Evidence for Cytokine Dysregulation in Schizophrenia Spectrum Disorders: A Comparison of Cerebrospinal Fluid and Blood Samples

Abstract

Abnormalities in immune function have been described in schizophrenia but few studies have investigated cytokines in cerebrospinal fluid (CSF) and their correlation with blood levels. In this cross-sectional study, cytokines were measured in CSF and plasma of 30 subjects with schizophrenia spectrum disorder (SSD) diagnosis and 23 healthy volunteers (HV). Results showed that CSF TNFα was increased in SSD subjects compared to HV and there were no correlations between CSF and plasma cytokine levels. The present findings provide evidence of dysregulation of TNFα in CSF of schizophrenia. These results identify elevated CSF TNFα levels as a potential biomarker in schizophrenia.

1. Introduction

One of the leading hypotheses in the pathophysiology of schizophrenia is the dysregulation of the immune system. Cytokines have been prominently investigated given their critical role in the regulation of the innate and adaptive immune system (Parkin et al., 2001). In the brain, microglial activation leading to cytokine release has been hypothesized to be linked to the pathophysiology of schizophrenia, at least in a subset of patients (Monji et al., 2009). In support of this hypothesis, increased levels of pro-inflammatory cytokines have been described in peripheral blood and cerebrospinal fluid (CSF) samples of patients with schizophrenia, as reported by individual studies, such as those by C. G. McAllister et al. 1995, Garver et al. 2003, Pandey et al. 2015, Boerrigter et al. 2017, and as summarized in meta-analyses by our group (Gallego et al., 2018) and other groups (Miller et al., 2011; Wang et al., 2018). Even though schizophrenia is considered a brain disorder, most studies examined cytokine levels in peripheral blood samples (Wang et al., 2018) and only few have been conducted using CSF samples (Gallego et al., 2018). Moreover, despite the critical importance of determining whether peripheral blood cytokine levels accurately reflect central cytokine levels, very few studies have directly investigated the relationship between blood and CSF cytokine levels in schizophrenia patients (Katila et al., 1994; Sasayama et al., 2013; Coughlin et al., 2016), and even though other investigators have conducted parallel CSF/Blood studies in schizophrenia, they did not compute a correlation coefficient that would allow us to compare the relationship between CSF and blood values for each cytokine (McAllister et al., 1995, Maxeiner et al., 2014). Therefore, additional studies are needed to better characterize peripheral and central cytokine levels in patients with schizophrenia.

2. Methods

2.1. Inclusion and exclusion criteria

Patients were included in the study if they (1) fulfilled DSM-IV criteria for schizophrenia, schizophreniform disorder, schizoaffective disorder, or Psychosis Not Otherwise Specified (NOS); (2) were 15–59 years old, and (3) were willing and capable of providing informed consent. Healthy controls were also recruited and fulfilled the same inclusion criteria except for the diagnosis’s criterion. They were excluded if they had an Axis I diagnosis or if they had a first-degree relative with a known or suspected Axis I disorder, based on family history questionnaire. Participants with an acute inflammatory or infectious disease were excluded from participation.

2.2. Recruitment and consent

SSD patients were recruited from the inpatient and outpatient departments at The Zucker Hillside Hospital, a tertiary psychiatric hospital from Northwell Health in Glen Oaks, NY. HV were recruited from the general population via word of mouth, newspaper and internet advertisements, and poster flyers. All subjects provided written informed consent to a protocol approved by the Institutional Review Board of the Northwell Health System. After providing informed consent subjects provided urine for a urine toxicology test, underwent baseline ratings, including demographic information and psychopathology ratings.

2.3. CSF and blood collection

Subjects underwent a lumbar puncture, performed using a standard technique with a 25 gauge, Whitacre point spinal needle after subcutaneous lidocaine was applied. The procedure was conducted with patient sitting up and all procedures took place at 2 pm. Approximately 15–25 cc of CSF were obtained from each subject. CSF samples with macroscopic blood as consequence of traumatic procedures were not included in these analyses. A blood draw to obtain 10 cc of peripheral blood was obtained prior to the lumbar puncture procedure. Both sample types were flash frozen with liquid nitrogen and stored at −80 Celsius within 30 minutes of collection at the biorepository of the Feinstein Institute for Medical Research, the research institute for Northwell Health located in Manhasset, NY. CSF and plasma samples from the subjects included in this report were collected between 2009 and 2016 and then sent for cytokine analyses at the University of Cincinnati in July 2017.

