Inflammatory genotype moderates the association between anxiety and systemic inflammation in adults at risk for cardiovascular disease

Introduction

Cardiovascular disease (CVD) is the leading cause of death globally.¹ In 2016, CVD accounted for over 17.6 million deaths, and by 2030, it is expected to contribute to over 23.6 million deaths annually.¹ Preventing CVD remains a major challenge largely due to the multifactorial nature of CVD risk factors that range from genetic, to psychological and behavioral, to environmental factors, and interactions among these factors. While behavioral and environmental factors have been well studied, less is known about associations among psychological factors and CVD risk. For example, while anxiety is associated with increased risk for CVD,² the exact mechanism(s) underlying this relationship remain unclear.

Among the proposed mechanisms is a link between high anxiety and engagement in unhealthy lifestyle behaviors that, in turn, increase CVD risk.³ In contrast, others have reported that CVD risk is increased in individuals with high anxiety due to overactivation of both the sympathetic nervous system and the hypothalamic-pituitary-adrenal axis, leading to increased plasma catecholamine release, and resulting in endothelial damage, atherosclerosis, and coronary artery disease.³ Yet another proposed mechanism suggests that anxiety increases CVD risk through inflammatory pathways, where symptoms of anxiety and anxiety-based disorders increase levels of systemic inflammation through activation of the immune system that occurs in response to feelings of anxiety.⁴ This hypothesis has been supported by the results of two recent studies that have demonstrated associations between anxiety disorders and prospective increases in systemic inflammatory biomarkers in a samples of adolescents and adults, providing further evidence of increased inflammation as a consequence of anxiety.^4,5 However, evidence supporting this mechanism has been conflicting. While outcomes from some studies suggest that anxiety is positively associated with levels of systemic inflammatory biomarkers including C-reactive protein (CRP) and interleukin-6 (IL-6),^6–8 results from other studies have not supported these findings.^9,10 Further investigation into associations between anxiety and inflammation is warranted, as a clearer understanding of systemic effects of anxiety can improve CVD risk reduction strategies.

Within the human genome, the largest source of genetic variation between individuals comes from single nucleotide polymorphisms (SNPs) that are inherited. These single nucleotide differences in the DNA code can be located within a gene (i.e., in the exon code of the gene), in the regulatory region(s) of the gene (i.e., within the promotor, the enhancers, the 5’- and 3’- untranslated regions (UTR) and/or the RNA-splice sites of a gene), and in the region between two genes. Based on the location of a particular SNP relative to a gene, and the difference nucleotide possibilities at that SNP location (i.e., termed alleles), certain SNP alleles can impact gene expression and/or protein production. Inflammation is known to have a strong genetic component.¹¹ Interestingly, several SNPs located within or near inflammatory response genes, have been shown to be associated with increased levels of circulating inflammatory biomarker proteins. Among these are SNPs in the CRP gene (rs1205),¹² IL-6 gene (rs1800797),¹³ and Interleukin-6 Receptor (IL-6R) gene (rs4129267).¹⁴ For example, rs1205 is a SNP located within the 3’-UTR region of the CRP gene. Referred to as the major allele, the nucleotide cytosine (C) is most frequently found at this location. Thymine (T) is less common, and inheritance of the minor T allele at this location has been associated with decreased levels of serum CRP protein levels.^15,16 Similarly, rs1800797 is located within the IL-6 gene promotor at position (−597), and inheritance of the minor A (adenine) allele has been associated with increased systemic levels of IL-6 and decreased systemic levels of CRP when compared to individuals who inherited two guanine (G) alleles, referred to as the GG genotype.^13,17 Finally, rs4129267 is located in the intron of the IL-6R gene (i.e., rs412967 is often a proxy for rs2228145 Asp358Ala due to the linkage disequilibrium between the markers; r²=0.99), and inheritance of the minor T allele has been reported to be associated with increased soluble IL-6R levels and increased IL-6 plasma levels when compared to inheritance of the more common C allele.^18–23

