Changes in adipose glucocorticoid metabolism before and after bariatric surgery assessed by direct hormone measurements




Increased intra-adipose cortisol is thought to promote obesity, but few human studies have investigated intra-adipose glucocorticoid hormones and none have demonstrated prospective changes with fat loss.

Design and Methods

Subcutaneous adipose tissue (SAT) was obtained from obese subjects before and 1-year after surgery-induced fat loss, and from nonobese controls. In a second similar cohort of obese subjects, adipocytes and stromal-vascular fraction were isolated. Intra-adipose cortisol and cortisone levels were analyzed by liquid chromatography mass spectrometry and HSD11B1/HSD11B2 mRNA by qPCR.


SAT cortisol/cortisone ratio before fat loss, median 4.8 (interquartile range, 4.1-5.7), was higher than after fat loss, 1.9 (1.0-2.7) (P = 0.001), and compared to nonobese controls, 3.2 (2.4-3.9) (P = 0.005). Cortisone before fat loss, 2.3 (1.2-2.9) nmol/kg, was lower than after fat loss, 5.8 (3.0-10.2) nmol/kg (P = 0.042), and compared to controls, 5.1 (3.8-6.7) nmol/kg (P = 0.013). HSD11B1 was predominantly expressed in mature adipocytes, whereas HSD11B2 was expressed at a higher level in stromal-vascular fraction.


The intra-adipose glucocorticoid metabolism was markedly altered in the extremely obese state with increased cortisol levels relative to cortisone, whereas fat loss restored this balance approximating nonobese subjects. Changes were more pronounced for cortisone than cortisol, suggesting an adaptive response to insufficient intra-adipose cortisol levels in obesity.


Obesity has reached epidemic proportions and is a major health-economic burden to society. Though the principal causes are increased consumption of energy-dense food and reduced physical activity, various factors may confer individual vulnerability to obesity. Glucocorticoid excess, such as in Cushing's disease, is associated with the development of central obesity, increased insulin resistance, and hallmarks of the metabolic syndrome.

Although obese subjects have normal or lower levels of circulating cortisol [1], the biological effects of cortisol in adipose tissues may be locally modulated. Most studies agree that in obesity the expression of 11β-hydroxysteroid dehydrogenase type 1 (11β-HSD type 1, HSD11B1) is increased in adipose tissue [2]. This intracellular enzyme activates cortisone to cortisol, increasing the availability of cortisol to the glucocorticoid receptor which mediates biological effects of cortisol via gene transcription. Evidence derived from transgenic mice selectively overexpressing HSD11B1 in adipose tissues indicates that increased active glucocorticoid hormone in fat produces visceral obesity, insulin resistance, and features of the metabolic syndrome [11]. Recently, 11β-HSD type 1 has become an attractive target in research on diabetes and obesity to investigate the potential for therapeutic modulation [12, 13].

A limitation of many previous studies is that they do not take into account factors that may influence the in vivo metabolism of cortisol. Commonly, the biological activity of cortisol in tissues has been estimated from the HSD11B1 gene expression, which correlates with 11β-HSD type 1 protein or from in vitro measurements of 11β-HSD type 1 activity [3, 14]. However, this enzyme may be competitively inhibited [15], and is also linked to glucose-6-phosphate dehydrogenase, which is affected by sex hormones and diet [16]. Moreover, other cortisol metabolizing enzymes, such as 11β-HSD type 2 (HSD11B2) [17] and 5α- and 5β-reductases [9, 18, 19], could be underappreciated as modulators of adipose tissue cortisol activity. Very recently, it has also been suggested that 11β-HSD type 1 could catalyze the reverse dehydrogenase reaction in adipose tissue [20].

Although evidence indicates that the adipose glucocorticoid metabolism is altered in human obesity, only a few studies [3, 14, 21] have actually measured in vivo intra-adipose cortisol, and these studies did not find increased levels. Additionally, prospective studies investigating the impact of weight loss on intra-adipose glucocorticoid hormones are lacking. The aim of this study was to assess the intra-adipose glucocorticoid metabolism by direct hormone measurements in a prospective setting with varying states of obesity. The impact of potential confounding interindividual factors was limited by a prospective design analyzing matched subcutaneous adipose tissue (SAT) biopsies before and after profound fat loss. Moreover, to explore the contributions from different adipose cells, we studied the expressions of HSD11B1 and HSD11B2 in isolated adipocytes and stromal-vascular fraction (SVF).


