Steroids 76 (2011) 1575–1581

Contents lists available at SciVerse ScienceDirect

Steroids

journal homepage: www.elsevier .com/locate /s teroids

Increased urinary glucocorticoid metabolites are associated with metabolicsyndrome, hypoadiponectinemia, insulin resistance and b cell dysfunction

Rene Baudrand a, Carmen Campino a, Cristian A. Carvajal a, Oliviero Olivieri b, Giancesare Guidi b,Giovanni Faccini b, Javiera Sateler c, Javiera Cornejo d, Betty San Martin d, Jose M. Dominguez a,Jaime Cerda e, Lorena M. Mosso a, Gareth I. Owen f, Alexis M. Kalergis g,h,i, Carlos E. Fardella a,h,⇑a Department of Endocrinology, School of Medicine, Pontificia Universidad Católica de Chile, Chileb Department of Medicine, University of Verona, Italyc Department of Family Medicine, School of Medicine, Pontificia Universidad Católica de Chile, Chiled Faculty of Veterinary Medicine, Universidad de Chile, Chilee Department of Public Health, School of Medicine, Pontificia Universidad Católica de Chile, Chilef Department of Physiology, Faculty of Biological Sciences, Pontificia Universidad Católica de Chile, Chileg Department of Molecular Genetics, Faculty of Biological Sciences, Pontificia Universidad Católica de Chile, Chileh Millennium Nucleus of Immunology and Immunotherapy, Pontificia Universidad Católica de Chile, Chilei Department of Rheumatology, Faculty of Medicine, Pontificia Universidad Católica de Chile, Chile

a r t i c l e i n f o

Article history:Received 3 August 2011Received in revised form 26 September2011Accepted 27 September 2011Available online 5 October 2011

Keywords:Glucocorticoid metabolitesMetabolic syndrome11b-Hydroxysteroid dehydrogenaseAdiponectinInsulin resistance

0039-128X/$ - see front matter � 2011 Elsevier Inc. Adoi:10.1016/j.steroids.2011.09.010

Abbreviations: MetS, metabolic syndrome; T2DM, thypothalamic–pituitary–adrenal; 11b-HSD1, 11b-hytype 1; GR, glucocorticoid receptor; THM, tetrahydadipose tissue; BMI, body mass index; HOMA-IR, hominsulin resistance index; HDL, high-density lipoprotetein; MHO, metabolically healthy obese; p25, percent⇑ Corresponding author at: Department of Endocr

Pontificia Universidad Católica de Chile, Lira 85, 5th FlTel.: +56 2 354 3095; fax: +56 2 638 5675.

E-mail address: cfardella@med.puc.cl (C.E. Fardella

a b s t r a c t

Metabolic syndrome (MetS) may have increased cortisol (F) production caused by 11b-hydroxysteroiddehydrogenase 1 (11b-HSD1) in liver and adipose tissue and/or by HPA axis dysregulation. F is then mainlymetabolized by liver reductases into inactive tetrahydrometabolites (THMs). We measured THM levels inpatients with or without MetS and evaluate the correlation between THMs and anthropometric andbiochemical parameters. We recruited 221 subjects, of whom 130 had MetS by ATP III. We evaluated F,cortisone (E), adipokines, glucose, insulin and lipid profiles as well as urinary (24 h) F, E and THM levels.b Cell function was estimated by the HOMA Calculator. We observed that patients with MetS showedhigher levels of THMs, HOMA-IR and leptin and lower levels of adiponectin and HOMA-b but no differ-ences in F and E in plasma or urine. THM was associated with weight (r = +0.44, p < 0.001), waist circum-ference (r = +0.38, p < 0.01), glycemia (r = +0.37, p < 0.01), and triglycerides (r = +0.18, p = 0.06) andnegatively correlated with adiponectin (r = �0.36, p < 0.001), HOMA-b (r = �0.21, p < 0.001) and HDL(r = �0.29, p < 0.01). In a logistic regression model, THM levels were associated with hypertension, hyper-glycemia and dyslipidemia. We conclude that MetS is associated with increased urinary THMs but notwith F and E levels in plasma or urine. Increased levels of THM, reflecting the daily cortisol productionsubsequently metabolized, are correlated with hypoadiponectinemia, hypertension, dyslipidemia, insulinresistance and b cell dysfunction. A subtle increased in glucocorticoid production may further account forthe phenotypic and biochemical similarities observed in central obesity and Cushing’s syndrome.

� 2011 Elsevier Inc. All rights reserved.

1. Introduction

The worldwide incidence of obesity and metabolic syndrome(MetS) continues to rise at an alarming rate. The WHO estimates

ll rights reserved.

ype 2 diabetes mellitus; HPA,droxysteroid dehydrogenaserometabolites; VAT, visceraleostasis model assessment ofin; LDL, low-density lipopro-ile 25; p75, percentile 75.inology, School of Medicine,oor, Santiago 8330074, Chile.

).

that approximately 1 billion people are obese, and about one-fourth of the world’s adult population suffer from MetS. The com-ponents of MetS (central obesity, hyperglycemia, reduced HDL lev-els, elevated triglycerides and hypertension) are associated withendothelial dysfunction, hypoadiponectinemia, an increased riskof insulin resistance, type 2 diabetes mellitus (T2DM) and cardio-vascular morbidity and mortality [1].

The phenotypic similarities between central obesity, MetS andpatients with endogenous or exogenous glucocorticoid excess sug-gest that cortisol (F) can contribute, at least in part, to the patho-genesis of obesity-related metabolic disorders despite the factthat MetS and obesity are not associated with excess plasma or uri-nary free cortisol [2].

1576 R. Baudrand et al. / Steroids 76 (2011) 1575–1581

Recent studies have shown that glucocorticoid activity dependson both the hypothalamic–pituitary–adrenal (HPA) axis and corti-sol intracellular regulation by 11b-hydroxysteroid dehydrogenasetype 1 and type 2 (11b-HSD1 and 11b-HSD2); 11b-HSD1 is anNADP(H)-dependent reductase that converts inactive cortisone tocortisol, up-regulating glucocorticoid receptor (GR) activation, andis expressed primarily in liver and adipose tissue and also withina pancreatic cells, where it downregulates b cell insulin secretion[3,4]. In contrast, 11b-HSD2 inactivates cortisol, preventing activa-tion of the mineralocorticoid receptor mainly in the kidney andcolon.

