Stress-related Development of Obesity and Cortisol in Women

Authors

  • Valentina Vicennati,

    1. Division of Endocrinology, Department of Internal Medicine, S. Orsola-Malpighi Hospital, University Alma Mater Studiorum of Bologna, Bologna, Italy
    Search for more papers by this author
  • Francesca Pasqui,

    1. Division of Endocrinology, Department of Internal Medicine, S. Orsola-Malpighi Hospital, University Alma Mater Studiorum of Bologna, Bologna, Italy
    Search for more papers by this author
  • Carla Cavazza,

    1. Division of Endocrinology, Department of Internal Medicine, S. Orsola-Malpighi Hospital, University Alma Mater Studiorum of Bologna, Bologna, Italy
    Search for more papers by this author
  • Uberto Pagotto,

    1. Division of Endocrinology, Department of Internal Medicine, S. Orsola-Malpighi Hospital, University Alma Mater Studiorum of Bologna, Bologna, Italy
    Search for more papers by this author
  • Renato Pasquali

    Corresponding author
    1. Division of Endocrinology, Department of Internal Medicine, S. Orsola-Malpighi Hospital, University Alma Mater Studiorum of Bologna, Bologna, Italy
    Search for more papers by this author

(renato.pasquali@unibo.it)

Abstract

Chronic exposure to environmental stress may play a role in the development of obesity, through hyperactivation of the hypothalamic–pituitary–adrenocortical (HPA) axis. This study investigated the dynamics of weight gain and the activity of the HPA axis in women who developed weight gain after a stressful event. This is a case–control retrospective study. Two groups of age-matched premenopausal women were selected. One (n = 14) included women characterized by a rapid weight gain following a stressful event, defined as the “stress-related obesity” (SRO) group, and the other (n = 21) women with nonstress-related development of obesity, defined as the “nonstress-related obesity” (NSRO) group. Twenty-one healthy premenopausal women served as normal-weight controls. Baseline hormonal and metabolic parameters, and 24-h urinary free cortisol (UFC/24 h) excretion rate (as a measure of HPA-axis activity) were measured in all women. Anthropometry, diet, and physical activity were similar in both obese groups. Both obese groups showed similar metabolic and hormonal profiles, but the SRO group had UFC/24 h values (41.1 ± 14.3 µg) significantly higher (P < 0.001) with respect to the NSRO (26.6 ± 17.6 µg) or the normal-weight control groups (21.1 ± 9.8 µg). Moreover, time (years) to achieve maximum Δweight gain (kg) and the Δweight gain/time ratio were significantly shorter (P < 0.001) and higher (P < 0.001) in the SRO group with respect to the NSRO group, respectively. In the SRO group, there was a tendency to a significant correlation between UFC/24 h and the Δweight gain/time ratio. These findings support the concept that SRO has distinct pathophysiological mechanisms, including hyperactivity of the HPA axis.

Introduction

The similarities with syndromes of endogenous or exogenous hypercortisolism provide the basis for a hypothetical role of glucocorticoids on the development of human obesity, particularly the abdominal phenotype, and associated metabolic and cardiovascular alterations (1,2). In obese patients, basal adrenocorticotropin hormone, and cortisol concentrations are usually normal or slightly reduced, and some studies reported altered salivary cortisol concentrations throughout the daytime (3). By contrast, dynamic studies following stimulation with different neuropeptides, psychological stress challenges, and mixed meal tests supported the presence of hyperresponsiveness of the hypothalamic–pituitary–adrenocortical (HPA) axis, as did suppression studies using low-dose dexamethasone and alprazolam, particularly in females (3). Notably, glucocorticoid receptor density in the visceral adipose tissue has been found to be higher than in the peripheral subcutaneous fat (4), which emphasizes the potential role of cortisol excess availability in the pathophysiology of abdominal obesity.