Study population

Adipose biopsies from two cohorts (main cohort and second cohort) were studied. All subjects were of Caucasian origin, and gave written consent about the study after full explanation of the purpose and nature of all procedures used. The study was approved by the Regional committee for Medical and Health Research Ethics, Western Norway (REK Vest) and was carried out in accordance with the principles of the Declaration of Helsinki and its amendments.

In the main cohort, obese subjects undergoing bariatric surgery at Førde Central Hospital, Norway, between 2003 and 2007 (biliopancreatic diversion with duodenal switch) were consecutively enrolled. The original study population (n = 16) was described previously [22], and in this study an additional subject was included. Five patients were excluded owing to insufficient adipose tissue, and one for current use of glucocorticoid medication (total, n = 11). Before fat loss, seven of the obese patients had type 2 diabetes necessitating insulin therapy (n = 4) or oral antidiabetic drugs (n = 3). Some subjects used statins (n = 4) and/or antihypertensive drugs (n = 4). After fat loss, three patients were treated for hypertension. The nonobese control subjects (n = 12) were patients undergoing elective hernia repair and complying with the inclusion criteria of BMI below 28 kg/m2, no history of inflammatory or metabolic disorders, and no current medication.

In the second cohort, six obese subjects undergoing laparoscopic sleeve gastrectomy were included for investigations of adipocytes and SVF isolated from SAT.

Sample collection

Fasting glucose, insulin, insulin C-peptide, triglycerides, total-cholesterol, high-density lipoprotein (HDL)- and low-density lipoprotein (LDL)-cholesterol, and high-sensitive CRP were measured in blood. The homeostatic model assessment (HOMA) was used as an index of insulin resistance; formula: fasting serum-insulin (mIU/l) × fasting-glucose (mmol/l)/22.5.

Whole abdominal SAT was obtained between 9 and 10 AM from obese patients (both cohorts) and between 9 and 11 AM for nonobese controls (main cohort), by surgical incision in the initial stage of surgery. Twelve months later, SAT from the same abdominal region was acquired from the obese patients under local anesthesia during visits to the outpatient clinic. Omental VAT was also acquired from seven of the obese individuals at surgery in the main cohort.

In the second cohort, adipocytes and SVF were isolated immediately after biopsy collection. About 700-800 mg of fresh tissue was rinsed in GIBCO™ Hanks' Balanced Salt Solution (HBSS) (Invitrogen, Paisley, UK) and sliced. The tissue was transferred to a 15-ml tube containing 4 ml of preheated HBSS and 2.6 Wunch units of Liberase Blendzyme 3 (Roche Diagnostics, Penzberg, Germany). The tissue/collagenase/thermolysin mixture was shaken at 37°C for 30 min, and sieved through a 25-mm polypropylene filter with pore size of 200 μm (148116, Spectrum Laboratories, Inc., Rancho Dominguez, CA, USA) using an Easy Pressure syringe filter holder (PALL 4320, Life Sciences, Munich, Germany) and 10 ml syringe. One milliliter of HBSS w/5% BSA (A1311, US Biological, Massachusetts, MA, USA) was flushed through the filter twice. After 1 min, floating adipocytes were transferred to a 2-ml Eppendorf tube and rinsed with 1 ml of HBSS w/5% BSA. The remaining solution (containing SVF) was centrifuged at 2,500 rpm for 5 min. The adipocytes and SVF were dissolved in 1 ml of Qiazol lysing buffer (Qiagen, Germantown, MD, USA), frozen in liquid nitrogen and stored at −80°C.

