Logo of nihpaAbout Author manuscripts Submit a manuscript HHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Psychoneuroendocrinology. Author manuscript; available in PMC 2014 Aug 1.
Published in final edited form as:
PMCID: PMC4104274
NIHMSID: NIHMS597267
PMID: 24882154

Chronic Stress Increases Vulnerability to Diet-Related Abdominal Fat, Oxidative Stress, and Metabolic Risk

Kirstin Aschbacher, Ph.D.,1,2 Sarah Kornfeld, M.A.,3 Martin Picard, Ph.D.,4 Eli Puterman, Ph.D.,1 Peter Havel, Ph.D.,5 Kimber Stanhope, Ph.D.,5 Robert H. Lustig, M.D., M.S.L.,6,7 and Elissa Epel, Ph.D.1
Author information Copyright and License information Disclaimer

Associated Data

Supplementary Materials

Summary

Background

In preclinical studies, the combination of chronic stress and a high sugar/fat diet is a more potent driver of visceral adiposity than diet alone, a process mediated by peripheral Neuropeptide Y (NPY).

Methods

In a human model of chronic stress, we investigated whether the synergistic combination of highly palatable foods (HPF; high sugar/fat) and stress was associated with elevated metabolic risk. Using a case-control design, we compared 33 post-menopausal caregivers (the chronic stress group) to 28 age-matched low-stress control women on reported HPF consumption (modified Block Food Frequency Questionnaire), waistline circumference, truncal fat ultrasound, and insulin sensitivity using a three-hour oral glucose tolerance test. A fasting blood draw was assayed for plasma NPY and oxidative stress markers (8-hydroxyguanosine and F2-Isoprostanes).

Results

Among chronically stressed women only, greater HPF consumption was associated with greater abdominal adiposity, oxidative stress, and insulin resistance at baseline (all p’s ≤.01). Furthermore, plasma NPY was significantly elevated in chronically stressed women (p<.01 and the association of hpf with abdominal adiposity was stronger among women high versus low npy. there were no significant predictions change over one-year likely due to stability in primary outcomes this period.>

Discussion

Chronic stress is associated with enhanced vulnerability to diet-related metabolic risk (abdominal adiposity, insulin resistance, and oxidative stress). Stress-induced peripheral NPY may play a mechanistic role.

Keywords: Psychological stress, obesity, abdominal adiposity, metabolic syndrome, pre-diabetes

Introduction

Metabolic syndrome has reached epidemic proportions, affecting 20–30% of adults worldwide (Grundy, 2008). The implications for the future burden of chronic disease are grave, as metabolic syndrome doubles the risk of cardiovascular disease and increases the risk for type 2 diabetes by five-fold (Grundy, 2008). Although metabolic syndrome is defined as a cluster of medical conditions, abdominal adiposity and insulin resistance are core features, even among those of normal weight (Abbasi et al., 2004; Voulgari et al., 2011). Chronic psychological stress is an emerging risk factor that prospectively predicts metabolic syndrome (Pyykkonen et al., 2010), abdominal fat (Marniemi et al., 2002), and obesity (Brunner et al., 2007). Furthermore, stress-reduction can improve glycemic control in type 2 diabetics (Ismail et al., 2004). Yet, stress reduction has remained essentially an afterthought within the Western medical model, representing a lost opportunity to improve prevention and management.

Stress may promote overeating and physical inactivity, thereby contributing to metabolic risk (Epel et al., 2004a). In addition, there appears to be another important, but underexplored, physiological pathway. Preclinical studies find that chronic stress activates peripheral mechanisms within adipose tissue, which augment the adverse effects of sugar and fat on visceral tissue accumulation (Kuo et al., 2007). In mice fed a high fat/high sugar diet, those mice exposed to chronic stress developed visceral adiposity and metabolic syndrome at a considerably faster rate than their non-stressed counterparts (Kuo et al., 2007). A key biological mechanism for this synergistic interaction is the peripheral action of neuropeptide Y (NPY), released from sympathetic nerve terminals innervating visceral adipose tissue, which stimulates adipocyte growth and rapid expansion of visceral fat mass in response to stress (Kuo et al., 2007).

Excess intake of fat and sugar leads to oversupply of energy substrate (Picard and Turnbull, 2013), which increases the production of reactive oxygen species (ROS) by mitochondria and causes oxidative stress (Anderson et al., 2009). One rodent study has shown that the combination of chronic stress and a high fat/high sugar diet led to greater oxidative stress markers of fatty liver disease (Fu et al., 2010). Elevated markers of oxidative stress are associated with human obesity (Keaney et al., 2003), diabetes (Keaney et al., 2003), and cardiovascular mortality (Roest et al., 2008). Further, 8-hydroxyguanosine (8-oxoG), a marker of RNA oxidation, prospectively predicts long-term mortality among patients with diabetes (Broedbaek et al., 2011). Chronic psychological stress (e.g., caring for a spouse with dementia) has also been associated with heightened levels of both 8-oxoG and the oxidative byproduct F2-Isoprostanes (Epel et al., 2004b; Aschbacher et al., 2013). In turn, oxidative stress can induce insulin resistance (Ceriello and Motz, 2004; Hoehn et al., 2009), fomenting the development of metabolic syndrome (Bremer et al., 2012). However, synergistic effects of psychological stress and diet on oxidative stress, insulin resistance and adiposity in humans have not yet been assessed.

