Perceived Stress, Hair Cortisol, and Hair Cortisone in Relation to Appetite-Regulating Hormones in Patients with Obesity

The increasing prevalence of obesity is a worldwide growing health problem [1]. In order to develop personalized prevention and treatment strategies, research progresses to elucidate the highly heterogeneous, interrelated biological and psychological correlates of appetite regulation in the context of overweight and obesity [2, 3].

A known risk factor in obesity’s pathogenesis is stress [3]. Both chronic psychological stress (e.g., self-reported in questionnaires) and biological stress (e.g., reflected by increased hair glucocorticoid levels) predict future weight gain, especially in people already having a high body mass index (BMI) [4‒6]. Evidence from laboratory studies in humans revealed that acute stress can induce a preference for sweet and calorie-dense “comfort” foods along with changes in levels of circulating hormonal appetite regulators [3, 7, 8], suggesting that the relationship between stress and overeating is likely to be partially mediated by altered signaling of appetite-regulating hormones [3, 9].

Indeed, exposure to stress or stress hormones can directly increase signaling of hunger-inducing hormones such as ghrelin and neuropeptide-y, and decrease signaling of satiety-inducing hormones such as leptin and peptide tyrosine-tyrosine (PYY) [3, 7, 9]. However, most available evidence stems from animal studies, while especially the relation between long-term stress and the hormonal appetite system in humans remains to be elucidated. Therefore, we investigated the associations between measures of both long-term biological (hair glucocorticoid levels) and psychological stress (Perceived Stress Scale [PSS]) with fasting serum levels of appetite-regulating hormones in adults with obesity. We hypothesized that higher chronic biological and psychological stress is associated with lower levels of satiety hormones PYY, cholecystokinin (CCK), leptin, insulin and pancreatic polypeptide (PP) and lower levels of hormonal regulators of insulin signaling (gastric-inhibitory polypeptide [GIP] and adiponectin). We also expected a positive association between stress measures and levels of hunger-inducing agouti-related peptide (AgRP).

Between October 2013 and October 2019, adults with lifestyle-induced obesity (and no other underlying endocrine or genetic diseases) were recruited via the Obesity Center CGG, an academic center for patient-tailored obesity treatment at the Erasmus MC Rotterdam, The Netherlands. Inclusion criteria were BMI ≥30 kg/m², age ≥18 years and presence of ≥1 obesity-related comorbidity (e.g., hypertension, type 2 diabetes, dyslipidemia, nonalcoholic fatty liver disease, or obstructive sleep apnea). Exclusion criteria were inability to speak Dutch, intellectual disability (IQ <80/special education), current wish for pregnancy, psychopathology that requires acute treatment, and use of systemic glucocorticoid medication in the past 6 months. Participants provided written informed consent. The study was approved by the Local Ethics Committee (MEC2012–257). The findings presented here result from secondary analyses of data collected to study factors affecting weight management in patients with obesity [10].

Psychological stress was measured using the 14-item self-report PSS questionnaire (Dutch version, range 0–56, higher scores indicate more severe stress) [11]. Items assessed stress levels of the past 1 year. Scalp hair cortisol (“HairF”) and cortisone (“HairE”) levels were measured as previously described to assess biological stress. First, hair was cut at the posterior vertex. Subsequently, HairF and HairE were measured in the first 3 cm closest to the scalp using liquid chromatography-tandem mass spectrometry, reflecting glucocorticoid exposure of the past 3 months [12].

For analyses of appetite-regulating hormones, blood was drawn from patients after an overnight fast. Serum was stored after centrifugation at −20 or −80°C for a maximum duration of 7 years. Measurements of appetite-regulating hormones were performed using automated or manual immunochemistry methods as described previously [13], according to the manufacturers specifications in an ISO15189:2012 accredited laboratory.

Anthropometric measurements were performed by outpatient clinic assistants. Height was measured via a stadiometer. Weight in kilograms (kg) was assessed using a calibrated scale (patients wearing clothes but no shoes). BMI was calculated as weight in kg divided by height in meters squared (kg/m²) [10, 14]. Waist circumference was assessed unclothed, halfway between the superior anterior iliac crest and the lowest rib (the average of two consecutive measurements was noted).