2.4. CSF and plasma cytokine levels

CSF and plasma concentrations of six cytokines (IL-1ß, IL-2, IL-4, IL-6, IL-8 and TNFα) were determined in duplicate by enzyme-linked immunosorbent assay (ELISA) using MilliplexTM Multiplex kits (Millipore, Billerica, MA) according to manufacturer’s protocol. Samples were processed by a technician blinded to group assignment.

2.5. Statistical analysis

Baseline characteristics between patients and healthy volunteers were compared using t-tests for normally distributed continuous variables or Wilcoxon rank sum tests for continuous variables not normally distributed. Raw cytokine values did not follow a normal distribution and values were square root transformed. This transformation was selected above others, including the log transformation, given that it provided the most appropriate normal distribution as tested by the Ladder and Gladder commands on Stata 15. Statistical analyses were conducted only for those cytokines with less than 10% of undetectable values. Using that threshold, statistical analyses were only conducted for IL-6, IL-8 and TNFα in CSF, and for IL-4, IL-6, IL-8 and TNFα in plasma. Cytokine comparisons between SSD patients and HV were conducted using t-tests or Wilcoxon rank sum test for non-normally distributed variables. To determine the association between cytokines and a diagnosis of schizophrenia, a multivariate linear regression model was built with cytokine values as the dependent variable and subject type (SSD vs. HV), sex, age, and BMI as covariates. In addition, we computed correlations between CSF and plasma levels for all individual cytokines. These correlation analyses were exploratory and not adjusted for multiple comparisons due to limitations secondary to the limited sample size of our study.

3. Results

3.1. Subjects

Thirty SSD patients and 23 HV were included in the CSF analysis and 22 SSD and 17 HV were included in the plasma analysis. There were no significant differences between SSD patients and HV in sex, age or race, and SSD patients had higher body mass index values (mean: 30.0, SD=8.1 vs. mean: 26.3, SD=2.9, p=0.04) and lower MCCB composite scores (mean=25.1, SD=16.1, p<0.0001) compared to healthy volunteers (Table 1).

Table 1.

Demographic and clinical characteristics

	All Subjects (N=53)	SSD (n=30)	HV (n=23)	P-value
Age, mean (SD)	37.4(11.0)	36.8 (11.9)	38.1 (10.1)	0.69
Male, n (%)	37(69.8)	23 (76.7)	14(60.9)	0.21
Race				0.68
Black, n (%)	26(49.1)	15(50.2)	11(47.8)
White, n (%)	14(26.4)	6(20.0)	8(34.8)
Hispanic, n (%)	4(7.6)	3(10.0)	1(4.4)
Asian, n (%)	2(3.8)	1(3.3)	1(4.4)
Other, n (%)	7(13.2)	5(16.7)	2(8.7)
BPRS total score, mean (SD)	NA	30.0 (9.6)[25]	NA	NA
Diagnoses
Schizophrenia, n (%)	NA	20(66.7)	NA	NA
Schizoaffective, n (%)	NA	6(20.0)	NA	NA
Psychotic Disorder NOS, n (%)	NA	4(13.3)	NA	NA
BMI, mean (SD)	28.3(6.5)[46]	30.0(8.1)[25]	26.3(2.9)[21]	0.04
THC+	4(8.0) [50]	4(14.3)[28]	0(0.0)[22]	0.07
Number of Hospitalizations, median (IQR)	NA	4(2–6.5)		NA
Length of Illness, median years (IQR)	NA	16.2(2.1–25–7)	NA	NA
Medications
Antipsychotics, n (%)	NA	22(100)[22]	NA	NA
Antidepressants, n (%)	NA	9(40.9)[22]	NA	NA
Mood stabilizers, n (%)	NA	3(13.6)[22]	NA	NA
Benzodiazepines, n (%)	NA	8(36.4)[22]	NA	NA
MCCM composite score, mean(SD)	34.0(17.0)	25.1(16.1)	44.5(11.3)	<0.0001

