Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Jan 3.
Published in final edited form as: Psychiatry Res. 2017 Dec 21;261:148–153. doi: 10.1016/j.psychres.2017.12.025

Joint evaluation of Serum C-Reactive Protein levels and Polygenic Risk Scores as risk factors for Schizophrenia

Vishwajit L Nimgaonkar 1,2,*, Faith Dickerson 3, Jennie G Pouget 4, Kodavali Chowdari 1, Colm O’Dushlaine 5, Joel Wood 1, Lambertus Klei 1, Bernie Devlin 1, Robert Yolken 6
PMCID: PMC6941903  NIHMSID: NIHMS932118  PMID: 29306175

Abstract

Several studies have indicated infectious and immune-related abnormalities in schizophrenia (Scz), including elevated serum C Reactive protein (CRP) - a well-known proxy for infections/immune abnormalities. A portion of the genetic risk for Scz can be estimated using the polygenic risk score (PGRS). It is not known whether there is an interaction in the risks traceable to CRP and PGRS. Patients with Scz and individuals without psychosis were evaluated systematically using DSM IV criteria (N=794, N=446, respectively). To estimate risk for Scz attributable to CRP and PGRS, serum from these participants was assayed for CRP levels using enzyme linked immunosorbent assays. PGRS was estimated from common DNA polymorphisms associated with Scz from genome wide association studies. CRP level and PGRS were not significantly correlated. Using a generalized linear logistic model, case/control status was evaluated in relation to the following predictors: CRP, PGRS, and demographic variables. CRP and PGRS were individually associated with case status; CRP: odds ratio (OR) 1.27, 95% confidence intervals (95%CI) 1.12, 1.43; p = 0.0001; PGRS: OR 1.66, 95% CI 1.47, 1.89; p = 1.28 ×10–15. There were no significant interactions between PGRS and CRP for predicting Scz versus control status.

Keywords: schizophrenia, C Reactive Protein, SNP, genetic association, polygenic risk score

1. Introduction

Schizophrenia (Scz) is a severe psychiatric disorder that exacts a considerable public health burden, with a lifetime morbid risk of approximately 1% in the United States (Gottesman, 1994). Therefore, it is important to understand the etiologic architecture and pathogenesis of Scz. A multi-factorial, polygenic threshold (MFPT) model that accommodates genetic and environmental risk factors, as well as possible interactions between genetic and environmental risk factors, provides the most plausible explanation for the etiology of Scz (Gottesman, 1991). Identifying Scz risk factors and assessing their relative effects presents an ongoing challenge.

Abnormalities in immune functions, particularly inflammatory factors have been investigated in relation to Scz for several decades. Evidence of a role for immunologic factors in Scz pathogenesis has accumulated from epidemiological studies that show an association between several prenatal infectious exposures and the development of schizophrenia in the offspring(Miller et al., 2013). Epidemiologic studies also show an increased Scz risk in persons who have a history of severe infection or autoimmune disease (Benros et al., 2014; Benros et al., 2011). A number of immune markers have also been associated with Scz in case-control studies. Of these, C-reactive protein (CRP) levels in the serum or plasma provide a reliable marker of acute or chronic inflammation caused by infectious and other inflammatory agents (Lowe, 2005). CRP is a pentameric protein that is generated in the liver and secreted in the blood. Because it plays a central role in human inflammation, it provides a composite estimate for inflammatory processes in the body. In a previous study, we found that individuals with established Scz had significantly elevated levels of C-reactive protein (CRP) compared with controls (Dickerson et al., 2013). This finding has been replicated in other investigations (Inoshita et al., 2016; Joseph et al., 2015; Miller et al., 2014) (Wium-Andersen et al., 2014). Thus, CRP serves as a useful proxy for inflammatory abnormalities in Scz (Dickerson et al., 2013).

A genetic etiology for Scz is supported by family and twin studies that demonstrated a substantial heritability (~70–80%) (Gottesman, 1991). Case-control associations have yielded over 100 independent single nucleotide polymorphisms (SNPs) as risk variants (Schizophrenia Working Group of the Psychiatric Genomics, 2014). The SNPs are relatively common in the population, but confer modest risks individually (individual odds ratios, OR ~ 1.03–1.3). Notably, the SNPs with the strongest statistical evidence for association with Scz risk are localized in the major histocompatibility complex (MHC) and other immune regulatory regions (Corvin and Morris, 2014). Even though the population attributable risk for individual genetic risk variants is relatively small, it is now possible to estimate the composite risk due to common SNPs using the polygenic risk score (PGRS) (Purcell et al., 2009) (Schizophrenia Working Group of the Psychiatric Genomics, 2014). Using published genome wide association studies (GWAS), the PGRS can be estimated for any individual by summing risk across the genome based on the number of risk conferring alleles. It is not yet known whether CRP level and PGRS combine additively or interact to predict risk. In the present study, we address this question by joint evaluation of PGRS and CRP as predictors of case-control status.

