Bibian van der Voorn

148 CHAPTER 10 0.01, I 2 = 8%) higher in boys between ages 8-18 yr. The sensitivity analyses did not significantly change the results, although it decreased the heterogeneity: boys <8 yr had 0.40 (0.11 to 0.69) nmol/L ( P < 0.01, I 2 = 55%) higher salivary, 0.45 (0.30 to 0.61) nmol/L ( P < 0.01, I 2 = 94%) higher serum and 0.28 (-0.04 to 0.61) µg/24h ( P = 0.08, I 2 = 33%) higher 24h-urine cortisol; boys 8–18 yr had 0.20 (0.13 to 0.26) nmol/L ( P < 0.01, I 2 = 47%) lower salivary, 0.10 (0.02 to 0.18) nmol/L ( P = 0.01, I 2 = 33%) lower serum and 0.24 (0.02 to 0.47) µg/24h ( P = 0.04, I 2 = 24%) higher 24h-urine cortisol. Appendix 3 shows the results of the random-effects meta-analyses. When analyzed by the random-effects method, the effect estimates of serum cortisol <8 yr and between 8-18 yr became non-significant ( P = 0.46 and P = 0.62, respectively). This also applied to salivary cortisol <8 yr ( P = 0.06) and urinary cortisol <8yr ( P = 0.12), although trends in the same direction were observed. Funnel plots showed no evidence of publication bias. ( Appendix 4 ) DISCUSSION The results from this meta-analysis suggest that gender-specific differences in HPA axis activity are already present early in life. They also support previous observations which show that cortisol metabolism diverges between genders at pubertal age. Before age 8 yr, cortisol in both serum and saliva was higher in boys compared to girls, at least in fixed-effect meta-analysis. These patterns were reversed after age 8 yr. In contrast, gender differences in 24h-urine cortisol remained consistent with age, with higher cortisol levels in urine for boys before and after age 8 yr. Total serum cortisol and free salivary cortisol reflect the balance between cortisol production and degradation, i.e., the bioavailability. Our meta-analysis suggests that puberty induces gender-specific changes in the bioavailability of cortisol, as reflected by similar changes in both total serum and free salivary cortisol levels, at least in fixed-effect models. Even though associations were absent for total serum cortisol in random-effects models, the change in free salivary cortisol could not be explained by an estrogen-induced increase in the production of CBG 4 . Moreover, the gender difference in cortisol in 24hr-urine (i.e., non-metabolized, free cortisol, representing cortisol production rate) remained consistent with age. Consequently, sex-hormone dependent effects on the hepatic metabolism of cortisol are more likely to explain our observations. Cortisol is metabolized reversibly by 11βHSD type 2,