Human studies have shown a potential pathogenic role for 11b-HSD1 in central fat distribution, metabolic syndrome, hypertriglyc-eridemia, low HDL and hypertension [5–8]. In addition, 11b-HSD1expression and chronic exposure to glucocorticoids are associatedwith an increased risk of glucose intolerance and T2DM [9,10].Moreover, a hyper-responsive HPA axis leading to slight butinappropriately elevated cortisol production and abnormal HPAsuppression has been described in individuals with central obesity,raising the question of whether MetS and Cushing’s syndrome couldbe part of the same continuum [11].

It has been described decades ago that obesity have increasedcortisol production rate measured by increased 17 hydroxycorticos-teroids [12]. The interpretation is that patients with MetS andcentral obesity may have an increase in local cortisol activationand/or HPA axis dysregulation, leading to this higher cortisol pro-duction [5,11,13]. The classic observation of normal plasma and uri-nary free cortisol levels may be explained by an increasedmetabolism of glucocorticoids, mainly by the hepatic 5a and 5breductases, that converts cortisol (F) and cortisone (E) to inactivetetrahydrometabolites (THMs) such as 5a-tetrahydrocortisol(aTHF), 5b-tetrahydrocortisol (bTHF) and tetrahydrocortisone(THE), as described recently by our group [14]. The urinary THM lev-els, described by Bradlow et al., represent a more accurate reflectionof overall daily cortisol production because subtle or transitoryincreases in cortisol can be counterbalanced by the increased activ-ity and/or overexpression of reductases [14–16].

The aim of this study was to evaluate urine THM levels inpatients with and without MetS and investigate their correlationwith anthropometric and several biochemical parameters such asadipokines, lipid profile, insulin resistance and pancreatic b cellfunction.

Table 1Description of recruited subjects categorized by MetS criteria.

Variable Metabolic syndrome

MetS (+) [n = 130] median (p25–75)

Gender (female) 98 (75.3%)Age (year) 53.1 (46.2–60.4)Hypertension (positive) 108 (83.1%)Body mass index (kg/m2) 29.2 (27.1–32.9)Waist circumference (cm) 100.1 (91.7–107.1)Adiponectin (lg/mL) 11.2 (8.2–16.1)Leptin (ng/mL) 18.3 (9.5–27.5)Glycemia (mg/dL) 95.1 (86.1–102.3)Insulin (lU/mL) 20.4 (14.6–25.1)HOMA-IRa 4.88 (3.3–6.7)%Bb 170.7 (127.8–207.9)%Sc 41.6 (30.8–53.1)HOMA-bd 67.2 (55.5–86.9)HDL (mg/dL) 40.7 (32.1–48.3)Triglycerides (mg/dL) 195.6 (126.1–224.3)

a Insulin resistance was determined by a homeostatic model assessment.b Pancreatic b cell secretion was determined by a homeostatic model assessment.c Insulin sensitivity was determined by a homeostatic model assessment.d Pancreatic b cell function was determined by a homeostatic model assessment.

* Statistically significant.

2. Subjects and measurements

2.1. Subjects

We recruited 221 Hispanic adult subjects from a governmentprimary care center. A standardized questionnaire was adminis-tered to all participants. Demographic data and information aboutcomorbid conditions and the use of medication (especially steroidcontaining pills, creams or injections) were obtained. Height,weight, waist and hip circumference, and blood pressure weremeasured in all cases.

Patients with known or recently detected clinical Cushing’ssyndrome or primary hyperaldosteronism were excluded. For thosepatients with elevated urinary free cortisol (UFF), we performed anovernight low dose dexamethasone suppression test or midnightsalivary cortisol to rule out Cushing’s syndrome. We excludedpatients with recent use of steroid medication, renal failure, livercirrhosis, cardiac failure, alcohol abuse and mood disorders becauseknown changes in cortisol production and metabolism.

The baseline characteristics of the recruited subjects are shownin Table 1.

Patients were classified as having MetS if they met at least threeof the following conditions according to modified ATP III criteria:(a) waist circumference P40 in. for males (102 cm) or P35 in. infemales (88 cm); (b) HDL cholesterol 61.03 mM in males (40 mg/dL), 61.3 mM in females (50 mg/dL) or taking medication forHDL cholesterol; (c) triglycerides P1.7 mM (150 mg/dL) or takingmedication to reduce triglycerides; (d) systolic blood pressure(BP) P130 mm Hg, diastolic BP P85 mm Hg or taking antihyper-tensive medication; and (e) fasting glucose P5.6 mM (100 mg/dL) or taking medication for hyperglycemia [17].

Subjects visited our outpatient clinic after a 12-h fasting period.Blood samples were obtained at 09:00 am. The patients rested for10 min before blood samples were drawn to measure severalhormones, the lipid profile and electrolytes. We collected 24-h ur-ine samples from the 221 subjects, and 50 mL aliquots were storedto measure F, E and THM levels. The serum, plasma and urine sam-ples were stored at �70 �C until analysis.

We subclassified 20 patients as metabolically healthy obese(MHO), defined as having a BMI above 30 kg/m2 in conjunction witheach of the following parameters: HDL cholesterol P1.03 mM(40 mg/dL), triglycerides 61.7 mM (150 mg/dL), fasting glucose

p-Value

MetS (�) [n = 91] median (p25–75)

65 (71.4%) 0.10550.3 (45.1–57.6) 0.33035 (38.5%) <0.001*

27.3 (25.5–29.9) 0.001*

93.7 (87.6–99.5) <0.001*

14.5 (9.7–18.4) <0.001*

10.4 (6.4–16.6) 0.001*

89.4 (83.1–95.3) 0.001*

13.1 (9.8–18.8) <0.001*

2.8 (1.9–3.9) <0.001*

138.3 (107.8–172.7) <0.001*

64.1 (43.8–81.7) 0.003*

85.8 (67.3–105.1) <0.001*

57.6 (48.3–64.7) <0.001*

93.5 (70.7–121.8) <0.001*

R. Baudrand et al. / Steroids 76 (2011) 1575–1581 1577

65.6 mM (100 mg/dL), systolic blood pressure (BP) 6130 mm Hgand diastolic BP 685 mmHg without the patient taking any medi-cation for these conditions [18]. These patients were matchedaccording to age, sex and weight with 62 patients who did not meetthe criteria for MHO and were classified as metabolically unhealthyobese (MUO) to evaluate differences in glucocorticoid and bio-chemical profiles.

