This observational study evaluated the potential of hair testosterone measurements in general and segmental hair analysis as a retrospective testosterone diary in particular. Testosterone patterns along the hair were preserved over time to some degree, but concentrations were generally higher more distally from the scalp. Testosterone concentrations in hair were correlated to testosterone concentrations in saliva and could on a group level reliably differentiate between men and women. In addition to sex, also hair colour and relationship status were associated with hair testosterone levels. Finally, the intra-individual variation in hair outgrowth over a three-month period identified a rarely considered confounder in the field of hormonal hair analysis.
To our knowledge, only one previous study has compared hormone concentrations (cortisol, washed in isopropanol and analysed with an immunoassay) in outgrown hair segments with previously sampled scalp-near segments [17]. Unlike the present study, they found no correlation between the related hair segments. Although we could demonstrate a preserved relation in concentrations between adjacent segments as the hair grew, the absolute concentrations were higher more distally from the scalp. A possible explanation could be continuous addition of hormones to hair not only in the hair follicle but also more distally, for example through contact with sebum which could contain androgens. Sebaceous glands and sweat glands contain androgen-synthesizing enzymes and influence the production of sebum, but whether the sebum itself contains hormones has not been studied [18]. In hair from pigs and cattle, Otten et al. described how distal hair segments show greater permeability compared to proximal hair segments to external contamination with urine, leading to markedly increased cortisol concentrations in contaminated hair [19]. Likewise, repeated immersion in water resulted in loss of cortisol in distal hair segments compared to proximal segments. This is also in line with our finding that increased hair wash frequency moderates the distal increase in testosterone concentrations, with the possible rationale being that testosterone from sebum and/or cortex is washed away and that the effect of hair washing has a cumulative effect in the more distal hair segments. Consequently, it is plausible that the previously described decreases in hair cortisol concentrations in distal hair segments, when a pre-extraction wash has been used, leads to increased removal of lipophilic substances from distal hair segments compared to scalp-near hair. Another possibility is that hair matrix degradation could affect the ratio between the total sample weight and the hormone content, which could in part explain the increasing hormone concentrations in distal hair segments. In vitro hair washing and exposure to visible light and UV light has been shown to damage hair, reducing the tensile strength of the hair and damaging the protective cuticle [20–22]. Increased photodamage in distal hair segments compared to proximal segments was described by Cao et al. and the same pattern of damage could be reproduced by exposing hair from the proximal segment to increasing amounts of UV light [23].
Validation of testosterone analysis in hair through correlation with salivary testosterone concentrations has only been attempted a few times with contradicting results [24, 25]. As mentioned above, there is a local production of steroid hormones in the skin, independent from the hypothalamo-pituitary-gonadal axis, that could potentially blur the relationship between the two matrices [26, 27]. Also, methodological differences such as different antibodies in the assays used for the hair and saliva analyses, respectively, could add some variance. Further, a recent study on radiolabelled cortisol administered to rhesus monkeys describes how cortisol was not only incorporated in hair in its native form but more readily incorporated in the form of cortisone as well as other yet unidentified radiolabelled metabolites in the hair samples [28]. Perhaps a true measure of mean retrospective concentrations of the unbound testosterone hormone fraction in serum should include testosterone metabolites in the hair, which needs to be explored in future research. However, the current results showed that mean salivary testosterone and testosterone in hair correlated significantly. Interestingly, hair testosterone analysis demonstrated a superior ability to discriminate between the sexes compared to mean salivary testosterone, which underpins the physiological relevance of the testosterone levels in hair. Hair washing frequency differed slightly between the sexes (women: mean 2.68 times per week, SD 1.18; men: mean 4.00 times per week, SD 2.05), with more frequent hair washing habits among men, meaning that the difference in hair testosterone concentrations between sexes could potentially be even more distinct if the analysis had been corrected for hair washing frequency. The salivary analysis was calculated with a mean value of 16 saliva samples per individual, which should correct for the intra-day and inter-day variation in salivary testosterone concentrations. Two possible explanations to the superior discrimination between sexes of hair testosterone are that the time frame for saliva sampling in the current may have been suboptimal and did not capture the time during the day when the difference between the sexes would have been the most pronounced. Also, a local production and incorporation of testosterone in the hair follicle could, perhaps, occur to a greater extent in the scalp hair of males.
Regarding the relation between background variables and testosterone in hair, the regression analysis showed significant effects for sex and natural hair colour, as well as significantly lower hair testosterone levels in individuals cohabitating with a partner compared to singles or persons not cohabitating with a partner, adjusted for potential age differences between the groups. To our knowledge, only Voegel et al have evaluated testosterone concentrations in hair with regard to pigmentation, where no intra-individual difference between pigmented and grey hair was found (n = 18). Several studies focusing on cortisol have addressed the issue of natural hair colour, with ambiguous results. Higher cortisol concentrations in individuals with black hair compared to individuals with lighter shades has been reported [29–31], as well as no differences in hair cortisol concentrations in relation to natural hair colour [32]. It has also been hypothesised that differences in hair hormone concentrations across categories of hair colour would be an effect of ethnicity [29, 33]. During year 2020 19.7 % of the populaion in Sweden was either born abroad or had parents that both were born abroad [34]. Unfortunately, there was no information collected on the ethnical background of our study participants. Ad hoc, we performed Spearman correlation on mean salivary testosterone with the different hair colour shades as an ordinal variable (men and women apart), and found no significant correlations between salivary testosterone and natural hair colour (rho=-0.122 to 0.117, p = 0.473 to 0.970). This could support the effect of hair pigmentation specifically on the hair testosterone concentrations. Regarding possible matrix degradation, the melanin pigment counteracts cortical damage from visible light and UV light, with the least photobleaching occurring in black hair [35, 36]. In an in vitro setting, irradiating a hydrocortisone solution and intact hair strands has been shown to decrease hydrocortisone concentrations in the solution as well as decrease the cortisol concentrations in hair [8]. Interestingly, in the same study by Grass et al. dehydroepiandrosterone and progesterone did not have the same pattern of decreasing concentrations as cortisol, which raises the question of whether steroid hormones continue to be metabolised within the hair as it remains attached to the scalp.
The current study confirms previous research regarding an average scalp hair growth rate of about 10 millimetres per month. Growth rate has generally has been studied in regrown hairs previously shaved or clipped, with a follow up of a few days to a fortnight, which also allowed for identification of growing and resting hair follicles at a certain time-point [37, 38]. With a longer follow up time, of weeks to months, the impact of hair follicles switching between growing, transitioning and resting phases (anagen, catagen and telogen) on the total hair growth should be detectable and influence the individual lengths of regrown hairs within a sampling area. Hair follicles transition between growth phases independently of each other, the proportion of growing hair follicles and mean growth rate varies during the year and towards the end of the anagen phase the growth has been shown to gradually slow down [39, 40]. This has the potential to influence the temporal relationship between a certain hair segment and a specific time period in the past [41]. Descriptions of intra- and inter-individual variability in hair growth rate have been surprisingly scarce, given its important implications for hair hormone analyses. To the best of our knowledge, a within-individual variation in hair growth rate impacting the length of regrown hairs during a few months has previously not been described. This finding highlights a confounder that is not commonly considered in hormonal hair analysis, hampering the temporal resolution when attempting to use hair as a retrospective hormone diary.