2. Methods

2.1. Design

The primary hypothesis tested whether CRP assays and PGRS scores were associated with risk for Scz. The study utilized a case-control design that excluded related individuals. The participants were individuals with schizophrenia and schizophrenia spectrum disorders or non-psychotic control individuals on whom both CRP and PGRS were measured.

2.2. Site

Participants were enrolled in the Stanley Research Program at Sheppard Pratt Hospital in Baltimore, Maryland or the Program for Genetics and Psychosis, Department of Psychiatry, University of Pittsburgh in Pittsburgh, PA. The procedures employed in the studies have been published (Dickerson et al., 2012; Dickerson et al., 2003; Talkowski et al., 2006).

2.3. Cases

The inclusion criteria for cases were either a diagnosis of schizophrenia or schizoaffective disorder or, at Baltimore, schizophreniform disorder (DSM IV criteria). The cases in Baltimore were recruited from inpatient and day hospital programs of Sheppard Pratt Hospital and from affiliated psychiatric rehabilitation programs. The cases in Pittsburgh were recruited from Western Psychiatric Institute and Clinic, Pittsburgh and additional psychiatric treatment facilities in a 500 mile radius of Pittsburgh. Patients referred by their therapists were evaluated using structured diagnostic instruments (Structured Clinical Interview for DSM-IV Axis 1 Disorders for Baltimore patients, Diagnostic Interview for Genetic Studies for Pittsburgh patients). This information was synthesized with available information from the patients’ medical records (both sites) and additional details from informants, usually relatives (Pittsburgh). The clinical information was synthesized and presented to board certified psychiatrists / psychologists, followed by consensus diagnosis based on DSM IV criteria.

2.4. Control individuals

The control group was recruited from posted announcements at local health care facilities and universities in the same geographic area and settings where the schizophrenia participants were recruited. At Baltimore, the primary inclusion criterion for the control individuals was the absence of a current or past psychiatric disorder as determined by screening with the DSM-IV Axis I Disorders, Non-patient Edition. The diagnosis of each control participant was made by a board-certified psychiatrist based on the Structured Clinical Interview for Diagnosis for Axis I disorders (SCID, First et al., 1996). At Pittsburgh, the control individuals were assessed in the same manner as the cases and were screened for absence of psychosis.

Participants in both groups fulfilled following additional criteria: age 20–60 years (Baltimore); proficient in English; absence of any history of intravenous substance abuse; absence of mental retardation; absence of self-reported HIV infection; absence of serious medical disorder that would confound a diagnosis of Scz; absence of a primary diagnosis of alcohol or substance use disorder.

All participants provided written informed consent and the study was approved by the Institutional Review Boards of Sheppard Pratt, the University of Pittsburgh School of Medicine, and the Johns Hopkins School of Medicine following established guidelines.

2.5. Additional evaluations

Participants were interviewed at the study visit and information was obtained about demographic variables including age, race, sex, and current smoking status (as a binary yes/no categorical variable).

2.6. Laboratory analyses

Each participant provided a venous blood sample. Aliquots were centrifuged and serum samples extracted, while other aliquots of whole blood were stored for later extraction of genomic DNA.

CRP assays

Serum samples were used to assay CRP levels using enzyme immunoassay at the Johns Hopkins Stanley Neurovirology Laboratory as previously described (Dickerson et al., 2013). CRP levels are presented as ratios based on a mean value of 1 in the control group (Dickerson et al., 2013).

Genotype assays

Genomic DNA samples were assayed using the Infinium PsychArray BeadChip, a high density DNA microarray that includes 571,054 DNA markers, including SNPs associated with Scz in prior GWAS excluding Scz associated SNPs in the HLA region (http://www.illumina.com/products/psycharray.html). The standardized Infinium HTS assay combines whole-genome amplification sample preparation with direct, array-based capture, and enzymatic scoring of the SNP loci.

PGRS estimates

For detailed information on quality control, imputation and genotype-based ancestry estimation, please see Supplement. In brief, after quality control at the level of individual subjects and individual SNP genotypes, the genetic data set contained 1288 samples with genotypes for 525, 644 SNPs. Additional autosomal genotypes were imputed using the Michigan Imputation Server (Das et al., 2016; McCarthy et al., 2016). After imputation, genotypes for 47,112,785 SNP were returned. Ancestry was estimated using Gemtools based on 9270 high quality SNPs (directly genotyped, 100% genotype completion rate, minor allele frequency MAF > 0.05, SNP set pruned using Plink to remove SNPs with alleles in high linkage disequilibrium).