This study was approved by our Institutional Review Board forHuman Studies, and all patients provided informed consentaccording to the guidelines of the Declaration of Helsinki.

2.2. Measurements

Serum lipid profile and serum fasting glucose concentrationswere measured using an automated Roche Hitachi Modularchemistry analyzer (Hitachi, Tokyo, Japan). Total serum adiponectinwas analyzed by radioimmunoassay using a commercial kit fromMillipore (Billerica, MA, US), and the intraassay and interassay coef-ficients of variation were 6.5% and 8.9%, respectively. Serum leptinwas measured by radioimmunoassay using a commercial kit fromDiasource (Nivelles, Belgium), and the intraassay and interassaycoefficients of variation were 4.6% and 6.2%, respectively. Fastinginsulin was measured by radioimmunoassay using a Coat-A-Countcommercial kit from Siemens (Siemens Medical Solutions Diagno-sis, Deerfield, IL, USA), and the intraassay and interassay coefficientsof variation were 5.7% and 6.7%, respectively [8]. We evaluatedinsulin resistance (HOMA-IR) and pancreatic b cell function(HOMA-b) with glucose and insulin values obtained by the OxfordUniversity HOMA Calculator 2.2� to estimate b cell secretion (%B)and insulin sensitivity (%S). This model has been validated prospec-tively as a reliable predictor of diabetes [19].

Serum F was measured by immunoassay using automatedequipment (IMMULITE 2000, Siemens Healthcare Diagnostics Inc.,Germany). Serum E was measured by RIA using highly specific anti-serum donated by Dr. Doris Haack, as previously described [20]. Weexcluded serum F and E levels of women using estrogen supple-mentation in our analysis due to the known effects on cortisol bind-ing globulin. For all patients, serum aldosterone (SA) and plasmarenin activity (PRA) were measured, and the ratio SA/PRA (ARR)was calculated. SA was measured by radioimmunoassay using acommercial kit (Diagnosis Products Corp.). PRA was determinedas previously described [8]. In patients who were taking antihyper-tensive drugs that affect the renin angiotensin system, such asb-blockers, angiotensin-converting enzyme (ACE) inhibitors, angio-tensin II receptor blockers, diuretics or spironolactone, a washoutfor at least 15 days was performed.

Table 2Hormone profiles of recruited subjects categorized by MetS criteria.

Variable Metabolic syndrome

MetS (+) [n = 130] median (p25–75)

Aldosterone (ng/dL) 6.06 (3.2–8.1)PRA (ng/mL* h)a 1.10 (0.4–1.4)Plasmatic F (lg/dL) 10.71 (7.6–13.1)Plasmatic E (lg/dL) 3.03 (2.5–3.5)Urinary F (lg/24 h)b 21.13 (11.3–28.1)Urinary E (lg/24 h)b 45.04 (30.2–61.3)bTHF (mg/24 h) 1.31 (0.9–1.7)aTHF (mg/24 h) 1.21 (0.6–1.6)THE (mg/24 h) 2.92 (1.9–3.5)Total THMs (mg/24 h)c 8.35 (6.1–10.3)(aTHF + bTHF)/THE 0.96 (0.7–1.1)aTHF/bTHF 0.95 (0.7–1.0)

a Plasma renin activity.b N = 137 [65 MetS (+), 72 MetS (�)].c Sum of tetrahydrometabolites: bTHF + a-THF + THE + cortol + b-cortol + cortolone + b

* Statistically significant.

We measured free F levels in the urine samples without preser-vatives by immunoassay using automated equipment (IMMULITE2000, Siemens Healthcare Diagnostics Inc., Germany), and theTHMs: aTHF, bTHF, THE, cortol, b-cortol, cortolone and b-cortolonewere measured by gas chromatography–mass spectrometry (GC–MS) using a Hewlett–Packard 7890A gas chromatograph coupledwith a Hewlett–Packard 5975C XL quadrupole mass selective detec-tor (Agilent, Avondale, PA, USA) as described previously [8]. The sumof the cortisol and cortisone metabolite levels (a-THF, bTHF, THE,cortol, b-cortol, cortolone and b-cortolone) was recorded as totalTHMs, as described by Tomlinson and collaborators [21].

In a subgroup of 137 subjects, free cortisol (F) and free cortisone(E) in 24-h urine samples were measured using high-performanceliquid chromatography (HPLC) (Agilent model 1200 with autosam-pler) equipped with an API 3200 QTRAP triple quadrupole massspectrometer (Applied Biosystem, Foster City, CA, USA).

2.3. Statistical analysis

A descriptive analysis was performed. Normality was assessedusing the Shapiro–Wilk test. Demographic, anthropometric andbiochemical characteristics of patients categorized by MetS criteriawere compared using the Mann–Whitney and chi-square tests.Crude and adjusted odds ratios for the association of THMs withMetS were assessed by logistic regression (modality ‘‘enter’’) withgoodness of fit determined by the Hosmer–Lemeshow test. Corre-lations between continuous variables were evaluated using theSpearman coefficient. Statistical significance was set at a p-value<0.05. All analyses were performed using SPSS 15.0.

3. Results

3.1. Clinical and biochemical characteristics categorized by metabolicsyndrome

A total of 221 subjects were analyzed. According to ATPIII criteria,130 patients (58.8%) were categorized as having MetS, and 91(41.2%) subjects were categorized as controls. When comparing pa-tients with MetS to subjects without MetS, we observed that BMI(p = 0.001), waist circumference (p < 0.001), leptin (p = 0.001), gly-cemia, (p < 0.001), HOMA-IR (p < 0.001), %B (p < 0.001) and triglyc-erides (p < 0.001) were significantly higher in the MetS subjectsthan in the control subjects. In contrast, adiponectin (p < 0.001),HDL cholesterol (p < 0.001), %S (p = 0.003) and HOMA-b (p < 0.001)levels were significantly lower in subjects with MetS. Gender, age,

p-Value

MetS (�) [n = 91] median (p25–75)

5.71 (2.9–7.1) 0.2310.92 (0.3–1.1) 0.392

11.41 (8.2–14.4) 0.5602.87 (2.4–3.4) 0.519

24.81 (13.8–31.2) 0.19646.85 (39.8–63.4) 0.837

0.97 (0.6–1.4) 0.016*

0.75 (0.4–1.2) 0.017*

1.71 (1.4–2.7) <0.001*

5.91 (4.2–8.0) <0.001*

0.99 (0.7–1.1) 0.1280.93 (0.7–1.0) 0.602

-cortolone.