Stress adaptation requires a coordinated series of adaptive responses, including an increased HPA axis and sympathetic nervous system activation to maintain homeostasis and protect against chronic diseases (5,6). A hypothetical factor for chronic hyperactivation of the HPA axis in obesity, particularly the abdominal phenotype, has been related to the individual inability to cope with long-term environmental adverse stressful events throughout the life span (6,7,8). This sequence of events has been clearly demonstrated in primates exposed to long-term physical and psychological stress, including enlarged visceral fat deposition, insulin resistance and hyperinsulinemia, impaired glucose tolerance, altered lipid profiles, and a very high incidence of coronary artery disease, together with adrenal hypertrophy and enhanced cortisol response to adrenocorticotropin hormone stimulation (9). Epidemiological and clinical studies performed in humans have in turn documented that abdominal obesity and its metabolic comorbidities are significantly correlated with stress-related conditions such as adverse life events, psychological disturbances, and psychosocial problems (10,11,12,13,14,15,16). Overall, these findings suggest that obesity may be the consequence of a chronic maladaptation to environmental stress factors in susceptible individuals (17). Moreover, other studies provided evidence for a significant positive association between cortisol levels and the salient features of the metabolic syndrome and insulin resistance (18,19).

Chronic stress can also lead to increased food intake in both animals and human beings, particularly women (20), and, again in women, may predict relapse and overeating after weight loss achieved by hypocaloric dieting (21). In unselected obese subjects, normal or slightly increased values of urinary free cortisol (UFC) excretion have been reported (3). Interestingly, some studies have also reported that UFC excretion rates during the night time may better distinguish obese individuals with different phenotypes of obesity (22,23).

With this background, we planned this retrospective study to investigate: (i) the dynamics of weight gain in women who developed the weight gain after a well-defined stressful event compared to an age-matched control group of women with onset and progressive development of obesity, (ii) their anthropometric, metabolic and hormonal status, including insulin, 24-h UFC (UFC/24 h) and sex hormones.

Methods and Procedures

Subjects

All women included in the study were selected from our database, including all patients referred to the division of Endocrinology of the Department of Clinical Medicine, University Alma Mater Studiorum of Bologna, seeking treatment for excess weight or obesity and related comorbidities. The criteria for selection were adult premenopausal age, regular menses, and a complete personal history, including weight fluctuation, weight before onset of excess weight gain, age at onset of excess weight gain, any factor associated with onset, and progression of weight gain, minimum (if different from preweight gain value) and maximum weight achieved, a complete metabolic and hormonal work-up, including routine biochemistry, lipids, fasting glucose–stimulated glucose and insulin, sex hormones, and UFC/24 h measurements.

Two groups of women were selected, one (n = 14) characterized by a rapid development of excess body weight following a well-defined stressful event, defined as the “stress-related obesity” (SRO) group, and the other (n = 21) including women with onset of weight gain during infancy or adolescence (n = 6), during pregnancy and/or lactation (n = 12), or after stopping smoking (n = 1) or practicing sport (n = 2), thereby with a progressively increased weight gain over time. This latter group was defined as the “nonstress-related obesity” (NSRO) group. Specific stress events, which were classified as clearly associated with the onset of weight gain in the SRO group, were family bereavement (such as the death of a husband, son, or parent) (n = 6), miscarriage (n = 1), job change or change of address (n = 4), or major surgery (n = 3). In this group, only four women had a pregnancy after the stressful event. Therefore, when classifying the subjects in the two groups, only clinical criteria were used and we were blinded with respect to any hormonal and metabolic parameters. Groups were matched for age. Information on weight history and characterization of patients with regard to stressful events and pregnancy-related weight gain was carefully collected by one of the authors (V.V.) and then reviewed by R.P. All women had stable body weight, with fluctuations <2–3 kg during the last year.