The current study utilized a dementia caregiving model of chronic stress exposure among post-menopausal women and age-matched, non-caregiving low-stress control women in a case-control design. We hypothesized that the synergistic combination of exposure to chronic stress (defined as being a caregiver) and greater highly palatable food (HPF) consumption would be associated with significantly higher waistline circumference and truncal fat (abdominal adiposity), markers of oxidative damage and insulin resistance. These relationships were examined both cross-sectionally at baseline and prospectively over 1 year. In addition, we hypothesized that the chronic stress group would have higher NPY levels relative to low-stress controls, supporting the peripheral mechanism of sympathetic innervation of adipocytes established in rodent models.

Methods

Participants

Sixty-three non-smoking, post-menopausal women participated in a larger study of chronic stress, metabolism, and cellular aging, as described previously (Epel et al., 2010). Women 50–80 years old were recruited from within the larger San Francisco Bay area using flyers, and community advertisements in newspapers, online and on radio stations. Data for HPF were not available for two women, precluding their inclusion in this investigation. Of the remaining 61 women, the chronic stress group (CS) consisted of 33 women providing care for a spouse or parent with dementia (average years of care = 4.7; range: 0.5–16.5 years). Antidepressant use was permitted among CS women, because excluding it would have biased the sample toward highly resilient individuals, whereas the overarching study sought to understand both the mental and physical effects of chronic stress (e.g., previous publications in this sample have explored the prospective relations between CS and depressive symptoms (Aschbacher et al., 2012)). The low stress group (LS) consisted of 28 women of similar age (within a 4-year range) who were not caregiving, who scored below the national mean on a well-validated measure of chronic stress, the 10-item Perceived Stress Scale (PSS <17) (Cohen and Janicki-Deverts, 2012), and were not taking antidepressants. Table 1 provides group comparisons on eligibility PSS scores, demographics, and other relevant health factors. General study exclusion criteria included chronic medical conditions (cardiovascular disease, cancer, diabetes, and autoimmune diseases), current smoking, or use of medications known to affect stress-responsive biomarkers. The mean age was 62 years (range: 51–79), the median income was $70,000–79,000, and 66% of participants had completed a Bachelor’s or advance degree. The sample was 80% Caucasian, 3% African American, 2% Latina/Hispanic, 10% Asian/Pacific Islander, 2% other, and 3% declined to answer. The study protocol was approved by the Committee for Human Research of the University of California, San Francisco.

Table 1

Group Mean Comparisons on Demographics, Psychosocial Factors, and Metabolic Risk

Chronic Stress Low Stress Statistical Test P-value
Demographic Characteristics
 Age, yearsa 62 (7) 62 (6) t(59) = −0.50 .62
 Non-Caucasian Ethnicityb 6 (19%) 4 (15%) FET .74
 College or Higher Educationb 20 (63%) 19 (70%) χ2(1) = 0.41 .53
 Below Median Incomeb 15 (50%) 14 (54%) χ2(1) = 0.08 .77
Lifestyle/Health Factors
 Perceived Stress Scalea 20.27 (5.71) 5.30 (3.84) t(53.15) = −11.56 .001*
 Highly Palatable Fooda 1.42 (0.75) 1.39 (0.65) t(59) = −0.16 .87
 Low Physical Activityb 22 (69%) 11 (41%) χ2(1) = 4.66 .03*
 NSAIDsb 14 (44%) 4 (15%) FET .02*
Cardiometabolic Risk
 Body Mass Indexa 26.35 (4.93) 26.29 (5.33) t(59) = −0.50 .96
 Triglyceridesa 99.67 (42.36) 102.96 (41.98) t(57) = 0.30 .77
 HDLa 66.87 (19.66) 65.29 (15.92) t(57) = −0.34 .74
 LDLa 117.71 (29.90) 106.61 (35.40) t(57) = −1.30 .20
 Fasting Glucose (mg/dl)a 94.34 (8.82) 97.43 (10.50) t(58) = 1.24 .22
 Systolic Blood Pressurea 133.58 (18.81) 131.25 (19.07) t(56) = −0.47 .64
 Diastolic Blood Pressurea 77.12 (8.70) 75.82 (11.37) t(56) = −0.49 .63
*p ≤.05, critical alpha = .05.
aMean (SD),
bn(%). FET = Fisher’s exact test, 2-sided. College or higher education was categorized as participants holding a Bachelor’s or an advanced educational degree.

Highly Palatable Food (HPF)

Self-reported dietary consumption was assessed at baseline using a modified version of the previously published 29-item version of the Block Food Frequency Questionnaire (FFQ) (Groesz et al., 2012). The 29-item FFQ is an abbreviated version of the 100-item FFQ, which has been validated against nutrient consumption (Block et al., 1990). HPF was assessed using a four-item subscale, inquiring about the frequency of similarly categorized foods – e.g., chocolate, candy bars, cakes, cookies and brownies, etcetera. The response options ask “how often” a food item was consumed and response categories range from “never” (0) to “more than once a day” (5). The scale is deemed appropriate for comparing relative individual differences in HPF consumption, but not absolute caloric intake.

Blood Draw Procedure

Between 0730h and 0800h, participants came into the Clinical Research Center at University of California San Francisco. A venous forearm catheter was inserted, participants rested for 20 minutes, and a fasting blood sample was drawn, which was assayed for peripheral NPY and oxidative stress measures.

Adiposity Measures

Dual-energy X-ray absorptiometry (DXA), a gold-standard method for assessing abdominal fat depots, was used to assess truncal fat (kg). Truncal fat and leg fat have been shown to have opposite relations with fasting and post-load glucose, suggesting differences in the underlying fat tissue(Snijder et al., 2004); therefore, we focused on truncal fat, which is most likely to contribute to or correlate positively with insulin resistance. Subjects underwent whole-body DXA scans on a Lunar Prodigy densitometer. Post-hoc manual analysis of fat in the arms, legs, and trunk was performed using skeletal and soft-tissue landmarks, as first described by Lo et al. (1998). The trunk region was defined by an upper horizontal border at the lower edge of the chin, lateral borders formed by vertical lines which bisected each axilla and which were oriented obliquely to include the waist, hip, buttock and thigh tissue, and a lower border formed by the intersection of oblique lines extending from the level of the superior aspect of the iliac crest and passing through the hip joint. The coefficients of variation for repeated analyses of the same scans are 0.8, and 1.3% for total and trunk fat, respectively. For comparison, we also included a simpler measure that can be easily integrated into large studies: waistline circumference (cm) at the narrowest point, which was measured twice by trained research assistants and averaged.

