Adipose tissue 11βHSD1 gene expression, βcell function and ectopic fat in obese African Americans versus Hispanics


  • Lauren E. Gyllenhammer,

    Corresponding author
    1. Department of Preventive Medicine, Diabetes and Obesity Research Institute, University of Southern California, Los Angeles, California, USA
    Search for more papers by this author
  • Tanya L. Alderete,

    1. Department of Preventive Medicine, Diabetes and Obesity Research Institute, University of Southern California, Los Angeles, California, USA
    Search for more papers by this author
  • Swapna Mahurka,

    1. Department of Preventive Medicine, Diabetes and Obesity Research Institute, University of Southern California, Los Angeles, California, USA
    Search for more papers by this author
  • Hooman Allayee,

    1. Department of Preventive Medicine, Diabetes and Obesity Research Institute, University of Southern California, Los Angeles, California, USA
    Search for more papers by this author
  • Michael I. Goran

    1. Department of Preventive Medicine, Diabetes and Obesity Research Institute, University of Southern California, Los Angeles, California, USA
    Search for more papers by this author

  • Funding agencies: Clinical trial reg. no. NCT00697580, This work was supported by the following: The Robert C. and Veronica Atkins Foundation Grant, and USC Graduate Provost Fellowship. The content is solely the responsibility of the authors and does not necessarily represent the official views of the sponsoring agencies.

  • Disclosure: The authors have no competing interests to report.

  • Author contributions: L.E.G. designed the study, analyzed data, wrote the manuscript, and reviewed and edited the manuscript. S.M., H.A., M.I.G. reviewed and edited the manuscript. T.L.A. contributed to data and sample collection and reviewed and edited the manuscript.



This study examined the contribution of subcutaneous adipose tissue (SAT) 11βHSD1 to obese African Americans' (AA) elevated metabolic risk, despite a protective obesity phenotype of reduced visceral adipose tissue (VAT) and hepatic fat fraction (HFF) relative to obese Hispanics with similar metabolic risk.

Design and Methods

Obese AA and Hispanic adults (N = 36(16AA); BMI 35.2 ± 0.6 kg/m2, 18-25y) participated, with VAT, SAT, and HFF measured by MRI, SAT gene expression measured by HT-12 microarray and insulin sensitivity (SI), disposition index (DI) by IVGTT. Multiple linear regression examined relationships/interactions of ethnicity and 11βHSD1 expression on outcomes (covariates: age, sex, total fat mass), with standardized β (stβ) reported.


SAT 11βHSD1 expression significantly associated with insulin parameters and this varied by ethnicity (Pinteraction<0.1). In AA, 11βHSD1 negatively associated with SI (stβ = -0.58, P = 0.03), DI (stβ = −0.62, P = 0.03) and positively associated with fasting insulin (stβ = 0.54, P = 0.04), with no significant relationship in Hispanics. SAT 11βHSD1 associated with HFF in the combined sample (stβ = 0.42, P = 0.008), with no difference between ethnicites (Pinteraction>0.1). After controlling for HFF, 11βHSD1 associations with metabolic risk in AA became nonsignificant.


These results suggested that in AA and not Hispanics, SAT 11βHSD1 is associated with SI and DI, and may be mediated by HFF.


The accumulation of body fat, particularly visceral adipose tissue (VAT) and hepatic fat fraction (HFF), is hypothesized to be associated with increased risk for type 2 diabetes (T2D) and cardiometabolic disease. This relationship, however, does not hold up when examined across ethnicities. African Americans (AA) and Hispanics are both at increased risk for obesity and its co-morbidities, but AA have markedly lower VAT and HFF compared to Hispanics but are not protected from insulin resistance and disease, thus establishing the “African American-Hispanic paradox” [1]. AA additionally have the highest levels of subcutaneous adipose tissue (SAT), but the quantity of this depot has not shown to significantly alter their T2DM risk. This paper will take a two-step approach to explore this paradox. First, we propose that while the quantity of SAT may not associate with T2DM risk, the functional activity of a greater SAT depot could still play an important role in AA. Secondly, while VAT and HFF accumulation is lower in AA, when elevations do occur, particularly in hepatic fat, they may be more metabolically potent. We have previously shown this [2], and therefore propose to examine how the functional activity of SAT associates with hepatic fat accumulation.