PGRS for each subject was estimated by summing risk across 102,636 SNPs significantly associated (<5×10−8) with Scz from GWAS studies using the PLINK software (Korn et al., 2008). We used 69,343 of the GWA-significant SNPs because they were either directly genotyped on the PsychArray BeadChip or imputable with high confidence from the genotyped SNPs on the array. PGRS was computed as a weighted score, which was a function of risk allele count and effect size per SNP. The scores were then adjusted for genetic ancestry by regressing the scores against the first 4 principal components of genetic ancestry to obtain the adjusted values. In the rest of the manuscript PGRS refers to these ancestry adjusted scores.

2.7. Statistical Analyses

The relationship of CRP level with demographic variables was evaluated using linear regression with CRP level as the outcome and demographic variables smoking status, sex, age, site and self-declared ancestry as predictors. The relationship of CRP and PGRS level was estimated by Pearson correlation. Case versus control status was fit to predictors CRP, PGRS and demographic variables using a generalized linear logistic model. This analysis also evaluated an interaction term for CRP and PGRS. Various other analyses were performed to ensure robustness of the results. First, residual CRP (rCRP) was calculated by adjusting for smoking status, sex, age, site and PGRS; then case status was fit to rCRP and PGRS. Second, some predictor variables were transformed to reduce any potential impact of extreme values and models with transformed predictors were evaluated. Third, following earlier literature, we binned CRP values and evaluated the within-bin association with case status. The data were analyzed using the statistical software R (function glm in R package base for all logistic regression analyses; function NagelkerkeR2 in package fmsb for calculating pseudo-R2).

3. Results

3.1. Demographic variables

The study sample consisted of 1240 participants (Pittsburgh, N = 946; Baltimore, N = 294; Table 1), the majority were cases (Scz or Scz spectrum disorders, N=794, 64.0%, Controls, N = 446, 36.0%). The participants were predominantly male (N= 688, 53.9 %) and reported Caucasian race (Caucasian: N = 870, 70.2%; African-American, N = 370, 29.8%). Less than half of the entire sample were cigarette smokers (N=550, 44.4%).

Table 1.

Demographic features of the samples.

Diagnostic Group
Schizophrenia (n=794) Control (n=446)
Site (Baltimore / Pittsburgh) 170/624 124/322
Age (mean years, SD)* 38.7 ± 10.6 40.4 ± 16.1
Gender: male/female (% female)** 494/300 (37.8) 194/252 (56.5)
Caucasian ancestry (%) 567 (71.4) 303 (67.9)
Cigarette smokers (% current smokers)** 56.7 22.4
Open in a new tab

The samples were recruited from the Baltimore or Pittsburgh sites. The majority of the participants reported Caucasian ancestry; the majority of the remainder reported African-American ethnicity.

SD: standard deviation.

Significant differences between patients with schizophrenia and controls were noted with regard to age (* p <0.05), gender and smoking history (** p <0.0001).

3.2. Variables associated with CRP levels

Univariate linear regression was used to determine variables associated with CRP levels. Variables significant in univariate analysis (Supplementary Table 1) were Scz diagnosis (p = 1.15 × 10−4), smoking status (p = 8.59 × 10−7), gender (p = 2.99 × 10−4), age (p = 1.41 × 10−6), and site (p = 0.0074). Self-reported race or PGRS were not significantly correlated with CRP. For the latter, this was true in the context of the linear model or as a Pearson correlation of these two variables (r = −0.006, p = 0.83). Age was transformed (age/10) to enhance comparability to other predictors. None of the variables individually explained more than 2% of the variance (see Supplementary Table 1 for R2 for each independent variable).

3.3. Variables associated with Scz

Univariate and multiple logistic regression was next used to evaluate case versus control status and its relationship with the following variables: CRP level, PGRS score, and demographic variables (smoking status, age/10, race and gender). Variables significant in univariate analysis (Supplementary Table 2), in which Scz was fit to each predictor variable individually, were then entered as predictors of case versus control status in a joint model (Table 2). After accounting for demographic variables, both CRP levels and PGRS were each significantly associated with Scz status (p = 4.96 × 10−6 and 1.86 × 10−12, respectively) (Table 2, see Supplementary information, Supplementary Figure 1). Notably, there was no significant interaction between CRP and PGRS to predict case status in the context of this model (p = 0.37).

Table 2.

Multiple regression analysis of clinical and demographic variables in relation to Schizophrenia risk.

Effect b se OR CI p
Sex 0.641 0.133 1.90 1.46–2.47 8.46 × 10−6
Smoking status (current) 1.387 0.141 4.00 3.04–5.27 5.55 × 10−23
Age/10 −0.153 0.051 0.86 0.78–0.95 0.0028
Site (Pittsburgh) 0.460 0.156 1.58 1.17–2.15 0.0023
PGRS 0.487 0.069 1.63 1.42–1.86 1.86 × 10−12
CRP 0.240 0.069 1.27 1.11–1.46 4.96 × 10−6
Open in a new tab

PGRS: polygenic risk score. rCRP: residual C Reactive protein

To estimate the effect of each of the covariates, while all others effects are in the model, on the risk for schizophrenia, logistic regression was performed with schizophrenia status as the binary outcome. Predictors included in the model were sex, smoking status, age (years/10), site, PGRS and CRP. Ancestry was not included in the model since PGRS is adjusted for ancestry while CRP and schizophrenia status did not show a significant relationship with ancestry.

