Brain, Behavior, and Immunity 42 (2014) 33–40

Contents lists available at ScienceDirect

Brain, Behavior, and Immunity

journal homepage: www.elsevier .com/locate /ybrbi

Named Series: Diet, Inflammation and the Brain

Measures of adiposity predict interleukin-6 responses to repeatedpsychosocial stress

http://dx.doi.org/10.1016/j.bbi.2014.07.0180889-1591/� 2014 Elsevier Inc. All rights reserved.

⇑ Corresponding author. Address: Department of Psychology and Volen NationalCenter for Complex Systems, Brandeis University, 415 South Street, Waltham, MA02453, United States. Tel.: +1 781 736 3319; fax: +1 781 736 3291.

E-mail address: rohleder@brandeis.edu (N. Rohleder).

Christine M. McInnis a, Myriam V. Thoma a, Danielle Gianferante a, Luke Hanlin a, Xuejie Chen a,Juliana G. Breines a, Suzi Hong b, Nicolas Rohleder a,⇑a Department of Psychology and Volen National Center for Complex Systems, Brandeis University, Waltham, MA, United Statesb Department of Psychiatry, University of California San Diego, La Jolla, CA, United States

a r t i c l e i n f o

Article history:Received 1 May 2014Received in revised form 22 July 2014Accepted 29 July 2014Available online 6 August 2014

Keywords:InflammationObesityAdiposityStressIL-6SensitizationTSST

a b s t r a c t

Objective: Overweight and obese individuals, who comprise approximately two-thirds of the U.S. popu-lation, are at increased risk for developing a range of diseases. This increased risk may be due in part tomaladaptive stress responses within this group, including heightened low-grade inflammation and HPAaxis non-habituation. In this study we tested the relationship between adiposity, plasma interleukin-6(IL-6) and HPA axis responses to repeated stress.Methods: Sixty-seven healthy participants were exposed to the Trier Social Stress Test (TSST) on two con-secutive days. We collected saliva for cortisol measurements at baseline and at 1, 10, 30, 60 and 120 minpost-TSST, and blood for plasma IL-6 measurements at baseline and 30 and 120 min post-TSST.Results: Stress exposure induced significant increases of cortisol and IL-6 on both days (cortisol: F = 38,p < 0.001; IL-6: F = 90.8; p < 0.001), and repeated exposure was related with cortisol habituation(F = 8.2; p < 0.001) and IL-6 sensitization (F = 5.2; p = 0.022). BMI and body fat were related with highercortisol responses to repeated stress (BMI: beta = 0.34; p = 0.014; body fat: beta = 0.29; p = 0.045), andwith higher IL-6 responses to repeated stress (BMI: beta = 0.27, p = 0.044; body fat: beta = 0.37;p = 0.006).Conclusions: Taken together, individuals with higher measures of adiposity showed less efficient HPA axishabituation as well as sensitization of IL-6 responses to repeated acute stress. These findings point tomaladaptive stress response patterns in overweight humans, which, through exposure to higher levelsof inflammatory mediators, might partially explain diseases related with overweight and/or obesity.

� 2014 Elsevier Inc. All rights reserved.

1. Introduction

Approximately two thirds of Americans are overweight orobese, and worldwide obesity has almost doubled since 1980(WHO, 2013). Obesity is a major risk factor for many conditionsincluding hyperlipidemia, hypertension, heart disease, stroke andType II Diabetes (Lapidus et al., 1984; Kissebah and Krakower,1994; Grundy, 2002; Lavie et al., 2009). Adipose tissue is recog-nized as an endocrine organ capable of regulating metabolic func-tion as well as secreting signaling molecules and cytokines(Trayhurn, 2005). Chronic low-grade inflammation is a hallmarkof obesity, and the dysregulated inflammation seen in obesity con-tributes to the pathology of a number of co-morbid conditions,

including atherosclerosis, type 2 diabetes and fatty liver disease(Danesh et al., 1998; Danesh, 1999; Black and Garbutt, 2002). Inaddition, obesity-related inflammation is emerging as a mecha-nism for increased cancer risk (Roberts et al., 2010).

While there is strong evidence supporting the association ofobesity with basal levels of inflammation (Rexrode et al., 2003;Panagiotakos et al., 2005; Himmerich et al., 2006; Thorand et al.,2006; Brydon, 2011), very little is known about how obesity affectspsychosocial stress-induced increases of IL-6 concentrations.Among normal weight individuals, acute psychosocial stressinduces an increase in plasma inflammatory molecules such asIL-6 (Steptoe et al., 2007), which does not typically habituate torepeated stress. Therefore, recurrent psychosocial stressors resultin repeated exposure to increased IL-6 (von Kanel et al., 2006;Rohleder, 2014). Due to its relatively slow response and recovery,and its failure to habituate to repeated stressors, it has been sug-gested that low-grade peripheral inflammation up-regulated bypsychosocial stress exposes individuals to sustained higher

34 C.M. McInnis et al. / Brain, Behavior, and Immunity 42 (2014) 33–40

concentrations of inflammatory mediators over time, which in turnmay increase disease risk (Rohleder, 2014). This risk might be evenmore exaggerated in overweight and obese individuals because ofthe relationship between adipose tissue and inflammation, as wellas increased basal IL-6 levels seen in overweight and obese individ-uals (Rexrode et al., 2003; Panagiotakos et al., 2005; Himmerichet al., 2006; Thorand et al., 2006; Brydon, 2011). One study exam-ined the effect of central adiposity on response to a single mildpsychological stressor in young women and found that while waistcircumference was related to baseline IL-6, there was no relation-ship between adiposity and IL-6 response to stressor (Brydon,2011). However, this study was done in a sample limited to youngwomen, with a mild psychological stressor administered only once.It is possible that differences in stress responses due to adiposityemerge with a more robust stressor administered repeatedly.

