Evidence for Disruption of Normal Circadian Cortisol Rhythm in Women with Obesity

Abstract

Hyperactivity of the hypothalamic-pituitary-adrenal (HPA) axis may play a role in the pathogenesis of comorbidities encountered in obesity, including the relative hypogonadotropic hypogonadism that we and others have observed. We sought to examine serum cortisol profiles throughout the day and evening in a sample of normal weight women and women with obesity. In this cross-sectional study, regularly cycling obese (n=12) and normal weight (n=10) women were recruited. Mean serum cortisol was measured by frequent blood sampling for 16 hours (8am-midnight) in the luteal phase of the menstrual cycle. Women with obesity had significantly higher overall cortisol levels when compared to normal weight women (6.2 [4.3, 6.6] vs 4.7 [3.7, 5.5] ug/dl, p=0.04). Over the two-hour post-prandial period, obese women displayed an almost two-fold greater (7.2 [6.5, 8.6] ug/dl) rise in cortisol than normal weight controls (4.4 [3.7, 6.2] ug/dl, p<0.01). In addition, obese women demonstrated a sustained evening cortisol elevation compared to normal weight women, who displayed the typical decline in cortisol (3.2 [2.3, 4] vs 2 [1.5, 3.2] ug/dl, p<0.05). Changes in the HPA axis in the setting of obesity may be related to risks of obesity-associated metabolic comorbidities and reproductive dysfunction often seen in these women.

INTRODUCTION

Over recent decades, the prevalence of obesity in the United States has increased to more than 35% of adults and more than 17% of youth [1], and does not differ between men and women. Obesity increases the risk of developing a number of health conditions, including hypertension, dyslipidemia, and type 2 diabetes, and augments their associated mortality [2].

The relationship between cortisol and obesity has been studied with profound interest due to the similarity of clinical features of hypercortisolism and obesity [3]. Data have been inconsistent regarding an association between obesity and hypothalamic-pituitary-adrenal axis (HPA) activation [4]. Physiologically, cortisol’s role in fat is very complex. Increased cortisol concentrations have been causally linked to fat accumulation and weight gain, since glucocorticoids promote conversion of pre-adipocytes to mature adipocytes [5]. However, cortisol also has other functions such as enhancing lipolysis and triglyceride uptake [6], so in some instances may potentially cause weight loss.

It is also unclear whether hypercortisolism is a cause or a consequence of the obese state. Abnormal cortisol secretion in obesity has been attributed to dysregulation of the HPA as well as peripheral modulation of cortisol action. Studies have shown various degrees of abnormalities, from elevated cortisol output in obesity measured by urinary free cortisol [7–9], to increased cortisol production [7–10], to abnormalities in the HPA axis, particularly in central obesity [11, 12].

The reasons behind the inconsistent findings between various studies on this subject may be related to different methods of assessing cortisol levels, as circulating cortisol can be measured through saliva, blood, or urine; these measures may be different and do not always correlate. Cortisol measurement may be tailored to measure total daily output, response to stimuli, assessing cortisol awakening response, and diurnal cortisol slope.

We sought to investigate this relationship in regularly cycling women of varying body mass index (BMI) by assessing serum cortisol levels throughout the day and early evening and determining its relationship to food intake, as well as correlation with central obesity and blood pressure.

METHODS

Participants

This was a secondary analysis of a previously published study [13]. The primary study was registered at ClinicalTrials.gov (NCT01457703). Regularly cycling women with obesity (n=12) and normal weight women (n=10) were recruited from the community through campus-wide advertisement. Inclusion criteria were: (a) age 18-40 years at the time of study; (b) obesity-range (≥30kg/m²) or normal (18-25kg/m²) BMI; (c) history of regular menses every 25–40 days; (d) normal baseline prolactin, thyroid stimulating hormone (TSH), and blood count. Participants were excluded if they had a history of chronic disease (such as hypercortisolism) or used medication known to affect reproductive hormones, used exogenous sex steroids within the last 3 months, exercised more than 4 hours weekly, or were attempting pregnancy. All participants had a baseline physical exam by study personnel and underwent all blood tests at the Clinical and Translational Research Center (CTRC) of the University of Colorado School of Medicine’s Clinical and Translational Sciences Institute. A comprehensive metabolic panel (CMP) and serum pregnancy test were performed, with the CMP repeated at the end of the study.

The study was approved by Colorado Multiple Institutional Review Board and all participants provided informed consent.

