A series of studies in adults and a few studies in selected groups of adolescents, have shown alterations in cortisol in individuals with cardiovascular risk factors (for review see [8, 9, 13]). In adults, abdominal obesity, high triglyceride and low HDL-cholesterol levels, hypertension, hyperglycemia, MetS and chronic stress have all been characterized by hyperactivity of the HPA axis leading to a functional hypercortisolism. It has also been suggested that inhibiting cortisol action could provide a novel approach for these conditions .
In the present study, although ACTH and cortisol levels were within the normal ranges, we observed higher ACTH and cortisol levels in obese children and adolescents with specific cardiovascular risk factors. In particular, ACTH levels were higher in those with higher glucose and triglyceride levels, while cortisol levels were higher in those with hypertension and higher LDL-cholesterol, thereby increasing the risk for these metabolic disturbances.
The first aim of our study was to determine whether cortisol and ACTH were associated with cardiovascular risk factors in obese Caucasian children and adolescents. We showed that ACTH and cortisol were directly associated with glucose, triglycerides, and BP independently of sex, age, puberty, BMI and insulin resistance. Moreover, cortisol was also associated with LDL-cholesterol. These data suggest that the link between HPA and comorbidities in obesity is present in very young children and that elevated ACTH and cortisol levels, although within normal ranges, are already associated with cardiovascular risk factors. The data regarding cortisol are in agreement with the study of Weigensberg and coworkers who demonstrated that cortisol is higher in obese Latino youths with MetS, independent of the degree of obesity and insulin sensitivity . Many studies evaluating cortisol in pediatric obesity have demonstrated an association between cortisol and insulin resistance, leading to hypothesis that a relationship between cortisol and metabolic disturbances would be mediated by insulin sensitivity [9, 11, 13]. However, both our data and those from Latino youths suggest that the relationship is complex and not only due to insulin resistance. It is well known that Hispanic people have a higher prevalence of type 2 diabetes and cardiovascular diseases [20, 23–25], thus such a selected sample of subjects could have quite a different phenotype linked to their genetic susceptibility. Despite this, however, our data and that of others have shown that the association with cardiovascular risk factors remains positive in non-specific populations of obese children and adolescents [9, 13].
The lack of association between BMI and cortisol was unexpected, particularly because an association was present for ACTH. However, a number of studies have failed to show an association [12, 26–28]. Similar to the lack of an association between BMI and plasma cortisol in the obese population, a lack of association between BMI or body fat levels and urinary free cortisol and free cortisone (in 24-h urine) has also been demonstrated in non-obese children . One explanation could be the homogenous population in terms of weight in our and other studies. However, it could be that ACTH is a better biomarker in childhood in relation to obesity and associated cardiovascular risk factors. Accordingly, major glucocorticoid metabolites in 24-h urine samples (reflecting ACTH-driven adrenocortical activity or cortisol secretion) were significantly associated with body fat in non-obese children . The latter findings, together with our cross-sectional study, suggest that adrenocortical activity (driven by ACTH) is related to body composition during growth whether children are lean/normal weight  or obese. However, in this context it is important that higher body fat levels do not always imply higher total blood cortisol although ACTH is increased. ACTH can be increased and total circulating cortisol concomitantly reduced . However, also in children reduced cortisol blood levels are not uncommon in case of elevated body fat levels [31, 32]. Moreover, our paper and that of Reinehr and Andler  indicate that children's total cortisol plasma levels are not necessarily reduced if body fat is higher. However, really elevated cortisol concentrations in obese children appear to emerge only if a marked insulin resistance is also present . Interestingly, we also observed changes in ACTH and cortisol levels at the last years and at the end of puberty. Healthy adults have higher blood cortisol levels than children. Adults also have clearly higher 24-h excretion rates of free cortisol and free cortisone after correction for body surface area than children. However 24-h excretion of free cortisol and free cortisone in healthy children up to an age of about 14 years is constant after body surface area correction [29, 33]. We showed in our obese cohort that cortisol rose in Tanner 4 to 5 stage participants, particularly in females. Conversely, ACTH was higher in males with respect to females in the same pubertal stage and at the age of 14.0 to 15.9 years in presence of still unmodified cortisol levels. These data, with respect to what age ACTH and cortisol start to increase, have to be considered in future works on cortisol in obese adolescents.