In addition to its relationship with chronic inflammation, thereis evidence that obesity modulates the glucocorticoid response tostress, although the literature is inconsistent with regard to thedirectionality of associations. Obesity has been found to be associ-ated with elevated baseline cortisol secretion and higher HPA-axisreactivity to psychological stress as well as physiological and phar-macological stimulation (Bjorntorp, 1993). In line with these ear-lier findings, greater cortisol responses to stress have recentlybeen noted in overweight women compared to leaner women(Benson et al., 2009). However, Jones et al. (2012) using a novel,magnetic resonance (MR) imaging-based method of body fat quan-tification found lower cortisol stress responses in individuals withadiposity (Jones et al., 2012). Only one study has addressed HPAaxis habituation to repeated stress. Individuals with a highWaist-to-Hip Ratio (WHR), indicative of increased central adipos-ity, showed less efficient cortisol habituation in response torepeated stress compared to lean individuals with a low WHR(Epel et al., 2000). Taken together, the relationships of measuresof adiposity with cortisol responses to psychosocial stressors aredocumented in a limited number of studies and findings areinconsistent.

To our knowledge, there have been no studies reporting theeffect of repeated psychosocial stress on low-grade inflammationin overweight individuals. In the present study we therefore aimedto examine whether measures of adiposity, including Body MassIndex (BMI), body fat percentage, waist circumference and waistto hip ratio, were associated with altered IL-6 and cortisolresponses to repeated stress. We hypothesized that greater adipos-ity will be associated with increased IL-6 responses as well asaltered cortisol responses, including less habituation to repeatedstress. We also expected to replicate previous findings of higherbaseline IL-6 and cortisol in overweight individuals.

2. Methods

2.1. Participants

Data were collected as part of a larger research project con-ducted over 2 years to investigate the effects of stress on endocrineand inflammatory parameters. Young adults (age 18–35 years) andolder adults (age 50–65 years) were recruited from the GreaterBoston area and the Brandeis University campus via newspaper,magazine, and Facebook advertisements. All participants under-went a brief medical and psychological screening by telephonebefore testing and were invited to participate only if they metthe following selection criteria: (a) body mass index (BMI) withinthe reference range between 18 and 35 kg/m2; (b) luteal phase ofmenstrual cycle at time of participation, for females; (c) absenceof psychiatric, endocrine, or cardiovascular diseases, or otherspecific chronic diseases; (d) no intake of psychoactive drugs,

beta-blockers, gonadal steroids (hormonal contraceptives), GCs;(e) non-smoker, and (f) no previous experience with the stress pro-tocol. Individuals were paid for their participation.

We recruited n = 72 individuals for participation in this study.Five participants were excluded from IL-6 analyses because theirIL-6 baseline or responses were greater than 2.5 standard devia-tions above the mean. In addition, one participant discontinuedafter the first session, and two participants displayed signs of infec-tion during session two, and therefore only their biological datafrom session one was used. This left a final sample of n = 67 forday 1 IL-6 and n = 64 for day 2 IL-6 analyses. N = 6 participantsincluded in the IL-6 analysis were excluded from cortisol analysesbecause they did not have cortisol data. N = 9 participants wereexcluded because their baseline cortisol was over 15 nmol/l. Thisleft a final sample of n = 56 participants for day 1 and 2 cortisolanalyses. Two participants were missing body fat percentage data.

2.2. Procedure

Eligible participants were scheduled for laboratory sessions ontwo consecutive days. All laboratory sessions were scheduled inthe afternoon (13:30–18:30 h) to control for circadian variationof cortisol, and participants came in at the same time for both ses-sions. Participants were instructed to refrain from eating or drink-ing anything but water for 1 h before the laboratory sessions.Written informed consent was obtained prior to participation.The Brandeis University Institutional Review Board approved allprocedures.

Each laboratory session lasted approximately 3 h and included a30-min resting period followed by exposure to the Trier SocialStress Test (TSST; Kirschbaum et al., 1993) Saliva samples for mea-surement of free cortisol were collected using Salivette collectiondevices (Sarstedt, Newton, NC) at baseline, as well as 1, 10, 30,60 and 120 min post TSST on both study days. Details on the salivacollection procedure are included below.

For assessment of IL-6 concentrations, blood was drawn via theantecubital vein using a peripheral venous catheter (BD Nexiva IVcatheter, Becton–Dickinson, Franklin Lakes, NJ) and collected inVacutainers (Becton–Dickinson, Franklin Lakes, NJ), containingEDTA. Initial placement of the catheter was followed by a restingperiod of 30 min to ensure recovery from potential stress responseto catheter placement or traveling to the laboratory. Because pre-vious research has found that IL-6 peaks 120 min post-stressor(von Kanel et al., 2006; Breines et al., 2014; Rohleder, 2014), bloodwas drawn at baseline, 30, and 120 min following the TSST on bothstudy days.

2.2.1. Stress induction paradigmAcute psychosocial stress was induced using the Trier Social

Stress Test (TSST, (Kirschbaum et al., 1993)), a widely used stan-dardized laboratory stress paradigm. The TSST used in the presentstudy consisted of a three-minute preparation period, a five-minute public speech, and a five-minute mental arithmetic taskin front of an audience of two judges wearing lab coats andmaintaining a neutral evaluative facial expression. The publicspeech involved describing how one’s personality makes onequalified for a dream job and the mental arithmetic ask involvedcounting backwards from 2043 by deduction of 17 on the firststudy day and from 2011 by 13 on the second study day. Partici-pants were informed that the judges were trained in analyzing ver-bal and non-verbal behavior and that their performance would bevideotaped. The TSST has demonstrated reliability and validity andhas been shown to produce strong biological responses to stress(Dickerson and Kemeny, 2004).

Table 1Characteristics of participants.

BMI (kg/m2) 24.9, SD = 3.5Body fat% 25.1, SD = 6.7WHR 0.85, SD = 0.08Waist (cm) 83.1, SD = 11.3Age (years) 37, SD = 17.8Sex 32M, 35F

Table 2Characteristics of participants included in IL-6 analysis.

Lean Overweight

BMI (kg/m2)*** 22.1, SD = 1.8 27.7, SD = 2.4 t(66) = �11.1, p < 0.001Body fat%*** 21.8, SD = 5.9 28.5, SD = 5.7 t(64) = �4.7, p < 0.001WHR* 0.82, SD = 0.07 0.87, SD = 0.09 t(66) = �2.4, p = 0.021Waist (cm)*** 76.2, SD = 7.5 89.8, SD = 10.5 t (66) = �6.3, p < 0.001Age (years) 36.1, SD = 18.2 37.6, SD = 17.6 t(66) = �0.4, nsSex 13M, 19F 19M, 16F X2 = 1.5, p = 0.22

Note: There were no differences in these characteristics between the larger sampleused for IL-6 analysis and the smaller sample for cortisol analysis.Note: ⁄⁄p < 0.01.