While studies have been conducted to better understand how inflammatory genotypes interact with CVD risk factors to influence systemic inflammation levels,^11,24 few studies have been conducted to examine whether inflammatory gene variants moderate associations between anxiety and systemic inflammation. Gaining a better understanding of the relationship between anxiety and systemic inflammation, as well as genes that may affect this relationship, would enlarge our current knowledge base of health outcomes associated with anxiety, and ultimately aid in the identification of the mechanism(s) by which anxiety can influence CVD risk. The purpose of this study was to (1) examine the relationship between anxiety and systemic inflammation; and (2) determine if SNPs associated with inflammation moderate this relationship. We hypothesized that anxiety would be a predictor of systemic inflammation. Further, we hypothesized that inflammatory genotypes would moderate this relationship, with those with an rs1205 CC genotype, an rs1800797 A allele, or an rs4129267 T allele having higher levels of systemic inflammation in comparison to participants without these characteristics.

Methods

Study Design

This was a secondary analysis of data collected at baseline in a primary research study entitled “Gene Environment Interactions Regulating CVD Inflammation and Success of Behavioral Therapies.” The data for the primary research study was leveraged on the HeartHealth in Rural Kentucky study that was conducted from 2009 to 2012 to test the efficacy of a 12-week CVD risk reduction intervention. Participants in the HeartHealth study were community-dwelling rural Kentucky adults, aged 18 years or older, with two or more CVD risk factors. These risk factors included sedentary lifestyle; poor diet quality; current smoker or tobacco user; abnormal lipid levels; hypertension; being overweight or obese; and diagnosis of diabetes. The HeartHealth study protocol has been previously published.²⁵ The “Gene Environment Interactions Regulating CVD Inflammation and Success of Behavioral Therapies” study was conducted to investigate the impact of the HeartHealth intervention on systemic levels of inflammation and the influence of inflammatory genotypes on the response. Extensive baseline and post-intervention data were collected including sociodemographic data, clinical history, blood pressure, height, weight, lipid profiles, and smoking status. For the purposes of the leveraged study, serum for measuring systemic levels of inflammatory biomarkers and saliva for DNA extraction were collected from participants who completed baseline and post-intervention data collection.

Sample and Setting

Participants in the HeartHealth study were community-dwelling adults in rural Kentucky, who were recruited using purposive sampling from four rural Kentucky areas. Participants were age 18 years or older with two or more CVD risk factors, including: age > 44 years for men and >55 years for women, a positive family history of CVD, history of hypertension, abnormal lipid levels or diabetes; current smoker or tobacco user, body mass index >25 kg/m², a sedentary lifestyle, or a diet high in saturated fat. Individuals were excluded if they were non-English speaking, had a history of chronic drug abuse; end-stage renal, liver, or pulmonary disease; were diagnosed with active cancer; were diagnosed with gastrointestinal disease requiring special diets; had cognitive impairment; had physical activity limitations; or were taking protease inhibitors or other medications that interfere with lipid metabolism. Approval from the University of Kentucky Institutional Review Board was obtained prior to all study activities, and all participants provided written informed consent prior to commencement of study protocols.

Measures

Anxiety.

Anxiety was measured using the Brief Symptom Inventory (BSI) Anxiety subscale.²⁶ The scale consists of 6 items that are used to quantitate the degree of anxiety experienced by an individual. Each item is rated on a scale from 0 (not at all) to 4 (extremely) based on the distress caused by the symptom²⁶; the items were summed and then the average was obtained, with higher total scores indicative of greater anxiety.²⁷ The instrument has demonstrated excellent reliability and validity in non-cardiac and cardiac samples.^26,28 The Cronbach’s alpha value for the current study was 0.77.

Systemic inflammation.

CRP and IL-6 are key inflammatory biomarkers.^29,30 C-reactive protein and IL-6 are two of the most commonly measured biomarkers to determine levels of systemic inflammation.^29,30 Fasting blood samples were obtained in the morning for the purpose of measuring high sensitivity CRP (hsCRP) and serum IL-6 protein levels at baseline. For hsCRP, the Cholestech LDX® analyzer (Cholestech LDX Diagnostics, Hayward, CA), a point of care methodology, was used. The Cholestech LDX® system uses reflectance photometry, and has well-established accuracy and reproducibility for the detection of serum hsCRP.³¹

To measure serum IL-6 protein levels, 10 ml of whole blood was collected using anti-coagulant free red top vacutainers. Samples were maintained at room temperature for at least 30 minutes and no longer than 60 minutes to allow for clotting. Samples were then centrifuged at 1100 rpm for 15 minutes, after which serum was removed, aliquoted, and stored in cryovials at −80°C until analyzed. Serum IL-6 was analyzed using a 6-plex Millipore kit (EMD Millipore, Billerica, MA), and results were read using a Luminex IS100 (Luminex, Austin, TX), following manufacturers’ protocols.

