The study of biological markers has been gaining increased attention with regard to personalized treatment of mental disorders. Biomarkers can simultaneously be moderators and mediators of treatment response, providing enormous insight for future targets of drug and therapeutic development.
Several studies have investigated the role of inflammatory biomarkers in major depressive disorder (MDD). Two meta-analyses pointed out the relevance of interleukin (IL)-1, IL-6, tumor necrosis factor α (TNF-α), and C-reactive protein (CRP) in patients with MDD compared to controls. Moreover, a dose-response relationship between MDD and inflammatory biomarkers has been proposed (1,2). An earlier report suggests that a composite score that consists of several measures of concentrations of inflammatory cytokines can provide useful information about the diagnosis of MDD (3). Finally, 4 large trials are assessing the clinical use of a range of inflammatory biomarkers in the evaluation of treatment response in participants with MDD (4).
Because panic disorder (PD) presents high comorbidity with MDD and shares similar first-line treatments, comparisons in these disorders are often studied, and the investigation of common inflammatory markers can be of value. Furthermore, because there are high rates of anxiety disorders (ADs) in clinical samples of people with autoimmune diseases and reports that anti-inflammatory therapy can alleviate anxiety symptoms, it is possible that inflammatory cytokines might also play a role as biological markers in PD (5–9). Finally, higher cytokine states are reversible by some of the same therapeutic strategies effectively used in PD, including antidepressant medications (10), cognitive-behavioral stress management (11), and relaxation techniques (12).
Hoge et al. (13) suggested that there are altered levels of inflammation markers (IL-6, TNF- α, IL-10) in patients with PD compared to controls, but other studies found no significant differences in IL-6 and TNF-α between patients with current PD compared to controls (14,15). The inconsistent results provide little information about the use of these inflammatory markers in the context of PD.
Therefore, with the present study, we set out to explore the levels of serum inflammatory markers in a cohort of outpatients with a lifetime PD diagnosis. Our main hypothesis is that PD status may be positively associated with serum levels of proinflammatory cytokines (IL-6 and TNF- α) as well as negatively associated with serum levels of anti-inflammatory cytokine IL-10.
Design and Participants
Patients with lifetime PD who have been treated with cognitive behavioral therapy and antidepressants since 1998 (n = 155) were followed by the research team of the Anxiety Disorder Outpatient Unit of Hospital de Clínicas de Porto Alegre, Brazil, and were recontacted by phone calls, e-mails, or invitation letters between November 2011 and December 2012 to participate in this study.
During reevaluation, the clinicians confirmed the psychiatric diagnosis via a clinical semistructured interview and screening questions about the patients' general health. For the present study, we excluded current smokers and patients who reported any inflammatory disease (e.g., rheumatic condition, atopy, acute/chronic infection, Alzheimer disease), or diabetes. Of the 125 participants enrolled in the initial interview, 45 declined to participate or did not attend the scheduled evaluation. One participant with Parkinson disease and one with alcohol dependence were excluded after interview because of impairing clinical conditions not reported on the screening evaluation.
All included individuals provided written informed consent to participate in this study. The Hospital de Clínicas Ethics Research Committee approved all aspects of this research study (protocol number 11–0376).
After recruitment, patients underwent a structured clinical interview by 2 psychiatrists or a psychologist from our research staff. All who conducted the interviews had attended the required training sessions on semistructured diagnostic interviews and rating scales. Interviewers were unaware of individual cytokine levels of the participants during the interview. Self-report questionnaires were given to participants to complete at home.
Inter-rater reliability was based on 6 interviews from patients with appointments at the Anxiety Disorders Outpatients Unit from the same hospital. All 3 interviewers were in the same room with each patient and each psychiatrist/psychologist was the main interviewer twice while the others rated the psychometric instruments. The anxiety disorders section of the Brazilian Portuguese version of the Mini-International Neuropsychiatric Interview 5.0.0 (MINI) (16) and the Panic Disorder Severity Scale (PDSS) (17) were considered to compute agreement and inter-rater reliability.