* p < 0.05.*** p < 0.001.

C.M. McInnis et al. / Brain, Behavior, and Immunity 42 (2014) 33–40 35

2.3. Measures

2.3.1. Self-reports of psychological healthDepressive symptoms were assessed using the 20-item Center

for Epidemiologic Studies Depression Scale (CES-D; (Radloff,1977)) by which respondents are asked to indicate how often theyhave felt or behaved in the stated manner over the past week,including statements such as ‘‘I felt depressed.’’ Ratings are madeon a 4-point scale (0 = Rarely or none of the time; 3 = Most or allof the time) and final scores are computed by summing scores onall items after reverse scoring four items. The CES-D has demon-strated reliability and validity (Radloff, 1977). In the present study,the CES-D showed good reliability (a = 0.89) and average scores fellbelow the clinical cut-off of 16 (M = 12.15; SD = 10.02, range 0–43)(Anderson et al., 2001).

Perceived stress over the past month was assessed using thePerceived Stress Scale (PSS, (Cohen et al., 1983)). The PSS isdesigned to measure how stressful an individual interprets theirlife to be. Respondents are asked how often they have experiencedcertain situations and how they have felt about them. Ratings aremade on a 5-point scale (0 = Never; 4 = Very often) and final scoresare computed by summing scores on all items after reverse scoringfour items. In the present study, the PSS showed good reliability(a = 0.87).

Positive and negative affect post-TSST was assessed using thePositive and Negative Affect Schedule (PANAS, (Watson et al.,1988)). Respondents are asked how strongly they feel positive(interested, excited, strong, enthusiastic, proud alert, inspired,determined, attentive and active) and negative (distressed, upset,guilty, scared hostile, irritable, ashamed, nervous, jittery andafraid) emotions. Ratings are made on a 5-point scale (1 = veryslightly or not at all; 5 = extremely) and greater scores on each sub-scale indicate greater positive or negative affect. In the presentstudy, the PANAS showed good reliability (positive affect Day 1:a = 0.93; negative affect Day 1: a = 0.90; positive affect Day 2:a = 0.93; negative affect Day 2: a = 0.87).

2.3.2. Measures of adiposityWeight and body fat measurements were taken using a Seca

Supra Plus 720 column scale (Seca, Hamburg, Germany), via a bio-electrical impedance analysis. Briefly, a current is passed throughthe participant and resistance is measured. The scale uses thisfeedback to calculate body fat and body water percentage (Fosterand Lukaski, 1996). Participants were asked to stand barefoot onthe scale and two measurements were taken per individual usinga normal body type and athletic body type measurement mode.Both measurements were averaged to obtain a measure that wasindependent of subjective judgment of participants’ body types.Height, waist and hip circumference measurements were takenwith a tape measure by the study nurse at the end of the first ses-sion. Waist measurements were taken midway between the ileaccrest and the lowest rib. Hip measurements were taken at the max-imum width of the buttocks or the gluteofemoral fold. BMI wasexpressed in kg/m2. WHR was calculated as waist circumferencedivided by hip circumference. Using established WHO guidelines,we split our sample into lean and overweight participants basedupon a BMI cutoff of 25 kg/m2 (WHO, 2013). This cut-off was usedfor graphing purposes only. The median BMI was 25. Characteris-tics of participants included in analyses are presented in Tables 1and 2.

2.3.3. Measurement of HPA axis and IL-6 stress responsesTo assess HPA axis responses to repeated acute stress, we mea-

sured salivary free cortisol. Saliva samples were collected at thetime points indicated above, salivettes were centrifuged, and saliva

was stored at �20 �C until processing. Cortisol was measured usinga competitive chemiluminescence immunoassay (CLIA; IBL-Inter-national, Toronto, ON, Canada). To assess systemic inflammationat baseline and in response to acute stress, we measured plasmainterleukin-6 (IL-6) at baseline (pre-TSST), as well as 30 and120 min post-TSST on both study days. Blood samples were centri-fuged immediately and plasma was aliquoted and stored at �80 �Cuntil batch processing. IL-6 concentrations were determined usinga commercial high-sensitivity ELISA (Quantikine HS; R&D Systems,Minneapolis, MN, USA), with a lower limit of detection for IL-6 of0.09 pg/ml. Inter- and intra assay coefficients of all assays werebelow 10%.

2.4. Statistical analyses

All statistical analyses were performed using SPSS 21 for MacOS X software packages (IBM, Chicago, IL, USA). Kolmogorov–Smirnov tests were computed prior to statistical analyses to testfor normal distribution as well as homogeneity of variance of alldependent variables. To test for stress-induced changes in cortisoland IL-6, we used analysis of variance (ANOVA) and analysis ofcovariance (ANCOVA) for repeated measures, with the within-subject factors ‘‘day’’ (day 1 vs. day 2) and ‘‘time’’ (three time pointsfor IL-6, and six time points for cortisol). We controlled for age andsex where appropriate, and we added the between-subjects factoroverweight vs. lean based on a BMI cut-off of 25 kg/m2 (see above)to test for differences in stress responses between overweight andlean participants. In all ANOVAs, Greenhouse–Geisser correctionswere applied if the sphericity assumption was violated (Vasey andThayer, 1987; Greenhouse and Junker, 1992). For an estimation ofcortisol stress reactivity to each of the TSSTs, we computed areaunder the curve with respect to increase (AUCi) indices (Pruessneret al., 2003). To estimate IL-6 stress reactivity, we computed deltascores by subtracting pre-stress IL-6 concentrations from IL-6 con-centrations 2 h post-TSST. To further examine the associations ofadiposity measures with cortisol and IL-6 stress responses, we per-formed linear regression analyses, controlling for age and sex in thefirst step, and entering each measure of adiposity as a predictor inthe second step. All reported results were considered to besignificant at the p6 0.05 level and were considered a trend at the

36 C.M. McInnis et al. / Brain, Behavior, and Immunity 42 (2014) 33–40

p6 0.1 level. Unless otherwise indicated, all reported values shownare untransformed means ± standard deviations (SD).