Next we evaluated the robustness of our results. First we evaluated whether covariate effects on CRP level were completely accounted for in our models. To do so, residual CRP (rCRP) was calculated from a regression model in which subjects’ CRP levels were fit by their smoking status, sex, age, site and PGRS; then case status was fit to rCRP and PGRS using logistic regression. Even after adjustment, there was a significant association between rCRP and Scz status (OR=1.21; p=0.0022; pseudo-R2=1.1%). Next, because the distribution of CRP and rCRP were skewed, we also evaluated the square root transformation of rCRP as a predictor (after adding a constant to make all values positive), but this adjustment made no material difference to the results; likewise, including a quadratic term for rCRP in the model had no predictive value (data not shown). Because other investigators have analyzed CRP estimates as categorical variables, the sample was also divided into three bins based on the CRP values (0–50%, 50–90% and 90–100%). The risk for Scz increased progressively for each bin: the respective odds ratios (95% confidence intervals) were 1.16 (0.98, 1.39), 1.49 (1.25, 1.78) and 1.73 (1.45, 2.06), respectively.

We next fitted a model in which Scz risk was predicted by both rCRP and PGRS jointly, as well as their interaction. Because the interaction term was non-significant, we report only the impact of main effects (Table 3). The fitted model accounted for a pseudo-R2 value of 8.5% (inclusion of the interaction term increased pseudo-R2 to 8.6%, a non-significant increase). Finally, a model including all significant predictors of diagnostic status results in a pseudo-R2 of 24.6% (Figure 1); all terms remained significant in the full model, which included effects for smoking status, sex, age, site, PGRS and rCRP. Using the same model without rCRP diminished the pseudo-R2 to 23.5%, demonstrating the independent predictive value of CRP in all model settings.

Table 3.

Risk attributable to CRP and PGRS in relation to Scz diagnosis.

Variable b OR 95%CI p-value
PGRS 0.515 1.67 1.48–1.90 9.32 × 10−16
rCRP 0.206 1.23 1.08–1.39 0.0015
Open in a new tab

PGRS: polygenic risk score. rCRP: residual value for C Reactive protein. OR: odds ratio. CI: confidence intervals.

To estimate risk due to CRP and PGRS, logistic regression was conducted with schizophrenia or control status as the binary outcome, and PGRS and rCRP as predictors. Residual CRP (rCRP) was determined from a subject’s CRP level after adjusting for their smoking status, sex, age, site. There was no significant interaction between rCRP and PGRS.

Figure 1. Risks for Scz in relation to CRP and PGRS.

Figure 1.

Open in a new tab

PGRS: polygenic risk score. CRP: C Reactive protein. rCRP: residual value for C Reactive protein. Risks associated with CRP and rCRP, indicated by pseudo-R2 values, were estimated singly or jointly with PGRS, along with age, sex, site and smoking status as indicated. From the left, the first four columns indicate risk due to CRP, rCRP or PGRS without correction for sex, age, smoking status or recruitment site. The last three columns indicate risk due to CRP, rCRP or PGRS in conjunction with variance attributable to these variables (i.e., sex, age, smoking status and recruitment site, shown as blue bars). Variance attributable to PGRS is indicated by yellow bars, while red bars indicate variance attributable to CRP or rCRP values.

Interpretation of the results in Table 3, in terms of increased risk for Scz, follows immediately from the relationship between the estimated effect for a predictor in the logistic model and its related odds ratio (i.e., odds ratio = eXβ). Thus, a unit increase in the PGRS increases risk for Scz by an odds ratio of 1.67 whereas a unit increase in rCRP increases risk by 1.23. The observed ranges for these variables (Table 3 and Figure 1) are, approximately, −3.5 to 3.1 and −1.7 to 3.0 for PGRS and rCRP, respectively. Thus, ignoring other risk factors, subjects with PGRS = −3.5 would be associated with small risk of Scz, ≈ 0.2, whereas those with PGRS = 3.1 have a risk of ≈ 4.9, both interpreted relative to the population prevalence. In the model predicting diagnostic status as a function of a multiplicity of covariates (Table 2), site, sex and race should be considered structural variables that account for the somewhat imbalanced representation of case and control subjects for these variables. None of these variables has a biological interpretation in the context of the model.

