The use of saliva for assessment of cortisol pulsatile secretion by deconvolution analysis

Cortisol is the key effector molecule of the HPA axis and is secreted in a pulsatile manner in all species studied. In order to understand cortisol signalling in health and disease, detailed analysis of hormone pulsatility is necessary. To dissect cortisol pulsatility in plasma deconvolution techniques have been applied. Blood sampling is a labour-intensive, expensive and invasive technique that causes stress and alters HPA axis activity. Therefore saliva has been extensively investigated as an alternative sample to measure cortisol. Here we use state of the art deconvolution algorithms to investigate cortisol pulsatility in saliva. Blood and saliva samples were obtained at 15-min intervals over an 8h period in 18 healthy men to analyse their diurnal cortisol levels. A multiparameter deconvolution technique was used to generate statistically significant models of cortisol secretion and elimination in plasma and saliva. The models consisted of estimates of the number, amplitude, duration and frequency of secretory bursts as well as the elimination half-life (t1/2) in a subject specific manner. No significant differences were noted between plasma and saliva with regard to the observed secretory bursts (7.8±1.5 vs. 7.0±1.4) and the interpeak interval (59.6±10.5 min vs. 61.0±11.5 min). Moreover a strong positive correlation between the numbers of peaks in both fluids was observed (r=0.83, P<0.0001). Monte Carlo simulations revealed an 84% temporal concordance between plasma and saliva peaks in all donors (P<0.05) with a mean of 1.3±0.8 plasma peaks unmatched in saliva. The percentage concordance increased to 90% when concording only the morning cortisol peaks in plasma and saliva up to 11:00 h. The deconvolution of the most distinct component of cortisol diurnal rhythm-cortisol awakening response (CAR), revealed an average 2.5±1.1 peaks based on the individual time for cortisol to return to baseline levels. In conclusion, deconvolution analysis of plasma and salivary cortisol concentration time series showed a close correlation and similar pulsatile characteristics between saliva and plasma cortisol. Similarly, Monte Carlo simulations revealed a high concordance between the peaks in these coupled time series suggesting that saliva is a suitable medium for subsequent deconvolution analysis yielding accurate and reliable models of cortisol secretion in particular during the morning hours.