There is a large body of evidence in animal and human studies supporting the role of the enzyme 11β-hydroxysteroid dehydrogenase type-1 (11βHSD1) in adipose tissue expansion, glycemic disturbance and cardiometabolic disease [3-8]. 11βHSD1 activates glucocorticoids within the adipocyte, producing a local elevation in cortisol, a catabolic and adipogenic hormone. Exploring novel mechanisms that predict insulin parameters and hepatic fat accumulation in AA and Hispanics are crucial in tackling the gap in suitable clinical care for these populations. Given the central role of 11βHSD1 in potentially linking adipose tissue metabolism to ectopic fat deposition and the difference between AA and Hispanics in SAT, we sought to explore the association of SAT 11βHSD1 expression with insulin parameters [insulin sensitivity (SI), disposition index (DI)] and HFF in obese AA and Hispanic young adults. Our goal was to explore 11βHSD1 as a novel mechanism to explain the AA-Hispanic paradox; hypothesizing that SAT 11βHSD1 expression will more strongly associate with these parameters in AA than in Hispanics.


This cross-sectional study included 36 obese (BMI≥30 kg/m2) non-T2DM AA (n = 16) and Hispanic (n = 20) young adults, aged 18-25 years. All participants signed informed consents and the IRB of USC approved this study. All participant data and methods description have previously been described in detail [9].

DEXA was used to quantify total body fat and lean tissue, and MRI was used to quantify abdominal fat depots: SAT and VAT volume, and HFF. SAT biopsies were performed via punch biopsy and mRNA isolated as previously reported [9]. mRNA expression from SAT was processed in Genome Studio and the signal intensities were background corrected and quantile normalized. An insulin modified frequently samples intravenous glucose tolerance test (IVGTT) was performed after an overnight fast, [10]. Glucose and insulin values obtained from the IVGTT were entered into the MINMOD Millenium 2003 program (version 5.16) to determine SI, acute insulin response (AIR), and DI, and fasting blood was used to determine basal insulin, glucose, and free fatty acids (FFA). Heart rate and blood pressure were measured in triplicate from a sitting position and the mean is reported.

Statistical methods

The data were assessed for normality, and the following transformations were applied to achieve normal distribution for non-normal variables: log transformation for AIR and fasting insulin; inverse transformation for HFF (i.e. 1/HFF). Participant characteristics are presented as means ± standard error (SE) and the unadjusted comparisons between ethnicities were computed by student t tests or chi-square (sex). Linear regression was used to examine how SAT 11βHSD1 related to insulin parameters (SI, DI, AIR) and abdominal fat depots (SAT, VAT, and HFF), with standardized betas (stβ) used to display relationships. Next, we examined how these relationships varied by ethnicity by examining the interaction between 11βHSD1 and ethnicity, and added the interaction parameter (11βHSD1*ethnicity) to existing models. If there was a significant interaction, then subsequent regressions were stratified by ethnicity, otherwise the regression for the combined sample is reported. Furthermore, since 11βHSD1 is known to associate with abdominal fat depots and metabolic traits, if SAT 11βHSD1 expression significantly associated with any abdominal fat depots, the depot was tested for mediation using the Baron and Kenny method [11]. The depot was then included as a covariate in any models showing significant associations with metabolic traits, and updated results were reported. A priori covariates included: age, sex, total fat mass, and ethnicity (if ethnicities combined). All analyses were performed using SPSS version 16.0, (SPSS, Chicago, IL), with P < 0.05 considered statistically significant, and P < 0.1 used for interaction parameters.


Table 1 shows the clinical characteristics of the participants combined and stratified by ethnicity. Combined, participants had a mean ± SE age of 21.5 ± 0.4 years and a BMI of 35.2 ± 0.6 kg/m2. Across ethnicities, the participants did not significantly differ by age, sex, BMI, body fat percent or insulin parameters, but as previously shown, AA had significantly less VAT and HFF [9].

Table 1. Clinical characteristics
 Combined (n = 36)African American (n = 16)Latino (n = 20)P-value
  1. Mean ± SE; * P < 0.05, ** P < 0.001; BMI, body mass index; SAT, subcutaneous adipose tissue; VAT, visceral adipose tissue; HFF, hepatic fat fraction; AIR, acute insulin response; SI, insulin sensitivity; DI, disposition index; FFA, free fatty acids.

