Circadian Rhythm of Salivary Cortisol, 17α-Hydroxyprogesterone, and Progesterone in Healthy Children

Abstract

Very few reference intervals for salivary steroids in children have been established to date (1). Even the manufacturers of salivary steroid assays do not provide sufficient reference data for their products. This lack of information is surprising because the measurement of salivary steroids has been accepted as being noninvasive and stress-free (2)(3). In particular, psychiatric and neuro-endocrinologic experiments are frequently designed with saliva as the medium of choice for steroid analysis (4)(5)(6). A large variety of stressors can rapidly affect the adrenal cortex, causing increased adrenal steroid concentrations. For example, hypoglycemia (7) or physical exercise (8) are potent physiologic stressors, whereas fear (9), feelings of inferiority (10)(11), or experiences at school (12)(13) can affect the adrenal cortex activity as psychologic stressors. The taking of blood can also influence adrenal steroid concentrations in children; saliva collection, however, is almost stress-free (14). The use of saliva for steroid analysis in children is therefore an excellent alternative to blood. The aim of our study was to establish age-dependent reference values for salivary cortisol, 17α-hydroxyprogesterone (17OHP), and progesterone in a large cohort of healthy children. The availability of such reference intervals will improve the applicability of saliva analysis as a diagnostic tool in pediatric endocrinology. We collected 252 saliva profiles from healthy children and adolescents (125 boys; age range, 4 days to 15 years; 127 girls; age range, 6 days to 13 years) with normal body length/height and weight. None of the girls had developed a regular menstrual cycle. The parents of the children gave informed consent. Saliva was collected either with the Salivette®, using polyester swabs (Sarstedt), from children >1 year of age or with modified medical pacifiers (Büttner-Frank; see Fig. 1 in the Data Supplement that accompanies the online version of this Technical Brief at http://www.clinchem.org/content/vol49/issue10/) from infants <1 year. The teats of the pacifiers were perforated, and strips of filter paper, as used for blood-spot collection, were inserted into the teats. The absorbent material had no effect on the steroid analysis in the RIAs, in contrast to the cotton Salivette, particularly for 17OHP (15) and progesterone (16). Samples were taken three times a day (0700, 1300, and 1900) at home. After sample collection, the saliva was immediately frozen and sent to the laboratory in a cooler. We tried to exclude any potential stressful events; thus the samples were taken only on so-called “quiet” days at home, e. g., days with no school-related stress or sports, or on weekends. All of the children participating in the study lived in an intact familial environment. We used adapted commercial RIAs (DSL), as described in detail elsewhere (17)(18), that require 50 μL of saliva per tube. The sensitivities and intra- and interassay CVs were 0.2 nmol/L (6% and 9.6%) for cortisol, 9 pmol/L (6% and 8.6%) for 17OHP, and 9 pmol/L (5.1% and 7.7%) for progesterone. Mean recoveries of analytes that had been added to saliva samples were 89–94% for the three steroids. The samples were measured in duplicate without extraction. The conditions during preanalysis guaranteed maximum stability of the samples (19). Variations in salivary steroid values in relation to age, gender, and time of day were tested using ANOVA (Bonferroni-adjusted post hoc test). For descriptive and statistical analysis, the children were divided into age groups of <4 weeks, 1–12 months, and 1–2, 2–4, 5–7, 8–10, 11–13, and 14–15 years. For graphical presentation, percentiles were calculated by transformation of the age-dependent data according to the recommendations of the IFCC (20). An appropriate (square-root or ln) transformation was used to adjust data to a gaussian distribution. The appropriate transformation was identified by an iterative process for each data set. Percentiles (3rd and 97th) were determined from the mean and the SD of the transformed data and then transformed back to the original scale. Polynomial regression of the percentiles on age was then performed using PRISM software (GraphPad Inc.). Several figures and a detailed table are presented in the online Data Supplement that accompanies the online version of this Technical Brief at http://www.clinchem.org/content/vol49/issue10/. We observed no gender-related differences in the salivary steroid concentrations in the different age groups included in this study (P = 0.45). The data for boys and girls were therefore combined. There were no significant differences between the age groups >1 year of age (P >0.05) for any of the three steroids. Percentiles for these data sets were calculated without further division into age groups according to: Percentile = SD ∗ 1.96 ± Mean (20). A detailed summary of the steroid concentrations is presented in Table 1 . Basic statistics for salivary cortisol, 17OHP, and progesterone in healthy children and adolescents in the morning, at noon, and in the evening. Basic statistics for salivary cortisol, 17OHP, and progesterone in healthy children and adolescents in the morning, at noon, and in the evening. In children <1 month of age, we found no circadian rhythm, whereas children between 1 and 12 months were developing a circadian rhythm. The morning values differed significantly (P <0.001) from the noon and evening values. There was a further decrease in noon and evening values in children >12 months. These concentrations differed significantly from the morning values (P <0.001). Regression analysis yielded no significant relationship of the morning values with age, but noon as well as evening values showed a highly significant correlation (P <0.001; for 1200, r2 = 0.70; for 1900, r2 = 0.81; Fig. 1 ). Nonlinear regression analysis of salivary cortisol concentrations vs age. The correlation was significant (P <0.001) for the noon (r2 = 0.70) and evening values (r2 = 0.81). In