SNP Genotyping.

Whole expectorated saliva was collected from participants using Oragene® DNA Collection Kits. From these samples, genomic DNA was isolated according to the manufacturer’s instructions.³² Purified DNA was suspended in 10mM Tris-HCl, 1 mM EDTA pH 8.0 (Thermo Fisher, Wilmington, DE), and DNA concentrations were measured using the NanoDrop-1000 spectrophotometer (Fisher Scientific, Fair Lawn, NJ). SNP genotyping of purified genomic DNA was performed utilizing Taqman® Genotyping Assay Kits and reagents (Applied Biosystems/Thermo Fisher Scientific. Carlsbad, CA) in the Roche LightCycler 480® Instrument (Roche Applied Science. Indianapolis, IN). For the purposes of this study, we examined a single SNP in each of the CRP (rs1205), IL-6 (rs1800797), and IL-6R (rs4129267) genes. All SNPs tests maintained Hardy-Weinberg equilibrium (p > 0.05). Genotypes were coded as follows: (1) for rs1205, participants homozygous for major C alleles were compared to those who were CT heterozygous and TT homozygous; (2) for rs1800797, participants who were major G allele homozygotes were compared to GA heterozygotes and minor allele A homozygotes; and (3) for rs4129267, participants homozygous for the major C allele were compared to those who were CT heterozygotes or were homozygous for the minor T alleles.

Body Mass Index.

Height and weight were measured with a professional grade stadiometer and a professional grade digital body weight scale at baseline for all participants. Body mass index (BMI) was calculated using the formula body weight (kilograms) divided by height (meters) squared, and entered into analyses as a continuous variable.

Smoking status.

Smoking status was determined from self-report. Participants were asked to identify if they were a never smoker, quit smoking 1 or more years, quit smoking less than one year ago, or were a current smoker. Responses were collapsed into 2 categories: participants who had never smoked or had quit 1 or more years prior were categorized as non-smokers; participants who had quit less than 1 year ago or were current smokers were categorized as current smokers.

Demographic characteristics.

Demographic information was collected via self-report questionnaire. Data included age in years, race/ethnicity, gender, marital status, education level, and socioeconomic status. Socioeconomic status was assessed using a measure of financial comfort in which participants were self-categorized into 3 groups: having more than enough to make ends meet, enough money to make ends meet, or having not enough money to make ends meet. For race/ethnicity, participants were asked to self-identify as being non-Hispanic white or Caucasian, non-Hispanic black or African American, Asian, Hispanic or Latino, American Indian or Alaskan Native, Native Hawaiian or other Pacific Islander, or other. Reflective of the rural Kentucky population, 89% of participants in the HeartHealth study self-identified as being non-Hispanic white or Caucasian. To control for population stratification, analyses in this study were limited to participants who self-identified as being non-Hispanic white.

Data Analysis

Descriptive analyses included frequency distributions or means and standard deviations, as appropriate for each variable of interest. Chi square analyses and independent t-tests were used to assess sociodemographic differences between genotypes.

Multiple linear regression modeling was conducted to examine whether anxiety was associated with serum CRP and IL-6, adjusting for age, gender, BMI, and smoking status. To test moderation effects, hierarchical multiple linear regression modeling was conducted to examine if rs1205, rs1800797, or rs4129267 genotypes moderated associations between anxiety and serum hsCRP or IL-6. In each of the models, block one included age, gender, BMI, smoking status, and anxiety score, and genotype for either rs1205, rs1800797, or rs4129267, and the outcome variable of either serum hsCRP levels or serum IL-6 protein levels. An interaction term for anxiety and each genotype (anxiety * genotype) was entered in the second block of the regression model to test whether the relationship between anxiety and systemic inflammation varied according to genotype. For each significant interaction term, a subsequent simple slope analysis was conducted to determine the nature of the moderation effect.