Age (yrs)21.5 ± 0.421.6 ± 0.521.4 ± 0.50.89
Sex (men/women)(13/23)(5/11)(8/12)0.73
BMI (kg/m2)35.2 ± 0.636.2 ± 1.134.4 ± 0.70.18
Body fat (%)37.8 ± 0.938.1 ± 1.237.6 ± 1.30.80
Total fat mass (kg)37.0 ± 1.438.2 ± 2.336.0 ± 1.70.42
SAT (L)16.2 ± 0.716.8 ± 1.015.7 ± 0.80.41
VAT (L)2.9 ± 0.31.9 ± 0.33.8 ± 0.3<0.001**
HFF (%)7.6 ± 1.05.4 ± 1.39.4 ± 1.50.05*
Fasting glucose (mg/dL)89.9 ± 1.489.6 ± 2.690.2 ± 1.30.87
Fasting insulin (mU/mL)14.2 ± 1.613.4 ± 2.414.9 ± 2.30.67
AIR (mU/mL × 10 min)1160 ± 1381331 ± 2761023 ± 1140.28
SI (× 10-4 min-1/mU/mL)1.8 ± 0.11.7 ± 0.21.9 ± 0.20.51
DI (× 10-4 min-1)1759 ± 1421782 ± 2471740 ± 1680.87
FFA (mmol/L)0.85 ± 0.030.84 ± 0.040.85 ± 0.040.81

SAT 11βHSD1 and metabolic traits

The mean level of SAT expression of 11βHSD1 did not differ between ethnicities (t = -1.6, P = 0.9), however, the association with insulin parameters did differ by ethnicity. Specifically, there was a significant interaction between 11βHSD1 and ethnicity in association with SI and DI (Pinteraction < 0.05). When stratified by ethnicity, 11βHSD1 only significantly associated with lower SI (stβ = -0.58, P = 0.026) and DI (stβ = -0.62, P = 0.031) in AA, with no significant association in Hispanic participants (SI: stβ = 0.04, P = 0.86; DI: stβ = 0.02, P = 0.93) (Figure 1). The relationships between 11βHSD1 and fasting insulin and glucose levels showed a trend for differing by ethnicity (Pinteraction < 0.1). When stratified by ethnicity, 11βHSD1 was only significantly associated with higher fasting insulin (stβ = 0.54, P = 0.044) in AA, with no relationship in Hispanics (stβ = 0.14, P = 0.56) and no significant association with glucose in either ethnicity. There was no significant association with AIR, FFA, heart rate, or blood pressure in the combined or stratified samples.

Figure 1.

SAT 11βHSD1 expression associates with insulin sensitivity (P = 0.026) and disposition index (P = 0.031) in African Americans. Model is adjusted for gender, age, and total body fat.

SAT 11βHSD1 and abdominal fat deposition

There was no significant difference between ethnicities in the association between SAT 11βHSD1 expression and the quantity of any abdominal depot (Pinteraction >0.1). However, in both AA and Hispanics 11βHSD1 expression significantly associated with higher HFF (stβ = 0.42, P = 0.008) (Figure 2), showed a trend association with VAT (stβ = 0.21, P = 0.073), and no significant association with SAT. Finally, we investigated whether the association between 11βHSD1 and the insulin parameters in AA remained once HFF was included in the model, but all of the associations became nonsignificant (SI: stβ = −0.306, P-value = 0.326; DI: stβ = −0.574, P-value = 0.140; fasting insulin: stβ = 0.620, P-value = 0.095), suggesting potential mediation by HFF.

Figure 2.

SAT 11βHSD1 expression associates with HFF (P = 0.008) in both ethnicities. Model is adjusted for gender, ethnicity, age, and total body fat.


Our data demonstrates that within obese AA young adults, the level of 11βHSD1 expression in SAT is linked to T2D risk through negative associations with SI and DI, and positively with fasting insulin levels. These significant relationships did not exist in obese Hispanic participants. Additionally, SAT 11βHSD1 expression was positively associated with HFF in both ethnicities. Given that the relationship of SAT 11βHSD1 expression with insulin parameters became nonsignificant with the addition of HFF to the regression models, it is possible that the level of hepatic fat accumulation may be an important mechanism through which 11βHSD1 expression in SAT leads to systemic metabolic effects.