newborns (<4 weeks), salivary 17OHP values were significantly higher compared with the other age groups at all times of day. In this age group, evening values were already significantly lower than in the morning and at noon (P <0.001). The circadian rhythm was established in children >12 months, with morning values being twice as high as noon values (P <0.05) and three times as high as in the evening (P <0.001). In children between 2 and 4 years of age, the circadian rhythm was fully developed (P <0.001). Regression analysis yielded a significant correlation (P <0.001; Fig. 2 in the online Data Supplement) between 17OHP concentrations and age in the morning (r2 = 0.63), at noon (r2 = 0.79), and in the evening (r2 = 0.79). We found no significant diurnal changes in progesterone concentration in healthy newborns. Salivary progesterone was highest (P <0.001) in this age group compared with the other groups. In children 1–12 months of age, progesterone was decreased to one-third of the newborn values. The circadian rhythm was fully developed in children older than 2 years. In pubertal adolescents, however, salivary progesterone concentrations were spread over a broader range. Regression analysis yielded a significant correlation (P <0.001; Fig. 3 in the online Data Supplement) between progesterone values and age in the morning (r2 = 0.42), at noon (r2 = 0.60), and in the evening (r2 = 0.66). Steroid measurements in saliva samples collected several times a day present an excellent alternative to using blood samples in children. To establish reference intervals for salivary steroids, especially for cortisol, it is important that samples are not taken under the influence of potential stressors. This aspect is even more important for saliva than for plasma, where the bound and unbound fractions of the steroid are measured. In saliva, concentrations reflect the unbound, physiologically active fraction of the steroid (2)(14). We therefore collected saliva samples in our study at home during so-called “stress-free” days. The main advantage is that saliva samples can be easily obtained several times from the same individual at short time intervals (21). Many clinical applications for salivary steroid measurements at different times of the day have been developed, e.g., monitoring the adequacy of glucocorticoid therapy in congenital adrenal hyperplasia (21)(22)(23) or diagnosing patients with glucocorticoid excess (24)(25)(26). In general, three or more saliva samples are collected at fixed times. To analyze the results correctly, however, knowledge of the physiologic course of salivary steroid values during the day is a prerequisite. This study underlines the potential usefulness of salivary profiles for an estimation of the onset of the circadian rhythm of adrenal steroid synthesis and confirms the results reported in previous studies (27)(28). The primary aim of our study was not, however, to examine the onset of the appearance of a circadian rhythm in infants, despite the fact that this topic still appears to be controversial in the literature. Some publications describe cortisol rhythmicity as early as the 2nd week of life (27), whereas others found no circadian rhythm in children before the 9th month (29). None of our participants showed a circadian rhythm before the age of 1 month. Our results therefore support the published data of Antonini et al. (28) and Castro et al. (24). We found only a weak correlation between morning salivary 17OHP and progesterone values and age. This correlation became stronger at noon and in the evening for all three steroids measured. All morning salivary steroid concentrations showed a high variability. Surprisingly, we found no age dependency at all for cortisol morning values, whereas 17OHP and progesterone showed age dependency at this time of day. Our data for salivary progesterone concentrations in children are the first reported in the literature. Salivary progesterone is widely accepted as a means of assessing ovarian function in obstetrics (30)(31)(32). The high progesterone values in the age group <1 month may be related to high plasma progesterone concentrations derived from the materno-placental circulation. The slight increase in adolescents can be related to the increasing gonadal activity, although the pubertal girls included in this study were still premenarcheal. Ovarian activity certainly precedes the development of an endometrial environment leading to measurable salivary progesterone concentrations in girls before the onset of menarche. Moreover, morning progesterone and 17OHP values were much lower in adolescent girls than the mean values measured with the same assays during the luteal phase in young women with regular menstrual cycles (31). The salivary cortisol concentrations that we obtained with a commercial method are in good accordance with data reported in reference studies and data from healthy controls from various clinical studies using in-house methods. Our mean data for cortisol and 17OHP also fit with previously reported data in neonates (33), although the variation in our values was smaller. One study (33) reported wide ranges for cortisol (up to 166 nmol/L) and 17OHP (>5160 pmol/L), and it is possible that the extreme morning values for cortisol and 17OHP may partially reflect a stress response to the Salivette used. From our own experience, the Salivette is not suitable for use in newborns because of the large size of the cotton swab (diameter, 1.3 cm; length, 3.5 cm). Apart from the danger of swallowing the swab, the infants are stressed by this procedure. They refuse to keep the swab inside their mouths because their cheeks are too small. Additionally, the mean saliva volume of ∼100 μL obtained by this method is only sufficient to determine salivary cortisol by currently developed commercial assays (34), and may not be sufficient for further saliva analysis. The inert polyester swab should be used to avoid falsely high values (15)(16). In addition, saliva collection in toddlers requires special adaptations of the sampling devices (35) to guarantee stress-free sample collection.