Salivary cortisol level was used as a proxy measure of adrenal, pituitary and hypothalamic function. Due to its liposoluble nature, unbound cortisol can easily enter nucleated cells via passive diffusion. Unbound cortisol appears in all bodily fluids including blood, cerebral spinal fluid, sweat, urine, semen, and saliva. Unbound cortisol enters saliva primarily via the acinar cells of the parotid, submandibular and sublingual salivary glands. However, the concentration of cortisol in saliva is independent of salivary flow rate (Gatti et al., 2009). A strong positive correlation between levels of unbound cortisol in saliva and plasma (approximately 80% of total variance [r ≥ .90]) has been reported (Arafah, 2006; Gozansky, Lynn, Laudenslager, & Kohrt, 2005; Kirschbaum & Hellhammer, 1989; Vining, McGinley, Maksvytis, & Ho, 1983). Salivary analysis is therefore considered a physiologically relevant and useful measure of the unbound fraction of cortisol. Salivary cortisol is highly correlated with total cortisol levels in blood (Kirschbaum & Hellhammer, 1989, 1994). However, absolute cortisol levels in saliva are lower due to the metabolising action of enzyme 11β-hydroxysteroid dehydrogenase type 2 (Van Uum et al., 2002).
The use of saliva to measure cortisol was considered to confer a number of advantages over alternative sampling methods. Collecting blood samples is invasive and associated
50
with stress-induced adrenal activation that would likely confound endocrine responses to stress protocols (Granger et al., 2007). Measurement of cortisol in urine is a useful method for measuring 24 hr aggregated basal activity but is not suitable for assessing rapid changes in cortisol response. Saliva for assay can be easily collected repeatedly over relatively short time intervals using non-invasive sampling methods.
All saliva samples were collected using a Salivette® saliva collection device - a widely utilised and validated method (Arafah, Nishiyama, Tlaygeh, & Hejal, 2007; Dorn, Lucke, Loucks, & Berga, 2007; Gatti et al., 2009; Gozansky et al., 2005; Hellhammer, Wust, & Kudielka, 2009). Cortisol assayed from Salivettes® has also been shown to be a better predictor of bound and unbound plasma cortisol compared to passive drool methods (Poll et al., 2007). In the studies presented in this thesis, samples were collected via the chewing of a roll-shaped synthetic saliva collector swab stored in a sample tube. Participants were asked to place the saliva collector swab directly into their mouths from the Salivette® tube and chew the swab gently for at least one minute to ensure adequate saliva absorption. The swab was then returned directly from the mouth into the Salivette® tube. Saliva collection was performed at least 1 hour after consumption of meals and drinking caffeine/acidic drinks or brushing teeth to ensure no contamination of the saliva by interfering substances (see Appendix 1 for cortisol collection standard operating procedures). The number and frequency of samples collected from participants was informed by prior research demonstrating typical acute cortisol responsivity to stress. Cortisol concentrations following acute stress peak 21 – 40 minutes after stressor onset (Dickerson & Kemeny, 2004) with a gradual return to baseline levels after approximately one hour (Kirschbaum & Foley, 2010). All studies outlined in the thesis collected at least one baseline measure prior to stress exposure and further samples at approximately 10 minutes intervals until approximately 40 – 50 minutes after stress onset to adequately capture the cortisol response trajectory.
Due to the diurnal variation in cortisol secretion, time of day is an important methodological consideration when sampling this steroid. Firstly, testing early in the morning can interfere with, and be confounded by, the CAR. Secondly, HPA diurnal activity follows a pronounced circadian rhythm characterised by an early morning peak of secretory bursts of cortisol, and decreasing bursts over the afternoon. Consequently, basal cortisol levels vary as a function of time of day. The pattern and net increase in cortisol response to acute psychosocial stress has been shown to be comparable when testing occurs between the hours of 0945 – 1900 hrs (Kudielka, Schommer, Hellhammer, & Kirschbaum, 2004). However, higher baseline pre-stress salivary cortisol
51
levels in the morning contribute to significantly higher morning area under the curve (AUC). Higher baseline cortisol levels have also been negatively associated with lower net increases in acute stress suggesting higher baseline levels may provide less “space” for an effect of stress (Decherney et al., 1985; Hermus, Pieters, Smals, Benraad, & Kloppenborg, 1984; Kudielka, Schommer et al., 2004; Schurmeyer et al., 1987). Moreover, Dickerson and Kemeny’s (2004) meta-analytic review reported that time of day significantly predicted cortisol response effect sizes (afternoon, d = 0.46; morning, d = 0.14).
Considering the potential for the influence of divergent baseline endogenous levels to affect measures of cortisol response, testing in the studies presented in the thesis was confined to the afternoon. All salivary cortisol measures across the reported studies were collected after 1200 hrs (with the exception of four participants tested at 1145 hrs in Study 1 due to constraints of the naturalistic testing context) and before 1720 hrs. In studies requiring participants to attend repeated experimental visits, the repeat visiting times were matched to within 1 hr of the first visit.
3.2.1.1
Assay of salivary cortisol
Saliva was extracted from cotton wool swabs by centrifugation (2500 rpm, five minutes) and frozen at - 20°C until assay. Salivary-free cortisol concentrations were determined using a commercially available Salimetrics Salivary Cortisol Enzyme Immunoassay kit (EIA; Sarstedt; Nümbrecht, Germany). The Salimetrics EIA kit is a competitive immunoassay with a high sensitivity (< 0.007 ug/dL) specifically designed and validated for the quantitative measurement of salivary cortisol. The assay uses a microtitre plate coated with monoclonal antibodies to cortisol. Cortisol in standards and unknowns compete with cortisol linked to horseradish peroxidase for antibody binding sites. Once unbound components are washed away, the reaction of peroxidise enzyme on the substrate tetramethylbenzidine (which produces a blue colour) is used to measure bound cortisol peroxidase. Sulphuric acid is used to stop the reaction, forming a yellow colour. The level of cortisol peroxidase is indicated by the intensity of colour which is inversely proportional to cortisol level (Chard, 1990).
Intra- and inter-assay coefficients of variability were below 9.5 and 12.5% respectively across all studies (below the respective < 10 and < 15% levels recommended by the assay kit manufacturer). Specific coefficients of variability values are reported in the method sections of each respective study.
52
3.2.1.2
Aggregated measures of cortisol
A number of aggregated indices of cortisol response were included in the statistical analyses across all studies. The delta increase in cortisol response was calculated by subtracting the baseline cortisol value from the peak post-stress induction level. Area under the curve formulae were used to calculate aggregated measures of cortisol response. Two formulae for calculating the area under the curve using the trapezoid method were employed (Pruessner et al., 2003): area under the curve with respect to ground (AUCg), and area under the curve with respect to increase (AUCi). In relation to endocrinological data AUCg is considered to give an indication of total hormonal output independent of changes over time, and AUCi indexes response change over time (Pruessner et al., 2003). The trapezoid method uses the measurement and time distance between measurements to calculate AUC values. Two formulae for each aggregated AUC measure were used dependent upon whether the time differences between measurement sampling points were equal or unequal. The following formulae were employed:
Equation 1.1 Formula for the calculation of AUCg with equal time differences between sampling time points
53
Equation 1.2 Formula for the calculation of AUCi with equal time differences between sampling time points
Equation 1.3 Formula for the calculation of AUCg with unequal time differences between sampling time points
Equation 1.4 Formula for the calculation of AUCi with unequal time differences between sampling time points