Neuropeptide Y (NPY) Assay

NPY levels in plasma obtained from peripheral blood were measured by ELISA (EMD Millipore, St. Charles MO). This assay is a Sandwich ELISA based on: 1) capture of NPY in the sample by anti-human NPY IgG and immobilization of the resulting complex to the wells of a microtiter plate coated by a pre-titered amount of anchor antibodies, 2) binding of a second biotinylated antibody to NPY after brief washings, 3) wash away of unbound materials, followed by conjugation of horseradish peroxidase to the immobilized biotinylated antibodies, 4) wash away of free enzyme, and 5) quantification of immobilized antibody-enzyme conjugates by monitoring horseradish peroxidase activities in the presence of the substrate 3,3′,5,5′-tetra-methylbenzidine. The enzyme activity is measured spectrophotometrically at 450 nm, and corrected from the absorbency at 590nm after acidification of formed products. Since the increase in absorbency is directly proportional to the amount of captured NPY in the unknown sample, the concentration of NPY is derived by interpolation from a reference curve generated in the same assay with reference standards of known concentrations of human NPY. The intra-assay and inter-assay coefficients of variation were 5% and 14% respectively.

Oxidative Damage Assays

Blood serum from fasting blood samples were assayed for 8-hydroxyguanosine (8-OxoG) and 8-iso-prostaglandin F2α (IsoP) using the API 4000™ QTRAP® LC/MS/MS System (Kronos Science Laboratory) per a previously published protocol (Aschbacher et al., 2013). Serum samples were combined with an internal standard of either O18-8-hydroxy-2′-deoxy-guanosine (for 8-OxoG) or 8-isoprostaglandin F2α-d4 (for IsoP) and an organic solvent was added to precipitate proteins. 8-OxoG (ng/mL) and IsoP (ng/mL) were quantified using the ion pairs of m/z 300/168 and 353/193 (respectively) investigated under multiple reaction-monitoring (MRM) detection mode. The coefficients of variation for ranged from 4% to 8% (intra) and 8% to 12% (inter) for 8-OxoG and from 4% to 7% (intra) and 6% to 11% (inter) for IsoP.

Insulin Sensitivity

Participants underwent a three-hour oral glucose tolerance test (OGTT), conducted immediately following the fasting blood draw, and the composite insulin sensitivity index was determined from multiple glucose and insulin measures taken over this three-hour period, per the following standard formula: 10,000/√[(fasting glucose x fasting insulin)(mean glucose x mean insulin)] (Matsuda and DeFronzo, 1999). A higher score indicates higher insulin sensitivity.

Health Factors

Participants reported whether they had previously been diagnosed by a physician with any of the following health conditions relevant to metabolic syndrome and cellular aging: hypertension, high blood cholesterol, arthritis/osteoarthritis, or chronic pain. Participants reported use of medications such as non-steroidal anti-inflammatory medications (NSAIDs), statins, antidepressants, and vitamin supplements that might contain anti-oxidants. Using a previously published method (Puterman et al., 2011), we assessed participants’ self-reports of whether they were sedentary or active, based on the Center for Disease Control’s recommendations of 75 minutes of vigorous activity per week.

Data Analysis

Variable distributions were inspected graphically, and a few outliers were winsorized to 2.6 SD to prevent biasing of the regression results1. The primary hypotheses were tested with regression, in which the main effect of HPF was z-scored so the corresponding coefficient would reflect the change in the outcome attributable to a one-standard deviation change in HPF. The interaction term represented the product of the standardized HPF score by Group. Two-tailed p-values and a critical alpha of .05 were used as the criteria for statistical significance. Longitudinal analyses were conducted by first examining predictions of the year 1 outcomes, uncorrected for baseline, and then examining predictions of delta change variables (e.g., outcome at year 1 – outcome at baseline). Age was used as a covariate in all analyses, as was antidepressant use (an exclusion criteria for the LS), and other demographic, health or medical factors that significantly different between the groups were additionally included as covariates in all full model analyses (see results). All final model analyses (i.e., the group by HPF interaction regression models including covariates presented in Table 3) were bootstrapped using residual resampling with 1000 iterations and bias-corrected confidence intervals using Matlab R2013b for Macintosh, which enhances the accuracy of these estimates. The following variables were missing data at baseline (number of participants missing data in parentheses) – waist circumference (4), truncal fat (1), insulin sensitivity (5), 8-OxoG (2), IsoP (2), and NPY (4). The following variables were missing data at year 1 –waist circumference (7), truncal fat (6), insulin sensitivity (7), IsoP (9), and 8-OxoG (20; 8-OxoG was analyzed in a subset of the full sample at Year 1). Analyses have varied ‘n’ dependent on the outcome.

