In men, abdominal obesity has usually been associated with low plasma testosterone levels in cross-sectional (123) as well as in longitudinal studies (124, 125). Waist circumference and waist-to-hip ratio are also inversely associated with plasma sex hormone-binding globulin (SHBG) levels (123-125). Many studies that have measured abdominal fat areas using imaging techniques such as computed tomography or magnetic resonance have confirmed that low plasma testosterone concentrations are often found with elevated visceral fat accumulation (126-131). The fact that both SHBG and total testosterone are reduced in abdominally obese men (132) and methodological limitations in free-testosterone measurement (133, 134) have made it difficult to detect obesity-related differences in free androgen levels. Nevertheless, androgen treatment in hypogonadal men generally leads to a decrease in abdominal fat accumulation (135). These effects appear to be dose-dependent (136) and lead to concomitant improvements of glucose-insulin homeostasis (135, 137) while having neutral effects on the lipid profile (138). These effects are observed when the androgen levels reached during treatment remain within the physiological range (106). Supraphysiological androgen treatment has a different set of effects including increased visceral fat accumulation and insulin resistance [e.g., in female-to-male transsexuals (139, 140)] and dramatic alterations of the lipid profile [e.g., anabolic steroid users (141)]. Thus, when examining physiologically relevant data in men, low circulating androgen levels are associated with increased abdominal adiposity, and restoration of physiological levels leads to reduced abdominal fat (106).
In women, the association between circulating androgens and abdominal obesity is more complex. In contrast to men, it is generally thought that abdominal obese women are hyperandrogenic (132).
This belief is largely based on the observation that women with the polycystic ovary syndrome (PCOS) show hyperandrogenism often associated with abdominal obesity and hyperinsulinemia (142). Interesting developments in our understanding of the pathophysiology of PCOS indicate that the elevated androgens of these women are not the cause of abdominal obesity, but the result of a hyperinsulinemic/insulin resistant state often associated with excess visceral fat (142). Indeed, insulin-sensitizing treatments such as diazoxide, metformin, troglitazone (143-145), or weight loss (146, 147) all lead to reduced androgen secretion through improved insulinemia in PCOS women. A specific sensitivity of ovarian steroidogenesis to insulin is thought to be present in PCOS women and is maintained even in insulin-resistant states (142). Outside PCOS-related hyperandrogenism, however, the association between circulating androgen levels and abdominal obesity in women remains unclear. Some reported high plasma testosterone (total or free) in women with visceral obesity (148, 149), but few studies are available and results are not always consistent (150,151). Some studies actually found negative associations between plasma testosterone levels and visceral fat accumulation (152-154). Further studies are needed to clarify the link between endogenous androgens and visceral obesity in non-PCOS women.
Significant androgen binding in adipose tissue (and adipocytes) has long been established (155-161). These findings are supported by our own analyses of fat tissue steroid content and androgen-receptor expression (162, 163). Numerous studies investigated the effects of androgens on adipocyte/adipose tissue function using various models. We recently performed a detailed review of these studies (106). Overall, the most consistent effect of androgens observed on fat cell function is a stimulation of lipolysis that is receptor-dependent and may affect the lipolysis cascade at different levels. These data are concordant with observations in androgen-receptor null mice, which show a late-onset obesity phenotype likely attributable to impaired lipolysis (164). Additional data seem to indicate that testosterone inhibits adipose tissue LPL activity in human fat cells (165), consistent with an inhibitory effect of androgens on lipid accumulation.
In rats (166) or in the 3T3-L1 cell line (167), androgens (testosterone and DHT) were found to inhibit preadipocyte differentiation. Castration was found to increase differentiation in preadipocytes from the perirenal fat depot in rats (168). However, castration was also found to inhibit differentiation of epididymal preadipocytes (169). In addition, it was reported in human fat cells that testosterone had no effect, in any region, on glycerol-3-phosphate dehydrogenase activity, a late marker of differentiation (170). Androgen responsiveness was found to be more pronounced in deep fat depots (visceral) in comparison to subcutaneous adipose compartments in other studies (161,166,168,171). Thus, available data seem to suggest that androgens inhibit fat cell differentiation, although further studies are needed to clearly establish this effect in humans.
Previous results from our group (163, 172-177) have suggested that prereceptor modulation of androgen action by aldo-keto reductases from the 1C family (AKR1C) may be related to abdominal obesity. Our original studies suggesting increased circulating androgen metabolite levels in abdominal obese males (172-174) have recently been confirmed in a large cohort study by a Swedish group (178). Our work has also shown that the conversion of dihydrotestosterone, the most potent natural androgen, to the inactive androgen metabolite 5a-androstane-3a,17b-diol (3a-diol) was detected in the fat tissue of both men and women (163, 176, 177). Activity was higher in subcutaneous fat than in omental fat, and, most importantly, androgen inactivation rates in omental fat were positively correlated with measures of obesity including BMI, fat cell size, and visceral adipose tissue area assessed by computed tomography (163, 176, 177). The enzyme responsible for most of the dihydrotestosterone-to-3a-diol conversion in humans is AKR1C2 or 3a-HSD-3. Despite having rather distinct substrate specificities (179) compared with other AKR1C enzymes of the same family such as AKR1C1 (20a-HSD) and AKR1C3 (170-HSD-5) (179-182), these enzymes are highly homologous (e.g., 97.8% amino acid identity between 1C1 and 1C2), suggesting that they have diverged very recently in evolution (181).
We have shown that omental AKR1C enzyme mRNA expression in adipose tissue was specifically and positively associated with visceral adipose tissue area in women (176). Depot differences were similar to those observed with activity measures, and significant positive correlations were found between AKR1C mRNA expression and androgen inactivation rates (176). We suggest that increased androgen inactivation by AKR1C enzymes, especially AKR1C2, may lead to reduced local exposure of fat cells to androgens, which could remove some of the inhibitory effects of this hormone on adipocyte differentiation. Additional experiments will further establish the importance of prereceptor androgen metabolism in the etiology of visceral obesity.
Was this article helpful?