Intrathecal Thyroid Autoantibody Synthesis in a Subgroup of Patients With Schizophreniform Syndromes

In the Department of Psychiatry and Psychotherapy at the University Medical Center Freiburg, we have traditionally analyzed cerebrospinal fluid (CSF) in patients with schizophreniform syndrome if they showed any signs of neuroinflammation (e.g., sudden onset, atypical clinical presentation, neurological symptoms, altered EEG and cMRI findings). Based on our experience with psychiatric patients suffering from autoimmune encephalitis, we have offered CSF analysis as part of routine diagnostics since June 2009 for all patients with schizophreniform syndrome. Therefore, all patients who consented were investigated since 2009. Moreover, we performed broad serum analyses (including thyroid hormone status), EEGs, and cMRI measurements in our routine work-up for these patients.^4,21 Based on these experiences, we retrospectively performed paired antithyroid antibody testing of CSF and serum samples and calculated specific autoantibody indices (AIs) for the anti-TPO and anti-TG antibodies.^22,23 We hypothesized that we would find increased AIs in a subgroup of antithyroid antibody seropositive patients. For these patients, we planned to describe our clinical, laboratory, and instrument-based findings in detail. We also wanted to compare these findings between patients with and without increased AIs.

We included 100 patients at our tertiary care hospital with schizophreniform syndrome. Patients diagnosed with schizophreniform syndrome included those with schizophrenia (N=56), schizoaffective disorder (N=29), acute polymorphic psychotic disorder (N=14), and substance-induced psychosis (N=1). Only CSF/serum samples from patients with complete thyroid hormone status (thyroid-stimulating hormone [TSH], triiodothyronine [t3], and thyroxine [t4]) were included. For pragmatic reasons, we selected the first 100 patients since 2006. The study included 13 patients who underwent lumbar puncture between 2006 and 2009 (i.e., CSF analysis was performed if there were signs of possible neuroinflammatory features). The other 87 patients underwent lumbar puncture since June 2009, when CSF analysis was offered as a routine procedure. The subgroups for the schizophreniform patient cohort as well as number of instrument-based diagnostics are presented in Table 1. The diagnostic classification was performed by a board-certified psychiatrist following the ICD-10 criteria.

Paired CSF and serum samples were analyzed using an ELISA to identify anti-TPO and anti-TG antibodies. We used commercially available Medizym anti-TPO and Medizym anti-TG ELISA kits (Medipan GmbH, 15827 Dahlewitz/Berlin, Germany). We performed the ELISA according to the order of manufacturer’s instructions; however, since this assay is not validated for analyzing CSF samples, and we expected the titers of the thyroid antibodies to be very low, we used undiluted CSF samples. The results of the measurements in the CSF were adjusted to match the serum dilution. Those serum samples exceeding the value limits were controlled in a second assay to confirm the first positive finding. The AIs were only calculated if the second test was also positive.

Measuring AIs allows the detection of intrathecal antibody synthesis by distinguishing brain-derived fraction thyroid antibodies from antibodies passively diffused out of the serum into the CSF.^22,23 The calculation of specific AIs depends on parallel antibody testing of CSF and serum samples and the reference to the relevant quotient of total CSF/serum IgG in relation to blood-brain barrier (BBB) function (the albumin quotient is used as a reference standard).²² AIs were calculated in the case of increased antithyroid autoantibodies in the serum. The calculation of AIs was described in earlier papers.^22–25 The normal range for AIs is 0.7–1.3.²² AI values >1.4 were assessed as indicating intrathecal-specific antithyroid autoantibody synthesis in our study.²³