3. Results

3.1. Preliminary analyses

Women had higher body fat (M = 27.3%, SD = 5.4) than men(M = 22.8%, SD = 7.3; t(64) = �2.9, p = 0.005), lower WHR(M = 0.80, SD = 0.06) than men (M = 0.90, SD = 0.07) t(66) = 5.9,p < 0.001, as well as a lower waist circumference (M = 77.8 cm,SD = 8.6 cm) than men (M = 90 cm, SD = 10.8 cm) t(66) = 5.0,p < 0.001). There was no significant difference in BMI betweenmen and women. There were no correlations between BMI, bodyfat, WHR and waist circumference with age, CES-D, and PSS exceptfor waist with age (r = 0.35, p = 0.004), and body fat with PSS(r = 0.35, p = 0.005). There was no correlation between CES-D orPSS scores with baseline or stress responses of cortisol or IL-6 oneither of the study days. Higher waist circumference was relatedwith lower negative affect on day one (r = �0.26, p = 0.033), butnot on day two (r = �0.21, p = 0.10).

3.2. Adiposity and HPA axis

3.2.1. Cortisol stress responseWe first examined whether repeated TSST exposure induced

increases in salivary cortisol using repeated measures ANOVA.There was a significant time effect (F2.4,118 = 38.0, p < 0.001) indi-cating cortisol responses to stress, and a significant day by timeinteraction (F3.0,144.7 = 8.2, p = <0.001) indicating habituation ofcortisol responses on the second day. Further, we found a signifi-cant day by time by sex interaction indicating that men andwomen showed differences in HPA axis response and habituation,such that men showed stronger cortisol responses to TSST1 andgreater habituation to TSST2 (F3.0,144.6 = 2.9, p = 0.038). Addition-ally, there was a negative relationship between age and cortisolincreases (AUCi) to TSST1 (beta = �0.31, p = 0.022; R2 = 0.01).

3.2.2. Adiposity and baseline cortisolAs shown in Table 3, WHR and waist circumference were signif-

icantly or marginally significantly related to baseline cortisol onday 1, while BMI and body fat percentage were not related to base-line cortisol on day 1, when controlling for age and sex. BMI, bodyfat percentage and waist circumference were significantly, or mar-ginally significantly, related with higher baseline cortisol concen-trations on day 2, controlling for age and sex.

3.2.3. Adiposity and cortisol responses to repeated stressAs shown in Table 3, BMI, body fat, and WHR were not signifi-

cant predictors of cortisol response on day 1, while waist circum-ference was, when controlling for age and sex. BMI, body fatpercentage and waist circumference were significant predictorsof cortisol increase on day 2, while WHR was not, when controllingfor age and sex.

Table 3Relationships between measures of adiposity and cortisol baseline and responses.

Cortisol baseline day 1 Cortisol baseline da

Body mass index (BMI) Beta = 0.15; p = 0.27; R2 = 0.09 Beta = 0.27; p = 0.04Body fat% Beta = 0.19; p = 0.18; R2 = 0.09 Beta = 0.26; p = 0.06Waist circumference Beta = 0.29; p = 0.09; R2 = 0.12 Beta = 0.33; p = 0.04Waist-to-hip ratio (WHR) Beta = �0.34; p = 0.038; R2 = 0.14 Beta = 0.15; p = 0.93

All regression analyses controlled for age and sex in the first step.

In order to better understand and visualize the relationship ofBMI with cortisol responses, we performed a repeated measuresANOVA using BMI-based obesity grouping as a factor after splittingthe sample into lean and overweight groups at BMI = 25 kg/m2. Asshown in Fig. 2A and B, both, lean and overweight participantsshowed a strong response to the initial TSST and habituation torepeated stress. However, a marginally significant group by dayby time interaction indicates less efficient habituation in over-weight participants (F3.0,151.9 = 2.5, p = 0.061).

3.3. Adiposity and plasma interleukin-6

3.3.1. Interleukin-6. stress responseRepeated measures ANOVA of plasma IL-6 on both study days

with the factors sex and age group revealed a significant time effect(F1.1,146.2 = 90.8, p < 0.001), indicating an overall response of IL-6 onboth days. We further found a day by time interaction(F1.1,67.6 = 5.2, p = 0.022), indicating differential responses on thetwo study days. Overall, the IL-6 increase on Day 2 was higher thanthe increase on Day 1, showing sensitization of IL-6 stressresponses (t(63) = �2.54, p = 0.012). There were no age or sexgroup differences in IL-6 responses (all F < 1.1; p > 0.30). Therewas no correlation between age and IL-6 baseline or responses(all p’s > 0.46).

3.3.2. Adiposity and baseline interleukin-6As shown in Table 4, BMI body fat percentage and waist circum-

ference were significantly, or marginally significantly, related withhigher baseline IL-6 concentrations on both study days, with orwithout controlling for age and sex. WHR was only related withday 1 baseline IL-6.

3.3.3. Adiposity and interleukin-6 responses to repeated stressAs shown in Table 4, none of the adiposity measures were pre-

dictors of IL-6 stress responses on day 1, with or without control-ling for age and sex. However, IL-6 responses on Day 2 werepositively related with BMI and with body fat percentage control-ling for age and sex (see Fig. 1B). Waist circumference and WHRwere not significant predictors of IL-6 response on Day 2 with orwithout controlling for age and sex.

Since depression has been found to be associated with IL-6responses to stress, and with adiposity (Roberts et al., 2000;Fagundes et al., 2013), we additionally controlled for CES-D toensure that increased depression in overweight individuals wasnot driving the relationship between adiposity and sensitizedIL-6 responses on day 2. BMI remained a marginally significantpredictor of IL-6 increase on day 2 (beta = 0.25; p = 0.07;R2 = 0.08) when controlling for age, sex, and CES-D (CES-D:beta = 0.09). Body fat remained a significant predictor of IL-6 increaseon day 2 (beta = 0.36; p = 0.009; R2 = 0.14) when controlling for age,sex, and CES-D (CES-D; beta = 0.08). Waist circumference and WHRremain unrelated with IL-6 responses when controlling for CES-D(lowest p = 0.44).