Current psychiatric diagnosis was assessed with a clinical interview and the Brazilian Portuguese version of the Mini-International Neuropsychiatric Interview 5.0.0 (MINI) (16), which is a structured diagnostic interview widely applied and cross validated worldwide. It is used to assess psychiatric diagnosis according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (18,19). Patients were categorized according to their current diagnostic status into current or remitted PD, depending on their MINI interview. Such categorization was later used to compare both groups with regard to systemic inflammation. There was perfect agreement among the 3raters for PD diagnosis according to the MINI interview (Krippendorff alpha = 1.0).
The severity of PD was measured with the Panic Disorder Severity Scale (PDSS), a clinician-rated instrument that provides information on several dimensions of PD, including avoidance and anticipation. Individual responses are scored on a scale of zero to 4, and total scores range from zero to 28. This instrument has adequate psychometric properties across different settings (17), and Krippendorff alpha reliability coefficient for the 3 raters (inter-rater reliability) was 0.73.
The severity of anxiety symptoms was broadly assessed using the Hamilton Anxiety Rating Scale (HAM-A), a consistently validated instrument with 14 items. The score of each item (0–4) is summed for a total score, ranging from zero to 56. The total HAM-A scores indicate mild (18–24 points), moderate (25–30 points) or severe (>30 points) anxiety (20). The Hamilton Depression Rating Scale (HAM-D) was used to measure the severity of depressive symptoms within the sample. The HAM-D has also been extensively applied and validated across different settings (21).
The clinical interview also consisted of a general assessment of previous medical conditions. We asked for available tests/examinations to confirm the clinical diagnoses. To calculate the duration of psychiatric disease, patients were also asked to estimate the time from the first episode of PD. To assign 1 of 4 income classes (A, B, C, and D), we used Brazil's Economic Classification Criterion, which takes into account the amount of durable goods, number of employees, and income within each individual's family (22).
Inflammation Biomarkers and Anthropometric Measures
Anthropometric measures were assessed with a standard scale, inspected by the national weight and measures agency, and recorded to the nearest 0.1 kg during the patients' clinical interview. Body mass index (BMI) was calculated as weight/height2 (kg/m2).
Blood samples were collected by venipuncture in the morning after fasting (between 7 and 9 AM) to measure glucose, high-sensitivity CRP and lipid profile through a colorimetric enzymatic method (Modular; Roche, Mannheim, Germany). Moreover, we evaluated IL-6, IL-10, and TNF-α serum levels by flow cytometry using the BD Cytometric Bead Array assay (BD Biosciences, San Diego, CA) with human TNF-α, IL-6, and IL-10 enhanced sensitivity flex sets. This is a well-established and sensitive method of quantifying cytokines, previously validated against the criterion-standard enzyme-linked immunosorbent assay, with the advantage of detecting levels at the range below 1 pg/mL (23). Furthermore, since each of the enhanced sensitivity sets was used to quantify a single cytokine, it was assured that undesirable interactions among different antibodies and cytokines did not take place, which is a potential disadvantage of multiplexed assays (23).
Samples processing and data analysis were conducted according to manufacturers' instructions. Briefly, serum samples were incubated with the cytokine capture beads related to cytokines analyzed and with respective phycoerythrin-conjugated detection antibodies at room temperature and protected from light. Afterward, samples were washed and sample data were acquired using a FACSCalibur flow cytometer (BD Biosciences). The standard curves were generated in graphical and tabular format using the BD Cytometric Bead Array Analysis Software FCAP Array 3.0 (BD Biosciences), and the sample data were calculated using the 5-parameter logistic equation. The lower detection limits for IL-6, TNF-α, and IL-10 were 0.27, 0.10, and 0.11 pg/mL, respectively. Intra-assay and interassay coefficients of variation for individual cytokines were less than 10 %.
Independent sample t tests were used for comparisons of 2 groups with continuous, normally distributed data; Mann-Whitney tests were used for asymmetrically distributed data. To compare means across 3 or more groups, we used one-way analysis of variance. We used analysis of covariance to adjust for confounders. For analysis of categorical variables, the χ2 and the Fisher exact test were used whenever applicable. Shapiro-Wilks was used for normality testing. Levels of IL-6, IL-10, and TNF-α were normalized through log transformation. Log-transformed data were then compared using independent samples t test for 2-group comparisons.