Previous studies in humans have shown that 11βHSD1 in SAT and VAT are positively correlated with adiposity and metabolic risk [5-8]. In our study, the obese AA and Hispanic participants were matched for total body fat but had marked differences in VAT and HFF. Despite these differences, AA and Hispanics were similar in regards to metabolic risk (fasting glucose and insulin, SI, DI, and AIR). This lends further evidence towards changing our focus from fat depot quantity driven analysis towards exploration of adipose function as a means of understanding the African American-Hispanic paradox. Our data show that the activity of 11βHSD1 in SAT may be an important functional pathway that requires future exploration in this population, particularly in association with HFF. Sixty percent of the lipids in steatotic livers are derived from adipose tissue lipolysis [12], and the majority of the fatty acids delivered to the liver through portal circulation are derived from upper body adipose tissue (which includes SAT) [13]. Increased 11βHSD1 expression in SAT associated with HFF in both ethnicities and this may be explained by increased lipolytic FFA turn-over and increased hepatic exposure. SAT 11βHSD1 expression associations with metabolic parameters (SI, DI, insulin) and potential mediation by HFF only occurred in AA, which may be an indication of the metabolic potency of HFF accumulation in AA. Our group [2] has shown that while HFF is significantly lower in AA, when elevations do occur, they show greater impact upon beta-cell function and SI. Furthermore, if the steatosis has more potent metabolic implications in AA, this may explain their atypical metabolic profile [1]. Hepatic specific insulin resistance, modeled by genetic ablation of hepatic insulin receptor leads to hyperinsulinemia, hyperglycemia, and peripheral insulin resistance, without concurrent increases in hepatic fat accumulation or release of triglycerides [14]; paralleling the profile exhibited in AA. Future studies should assess the relationship between adipose 11βHSD1 expression, hepatic steatosis, and hepatic specific insulin resistance in AA.

11βHSD1 increases local concentrations of cortisol in adipose tissue. Cortisol can enhance FFA production, alter insulin action and has been shown to serve an immune modulatory role [15]. Cortisol may work to cause insulin-resistant adipocytes to further expel FFA into circulation, leading to ectopic fat deposition [16], such as increased HFF. We, however, did not see an association between SAT 11βHSD1 expression and circulating FFA levels. This may be because of potential differences in cortisol activity in adipocytes because of diurnal patterns or co-exposure with insulin or differences in FFA turnover not detected through circulating concentrations. While exposure to cortisol causes insulin resistance at the whole body level, there is recent evidence that co-exposure of cortisol (or dexamethasone) and insulin in adipocytes causes insulin sensitization [17, 18]. However, during the evening or periods with low exposure to insulin, cortisol still induces adipocyte lipolysis and increased FFA release [17, 19], which is of importance as nocturnal FFA, and not morning fasting levels, have been shown to be a key signal for the development of systemic insulin resistance [20]. We measured SAT 11βHSD1 and FFA in the morning at fasting, and there are potential nocturnal interactions that should be explored in future studies. Furthermore, we did not measure VAT 11βHSD1 expression and VAT exposure to cortisol does not sensitize insulin activity, and thus the unmitigated lipolysis of VAT may still be important [17].

Given that this study is cross-sectional, with a relatively small number of subjects, we are unable to determine causality. However, 11βHSD1 has been manipulated in animal models through knockout and over-expression studies [3, 4], and human models of increased cortisol exposure (Cushing syndrome and pharmaceutical glucocorticoids) do support the purported direction of association in our study. Additionally, we quantified 11βHSD1 levels through gene expression methods and did not measure 11βHSD1 protein levels or enzyme activity. Despite these limitations, it is important to note that a previous study has shown that in humans 11βHSD1 mRNA levels significantly associate with the enzyme activity [5]. This study was conducted in obese healthy young adults and it is unknown if these results generalize to other populations. Overall, our study is strengthened by the novel hypothesis, which we put forward in an effort to understand the paradoxical health risk seen in AA. Our preliminary data are first step in exploring pathways underlying potential adipose specific dysfunction in AA.

In conclusion, higher expression of 11βHSD1 in SAT is associated with metabolic risk in AA, but not Hispanics. These results suggest that further work should be done to explore the 11βHSD1 and cortisol pathway in AA, with an emphasis on adipocyte specific dysfunction rather than adipose quantity driven analysis. Potential mediation by hepatic accumulation and hepatic specific insulin resistance should also be explored. Furthermore, 11βHSD1-inhibition should be explored as a potential therapeutic target that might be more effective in AA than Hispanics, particularly nocturnal inhibition.


We would like to thank all the Childhood Obesity Research Core (CORC) research team, as well as the nursing staff at the CTU. In addition, we are grateful for our study participants for their involvement.