Table 3Logistic regression model for the association of glucocorticoid metabolites withmetabolic syndrome and its components.

Variable Total tetrahydrometabolites (THMs)a

Crude OR (CI 95%) Adjusted ORb (CI 95%)

Metabolic syndrome OR 1.14 (1.04–1.25)* OR 1.18 (1.07–1.31)*

Hypertension OR 1.12 (1.04–1.21)* OR 1.12 (1.03–1.22)*

Glycemia >100 mg/dL OR 1.09 (1.01–1.18)* OR 1.03 (0.97–1.13)Triglycerides >150 mg/dL OR 1.10 (1.02–1.19)* OR 1.09 (1.01–1.19)*

HDLc OR 1.07 (1.00–1.16)* OR 1.15 (1.05–1.30)*

Waist circumferenced OR 1.03 (0.96–1.12) OR 1.06 (1.00–1.20)*

a Sum of tetrahydrometabolites: bTHF + a-THF + THE + cortol + b-cortol + corto-lone + b-cortolone.

b Adjusted odds ratio by age, sex and body mass index.c HDL cholesterol 61.03 mM in males (40 mg/dL), 61.3 mM in females (50 mg/

dL).d Waist circumference P40 in. in males (102 cm) or P35 in. in females (88 cm).

* Statistically significant.

1578 R. Baudrand et al. / Steroids 76 (2011) 1575–1581

LDL cholesterol, plasmatic and urinary electrolytes, aldosterone andPRA did not significantly differ between the groups(Tables 1 and 2).

Urinary glucocorticoid metabolite levels, including aTHF, bTHF,THE, cortol, cortolone and THM (the sum of all metabolites), weresignificantly higher in patients with MetS compared to patientswithout MetS (p < 0.001) (Table 2). However, we did not findsignificant differences in plasma or urine levels of F or E betweenpatients with and without MetS. We also did not find differencesin F/E ratio, aTHF/bTHF ratio or aTHF + bTHF/THE ratio betweenthese groups (Table 2). Gender analysis showed no differences inbiochemical parameters.

Using a logistic regression model, we determined that theprobability of developing MetS is positively associated with THMexcretion, increasing 14% for every unit of total THMs (crude OR1.14, CI 95% 1.04–1.25). In the same model, this percentage in-creased to 18% when adjusted for age, sex and body mass index (ad-justed OR 1.18, CI 95% 1.07–1.31). When analyzing the association ofTHMs with individual MetS components, we observed an increasedprobability of developing hypertension according to THM excretionadjusted for age, sex and body mass index (adjusted OR 1.12, CI 95%1.03–1.22). We also observed that the probability of having glyce-mia >100 mg/dL (crude OR 1.09, CI 95% 1.01–1.18), triglycerides>150 mg/dL (adjusted OR 1.09, CI 95% 1.01–1.19), and low HDL (ad-justed OR 1.15, CI 95% 1.05–1.26) increased with THM excretion bythis logistic regression model (Table 3). The Hosmer–Lemeshow testp-value was 0.396, which demonstrates an adequate goodness of fit.

3.2. Anthropometric and biochemical characteristics of recruitedsubjects

When analyzing all subjects (n = 221), we observed a positivecorrelation between total THM excretion and anthropometric

Table 4Correlation of glucocorticoid profile with anthropometric and biochemical parameters.

N = 221 Plasma F (lg/dL) Urine F

Weight (kg) r = �0.34, p = NS r = +0.19Waist circumference (cm) r = +0.07, p = NS r = +0.09HOMA-IRb r = �0.05, p = NS r = +0.09HOMA-bc r = �0.18, p = 0.005* r = +0.21Leptin (ng/mL) r = �0.09, p = NS r = +0.08Adiponectin (lg/mL) r = �0.02, p = NS r = �0.1HDL-C (mg/dL) r = +0.09, p = NS r = �0.0Triglycerides (mg/dL) r = �0.03, p = NS r = �0.1

a Sum of tetrahydrometabolites: bTHF + aTHF + THE + cortol + b-cortol + cortolone + b-b Insulin resistance was determined by a homeostatic model assessment.c Pancreatic b cell function was determined by a homeostatic model assessment.

* Statistically significant; NS = not significant.

variables such as weight (r = +0.44, p < 0.001), BMI (r = +0.28,p < 0.001), and waist circumference (r = +0.38, p < 0.001). We didnot observe a correlation between anthropometric variables andplasma or urine levels of F or E (Table 4).

In addition, the analysis of all recruited subjects showed thatTHM levels were associated with multiple biochemical parameters.We observed a significant inverse correlation of THMs with totaladiponectin (r = �0.37, p < 0.001), but neither F nor E in plasmaor urine correlated with adiponectin. We observed a positive corre-lation between total THMs and glycemia, insulin and triglyceridesas well as an inverse correlation with HDL, but these correlationswere not observed with free F or E (Table 4). When analyzing theassociation of THM levels with insulin resistance parameters, weobserved a positive correlation of total THMs with HOMA-IR(r = +0.23, p < 0.001) and %B (r = +0.14, p = 0.05) and an inverse cor-relation with %S (r = �0.23, p < 0.001) and HOMA-b (r = �0.21,p < 0.001); no correlations were observed with urinary free F or E(Table 4). We found that plasmatic F was inversely associated withHOMA-b but not other insulin resistance parameters (Table 4).

We did not observe a significant correlation of total THMs withleptin, aldosterone, PRA, LDL cholesterol, plasmatic or urinary elec-trolytes. In addition, F/E ratios in plasma or urine were not signif-icantly associated with biochemical parameters, except for acorrelation between plasmatic F and glycemia. We observed nogender-related differences in the anthropometric or biochemicalparameters.

3.3. Clinical and biochemical characteristics categorized by themetabolically healthy obese phenotype

According to the established criteria, we subclassified 20 pa-tients as MHO and compared them with 62 matched patients withan unhealthy phenotype (MUO). As expected, gender, age, BMI andwaist circumference were not significantly different between thegroups (Table 5).

