Schizophrenia Is Associated With an Aberrant Immune Response to Epstein–Barr Virus

Faith Dickerson; Lorraine Jones-Brando; Glen Ford; Giulio Genovese; Cassie Stallings; Andrea Origoni; Colm O'Dushlaine; Emily Katsafanas; Kevin Sweeney; Sunil Khushalani; Robert Yolken

Disclosures

Schizophr Bull. 2019;45(5):1112-1119. 

In This Article

Results

The demographic and clinical characteristics of the 743 individuals in the study, 432 individuals with schizophrenia and 311 nonpsychiatric controls, are presented in Table 1. Within the schizophrenia group, participants had the following diagnoses per DSM-IV criteria: schizophreniform disorder (n = 17, 4%); paranoid subtype (n = 51, 12%); undifferentiated subtype (n = 126, 29%); other schizophrenia subtype (n = 10, 2%); schizoaffective disorder (n = 228, 53%). A total of 124 (29%) of the schizophrenia participants met the criteria for the deficit syndrome. The following antipsychotic medications were the most commonly received at the time of the study assessment by the persons in the schizophrenia group: risperidone (n = 115, 27%); olanzapine (n = 76, 18%); clozapine (n = 68, 16%); ziprasidone (n = 26, 6%). The schizophrenia participants also received additional types of psychotropic medications including antidepressants (n = 169, 39%) and valproate (n = 87, 20%).

Initial analyses were performed to compare the quantitative levels of antibodies between the diagnostic groups. As shown in figure 1 there were significantly elevated levels of IgG antibodies to EBV virions in the schizophrenia group vs the control group (effect size = 0.356; 95% CI 0.168, 0.543; P < .0002). This association was confirmed by analysis employing a nonparametric interquartile regression analysis (effect size = 0.467; 95% CI 0.240, 0.693; P < .0001). There was also a trend towards increased levels of IgG antibodies to VCA in the schizophrenia group (effect size = 0.197; 95% CI 0.025, 0.370; P = .025). However, the levels of antibodies to EBNA-1 did not differ significantly between the groups. Histograms of the distribution of values for these EBV antibodies in individuals with schizophrenia and controls are shown in supplementary figures 1–3 and the prevalence of antibodies in the case and control population are displayed in Supplementary Table 1. We also measured antibodies to the other human herpesviruses HSV-1, HSV-2, CMV, VZV, and HHV-6. There were no significant differences between the schizophrenia and the control group in any of these antibody levels. There was a trend toward decreased levels of antibodies to CMV in individuals with schizophrenia (effect size= −0.192; 95% CI −0.357, 0.026; P = .023).

Figure 1.

Effect sizes of immunoglobulin G (IgG) antibody reactivity to Epstein–Barr virus (EBV) and other human herpesviruses by solid phase immunoassays in individuals with schizophrenia as compared to controls calculated using logistic regression models. **P < .001; #P < .05.

Supplemental Figure 1.

Histogram distribution of antibodies to EBV Virions in Individuals with Schizophrenia and Controls. Effect size = .356; 95% CI= .168 - .543; p<.0002 adjusted for age, gender, race, maternal education and cigarette smoking.

Supplemental Figure 2.

Histogram distribution of antibodies to EBV Viral Capsid Antigen (VCA) in Individuals with Schizophrenia and Controls. Effect size =.197; 95% CI= .025 – .370; p=.025 adjusted for age, gender, race, maternal education, and cigarette smoking.

Supplemental Figure 3.

Histogram distribution of antibodies to EBV Nuclear Antigen 1 (ENV NA1 , EBNA-1 ) in Individuals with Schizophrenia and Controls. The differences between the populations was not statistically significant.

We also examined the odds ratios associated with elevated levels of antibodies defined by values greater than pre-defined percentile levels of the control group adjusted for age, sex, race, cigarette smoking, and maternal education. As depicted in figure 2 we found increased odds of elevated antibodies to EBV virions in the schizophrenia group relative to cutoffs greater than the 50th (OR = 1.71; 95% CI 1.18, 2.48; P = .005), the 75th (OR = 2.22; 95% CI 1.50, 3.28; P < .001) and the 90th (OR = 2.31; 95% CI 1.39, 3.84; P = .001) percentile of the levels of antibodies in the controls. Further, we found a trend to increased odds of antibodies to VCA in the schizophrenia group vs the control group for cutoffs greater than the 50th (OR = 1.45; 95% CI 1.00, 2.08; P = .048), and the 75th (OR = 1.54;, 95% CI 1.04, 2.28; P = .032) percentile, as well as a significant effect at the 90th (OR= 2.03; 95% CI 1.23, 3.37; P = .007) percentile of the levels of the antibodies in the controls. There were no significant differences in the odds associated with increased levels of antibodies to EBNA-1.

Figure 2.