4. Discussion

Though several investigators have previously analyzed PGRS or CRP individually in relation to Scz risk, to our knowledge, most have analyzed one or the other variable. Concurrent analysis in two independent samples reveals association between each of these variables and Scz. The risk attributable to CRP and PGRS is independent and additive, without significant statistical evidence for added risk due to an interaction. The risk embodied by the PGRS is due to additive effects of many common genetic variants or SNPs, and our results for this variable are consistent with results from larger samples (Schizophrenia Working Group of the Psychiatric Genomics, 2014) (Purcell et al., 2009). CRP, on the other hand, is taken to represent an outcome related to inflammatory responses, although it could also be genetically or environmentally driven, or both. Our results, which show elevated CRP levels to be associated with an increase in risk for SCZ, are consistent with prior reports, including two meta analyses (Fernandes et al., 2016) (Dickerson et al., 2007; Dickerson et al., 2013). Although some studies report a non-linear association with CRP levels (Chokshi et al., 2015), we detected only a linear relationship between CRP levels and Scz risk. Consistent with the linear relationship, analysis of CRP levels as categorical variables indicated progressively increasing risk.

It has been speculated that immune abnormalities could partially explain the pathogenesis of Scz (Fillman et al., 2013). If this were true, then it is reasonable to anticipate that some of the SNPs associated with Scz regulate immune function. Some of these SNPs could be incorporated into scores, such as the PGRS. Scz associated SNPs in the HLA region are obvious candidates, in view of the substantial number of immune related genes in the HLA region. Exploratory analyses also indicated that even SNPs not localized to the HLA region are over-represented in genomic sequences that encode enhancer elements active in tissues with important immune functions, such as certain B-lymphocyte lineages (Schizophrenia Working Group of the Psychiatric Genomics, 2014). On the other hand, several lines of evidence argue against a GWAS based explanation for the immunopathogenic hypothesis of Scz. Using an specialized imputation approach (Jia et al., 2013), one of the authors (JP) and her colleagues found that the SNPs associated with functional variation in canonical immune genes such as the classical HLA genes do not account for the Scz associations in the HLA region (Pouget et al., 2016). Second, although the associations with Scz in the HLA region are partially explained by a repeat variation in Complement C4, which is an immune related gene, the role of C4 in Scz risk could be regulation of neural dendritic spine density, rather than immunity per se (Sekar et al., 2016). Pouget and colleagues further argued against an immune based explanation for the Scz associated SNPs, as SNPs associated with six common disorders with immune-based pathogenesis do not overlap significantly with Scz associated SNPs (Pouget et al., 2016). Insofar as the risks associated with PGRS and CRP are independent in the present studies, our analyses are consistent with Pouget’s proposition.

There are a number of limitations in the present study. First, our analyses assume that CRP level is a proxy for inflammatory processes that confer risk for Scz, whereas CRP levels could also represent consequences of SCZ pathology. One possible route to shed light could be with Mendelian Randomization (MR), an analytic model in which the risk for Scz conferred by SNPs associated with CRP levels would be assessed (Lawlor et al., 2008). Two recent MR studies using pooled datasets from published studies to estimate risk conferred by CRP genetic risk scores (GRSs) (Inoshita et al., 2016) (Prins et al., 2016). Both studies indicated relatively small, statistically significant causal relationships, albeit in opposite directions. We did not report on MR analyses due to the limited sample size. Second, although significant associations between CRP and smoking status, sex, age and site (possibly related to site-related differences in gender, age and Scz composition) were noted by adjusting for these variables, we obtained residuals denoted as rCRP that provided the most suitable predictor in a model to predict diagnostic status. Variables that might impact CRP levels, such basal metabolic index could not be analyzed. Similarly, antipsychotic medication should be considered in relation to CRP, though our earlier analyses (Dickerson et al., 2013) and a recent meta-analysis indicates that antipsychotic medications are not correlated with CRP levels (Fernandes et al., 2016). Third, the Infinium PsychArray BeadChip does not incorporate all the SNPs associated with Scz in a recent mega analysis (Fernandes et al., 2016), although the risk accounted with the present set of SNPs is statistically significant and its magnitude is consistent with other studies utilizing larger numbers of SNPs (Agerbo et al., 2015). Finally, the estimates related to smoking in the present studies are also somewhat imprecise; more refined variables are desirable.

The present results suggest several new lines of investigation. Importantly, replication studies to evaluate the independent associations with CRP and PGRS noted here are indicated. Indeed, our studies motivate similar analyses in systematically ascertained samples to derive clinical relevant population based risk estimates of PGRS and CRP. Whether cigarette smoking plays a role in the etiology of Scz is unclear; in any case, smoking is common among subjects diagnosed with Scz (Zammit et al., 2003) (Dickerson et al., 2006). Because smoking is typically correlated with CRP, as observed in our data, it is important to account for its effect on CRP level as we have done here (i.e., rCRP). In a similar vein, it would also be desirable to estimate population based risks for environmental variables such as cannabis abuse (Di Forti et al., 2009).