Controlling for PSS, our main findings were also remainedsignificant or marginally significant. BMI remained a marginally

y 2 Cortisol response day 1 Cortisol response day 2

3; R2 = 0.17 Beta = 0.16; p = 0.22; R2 = 0.23 Beta = 0.34; p = 0.014; R2 = 0.174; R2 = 0.12 Beta = 0.12; p = 0.37; R2 = 0.22 Beta = 0.29; p = 0.045; R2 = 0.167; R2 = 0.17 Beta = 0.33; p = 0.04; R2 = 0.27 Beta = 0.39; p = 0.02; R2 = 0.17; R2 = 0.10 Beta = 0.22; p = 0.17; R2 = 0.22 Beta = �0.20; p = 0.22; R2 = 0.11

Table 4Relationships between measures of adiposity and IL-6 baseline and responses.

IL-6 baseline day 1 IL-6 baseline day 2 IL-6 response day 1 IL-6 response day 2

Body mass index (BMI) Beta = 0.41; p = 0.001; R2 = 0.16 Beta = 0.34; p = 0.011; R2 = 0.12 Beta = 0.09; p = 0.50; R2 = 0.02 Beta = 0.27; p = 0.044; R2 = 0.073Body fat% Beta = 0.39; p = 0.004; R2 = 0.13 Beta = 0.40; p = 0.003; R2 = 0.16 Beta = 0.17; p = 0.21; R2 = 0.03 Beta = 0.37; p = 0.006; R2 = 0.13Waist circumference Beta = 0.39; p = 0.017; R2 = 0.09 Beta = 0.33; p = 0.057; R2 = 0.08 Beta = 0.07; p = 0.69; R2 = 0.01 Beta = 0.17; p = 0.34; R2 = 0.02Waist-to-hip ratio (WHR) Beta = 0.42; p = 0.01; R2 = 0.10 Beta = 0.19; p = 0.28; R2 = 0.04 Beta = 0.10; p = 0.55; R2 = 0.02 Beta = 0.05; p = 0.77; R2 = 0.01

All regression analyses controlled for age and sex in the first step; neither age nor sex were significant predictors of baseline IL-6 concentrations except for sex on day 2 whenincluding body fat in the model (all other p > 0.08); all regressions were significant also without controlling for age and sex.All regression analyses controlled for age and sex in the first step; neither age nor sex were significant predictors of IL-6 increases (all p > 0.28); all regressions were significantalso without controlling for age and sex.

Fig. 1. Scatterplots showing the relationship of BMI with cortisol response to TSST2(A) and IL-6 response to TSST2 (B).

C.M. McInnis et al. / Brain, Behavior, and Immunity 42 (2014) 33–40 37

significant predictor of IL-6 increase on day 2 (beta = 0.25; p = 0.07;R2 = 0.08) when controlling for age, sex and PSS (PSS beta = 0.10).Body fat remained a significant predictor of IL-6 increase on day2 (beta = 0.36; p = 0.011; R2 = 0.13) when controlling for age, sexand PSS (PSS beta = 0.033). Waist circumference and WHR remainunrelated with IL-6 responses when controlling for PSS (lowestp = 0.44).

In order to better understand and visualize the relationship ofBMI with IL-6 responses to repeated stress, we again used the splitinto lean and overweight participants using a BMI cutoff of 25kg/m2. As shown in Fig. 2C and D, overweight individuals showedsimilar IL-6 increases on day 1, but higher increases than normalweight individuals on day 2. While repeated measures ANOVAdid not reveal any significant day, time, or day by time effects withour grouping factor (p > 0.20), paired samples t-tests comparingIL-6 increases from baseline to peak revealed that IL-6 responseswere significantly larger on Day 2 compared to Day 1 in theoverweight group (t(32) = �2.54, p = 0.016), but not the lean group(t(30) = �0.95, p = 0.35, see Fig. 3), indicating sensitization of IL-6stress responses.

3.4. Relationship between HPA axis and IL-6 responses to repeatedstress

To test whether greater cortisol responses were associated withlower IL-6 increases, we used linear regression. The relationshipbetween cortisol responses and IL-6 responses were not significanton either study day (p > 0.30).

4. Discussion

In this study, we aimed to investigate whether measures of adi-posity would be related with HPA axis and IL-6 responses, as wellas habituation or sensitization of these systems, in response torepeated acute psychosocial stress. Results confirmed previousfindings of stress response of the HPA axis and habituation, as wellas stress responses and non-habituation of IL-6 responses torepeated stress in the total sample (Schommer et al., 2003; vonKanel et al., 2006). Measures of adiposity were related with higherIL-6 baselines on both study days, and with cortisol baseline on theday of repeated stress exposure. With regard to our main hypoth-eses, we found that higher BMI and body fat percentage wererelated with higher cortisol stress responses to repeated stress,indicating less efficient habituation. Similarly, IL-6 responses torepeated stress were positively related with BMI and body fat, indi-cating sensitization of stress responses, in line with our hypothesesthat higher measures of adiposity would be related with alteredendocrine and IL-6 stress responses.

Our findings of less efficient HPA axis habituation in overweightindividuals, as well as the higher day 2 baseline cortisol values,suggest maladaptive alterations of HPA axis stress responses inoverweight individuals. Elevated basal cortisol secretion has beenlinked with excess central fat, in some (Mayo-Smith et al., 1989;Trayhurn, 2005), but not all studies (Epel et al., 2000). We foundhere that individuals with higher WHR had increased baseline cor-tisol only on day 1. However, the fact that in our study, we foundrelationships of baseline cortisol with BMI, body fat and waist cir-cumference only on the second day of stress testing supports thealternative interpretation that overweight individuals might comeinto repeated stress testing with higher anticipatory stress. Withregard to stress responses, we found that waist circumferencewas positively related to cortisol stress responses on both days,while BMI and body fat were predictors of cortisol responses torepeated exposures only. Earlier studies have revealed mixed find-ings, with some studies reporting positive associations of measuresof central (Marin et al., 1992; Moyer et al., 1994) and total adipos-ity (Benson et al., 2009) with cortisol stress responses, and othersreporting lower cortisol responses to one-time stress exposure inindividuals with higher body fat (Jones et al., 2012). Our findingsof a correlation of initial cortisol stress responses with waist cir-cumference, but not with WHR, BMI, and body fat percentage, isonly partially in line with this literature. These differences mightbe explained by different gender and age distribution, as well asdifferent ranges of adiposity between these studies. Further, there

Fig. 2. HPA axis responses to repeated stress in participants below (A: lean, n = 31) and above (B: overweight, n = 25) the BMI cut-off of 25 kg/m2. Graphs show means andstandard errors of the mean (SEM) of salivary cortisol concentrations at baseline as well as 1, 10, 30, 60, and 120 min post-TSST; inflammatory responses to repeated stress inparticipants below (C: lean, n = 32) and above (D: overweight, n = 35) the BMI cut-off. Graphs show means and SEM of plasma IL-6 concentrations at baseline as well as 30 and120 min post-TSST.

