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


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

In This Article


Study Population

All participants met the criteria for schizophrenia1 and the following additional criteria: absence of current substance abuse or dependence over the past one month, and of any history of intravenous substance abuse; absence of mental retardation; absence of clinically significant medical disorder that would affect cognitive performance such as epilepsy, history of encephalitis or head trauma, or any other reported neurological disorder of the central nervous system that had resulted in past or current treatment. The participants with schizophrenia were recruited from psychiatric programs affiliated with Sheppard Pratt Health System and at other outpatient treatment sites in central Maryland initially for a study on the association between antibodies to infectious agents and schizophrenia 2. The control individuals without a history of psychiatric disorder were recruited from posted announcements at local health care facilities and universities in the same geographic area as the sites where the individuals with schizophrenia were drawn.

At the initial visit, demographic and background information was obtained by interview. A review of systems was conducted including a past history of autoimmune disorders. Body Mass Index (BMI) was calculated based on height and measured weight. Participants were also asked about current cigarette smoking status. All participants were individually administered the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS)3. The schizophrenia individuals were interviewed and rated on the Positive and Negative Syndrome Scale (PANSS) 4 and their medication data were recorded from the clinical chart. Deficit syndrome, a putative schizophrenia subtype comprised of individuals with schizophrenia who have primary and enduring negative symptoms such as restricted affect and diminished social drive 5, was measured based on PANSS scores reflecting a distinctive combination of high negative symptom scores and an absence of dysphoria 6.

Follow-up blood samples were also available from 183 individuals, 131 of whom were individuals with schizophrenia and 52 controls. The follow-up samples from the schizophrenia individuals were obtained as part of their participation in subsequent clinical trials. The follow-up samples from the non-psychiatric controls were obtained as part of the planned follow-up evaluations. The median interval between the initial and follow-up was 182 days (interquartile range 112–578 days).

Measurement of IgG Antibodies to EBV Virions

Microtiter plates from Thermo Fisher (Immulon 2) were coated with 100 μL of Epstein Barr Virus Antigen Lysate (Meridian Catalog 7420); the EBV virus was diluted to 4 μg/mL in 0.05 M Potassium Carbonate buffer, pH 9.6. Plates were stored at 4°C for 16–20 hours and subsequently washed with high purity water. Plates were then blocked with 200 μL Non-Mammalian blocker (ImmunoO4, Catalog NMPCC diluted 1:10 in 5% sucrose) for two hours at room temperature and stored desiccated. On the day of testing Blood samples were diluted 1:300 in chaotrope buffer (0.25% non-fat dry milk; 1% BSA, 50 mM Tris pH 8.1, 0.5% Triton X 100, 0.5% Tween 20; 0.5 M sodium chloride; 0.2% casein). Diluted serum (100 μL) was added to each well and allowed to react at room temperature for four (4) hours. After four (4) washings with PBS/0.05%Tween 20, 100 μL of HRP goat anti-human IgG (KPL Catalog 074–1006) diluted 1:5000 in Super Shield (Immun04, Catalog HRPZR) was added to each well. The diluted conjugate was incubated at room temperature for two hours and then washed with PBS-Tween. ABTS substrate (Moss Catalog ABTS-1000) was added to each well and after 20 minutes the color reaction was read at 405 nm in a BioTek microplate reader.

Western Blot Analysis

Western blot analysis was performed employing kits purchased from Euroimmun, Luebeck, Germany following the manufacturer's instructions. Total IgG reactivity was also quantitated by measuring reactivity to a control band included on each blot. This analytic system consists of kits containing 16 nitrocellulose blot strips onto which proteins from EBV viral particles have been transferred following separation by SDS polyacrylamide gel electrophoresis. Each kit contains a control/calibrator strip against which analysis software (EuroLineScan, Euroimmun) compares each test strip. Briefly, each blot strip was incubated with a 1:50 dilution of a human serum sample, washed and then incubated with alkaline-phosphatase-conjugated goat anti-human IgG. Following washing, strips were exposed to enzyme substrate solution (NBT/BCIP) and the reaction stopped after ten minutes. Strips were allowed to air dry, scanned on a flatbed scanner and the resulting images analyzed using EuroLineScan software. This software performs a digital evaluation of each test strip thus allowing quantitative evaluation of the western blot results. The software determined numerical antigen/antibody reactivity band intensities for each of twelve EBV antigens (Supplementary Table S1) on the strip. The band intensities for each EBV antigen were compared between cases and controls.

Genetic Analyses

Genotyping was performed using the Infinium Omni 2.5 (, the Illumina Global Screening Array, or the Infinium PsychArray BeadChip ( Genotype array data was analyzed using plink 1.97 and principal components were computed with GCTA using a freely available protocol (see using methods which have been previously described.8,9. The protocol listed in the supplement ( includes the following quality control measures:1) Removal of variants with minor allele frequency less than 0.5%; 2) Removal of variants with missingness above 2%; 3) Removal of variants with excess heterozygosity (p<1e-6); 4) Removal of autosomal variants associating with sex (p<1e-6). Further removal of variants in long-range linkage disequilibrium regions* was performed for principal component analysis as previously described10.

Genotypes were phased using referenced based phasing 11 and imputed genotype dosages were computed using Minimac 12 and the 1000 Genomes project phase 3 reference panel 13. The schizophrenia polygenic score was computed using available summary statistics 14 retaining markers with an association p-value p<0.05 and using plink 1.9 to compute the score from genotype dosages. Computed scores were further adjusted against the first 10 principal components to avoid potential stratification issues. Sufficient DNA samples were available from 377 individuals, 235 of whom were individuals with schizophrenia and 142 were control individuals without a psychiatric disorder.