In conclusion, joint analyses of CRP and PGRs in two independent case-control samples indicate significant risk for Scz attributable to both variables. The magnitude of both associations are modest and the associations are additive. If the present results are replicated, they will set the stage for analysis of additional risk variables as part of ongoing efforts to identify clinically relevant risk predictors for Scz, as well investigations of Scz pathogenesis.

Supplementary Material

1

Highlights.

Serum C Reactive protein (CRP) is higher in Schizophrenia (Scz) cases than controls.

Polygenic risk scores (PGRS) from Scz genome wide associations predicted risk too.

Among Scz cases and controls, CRP and PGRS were not significantly correlated.

CRP and PGRS independently confer risk for Scz.

Acknowledgements

This study was funded by grants from the Stanley Medical Research Institute to HRY, grant 07R-1712 to VLN, and grant 07R-1690 to FD. Additional support was provided by the National Institute of Health (MH63480, MH93246 and D43 TW009114 to VLN). We acknowledge the support of the Broad Institute of MIT and Harvard University for DNA analyses, particularly Steve McCarroll PhD. We thank our research participants and the following colleagues for their help with clinical evaluations and with data analyses: Dr Jennifer Schroeder, Cassie Stallings, Andrea Origoni, Emily Katsafanas, Lucy A. B. Schwienfurth, Christina L. G. Savage, Dr Sunil Khushalani.

Footnotes

Conflict of Interest

The authors declare no competing financial interests in relation to the work described.