Fig. 3. No difference in IL-6 response to TSST1 and TSST2 in lean individuals, butoverweight (OW) individuals had a significantly greater increase in IL-6 in responseto TSST2 than TSST1 in overweight individuals.

38 C.M. McInnis et al. / Brain, Behavior, and Immunity 42 (2014) 33–40

seems to be a complex relationship between different measures ofadiposity. Jones et al. (2012) used a MR imaging based method ofbody fat assessment and found lower HPA axis responses in

relation to both BMI and visceral fat volume, while Epel et al.(2000) found higher HPA axis responses related with higher centraladiposity, but no significant differences between high and low BMIgroups. Epel et al.’s study is also the only other study employingrepeated stress exposure and our findings of less efficient HPA axishabituation in individuals with higher BMI and body fat percentageis in line with their results, but extends it to both genders and a lar-ger age range (Epel et al., 2000). Since our study was cross-sectional, we cannot determine whether obesity caused the alteredHPA axis activity and reactivity found here, or whether altered HPAaxis reactivity affects eating patterns and thereby might causeobesity. Studies showing that individuals with higher cortisolresponses to stress consumed more snack foods than individualswith lower cortisol responses might support this alternative path-way or a bi-directional relationship between adiposity and HPAaxis stress responses (Macht and Simons, 2000; Newman et al.,2007).

Our findings of relationships of baseline IL-6 levels with mea-sures of adiposity are in agreement with the well-documentedrelationship of adiposity with chronic low-grade inflammation(Rexrode et al., 2003; Panagiotakos et al., 2005; Himmerich et al.,2006; Thorand et al., 2006; Brydon, 2011). Only one prior studyfound that waist circumference was related to baseline IL-6, butnot with IL-6 responses to a single laboratory stress task (Brydon,2011). Consistent with this, we found in our study that IL-6 stressresponses were not related with waist circumference or WHR.Above and beyond these earlier results, however, we found herethat when exposed to the same stressor repeatedly, individuals

C.M. McInnis et al. / Brain, Behavior, and Immunity 42 (2014) 33–40 39

with higher BMI and body fat percentage showed significantlyhigher IL-6 responses to this repeated stressor, indicating sensiti-zation of acute stress responses.

While the current findings clearly document that individualswith higher body fat percentage have higher IL-6 responses whenexposed to the same stressor repeatedly, it remains an open ques-tion where the additional stress-induced plasma IL-6 protein isoriginating, and how this release is stimulated. Adipose tissue isthe likely source, as adipocytes secrete IL-6 and other pro-inflammatory cytokines (Trayhurn and Wood, 2004). Furthermore,adipose-tissue infiltrating macrophages secrete inflammatorycytokines, and adipose expression of cytokines is correlated withperipheral levels of these plasma proteins (Weisberg et al., 2003;Trayhurn and Wood, 2004; Lasselin et al., 2014). Increased leptinlevels in response to stress may drive synthesis of cytokines bymacrophages, as women with larger waists showed higher leptinresponses, and higher basal leptin was associated with greater IL-6 responses (Brydon, 2011). Leptin receptors are present on macro-phages, and binding stimulates macrophage production of IL-6(Gabay et al., 2001; La Cava and Matarese, 2004). It is thereforeconceivable that larger leptin responses in overweight individualsresult in greater IL-6 responses to stress. Analysis of stress-inducedleptin levels in a group with an established relationship betweenadiposity and plasma IL-6 response might shed light on whetherthe greater stress-induced IL-6 increase in overweight individualsis due to an increased leptin response to stress.

Higher IL-6 responses could theoretically be the result of lowerHPA axis responses, due to the well-described inhibitory effects ofglucocorticoids on the inflammatory cascade in general (Muncket al., 1984; Sapolsky et al., 2000), and because intracellular inflam-matory responses to stress have been found to be inversely relatedwith cortisol responses in particular (Wolf et al., 2009). However,our data do not support this possibility, because we find higher,not lower cortisol responses to repeated stress in individuals withhigher measures of adiposity, and we also do not find cortisol andIL-6 responses to be related on either of the two days of stressexposure.

Chronic low-grade inflammation, indicated by increased con-centrations of IL-6 and C-reactive protein, contributes to conditionsresponsible for some of the most common causes of morbidity andmortality, including Type 2 Diabetes, atherosclerosis, heart diseaseand cancer (Danesh, 1999; Roberts et al., 2010; Kotas et al., 2013),which provides an important link in obesity-related chronic dis-eases (Rexrode et al., 2003; Panagiotakos et al., 2005; Himmerichet al., 2006; Thorand et al., 2006; Brydon, 2011). The finding thatoverweight individuals also display sensitization of stress-inducedincreases in IL-6 is especially concerning in conjunction with theirwell-documented systemic low-grade inflammation. We see herethat a dysregulated IL-6 response to stress compounds thepre-existing condition of increased peripheral inflammation inobesity. Our work thereby underscores the importance of weightmaintenance in regards to the ability to mount adaptive stressresponses, and reducing risk for stress- and inflammation-drivenco-morbidities of obesity.

The current findings have to be interpreted in light of some lim-itations. First of all, the findings were based on the correlationalrelationships between measures of adiposity and endocrine andIL-6 stress responses. To draw causal conclusions, a prospectiveinvestigation or time-dependent experimental manipulationwould be necessary, for example employing a weight-loss inter-vention. However, given the current state of our knowledge, thiscross-sectional design provides a significant improvement andgroundwork for future studies, especially findings of acute stressresponses. Additionally, it may be possible that the stress effectsof session one have not completely recovered before session two.This does not seem to be an issue for plasma IL-6, but may in fact

be occurring for cortisol in the overweight group as their baselinecortisol on session two was elevated. Also, our sample size was rel-atively small, and a larger sample size may have allowed us todetect more subtle relationships. Finally, as our screening criteriarequired that individuals be free of potentially confounding healthconditions, we had a limited number of obese individuals includedin our analyses. Having a larger sample of obese participants mayfurther increase the generalizability of these findings to obese indi-viduals and may have allowed us to detect larger group differencesand greater effects of adiposity.