Odds ratios of immunoglobulin G (IgG) anti-EBV antibody levels in schizophrenia as compared to controls by percentile values of IgG antibodies to Epstein–Barr virus (EBV) virions, EBV Viral Capsid Antigen (VCA) and EBV nuclear antigen (NA). The odds ratios were calculated by the use of logistic regression models **P < .001, *P < .012, #P < .05.

We further measured the reactivity of samples towards EBV proteins employing a quantitative western blot assay system. As shown in figure 3, there was a significant increase related to schizophrenia diagnosis in the levels of antibodies to VCA p33 (effect size = 0.363; 95% CI 0.113, 0.614; P < .005), VCA p22 (effect size = 0.326; 95% CI 0.085, 0.568; P < .008), VCA p41 (effect size = 0.392; 95% CI 0.088, 0.696; P < .012).and viral protein p27 (effect size = 0.507; 95% CI 0.116, 0.898; P < .011). There was also a trend towards increased levels in the schizophrenia group associated with antibodies to VCA p65 (effect size = 0.381; 95% CI 0.033, 0.729; P = .032) as well as the early antigen EA-D p43 (effect size = 0.290; 95% CI 0.015, 0.565; P < .039). There were no schizophrenia-associated increases in antibodies to the other antigens including EBNA-1 p79 and the other early antigens.

Figure 3.

Effect sizes of immunoglobulin G (IgG) antibody reactivity to individual Epstein–Barr virus (EBV) proteins as measured by quantitative western blot comparing reactivity in individuals with schizophrenia and controls. The effect sizes were calculated using logistic regression models employing *P < .012; #P < .05.

We examined the bivariate relationship between antibodies to EBV virions and clinical and demographic variables within the group of individuals with schizophrenia. In terms of basic demographic variables, the levels of antibodies to EBV virions were positively associated with increased age (correlation coefficient = .023; 95% CI .020, .038; P < .0001), female sex (F = 9.93, P < .0.0017), lower levels of maternal education (correlation coefficient = −.070; 95% CI −.115, .024; P < .003), cigarette smoking (F = 8.31, P < .0042) and non-Caucasian race (F = 15.75, P < .0001) but not with participant educational level, age of onset, illness duration, or birth outside of the United States or Canada (all P > .1).

We employed regression models to examine the relationship between levels of antibodies to EBV virions and clinical variables as described in the supplementary methods employing age, sex, race, cigarette smoking and maternal education. We found that levels of antibodies to EBV virions were significantly associated with the presence of deficit syndrome (effect size = 0.363; 95% CI 0.111, 0.616; P < .005) and the administration of the medication valproate (effect size = 0.399; 95% CI 0.119, 0.680; P < .005). There were no significant associations with the PANSS symptom score, RBANS cognitive score, BMI, or other medications (all P > .05).

We also examined the interaction between EBV antibodies and the genetic risk for schizophrenia as measured by the polygenic risk score employing regression models adjusted for age, sex, race, cigarette smoking, and maternal education. The polygenic risk score was associated with an increased risk of schizophrenia in the study population (effect size = 0.498; 95% CI 0.245, 0.751; P < .001) but was not significantly associated with the level of anti-EBV virion, anti-EBV VCA, or anti-EBNA-1 antibodies (P > .1). Analyzed in terms of percentiles, a polygenic risk score of ≥50th percentile was associated with a schizophrenia diagnosis with an odds ratio of 2.18 (95% CI 1.27, 3.74; P < .005) and a polygenic risk score of ≥75th percentile was associated with a schizophrenia diagnosis with an odds ratio of 1.61 (95% CI 0.904, 2.88; P > .1).

There was an apparent additive effect of odds ratios associated with the polygenic risk score and EBV virion antibodies. The odds ratio for schizophrenia diagnosis associated with having both EBV virion antibody levels and the schizophrenia polygenic risk score ≥50th percentile was 3.41 (95% CI 1.47, 7.95; P < .004 adjusted for age, sex, race, maternal education, cigarette smoking, and genotyping array). The odds ratio for schizophrenia diagnosis associated with having both EBV virion antibody levels and the schizophrenia polygenic risk score greater than the 75th percentile was 8.86 (95% CI 2.59, 30.37; P < .001). There was no significant statistical interaction between the levels of EBV virion antibody and the polygenic risk score in relation to their association with schizophrenia (P > .1).

Follow-up samples were obtained and analyzed from 183 individuals: 131 individuals with schizophrenia and 52 controls. The median interval between the initial and follow-up was 182 days (interquartile range 112–578 d). The individual levels of EBV antibodies to virions VCA and EBNA-1 at the first and last sampling periods were highly correlated (effect size = 0.88, 0.76, and 0.84, respectively). None of the differences between the levels measured at the first and second sampling were statistically significant (P > .05 adjusted for age, sex, race, diagnosis, maternal education and time interval between first and second sampling).

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....