Protocol

A frequent blood sampling study was scheduled 6 to 10 days after a commercially available urinary LH kit indicated that an ovulatory LH surge was underway, thus assuring that all women were studied in the luteal phase of the menstrual cycle. Unstimulated frequent blood sampling was obtained at 10 minutes intervals during the first 12 hours. GnRH (Ferring, IND #74202) 25 ng/kg intravenously was given at 12 hours, while continuing blood sampling. Two hours later, a second dose of GnRH 150 ng/kg was given followed by 2 more hours of frequent blood sampling for a total of 16 hours of frequent blood sampling (8am- midnight). Mealtimes were consistent at the CTRC with lunch being served at noon, and dinner at 5 pm. This secondary analysis was performed on samples at 20 minutes intervals.

Hormone assays

Cortisol was measured by a direct chemiluminescent assay (Advia Centaur; Siemens). The cortisol intra-assay coefficient of variation (CV) ranged from 2.98-3.82% and the inter-assay CV ranged from 1.86-5.45%.

Insulin was measured using the same Siemens platform with reagents specific for insulin. The insulin inter-assay and intra-assay coefficients of variation were 6.2% and 1.6% respectively.

Corticosteroid-binding globulin (CBG) was measured from the first serum sample intra- and inter-assay CVs were <10% and <12%, respectively.

Statistical considerations

No sample size calculation was performed for this study as it was a secondary analysis of data collected to study characteristics of LH and progesterone secretion in women with obesity [13]. Mean cortisol levels were compared between women with obesity and normal weight women for the entire study and for selected time intervals using t tests or Mann-Whitney tests as appropriate, on a SAS software (v9.2 x64) platform. Time intervals that were compared included post lunch (noon-2 pm), post dinner (5-7 pm), and evening (8 pm-midnight). Results of statistical analysis are reported as mean ± standard deviation if a t test was used and as median [interquartile range] if a Mann-Whitney test was used. Pearson correlations for morphometric data and cortisol were computed. P<0.05 was considered statistically significant.

RESULTS

Main results

Demographic and hormonal data for the study participants are shown in Table 1. By design, the obesity group had a significantly greater BMI and, as expected, a significantly greater waist and hip circumference than the normal weight group. The women in the obesity group were significantly older than the normal weight women. The groups did not differ in terms of race or ethnicity, with the majority of participants being Caucasian and non-Hispanic.

Table 1.

Demographic and hormonal characteristics of the two groups¹

	Obese women N= 12	Normal weight N=10	p
Age (yrs)	32 [28.5, 35]	28 [26, 29]	0.02
BMI (kg/m2)	34.3 [31.8, 37.5]	22.4 [21, 22.7]
Insulin (mU/L)	18.5 [13, 23.1]	11.2 [9.6, 14.3]	0.02
Cortisol (ug/dl)	6.2 [4.3, 6.6]	4.7 [3.7, 5.5]	0.04
Post prandial cortisol (ug/dl) [noon–2 PM]	7.2 [6.5, 8.6]	4.4 [3.7, 6.2]	<0.01
Evening cortisol (ug/dl) [8 PM–midnight]	3.2 [2.3, 4]	2 [1.5, 3.2]	0.049
Cortisol binding globulin (mg/dl)	2.7 ± 0.4	2.5 ± 0.3	0.16

¹Data are reported as median [interquartile range] if a Mann-Whitney test was used or mean ± standard deviation if a t test was used.

BMI, body mass index

Participants with obesity had a higher mean cortisol over the entire study period when compared to normal weight women, (6.2 [4.3, 6.6] vs 4.7 [3.7, 5.5] ug/dl respectively, p=0.04) (Fig 1). Insulin was also significantly higher (18.5 [13, 23.1] vs 11.2 [9.6, 14.3] mU/L, p=0.02). When cortisol levels were analyzed by different time intervals, post-lunch (noon- 2 pm) mean cortisol was higher in women with obesity vs normal weight women, (7.2 [6.5, 8.6] vs (4.4 [3.7, 6.2] ug/dl, p<0.01). Evening cortisol (8 pm- midnight) was also maintained at higher levels in participants with obesity when compared to normal weight women, who demonstrated a physiologic drop in cortisol in the later afternoon and early evening, (3.2 [2.3, 4] vs 2 [1.5, 3.2] ug/dl, p=0.05). Post-dinner cortisol (5 pm- 7 pm) did not significantly differ between the two groups.

Figure 1.

Mean CBG levels measured from the first serum sample were not statistically different between women with obesity and normal weight women (2.7 ± 0.4 vs 2.5 ± 0.3 mg/dl).