In earlier projects,^4,26 we established a CSF database including all CSF findings since 2006, socio-demographic information, diagnoses, laboratory results, and cMRI and EEG ratings. The procedures we used to obtain CSF-based diagnostics and cMRI and EEG ratings were described earlier.^4,26 In our CSF-database, antithyroid-autoantibody findings from serum and CSF were added. Statistical analyses were performed using Statistical Package for the Social Sciences, version 22 (SPSS 22 [www-01.ibm.com/software/analytics/spss]). The AIs of both groups were described in detail. Group comparisons for continuous variables (age, protein concentration, etc.) were performed using two-sided independent sample t tests. Group comparisons for categorical variables (gender, general cMRI alterations, AIs, etc.) were carried out using Pearson’s chi-squared test. Correlation analyses were performed using the Pearson correlation coefficient. For all statistical analyses, a p value <0.05 served as the criterion of significance. For the correlation analyses, we did not perform a Bonferroni correction for multiple tests because our analyses were exploratory.

The patients’ mean age was 33.32 years (±12.35 years) (ranging from 18 to 66 years), and more were female than male (females, N=59; males, N=41).

The patients’ thyroid hormone statuses are presented in detail in Table 2, and the autoantibody findings are presented in Table 3. In our cohort of schizophreniform patients, we found increased anti-TPO antibodies in the sera of 18 of the 100 patients (18%). Twelve of these 18 seropositive patients (67%) showed increased AIs for anti-TPO antibodies (ranging from 1.54 to 8.16). Anti-TG antibodies were found in the sera of two of 100 patients (2%). One of these two patients showed an increased AI for anti-TG-antibodies (50%; AI: 1.84). Overall, we found increased thyroid autoantibodies in 19 of 100 (19%) of the patients. One patient showed both increased anti-TPO and anti-TG antibodies in the serum and increased AI only for anti-TPO antibodies (patient 3) (Table 4). Thirteen patients showed increased AIs. Increased AIs were found in 68% of the seropositive patients.

The age range of the sample with increased AIs varied from 19 to 61 years, and mostly women were affected (11/13 patients [84.6%]). Increased AIs were found in one of 13 (7.7%) patients in the nonscreened cohort (before June 2009) and in 12 of the 87 (13.8%) that underwent routine lumbar puncture (χ²=0.263, p=0.608). Most patients with increased AIs suffered from paranoid-hallucinatory schizophrenia (6/13 [46%]) or schizoaffective disorder (6/13 [46%]). One patient with an increased AI presented with an acute polymorphic psychotic disorder (1/13 patients [8%]). Six of the 56 patients with schizophrenia showed increased AIs (10.7%), while the other 51 patients had normal thyroid antibodies or normal AIs (91.1%). In the schizoaffective group, increased AIs were found in six out of 29 patients (20.7%), while 22 patients (75.9%) had normal thyroid antibodies or normal AIs. Of the 14 patients with acute polymorphic psychotic disorder, only one showed increased AIs (7.1%), while 13 had normal thyroid antibodies or normal AIs (92.9%). The single patient with drug-induced psychosis was antibody-negative. The chi-square test showed no significant differences in increased AIs between the diagnostic categories (χ²=2.349, p=0.503). Thyroid hormone status was normal in 10 of the 13 AI-positive cases (76.9%). Further CSF analyzes showed BBB dysfunction in three samples (23%) and displayed unspecific (identical pattern in CSF and serum) oligoclonal bands in two of 13 cases (15%). The cMRI showed alterations in nine of 13 cases (69%). The alterations were most likely unspecific white matter lesions (in 38%) (Table 5). EEGs were abnormal with intermittent slowing in five of 13 cases (38%). One patient with slowing had additional spike waves in the EEG (8%) (Table 5).

Females were significantly more likely to have increased antithyroid AIs (p=0.044). Patients with and without increased antithyroid AIs cannot be distinguished by the thyroid hormone status, CSF, and EEG findings separately. The cMRI alterations tended to be more frequent in the group with increased AIs (p=0.060) (Table 5).