Notably, we observed that HOMA-IR and THMs, variables thatare not included in the definition for healthy obese, were signifi-cantly lower in patients categorized as MHO vs. MUO (HOMA-IR2.4 ± 1.8 vs. 4.7 ± 2.4, p < 0.001; THM 5.82 ± 2.3 mg/24 h vs.8.8 ± 3.1 mg/24 h, p < 0.001).

We observed that patients with the MHO phenotype had lowerlevels of triglycerides (p = 0.035) and higher levels of HDL(p = 0.002) and HOMA-b (p = 0.011). We did not find differencesin leptin, adiponectin, aldosterone, PRA or plasma or urine F/E ra-tios between the groups (Table 5).

4. Discussion

In this study, we report that patients with MetS present in-creased urine excretion of glucocorticoid tetrahydrometabolites,yet they posses normal glucocorticoid levels in plasma and urine.

(lg/24 h) aTHF (mg/24 h) THMa (mg/24 h)

, p = NS r = +0.32, p < 0.001* r = +0.41, p < 0.001*

, p = NS r = +0.26, p < 0.001* r = +0.38, p < 0.001*

, p = NS r = +0.22, p = 0.001* r = +0.23, p = 0.001*

, p = NS r = �0.19, p = 0.005* r = �0.21, p = 0.003*

, p = NS r = �0.08, p = NS r = �0.06, p = NS2, p = NS r = �0.26, p < 0.001* r = �0.36, p < 0.001*

7, p = NS r = �0.20, p = 0.004* r = �0.29, p < 0.001*

0, p = NS r = +0.14, p = 0.05* r = +0.16, p = 0.02*

cortolone.

Table 5Clinical and biochemical characteristics categorized by the metabolically healthy obese phenotype.

Variable Healthy obese (MHOa) [n = 20] median (p25–75) Unhealthy obese (MUOa) [n = 62] median (p25–75) p-Value

Age (year) 52 (45–59) 54 (44–61) 0.872Hypertension (positive) 0% 89% <0.001*

Body mass index (kg/m2) 31.91 (30.7–33.1) 32.20 (30.5–35.0) 0.670Waist circumference (cm) 101.0 (95.5–104.0) 104.5 (98.5–109.1) 0.730HOMA-IRb 2.41 (1.2–3.0) 4.2 (3.1–6.7) <0.001*

HOMA-bc 84.8 (74.5–101.1) 67.0 (50.8–89.6) <0.011*

HDL (mg/dL) 54.5 (49.3–58.0) 47.0 (34.0–52.2) <0.002*

Triglycerides (mg/dL) 108.0 (85.5–145.0) 145.5 (98.2–196.3) <0.035*

Adiponectin (lg/mL) 12.0 (10.1–16.7) 11.2 (8.6–14.9) 0.335Leptin (ng/mL) 19.9 (14.6–29.6) 22.3 (11.6–29.5) 0.731Aldosterone (ng/dL) 7.06 (4.8–9.2) 5.71 (3.2–7.7) 0.131PRA (ng/mL � h)d 1.21 (0.6–1.2) 0.96 (0.3–1.5) 0.172Plasmatic F (lg/dL) 11.71 (8.9–14.1) 11.11 (7.8–13.4) 0.302Total THMs (mg/24 h)e 5.82 (3.7–7.3) 8.69 (6.1–10.9) <0.001*

a MHO: metabolically healthy obese; MUO: metabolically unhealthy obese.b Insulin resistance was determined by a homeostatic model assessment.c Pancreatic b cell function was determined by a homeostatic model assessment.d Plasma renin activity.e Sum of tetrahydrometabolites: bTHF + a-THF + THE + cortol + b-cortol + cortolone + b-cortolone.

* Statistically significant.

R. Baudrand et al. / Steroids 76 (2011) 1575–1581 1579

We found higher THM levels in patients with MetS than in patientswithout MetS, which suggests that an increased production andmetabolism of cortisol is associated with this prevalent clinical con-dition. Notably, the associations between increased THM and MetScomponents such as hypertension, hypertriglyceridemia, low HDLand waist circumference persisted after performing logistic regres-sion adjusting for age, gender and BMI, demonstrating that our re-sults are not only explained by higher cortisol secretion ratesrelated to weight or sex. These data suggest that increased cortisolproduction may explain, at least in part, the similarities betweenobesity-related metabolic disorders and Cushing’s syndrome.

We hypothesize that patients with MetS and central obesitymay have increased cortisol production, which may be counterbal-anced by increased hepatic metabolism by 5a and 5b reductasesthat convert cortisol to inactive metabolites such as aTHF andbTHF. As observed in our cohort, this balance could be a compen-satory phenomenon to avoid excessive glucocorticoid receptoractivation [22]. The action of reductases may explain why subtleor transient hypercortisolism is not frequently reflected in plasmaor urine cortisol levels in MetS patients, and is consistent with ourobservation of increased THM level in MetS subjects. Our group hasrecently published that in morbidly obese patients, hepatic 5a-and 5b-reductases mRNA expressions were positively correlatedwith hepatic 11b-HSD1 mRNA expression suggesting that localhypercortisolism could upregulate the expression of hepatic reduc-tases [14]. Moreover, we observed that cortisone metabolite levelswere also increased in MetS, which explains the similarity in the(aTHF + bTHF)/THE ratio between the groups. Recent studies haveshown that viscera can release a substantial amount of cortisoneinto the portal vein in obese patients, thereby providing the sub-strate for the increased THE found in our patients [23]. In addition,the recently described presence of 11b-HSD2 in adipose tissue,which is elevated in obese compared with lean subjects, could alsoexplain partially our observations [24], although 11b-HSD2 expres-sion in fat is low and its importance controversial.

When analyzing all the recruited subjects, not categorizing byMetS, we also observed that urinary THM levels, but not plasmaticor urinary free cortisol, were significantly correlated with anthro-pometric measurements such as weight, BMI and waist circumfer-ence. These findings suggest that increased cortisol output iscorrelated with central obesity and is subsequently metabolizedto THMs, but cannot differentiate between the contribution of li-ver, visceral and subcutaneous fat. The association of weight andTHMs is also supported by Tomlinson’s data showing that

increased adipose tissue correlated with glucocorticoid secretionrate and that THM levels decreased after weight loss [21].

