Bibian van der Voorn

149 GENDER-SPECIFIC HPA AXIS ACTIVITY IN CHILDHOOD and irreversibly by α- and β-ring reductases, and CYP3A. Animal studies showed a lower bioavailability of glucocorticoids in females due to decreased 11βHSD type 1 16–18 and relatively increased 11βHSD type 2 activity 18 , as compared to males. In addition, previous observations in humans suggest that estrogens could alter hepatic cortisol metabolism through increased CYP3A activity 19, 20 , and decreased A-ring reduction 3, 21 . In contrast, sex-specificity in the activities of 11βHSD isozymes is debated in humans 3,21,22 . Since analyses of gender-specific differences in total serum cortisol were inconclusive in random-effects models ( Appendix 3 ) and only one of the included studied had assessed CBG levels next to cortisol, we cannot exclude a gender-specific influence of CBG 4 on the serum cortisol level. The HPA axis set point can be modified through an altered balance between mineralocorticoid and glucocorticoid receptor expression 23 . Animal studies have suggested that patterns in receptor expression develop in a gender-specific manner from birth onwards 24 . In humans, behavioral patterns that impact a child’s stress vulnerability have been associated with gender-specific changes in cortisol levels from age 1.5 yr onwards 11,25 . Therefore, even in our sample of normal children, gender-specific effects of stress exposure could be an explanation for our results 9 . Even subtle disturbances inHPA axis activity have been associatedwith cardiovascular disease and its risk factors 26–28 . Cardiovascular disease susceptibility is gender- specific 7, 29 , which has been suggested to be due to gender differences in HPA axis activity, stress vulnerability and responsivity 4,30–32 . Early in life, developmental plasticity offers the child the capacity to change his HPA axis set point based on stress experiences 9,33 . This ability offers opportunities to withstand early-life challenges, but it has also been suggested to affect disease risk later in life. Accordingly, although the gender differences found in our study were small, these patterns might contribute to gender-specific origins of health and disease 9 . The major strength of this study is our systematic approach and the effort to contact all authors of eligible publications, enabling us to include the data on 16,551 healthy children. Moreover, articles with a lack of quantitative data were included in our descriptive analysis with the aim to be as complete as possible. The large sample size enabled us to perform a sensitivity analysis, which decreased the heterogeneity between studies. Furthermore, we accounted for this heterogeneity by calculating standardized mean differences, based on the intervention effects relative to the variability observed 13 . Additionally, we chose fixed-effect meta-analysis, because the studies with a large sample size were most likely conducted with greater