Table 3

Multivariate Associations Between Highly Palatable Food Consumption and Metabolic Outcomes by Group

Outcome
Chronic Stress (CS)
Low Stress (LS)
Group* HPF Interaction
Unstandardized Coefficient 95% CI Unstandardized Coefficient 95% CI Unstandardized Coefficient 95% CI
Basal Waistline Circumference (WC) 5.22** (1.47, 8.73) −0.24 (−5.34, 4.40) −5.50† (−12.03, 0.04)
1 Year WC 4.89* (0.82, 9.23) −1.89 (−7.24, 3.59) −6.86* (−13.74, −0.36)
Change in WC from Baseline to 1 Year 0.40 (−1.03, 1.75) −1.14 (−2.81, 0.70) −1.54 (−3.64, 0.63)
Basal Trunk Fat (TF), kg 2.54** (0.58, 4.21) 0.07 (−2.12, 2.08) −2.44† (−5.07, 0.61)
1 Year TF 2.26* (0.27, 4.61) −0.92 (−3.50, 2.17) −3.12† (−6.54, 0.07)
Change in TF from Baseline to 1 Year −0.25 (−0.81, 0.26) −0.85* (−1.51, −0.21) −0.59 (−1.42, 0.21)
Basal Insulin Sensitivity (IS) −1.21** (−1.95, −0.44) 0.43 (−0.47, 1.41) 1.66** (0.49, 2.88)
1 Year IS −1.16** (−1.90, −0.44) 0.60 (−0.35, 1.64) 1.78** (0.56, 2.98)
Change in IS from Baseline to 1 Year 0.52 (−0.26, 1.23) 0.22 (−0.70, 1.20) −0.30 (−1.58, 0.91)
Basal 8-hydroxyguanosine (8-OxoG) 0.007** (0.004, 0.012) −0.001 (−0.005, 0.004) −0.008** (−0.014, −0.002)
1 Year 8-OxoG 0.003 (−0.009, 0.013) 0.005 (−0.009, 0.019) 0.003 (−0.015, 0.019)
Change in 8-OxoG from Baseline to 1 Year −0.004 (−0.015, 0.009) 0.008 (−0.006, 0.024) 0.011 (−0.007, 0.031)
Basal F2-Isoprostanes (IsoP) 0.010* (0.004, 0.017) 0.005 (−0.002, 0.013) −0.005 (−0.015, 0.006)
1 Year IsoP 0.005† (−0.001, 0.011) 0.001 (−0.007, 0.007) −0.004 (−0.013, 0.004)
Change in IsoP from Baseline to 1 Year −0.005 (−0.012, 0.002) −0.001 (−0.010, 0.007) −0.004 (−0.008, 0.014)

Note:

**p ≤.01,
*p ≤.05,
†p ≤.10.

All models were adjusted for age, antidepressant use, NSAID use and exercise. All statistical estimates provided herein are based on bootstrapping using residual resampling with 1000 iterations and bias-corrected confidence intervals. Change outcomes were calculated using delta scores. The interaction term was formed by calculating the product of caregiver status and the standardized variable for Highly Palatable Food (HPF).

Results

Group Differences on Demographics, Psychological Factors and Cardiometabolic Risk

T-tests or Chi-Squared tests were used to compare the chronic stress group (CS) and low stress group (LS) on demographics, psychological factors, and cardiometabolic risk profiles at the study onset (Table 1). As per the inclusion criteria, the CS group self-reported significantly higher levels of chronic stress on the PSS than did the LS group (Table 1). The mean hours of care provided per day by the CS group was 13.6 hours (range: 1–24, mode: 24), and all LS women, by definition, provided 0 hours of care. The CS group endorsed significantly lower physical activity, and were more likely to take NSAIDS (Table 1). Hence, the final statistical model controlled for the following covariates: age, antidepressant use, NSAID use, and reduced levels of physical activity. The CS and LS groups were compared on the mean levels of the primary metabolic outcomes at baseline and one year later, and significantly differed only on basal 8-OxoG, which, as previously reported (Aschbacher et al., 2013), was significantly higher among the CS group (p<.01 href="/ncbinlm/pmc/articles/pmc4104274/table/t2/" target="table" class="fig-table-link figpopup" rid-figpopup="T2" rid-ob="ob-T2" co-legend-rid="">Table 2). Waistline circumference, truncal fat and insulin sensitivity exhibited high stability over one year (.78< r’s <.97 indicating that little change occurred on these outcomes from baseline through year>

Table 2

Group Mean Differences on Metabolic Risk Indices

Chronic Stress Low Stress Statistical Comparison
Baseline t(df) P-value
 Waistline Circumference (cm) 86.91 (12.78) 86.63 (12.77) −0.09 (55) .93
 Dex Trunk fat (kg) 14.45 (6.09) 14.37 (5.95) −0.05 (58) .96
 Insulin Sensitivity (composite) 4.35 (2.72) 4.49 (2.12) 0.21 (54) .83
 8-oxoG (ng/mL) 0.04 (0.01) 0.02 (0.01) −3.31 (57) <.01>**
 F2-Isoprotanes (ng/mL) 0.05 (0.02) 0.04 (0.02) −0.66 (55) .51
1 Year
 Waistline Circumference (cm) 87.17 (13.50) 84.56 (12.17) −0.74 (52) .46
 Dex Trunk fat (kg) 14.26 (6.38) 13.95 (6.34) −0.18 (53) .86
 Insulin Sensitivity (composite) 4.63 (3.20) 4.64 (2.15) 0.02 (52) .99
 8-oxoG (ng/mL) 0.05 (0.03) 0.03 (0.02) −1.64 (36.67‡) .11
 F2-Isoprotanes (ng/mL) 0.04 (0.02) 0.04 (0.02) −0.67 (50) .51
**p ≤.01, critical alpha = .05.
‡Unequal variance assumed per significant Levine’s test.

Means and standard deviations given in table. FET = Fisher’s exact test, 2-sided.