Today, serum thyroid antibodies are the main indicator for considering the diagnosis of HE. However, elevated serum antithyroid antibodies are not disease-specific and are also present in healthy subjects, as well as in patients with other autoimmune diseases.^10,28 In a large sample of U.S. citizens (17,353 people), serum thyroid antibodies were increased in 13% of healthy individuals (more often in females and older patients).^10,20 Therefore, they are unreliable as the sole criterion for detecting HE. CSF antibody synthesis (represented by AI) might be a better diagnostic marker. In a previous study including six patients with HE and neurological manifestations (e.g., confusion, paresis, ataxia, falls, myoclonic jerks), the authors detected antithyroid autoantibodies in the CSF of six patients with HE (independent of BBB function), but not in a control group of 21 patients with different neurological conditions.²⁹ High levels of CSF antithyroid autoantibodies were also detected in the HE group of another study with 10 neurological HE patients and 33 controls.³⁰ From these observations, one might conclude that it is possible for autoantibodies to reach the CSF compartment in patients with normal BBB function. However, it remains unclear whether intrathecal antibody synthesis or only passive transfer of blood-derived antibodies to CSF is the origin of elevated antithyroid antibodies in CSF. Therefore, it might be necessary to calculate patients’ AIs to investigate an intrathecal synthesis indicating presence of antithyroid specific autochthonous B-cell clones within the central nervous system (CNS). In a second study conducted by the Ferracci et al. group, they found increased specific antithyroid autoantibodies indices in nine antibody-seropositive patients with neurological manifestations (e.g., confusion, impaired consciousness, ataxia, paresis, paresthesias) of HE.³¹ However, the authors did not systematically compare their findings to the albumin quotient. In patients with increased polyspecific intrathecal total IgG synthesis, these procedures would lead to overestimation of the disruption of BBB function and underestimation of AIs,²³ although the authors still found unexpectedly high AI levels (varying from 5.8 to 314.5).³¹ It remains unclear whether these results can be applied to purely psychiatric (e.g., schizophreniform) cohorts. In our cohort of schizophreniform patients, we found increased AIs (varying from 1.54 to 8.16) in a relevant subgroup of 68% of antithyroid antibody seropositive patients. The increased AIs we found were lower than those reported by Ferracci et al.³¹ Patients with increased AIs were mostly female and showed more frequent cMRI alterations. EEGs did not help to distinguish patients with and without increased AIs. Although this finding is remarkable given the frequent EEG alterations reported in earlier publications,^11,32we recently published a case report of a female patient with a normal EEG that suffered from a purely psychiatric (i.e., affective) syndrome that turned out to be caused by SREAT. We speculated that the normal EEG might be the consequence of a less severe variant of SREAT without neurological deficits. The rate of increased CSF protein concentration in only 23.1% of patients is also lower, compared with a large literature review on SREAT characteristics, in which increased protein levels were found in 82%.³³ However, if increased AIs are a marker for corticosteroid responsiveness in schizophreniform HE patients, they would be more specific than serum autoantibody levels. Our patients were not treated with corticosteroids; therefore, this question cannot be answered at present.

Increased AIs can be found in inflammatory neurological conditions (e.g., herpes zoster, Lyme neuroborreliosis).^22,27 For antibodies against intracellular synaptic antigens (e.g., anti-GAD antibodies) and intracellular onconeural antigens (e.g., anti-Yo antibodies), AIs have also been shown to be effective in the detection of intrathecal antibody synthesis.^24,25 Increased AIs allow semi-quantitative evaluation of the intrathecal antigen-specific humoral immune reaction. Intrathecal antibody synthesis occurs as a consequence of activated plasma cell clones within the CNS. Therefore, our results support the idea of central autoimmunity in the pathophysiology of schizophreniform patients with thyroid autoantibodies. However, it is still unclear whether the thyroid anti-TPO and anti-TG antibodies play an etiopathogenic role (i.e., cytopathic or functional effects) or whether they are only an epiphenomenon of underlying autoimmune processes, comparable with the “MRZ reaction” in patients with multiple sclerosis.^34,35 The hypothesis of a direct etiopathogenic role of antithyroid autoantibodies could be supported by cross-reactivity between the thyroid gland and CNS epitopes.³⁵ Findings in line with this were provided by Blanchin et al.,³⁰ who demonstrated that anti-TPO antibodies in patients with HE can bind to cerebellar astrocytes. In contrast, serum antibodies from patients with only Hashimoto’s thyroiditis did not bind to cerebellar tissue.³⁰ In turn, astrocytes are important for immunomodulation via cytokines and for glutamatergic signaling, which might be associated with neuronal dysfunction and schizophreniform symptoms in a subgroup of patients with HE.^35,36 To our knowledge, similar results have not been found for anti-TG antibodies. Clearly, further research is necessary. In particular, it would be interesting to clarify whether passive transfer of anti-TPO antibodies (from HE patients) produces the same symptoms in animal studies, which would contradict the hypothesis that such antibodies are a pure epiphenomenon. Moreover, researchers should analyze the binding sites of antithyroid autoantibodies on brain slices and whether immunization with putative antigens allows the development of an animal model for HE.³⁰