Univariate linear regression models were built with variables empirically and/or theoretically associated with the dependent variable. We included in the final multivariate linear model only variables with p < .25 after univariate analysis (24).
Statistical analysis was performed using SPSS (version 20.0). A probability value of p < .05 was considered statistically significant.
For power calculations, we used g*Power version 18.104.22.168 (Düsseldorf University, Germany). Considering the naturalistic nature of the present study, power analysis was performed a posteriori. Because the only positive study on the association between inflammation and ADs did not provide sufficient data for computing effect size (13), we based power analysis on previous reports on inflammation markers and depression (1,2). Our study had 80% power to detect a medium effect size (d = 0.7) for all cytokines (considering unequal groups). Power calculation considered α = .05 and 2-tailed t tests with unequal groups (n = 22 and n = 54 for current vs remitted PD, respectively).
Descriptive characteristics of the sample (n = 78) are presented in Table 1. Patients were reassessed a median of 9 years (interquartile range, 4–12) after their initial psychiatric interview at our Anxiety Disorder Outpatient Clinic. Approximately two thirds of our sample (65%) was overweight or obese; approximately a fourth had elevated levels of fasting glucose (25%).
Comorbidity was somewhat frequent within the sample. Major depressive disorder was the most prevalent comorbid disorder (50%), followed by agoraphobia (44%), generalized anxiety disorder (25%), social anxiety disorder (12%), obsessive-compulsive disorder (6%), and posttraumatic stress disorder (3%).
Patients were categorized by their current diagnostic status of current or remitted PD. There were no significant demographic or clinical differences between the 2 groups (Table 1).
Panic Disorder and Inflammation Markers
Comparisons regarding inflammation markers between both PD groups (current vs remitted) are depicted in Figure 1A (IL-6), Figure 1B (TNF-α), and Figure 1C (IL-10). Inflammation was significantly associated with current PD when compared to remitted PD as evidenced by IL-6 (t  = 2.37; 95% confidence interval [CI] for the log-transformed mean difference, −0.41 to −0.57; p = .008; d = 0.6) but not by TNF-α ( t = −1.63; 95% CI, −1.12 to 0.11; p = 0.53; d = 0.4). We found no significant differences between PD groups (current vs remitted) for anti-inflammatory cytokine IL-10 ( t, 0.73; 95% CI, −0.20 to 0.44; p = 0.16; d = 0.2). Log-transformed levels of IL-6 remained significantly higher in patients with current PD when compared with those with PD in remission even after adjusting for BMI and fasting blood glucose of 5.6 mmol/L or greater [F(1,62) = 6.69, p = .012, d = 0.7, Bonferroni corrected].
No significant differences were found in log-transformed IL-6 levels between participants with and without comorbid MDD (t = 0.42, p = 0.68, d = 0.2), even after adjusting for PDSS, BMI, and fasting blood glucose of 5.6 mmol/L or greater [F(1, 61) = 0.17, p = 0.68, d = 0.1, Bonferroni corrected]. Similarly, no differences were found between both groups considering log-transformed levels of TNF-α (t  = 0.72, p = 0.72, d = 0.2) and IL-10 ( t  = −0.99, p = 0.33, d = 0.2), even after adjustment (data upon request).
To test the hypothesis that psychiatric comorbidity could play a role in the levels of inflammation markers, we categorized the sample according to the number of psychiatric diagnoses, including PD. There were 5 groups, ranging from 0 to 4 diagnoses. No statistically significant differences were found across groups for log-transformed levels of IL-6 [F(4, 71) = 1.22, p = 0.31, d = 0.2), TNF-α [F(4, 66) = 1.04, p = 0.39, d = 0.4), nor for IL-10 [F(4, 73) = 0.32, p = 0.87, d = 0.4].