Measurements of cortisol and cortisone in fat tissue

Measurements of cortisol and cortisone in whole-adipose tissue (main cohort) were performed by liquid chromatography tandem mass spectrometry based on a previously published method [23]. Briefly, we improved the method by including cortisone-d2 (4-Pregnen-17α,21-diol-3,11,20-trione-1,2-d2, 98 atom %D, CDN Isotopes Inc., Quebec, Canada) as a stable isotope-labeled internal standard for cortisone. A Dionex Ultimate 3000 ×2 Dual Analytical LC system (Dionex, Sunnyvale, CA, USA) was used for on-line sample cleanup and chromatographic separation, and this was coupled to an API 5500 mass spectrometer (Applied Biosystems/MDS, Foster City, CA, USA) with a TurboIonSpray™ source operating in multiple reaction monitoring (MRM) mode. Ion-source conditions were as follows: temperature, 600°C; ion spray voltage, +5,500 V; gas-1, 50 psig; gas-2, 40 psig; and curtain gas, 20 psig. Entrance potential was 10 V and collision gas 7. For cortisol and cortisone, respectively, the quantifier MRM transitions were 363/121 and 361/163, and qualifier MRM transitions were 363/327 and 361/105. Analogously, for cortisol-d4 and cortisone-d2, the quantifier transitions were 367/121 and 363/165, and qualifier transitions were 367/331 and 363/106, respectively. Specificity was confirmed if the ratio of quantifier and qualifier peak areas was within ±30% of the corresponding ratio of the highest working standard. For glucocorticoid measurements, approximately 100 mg of frozen SAT was gravimetrically weighed and analyzed. Two large SAT biopsies (more than 5 g) served as quality controls and duplicate pieces ran in each batch.

We validated the method precision by analyzing three large SAT biopsies. Individually processed pieces were analyzed in quintlicates, and the mean levels were 1.20, 1.56, and 17.1 nmol/l for cortisol and 0.37, 1.56, and 7.6 nmol/l for cortisone. The within-batch coefficients of variations (CV) ranged from 13.6 to 16.3% for cortisol, from 10.3 to 21.4% for cortisone, and from 10.9 to 15.1% for the cortisol/cortisone ratio. Levels in replicate samples (n = 5) individually processed from tissue homogenates were more precisely determined; the CV for cortisol and cortisone concentrations and the cortisol/cortisone ratio ranged between 1.6 and 4.6%. This indicated that tissue heterogeneity within each biopsy contributed most to the variations in measurements. Accuracy, determined by adding cortisol and cortisone to tissue homogenates, ranged from 95-106 to 95-116%, respectively. For both substances, the method was linear in the measuring range of 0.2-200 nmol/l and the limit of quantification was 0.2 nmol/kg (signal-to-noise-ratio, >10). All study samples were run in one batch, and the recoveries of the internal standards were consistently high with a CV of <6.5% considering all samples. Relative to samples from lean subjects, recoveries in biopsies obtained before fat loss were 97 ± 6% for cortisol-d4 and 96 ± 7% for cortisone-d2. Correspondingly, the recoveries from SAT obtained after weight reduction were 95 ± 14% for cortisol-d4 and 97 ± 16% for cortisone-d2.

Gene expression analysis

Whole-adipose tissue (200-300 mg) was sliced frozen and homogenized (main cohort), and RNA was extracted as reported previously [22]. qPCR was performed using the LightCycler480 Probes Master kit and the LightCycler480 rapid thermal cycler system (Roche Diagnostics, Penzberg, Germany). cDNA was synthesized from 0.1 μg total RNA per sample using the SuperScript® VILOTM cDNA Synthesis Kit (Invitrogen, Paisley, UK). HSD11B1 and HSD11B2 expression was quantified relative to TATA-binding protein (TBP) using target-specific primers and Universal ProbeLibrary (UPL) probes. The forward and reverse primers and UPL probe used were HSD11B1, F:tctctgtgttcttggcctca, R:gagctgcttgcatatggactatc (probe 8); HSD11B2, F:gtcaaggtcagcatcatcca, R:cactgacccacgtttctcac (probe 71). TBP was used as reference gene as it showed consistent expression in all samples (UPL Human TBP Gene Assay, Roche). HSD11B1 was amplified in duplex with TBP. As duplex with TBP reduced HSD11B2 amplification efficiency, HSD11B2 expression was calculated relative to TBP from a separate run. Amplification efficiency was measured by standard curves of concentrated cDNA from human SAT.