To limit the overall Type I error rate, serum CRP and IL-6 protein levels were selected as the only outcome variables (i.e., indicators of the systemic inflammatory status) based on the reviewed literature.^2,6–8 For each outcome variable (CRP and IL-6 protein levels), three models were considered (one for each genotype). Given the purposeful limit imposed on number of cytokines and number of genotypes, we maintained an a priori alpha level of .05 throughout. Consistent with prior analyses and as a correction for skewed distributions, both hsCRP and IL-6 variables were log-10 transformed prior to analysis.

All data analyses were performed using SPSS Statistics for Windows version 26 (IBM Corp., Armonk, N.Y., USA) except for simple slope analysis, for which ModGraph Internet Version, version 3.0 (Victoria University of Wellington, Wellington, New Zealand) was used.^33,34

Results

Sample Characteristics

Characteristics of the sample (N = 398) are summarized in Table 1. The majority of participants were female (73.4%) with a mean age of 51.4 ± 13.3 years. Most of the sample were non-smokers (88.7%). The mean BMI for this sample was 32.8 ± 7.4 kg/m². There were no significant differences in sociodemographic characteristics found when comparing genotypes for all three SNPs examined (data not shown).

Table 1.

Participant Characteristics.

	Total Sample (N=398)
Age (years)	51.4 ± 13.4
Gender (male)	106 (26.6%)
Marital status
Single, divorced, widowed	94 (23.6%)
Married or cohabitating	304 (76.4%)
Education
High school or less	136 (34.2%)
More than high school	261 (65.6%)
Financial comfort
Not enough to make ends meet	8 (32.0%)
Enough to make ends meet	211 (53.0%)
More than enough to make ends meet	155 (38.9%)
Never or former smoker	353 (88.6%)
Body mass index (kg/m2)	32.8 ± 7.4
Anxiety score	0.5 ± 0.5
hsCRPa (mg/L)	3.3 ± 3.7
IL-6b (pg/L)	16.4 ± 34.3

Values reported as either mean ± SD or n (%).

^ahigh sensitivity C-reactive protein

^bInterleukin-6

Associations between anxiety and levels of inflammatory biomarkers

Table 2 shows results of the regression analysis conducted to examine if anxiety predicted serum hsCRP protein levels controlling for age, gender, BMI, and smoking status. While the overall model was significant (F(5, 392) = 24.45, p < 0.001), this was due to the significant associations of gender, smoking status and BMI with serum hsCRP protein levels; anxiety was not a significant predictor of hsCRP protein levels. Additionally, the regression model examining whether anxiety predicted serum IL-6 protein levels was not significant (Table 3).

Table 2.

Summary of Multiple Linear Regression Predicting Serum High Sensitivity C-reactive Protein Level (N = 398)

Variable	B	β	95% CI for B	p value
Age	0.002	0.057	(−0.001, 0.006)	0.209
Gender (male)*	0.218	0.191	(0.116, 0.319)	<0.001
Body mass index	0.029	0.425	(0.023, 0.035)	<0.001
Smoking status (never or former)*	0.163	0.102	(0.022, 0.304)	0.023
Anxiety score	0.064	0.069	(−0.019, 0.148)	0.128

Overall: R² = 0.238, p <0.001

^*denotes reference group

Table 3.

Summary of Multiple Linear Regression Predicting Serum Interleukin-6 Protein Level.

Variable	B	β	95% CI for B	p value
Age	0.001	0.016	(−0.003, 0.005)	0.753
Gender (male)*	−0.072	−0.061	(−0.190, 0.046)	0.232
BMI	0.004	0.062	(−0.003, 0.011)	0.216
Smoking status (never or former)*	−0.133	−0.081	(−0.297, 0.031)	0.111
Anxiety score	0.100	0.103	(0.003, 0.197)	0.043

Overall: R² = 0.022, p = 0.116

^*denotes reference group

Moderating effects of inflammatory genotypes on associations between anxiety and levels of inflammatory biomarkers