Taken together we show here that measures of adiposity arerelated with higher sensitization of IL-6 responses to acute stress,and with altered HPA axis responses to repeated stress, character-ized by higher baseline cortisol before repeated stress exposure,and most importantly, by less efficient habituation. Future studiesshould investigate whether overweight individuals showdecreased GC sensitivity on the second day, which could explaintheir sensitized IL-6 response without corresponding decreasedcortisol response and in fact decreased glucocorticoid sensitivityhas been reported in individuals with higher BMI (Wirtz et al.,2008). Additionally, examining leptin response in conjunction withplasma IL-6 response to stress in a population with a detectableinfluence of adiposity on IL-6 response would begin to help addressthe question of whether increased leptin response drives increasedimmune response in overweight and obese individuals.

Conflict of interest statement

All authors declare that there are no conflicts of interest.

Acknowledgments

This research was supported by the American Federation ofAging Research (NR), and by training grants from the NationalInstitute of Health (T32 MH 019929: CMM, T32-084907: DG, andT32-AG000204: JGB). MVT acknowledges funding from the SwissNational Science Foundation (SNF). SH acknowledges funding,R01 HL90975 from the National Institutes of Health.

References

Anderson, R.J., Freedland, K.E., Clouse, R.E., Lustman, P.J., 2001. The prevalence ofcomorbid depression in adults with diabetes: a meta-analysis. Diabetes Care 24(6), 1069–1078.

Benson, S., Arck, P.C., Tan, S., Mann, K., Hahn, S., Janssen, O.E., Schedlowski, M.,Elsenbruch, S., 2009. Effects of obesity on neuroendocrine, cardiovascular, andimmune cell responses to acute psychosocial stress in premenopausal women.Psychoneuroendocrinology 34 (2), 181–189.

Bjorntorp, P., 1993. Visceral obesity: a ‘‘civilization syndrome’’. Obes. Res. 1 (3),206–222.

Black, P.H., Garbutt, L.D., 2002. Stress, inflammation and cardiovascular disease. J.Psychosom. Res. 52 (1), 1–23.

Breines, J.G., Thoma, M.V., Gianferante, D., Hanlin, L., Chen, X., Rohleder, N., 2014.Self-compassion as a predictor of interleukin-6 response to acute psychosocialstress. Brain Behav. Immun. 37, 109–114.

Brydon, L., 2011. Adiposity, leptin and stress reactivity in humans. Biol. Psychol. 86(2), 114–120.

Cohen, S., Kamarck, T., Mermelstein, R., 1983. A global measure of perceived stress.J. Health Soc. Behav. 24 (4), 385–396.

Danesh, J., 1999. Smoldering arteries? Low-grade inflammation and coronary heartdisease. JAMA 282 (22), 2169–2171.

Danesh, J., Collins, R., Appleby, P., Peto, R., 1998. Association of fibrinogen, C-reactive protein, albumin, or leukocyte count with coronary heart disease:meta-analyses of prospective studies. JAMA 279 (18), 1477–1482.

Dickerson, S.S., Kemeny, M.E., 2004. Acute stressors and cortisol responses: atheoretical integration and synthesis of laboratory research. Psychol. Bull. 130(3), 355–391.

Epel, E.S., McEwen, B., Seeman, T., Matthews, K., Castellazzo, G., Brownell, K.D., Bell,J., Ickovics, J.R., 2000. Stress and body shape: stress-induced cortisol secretion isconsistently greater among women with central fat. Psychosom. Med. 62 (5),623–632.

40 C.M. McInnis et al. / Brain, Behavior, and Immunity 42 (2014) 33–40

Fagundes, C.P., Glaser, R., Hwang, B.S., Malarkey, W.B., Kiecolt-Glaser, J.K., 2013.Depressive symptoms enhance stress-induced inflammatory responses. BrainBehav. Immun. 31, 172–176.

Foster, K.R., Lukaski, H.C., 1996. Whole-body impedance–what does it measure?Am. J. Clin. Nutr. 64 (3 Suppl.), 388S–396S.

Gabay, C., Dreyer, M., Pellegrinelli, N., Chicheportiche, R., Meier, C.A., 2001. Leptindirectly induces the secretion of interleukin 1 receptor antagonist in humanmonocytes. J. Clin. Endocrinol. Metab. 86 (2), 783–791.

Greenhouse, J.B., Junker, B.W., 1992. Exploratory statistical methods, withapplications to psychiatric research. Psychoneuroendocrinology 17 (5), 423–441.

Grundy, S.M., 2002. Obesity, metabolic syndrome, and coronary atherosclerosis.Circulation 105 (23), 2696–2698.

Himmerich, H., Fulda, S., Linseisen, J., Seiler, H., Wolfram, G., Himmerich, S., Gedrich,K., Pollmacher, T., 2006. TNF-alpha, soluble TNF receptor and interleukin-6plasma levels in the general population. Eur. Cytokine Netw. 17 (3), 196–201.

Jones, A., McMillan, M.R., Jones, R.W., Kowalik, G.T., Steeden, J.A., Deanfield, J.E.,Pruessner, J.C., Taylor, A.M., Muthurangu, V., 2012. Adiposity is associated withblunted cardiovascular, neuroendocrine and cognitive responses to acutemental stress. PLoS ONE 7, e39143.

Kirschbaum, C., Pirke, K.M., Hellhammer, D.H., 1993. The ‘Trier Social Stress Test’–atool for investigating psychobiological stress responses in a laboratory setting.Neuropsychobiology 28 (1–2), 76–81.

Kissebah, A.H., Krakower, G.R., 1994. Regional adiposity and morbidity. Physiol. Rev.74 (4), 761–811.

Kotas, M.E., Gorecki, M.C., Gillum, M.P., 2013. Sirtuin-1 is a nutrient-dependentmodulator of inflammation. Adipocyte 2 (2), 113–118.

La Cava, A., Matarese, G., 2004. The weight of leptin in immunity. Nat. Rev.Immunol. 4 (5), 371–379.

Lapidus, L., Bengtsson, C., Larsson, B., Pennert, K., Rybo, E., Sjostrom, L., 1984.Distribution of adipose tissue and risk of cardiovascular disease and death: a12 years follow up of participants in the population study of women inGothenburg, Sweden. Br. Med. J. (Clin. Res. Ed.) 289 (6454), 1257–1261.