Statistical methods

All data are reported as median (interquartile range), unless otherwise indicated. Groups were compared by Kruskal-Wallis test, and Wilcoxon tests with Bonferroni correction were used for post hoc analyses. To check for possible confounders, ANCOVA analysis was performed on continuous and ranked dependent variables (glucocorticoid levels, cortisol/cortisone ratio, HSD11B1, and HSD11B2 mRNA) with age and sex as covariates. Adjusted data are reported if these tests changed the statistical outcome. Correlation between HSD11B1, HSD11B2, cortisol/cortisone ratio, and BMI were performed by Spearman's partial rank-order correlation, controlling for age, sex, and where appropriate, BMI. A two-tailed P value of <0.05 was considered significant. Statistical analysis was performed using R version 2.13.0 (R Foundation for Statistical Computing, Vienna, Austria).


Study population

Characteristics of the subjects included in the main cohort are summarized in Table 1. The obese group (n = 11) included more males and was nonsignificantly older than the nonobese group (n = 12). One year after bariatric surgery, the median BMI was reduced from 53.0 to 35.0 kg/m2, which was still higher than in the nonobese group (BMI, 23.7 kg/m2). Additionally, most biochemical markers of metabolic syndrome were largely improved and comparable to the nonobese subjects. There were similar trends in the second cohort (n = 6) as summarized in Table 2.

Table 1. Characteristics of the obese subjects and of the nonobese controls (main cohort)
 Extremely obese patientsNonobese controls
 Before fat loss1-Year after fat loss    
 MedianQuartilesMedianQuartilesP valueaMedianQuartilesP valueaP valueb
  1. Data expressed as median (interquartile range). P-levels < 0.05 are reported in bold.
  2. aStatistical significance compared to data before bariatric surgery.
  3. bStatistical significance compared to data before bariatric surgery 1 year later. Wilcoxon test, two sided, P < 0.05.
n (♀)11 (6)    12 (4)   
Age (years)36(31– 44)37(32–45)45.5(39–52)0.3100.388
BMI (kg/m2)53.0(50.0–57.5)35.0(29.5–35.5)0.00423.7(22.2–27.3)<0.001<0.001
Fasting glucose (mmol/l)6.2(5.7–8.2)4.9(4.7–5.2)0.0015.1(4.7–5.3)<0.0010.644
Fasting insulin (mU/l)20.2(14.3–41.5)4.4(3.9–9.3)0.0035(2.8–7.4)<0.0010.478
Insulin C-peptide (nmol/l)1.2(1.1–2.1)0.9(0.6–1.1)0.2780.7(0.4–0.9)0.0070.185
Total cholesterol (mmol/l)5.0(4.2–5.2)3.5(2.9–4.0)0.0025.2(4.2–6.0)0.4050.003
HDL-cholesterol (mmol/l)1.0(0.9–1.1)1.0(0.8–1.1)0.4431.3(1.1–1.6)0.0090.004
LDL-cholesterol (mmol/l)3.0(2.2–3.2)1.8(1.3–2.2)0.0103.1(2.6–3.9)0.053<0.001
Triglycerides (mmol/l)1.9(1.4–2.3)1.5(1.0–2.1)0.0541.2(0.8–1.7)0.0560.406
High-sensitive CRP (mg/dl)14(9–19)2(1–5)0.0051(1–1)<0.0010.0323
Table 2. Characteristics of the obese subjects before and after fat loss (second cohort)
 Before fat loss1 Year after fat loss
 MedianQuartilesMedianQuartilesP valuea
  1. Data expressed as median (interquartile range).
  2. aStatistical significance compared to data before bariatric surgery. Wilcoxon test, two sided, P < 0.05.
n (♀)6 (3)    
Age (years)47.0(38.3–58.0)  
BMI (kg/m2)46.9(45.0–51.7)30.2(30.0–34.5)0.003
Fasting glucose (mmol/l)7.2(6.3–8.3)5.0(4.8–5.0)0.125
Fasting insulin (μE/l)8.6(4.5–10.0)5.7(2.1–7.8)0.063
Insulin C-peptide (nmol/l)1.0(1.0–1.4)0.9(0.8–0.9)0.063
Total cholesterol (mmol/l)4.8(4.4–5.3)5.3(4.8–5.6)0.181
HDL-cholesterol (mmol/l)0.9(0.9–1.0)1.3(1.2–1.4)0.125
LDL-cholesterol (mmol/l)3.5(3.1–4.0)3.9(3.2–4.5)0.625
Triglycerides (mmol/l)1.8(1.6–2.2)1.0(0.8–1.2)0.181
High-sensitive CRP (mg/dl)7.0(5.3–8.5)4.5(3.0–8.3)1.000