Table 2 describes results for univariate and multivariate linear regressions with IL-6 as the dependent variable. Individual covariates were chosen according to consistent associations with IL-6 in previous studies. After univariate linear regression, BMI, reported clinical disease, fasting blood glucose of 5.6 mmol/L or greater, high-density lipoprotein, triglycerides, CRP of 4 mg/dL or greater, HAM-A, PDSS, and antidepressant use remained at the final multivariate model. Only BMI, fasting blood glucose of 5.6 mmol/L or greater, and PDSS remained significantly associated with higher levels of IL-6 in the final model, which explained nearly 42% of total IL-6 variance (adjusted R2 = 0.42). Both IL-6 and IL-10 were significantly correlated (r = 0.43, p = .048) in participants with current PD.
To assure that our results were not driven by the presence of outliers, we performed a sensitivity analysis excluding cases that exceeded the expected frequency considering Gaussian distribution (after log transformation). According to such criteria, a single case (case 36) had an unexpectedly high level of IL-6 (e.g., was situated more than 3 SDs greater than the mean) and was thus excluded in sensitivity analysis. We therefore repeated the 2-tailed t test without case 36 for comparing log-transformed levels of IL-6 between participants with current and remitted PD; results remained significant (t = −2.33, p = .023, d = 0.6). Similarly, the multivariate linear regression model of log-transformed levels of IL-6 was largely unchanged. Results of sensitivity analysis are illustrated in Supplementary Table S1 and Table S2, Supplemental Digital Content 1, http://links.lww.com/PSYMED/A319.
In this study, we showed that current PD was associated with higher mean levels of IL-6 compared to remitted PD in a sample of chronically affected outpatients. Furthermore, along with BMI and fasting blood glucose, PD severity as measured by PDSS was also associated with higher levels of IL-6 in the multivariate linear regression model. However, no significant differences were found between groups for TNF-α (proinflammatory) and IL-10 (anti-inflammatory) cytokines. To date, few studies have reported positive findings for the relationship between PD and IL-6. Therefore, our results corroborate a proinflammatory state within AD.
Hoge et al. (13), although using a different method for quantifying cytokines (multiplex assay, Luminex) and a different statistical approach, possibly due to highly asymmetrical data distribution, also found increased levels of IL-6 in patients with current Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition diagnoses of PD as compared to controls. However, contrary to our findings, these authors also found increased levels of TNF-α and IL-10 cytokines. Another study (25) using a similar methodology of high-sensitivity enzyme-linked immunosorbent assay, found no associations among anxiety symptoms, IL-6, and TNF-α after adjusting for unhealthy lifestyle factors. Despite a representative sample and adequate power, negative results may be explained by less severe psychiatric symptoms, since patients were recruited in community settings, whereas our sample was drawn from a specialized anxiety outpatient unit. Similarly, Tükel et al. (14) found no significant differences in IL-6 and TNF-α levels between patients with current PD and controls. However, the authors acknowledge that their study may have been underpowered to detect such differences. Another drawback of that study might have been a relatively low sensitivity to detect cytokine elevations, which was greater than 1 pg/mL range.
Our study shows that PD severity was associated with higher levels of serum IL-6 in the multivariate linear regression model, which represents a dose-response gradient between panic symptoms and inflammation. This is in line with a previous study by Murphy et al. (26) in which a significant correlation between IL-6 messenger RNA levels, and anxiety severity was reported. Additionally, BMI was strongly associated with higher IL-6 levels within our sample, which is consistent with other studies (27) and corroborates the representativeness of our sample.
Increased levels of IL-6, along with the absence of significant TNF-α changes compared to control participants, does not seem to be specific to PD. A meta-analysis by Steptoe et al. (28) showed that individuals after psychological stress show robust increases in levels of IL-6, but not TNF-α. No meta-analyses could be performed regarding IL-10 levels; in fact, the 3 included studies that reported results for IL-10 showed inconsistent results. On the other hand, another meta-analysis by Hiles et al. (10) showed that increased levels of IL-6, but not IL-10, were present in patients with MDD compared to controls. The proposed mechanism was that depressive patients might present an impaired leukocyte production of IL-10 in response to higher IL-6 levels, since both biomarkers are usually strongly correlated, but no such correlation was found in their study. However, the same cannot be generalized to PD, considering that both biomarkers were significantly correlated within our sample. Notwithstanding, although there is important overlap of common symptoms and the presence of comorbidity, both MDD and psychological stress are distinct from PD with regard to neurobiology. Therefore, similarity of inflammation patterns should be interpreted with caution.