No women were pregnant or lactating, were dieting, or had been taking any medication in the past 6 months before clinical examination and biochemical tests. Based on clinical examination, routine biochemistry, and dynamic tests, when needed, none of them had diabetes, thyroid dysfunction, hyperandrogenism, Cushing-related hypercortisolism, or other endocrine or metabolic diseases, or had significant psychiatric disorders (based on clinical evaluation), or major cardiovascular, renal, hepatic, and systemic diseases.

A total of 21 healthy premenopausal normal-weight ovulatory women served as a reference group for basal hormone and metabolic parameters, and UFC/24 h.

All subjects gave informed consent to participate in the study, which was approved by the local Ethics Committee.

Anthropometry, blood pressure, diet history, and physical activity

Body height was measured without shoes to the nearest 0.5 cm and body weight was measured to the nearest 0.1 kg and without clothes. Retrospective parameters were reported by patients. The waist circumference was also measured, with the subjects standing, using a 1-cm-wide metal measuring tape, as the minimum value between the iliac crest and the lateral costal margin. BMI was calculated as weight in kilograms divided by height in meters squared. Based on information taken from weight history, total weight gain (maximum weight achieved minus weight before onset of weight gain (Δweight gain)), and the ratio between Δweight gain with the time needed to achieve it were calculated and expressed as “Δweight gain (kg)/time (years) ratio”. Systolic and diastolic blood pressure (BP) was measured while the women had been lying down for at least 3 min under basal condition. The presence of the metabolic syndrome was assessed by using the definition of the International Diabetes Federation (24).

Habitual energy intake was calculated by means of the diet history method and a 3-day recall questionnaire and physical activity was investigated by Baeke's questionnaire, as previously described (25).

Metabolic and hormonal assessment

Blood samples for routine analysis, metabolic parameters, and hormones and sex hormone-binding globulin (SHBG) were drawn in all women from 8.00 to 9.00 am after overnight fasting, in the follicular phase. All women had been invited to follow a 3-day diet containing at least 250–300 g carbohydrate before testing. An oral glucose tolerance test (75 g Curvosio; Sclavo, Cinisello Balsamo, Italy) was also performed, taking blood samples after 30, 60, 90, 120, and 180 min for glucose and after 60, 120, and 180 min for insulin determinations. The day before, all patients were carefully instructed to collect 24-h urine for the measurement of free cortisol (UFC/24 h), as an integrated measurement of cortisol production rates.

Metabolic and hormonal assessment was performed in all patients in the context of a protocol used in our institution for obesity and related disorders. Metabolic parameters, hormones, and UFC/24 h were therefore not measured relative to weight gain, but on admission to the outpatient unit.

Blood and urine samples for hormone assay were immediately chilled on ice and centrifuged, and serum aliquots were collected and frozen at −20 °C until assayed.

Assays

Plasma glucose levels were determined by the glucose oxidase technique immediately after blood drawing. Hormones and lipids, including total cholesterol, high-density lipoprotein cholesterol, and triglycerides, SHBG, and UFC/24 h were measured as previously described (26). The free testosterone index was calculated as the ratio between total testosterone and SHBG, according to Vermeulen et al. (27). To investigate insulin sensitivity in basal condition, the homeostasis model assessment for insulin resistance was calculated (28) and the composite insulin sensitivity index was determined from glucose and insulin values during the oral glucose tolerance test (29). The intra-assay coefficients of variation in our laboratory were 3.0% for insulin, 7.0% for testosterone, 6.0% for androstenedione, and 6.5% for SHBG.

Statistics

All results are reported as the mean ± s.d. Glucose and insulin response to the oral glucose tolerance test was expressed as area under the curve, which was calculated by the trapezoidal method. Normal distribution and homoscedasticity of continuous variables were tested by means of the Kolmogorov–Smirnov and the Levene tests. Variables that did not fulfill these tests were log-transformed before analysis. Statistical differences between the two groups was evaluated by ANOVA. Simple correlation analyses were analyzed using log-transformed variables, if needed. Statistical evaluations were performed by running the SPSS/PC+ statistical package version 8.0 (SPSS, Chicago, IL). P values <0.05 were regarded as statistically significant.