Glucocorticoid hormones, and HSD11B1 and HSD11B2 mRNA expressions in whole fat

In the main cohort, we assessed the glucocorticoid hormones and their metabolism in SAT by direct measurements on biopsies obtained from subjects with different states of obesity and metabolic disease. The cortisol/cortisone ratio was calculated as an index of in vivo cortisol activation. This ratio was highest in the obese subjects before fat loss compared to the same individuals after fat loss (P = 0.001), and to the nonobese group (P = 0.005) (Figure 1 and Table 3). Specifically, following profound fat loss the median cortisol/cortisone ratio declined 2.5-fold in the obese patients. Considering this shift in cortisol/cortisone ratio concurrent with fat loss, we found that the levels of cortisone, not cortisol, showed the most prominent change. Median cortisone level was more than a twofold lower in the obese patients compared with both after fat loss and with nonobese controls (P ≤ 0.042).

Figure 1.

Cortisol/cortisone ratio and levels of HSD11B1 and HSD11B2 expression in SAT. Box plots of cortisol/cortisone ratio (left), HSD11B1 mRNA levels (center), and HSD11B2 mRNA levels (right). Measurements were performed on SAT obtained in the main cohort from obese patients before and after fat loss (paired samples, n = 11), and from nonobese subjects (n = 12). Expression of HSD11B2 mRNA was determined in 9 out of the 11 obese patients after fat loss owing to insufficient SAT biopsies. Groups were compared by Kruskal–Wallis test applying Wilcoxon tests with Bonferroni correction for post hoc analyses; level of significance, P < 0.05. *The cortisol/cortisone ratio comparing obese patients after fat loss with nonobese controls was not statistically significant after adjustment for age and sex (ANCOVA, p = 0.073). Box plots: percentiles, 25; median, 75; O, outliers. HSD11B1, gene encoding 11β-hydroxysteroid dehydrogenase type 1; HSD11B2, gene encoding 11β-hydroxysteroid dehydrogenase type 2.

Table 3. Glucocorticoids levels, and HSD11B1 and HSD11B2 mRNA in adipose tissues (main cohort)
 Extremely obese patients (n = 11)Nonobese controls (n = 12)
 Before fat loss1 Year after fat loss    
 MedianQuartilesMedianQuartilesP valueaMedianQuartilesP valueaP valueb
  1. Data expressed as median (interquartile range). P-levels reported in bold were significant after adjustment for age and sex.
  2. mRNA levels are relative to the TATA-binding protein. VAT was only available from seven extremely obese patients.
  3. aStatistical significant compared to data obtained at bariatric surgery
  4. bStatistical significant compared to data obtained at bariatric surgery 1 year later. Wilcoxon test, two sided, P < 0.05.
  5. cNot significant after adjustment for age and sex; ANCOVA, adjusted means (95% CI) 14.9 (11.9–17.9) versus 12.4 (9.3–15.5) nmol/kg, P = 0.250.
  6. dNot significant after adjustment for age and sex; ANCOVA, adjusted means (95% CI) 3.1 (2.3–3.8) versus 2.1 (1.4–2.8) nmol/kg, P = 0.073.
SAT cortisol (nmol/kg)8.1(6.6–15.2)11.9(9.4–13)0.89815(13.2–19.8)0.0790.045c
SAT cortisone (nmol/kg)2.3(1.2–2.9)5.8(3.0–10.2)0.0425.1(3.8–6.7)0.0131.000
SAT cortisol/cortisone4.8(4.1–5.7)1.9(1.0–2.7)0.0013.2(2.4–3.9)0.0050.044d
SAT HSD11B1 mRNA1.5(1.4–1.6)0.5(0.3–0.9)0.0010.7(0.6–1.6)0.0450.166
SAT HSD11B2 mRNA0.014(0.003–0.02)0.083(0.07–0.11)0.0040.066(0.05–0.09)0.0020.422
VAT cortisol (nmol/kg)10.6(10.2–20.3)       
VAT cortisone (nmol/kg)2.6(2.2–4.5)       
VAT cortisol/cortisone4.7(3.5–6.1)       
VAT HSD11B1 mRNA1.4(1.1–1.4)       
VAT HSD11B2 mRNA0.05(0.03–0.06)       

The metabolism of cortisol in the same biopsies was then assessed using a fundamentally different methodological approach, by measuring mRNA expression of 11β-HSD type 1 and 2. Before fat loss, HSD11B1 mRNA levels were threefold higher (P = 0.001) and HSD11B2 mRNA levels were sixfold lower (P = 0.004, n = 9) in the obese patients than at 1-year follow-up.