No association was found between MDD and IL-6 levels in the present study. Moreover, increased severity of depressive symptoms was not associated with higher levels of serum IL-6. Both findings are somewhat surprising but confirm that our results are associated with PD and not depressive symptoms.
Higher scores in PDSS were significantly associated with higher blood levels of IL-6, which represents a biological gradient in the present sample. Using a pre-post test design, Engler et al. (29) injected bacterial endotoxin intravenously in 28 healthy volunteers. They measured mood and anxiety symptoms, as well as IL-6 blood levels and several other markers of inflammation response after the intervention. The authors reported that mean mood and anxiety symptoms increased significantly compared to baseline and that their peak levels overlapped IL-6 highest levels between 4 and 6 hours after endotoxin injection. Although the study used a different design and a healthy sample, results are consistent with ours. Neither causality nor directionality can be inferred from our results; however, we show for the first time a dose-response relationship in a sample of individuals chronically affected by anxiety disorders.
A neurovisceral integration model has been proposed by some authors to explain increased inflammation during stress; it postulates that a top-down central autonomic network (CAN), including structures within the forebrain, midbrain, and hindbrain, may be disinhibited across different anxiety conditions, resulting in impaired autonomic nervous system responses (30). Decreased vagal tone is a result of top-down CAN disinhibition, leading to decreased heart rate variability (HRV). Both impaired CAN inhibition by prefrontal cortex (31) and decreased HRV (32) have been observed, respectively, in participants with social anxiety disorder and PD. Furthermore, vagal activity and cytokine production may be closely linked. Marsland et al. (33) have shown an inverse association between vagal tone measured by HRV and lipopolysaccharide-induced IL-6 production in healthy volunteers. Moreover, Pellissier et al. (34) reported that patients with Crohn disease and elevated anxiety levels had a somewhat blunted vagal anti-inflammatory control over the production of TNF-α. Although no studies have yet investigated the interactions between PD and vagal inhibitory activity in cytokine production, we can speculate that such a phenomenon might play a role.
Several limitations should be considered when interpreting the present findings. First, sample size limitations did not allow us to test for more complex models, and effect sizes were small to moderate. Second, the cross-sectional nature and the lack of a HRV measure within the present research also limit further inferences.
Quantifying a limited number of serum inflammation markers, which were carefully selected based on an a priori hypothesis, is an important strength of the present study. By means of an accurate and validated method of quantifying the selected cytokines, using cytometric bead array with enhanced sensitivity flex sets—seldom applied in previous AD studies—this paper supports a positive dose-response association between IL-6 levels and current PD, reinforcing the idea that AD per se may also be associated with broad systemic dysfunctions. Because little is known about such associations, it may encourage replication in different settings.
The authors thank Giovanni Salum, MD, PhD, for his kind contribution when discussing our results.
Source of Funding and Conflicts of Interest: This work was supported by grants from the Research Incentive Fund (FIPE; Fundo de Incentivo a Pesquisa e Eventos) of Hospital de Clínicas de Porto Alegre, the Coordination for the Improvement of Higher Education Personnel (CAPES; Coordenação de Aperfeiçoamento de Pessoal de Nivel Superior), and the National Council for Scientific and Technological Development (CNPq; Conselho Nacional de Desenvolvimento Científico e Tecnológico). The authors report no conflicts of interest.
1. Howren MB, Lamkin DM, Suls J. Associations of depression with C-reactive protein, IL-1, and IL-6: a meta-analysis. Psychosom Med 2009;71:171–86.