Assay Standardization. To allow for comparison across assays, the values measured at the completion of the solid phase assays and western blot assays were converted to normalized scores whereby the control samples on each microplate had a mean value of 1.0 and a standard deviation of 1.0 as described in reference 24 in the main text.

Determinations of Antibody Prevalence. Standard samples corresponding to low levels of reactivity to the target virus were run in with each assay in at least 2 replicated per reaction microplate. An individual was considered to be exposed to the target antigen is the sample gave a reaction value which was at last 80% of the mean value of the standard samples

Additional Statistical Methods

The levels of antibodies to EBV virions and defined EBV proteins in individuals with schizophrenia and non-psychiatric controls were compared using linear regression models employing age, sex, race, cigarette smoking, and maternal education as covariates with maternal education serving as an indication of socio-economic status. The resulting coefficients were used as estimates of effect sizes. Significant relationships between antibodies and schizophrenia diagnosis determined by linear regression models were confirmed by the use of non-parametric bootstrapped quantile regression models examining the 50th quantile and employing the same covariates.15 Groups were also compared employing logistic regression models to calculate the odds of elevated antibody levels in the schizophrenia group compared to the levels in the control group. For these analyses, elevated levels of antibodies were defined as antibody levels that were ≥50th, ≥75th, and ≥90th percentile of the levels of the controls. Percentile cutoffs were employed to allow for the calculation of individual and combined adjusted odds ratios as well as to account for possible non-normal distribution of some of the data. We further examined the relationship between genetic susceptibility to schizophrenia and EBV antibodies to virions employing linear and logistic regression models in which EBV antibodies to virions and the schizophrenia polygenic risk score were examined as independent and as interactive variables. For these regressions, an additional variable was added relating to the type of sequencing array used for the determinations.

The association between the antibody levels at baseline and at follow-up was assessed by means of mixed effects models employing the covariates age, sex, race, maternal education, follow-up duration, and schizophrenia vs. control group status.

Within the schizophrenia group, basic demographic and clinical data such as age, sex, race, place of birth, cigarette smoking, and education were related to levels of antibodies using bivariate associations. The association between EBV antibody levels and additional demographic, clinical, and medication variables were also analyzed using regression models employing age, sex, race, cigarette smoking, and maternal education as covariates.

The results of the Western Blot analyses were analyzed using linear regression models with the covariates of age, sex, race, cigarette smoking, and maternal education as well as reactivity to the control band described above.

In light of the performance of three sets of immunoassay measurements (EBV Virions, EBV VCA, EBNA-1) a critical value of p<.05/3 = .016 was employed to indicate statistical significance for assays using these measures. A value of .016 ≤ p ≤ .05 was considered to represent a trend. All statistical analyses were performed with STATA version 15, College Station, Texas

Supplemental References

  1. First M, Gibbon, M, Spitzer, RL, Williams, JBW. User's Guide for the SCID-I, Structured Clinical Interview for DSM IV Axis I Disorders. . New York, NY: Biometrics Research; 1996.

  2. Dickerson FB, Boronow JJ, Stallings C, Origoni AE, Ruslanova I, Yolken RH. Association of serum antibodies to herpes simplex virus 1 with cognitive deficits in individuals with schizophrenia. Archives of general psychiatry May 2003;60(5):466–472.

  3. Randolph C, Tierney MC, Mohr E, Chase TN. The Repeatable Battery for the Assessment of Neuropsychological Status (RBANS): preliminary clinical validity. J Clin Exp Neuropsychol Jun 1998;20(3):310–319.

  4. Kay SR, Fiszbein A, Opler LA. The positive and negative syndrome scale (PANSS) for schizophrenia. Schizophrenia bulletin 1987;13(2):261–276.

  5. Kirkpatrick B, Buchanan RW, McKenney PD, Alphs LD, Carpenter WT, Jr. The Schedule for the Deficit syndrome: an instrument for research in schizophrenia. Psychiatry research Nov 1989;30(2):119–123.

  6. Kirkpatrick B, Buchanan RW, Breier A, Carpenter WT, Jr. Case identification and stability of the deficit syndrome of schizophrenia. Psychiatry research Apr 1993;47(1):47–56.

  7. Chang CC, Chow CC, Tellier LC, Vattikuti S, Purcell SM, Lee JJ. Second-generation PLINK: rising to the challenge of larger and richer datasets. Gigascience 2015;4:7.

  8. Teo YY, Inouye M, Small KS, Gwilliam R, Deloukas P, Kwiatkowski DP, Clark TG. A genotype calling algorithm for the Illumina BeadArray platform. Bioinformatics (Oxford, England) Oct 15 2007;23(20):2741–2746.

  9. Korn JM, Kuruvilla FG, McCarroll SA, et al. Integrated genotype calling and association analysis of SNPs, common copy number polymorphisms and rare CNVs. Nature genetics Oct 2008;40(10):1253–1260.

  10. Price AL, Weale ME, Patterson N, et al. Long-range LD can confound genome scans in admixed populations. American journal of human genetics Jul 2008;83(1):132–135; author reply 135–139.

  11. Loh PR, Danecek P, Palamara PF, et al. Reference-based phasing using the Haplotype Reference Consortium panel. Nat Genet Nov 2016;48(11):1443–1448.

  12. Das S, Forer L, Schonherr S, et al. Next-generation genotype imputation service and methods. Nat Genet Oct 2016;48(10):1284–1287.

  13. Genomes Project C, Auton A, Brooks LD, et al. A global reference for human genetic variation. Nature Oct 1 2015;526(7571):68–74.

  14. Pardinas AF, Holmans P, Pocklington AJ, et al. Common schizophrenia alleles are enriched in mutation-intolerant genes and in regions under strong background selection. Nat Genet Mar 2018;50(3):381–389.

  15. Koenker R. Quantile Regression. New York: Cambridge University Press; 2005.