Lasselin, J., Magne, E., Beau, C., Ledaguenel, P., Dexpert, S., Aubert, A., Laye, S.,Capuron, L., 2014. Adipose inflammation in obesity: relationship withcirculating levels of inflammatory markers and association with surgery-induced weight loss. J. Clin. Endocrinol. Metab. 99 (1), E53–61.

Lavie, C.J., Milani, R.V., Ventura, H.O., 2009. Obesity and cardiovascular disease: riskfactor, paradox, and impact of weight loss. J. Am. Coll. Cardiol. 53 (21), 1925–1932.

Macht, M., Simons, G., 2000. Emotions and eating in everyday life. Appetite 35 (1),65–71.

Marin, P., Darin, N., Amemiya, T., Andersson, B., Jern, S., Bjorntorp, P., 1992. Cortisolsecretion in relation to body fat distribution in obese premenopausal women.Metabolism 41 (8), 882–886.

Mayo-Smith, W., Hayes, C.W., Biller, B.M., Klibanski, A., Rosenthal, H., Rosenthal,D.I., 1989. Body fat distribution measured with CT: correlations in healthysubjects, patients with anorexia nervosa, and patients with cushing syndrome.Radiology 170 (2), 515–518.

Moyer, A.E., Rodin, J., Grilo, C.M., Cummings, N., Larson, L.M., Rebuffe-Scrive, M.,1994. Stress-induced cortisol response and fat distribution in women. Obes.Res. 2 (3), 255–262.

Munck, A., Guyre, P.M., Holbrook, N.J., 1984. Physiological functions ofglucocorticoids in stress and their relation to pharmacological actions.Endocr. Rev. 5 (1), 25–44.

Newman, E., O’Connor, D.B., Conner, M., 2007. Daily hassles and eating behaviour:the role of cortisol reactivity status. Psychoneuroendocrinology 32 (2), 125–132.

Panagiotakos, D.B., Pitsavos, C., Yannakoulia, M., Chrysohoou, C., Stefanadis, C.,2005. The implication of obesity and central fat on markers of chronicinflammation: the ATTICA study. Atherosclerosis 183 (2), 308–315.

Pruessner, J.C., Kirschbaum, C., Meinlschmid, G., Hellhammer, D.H., 2003. Twoformulas for computation of the area under the curve represent measures oftotal hormone concentration vs. time-dependent change.Psychoneuroendocrinology 28 (7), 916–931.

Radloff, L.S., 1977. The CES-D Scale: a self-report depression scale for research in thegeneral population. Appl. Psychol. Meas. 1 (3), 385–401.

Rexrode, K.M., Pradhan, A., Manson, J.E., Buring, J.E., Ridker, P.M., 2003. Relationshipof total and abdominal adiposity with CRP and IL-6 in women. Ann. Epidemiol.13 (10), 674–682.

Roberts, D.L., Dive, C., Renehan, A.G., 2010. Biological mechanisms linking obesityand cancer risk: new perspectives. Annu. Rev. Med. 61, 301–316.

Roberts, R.E., Kaplan, G.A., Shema, S.J., Strawbridge, W.J., 2000. Are the obese atgreater risk for depression? Am. J. Epidemiol. 152 (2), 163–170.

Rohleder, N., 2014. Stimulation of systemic low-grade inflammation bypsychosocial stress. Psychosom. Med.

Sapolsky, R.M., Romero, L.M., Munck, A.U., 2000. How do glucocorticoids influencestress responses? Integrating permissive, suppressive, stimulatory, andpreparative actions. Endocr. Rev. 21 (1), 55–89.

Schommer, N.C., Hellhammer, D.H., Kirschbaum, C., 2003. Dissociation betweenreactivity of the hypothalamus–pituitary–adrenal axis and the sympathetic–adrenal–medullary system to repeated psychosocial stress. Psychosom. Med. 65(3), 450–460.

Steptoe, A., Hamer, M., Chida, Y., 2007. The effects of acute psychological stress oncirculating inflammatory factors in humans: a review and meta-analysis. BrainBehav. Immun. 21 (7), 901–912.

Thorand, B., Baumert, J., Doring, A., Herder, C., Kolb, H., Rathmann, W., Giani, G.,Koenig, W., Group, K., 2006. Sex differences in the relation of body compositionto markers of inflammation. Atherosclerosis 184 (1), 216–224.

Trayhurn, P., 2005. Endocrine and signalling role of adipose tissue: newperspectives on fat. Acta Physiol. Scand. 184 (4), 285–293.

Trayhurn, P., Wood, I.S., 2004. Adipokines: inflammation and the pleiotropic role ofwhite adipose tissue. Br. J. Nutr. 92 (3), 347–355.

Vasey, M.W., Thayer, J.F., 1987. The continuing problem of false positives inrepeated measures ANOVA in psychophysiology: a multivariate solution.Psychophysiology 24 (4), 479–486.

von Kanel, R., Kudielka, B.M., Preckel, D., Hanebuth, D., Fischer, J.E., 2006. Delayedresponse and lack of habituation in plasma interleukin-6 to acute mental stressin men. Brain Behav. Immun. 20 (1), 40–48.

Watson, D., Clark, L.A., Tellegen, A., 1988. Development and validation of briefmeasures of positive and negative affect: the PANAS scales. J. Pers. Soc. Psychol.54 (6), 1063–1070.

Weisberg, S.P., McCann, D., Desai, M., Rosenbaum, M., Leibel, R.L., Ferrante Jr., A.W.,2003. Obesity is associated with macrophage accumulation in adipose tissue. J.Clin. Invest. 112 (12), 1796–1808.

WHO, 2013. Obesity and Overweight. Fact Sheet No. 311. World HealthOrganization, Geneva.

Wirtz, P.H., Ehlert, U., Emini, L., Suter, T., 2008. Higher body mass index (BMI) isassociated with reduced glucocorticoid inhibition of inflammatory cytokineproduction following acute psychosocial stress in men.Psychoneuroendocrinology 33 (8), 1102–1110.

Wolf, J.M., Rohleder, N., Bierhaus, A., Nawroth, P.P., Kirschbaum, C., 2009.Determinants of the NF-kappaB response to acute psychosocial stress inhumans. Brain Behav. Immun. 23 (6), 742–749.