Results

Anthropometry, blood pressure, diet history, and physical activity

There were no significant differences in anthropometric parameters between the two groups. In addition, no differences were present in body weight at 20 years, age of onset of weight gain and body weight before onset of weight gain, Δweight gain, and maximal body weight achieved. By contrast, time to achieve Δweight gain and the Δweight gain (kg)/time (years) ratio were significantly shorter (P < 0.001) and higher (P < 0.001) in the SRO group with respect to the NSRO group, respectively (Table 1). There were no significant differences in Baeke's physical activity scores (work: 2.86 ± 0.86 vs. 2.42 ± 0.83 leisure time 2.50 ± 0.78 vs. 2.42 ± 0.62, sport 1.73 ± 0.58 vs. 2.01 ± 0.60, respectively) and daily energy intake (kilocalories: 2463 ± 425 vs. 2335 ± 391, carbohydrates 303 ± 74 g vs. 289 ± 63 g, proteins 87 ± 16 g vs. 80 ± 13 g, lipids 100 ± 19 vs. 93 ± 20, respectively) between the SRO group and the NSRO group.

Table 1.  General characteristics, anthropometry, and weight history parameters (mean ± s.d.) of women with stress-related obesity and nonstress-related obesity
inline image

Metabolic and hormonal parameters

Women of the SRO and NSRO groups had similar metabolic and hormonal parameters, but both had higher fasting and glucose-stimulated (as area under the curve) insulin levels, free testosterone index values, and systolic and diastolic blood pressure, but significantly lower SHBG values in comparison to values of the normal-weight reference group (Table 2).

Table 2.  Metabolic parameters, sex hormones, and blood pressure (mean ± s.d.) of women with stress-related obesity and nonstress-related obesity
inline image

UFC/24 h

In the SRO group, values of UFC/24 h (41.1 ± 14.3 µg) were significantly higher (P < 0.001) with respect to those observed in the NSRO group (26.6 ± 17.6 µg) or the reference control group (21.1 ± 9.8 µg), without any significant difference between the latter two (Figure 1).

Figure 1.

Daily 24-h urinary free cortisol (UFC/24 h) concentrations (mean ± s.d.) in women with “stress-related obesity” (black bar) and “nonstress-related obesity” (white bar). Values taken from premenopausal normal-weight women are also reported for reference (gray bar).

There were no significant relationships between UFC/24 h values and anthropometric, metabolic, or hormonal parameters. However, in the SRO group a tendency to a significant correlation with the Δweight gain (kg)/time (years) ratio (r = 0.303; P = 0.064) was found.

Discussion

This is the first study showing that obese women with onset of weight gain after an important stressful event were characterized by rapid onset of obesity and overactivated adrenocortical function, as documented by higher-than-normal UFC/24 h values, which persisted over time even after the exposure to a well-defined stressful event following which weight gain rapidly started to develop. Metabolic and hormonal assessment was performed in all patients in the context of a protocol used in our institution for obesity and related disorders; therefore, UFC/24 h was not measured relative to weight gain, but on admission to the outpatient unit.

Previous studies have shown a significant association between hyperactivity of the HPA axis and abdominal fat distribution and/or the metabolic syndrome (3,12,23). Although these findings need to be confirmed in larger studies, they nonetheless raise several general and pathophysiological considerations on the potential role of chronic stress in the development of obesity in women.

Many studies have been published supporting the concept that long-term maladaptation to chronic environmental stressors may have an important impact on the development of human obesity, particularly in women, who are more responsive to stress exposure (5), the HPA axis being subject to the influence of sex hormones (23).

From the pathophysiological point of view, mechanisms by which chronic excess cortisol may favor obesity may be direct or indirect. Glucocorticoids do in fact have a direct impact on adipose tissue metabolism, energy partitioning and insulin action (30). Specifically, they downregulate hormone sensitive lipase and increase lipolysis, favor preadipocyte differentiation, stimulate substrates to gluconeogenesis and free fatty acids to central fat (30), and inhibit glucose uptake by peripheral tissues (30). In addition, glucocorticoids suppress thermogenesis and induce leptin resistance (31), thereby opposing the metabolic action of leptin. Although cortisol is lipolytic, chronic hypercortisolemia nevertheless results in increased fat mass, particularly in the visceral depots (31).

Glucocorticoids are also important regulators of feeding behavior and choice. High cortisol levels after stress exposure can predict stress-induced eating (32). Other studies have shown that a hyperactive HPA-axis can be detected in women with binge-eating disorders (33). However, the relationship between the HPA axis and food intake represents a still undefined and complex network potentially involved in the pathophysiology of stress-dependent obesity. Based on experimental animal studies, Dallman et al. (34) provided solid evidence supporting the concept that, whereas glucocorticoids acutely and directly inhibit further activity in the HPA axis, chronic actions of glucocorticoids on the brain may conversely be excitatory, by increasing the expression of CRF mRNA in the brain, which in turn enables recruitment of a chronic stress–response network, and by stimulating, together with insulin, the salience of pleasurable or compulsive activities (so-called “comfort foods”), therefore increasing body fat, particularly in the visceral depots. A still undefined molecular signal associated with visceral fat depots appears, as with eating “comfort foods,” to reduce the influence of the chronic stress network on behavior, autonomic, and neuroendocrine outflow. Recent evidence suggests that, among other factors, neuropeptide-Y seems to play a role in this context (35).

As with animals, it has been postulated that reactive food intake may also occur in humans, and that this may ultimately represent an attempt to reduce the activity in the chronic stress–response network. Rapid onset of excess body fat and, ultimately, obesity may therefore represent the price paid to compensate for chronic stress exposure. Whether this behavior is transient or may conversely continue in the long term still represents an unanswered question. We speculate that, during the dynamic phase of rapid weight gain following a stressful event, excess eating may develop, whereas when obesity is stabilized over time, adaptative changes in eating behavior may probably disappear, although some specific preferences for foods, specifically those containing carbohydrates, may persist for a long time (unpublished data). This could explain why we did not find any significant differences in energy intake and macronutrient composition between our study groups. Accordingly, we did not find any significant differences in fat distribution, metabolic parameters and hormones between the two obese groups, although a tendency to higher androstenedione levels was present in the SRO group. Compared to normal weight controls, however, both obese groups had higher glucose-stimulated insulin responses, higher testosterone (free and total) and lower SHBG concentrations, and higher arterial blood pressure values. The lack of any significant difference in fat distribution and metabolism in the SRO group is not surprising, given the small number of subjects investigated and similar BMI values to the nonstressed group. However, although direct evidence is still lacking in humans, there are epidemiological data and clinical studies providing evidence for a significant positive association between blood cortisol levels, either basal or stimulated by specific neuropeptide standardized stressful challenges, with alterations of the glucose–insulin system and altered lipid profiles (23). It should be noted that all these studies included subjects presenting with different patterns of fat distribution and with or without metabolic abnormalities, but did not categorize subjects for cause of weight gain including a well-defined stressful event. On the other hand, the lack of association between anthropometry and metabolic parameters with UFC/24 h somewhat weakens our conclusions, implying the need for further careful well characterized prospective studies.

In conclusion, we have demonstrated that obese women with SRO are characterized by a rapid onset of obesity and overactivated adrenocortical function, as documented by higher-than-normal UFC/24 h values. These findings support the concept that SRO has distinct pathophysiological mechanisms and potentially implies specific preventive or therapeutic strategies.

Acknowledgments

We thank Antonio Maria Morselli-Labate for his assistance in the statistical evaluation and Mrs Susan West for reviewing the English language.

Ancillary