Three models were analyzed to examine if the rs1205, rs1800797, or rs4129267 SNPs moderated the association between anxiety and serum IL-6 protein level. Of the three models, the two hierarchical linear regression models examining whether the rs1205 or rs1800797 SNPs moderated the association between anxiety and serum IL-6 protein level were not found to be significant (F(7, 390) = 1.607, p = 0.132 and F(7, 390) = 1.560, p = 0.146, respectively). The third model, in which the moderation effect of the IL-6R gene SNP, rs4129267, was examined was found to be significant (R² = 0.042, F(7, 390) = 2.42, p = 0.019). In this final model, the interaction term between anxiety and rs4129267 genotype was significant (β = −0.235, p = 0.010), indicating a significant moderation effect. The subsequent simple slope analysis revealed that anxiety was only significantly associated with IL-6 protein levels for those with an rs4129267 CC genotype (b = 0.243, SE = 0.04, p <0.001) and not for those with a CT or TT genotype (p = 0.770) (Figure 1).

Figure 1.

Association between anxiety and IL-6 by rs4129267 genotype

Three additional models examined whether rs1205, rs1800797, or rs4129267 SNPs moderated the association between anxiety and serum CRP level. In these models, there was no evidence that rs1205, rs1800797, or rs4129267 genotypes moderated the association between anxiety and serum CRP level.

Discussion

The relationship between anxiety and inflammation has been inconsistently supported in the literature,^9,10 and was not supported by the findings from our study. Results from previous studies have indicated that anxiety is associated with increased levels of systemic inflammation,^3,6,35,36 supporting the hypothesis that symptoms of anxiety activate the sympathetic nervous and immune systems which, in turn, stimulate the secretion of pro-inflammatory cytokines leading to increased systemic inflammation levels.^6,37,38 However, in accord with the results of our study, anxiety was not significantly associated with either serum hsCRP or IL-6 protein levels. Similar to our findings, the results of a study conducted to examine associations between anxiety and hsCRP, IL-6 and Tumor Necrosis Factor alpha (TNFα) in 2,861 adults were non-significant when lifestyle factors were controlled for, but path analysis did show a mediating effect of BMI.⁹ This and other similar studies³⁹ suggest that the inconsistency of associations between anxiety and inflammation may be due to other factors that may influence this relationship.

Genetic variants previously associated with increased systemic levels of inflammatory biomarkers were examined as potential moderators of the association between anxiety and inflammation.^15,24,40 Of the three SNPs (rs1205, rs1800797, and rs4129267) examined in this study, the rs4129267 SNP, located in intron 8 of the the IL-6R gene, was found to have a moderating effect on the association between anxiety symptoms and serum IL-6 levels. Results of the simple slope analysis indicated that the association between anxiety and inflammation was only significant for individuals with the rs4129267 CC genotype. The IL-6R SNP, rs4129267, examined in this study has been reported to be in complete linkage disequilibrium with a second IL-6R SNP, rs2228145, with the minor C allele of rs2228145 co-inherited with the minor T allele of rs4129267.¹⁸ The rs2228145 SNP (previously referred to a rs8192284⁴¹) is responsible for an amino acid change, Asp(358)Ala, that results in a modification to the structure of the IL-6R.¹⁸ This amino acid change increases the cleavage of the IL-6R from the cell surface into the extracellular space, where the soluble IL-6R can then associate with free IL-6. This resulting complex is then recognized by the glycoprotein 130 structures that exists on the membrane of most cells in the body, providing an alternative IL-6 signaling pathway, known as IL-6 trans-signaling.^19,20,42 As compared to classic IL-6 signaling, where IL-6 associates with membrane-bound IL-6R that are expressed by only a few types of cells in the body, IL-6 trans-signaling is reported to be a pro-inflammatory process, and may explain findings from this study.¹⁹

The pro-inflammatory nature of the T allele of rs4129267 is reflected in our findings. There was no significant association between anxiety and IL-6 levels for those with the rs4129267 CT or TT genotype, but rather, as demonstrated in the simple slope analysis results, those with the CT or TT genotype had higher IL-6 levels, with no regard to level of anxiety. This suggests that individuals with CT or TT rs4129267 genotypes may experience persistent elevated systemic inflammation regardless of anxiety levels. Conversely, our findings suggest that those with an rs4129267 CC genotype, which has not been reported to be associated with increased systemic inflammation, may experience an increased inflammatory response to stressors, such as anxiety.