We also demonstrated that ACTH levels were associated with metabolic alterations in pediatric obesity, and that some associations were stronger with respect to those of cortisol, in particular insulin resistance. The characteristics of this association suggest that higher ACTH levels could better reflect the interplay between obesity and the HPA axis, and that cortisol-binding globulin (CBG) may be important. Higher CBG levels reduce the rate of cortisol clearance, and thus reflect cortisol levels in plasma . In a large population study, CBG levels were negatively correlated with BMI, BP and insulin resistance, perhaps indicating suppression of CBG synthesis, or a CBG gene polymorphism in obesity . Since stress-induced cortisol pulses are elevated in obesity, lower CBG levels may enhance the glucocorticoid action on tissues and also increase the cortisol clearance. Because CBG is costained with ACTH in corticotrophs and is colocalized with vasopressin in the hypothalamus , lower CBG levels may result in higher ACTH levels in obesity by regulating the HPA stress response. However, ACTH levels also represent expression of the negative feedback loop formed by corticotropin-releasing hormone and cortisol, which is influenced by genetic differences in the glucocorticoid receptor .
The highest tertiles of the normal ranges of ACTH and cortisol levels in this study were associated with an increased risk of higher triglyceride and LDL-cholesterol levels, respectively. The specific associations for ACTH and cortisol were of interest. The association between cortisol and LDL-cholesterol could be a consequence of multifactorial mechanisms, including direct and indirect effects on lipolysis, free fatty acid production and turnover, and very-low-density lipoprotein synthesis and fatty acid accumulation in the liver (for review see [8, 37]). The association between ACTH and triglycerides may be secondary to the strong association between ACTH and insulin resistance in our study. Moreover, ACTH has been shown to increase apolipoprotein E levels in humans, a key protein in determining triglyceride metabolism . However, a higher ACTH-driven adrenocortical activity could have consequences for the hepatic fat and triglyceride metabolism likely through a higher hepatic glucocorticoid metabolism [8, 37].
Higher ACTH levels, although within the normal range, were also associated with fasting and post-challenge glucose, and ACTH levels were strong predictors of hyperglycemia. These data are in line with findings in obese Latino youths [11, 12], suggesting that changes in the HPA in altered glucose conditions are present also in a broader population. The relationship between glucose and cortisol is in line with glucocorticoid effects on hepatic gluconeogenesis, insulin secretion and resistance [8, 39]. However, only ACTH increased the risk of high glucose levels. These findings are concordant with the evidence that lower daily cortisol levels and normal CBG concentrations have been shown in childhood obesity as an age-dependent mechanism to prevent type 2 diabetes . Because ACTH-driven adrenocortical activity or cortisol secretion has been shown associated with body fat in childhood, as previously discussed , ACTH might also associate with higher glucose levels before overt disease. In agreement with this hypothesis is that none of the subjects in the present study had overt type 2 diabetes, but they had impaired fasting glucose or glucose tolerance. We also found higher cortisol levels, although within the normal range, in subjects with higher BP, reflecting the data in Latino youths . It is interesting to note that when a cut-off was imposed, cortisol levels were significantly higher only with respect to the 95th percentile of BP, whereas there was no significant association with the 90th percentile, which is suggested as pathological in the MetS definition , and is prehypertensive according to the NHBPEP Working Group definition . The lack of an association between cortisol and hypertension using the lower BP cut-off suggests that cortisol levels may be increased only in overt disease. HPA alterations are thus likely to be a consequence of obesity comorbidities, as is also suggested by normalization of cortisol levels after weight reduction .
There are limitations in the present study. First is the cross-sectional design, in which we could not determine whether slightly higher ACTH and cortisol levels were a consequence rather than a cause of cardiovascular risk factors in pediatric obesity. Longitudinal studies might clarify this aspect. The second limitation was the inability to define the length of exposure to HPA alterations. The third limitation was evaluation of the HPA without the evaluation of urinary free cortisol. It is difficult to collect daily urine samples properly in pediatric cases, in particular in younger children. In fact, urinary samples were incomplete in most of the children who also took the dexamethasone test for exclusion of Cushing syndrome. However, a single morning fasting cortisol measurement has been shown to be associated with chronic stress and metabolic disturbances . The fourth limitation was the lack of precise data on socioeconomic status due to the refusal of many parents. Socioeconomic status has been found to affect chronic stress and cortisol levels and its role needs to be explored further. The fifth limitation was the absence of true body fat measurements through radiological techniques. However, BMI is a good surrogate for body fat in obesity in large epidemiological sample sizes . The final limitation was the lack of a control group. It would, however, be difficult to choose a good control group for our purpose. Our population was followed in a tertiary care center, and a healthy population of schoolchildren would not be completely comparable in terms of chronic stress. Conversely, the strength of the study was the large sample size, the measurement of post-challenge glucose levels, and the evaluation of many confounding factors.