We also report that THM levels, but not plasmatic or free corti-sol, positively correlated with biochemical parameters, addingmore evidence to the role of cortisol in the pathogenesis of meta-bolic disorders associated with central obesity. Notably, we reportthat THM levels are associated with hypoadiponectinemia. Adipo-nectin, which is produced exclusively by adipocytes, is considereda major pathogenic factor in obesity-related insulin resistance anda predictor of diabetes, hypertension and metabolic syndrome, andits secretion and pulsatility are restored after weight loss [25–27].Although there is little information about glucocorticoids andadiponectin in humans, our results are supported by increasedadiponectin levels after adrenalectomy for Cushing’s syndromeand a recent report showing an association of higher omental fat11b-HSD1 activity with hypoadiponectinemia [28,29]. Addition-ally, murine studies have shown that cortisol decreased adiponec-tin mRNA expression and that pharmacologic inhibition of 11b-HSD1 increased adiponectin concentrations and improved glucosehomeostasis in diabetic mice [30,31].

We found that THM levels were positively associated with fast-ing glucose levels, fasting insulin concentration, pancreatic b cellsecretion and HOMA-IR. A local diabetogenic effect by liver 11b-HSD1 may induce hepatic insulin resistance that may impair thesuppression of liver glucose production by insulin [32]. These sug-gest that THM excretion could be a marker of augmented glucocor-ticoid inactivation used as a protective mechanism to preserveinsulin sensitivity [22]. We also observed that THM levels, butnot urinary free cortisol levels, were inversely associated withinsulin sensitivity and b cell function (HOMA-b). Glucocorticoidsmay cause progressive loss of b cell activity through a pancreaticcells 11b-HSD1 activity and by elevation of triglycerides and fattyacids, which generate lipotoxicity and secondary cell damage[4,10,33]. The findings that THMs correlated with triglyceridesand were inversely associated with HDL are consistent with datafrom transgenic murine models of 11b-HSD1 overexpression andwith the alterations in lipid metabolism described in endogenoushypercortisolism [34,35]. In addition, a previous report postulatedthat higher 11b-HSD1 activity may be associated with increasedlipolysis in omental tissue, increased lipoprotein lipase activityand decreased HDL cholesterol [29].

The observation that metabolically healthy obese patients had alower THM excretion compared with matched unhealthy controlssuggests that the unfavorable metabolic profile in MUO subjects

1580 R. Baudrand et al. / Steroids 76 (2011) 1575–1581

may be due to a dysfunction of adipose tissue rather than itsexcess. The concept of ‘‘healthy fat’’ is characterized by increasedadipogenic capacity, smaller adipocyte size, lower levels of prea-dipocyte differentiation inhibitor (Pref-1) and a more favorableinflammatory profile that may be associated with decreasedcortisol production [36]. The finding of lower THM excretion inMHO patients also supports the role of cortisol production andmetabolism as key pathological processes linking adipose tissuedysfunction in obesity and related metabolic disorders [37]. Inaddition, the importance of localization and function of adipose tis-sue is highlighted by the removal of abdominal subcutaneous fatby liposuction, that do not improve cardiovascular metabolic riskfactors, despite a long-term reduction in body fat [38].

We hypothesize that the phenotypic similarities betweencentral obesity and Cushing’s syndrome could be due to increasedcortisol production and metabolism in local tissues combined witha subtle dysregulation of the HPA axis. The features of metabolicdisorders related to obesity have been investigated in transgenicmurine models by overexpressing 11b-HSD1 in local tissues. Thesemice developed increased visceral fat content, hypertension, hyp-oadiponectinemia, dyslipidemia and insulin resistance, similar tothe metabolic disorders associated with THMs levels in our cohort[34]. Opposite effects were seen in 11b-HSD1 knock-out mice,which showed beneficial effects on their metabolic profiles [39].Subsequent studies in humans have shown an increase of 11b-HSD1 activity in adipose tissue in overweight compared to leansubjects and in patients with type 2 diabetes compared to patientswith normal glucose tolerance [9]. We previously reported that inobese patients, 11b-HSD1 expression was higher in the liver thanin adipose tissue and observed positive correlations of 11b-HSD1expression with dyslipidemia and insulin resistance [32]. In addi-tion, although HPA axis dysregulation in obesity is difficult to as-sess clinically, there is growing evidence that obesity is alsoassociated with impaired negative feedback by endogenous corti-sol [40]. Moreover, overconsumption, sedentary lifestyle, and sleepdeprivation generate a hyperresponsive HPA axis, leading toslightly but inappropriately elevated cortisol production [13].

We must highlight that our results support an increased overallcortisol production rate in MetS that is subsequently metabolizedto THMs, but our methodology cannot determine whether cortisolis produced by increased secretion by adrenals or increasedconversion in liver and/or adipose tissue. Recent publications withstable isotopes have suggested that the liver and not visceral adi-pose tissue accounts for a substantial portion of cortisol conver-sion, adding some controversy regarding splanchnic cortisolproduction in humans [41,42]. It has yet to be determined whethercortisol production is enhanced by (i) a greater supply of cortisonefrom the kidneys and the intraabdominal viscera that increases li-ver conversion to cortisol, (ii) the effect of adipokines secreted toportal circulation in liver cortisol production and/or (iii) the pro-duction of adrenal cortisol due to HPA axis dysregulation [13,23].Using selective 11b-HSD1 inhibitors for specific tissues in patientswith obesity and metabolic syndrome may help to dilucidate thiscontroversy. Another limitation in the interpretation of our resultsis that we could not evaluate 11b-HSD1 mRNA expression in thesepatients. A previous study by our group addressed this issue andfound a significant correlation of hepatic and VAT 11b-HSD1expression with metabolic disorders of obesity [32].

In conclusion, we observed that MetS is associated withincreased urinary THMs but not with F and E levels in plasma orurine. The action of reductases in glucocorticoid metabolism mayexplain why a subtle increase in the overall production rate of glu-cocorticoids, either locally in liver and fat tissue or by HPA axisdysregulation, may only be reflected in urinary metabolites. Theassociation between increased glucocorticoid metabolites andobesity-related metabolic disorders such as hypoadiponectinemia,

hypertension, dyslipidemia, insulin resistance and b cell dysfunc-tion may explain the phenotypic similarities observed in centralobesity and Cushing’s syndrome.