Synergistic Effects of Chronic Stress and Consumption of Highly Palatable Food

The primary hypothesis that chronic stress increases vulnerability to diet-related metabolic risk was investigated with bias-corrected bootstrapped regression tests of each group-by-HPF interaction term on all five outcomes (waistline circumference, truncal fat, insulin sensitivity, 8-OxoG, and IsoP) at baseline, one-year, and the longitudinal change terms. In addition, we investigated the HPF main effect for each group (chronic versus low stress) to indicate the direction and magnitude of the association, and to investigate the pattern across outcomes, even when the interaction might be of borderline significance in this relatively small sample. Chronological age, NSAIDs, antidepressants, and exercise were included as covariates in all analyses (rationale described above). The group by HPF interaction effects were significant for insulin sensitivity at baseline and Year 1, and for 8-OxoG at baseline (Table 3; Figures 1 and ​and2).2). The stability coefficients for waistline circumference, truncal fat, and insulin sensitivity were high (all r’s > 0.78; Table 3), which indicates that very little change from baseline occurred. Indeed, the full model did not predict significant change over one-year for any outcome. Simple effect analyses were conducted to additionally examine the association of HPF with all outcomes separately in the CS and LS groups. Among the CS group at baseline, HPF was significantly associated with all outcomes in the expected direction (all p’s ≤ .05; Table 3), whereas no significant relationships were present among the LS group. This pattern persisted at Year 1, with similar effect sizes, smaller samples, and decreased significance.

Synergistic Effects of Chronic Stress and Highly Palatable Food on Waistline Circumference

**p ≤.01, *p ≤.05. Note: For the sake of data visualization, p-values given here are derived from t-tests; however, Table 3 provides bootstrapped regression coefficients adjusted for the covariates, which provide the more appropriate, final statistical tests of synergistic (interaction) and main effects.

Synergistic Effects of Chronic Stress and Highly Palatable Food on Oxidative Stress and Insulin Sensitivity

**p ≤.01, *p ≤.05. Note: For the sake of data visualization, p-values given here are derived from t-tests; however, Table 3 provides bootstrapped regression coefficients adjusted for the covariates, which provide the more appropriate, final statistical tests of synergistic (interaction) and main effects.

Investigation of Potential Explanatory Factors

We investigated several non-exclusive pathways that might explain why chronic stress was associated with greater diet-related metabolic risk: 1) stress promotes heightened consumption of HPF, and/or 2) stress alters peripheral stress-arousal physiology, which augments adipocyte responses to HPF. As a group, the CS did not report significantly more HPF consumption than the LS (p = .87; Table 1). However, providing more hours of care per day (an objective index of burden) was associated with significantly greater HPF consumption among the CS (Spearman’s rho: r = .40, p = .03).

We analyzed whether, in humans, as in rodents (Kuo et al., 2007), peripheral NPY might be a mechanism that could explain how chronic stress enhances vulnerability to diet-related adiposity. Consistent with this hypothesis, the CS group had significantly higher levels of basal NPY relative to the LS group (p = .001; Figure 3).

Peripheral Neuropeptide Y (NPY) is Elevated under Chronic Stress

**p ≤.01. Mean + SEM depicted. An independent t-test using a critical alpha of .05 was conducted to obtain the p-value for the group comparison.

Further, we reasoned that if NPY is one of the key stress-arousal mechanisms contributing to the accumulation of fat in the CS group, then the strength of the association between HPF and indices of abdominal adiposity should be greatest among women with high NPY levels. This hypothesis was tested by assessing the HPF by NPY interaction, which was significantly associated with basal waistline circumference (p = .02), truncal fat (p = .01; supplementary figure), and insulin sensitivity (p = .04), but not 8-OxoG (p = .40), or IsoP (p = .12), while controlling for age, physical activity, NSAIDs and antidepressant use (as in previous final models).

Discussion

Traditionally, the maintenance of body weight is thought to be determined by the balance of food intake and expenditure. However, psychological stress has been associated with both weight gain and weight loss (Nyberg et al., 2012), and may therefore alter energy homeostasis. There is a missing link in the current understanding of how and when stress contributes to obesity and metabolic dysfunction. In animal studies, when stress becomes chronic (defined by duration and intensity), sympathetic nerve activity upregulates peripheral expression of NPY, which promotes diet-induced abdominal adiposity and metabolic dysregulation (Kuo et al., 2007). Our data are the first to demonstrate in humans that the synergistic combination of chronic stress along with consuming more high fat/high sugar foods was associated with significantly worse metabolic outcomes and greater waist circumference. This study underscores the importance of chronic stress as a modifier of peripheral metabolic and adipose physiology. If confirmed in interventional studies, these data suggest the possibility that increasing stress-resilience skills could improve the efficacy of interventions to treat metabolic syndrome and obesity, at least for patients with chronically stressful life circumstances.

Our data, collected among a group of post-menopausal women who are overweight on average, demonstrate that more frequent HPF consumption (foods high in sugar and fat) significantly predicted increased waistline, truncal fat, and insulin resistance, but only among the group of women exposed to chronic stress. This association was observed both at the time of initial assessment and one year later (unadjusted for baseline), while statistically controlling for physical activity and other possible confounders. In addition, at baseline, the combination of psychological stress and HPF was significantly associated with higher basal levels of 8-OxoG, a marker of oxidative damage to the genome that predicts long-term mortality among patients with type 2 diabetes (Broedbaek et al., 2011). However, no prediction of changes in metabolic outcomes over 1-year was found, which may be due to the high stability of these measures across the study. One previous rodent study also reported synergistic stress-HPF effects on oxidative stress markers in the context of fatty liver disease (Fu et al., 2010). While that study did not show changes in visceral fat (Fu et al., 2010) in contrast to the study by Kuo et al.(2007), the diet used was considerably lower in fat and the chronic stressor model did not involve social threat, a key factor in human stress hormone reactivity. Although the stress by HPF synergism (i.e., the interaction term) did not reach statistical significance for every outcome in this clinical study, there was a highly consistent pattern whereby HPF consumption significantly predicted all metabolic markers in the chronically stressed group, controlling for age, physical activity, and medical factors (Table 3). Whereas high HPF consumption and a hypercaloric diet may lead to obesity regardless of stress, these data suggest that chronic stress may decrease the threshold of metabolic resilience, so that even low to moderate levels of HPF consumption proffer greater risks.