There was a strong agreement between these two methods as SAT cortisol/cortisone ratio and HSD11B1 expression correlated positively in the obese subjects as well as in the nonobese controls (Figure 2). Moreover, this correlation also held true in the prospective analysis of obese patients undergoing fat loss; the degree of change in cortisol/cortisone ratio correlated positively with the degree of change in HSD11B1 mRNA (ρ = 0.60, P = 0.034, Figure 3).

Figure 2.

Correlations of adipose cortisol/cortisone ratio to HSD11B1 mRNA expression and to BMI. Correlation plots show the associations of cortisol/cortisone ratio (x-axis) to HSD11B1 mRNA levels (upper three plots, y-axis) and to BMI (lower three plots, y-axis). All measurements were performed in SATs obtained in the main cohort from obese patients before and after fat loss (paired samples, n = 11), and from nonobese subjects (n = 12). Correlation coefficients, adjusted for age and sex, are reported as Spearman's ρ with P values (significance level, P < 0.05). Lines are Lowess (locally weighted scatter plot smoothing) curves. BMI, body mass index (kg/m2); HSD11B1, gene encoding 11β-hydroxysteroid dehydrogenase type 1.

Figure 3.

Prospective correlations of changes in adipose cortisol/cortisone ratio to HSD11B1 mRNA expression and to BMI. Correlation plots show the associations of percentage reduction in cortisol/cortisone ratio (y-axis, both plots) to percentage reduction in HSD11B1 mRNA levels (x-axis, upper plot), and to percentage reduction in BMI (x-axis, lower plot). All measurements were performed in SATs obtained in the main cohort from obese patients before and 1 year after fat loss (paired samples, n = 11). Correlation coefficients, adjusted for age and sex, are reported as Spearman's ρ with P values (significance level, P < 0.05). Lines are Lowess (locally weighted scatter plot smoothing) curves. BMI, body mass index; HSD11B1, gene encoding 11β-hydroxysteroid dehydrogenase type 1.

If glucocorticoid metabolism in adipose tissue is linked with obesity, an association between glucocorticoids and BMI would be expected. The SAT cortisol/cortisone ratio strongly correlated with BMI within the nonobese group and within the obese subjects after fat loss, whereas not reaching statistical significance before fat loss (Figure 2). The correlation statistics for HSD11B1 mRNA with BMI for the obese subjects was ρ = 0.57 (P = 0.068) before fat loss and ρ = 0.65 (P = 0.029) after fat loss, and for the nonobese subjects ρ = 0.38 (P = 0.242). Interestingly, in the prospective analysis the degree of change in BMI after fat loss correlated with the degree of change in cortisol/cortisone ratio (ρ = 0.599, P = 0.034, Figure 3).

HSD11B1 and HSD11B2 in isolated adipocytes and SVF

As adipose tissue comprises several cell types other than adipocytes, we wanted to evaluate whether HSD11B1 and HSD11B2 may primarily modulate glucocorticoid metabolism in SVF or adipocytes (second cohort, Table 2, n = 6). We found that HSD11B1 was predominantly expressed in adipocytes, and analogous to whole fat, showed a trend toward a twofold decrease in expression after weight loss; 2.8 (2.5-3.2) versus 1.5 (1.3-2.5) (P = 0.063) (Figure 4). In contrast, HSD11B2 was predominantly expressed in SVF over adipocytes. HSD11B2 mRNA in SVF did not change after fat loss; 0.30 (0.26-0.35) versus 0.25 (0.25-0.28) (P = 0.156), whereas in adipocytes HSD11B2 mRNA increased from nondetectable levels in extreme obesity to 0.03 (0.03-0.04).

Figure 4.

