| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Center for Pediatric Epidemiology and Biostatistics (C.P., L.L.), Institute of Child Health, London WC1N 1EH, United Kingdom; and Human Early Learning Partnership (C.H.), University of British Columbia, Vancouver, Canada V6T 1Z3
Address all correspondence and requests for reprints to: Chris Power, Professor of Epidemiology and Public Health, Institute of Child Health, Center for Pediatric Epidemiology and Biostatistics, 30 Guilford Street, London WC1N 1EH, United Kingdom. E-mail: C.Power{at}ich.ucl.ac.uk.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Objective: The objective of the study was to assess whether prenatal and postnatal growth and adiposity are associated with adult cortisol levels, and whether early growth and adiposity are related to later cortisol through adult body size.
Design: Weight, head circumference (birth), height, and body mass index (BMI) (7 yr); and height, BMI (33 yr), and waist-hip ratio (WHR) (45 yr) were measured in the 1958 British birth cohort.
Setting: All study subjects were born in England, Scotland, and Wales in 1 wk in March 1958.
Participants: A total of 6,470 participants with salivary cortisol were gathered from 12,069 invitees (54%) at 45 yr.
Main Outcome Measures: Two saliva samples on 1 d were collected: 45 min postwaking (t1) and 3 h later (t2). Three cortisol outcomes were measured: t1 level, area-under-curve, and abnormal t1–t2 pattern.
Results: WHR was associated with all cortisol measures: among men over the WHR range 0.81–1.05, t1 cortisol decreased by approximately 3 nmol/liter, and the risk of an abnormal t1–t2 pattern increased by 77%; for women, over the WHR range 0.69–0.93, the risk of an abnormal t1–t2 pattern increased by 74%. For childhood measures, among males, increasing 7-yr BMI was associated with decreased t1 cortisol and increased risk of an abnormal t1–t2 pattern. Poorer prenatal growth in women, and postnatal growth in both sexes, was associated with increasing area-under-curve.
Conclusions: Smaller head circumference, shorter stature, lower BMI, and WHR are associated with higher cortisol levels.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Birth weight was negatively correlated with fasting plasma cortisol concentrations in three adult cohorts (5), whereas elsewhere, adults who were small for gestational age had higher early morning plasma cortisols at 20 yr than others (4). Among children, higher 24-h urinary cortisol was seen in those who were high and low birth weight compared with the middle range (7); whereas others show no association for birth weight and diurnal serum cortisol (8). Similarly, no relationship was found for several cortisol measures among elderly women (3, 9) and men (3). With regard to postnatal growth, stunted 8 to 10 yr olds had higher baseline salivary cortisol concentrations than nonstunted children (10), and among 2 yr olds reared in Romanian orphanages, all stunted in growth, none exhibited a normal diurnal cortisol decline (11). However, cortisol was unrelated to height in two further studies of children (7, 8) or in three adult cohorts (5). In respect of adult adiposity, a laboratory study showed that women with a high waist-hip ratio (WHR) had higher cortisol levels than those with normal WHR when exposed to uncontrollable stressors (12). In nonexperimental studies, postwaking cortisol was higher in men, but not women, with greater WHR (13, 14) as were 24-h urinary excretion levels (15). Also, men with higher WHR were found to have lower basal morning cortisol values (16, 17). For body mass index (BMI), some studies show higher (15) or lower (3, 5, 18) cortisol levels with increasing BMI; others show no association (3, 13). Lower cortisol levels have the greatest biological plausibility because it has long been recognized that obesity is associated with increased clearance of cortisol by the liver (19).
It is now well known that secretion typically follows a diurnal rhythm, with a peak after waking and gradual decline throughout the day. Alternative patterns are observed, including an absence of the early morning peak, or prolongation of the high awakening level, or rises later in the day. These alternative patterns may include both hyposecretion and hypersecretion (11, 20, 21), representing dysregulation of the HPA axis. It has also been established that salivary measures are well-suited for population studies, partly because cortisol levels in saliva are closely correlated with the "free" cortisol fraction in serum (22).
Building on these insights, we examine whether prenatal and postnatal growth and adiposity are associated with salivary cortisol in midadulthood in the 1958 British birth cohort. Inherently, growth and adiposity develop over the life-course; therefore, our aim is to investigate whether growth and adiposity in early life are related to later cortisol patterns through a link with adult body size or whether they make a separate contribution. Rather than focusing on extreme groups, such as low birth weight or stunted postnatal growth, we examine the full population distribution. In view of gender differences seen in previous studies, we investigate relationships separately for men and women.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The 1958 cohort includes all children born in England, Scotland, and Wales in 1 wk in March 1958. About 17,000 live births were followed-up at ages 7, 11, 16, 23, 33, and 42 yr (23). More recently, at 44–45 yr, a target of 12,069 participants still in contact with the study, and who, at 42 yr, had not required a proxy interview (e.g. due to learning disability) were invited to a clinical examination undertaken in their home by a trained nurse; 9377 (78%) participants were seen. Ethical approval for the 45-yr survey was given by the South East Multi-Centre Research Ethics Committee; participants gave consent for studies of cortisol.
Compared with the original birth sample, those with cortisol data at 45 yr (3187 males, 3283 females) were heavier at birth (3352 vs. 3295 g), taller at 7 yr (123.1 vs. 122.8 cm for males, 122.1 vs. 121.9 cm for females) and less likely to have unskilled manual (IV&V) class at birth (21.0 vs. 24.3%).
Measures
Salivary cortisol. At 45 yr, participants were asked to collect two saliva samples on the next convenient day, the first at 45 min after awakening (time 1) and the second, 3 h later on the same day (time 2). A reminder was sent to 53% of those consenting to return a saliva sample, if they had not done so within 2 wk of the nurse visit. Samples were received from 6568 participants: of these, 6527 yielded usable information on cortisol; 6452 had information at both time 1 and time 2.
Participants were instructed to avoid brushing or flossing their teeth, eating, or drinking for 15 min before taking each sample. They were asked to chew on a salivette until it was soaked, record the date and time of collection, and store the sample at room temperature until mailed to the laboratory. Salivary cortisol is stable at room temperature for up to 30 d, but samples were frozen after reaching the laboratory to reduce microbial growth. Cortisol levels were measured at the University of Dresden with a commercial immunoassay kit with chemiluminescence detection (CLIA; IBL-Hamburg, Hamburg, Germany). The lower sensitivity of this assay is 0.16 ng/ml, with intraassay and interassay precision of less than 10% for a wide range of cortisol concentrations. High cortisol levels (>50 nmol/liter) were rerun in a second assay for confirmation. Participants also reported 1) whether they regularly worked at night (shift-work), 2) wakefulness during the previous night, 3) dental work within the last 3 d, 4) cuts inside their mouth that may bleed, and 5) current medications.
Anthropometry. Birth-weight measured in ounces was converted into grams. At birth, gestational age was recorded and head circumference was measured. Height was measured without shoes to the nearest inch at 7 yr, the nearest centimeter at 33 yr, and the nearest 0.1 cm at 45 yr. Leg length was obtained (standing height minus sitting height) at 45 yr. All height measures were converted into centimeters. Weight was measured in underclothes to the nearest pound at 7 yr, and weight was measured with indoor clothing without shoes to the nearest 0.1 kg at 33 and 45 yr. BMI was calculated as weight/height2 (kilograms per square meter). Waist (midway between the lower ribs and iliac crest in the midaxillary line) and hip (widest part of the body below the waist) circumferences were measured to the nearest millimeter at 45 yr. WHR was derived from these measures. All anthropometric measures, except head circumference and WHR, were converted to SD scores.
Socioeconomic position (SEP). Separate measures of SEP were constructed for childhood (based on father’s occupation at birth and at 7 yr) and adolescence to adulthood (based on father’s occupation when the participant was 16 yr, and on the participant’s own occupation at 23 and 33 yr).
Data analysis
As expected, the most common pattern was of relatively high postwaking cortisol with a steep decline to prelunch, but other patterns were also observed. Therefore, the postwaking peak (t1) is a notable variable in its own right, whereas t2 relative to t1 represents a total 3-h cortisol level as well as the presence of a "normal" morning decline. We used three outcomes to differentiate these patterns: 1) t1 cortisol, 2) area-under-curve (AUC), and 3) typologies for t1–t2 change in cortisol (decline, rise, low flat, other flats).
t1 and t2 cortisol values were truncated at 2 nmol/liter (for <2 nmol/liter), the lower limit of the detection range, and at 100 nmol/liter (for >100 nmol/liter) because higher values are physically implausible. Times of sample collection varied around the specified target for t1 [mean (SD) of 49 (15) min after waking] and t2 [mean (SD) of 3 h, 5 min (23 min)]. Cortisol values were skewed and, therefore, transformed using log 10. Cortisol level was influenced by both the time of awaking and time since awaking. Therefore, the transformed t1 values were centered at 0808 h (45 min after mean awakening time of 0723 h) and t2 values at 1108 h (3 h, 45 min after mean awakening time), using predictions from linear regression models, then back transformed to the original scale (nmol/liter). For AUC, the back-transformed t1 and t2 values were summed, multiplied by 3 h, and divided by 2. Thus, AUC represents the 3-h average of t1 and t2 after allowing for the variation in sample collection times. AUC was log 10 transformed to reduce skewness of the distribution. For the four typologies, "normal decline" is defined as having a t1 cortisol value more than 7.5 nmol/liter and a t2 value that is 20% lower than the t1 value; "low flat" is defined as t1 cortisol values 7.5 nmol/liter or less and a t2 cortisol within 20% of the t1 value; "other flat" is t1 cortisol more than 7.5 nmol/liter and a t2 cortisol within 20% of the t1 value; "rise" is a t2 cortisol that is 20% higher than the t1 value. In analyses, the typologies are combined into two categories: decline and "abnormal" pattern (including rise, low flat, and other flats).
Potential confounding factors were examined in relation to t1 cortisol. Regular shift working, recent dental treatment, cuts inside the mouth, and current medication were not associated with t1 cortisol. Sleep disturbance during the previous night was weakly associated with a reduced t1 cortisol, but only in women. No adjustments for these factors were made. All analyses were conducted for men and women separately.
Associations of growth and adiposity with t1 cortisol and AUC were examined using linear regression. Because t1 cortisol and AUC were log 10 transformed, percentage change in these measures was calculated from the regression coefficient (ß) as 100 x (10ß – 1). Due to log transformation, the coefficient (ß) is interpreted as percentage change, i.e. t1 cortisol changes from value X1 to (100 + ß)% X1 for a unit increase in anthropometric measure. Thus, the increase will differ at different t1 values. Associations with an abnormal t1–t2 change were assessed using logistic regression. Stages of analysis were as follows. First, effects of each anthropometric measure were estimated in univariate models; nonlinear relationships were tested with inclusion of a quadratic term. Next, effects of each measure were adjusted for SEP: for prenatal/childhood growth and adiposity, adjustments were made for childhood SEP; for adult adiposity, there was an additional adjustment for adult SEP. Where anthropometric measures were robust to adjustment for SEP, we examined all combinations in mutually adjusted models, namely prenatal/childhood, childhood/adulthood, prenatal/adulthood, and prenatal/childhood/adulthood. The purpose was to identify the contribution of each factor and whether it operated through other factors occurring later in life. In addition to results reported in the tables, we illustrate associations by presenting cortisol outcomes for a range, equivalent to ±2 SD from the mean, in the anthropometric measure. Because there is particular interest in individuals with low birth weight and adult obesity (2), we tested the interaction of these factors on cortisol outcomes; we also examined the association between adult adiposity and cortisol outcomes stratified by birth-weight (tertiles).
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
). Some participants had a higher t2 measure than t1, but on average, cortisol levels declined, with women having a greater decline than men. AUC was similar for men and women; for both, the range of AUC values is wide, indicating large differences in total 3 h free cortisol. Most participants had a normal decline (82.4% for men, 86.6% for women) (Table 1). Table 2 presents descriptive statistics for the growth and adiposity measures by life stage (prenatal, childhood, and adult) to highlight their timing for later analyses.
|
|
Among men, increasing BMI at 7 and 33 yr and WHR at 45 yr are associated with decreasing t1 cortisol, before and after adjustment for SEP (Table 3). In further analyses (data not shown), effects of WHR were also robust to adjustment for BMI at 7 yr. BMI at 7 yr made an additional contribution to the effect of WHR, but not to that for BMI at 33 yr, suggesting that BMI in childhood operates through BMI in adulthood. For WHR, over the range 0.81–1.05 (equivalent to ±2 SD from the mean), t1 cortisol decreased by 2.7 nmol/liter. The estimated difference in t1 cortisol over a BMI range 12.9–18.9 kg/m2 (±2 SD from the mean) was 1.84 nmol/liter. Among women, increasing head circumference at birth, height at 7 and 33 yr, BMI at 33 yr, and WHR at 45 yr were associated with decreasing t1 cortisol, before and after adjustment for SEP. The association for WHR persisted after adjustment for head circumference, height at 33 yr, and SEP, with a decrease of approximately 2 nmol/liter over the range 0.69–0.93 (±2 SD from the mean) (data not shown). In further models, head circumference and 33-yr height had a borderline effect, in addition to adult WHR.
|
In univariate analyses, AUC increased with increasing birth weight for gestational age for men, but decreased with increasing height at 7 yr, BMI at 33 yr, and WHR at 45 yr (Table 4). Importantly, the effect of height at 7 yr was reduced after adjustment for SEP. Nonetheless, height at 7 yr and WHR at 45 yr were associated with AUC after adjustment for other anthropometric measures and SEP (Table 4). BMI at 33 yr was the only factor with a significant nonlinear relationship: AUC decreased from the lowest BMI to the 70th percentile, increasing thereafter among the most overweight adults (data not shown). For women, increasing head circumference at birth, height at 7 and 33 yr, and BMI at 33 yr were all associated with decreasing AUC. Mutual adjustment showed that head circumference and height at 7 yr each contributed to AUC. As height at 7 yr decreased over the range 134–110 cm (±2 SD from the mean), AUC increased by 8% (from 38.2–41.5 nmol/liter); as head circumference decreased from 55.4–48.7 cm (±2 SD from the mean), AUC increased by 7.4% (from 38.3–41.6 nmol/liter).
|
In both sexes, increasing WHR and BMI at 33 yr, and for men only, increasing BMI at 7 yr, were associated with an increased risk of abnormal pattern, before and after adjustment for SEP (Table 5). The association for BMI at 33 yr was abolished after adjustment for BMI at 7 yr. Thus, BMI at 7 yr was an independent predictor in men, with an increased risk of abnormal pattern of 11% over a SD of BMI at 7 yr. Adult WHR also independently predicted abnormal patterns in both sexes: the risk increased by 25 and 23%, respectively, for a 0.1 increase in WHR. WHR was positively associated with all three abnormal cortisol typologies (rise, low flat, other flat); results were less consistent for BMI at age 7.
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This is a large population-based study, providing reference values for cortisol secretion patterns in midadult life: 82.4% of men and 86.6% of women showed a postwaking level of more than 7.5 nmol/liter and a decline of more than 20% from t1–t2; whereas the remaining 17.6 and 13.4%, respectively, had abnormal patterns. Nonetheless, mean 45 min postwaking cortisol values for our 45 yr olds (21.01 for men; 21.88 nmol/liter for women) are comparable with other studies, for example, adults aged 18–71 yr (22.3 nmol/liter for men and women combined) (24).
Strengths and limitations
Absolute levels of cortisol in saliva are significantly lower than in blood, but are strongly correlated with serum cortisol (r = 0.71–0.96) and are more closely correlated with the free cortisol fraction (22). Although free cortisol is available to cross cell membranes, influences on activity of cell receptors is not well understood, and therefore, the relationship between free cortisol and intracellular activity is uncertain. Because salivary measures are relatively easy and inexpensive to collect, they are well-suited for population studies. Ideally, an individual’s diurnal cortisol rhythm is obtained from multiple saliva collections throughout a day, repeated over several days to address intraindividual variability. In our large population study, a maximum of two samples on 1 d was feasible and affordable. Because of the study size, precision in the estimation of effects is gained at the group level even though our estimates for individuals may be less reliable. The usual pattern is for cortisol levels to decline within a few hours from a postwaking peak. Thus, the timing of our samples should capture change over this crucial interval. We were also able to examine several potential confounding factors, such as SEP and medications. Due to attrition, the sample with complete data was less than half of the original birth cohort, leading to an underrepresentation of participants from classes IV&V at birth. We have no reason to suspect that relationships for growth and adiposity are biased by loss to follow-up, but we cannot discount the possibility of differential adherence to the saliva collection protocol.
Interpretation of findings
Our results demonstrate that growth and adiposity across the life course are related to cortisol levels in midlife. Childhood and adulthood growth and adiposity were associated with all cortisol measures, but prenatal factors were related only to AUC and, to a lesser extent, t1 cortisol. This study underscores the importance of adult adiposity, particularly central adiposity, regardless of the cortisol measure examined. Without anthropometric measures over the life span, cross-sectional associations are difficult to interpret, and inconsistencies may occur between studies.
However, there are complexities in the associations even for adult adiposity: greater adiposity (BMI and WHR) was associated with lower t1 values, lower AUC, and a higher risk of an abnormal pattern. These trends are consistent with the observation that cortisol clearance is increased among the obese (19) and with several other (3, 5, 16, 17, 18), but not all, previous studies (13, 14, 15). Inconsistencies among studies suggest that further research should establish whether there are extremely low and high cortisol levels at extremes of the adiposity distribution that have not been detected to date. Our study is one of the few to examine childhood growth/adiposity and cortisol in adulthood. Interestingly, for BMI at 7 yr for males, the pattern is similar to that for WHR at 45 yr, wherein increasing BMI is consistently associated with decreasing t1 and increasing risk of abnormal t1–t2 patterns. Despite the observation shown elsewhere that childhood obesity predicts adult obesity in this cohort (25), the association of BMI at 7 yr with cortisol levels at 45 yr is independent of adult adiposity.
For both males and females, shorter stature at age 7 yr was associated with greater cortisol levels (higher AUC). This finding is consistent with studies of Romanian orphans (11) and Jamaican children (10) showing that stunting was associated with cortisol hypersecretion. Our finding that associations with height at 7 yr reduced after adjustment for SEP suggests that delayed growth resulting from early life deprivation may have long-lasting effects on cortisol metabolism. Growth in children treated with glucocorticoids has been shown to be inhibited (26), suggesting that the direction of association may be from cortisol to growth. We also show that decreasing head circumference at birth for females is associated independently with increasing cortisol levels, a similar pattern to height at 7 yr. Once again, the underlying mechanisms are uncertain. One possible mechanism is that the same prenatal factors that influence head circumference also "program" the HPA axis, which, in turn, influences postnatal growth, but other mechanisms cannot be excluded. Regardless of the mechanism, in this large population study, the predominant, possibly underlying, association appears to be a gradient between increasing underdevelopment and cortisol dysregulation: smaller head circumference, shorter stature, lower BMI and WHR having higher cortisol levels and lower odds of abnormal t1–t2 patterns. These findings are counter to the claim that cortisol hypersecretion is more prevalent in central adiposity (27).
Cortisol dysregulation has been investigated in relation to several outcomes, including blood pressure (5, 28), glucose tolerance (28), type-2 diabetes (29), stroke and cardiovascular disease risk (6, 29), memory loss (30), and breast cancer survival (31). We do not understand fully yet which measures of cortisol are most closely associated with specific outcomes, nor the direction or magnitude of dysregulation that would be clinically significant over the long term. Also, the importance of population differences in cortisol secretion has yet to be established: small differences between groups may have important effects at the population level if they represent a shift in the entire distribution, thereby increasing the proportion of the population above or below a cortisol threshold of clinical significance (32).
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Disclosure statement: The authors have nothing to disclose.
First Published Online August 15, 2006
Abbreviations: AUC, Area-under-curve; BMI, body mass index; HPA, hypothalamic-pituitary-adrenal; SEP, socioeconomic position; WHR, waist-hip ratio.
Received March 22, 2006.
Accepted August 9, 2006.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  C Thomas, C Hertzman, and C Power Night work, long working hours, psychosocial work stress and cortisol secretion in mid-life: evidence from a British birth cohort Occup. Environ. Med., December 1, 2009; 66(12): 824 - 831. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K Atherton, E Fuller, P Shepherd, D P Strachan, and C Power Loss and representativeness in a biomedical survey at age 45 years: 1958 British birth cohort J Epidemiol Community Health, March 1, 2008; 62(3): 216 - 223. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