Levels of hair glucocorticoids, adiponectin, insulin, PP, GIP, CCK, and AgRP were logarithmic-10-transformed to obtain normal distribution. We used linear regressions to examine the association between chronic stress measures and appetite-regulating hormones (crude, adjusted for age and sex [Model 1], and with additional adjustment for BMI [Model 2]). Sensitivity analyses were performed (1) excluding patients using diabetes medication (metformin, n = 5) and (2) including women only. We refrained from men-only sensitivity analyses due to the small sample size (n = 21). When both predictor (hair glucocorticoids) and outcome (appetite-regulating hormones) were log-transformed, the Beta coefficient signifies the percentage rise in the dependent variable for each 1% increase in the independent variable [15], Our a-priori power calculation using gPower was based on previous findings from human laboratory studies [7, 9], indicating a necessary sample size of 44 participants to detect moderate associations of r = 0.4, with 1 – β = 0.80 and α = 0.05. Analyses were performed in R 4.3.2.

Table 1 shows baseline characteristics of the 65 included patients (online suppl. Material S1; for all online suppl. material, see https://doi.org/10.1159/000542079 for details regarding inclusion). Briefly, 44 patients (67.7%) were female, mean age was 41 (±11.7) years, and mean BMI was 39.6 (±5.25) kg/m². Men had a higher waist circumference, lower leptin levels (both p < 0.001), lower adiponectin levels, and higher insulin and AgRP levels than women (p < 0.05). In online supplementary Tables S2 and S3, details are provided regarding the two subsamples included in the analyses based on data of psychological stress (n = 62) and hair glucocorticoids (n = 50), respectively.

HairF and HairE strongly correlated with each other in all models (all p < 0.001). In contrast, we did not see significant associations of HairF or HairE with PSS (online suppl. Table S4).

We did not see significant associations of stress measures (HairF, HairE, or PSS) with BMI or waist circumference. Results did not change notably after correction for age and sex (see online suppl. Table S5).

Of all tested appetite-regulating hormones, only CCK levels were significantly positively associated with HairE (p < 0.01 in all models), and in trend, with HairF levels (p < 0.06 in all models). For details, see Table 2, Figure 1a (HairF), and Figure 1b (HairE). The removal of a CCK outlier (>3SD above mean [CCK-log]) resulted in a nonsignificant association with HairF, but the association with HairE remained (data not shown). No other associations were observed between levels of hair glucocorticoids and appetite-regulating hormones. Neither did we observe any associations between PSS and levels of appetite-regulating hormones. Sensitivity analyses for diabetes medication (excluding n = 5 participants) did not yield different results compared to the main analyses.

The sensitivity analyses in women revealed a significant positive association of CCK with both HairE (p < 0.01 for all models) and HairF (p < 0.05 for all models). In addition, in women PSS tended to correlate negatively with PYY and positively with GIP levels (all p < 0.10, see online suppl. Table S6).

We investigated the association of chronic biological and psychological stress (hair glucocorticoid levels and perceived stress, respectively) with fasting serum levels of appetite-regulating hormones in patients with obesity. A 10% increase in HairE was associated with a 2.7% increase in circulating CCK levels. Additionally, a 10% increase in HairF tended to correlate with a 1.4% increase in CCK levels. Sensitivity analyses in women suggest additional potential associations of psychological stress with PYY and GIP levels.

From a behavioral perspective, the positive association between hair glucocorticoids and CCK levels is unexpected, as frequently described stress-induced eating tendencies in populations with obesity would suggest decreased levels of satiety hormones such as CCK in response to stress [3, 16, 17]. However, our study samples were taken in the fasted state, while CCK levels are especially modulated following a meal. Thus, it is possible that under chronic stress conditions the normally occurring postprandial rise in CCK levels is diminished, resulting in decreased meal-induced satiation [17]. Additionally, the correlation between chronic biological stress and CCK may reflect increased systemic inflammation; a state which is frequently observed in obesity, as well as related to hair glucocorticoid levels [4]. This hypothesis is in line with previous evidence regarding CCK responsiveness to inflammatory states and could be of relevance in view of suspected anti-inflammatory actions of CCK [18, 19]. Moreover, although we assessed chronic stress, our findings do partially align with previous studies demonstrating responsiveness of the brain-CCK-system to acute stress exposure [20‒22]. In turn, exogenous CCK receptor agonism induces HPA-axis activation, anxiety and panic symptoms through effects on the amygdala, the hypothalamus and the adrenal glands [23‒25]. While it remains unclear whether circulating endogenous CCK can also act on the brain or the adrenals to induce feelings of stress [25], our findings point toward this possibility. We note that the cross-sectional observational nature of our study does not allow to derive directionality, but suggests an interesting field for further research. Altogether, our findings expand previous evidence regarding CCK’s role in the context of obesity-related complications such as emotional dysregulation and inflammation.