Animal models suggest that one mechanism of action by which chronic stress exacerbates the adverse metabolic effects of HPF is via upregulation of stress-induced NPY secretion from sympathetic nerve terminals innervating visceral adipose tissue (Kuo et al., 2007). In our data, chronically stressed women had significantly higher levels of plasma NPY, measured in peripheral blood, compared to low stress women. Furthermore, HPF consumption was a significantly better predictor of trunk fat and waistline circumference among those individuals with high levels of plasma NPY. Hence, these results are consistent with the animal models in which chronic stress increased NPY secretion, which in turn, altered adipose tissue sensitivity to HPF. This raises the intriguing question of whether reducing psychological stress or peripheral NPY signaling could potentially help individuals lose fat more successfully, even if diet and activity levels remained the same.

This study used a well-accepted and oft-studied human model of chronic stress, caring for a loved one with dementia, which has been previously associated with weight gain (Fredman and Daly, 1997) and with alterations in a host of cardiometabolic risk markers (Vitaliano et al., 2002; Aschbacher et al., 2008; Mausbach et al., 2012). As a group, chronically stressed women did not report greater HPF consumption than low stress women. However, within the high stress group, the objective indicator of the degree of chronic stress exposure (hours of care provided per day) was significantly related to greater HPF consumption. Future studies should investigate whether other socially relevant models of chronic stress (e.g., low socioeconomic status, unemployment, traumatic stress) moderate the effects of HPF on metabolic risk, as well as extending these findings to males, children, and other age groups.

As a limitation, this study used a case-control design, as it would be unethical to assign humans to chronically stressful conditions. Self-reports (such as our HPF measure) may be subject to social desirability and recall biases, and provide stronger indications of relative than absolute intake. Nonetheless, this study represents a critical step needed to justify considerably more resource-intensive studies involving feeding paradigms, multiple daily dietary assessments, and biomarkers for macronutrient consumption. Future studies might randomly assign participants with a range of stress levels to a feeding study comparing diets with relatively greater or lesser amounts of sugar/fat, and obtain absolute measures of overall caloric, macro- and micronutrient intake. Waistline and truncal fat are proxies but not precise measures of visceral fat; nonetheless, both visceral and subcutaneous fat are associated with heightened metabolic risk (Fox et al., 2007). There are limits to the generalizability of these findings, given our sample was composed of post-menopausal women who were predominantly Caucasian. Menopause is a risk factor for an increase in visceral fat specifically, as well as total body fat (Lovejoy et al., 2008). The high stability of metabolic outcomes over one year in this sample made it difficult to detect prospective changes in metabolic risk. Future studies should reexamine prospective relations in samples with greater longitudinal variability, over longer periods, or in which an intervention has been used to experimentally elicit variability.

People under chronic stress are already known to be at higher risk of metabolic disease, although the mechanisms underlying this link are unclear. This study illuminates an important pathway by which stress-induced changes in peripheral physiology increase the likelihood that consuming HPF will lead to the preferential accumulation of abdominal fat and the worsening of metabolic risk factors. The implication of this work is that chronically stressed individuals are more vulnerable to a high fat/high sugar diet. If confirmed, these data invite the exciting possibility that increasing resilience skills could improve the efficacy of interventions to help individuals lose weight and manage metabolic syndrome, even if diet and activity levels remained the same.

Supplementary Material

Supplementary Data

Supplementary Figure. Combined Effects of Highly Palatable Food (HPF) and Peripheral Neuropeptide Y (NPY) on Trunk Fat

The interaction of the standardized continuously assessed NPY and HPF scores in association with trunk fat at baseline was significant at p=.01, controlling for age, physical activity, NSAID and antidepressant use (as in previous final model analyses). For illustrative purposes only, the depicted scatterplot uses a median split NPY.

Click here to view.(104K, pdf)

Acknowledgments

We thank Aoife O’Donovan and Yali Su for their technical and intellectual assistance.

Funding Sources

NIH/NIA R01 AG030424-01A2, NIH/NHLBI grant K23 HL112955, NIH/NCRR UCSF-CTSI Grant No. UL1 RR024131, NIH/NCATS, UCSF-CTSI UL1 TR000004, the Marchionne Foundation and The Institute for Integrative Health. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.

Footnotes

1One point for IS at baseline and one for IS at 1 year with respective z-scores of 4.3 and 4.7 were winsorized. One point for NPY with a z-score of 4.1 was winsorized. We additionally confirmed that regression analyses of the full model using the raw IS outcomes and all covariates (as specified in Table 3) did not change the pattern of significance. Winsorizing was not necessary to obtain the reported results, merely desirable because conclusions from regression are less subject to potential bias by extreme points.

Contributors

KA, MP, SK, EP, KS and EE wrote the manuscript. KA, EE, PH, and RL contributed to the design of the study. All authors contributed to and have approved the final manuscript.

Conflict of Interest Statements

No authors have conflicts of interest to declare.