Funding

FONDEF D08I1087, Fondecyt 110036 and Millennium Nucleusin Immunology and Immunotherapy P07/088-F.

Disclosure statement

The authors have nothing to declare.

Author contributions

René Baudrand: Design of the study, lab work; data collection,analysis and interpretation; manuscript writing.

Carmen Campino: Lab work; data collection, analysis and inter-pretation; manuscript writing.

Cristian A. Carvajal: Lab work; data collection, analysis andinterpretation; manuscript writing.

Oliviero Olivier: Data collection, lab work, analysis and inter-pretation; manuscript writing.

Giancesare Guidi: Data collection, lab work, analysis andinterpretation.

Giovanni Faccini: Data collection, lab work, analysis andinterpretation.

Javiera Sateler: Lab work; data collection.Javiera Cornejo: Lab work; data collection.Betty San Martin: Lab work; data collection.Jose M. Dominguez: Data analysis and interpretation; manu-

script writing.Jaime Cerda: Data analysis and interpretation; manuscript

writing.Lorena M. Mosso: Data collection.Gareth I. Owen: Data analysis and interpretation; manuscript

writing.Alexis M. Kalergis: Data analysis and interpretation; manuscript

writing.Carlos Fardella: Design of the study; data analysis and interpre-

tation; manuscript writing.

Acknowledgments

We thank Dr. Doris Haack from the Pharmacology Departmentof Heidelberg University, Germany for providing the antibodyagainst cortisone and Mr. Yester Nuñez from the Laboratory ofPharmacology, Faculty of Veterinary Medicine, University of Chilefor his valuable assistance in the determination of urinary cortisoland cortisone by HPLC–MS/MS.

Cristian A. Carvajal is a fellow of the Comision Nacional deInvestigation Cientifica y Tecnologica de Chile (CONICYT).

References

[1] Athyros VG, Ganotakis ES, Elisaf M, Mikhailidis DP. The prevalence of themetabolic syndrome using the National Cholesterol Educational Program andInternational Diabetes Federation definitions. Curr Med Res Opin2005;21:1157–9.

[2] Tomlinson JW, Stewart PM. Mechanisms of disease: selective inhibition of11beta-hydroxysteroid dehydrogenase type 1 as a novel treatment for themetabolic syndrome. Nat Clin Pract Endocrinol Metab 2005;1:92–9.

[3] Tomlinson JW, Walker EA, Bujalska IJ, Draper N, Lavery GG, Cooper MS, et al.11Beta-hydroxysteroid dehydrogenase type 1: a tissue-specific regulator ofglucocorticoid response. Endocr Rev 2004;25:831–66.

[4] Swali A, Walker EA, Lavery GG, Tomlinson JW, Stewart PM. 11Beta-hydroxysteroid dehydrogenase type 1 regulates insulin and glucagonsecretion in pancreatic islets. Diabetologia 2008;51:2003–11.

R. Baudrand et al. / Steroids 76 (2011) 1575–1581 1581

[5] Andrew R, Phillips DI, Walker BR. Obesity and gender influence cortisolsecretion and metabolism in man. J Clin Endocrinol Metab 1998;83:1806–9.

[6] Walker BR. Cortisol – cause and cure for metabolic syndrome? Diabet Med2006;23:1281–8.

[7] Anagnostis P, Athyros VG, Tziomalos K, Karagiannis A, Mikhailidis DP. Clinicalreview: the pathogenetic role of cortisol in the metabolic syndrome: ahypothesis. J Clin Endocrinol Metab 2009;94:2692–701.

[8] Campino C, Carvajal CA, Cornejo J, Martin BS, Olivieri O, Guidi G, et al. 11Beta-hydroxysteroid dehydrogenase type-2 and type-1 (11beta-HSD2 and 11beta-HSD1) and 5beta-reductase activities in the pathogenia of essentialhypertension. Endocrine 2009;37:106–14.

[9] Alberti L, Girola A, Gilardini L, Conti A, Cattaldo S, Micheletto G, et al. Type 2diabetes and metabolic syndrome are associated with increased expression of11beta-hydroxysteroid dehydrogenase 1 in obese subjects. Int J Obes (Lond)2007;31:1826–31.

[10] van Raalte DH, Ouwens DM, Diamant M. Novel insights into glucocorticoid-mediated diabetogenic effects: towards expansion of therapeutic options? EurJ Clin Invest 2009;39:81–93.

[11] Krikorian A, Khan M. Is metabolic syndrome a mild form of Cushing’ssyndrome? Rev Endocr Metab Disord 2010;11:141–5.

[12] Bradlow HL, Zumoff B, Fukushima DK, Hellman L, Gallagher TF.Biotransformations of the C-20-dihydro metabolites of cortisol in man. J ClinEndocrinol Metab 1972;34:997–1002.

[13] Bose M, Olivan B, Laferrere B. Stress and obesity: the role of the hypothalamic–pituitary–adrenal axis in metabolic disease. Curr Opin Endocrinol DiabetesObes 2009;16:340–6.

[14] Baudrand R, Dominguez JM, Carvajal CA, Riquelme A, Campino C, MacchiavelloS, et al. Overexpression of hepatic 5alpha-reductase and 11beta-hydroxysteroid dehydrogenase type 1 in visceral adipose tissue is associatedwith hyperinsulinemia in morbidly obese patients. Metabolism. doi:10.1016/j.metabol.2011.05.001.

[15] Zumoff B, Monder C, Bradlow HL. Studies in the biotransformation of cortisolto the cortoic acids in man. II. The central role of tetrahydrocortisol andtetrahydrocortisone as intermediates. J Clin Endocrinol Metab1977;44:647–50.

[16] Bradlow HL, Zumoff B, Gallagher TF, Hellman L. Tetrahydrocortisol metabolismin man. Steroids 1968;12(3):303–8.

[17] Grundy SM, Cleeman JI, Daniels SR, Donato KA, Eckel RH, Franklin BA, et al.Diagnosis and management of the metabolic syndrome: an American HeartAssociation/National Heart, Lung, and Blood Institute Scientific Statement.Circulation 2005;112:2735–52.

[18] Aguilar-Salinas CA, Garcia EG, Robles L, Riano D, Ruiz-Gomez DG, Garcia-UlloaAC, et al. High adiponectin concentrations are associated with themetabolically healthy obese phenotype. J Clin Endocrinol Metab2008;93:4075–9.