HSD11B1 and HSD11B2 expression levels in isolated adipose tissue fractions. Box plots of HSD11B1 mRNA levels (upper) and HSD11B2 mRNA levels (lower) measured in isolated adipocytes and isolated stromal-vascular cells. Measurements were performed on SATs obtained in the second cohort from obese patients before and after fat loss (paired samples, n = 6). Box plots: percentiles, 25; median, 75; O, outliers. HSD11B1, gene encoding 11β-hydroxysteroid dehydrogenase type 1; HSD11B2, gene encoding 11β-hydroxysteroid dehydrogenase type 2.


In this study, we prospectively assessed changes in the adipose glucocorticoid metabolism during fat loss by measurements of cortisol and cortisone levels in vivo. By investigating enzyme substrate and product, the cortisol/cortisone ratio provides an index of adipose 11β-HSD type 1 activity. As opposed to the studies measuring HSD11B1 mRNA expression or activity in vitro, however, the cortisol/cortisone ratio reflects any factor potentially impacting the local hormone metabolism. We found that the cortisol/cortisone ratio in SAT was increased more than twofold before compared to after fat loss, supporting a local upregulation of the cortisone to cortisol conversion in the obese state. Moreover, our data support a link between the local cortisol metabolism and the level of obesity because SAT cortisol/cortisone ratio correlated positively with BMI within the groups of lean and obese subjects after fat loss. Interestingly, this association was not only confined to the cross-sectional analysis, as there was also a prospective correlation between the degree of change in BMI and the degree of change in cortisol/cortisone ratio in subjects undergoing profound fat loss. Thus, our study indicates that with increasing BMI the glucocorticoid metabolism in SAT shifts, favoring increased levels of active cortisol relative to inactive cortisone. In line with this, profound fat loss had the opposite effect by promoting a restoration of the cortisol/cortisone balance.

Studies on in vivo hormone metabolism in tissues are challenging, which is why we applied two fundamentally different analytical techniques to assess the glucocorticoid metabolism in adipose tissue. The high agreement between these techniques, as indicated by the strong correlations between HSD11B1 mRNA and SAT cortisol/cortisone ratio, serves to verify that the measurements were reliable and accurately reflected the adipose glucocorticoid metabolism. Moreover, the prospective correlations between percentage changes in cortisol/cortisone ratio and percentage changes in both HSD11B1 mRNA and BMI further support that 11β-HSD type 1 is the main factor driving the marked shift in adipose glucocorticoid metabolism.

The increased expression of 11β-HSD type 1 in obese subjects is in line with most previous reports measuring HSD11B1 mRNA and enzyme activity in vitro [2]. Our in vivo investigations extend this observation to include an altered intra-adipose balance, favoring active cortisol over inactive cortisone. The increased SAT cortisol/cortisone ratio in the obese state is also in accordance with a sophisticated study on SAT microdialysate using labeled cortisone infusion techniques [24]. An important novel finding is the prospective change in the adipose glucocorticoid metabolism, as assessed by direct hormone measurements, demonstrating normalization after profound fat loss.

A recent study [10] reported that the linear relationship between HSD11B1 mRNA and BMI disappeared in the morbidly obese state. This is line with our study because neither cortisol/cortisone ratio nor HSD11B1 mRNA correlated significantly with BMI in the extremely obese subjects. As BMI reaches a certain threshold, it may be that HSD11B1 expression is increasingly influenced by factors independent of weight, as noted by Simonyte et al.

Although adipose cortisol metabolism has been a field of intense research for more than a decade, we are not aware of any reports that have shown increased adipose cortisol levels in human obesity. Lindsay et al. [3] studied 32 overweight subjects, and found that intra-adipose cortisol was positively associated with fasting insulin, but not with BMI or other metabolic parameters. In a cohort of 27 subjects, Wake et al. [14] found no correlation between SAT cortisol and cortisone levels, and BMI. Finally, in the study of Koska et al. [21], which included 20 subjects, baseline SAT cortisol concentrations and cortisol/cortisone ratio were not associated with, or could not predict changes in, BMI, anthropometric, or metabolic variables. Our prospective study design, comparing cortisol concentrations in matched SAT for two time points representing large changes in BMI and parameters of metabolic disease, should offer increased power to detect the differences in cortisol adipose levels. As such, our data showing indifferent adipose cortisol levels in the obese and nonobese state are in accordance with the few studies that have directly investigated in vivo tissue concentrations, albeit using less specific methods.