Additionally, our results suggest that the association between hair glucocorticoids and CCK levels may be stronger in women. Together with the observed trend toward associations of psychological stress with PYY and GIP levels in women, this indicates that exploring sex differences in this context would be highly interesting; especially in view of studies showing that stress‐related eating is more prevalent in women than in men [3, 16]. Moreover, GIP’s crucial role for insulin action and glucose metabolism [26] indicates that (stress-related) decreased GIP levels are likely unfavorable for metabolism, which is in line with our hypothesis of higher chronic stress to be associated with lower levels of hormonal regulators of insulin signaling. Interestingly, stress-induced decreases of PYY levels in women have been reported previously in a laboratory setting using the Trier Social Stress Test [7]. However, the effect did not mediate stress-induced snacking. To our knowledge, data regarding men is not available in this regard.

However, we did not observe expected associations of stress measures with BMI and waist circumference [4, 5]. Additionally, we did not see the link of stress or glucocorticoid excess with dysregulations in leptin, insulin, and adiponectin levels, as previously reported in human and animal laboratory settings and disease states such as Cushing’s disease [3]. Possibly, the lack of previously documented associations may be attributed to the high BMI within our study population (BMI ≈ 40 kg/m², corresponding to severe obesity). Obesity itself can cause major dysregulations in both the hormonal appetite system and the HPA-axis [2, 3]. Thus, the homogeneously high BMI of our study population may have created a ceiling effect which obscured potential associations. Future studies regarding the association between chronic stress and levels of appetite-regulating hormones should therefore include participants across multiple weight categories.

Altogether, our study suggests an association between hair glucocorticoid levels and circulating levels of CCK in patients with obesity. No other associations were seen with appetite-regulating hormones. Future studies are needed to pinpoint the relation between long-term stress, emotional regulation, and appetite signaling.

We would like to thank all patients who participated in the study. We would also like to thank Annemieke van der Zwaan for her support regarding data management and study coordination.

Participants provided written informed consent. The study was approved by the Local Ethics Committee (called “Medisch Ethische Toetsings Commissie [METC]” affiliated with the Erasmus MC, University Medical Center Rotterdam, Rotterdam, the Netherlands; Approval No. “MEC2012–257”). The privacy rights of all subjects have been protected throughout all stages of the project.

The authors have no conflicts of interest to declare.

This article was written as part of the research project “Stress in Action”: www.stress-in-action.com. Stress in Action is financially supported by the Dutch Research Council and the Dutch Ministry of Education, Culture and Science (NWO gravitation Grant No. 024.005.010). Part of this research was also funded by an NWO Vidi Grant (number 91716453) and the Elisabeth Foundation.

S.K. and R.L. had a leading role in the conception and design of the work, as well as data acquisition, analysis, and interpretation. They equally contributed to the drafting and reviewing of the manuscript. R.E.H.M. and E.S.V. contributed to the conceptualization as well as data management, and critically reviewed the manuscript. M.H.J.H., B.W.J.H.P., and M.K., had leading roles in the funding acquisition. They also contributed to the interpretation of the data and critically reviewed the manuscript. J.A.V., M.R.B., and S.A.A.B. had a supporting supervising role in the data interpretation and reviewed the manuscript critically. E.F.C.R. had a leading role in the funding acquisition and supervision of all work related to this research, including the conception and design of the study, as well as data analysis and interpretation, and the writing and reviewing of the manuscript. All authors reviewed the final version of the manuscript and provided approval prior to submission.

Susanne Kuckuck and Robin Lengton shared first authorship.

The data that support the findings of this study are not publicly available due ongoing research on variables related to the dataset but are available from the corresponding author (EFCvR).