SSD: Schizophrenia spectrum disorder; HV: healthy volunteer; NOS: Not Otherwise Specified; BPRS: Brief Psychiatric Rating Scale; THC: tetrahydrocannabinol; MCCB: MATRICS Consensus Cognitive Battery; ^Wilcoxon rank sum test; [ ] = Number of subjects with available data if different from group total

3.2. Cerebrospinal fluid and plasma analysis

3.2.1. Cytokine levels by group

Bivariate analysis showed that CSF TNFα levels were significantly higher in SSD patients compared to HV (mean=6.91 pg/mL, SD=0.52 vs 3.76 pg/mL SD 0=53; p=0.0001, Cohen’s d=1.16) (Figure 1). We did not find statistically significant differences between groups in CSF IL-6, or CSF IL-8 levels. In addition, there were no significant differences between groups in plasma levels of IL-4, IL-6, IL-8 and TNFα.

Figure 1.

CSF levels of TNFα in SSD vs. HV

3.2.2. Multivariable regression analysis

In a series of multivariable linear regression analysis using individual cytokines as the dependent variable, an SSD diagnosis was associated with TNFα (p=0.001) after adjusting for sex, age and BMI.

3.2. 3. Correlations between CSF and blood for all cytokines

Thirty-nine subjects had both CSF and plasma samples available for analyses. No statistically significant correlations were detected between CSF and plasma levels of TNFα (r=0.07, p=0.67); plasma IL-6 (r=0.10, p=0.23) and IL-8 (r=0.16, p=0.33).

4. Discussion

In this paper, we report elevated levels of TNFα in CSF in schizophrenia patients compared to healthy volunteers. Importantly, we did not find any significant differences between groups in plasma cytokine levels and did not find any significant correlations between CSF and plasma levels. Together, these findings provide important insight about cytokine dysregulation in schizophrenia and suggest that peripheral cytokine levels are not reflective of CSF levels.

Our CSF findings contrast with those by Zhu and colleagues (2016) who found decreased CSF TNFα levels in schizophrenia, though their control group consisted of patients with non-suppurative appendicitis instead of healthy volunteers. Our findings are in agreement with those by Maxeiner and colleagues (2014) which observed marginally elevated CSF TNFα levels in 16 schizophrenia patients compared to affective disorder patients. Therefore, our data provides support for elevated TNFα as a CSF biomarker that is relevant to the pathophysiology of schizophrenia and potentially serve as diagnostic biomarker or as a predictor of response to treatment in future studies.

Strengths of our study include the concurrent collection of blood and CSF samples in a well characterized group of schizophrenia patients and healthy volunteers. Importantly, our findings are not a result of substance use since most subjects had negative urine toxicology tests. On the other hand, our study was limited due to the relatively small sample size, as usually observed in single-site CSF studies, though our sample size was comparable to most prior CSF studies (Gallego et al., 2018). Also, our SSD participants were chronic, multi-episode patients on treatment with antipsychotic medications, therefore, these findings cannot be generalized to antipsychotic naïve, early psychosis patients”. In addition, we lacked smoking data, a variable often associated with increased inflammatory markers. However, we were able to investigate other variables associated with inflammatory markers such as sex, age and BMI, which were appropriately included in the regression model. Finally, we were unable to directly determine the impact of antipsychotic exposure on the outcomes since all patients included in our study were receiving antipsychotic treatment, but we did not find a correlation between CSF TNFα levels and antipsychotic dose expressed as chlorpromazine equivalents.

In summary, the present findings provide important insight concerning cytokine dysregulation in CSF in schizophrenia, particularly of TNFα, and the lack of correlation between CSF and blood cytokine levels of TNFα, IL-6 and IL-8. Our data provides initial support for the potential use of CSF TNFα as a diagnostic biomarker in schizophrenia, which could help us identify a group of subjects with an inflammatory subtype of the disease. Future studies, ideally using a longitudinal study design, are needed to validate these findings in an independent sample and to determine the impact of antipsychotic exposure e.g., in antipsychotic-naïve or antipsychotic-free participants. Finally, the simultaneous collection of genomic, neuroimaging, and other relevant measures would facilitate the integration and understanding of the inflammatory abnormalities in psychosis across several domains.

Highlights.

• TNFα is elevated in cerebrospinal fluid samples of patients with schizophrenia.

• There were no differences in plasma cytokine levels between patients and controls.

• Cerebrospinal fluid cytokine levels were not correlated with plasma levels.