Thyroid antibodies should be measured independently of thyroid hormone status, as we found normal thyroid hormone levels in 10 of 13 patients with thyroid autoantibodies. Based on the results of this study and earlier studies by our group,^4,21 we suggest a broad routine work-up for schizophreniform patients with increased thyroid antibodies. The routine work-up should include a cMRI (to detect white matter lesions or other nonspecific alterations), an EEG (to detect frequent slowing or epileptic activity), neuropsychological testing (to identify frequent cognitive impairments and as a follow-up parameter during corticosteroid treatment), CSF diagnostics (to detect BBB dysfunction, increased white blood cell count, or oligoclonal bands), and autoantibody measurements (onconeural/synaptic and cell surface antigens). In cases where SREAT is suspected, an analysis of anti-TPO and anti-TG antibodies in CSF with additional calculation of respective AIs could produce additional information in the diagnostic process. Earlier studies showed that a specific AI seems to be a marker for HE.^29,31 We also found increased AIs in a subgroup of schizophreniform patients. However, further research is needed to determine whether an elevated AI is a prognostic marker for steroid responsiveness and/or long-term prognosis of the disease course in such patients.

The main limitation of our study is that it was uncontrolled, as we did not analyze a healthy control group for comparison. However, it would not be ethically justifiable to perform lumbar punctures on healthy controls due to potential complications. Since we included patients since 2006 (before we introduced CSF analyses as a standard procedure), the cohort is not representative of all schizophreniform patients. There was a potential selection bias, as we included only patients at a tertiary care hospital for whom CSF and serum material, as well as full thyroid hormone status, were available. Another limitation is that we focused on thyroid antibodies; however, patients with autoimmune disorders may harbor a panoply of antibodies that might also play a pathophysiological role. Future studies should combine measurements of different, potentially relevant antibodies. One patient with negative antibodies suffered from drug-induced psychosis, and substance abuse could sufficiently explain this patient’s symptoms. Future studies should analyze more homogenous cohorts without obvious explanations for their psychiatric symptoms. In addition, it could be interesting to investigate whether drug consumption might lead to BBB dysfunction, allowing thyroid antibodies to transfer into the CNS. With regard to methodology, one criticism may be that the assays were not established for CSF. However, since there are no established measurements for thyroid antibody in CSF, no validated assays were available. Future projects should close this gap. The patients were treated primarily with antipsychotics, and sometimes with anticonvulsants, following the guidelines for schizophrenia. Only recently have we started to do treatment attempts with intravenous steroids in such constellations. Therefore, none of the patients in our sample had received such a steroid treatment, and thus we were unable to analyze difference in response to steroids between patients with and without abnormal AIs. Our results might be influenced by the effects of antipsychotic/anticonvulsant medications, for which no correction was performed. Additionally, the statistical analysis of possible group effects between patients with and without elevated AIs might have produced false negative results because of the small size of the AI positive group (N=13).