstrated (Tables 6C,D). Furthermore, we performed an analysis of only NMDAR antibodies compared to PsychAb – PB patients as NMDAR depicted the largest size of one autoantibody subtype. We found less attention and memory impairment in patients with NMDAR autoantibodies (n = 3) than in PsychAb – PB patients (n = 96) (Supplementary Table 1A, Fisher's exact test: p < 0.05, Bayesian BF Poisson = 2.09). In addition, patients with NMDAR showed fewer formal thinking disorders (n = 2) compared to PsychAb-PB patients (n = 83) (Supplementary Table 1A, Fisher's exact test: S < 0.05, Bayesian BF Poisson = 2.09). After an in-depth analysis of affective psychopathology, it has been shown that patients with NMDAR autoantibodies had a greater percentage of the feeling of loss of vitality (Supplementary Table 1B, Fisher's exact test: p < 0.05, Bayesian BF Poisson = 0.85) and ambivalence (Supplementary Table 1B, Fisher's exact test: p < 0.05, Bayesian BF Poisson = 0.58) compared to PsychAb – PB patients. No relevant differences emerged between NMDAR antibody-positive patients and PsychAb – PB patients concerning HiTOP subfactors and spectra (Supplementary Tables 1C,D).

Our subgroup analysis showed that patients with intracellular autoantibodies were much more likely to have attention and memory impairments compared with patients with membrane surface autoantibodies. These results are consistent with previous studies showing cognitive impairment in patients with subsets of intracellular antibodies such as recoverin antibodies, published recently (16, 17). It is likely that temporal aspects of the presence of autoantibodies and the development of cognitive dysfunction are the reason that patients with autoantibodies to intracellular antigens are more affected than those with autoantibodies to membrane surface antigens. In addition, we examined the largest group of autoantibodies to a single antigen relative to our study rather than the literature: patients with NMDAR autoantibodies (n = 7). Analyses of the homogenous group of patients with NMDAR antibodies compared to antibody-negative patients also showed that although there were changes in psychopathology (less memory and attention disturbances, less formal thought disorder, more ambivalence, and more feelings of loss of vitality), these cannot be regarded as meaningful due to the small number of cases.

No differing psychopathological features emerged when symptoms were analyzed via the HiTOP classification being in agreement with a recent controversy on HiTOP measurement system (18, 19) postulating disadvantages in HiTOP classification system in its current version for clinical use (18). The AMDP thus seems more suitable for assessing psychiatric symptoms potentially prominent in autoantibody-associated psychiatric syndromes. However, we could not confirm any relationship between neural autoantibody intensity and any main psychopathological feature in any of the measurement systems, as was recently demonstrated in terms of cognitive dysfunction in a subgroup of patients with psychiatric syndromes and associated autoantibodies—namely autoimmune psychosis—revealing a correlation between NMDAR autoantibody titer and cognitive dysfunction (20). Thus, our findings show that the AMDP measurement scale is superior to the HiTOP scale in detecting differences in the psychopathology of psychiatric patients between those with and those without autoantibodies. We believe that a detailed, single symptom-oriented approach is more fruitful than an overstretched grouping of symptom domains, as small differences might prove to be extremely helpful for differentiating disease entities (autoantibody-related vs. non-autoantibody-related).

The evidence of prominent orientation deficits has been reported in autoantibody-associated psychiatric syndromes (17, 21). The limbic system plays a key role in spatial orientation, which is often affected in the autoimmune encephalitis called limbic encephalitis, but it might be also functionally impaired in autoantibody-associated psychiatric syndromes leading to orientation deficits. The possibility of a different affective psychopathology in autoantibody-associated psychiatric syndromes should be kept in mind: A blunted affect is more pronounced in antibody-negative than antibody-positive psychiatric syndromes, underlying a somewhat maintained sensation capability in an autoantibody-associated psychiatric syndrome. The psychopathology of 103 patients with Sjögren's syndrome was assessed extensively in association with autoantibodies against neuropeptides in comparison with 110 healthy controls (22). In their study, the patients with Sjögren's syndrome and autoantibodies against neuropeptides revealed different degrees of psychopathological characteristics, indicating that assessing personality traits is a potentially promising research field to target also in conjunction with autoantibody-associated psychiatric syndromes. This point was not investigated in detail in our current study. Thus, we confirm in our study patients a specific psychopathology revealing more orientation deficits, less blunted affect, and more loss of vitality, which point toward a phenotype entailing lost energy and orientation deficits in autoantibody-associated psychiatric syndromes. Our findings with mild psychopathological abnormalities in individual AMDP items, but not major AMDP domains in psychiatric patients with autoantibodies, should be investigated in further studies with larger numbers of patients with specific autoantibody classes. Another important aspect to consider is that differences in psychopathology might not depend so much on the underlying psychiatric syndrome, as we found no differences between main syndromes apart from the manic syndrome of antibody-positive patients compared to antibody-negative patients. Thus, it is reasonable to measure a wide spectrum of psychopathologic items to be certain to detect small differences that might be of additional differential diagnostic value.

The heterogeneous composition of our autoantibody-positive group is a potential limitation as the psychopathology associated with specific autoantibodies might be unique. Further studies with homogeneous autoantibody groups are needed to clarify this issue. Furthermore, neural autoantibody groups should be examined in separate groups, and their psychiatric diagnoses should likewise be differentiated to discover whether certain autoantibodies play a role in specific psychiatric diseases. It would also be worthwhile to address this question in a future study with more homogeneous psychiatric-symptom constellations, and similar groups of neural autoantibodies (chosen for their supposed mechanism of action, and intra- or extracellular location of neural autoantibodies) in the hope that an immunotherapy approach would help to alleviate the symptoms of patients with a probable autoimmune basis. Another limitation is that patients with comorbidities such as other non-psychiatric disorders, developmental anomalies, or neurologic conditions were not excluded, as these are possible confounding factors. Furthermore, we cannot exclude that other factors like substance abuse or use, or anxiety might have an impact on psychopathology, not just the neural autoantibody biomarker. We thus recommend considering all these limitations in a future study to exclude confounding factors in a large cohort of autoantibody-positive psychiatric patients. Note that we refrained from multivariant testing and refer to our study's exploratory nature (we also dispensed with multivariant models because of its exploratory nature). A MANOVA is a suitable method that could be useful for a larger cohort to detect differences in psychopathology between antibody-positive and antibody-negative psychiatric patients. Another caveat is the timing of symptoms, which cannot be optimally assessed via a cross-sectional approach. In another study, we would prefer to assess psychopathology across its spectrum longitudinally, since in neural autoantibody diseases, symptoms often have a relapsing remitting character. A prospective study with fixed time points is necessary in future to avoid test biases regarding the timing of psychometric evaluation. Another limitation is that we only have data on the semiquantitative intensity assessment of neuronal autoantibodies and no endpoint titers of all autoantibodies, which is not a suitable biomarker to support the clinical relevance of autoantibody findings. In addition, as a further limitation of our study, it is worth noting that in the evaluation of psychopathology, the notes of doctors are used. However, this is not a prospective assessment that is carried out the same for everyone, so our study hereby draws attention to this limitation.