Bibian van der Voorn

79 MILK GLUCOCORTICOID DIURNAL RHYTHMICITY collected with a Salivette (Sarstedt). Participants were instructed to rinse their mouth with tap water to remove potential food residues, before saliva collection at home, which they did ≥30 min after brushing their teeth. FIGURE 1. Study design of milk and saliva collections over 24h by healthy mothers who delivered at term. Mothers were asked to collect samples at each feeding occasion, every 3h on average ( ). In addition, we asked them to collect 2 extra samples in between the morning feedings ( ). Mothers breastfed their children on demand; therefore times shown in the figure are only representative (Study 2). DETERMINATION OF CORTISOL AND CORTISONE CONCENTRATIONS IN BREAST MILK AND SALIVA We used 0.5 mL of mother’s milk to assess cortisol and cortisone concentrations with the use of an isotope dilution LC-MS/MS as described previously 24 . In short, internal standards ( 2 H 4 -labeled cortisol and 2 H 8 -labeled cortisone) were added to the samples. Lipids were removed by washing the milk 3 times with 2 mL hexane. Samples were extracted and analyzed by online solid phase extraction coupled to LC-MS/MS (XLC- tandem MS [Spark Holland, Emmen, Netherlands]), coupled to a Quattro Premier XE tandem MS (Waters Corporation). The lower limit of quantitation was 0.5 nmol/L for cortisol and 0.3 nmol/L for cortisone. The intra-assay CVs were 4% and 5%at cortisol concentrations of 7 and 23 nmol/L and 5% at cortisone concentrations of 8 and 33 nmol/L, respectively. The interassay CVs were <9% for cortisol and cortisone. Cortisol and cortisone concentrations in saliva were determined with the same method as that for breast milk, but without the hexane-washing procedure. DATA ANALYSIS Continuous data were compared by using Mann-Whitney U tests, and dichotomous data were compared by using Fisher’s exact tests. Generalized estimating equations (GEEs) were used for analyses of longitudinal data. GEEs adjust for grouped samples, collected from the same participant at different times, by using a correlation structure. For our data analyses, we chose an exchangeable correlation structure, in which 1 average within subject correlation between samples over time is assumed. Cortisol and cortisone concentrations were skewed to the right and logarithmically transformed before analysis. Therefore, GEE results are presented as exponentiated βs (95% CIs). P < 0.05 was defined as being significant.