[19] Song Y, Manson JE, Tinker L, Howard BV, Kuller LH, Nathan L, et al. Insulinsensitivity and insulin secretion determined by homeostasis modelassessment and risk of diabetes in a multiethnic cohort of women: theWomen’s Health Initiative Observational Study. Diabetes Care2007;30:1747–52.

[20] Krall P, Carvajal C, Ortiz E, Munoz C, Garrido JL, Mosso L, et al. Urinary freecortisol is not a biochemical marker of hypertension. Am J Hypertens2007;20:459–65.

[21] Tomlinson JW, Finney J, Hughes BA, Hughes SV, Stewart PM. Reducedglucocorticoid production rate, decreased 5alpha-reductase activity, andadipose tissue insulin sensitization after weight loss. Diabetes2008;57:1536–43.

[22] Tomlinson JW, Finney J, Gay C, Hughes BA, Hughes SV, Stewart PM. Impairedglucose tolerance and insulin resistance are associated with increased adipose11beta-hydroxysteroid dehydrogenase type 1 expression and elevated hepatic5alpha-reductase activity. Diabetes 2008;57:2652–60.

[23] Basu R, Basu A, Grudzien M, Jung P, Jacobson P, Johnson M, et al. Liver is thesite of splanchnic cortisol production in obese nondiabetic humans. Diabetes2009;58:39–45.

[24] Mussig K, Remer T, Haupt A, Gallwitz B, Fritsche A, Haring HU, et al. 11Beta-hydroxysteroid dehydrogenase 2 activity is elevated in severe obesity andnegatively associated with insulin sensitivity. Obesity (Silver Spring)2008;16:1256–60.

[25] Yun JE, Sull JW, Lee HY, Park E, Kim S, Jo J, et al. Serum adiponectin as a usefulmarker for metabolic syndrome in type 2 diabetic patients. Diabetes MetabRes Rev 2009;25:259–65.

[26] Shankar A, Marshall S, Li J. The association between plasma adiponectin leveland hypertension. Acta Cardiol 2008;63:160–5.

[27] Calvani M, Scarfone A, Granato L, Mora EV, Nanni G, Castagneto M, et al.Restoration of adiponectin pulsatility in severely obese subjects after weightloss. Diabetes 2004;53:939–47.

[28] Ashizawa N, Takagi M, Seto S, Suzuki S, Yano K. Serum adiponectin and leptinin a patient with Cushing’s syndrome before and after adrenalectomy. InternMed 2007;46:383–5.

[29] Veilleux A, Rheaume C, Daris M, Luu-The V, Tchernof A. Omental adipose tissuetype 1 11 beta-hydroxysteroid dehydrogenase oxoreductase activity, body fatdistribution, and metabolic alterations in women. J Clin Endocrinol Metab2009;94:3550–7.

[30] Shi JH, Du WH, Liu XY, Fan YP, Hu XL, Zhou HY, et al. Glucocorticoids decreaseserum adiponectin level and WAT adiponectin mRNA expression in rats.Steroids 2010;75:853–8.

[31] Sundbom M, Kaiser C, Bjorkstrand E, Castro VM, Larsson C, Selen G, et al.Inhibition of 11betaHSD1 with the S-phenylethylaminothiazolone BVT116429increases adiponectin concentrations and improves glucose homeostasis indiabetic KKAy mice. BMC Pharmacol 2008;8:3.

[32] Baudrand R, Carvajal CA, Riquelme A, Morales M, Solis N, Pizarro M, et al.Overexpression of 11beta-hydroxysteroid dehydrogenase type 1 in hepaticand visceral adipose tissue is associated with metabolic disorders in morbidlyobese patients. Obes Surg 2010;20:77–83.

[33] Davani B, Portwood N, Bryzgalova G, Reimer MK, Heiden T, Ostenson CG, et al.Aged transgenic mice with increased glucocorticoid sensitivity in pancreaticbeta-cells develop diabetes. Diabetes 2004;53(Suppl. 1):S51–9.

[34] Masuzaki H, Paterson J, Shinyama H, Morton NM, Mullins JJ, Seckl JR, et al. Atransgenic model of visceral obesity and the metabolic syndrome. Science2001;294:2166–70.

[35] Friedman TC, Mastorakos G, Newman TD, Mullen NM, Horton EG, Costello R,et al. Carbohydrate and lipid metabolism in endogenous hypercortisolism:shared features with metabolic syndrome X and NIDDM. Endocr J1996;43:645–55.

[36] O’Connell J, Lynch L, Hogan A, Cawood TJ, O’Shea D. Preadipocyte factor-1 isassociated with metabolic profile in severe obesity. J Clin Endocrinol Metab2011;96:E680–4.

[37] Karelis AD, Faraj M, Bastard JP, St.-Pierre DH, Brochu M, Prud’homme D, et al.The metabolically healthy but obese individual presents a favorableinflammation profile. J Clin Endocrinol Metab 2005;90:4145–50.

[38] Mohammed BS, Cohen S, Reeds D, Young VL, Klein S. Long-term effects oflarge-volume liposuction on metabolic risk factors for coronary heart disease.Obesity (Silver Spring) 2008;16:2648–51.

[39] Morton NM, Paterson JM, Masuzaki H, Holmes MC, Staels B, Fievet C, et al.Novel adipose tissue-mediated resistance to diet-induced visceral obesity in11 beta-hydroxysteroid dehydrogenase type 1-deficient mice. Diabetes2004;53:931–8.

[40] Mattsson C, Reynolds RM, Simonyte K, Olsson T, Walker BR. Combinedreceptor antagonist stimulation of the hypothalamic–pituitary–adrenal axistest identifies impaired negative feedback sensitivity to cortisol in obese men.J Clin Endocrinol Metab 2009;94:1347–52.

[41] Basu R, Singh RJ, Basu A, Chittilapilly EG, Johnson CM, Toffolo G, et al.Splanchnic cortisol production occurs in humans: evidence for conversion ofcortisone to cortisol via the 11-beta hydroxysteroid dehydrogenase (11beta-hsd) type 1 pathway. Diabetes 2004;53:2051–9.

[42] Stimson RH, Andersson J, Andrew R, Redhead DN, Karpe F, Hayes PC, et al.Cortisol release from adipose tissue by 11beta-hydroxysteroid dehydrogenasetype 1 in humans. Diabetes 2009;58:46–53.