In the previous studies, the lack of an association between intra-adipose cortisol and obesity was explained by exaggerated cortisol response in subjects with metabolic syndrome and analytical difficulties in steroid tissue extraction. If an exaggerated cortisol response caused by general anesthesia or surgery was important in our study, this stress would increase, rather than cancel out, intra-adipose cortisol differences between samples from the extremely obese subjects and samples acquired at the outpatient clinic after fat loss. Moreover, a recent article indicated that rapid fluctuations in circulating cortisol levels are not reflected in adipose tissue because uptake into adipose tissue and local turnover is slow [25]. As for analytical bias, we have no indication that the efficacy of steroid extraction varied with BMI as the isotopic-labeled internal standards were recovered highly consistently in our different groups.

The most noticeable difference in SAT glucocorticoid levels was decreased cortisone in the obese compared to the nonobese state. Conceivably, this may suggest that activity of 11β-HSD type 1 increases in obesity as an adaptive response to insufficient levels of cortisol in SAT. This could result from either decreased glucocorticoid delivery or increased glucocorticoid elimination. Factors impacting delivery may include a reduced glucocorticoid exchange rate between plasma and adipose tissue, which has recently been shown to be low [25], reduced blood flow, and increased diffusion distances [26], and decreased local release of cortisol from cortisol-binding globulin [27]. Factors potentially increasing elimination of glucocorticoids from adipose tissues could include 5α-reductase which is expressed in both subcutaneous and visceral fat depots [18, 19], and various steroid transporters [28, 29]. Mechanisms pertaining to all these factors in adipose tissue should be further explored to better understand the reason why glucocorticoid metabolism is altered in obesity and the putative causal effect of this alteration.

Although adipose tissue mostly comprises adipocytes, the SVF includes a range of immunoactive cells that are crucial regulators of the chronic low-grade inflammatory state in obesity [30]. Studies on whole SAT may overlook the features of intracellular cortisol metabolism specific to these cells, which is why we investigated 11β-HSD type 1 and 2 in isolated adipose fractions. We found that HSD11B2 was predominantly expressed and HSD11B1 reciprocally lower expressed in SVF compared to adipocytes. Notably, 11β-HSD type 2 exerts the opposite function to 11β-HSD type 1 by inactivating cortisol, and has considerably higher substrate affinity than 11β-HSD type 1 [31, 32]. In immunoactive cells this enzyme may decrease glucocorticoid sensitivity by reducing cortisol available to the glucocorticoid receptors and possibly mineralocorticoid receptors [33]. However, our data do not support a link between 11β-HSD type 2 in immunoactive cells and obesity because we did not observe a change in HSD11B2 expression in isolated SVF after fat loss. Indeed, in whole SAT HSD11B2 mRNA increased after fat loss, but this change may have primarily occurred in adipocytes as isolated adipocytes exhibited a trend toward increased expression. The role of 11β-HSD type 2 in obesity is further obscured by the lack of correlation to BMI and glucocorticoid hormones. Further studies are needed to dissect the adipose glucocorticoid metabolism within specific subtypes of immunoactive cells.


In summary, we performed a prospective human study investigating the in vivo glucocorticoid metabolism in matched SAT obtained before and after profound fat loss, as well in nonobese subjects. Overall, the glucocorticoid metabolism was strongly linked to BMI, and favored increased intra-adipose cortisol levels relative to inactive cortisone in the obese state. Profound fat loss restored this balance approximating the glucocorticoid metabolism in nonobese subjects. Intra-adipose levels of cortisone, rather than cortisol, showed the most prominent change, suggesting that the altered glucocorticoid metabolism may be an adaptive response to insufficient levels of adipose cortisol. We propose that more detailed investigations on isolated cell fractions are merited to assess cell specific as well as possible paracrine and autocrine effects of glucocorticoids in human adipose tissue.


The authors thank Christine Haugen, Carol Cook, and Anita Ivarsflaten for expert technical assistance. The authors are grateful to Dr. Kjell Petersen, Rita Holdhus, Anne-Kristin Stavrum, and others at the Norwegian Bioinformatics Platform and the Norwegian Microarray Consortium (national FUGE technology platforms,