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

References

  • Abbasi F, Chu JW, Lamendola C, McLaughlin T, Hayden J, Reaven GM, Reaven PD. Discrimination between obesity and insulin resistance in the relationship with adiponectin. Diabetes. 2004;53:585–590. [PubMed] [Google Scholar]
  • Anderson EJ, Lustig ME, Boyle KE, Woodlief TL, Kane DA, Lin CT, Price JW, 3rd, Kang L, Rabinovitch PS, Szeto HH, Houmard JA, Cortright RN, Wasserman DH, Neufer PD. Mitochondrial H2O2 emission and cellular redox state link excess fat intake to insulin resistance in both rodents and humans. J Clin Invest. 2009;119:573–581. [PMC free article] [PubMed] [Google Scholar]
  • Aschbacher K, Epel E, Wolkowitz OM, Prather A, Puterman E, Dhabhar F. Maintenance of a positive outlook during acute stress protects against pro-inflammatory reactivity and future depressive symptoms. Brain Behav Immun. 2012;26:346–352. [PMC free article] [PubMed] [Google Scholar]
  • Aschbacher K, Mills PJ, von Kanel R, Hong S, Mausbach BT, Roepke SK, Dimsdale JE, Patterson TL, Ziegler MG, Ancoli-Israel S, Grant I. Effects of depressive and anxious symptoms on norepinephrine and platelet P-selectin responses to acute psychological stress among elderly caregivers. Brain Behav Immun. 2008;22:493–502. [PMC free article] [PubMed] [Google Scholar]
  • Aschbacher K, O’Donovan A, Wolkowitz OM, Dhabhar FS, Su Y, Epel E. Good stress, bad stress and oxidative stress: Insights from anticipatory cortisol reactivity. Psychoneuroendocrinology. 2013;38:1698–1708. [PMC free article] [PubMed] [Google Scholar]
  • Block G, Hartman AM, Naughton D. A reduced dietary questionnaire: development and validation. Epidemiology. 1990;1:58–64. [PubMed] [Google Scholar]
  • Bremer AA, Mietus-Snyder M, Lustig RH. Toward a unifying hypothesis of metabolic syndrome. Pediatrics. 2012;129:557–570. [PMC free article] [PubMed] [Google Scholar]
  • Broedbaek K, Siersma V, Henriksen T, Weimann A, Petersen M, Andersen JT, Jimenez-Solem E, Stovgaard ES, Hansen LJ, Henriksen JE, Bonnema SJ, de Olivarius NF, Poulsen HE. Urinary markers of nucleic acid oxidation and long-term mortality of newly diagnosed type 2 diabetic patients. Diabetes Care. 2011;34:2594–2596. [PMC free article] [PubMed] [Google Scholar]
  • Brunner EJ, Chandola T, Marmot MG. Prospective effect of job strain on general and central obesity in the Whitehall II Study. Am J Epidemiol. 2007;165:828–837. [PubMed] [Google Scholar]
  • Ceriello A, Motz E. Is oxidative stress the pathogenic mechanism underlying insulin resistance, diabetes, and cardiovascular disease? The common soil hypothesis revisited. Arterioscler Thromb Vasc Biol. 2004;24:816–823. [PubMed] [Google Scholar]
  • Cohen S, Janicki-Deverts D. Who’s Stressed? Distributions of psychological stress in the United States in probability samples from 1983, 2006, and 2009. J Appl Soc Psychol. 2012;42:1320–1334. [Google Scholar]
  • Epel E, Jimenez S, Brownell K, Stroud L, Stoney C, Niaura R. Are stress eaters at risk for the metabolic syndrome? Ann N Y Acad Sci. 2004a;1032:208–210. [PubMed] [Google Scholar]
  • Epel ES, Blackburn EH, Lin J, Dhabhar FS, Adler NE, Morrow JD, Cawthon RM. Accelerated telomere shortening in response to life stress. Proc Natl Acad Sci U S A. 2004b;101:17312–17315. [PMC free article] [PubMed] [Google Scholar]
  • Epel ES, Lin J, Dhabhar FS, Wolkowitz OM, Puterman E, Karan L, Blackburn EH. Dynamics of telomerase activity in response to acute psychological stress. Brain Behav Immun. 2010;24:531–539. [PMC free article] [PubMed] [Google Scholar]
  • Fox CS, Massaro JM, Hoffmann U, Pou KM, Maurovich-Horvat P, Liu CY, Vasan RS, Murabito JM, Meigs JB, Cupples LA, D’Agostino RB, Sr, O’Donnell CJ. Abdominal visceral and subcutaneous adipose tissue compartments: association with metabolic risk factors in the Framingham Heart Study. Circulation. 2007;116:39–48. [PubMed] [Google Scholar]
  • Fredman L, Daly MP. Weight change: an indicator of caregiver stress. J Aging Health. 1997;9:43–69. [PubMed] [Google Scholar]
  • Fu JH, Sun HS, Wang Y, Zheng WQ, Shi ZY, Wang QJ. The effects of a fat- and sugar-enriched diet and chronic stress on nonalcoholic fatty liver disease in male Wistar rats. Dig Dis Sci. 2010;55:2227–2236. [PubMed] [Google Scholar]
  • Groesz LM, McCoy S, Carl J, Saslow L, Stewart J, Adler N, Laraia B, Epel E. What is eating you? Stress and the drive to eat. Appetite. 2012;58:717–721. [PMC free article] [PubMed] [Google Scholar]
  • Grundy SM. Metabolic syndrome pandemic. Arterioscler Thromb Vasc Biol. 2008;28:629–636. [PubMed] [Google Scholar]