Correlations with morphometrics

Mean cortisol was moderately correlated with BMI (r= 0.44, p=0.05), and waist-hip ratio (r= 0.48, p=0.03) (Figure 2), with an accentuation of the relationship when evening cortisol was correlated with waist-hip ratio (r= 0.57, p<0.01). Mean postprandial cortisol was also strongly correlated with BMI (r=0.59, p<0.01).

Figure 2.

When correlating cortisol with risk factors for adverse cardio-metabolic function, mean cortisol was strongly correlated with both systolic and diastolic blood pressures (r=0.55, p<0.01 and r=0.44, p=0.04, respectively), as was higher evening cortisol (r=0.55, p<0.01 and r=0.45, p=0.04, respectively) (Figure 2).

DISCUSSION

In this cross-sectional study in patients with no history of Cushing’s syndrome, we found that women with obesity had a higher mean serum cortisol and a greater cortisol response to food intake when compared to normal weight women. We also observed that evening serum cortisol remained elevated in the obesity group relative to the normal weight group, and did not demonstrate the physiologic drop that is characteristic of the normal circadian rhythm. These altered dynamics of the HPA axis shed more light on the association of cortisol and obesity, as the cortisol sampling and measurements in this study were more extensive than what has been reported in most previous studies, and mark important relations with stimuli such as food intake.

Previous studies have examined the association of obesity and cortisol with a variety of measurements. Measuring the total daily output had inconsistent results, possibly due to different designs and measurements for cortisol. Four studies, which measured urinary free cortisol metabolites [14–17], showed higher cortisol output in relation to generalized and/or abdominal obesity. In contrast, other studies found no evidence of high cortisol output in relation to obesity [18–21]. One of these studies only measured pooled serum cortisol from 2 morning samples [20]; another measured six samples of salivary cortisol over three days [21].

Our findings show higher mean serum cortisol levels in women with obesity by frequent blood measurements, every 20 minutes, for 16 hours (48 measurements per individual) that may give a better picture for the daily cortisol output.

The normal pattern of daily cortisol concentrations follows a negative slope [22], as cortisol concentrations increase sharply immediately upon waking and then show an attenuated decline throughout the day [23]. We examined cortisol diurnal pattern as a measure of normal HPA axis function, as it was previously found that flat slopes, failing to reach a low level by evening, are indicative of HPA axis dysregulation, and are associated with negative physical and mental health outcomes [24]. It is unclear from previous studies whether a consistent relationship exists between cortisol slope and obesity, but some studies have shown that a flatter, less sharply declining slope was correlated to BMI [21]. In agreement with these findings, our study’s multiple cortisol measurements showed that women with obesity had failure of decline in evening cortisol compared to normal weight women, who displayed the typical decline in cortisol. This pattern may be indicative of abnormal HPA activation, and also may suggest that an increase in nadir cortisol levels contributes to obesity. In agreement with our findings, bedtime salivary cortisol was found to increase with BMI in a cross-sectional study [25].

Finally, measuring cortisol in response to food intake represents an assessment of cortisol reactivity and HPA axis response to physiologic stimuli, as it was noted that food consumption could lead to cortisol secretion [26]. Our findings are in agreement with studies showing higher cortisol levels after lunch in individuals with obesity [12], which correlated with abdominal obesity in women [18].

There are several strengths of this study. They include the frequent sampling design, synchronization of all participants to the same phase of the menstrual cycle, and robust assays. However, there are weaknesses that limit our ability to make firmer conclusions. The secondary nature of the analysis precluded overnight sampling, which, in retrospect, may have provided more information. The cross sectional design of the study allows us to determine associations only, with minimal ability to make causal inference. Age differences between the two groups may have had an effect on the cortisol levels, as it has been shown in a study that included 54 men and women of different age groups, where age was positively correlated with 24-hour plasma free cortisol levels and increased cortisol production rate, independent of body size [27]. However, differences in cortisol with age in that study were only seen when comparing the oldest subjects (age 54-70) with the youngest (age 19-41). It seems unlikely that an age-related difference in our smaller study, when all participants spanned a range of 22-39 years, could wholly account for our findings; furthermore, the BMI range was very wide in that study (19-64 kg/m²), with a mean of 32 kg/m². This may contribute to their findings of non-association between free cortisol and body composition.

CONCLUSIONS

We have shown that women with obesity have a higher mean serum cortisol than normal weight controls, along with increased cortisol response to food intake and a dysregulated HPA axis, resulting in sustained flat slope of cortisol concentrations. These abnormalities were correlated with measures of abdominal obesity and blood pressure. A future study should be designed specifically for the purpose of investigating this association of cortisol with obesity to have the proper sample size calculation, and potentially correlate these levels with central adiposity measurements.