Supplementary information is available at the Journal’s website.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Agerbo E, Sullivan PF, Vilhjálmsson BJ, Pedersen CB, Mors O, Børglum AD, Hougaard DM, Hollegaard MV, Meier S, Mattheisen M, Ripke S, Wray NR, Mortensen PB, 2015. Polygenic Risk Score, Parental Socioeconomic Status, Family History of Psychiatric Disorders, and the Risk for Schizophrenia: A Danish Population-Based Study and Meta-analysis. JAMA Psychiatry 72 (7), 635–641. [DOI] [PubMed] [Google Scholar]
  2. Benros ME, Eaton WW, Mortensen PB, 2014. The epidemiologic evidence linking autoimmune diseases and psychosis. Biol Psychiatry 75 (4), 300–306. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Benros ME, Nielsen PR, Nordentoft M, Eaton WW, Dalton SO, Mortensen PB, 2011. Autoimmune diseases and severe infections as risk factors for schizophrenia: a 30-year population-based register study. Am J Psychiatry 168 (12), 1303–1310. [DOI] [PubMed] [Google Scholar]
  4. Chokshi DA, El-Sayed AM, Stine NW, 2015. J-Shaped Curves and Public Health. Jama 314 (13), 1339–1340. [DOI] [PubMed] [Google Scholar]
  5. Corvin A, Morris DW, 2014. Genome-wide association studies: findings at the major histocompatibility complex locus in psychosis. Biol Psychiatry 75 (4), 276–283. [DOI] [PubMed] [Google Scholar]
  6. Das S, Forer L, Schönherr S, Sidore C, Locke AE, Kwong A, Vrieze SI, Chew EY, Levy S, McGue M, Schlessinger D, Stambolian D, Loh PR, Iacono WG, Swaroop A, Scott LJ, Cucca F, Kronenberg F, Boehnke M, Abecasis GR, Fuchsberger C, 2016. Next-generation genotype imputation service and methods. Nat Genet 48 (10), 1284–1287. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Di Forti M, Morgan C, Dazzan P, Pariante C, Mondelli V, Marques TR, Handley R, Luzi S, Russo M, Paparelli A, Butt A, Stilo SA, Wiffen B, Powell J, Murray RM, 2009. High-potency cannabis and the risk of psychosis. Br J Psychiatry 195 (6), 488–491. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Dickerson F, Lillehoj E, Stallings C, Wiley M, Origoni A, Vaughan C, Khushalani S, Sabunciyan S, Yolken R, 2012. Antibodies to retroviruses in recent onset psychosis and multi-episode schizophrenia. Schizophr Res 138 (2–3), 198–205. [DOI] [PubMed] [Google Scholar]
  9. Dickerson F, Stallings C, Origoni A, Boronow J, Yolken R, 2007. C-reactive protein is associated with the severity of cognitive impairment but not of psychiatric symptoms in individuals with schizophrenia. Schizophr Res 93 (1–3), 261–265. [DOI] [PubMed] [Google Scholar]
  10. Dickerson F, Stallings C, Origoni A, Vaughan C, Khushalani S, Yang S, Yolken R, 2013. C-reactive protein is elevated in schizophrenia. Schizophr Res 143 (1), 198–202. [DOI] [PubMed] [Google Scholar]
  11. Dickerson FB, Boronow JJ, Stallings C, Origoni AE, Ruslanova I, Yolken RH, 2003. Association of serum antibodies to herpes simplex virus 1 with cognitive deficits in individuals with schizophrenia. Arch Gen Psychiatry 60 (5), 466–472. [DOI] [PubMed] [Google Scholar]
  12. Dickerson FB, Brown CH, Daumit GL, Fang L, Goldberg RW, Wohlheiter K, Dixon LB, 2006. Health status of individuals with serious mental illness. Schizophr Bull 32 (3), 584–589. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Fernandes BS, Steiner J, Bernstein HG, Dodd S, Pasco JA, Dean OM, Nardin P, Gonçalves CA, Berk M, 2016. C-reactive protein is increased in schizophrenia but is not altered by antipsychotics: meta-analysis and implications. Mol Psychiatry 21 (4), 554–564. [DOI] [PubMed] [Google Scholar]
  14. Fillman SG, Cloonan N, Catts VS, Miller LC, Wong J, McCrossin T, Cairns M, Weickert CS, 2013. Increased inflammatory markers identified in the dorsolateral prefrontal cortex of individuals with schizophrenia. Mol Psychiatry 18 (2), 206–214. [DOI] [PubMed] [Google Scholar]
  15. Gottesman I, 1991. Schizophrenia Genesis: The Origins of Madness. WH Freeman, New York. [Google Scholar]
  16. Gottesman II, 1994. Complications to the complex inheritance of schizophrenia. [Review]. Clin Genet 46 (1 Spec No), 116–123. [DOI] [PubMed] [Google Scholar]
  17. Inoshita M, Numata S, Tajima A, Kinoshita M, Umehara H, Nakataki M, Ikeda M, Maruyama S, Yamamori H, Kanazawa T, Shimodera S, Hashimoto R, Imoto I, Yoneda H, Iwata N, Ohmori T, 2016. A significant causal association between C-reactive protein levels and schizophrenia. Sci Rep 6, 26105. [DOI] [PMC free article] [PubMed] [Google Scholar] [Retracted]
  18. Jia X, Han B, Onengut-Gumuscu S, Chen WM, Concannon PJ, Rich SS, Raychaudhuri S, de Bakker PI, 2013. Imputing amino acid polymorphisms in human leukocyte antigens. PLoS ONE 8 (6), e64683. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Joseph J, Depp C, Martin AS, Daly RE, Glorioso DK, Palmer BW, Jeste DV, 2015. Associations of high sensitivity C-reactive protein levels in schizophrenia and comparison groups. Schizophr Res 168 (1–2), 456–460. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Korn JM, Kuruvilla FG, McCarroll SA, Wysoker A, Nemesh J, Cawley S, Hubbell E, Veitch J, Collins PJ, Darvishi K, Lee C, Nizzari MM, Gabriel SB, Purcell S, Daly MJ, Altshuler D, 2008. Integrated genotype calling and association analysis of SNPs, common copy number polymorphisms and rare CNVs. Nat Genet 40 (10), 1253–1260. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Lawlor DA, Harbord RM, Sterne JA, Timpson N, Davey Smith G, 2008. Mendelian randomization: using genes as instruments for making causal inferences in epidemiology. Stat Med 27 (8), 1133–1163. [DOI] [PubMed] [Google Scholar]
  22. Lowe GD, 2005. Circulating inflammatory markers and risks of cardiovascular and non-cardiovascular disease. J Thromb Haemost 3 (8), 1618–1627. [DOI] [PubMed] [Google Scholar]
  23. McCarthy S, Das S, Kretzschmar W, Delaneau O, Wood AR, Teumer A, Kang HM, Fuchsberger C, Danecek P, Sharp K, Luo Y, Sidore C, Kwong A, Timpson N, Koskinen S, Vrieze S, Scott LJ, Zhang H, Mahajan A, Veldink J, Peters U, Pato C, van Duijn CM, Gillies CE, Gandin I, Mezzavilla M, Gilly A, Cocca M, Traglia M, Angius A, Barrett JC, Boomsma D, Branham K, Breen G, Brummett CM, Busonero F, Campbell H, Chan A, Chen S, Chew E, Collins FS, Corbin LJ, Smith GD, Dedoussis G, Dorr M, Farmaki AE, Ferrucci L, Forer L, Fraser RM, Gabriel S, Levy S, Groop L, Harrison T, Hattersley A, Holmen OL, Hveem K, Kretzler M, Lee JC, McGue M, Meitinger T, Melzer D, Min JL, Mohlke KL, Vincent JB, Nauck M, Nickerson D, Palotie A, Pato M, Pirastu N, McInnis M, Richards JB, Sala C, Salomaa V, Schlessinger D, Schoenherr S, Slagboom PE, Small K, Spector T, Stambolian D, Tuke M, Tuomilehto J, Van den Berg LH, Van Rheenen W, Volker U, Wijmenga C, Toniolo D, Zeggini E, Gasparini P, Sampson MG, Wilson JF, Frayling T, de Bakker PI, Swertz MA, McCarroll S, Kooperberg C, Dekker A, Altshuler D, Willer C, Iacono W, Ripatti S, Soranzo N, Walter K, Swaroop A, Cucca F, Anderson CA, Myers RM, Boehnke M, McCarthy MI, Durbin R, Consortium, H.R., 2016. A reference panel of 64,976 haplotypes for genotype imputation. Nat Genet 48 (10), 1279–1283. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Miller BJ, Culpepper N, Rapaport MH, 2014. C-reactive protein levels in schizophrenia: a review and meta-analysis. Clin Schizophr Relat Psychoses 7 (4), 223–230. [PubMed] [Google Scholar]
  25. Miller BJ, Culpepper N, Rapaport MH, Buckley P, 2013. Prenatal inflammation and neurodevelopment in schizophrenia: a review of human studies. Prog Neuropsychopharmacol Biol Psychiatry 42, 92–100. [DOI] [PubMed] [Google Scholar]
  26. Pouget JG, Gonçalves VF, Spain SL, Finucane HK, Raychaudhuri S, Kennedy JL, Knight J, Consortium, S.W.G.o.t.P.G., 2016. Genome-Wide Association Studies Suggest Limited Immune Gene Enrichment in Schizophrenia Compared to 5 Autoimmune Diseases. Schizophr Bull. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Prins BP, Abbasi A, Wong A, Vaez A, Nolte I, Franceschini N, Stuart PE, Guterriez Achury J, Mistry V, Bradfield JP, Valdes AM, Bras J, Shatunov A, Lu C, Han B, Raychaudhuri S, Bevan S, Mayes MD, Tsoi LC, Evangelou E, Nair RP, Grant SF, Polychronakos C, Radstake TR, van Heel DA, Dunstan ML, Wood NW, Al-Chalabi A, Dehghan A, Hakonarson H, Markus HS, Elder JT, Knight J, Arking DE, Spector TD, Koeleman BP, van Duijn CM, Martin J, Morris AP, Weersma RK, Wijmenga C, Munroe PB, Perry JR, Pouget JG, Jamshidi Y, Snieder H, Alizadeh BZ, Consortium, P., Consortium, I.S.G., consortium, S.S., consortium, T.O., Consortium, D., Consortium, C.D., consortium, A., Consortium, I.P.s.D.G., Consortium, A.S.D.W.G.o.t.P.G., consortium, C., Consortium, G., Pressure, I.C.f.B., Consortium, S.W.G.o.t.P.G., Consortium, I.W.G.o.t.C., 2016. Investigating the Causal Relationship of C-Reactive Protein with 32 Complex Somatic and Psychiatric Outcomes: A Large-Scale Cross-Consortium Mendelian Randomization Study. PLoS Med 13 (6), e1001976. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Purcell SM, Wray NR, Stone JL, Visscher PM, O’Donovan MC, Sullivan PF, Sklar P, 2009. Common polygenic variation contributes to risk of schizophrenia and bipolar disorder. Nature 460 (7256), 748–752. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Schizophrenia Working Group of the Psychiatric Genomics, C., 2014. Biological insights from 108 schizophrenia-associated genetic loci. Nature 511 (7510), 421–427. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Sekar A, Bialas AR, de Rivera H, Davis A, Hammond TR, Kamitaki N, Tooley K, Presumey J, Baum M, Van Doren V, Genovese G, Rose SA, Handsaker RE, Daly MJ, Carroll MC, Stevens B, McCarroll SA, 2016. Schizophrenia risk from complex variation of complement component 4. Nature 530 (7589), 177–183. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Talkowski ME, Mansour H, Chowdari KV, Wood J, Butler A, Varma PG, Prasad S, Semwal P, Bhatia T, Deshpande S, Devlin B, Thelma BK, Nimgaonkar VL, 2006. Novel, Replicated Associations Between Dopamine D3 Receptor Gene Polymorphisms and Schizophrenia in Two Independent Samples. Biol Psychiatry. 60 (6), 570–577. [DOI] [PubMed] [Google Scholar]
  32. Wium-Andersen MK, Ørsted DD, Nordestgaard BG, 2014. Elevated C-reactive protein associated with late- and very-late-onset schizophrenia in the general population: a prospective study. Schizophr Bull 40 (5), 1117–1127. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Zammit S, Allebeck P, Dalman C, Lundberg I, Hemmingsson T, Lewis G, 2003. Investigating the association between cigarette smoking and schizophrenia in a cohort study. Am J Psychiatry 160 (12), 2216–2221. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS932118-supplement-1.docx (224.1KB, docx)

RESOURCES