  • Hoehn KL, Salmon AB, Hohnen-Behrens C, Turner N, Hoy AJ, Maghzal GJ, Stocker R, Van Remmen H, Kraegen EW, Cooney GJ, Richardson AR, James DE. Insulin resistance is a cellular antioxidant defense mechanism. Proc Natl Acad Sci U S A. 2009;106:17787–17792. [PMC free article] [PubMed] [Google Scholar]
  • Ismail K, Winkley K, Rabe-Hesketh S. Systematic review and meta-analysis of randomised controlled trials of psychological interventions to improve glycaemic control in patients with type 2 diabetes. Lancet. 2004;363:1589–1597. [PubMed] [Google Scholar]
  • Keaney JF, Jr, Larson MG, Vasan RS, Wilson PW, Lipinska I, Corey D, Massaro JM, Sutherland P, Vita JA, Benjamin EJ. Obesity and systemic oxidative stress: clinical correlates of oxidative stress in the Framingham Study. Arterioscler Thromb Vasc Biol. 2003;23:434–439. [PubMed] [Google Scholar]
  • Kuo LE, Kitlinska JB, Tilan JU, Li L, Baker SB, Johnson MD, Lee EW, Burnett MS, Fricke ST, Kvetnansky R, Herzog H, Zukowska Z. Neuropeptide Y acts directly in the periphery on fat tissue and mediates stress-induced obesity and metabolic syndrome. Nat Med. 2007;13:803–811. [PubMed] [Google Scholar]
  • Lo JC, Mulligan K, Tai VW, Algren H, Schambelan M. “Buffalo hump” in men with HIV-1 infection. Lancet. 1998;351:867–870. [PubMed] [Google Scholar]
  • Lovejoy JC, Champagne CM, de Jonge L, Xie H, Smith SR. Increased visceral fat and decreased energy expenditure during the menopausal transition. Int J Obes (Lond) 2008;32:949–958. [PMC free article] [PubMed] [Google Scholar]
  • Marniemi J, Kronholm E, Aunola S, Toikka T, Mattlar CE, Koskenvuo M, Ronnemaa T. Visceral fat and psychosocial stress in identical twins discordant for obesity. J Intern Med. 2002;251:35–43. [PubMed] [Google Scholar]
  • Matsuda M, DeFronzo RA. Insulin sensitivity indices obtained from oral glucose tolerance testing: comparison with the euglycemic insulin clamp. Diabetes Care. 1999;22:1462–1470. [PubMed] [Google Scholar]
  • Mausbach BT, Chattillion E, Roepke SK, Ziegler MG, Milic M, von Kanel R, Dimsdale JE, Mills PJ, Patterson TL, Allison MA, Ancoli-Israel S, Grant I. A longitudinal analysis of the relations among stress, depressive symptoms, leisure satisfaction, and endothelial function in caregivers. Health Psychol. 2012;31:433–440. [PMC free article] [PubMed] [Google Scholar]
  • Nyberg ST, Heikkila K, Fransson EI, Alfredsson L, De Bacquer D, Bjorner JB, Bonenfant S, Borritz M, Burr H, Casini A, Clays E, Dragano N, Erbel R, Geuskens GA, Goldberg M, Hooftman WE, Houtman IL, Jockel KH, Kittel F, Knutsson A, Koskenvuo M, Leineweber C, Lunau T, Madsen IE, Hanson LL, Marmot MG, Nielsen ML, Nordin M, Oksanen T, Pentti J, Rugulies R, Siegrist J, Suominen S, Vahtera J, Virtanen M, Westerholm P, Westerlund H, Zins M, Ferrie JE, Theorell T, Steptoe A, Hamer M, Singh-Manoux A, Batty GD, Kivimaki M. Job strain in relation to body mass index: pooled analysis of 160 000 adults from 13 cohort studies. J Intern Med. 2012;272:65–73. [PMC free article] [PubMed] [Google Scholar]
  • Picard M, Turnbull DM. Linking the metabolic state and mitochondrial DNA in chronic disease, health, and aging. Diabetes. 2013;62:672–678. [PMC free article] [PubMed] [Google Scholar]
  • Puterman E, O’Donovan A, Adler NE, Tomiyama AJ, Kemeny M, Wolkowitz OM, Epel E. Physical activity moderates effects of stressor-induced rumination on cortisol reactivity. Psychosom Med. 2011;73:604–611. [PMC free article] [PubMed] [Google Scholar]
  • Pyykkonen AJ, Raikkonen K, Tuomi T, Eriksson JG, Groop L, Isomaa B. Stressful life events and the metabolic syndrome: the prevalence, prediction and prevention of diabetes (PPP)-Botnia Study. Diabetes Care. 2010;33:378–384. [PMC free article] [PubMed] [Google Scholar]
  • Roest M, Voorbij HA, Van der Schouw YT, Peeters PH, Teerlink T, Scheffer PG. High levels of urinary F2-isoprostanes predict cardiovascular mortality in postmenopausal women. J Clin Lipidol. 2008;2:298–303. [PubMed] [Google Scholar]
  • Snijder MB, Dekker JM, Visser M, Bouter LM, Stehouwer CD, Yudkin JS, Heine RJ, Nijpels G, Seidell JC, Hoorn S. Trunk fat and leg fat have independent and opposite associations with fasting and postload glucose levels: the Hoorn study. Diabetes Care. 2004;27:372–377. [PubMed] [Google Scholar]
  • Vitaliano PP, Scanlan JM, Zhang J, Savage MV, Hirsch IB, Siegler IC. A path model of chronic stress, the metabolic syndrome, and coronary heart disease. Psychosom Med. 2002;64:418–435. [PubMed] [Google Scholar]
  • Voulgari C, Tentolouris N, Dilaveris P, Tousoulis D, Katsilambros N, Stefanadis C. Increased heart failure risk in normal-weight people with metabolic syndrome compared with metabolically healthy obese individuals. J Am Coll Cardiol. 2011;58:1343–1350. [PubMed] [Google Scholar]

Formats: