Intraindividual stability of hair cortisol concentrations

Summary

The analysis of cortisol in human hair constitutes a promising method for the retrospective assessment of cumulative cortisol secretion over extended periods of time. An implicit assumption underlying the use of this method is that in the absence of major life changes hair cortisol concentrations show a high level of intraindividual stability, i.e. single hair cortisol assessments exhibit considerable trait-specificity and are only to a smaller extent influenced by state-dependent factors. Here, we present data from two independent studies examining patterns of intraindividual stability in hair cortisol levels. In study I, 45 participants were examined at two sampling points carried out one year apart from each other. In study II, 64 individuals provided data at three sampling points which occurred at two-month intervals. In both studies, at each time point hair was sampled and relevant psychosocial and hair-related variables were assessed. Results of both studies consistently revealed strong test–retest associations for repeated hair cortisol measurements (‘r's between 0.68 and 0.79, ‘p's <0.0001). Findings of structural equation modelling applied to data of study II showed that single hair cortisol assessments comprise a strong trait component, explaining between 59 and 82% of variance, and are only to a lesser extent influenced by state-related factors. Only inconsistent evidence for covariation of changes in hair cortisol concentrations and simultaneous changes in perceived stress or other relevant variables was seen across the two studies. The current findings suggest a considerable degree of intraindividual stability in hair cortisol levels which highlights the utility of this method for obtaining trait estimates of long-term cortisol secretion in psychoneuroendocrinological research.

Introduction

Activity of the hypothalamus–pituitary–adrenal (HPA) axis and secretion of cortisol are frequently examined parameters in psychoneuroendocrinological research. Whilst acute cortisol reactivity constitutes an important part of the adaptive response to challenge (see Sapolsky et al., 2000), repeated activation of the HPA axis and long-term changes to basal cortisol secretion, such as under conditions of chronic stress, can have harmful consequences and have been linked to an increased susceptibility to various diseases (see Chrousos and Kino, 2007, Chrousos, 2009). A methodological challenge often faced by researchers aiming to understand the intermediary role of cortisol in linking chronic adversity and pathogenetic processes relates to difficulties in obtaining reliable estimates of long-term cortisol secretion using previously available assessment methods.

In most human research, cortisol levels have been measured using plasma, salivary or urinary samples. Whilst these are well-established methods, it is important to note that they only reflect acutely circulating cortisol levels (plasma or saliva) or cumulative cortisol secretion over somewhat longer periods, usually 12 or 24 h (urine). The HPA axis, however, is a highly reactive system which is influenced by a wide range of factors and exhibits marked circadian rhythmicity (Weitzman et al., 1971) as well as considerable day-to-day variability within an individual (Hellhammer et al., 2007, Stalder et al., 2010a). Together this suggests that single cortisol assessments using previous methods are strongly affected by the situational context and thus only provide relatively poor reflections of normal long-term cortisol secretion.

This notion is also clearly supported by findings showing only weak to moderate test–retest associations for cortisol concentrations assessed by a variety of methods, such as morning plasma, nocturnal urinary and 24-h urinary assessments (Coste et al., 1994), repeated plasma sampling during the late morning (Schulz & Knabe, 1994) or salivary assessments at different times across the day (Kirschbaum et al., 1990). Only weak to moderate levels of test–retest stability have also been reported for frequently used composite measures of cortisol secretion, such as the cortisol awakening response (CAR; Pruessner et al., 1997, Wüst et al., 2000, Stalder et al., 2011) or the diurnal cortisol profile (Edwards et al., 2001, Oskis et al., 2009). An apparent exception to this evidence can be seen in the results of a recent study which reported high to very high test–retest associations for daytime salivary cortisol assessments (Liening et al., 2010). However, as in this study the timing of saliva sampling – a crucial influence on cortisol findings due to the marked circadian cortisol rhythm – was only kept stable within-participants but not between-participants, it is likely that stability estimates were artificially inflated in this study.

A considerable state-dependence of previous cortisol results is also suggested by the findings of two studies showing that one-day salivary cortisol assessments across the day (Kirschbaum et al., 1990) and single assessments of the CAR (Hellhammer et al., 2007) are determined to a greater extent by state than by trait influences. A strategy suggested by the latter authors in order to gain more valid estimates of trait cortisol secretion lies in employing repeated measurements across several days. However, it must be noted that such a strategy is highly laborious and thus practically infeasible in most larger studies. In addition, the validity of repeated measurement data may also be compromised since problems with participant non-adherence to the sampling regime are known to increase with the length of the study period (Kudielka et al., 2003, Broderick et al., 2004).

Given these difficulties in obtaining reliable and valid estimates of long-term cortisol secretion using previous methods, the recent observation by Raul et al. (2004) that endogenous cortisol concentrations can be detected in human hair may constitute a major methodological advancement. Importantly, as it is assumed that cortisol is incorporated into the hair shaft during hair growth, the examination of cortisol in a specific hair segment is believed to provide a retrospective index of cumulative cortisol secretion over the time period during which the hair segment has grown (see Gow et al., 2010). Given an average hair growth rate of ∼1 cm/month (Wennig, 2000), the examination of a 3 cm hair segment allows the assessment of cumulative cortisol exposure over a period of three months – a window of time which would have been virtually impossible to cover adequately using previous measures.

The validity of hair cortisol as a retrospective index of long-term cortisol secretion has now been supported by research using a range of different paradigms, both in animals (Davenport et al., 2006, Accorsi et al., 2008) as well as in human participants (Sauve et al., 2007, Kirschbaum et al., 2009, Thomson et al., 2009, Stalder et al., 2010b, D’Anna-Hernandez et al., 2011, Manenschijn et al., 2011). In addition, evidence has been reported confirming marker qualities of hair cortisol levels with regard to chronic stress and anxiety-related measures (Kalra et al., 2007, Yamada et al., 2007, Van Uum et al., 2008, Dettenborn et al., 2010, Fairbanks et al., 2011, Laudenslager et al., 2011, Dettmer et al., in press). Finally, hair samples are also conveniently obtained, can be stored at room temperature and through their retrospective nature are insensitive to participant non-adherence which makes this method particularly well-suited for larger field studies.

Whilst overall these findings indicate the great potential of hair cortisol analysis as a retrospective measure of stable patterns of long-term cortisol secretion, to date only scarce information on fundamental characteristics of hair cortisol levels are available. Specifically, the implicit assumption underlying the use of this method that in the absence of major life changes hair cortisol concentrations show a high level of intraindividual stability – i.e. single hair cortisol assessments are determined to a large part by trait influences and only to a smaller extent by state-dependent factors – has to our knowledge not been tested to date. Given the importance of this assumption, in the current report we explicitly set out to examine patterns of intraindividual stability in hair cortisol concentrations. For this, we present results of two independent studies in which repeated hair cortisol assessments in the same individuals (test–retest) were carried out twice over a one-year interval (study I) as well as three times with two-month intervals between measurements (study II). From results of the latter study we also specifically aim to deduce information on the extent to which variance in hair cortisol concentrations can be decomposed into state and trait components.