Other SNPs examined did not have a moderating effect on associations between anxiety and levels of serum hsCRP or IL-6. Among these was rs1205. Located within the CRP gene, the minor T allele of rs1205 has been associated with lower circulating levels of CRP.^15,16,43 In the two models which examined rs1205 genotype, genotype was not a significant predictor of CRP or IL-6 levels, nor was there evidence that genotype moderated the relationship between anxiety and systemic inflammation. The second SNP examined in this study was rs1800797, located on the IL-6 gene. The minor A allele of the rs1800797 SNP has been associated with increased IL-6 levels and decreased CRP levels when compared to GG genotypes.^13,17 Like rs1205, rs1800797 was not found to be associated with IL-6 or CRP levels, nor did it have a moderating effect on the relationship between anxiety and inflammation.

Of note, in the two models examining the moderation of rs1205 and rs1800797 on the association between anxiety and CRP, gender, BMI, and smoking status were found to be significant predictors of CRP. These are expected findings, as female gender, increased BMI, and smoking are well-established predictors of systemic inflammation, and are commonly adjusted for in analyses examining novel associations with systemic inflammation.⁴⁴

Our study adds to the current literature that reports an association between anxiety and systemic inflammation by suggesting that certain inflammatory genotypes may be more susceptible to the negative effects of anxiety, resulting in increased inflammation. Few studies have been conducted to examine associations between anxiety and inflammation, and fewer still to test moderating effects of inflammatory genotypes on these associations. The findings from our study add to the knowledge of how specific genotypes might place some individuals at risk for increased systemic inflammation that can, in turn, increase risk for CVD. Additional research is needed to examine these relationships further, and to replicate and validate the findings of this preliminary study.

Strengths and Limitations

To control for population stratification, the analyses were limited to participants who self-identified as non-Hispanic white or Caucasian. Second, while we had a large sample representative of rural populations at high risk for CVD, it will be of interest in future studies to examine these relationships in more racially and geographically diverse samples, as there may be significant differences in SNPs based on race, ethnicity, and geographical regions. Additional studies with larger samples will be critical both to verify our preliminary findings, and to support conducting a larger number of analyses with correction for multiple comparisons.

Conclusion

Anxiety is associated with increased risk for CVD. The mechanism that underlies this relationship is not known, although inflammatory pathways have been hypothesized to be a potential mechanism. In this study, anxiety was positively associated with serum IL-6 protein levels, but a moderation analysis indicated that this association was significant only for individuals with the rs4129267 CC genotype. These findings suggest that there may be genotypic differences in individuals’ responses to anxiety, which may place certain individuals at higher risk than others for inflammation, and subsequently CVD. Further studies are needed to provide more robust evidence of these relationships, from which interventions can be tailored based on genotypic differences to improve CVD prevention efforts.

Table 4.

Summary of hierarchical multiple linear regression to examine moderating effects of rs4129267 genotype on associations between anxiety and serum IL-6 protein level.

Variable	B	β	95% CI for B	p value	R	R2	R2 change
Step 1					0.159	0.025	0.003
Age	0.001	0.015	(−0.003, 0.004)	0.772
Gender (male)*	−0.073	−0.062	(−0.191, 0.045)	0.227
BMI	0.004	0.062	(−0.003, 0.011)	0.222
Smoking status (never or former)*	−0.133	−0.081	(−0.296, 0.031)	0.112
Anxiety score	0.100	0.103	(0.003, 0.196)	0.044
rs4129267 (CC)*	0.057	0.054	(−0.047, 0.160)	0.281
Step 2					0.204	0.042	0.017
Age	0.000	0.008	(−0.004, 0.004)	0.867
Gender (male)	−0.075	−0.064	(−0.192, 0.042)	0.207
BMI	0.005	0.065	(−0.002, 0.011)	0.197
Smoking status (never or former)*	−0.143	−0.087	(−0.306, 0.020)	0.085
Anxiety score	0.243	0.251	(0.098, 0.388)	0.001
rs4129267 (CC)*	0.191	0.182	(0.047, 0.336)	0.010
Anxiety * rs4129267	−0.252	−0.235	(−0.443, −0.061)	0.010

^*denotes reference group