2. Dowlati Y, Herrmann N, Swardfager W, Liu H, Sham L, Reim EK, Lanctôt KL. A meta-analysis of cytokines in major depression. Biol Psychiatry 2010;67:446–57.
3. Papakostas GI, Shelton RC, Kinrys G, Henry ME, Bakow BR, Lipkin SH, Pi B, Thurmond L, Bilello JA. Assessment of a multi-assay, serum-based biological diagnostic test for major depressive disorder: a Pilot and Replication Study. Mol Psychiatry 2013;18:332–9.
4. Oquendo MA, McGrath P, Weissman MM. Biomarker studies and the future of personalized treatment for depression. Depress Anxiety 2014;31:902–5.
5. Isik A, Koca SS, Ozturk A, Mermi O. Anxiety and depression in patients with rheumatoid arthritis. Clin Rheumatol 2007;26:872–8.
6. Kheirandish M, Faezi ST, Paragomi P, Akhlaghi M, Gharibdoost F, Shahali A, Fini ME, Akbarian M. Prevalence and severity of depression and anxiety in patients with systemic lupus erythematosus: an epidemiologic study in Iranian patients. Mod Rheumatol 2015;25:405–9.
7. Bae SC, Gun SC, Mok CC, Khandker R, Nab HW, Koenig AS, Vlahos B, Pedersen R, Singh A. Improved health outcomes with etanercept versus usual DMARD therapy in an Asian population with established rheumatoid arthritis. BMC Musculoskelet Disord 2013;14:13.
8. Favreau H, Bacon SL, Labrecque M, Lavoie KL. Prospective impact of panic disorder and panic-anxiety on asthma control, health service use, and quality of life in adult patients with asthma over a 4-year follow-up. Psychosom Med 2014;76:147–55.
9. Rohleder N. Stimulation of systemic low-grade inflammation by psychosocial stress. Psychosom Med 2014;76:181–9.
10. Hiles SA, Baker AL, de Malmanche T, Attia J. Interleukin-6, C-reactive protein and interleukin-10 after antidepressant treatment in people with depression: a meta-analysis. Psychol Med 2012;42:2015–26.
11. Antoni MH, Lutgendorf SK, Blomberg B, Carver CS, Lechner S, Diaz A, Stagl J, Arevalo JM, Cole SW. Cognitive-behavioral stress management reverses anxiety-related leukocyte transcriptional dynamics. Biol Psychiatry 2012;71:366–72.
12. Bhasin MK, Dusek JA, Chang BH, Joseph MG, Denninger JW, Fricchione GL, Benson H, Libermann TA. Relaxation response induces temporal transcriptome changes in energy metabolism, insulin secretion and inflammatory pathways. PLoS One 2013;8:e62817.
13. Hoge EA, Brandstetter K, Moshier S, Pollack MH, Wong KK, Simon NM. Broad spectrum of cytokine abnormalities in panic disorder and posttraumatic stress disorder. Depress Anxiety 2009;26:447–55.
14. Tükel R, Arslan BA, Ertekin BA, Ertekin E, Oflaz S, Ergen A, Kuruca SE, Isbir T. Decreased IFN-γ and IL-12 levels in panic disorder. J Psychosom Res 2012;73:63–7.
15. Vogelzangs N, Beekman AT, de Jonge P, Penninx BW. Anxiety disorders and inflammation in a large adult cohort. Transl Psychiatry 2013;23:e249.
16. Amorim P. Mini International Neuropsychiatric Interview (MINI): validação de entrevista breve para diagnóstico de transtornos mentais. Rev Bras Psiquiatr 2000;22:106–15.
17. Shear MK, Brown TA, Barlow DH, Money R, Sholomskas DE, Woods SW, Gorman JM, Papp LA. Multicenter collaborative panic disorder severity scale. Am J Psychiatry 1997;154:1571–5.
18. Sheehan DV, Lecrubier Y, Sheehan KH, Amorim P, Janavs J, Weiller E, Hergueta T, Baker R, Dunbar GC. The Mini-International Neuropsychiatric Interview (M.I.N.I.): the development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10. J Clin Psychiatry 1998;59:22–33.
19. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision. Washington, DC: American Psychiatric Association; 2000.
20. Hamilton M. The assessment of anxiety states by rating. Br J Med Psychol 1959;32:50–5.
21. Hamilton M. A rating scale for depression. J Neurol Neurosurg Psychiatry 1960;23:56–62.
22. IBGE: DemographicCensus [Internet]. 2000. Rio de Janeiro (Brasil): Instituto Brasileiro de Geografia e Estatística. [updated 2014 may 26; cited 2015 may 17]. Available from: www.ibge.gov.br/censo/
. Accessed May 17, 2015.
23. Leng SX, McElhaney JE, Walston JD, Xie D, Fedarko NS, Kuchel GA. ELISA and multiplex technologies for cytokine measurement in inflammation and aging research. J Gerontol A Biol Sci Med Sci 2008;63:879–84.
24. Hosmer DW, Lemeshow S. Applied Logistic Regression. New York: Wiley; 2000.
25. Duivis HE, Vogelzangs N, Kupper N, de Jonge P, Penninx BWJH. Differential association of somatic and cognitive symptoms of depression and anxiety with inflammation: findings from the Netherlands Study of Depression and Anxiety (NESDA). Psychoneuroendocrinology 2013;38:1573–85.
26. Murphy TM, O'Donovan A, Mullins N, O'Farrelly C, McCann A, Malone K. Anxiety is associated with higher levels of global DNA methylation and altered expression of epigenetic and interleukin-6 genes. Psychiatr Genet 2015;25:71–8.
27. Schmidt FM, Lichtblau N, Minkwitz J, Chittka T, Thormann J, Kirkby KC, Sander C, Mergl R, Faßhauer M, Stumvoll M, Holdt LM, Teupser D, Hegerl U, Himmerich H. Cytokine levels in depressed and non-depressed subjects, and masking effects of obesity. J Psychiatr Res 2014;55:29–34.
28. Steptoe A, Hamer M, Chida Y. The effects of acute psychological stress on circulating inflammatory factors in humans: a review and meta-analysis. Brain Behav Immun 2007;21:901–12.
29. Engler H, Benson S, Wegner A, Spreitzer I, Schedlowski M, Elsenbruch S. Men and women differ in inflammatory and neuroendocrine responses to endotoxin but not in the severity of sickness symptoms. Brain Behav Immun 2016;52:18–26.
30. Friedman BH. An autonomic flexibility-neurovisceral integration model of anxiety and cardiac vagal tone. Biol Psychol 2007;74:185–99.
31. Prater KE, Hosanagar A, Klumpp H, Angstadt M, Phan KL. Aberrant amygdala-frontal cortex connectivity during perception of fearful faces and at rest in generalized social anxiety disorder. Depress Anxiety 2013;30:234–41.
32. Martinez JM, Garakani A, Kaufmann H, Aaronson CJ, Gorman JM. Heart rate and blood pressure changes during autonomic nervous system challenge in panic disorder patients. Psychosom Med 2010;72:442–9.
33. Marsland AL, Gianaros PJ, Prather AA, Jennings JR, Neumann SA, Manuck SB. Stimulated production of proinflammatory cytokines covaries inversely with heart rate variability. Psychosom Med 2007;69:709–16.
34. Pellissier S, Dantzer C, Mondillon L, Trocme C, Gauchez AS, Ducros V, Mathieu N, Toussaint B, Fournier A, Canini F, Bonaz B. Relationship between vagal tone, cortisol, TNF-alpha, epinephrine and negative affects in Crohn disease and irritable bowel syndrome. PLoS One 2014;9:e105328.
Key words/Abbreviations: anxiety disorders; biochemical markers; neuroinflammation; panic disorder; interleukin-6; AD = anxiety disorders; CAN = central autonomic network; CBT = cognitive-behavioral therapy; CI = confidence interval; DSM-IV = Diagnostic and Statistical Manual of Mental Disorders, Fourth edition; HRV = heart rate variability; IQR = Interquartile range; MDD = major depressive disorder; PD = panic disorder