Association between hair cortisol, hair cortisone, and fatigue in people living with HIV

Abstract

Cumulative evidence to date largely supports an association between dysregulation of the activity of the hypothalamic-pituitary-adrenal (HPA) axis and fatigue. People living with HIV (PLHIV), in particular, are vulnerable to both HPA axis dysregulation and fatigue. Few investigations have examined the possible role of HPA-axis dysfunction in the occurrence of fatigue in PLHIV. This cross-sectional study aimed to investigate the association between glucocorticoids in hair, retrospective indicators of long-term HPA axis activity and biomarkers of chronic stress, and fatigue in PLHIV. A total of 446 PLHIV from Guangxi China provided hair samples for cortisol and cortisone assay and provided information on fatigue levels, sociodemographic, lifestyle, and HIV-related characteristics. Results showed that before and after controlling sociodemographic, lifestyle, and HIV-related characteristics, hair cortisone levels, but not hair cortisol levels, were associated with fatigue levels in PLHIV. In conclusion, we found that higher cortisone levels are associated with greater fatigue levels in a large cohort of Chinese PLHIV.

LAY SUMMARY

• We found that hair cortisone levels were significantly associated with fatigue levels in a large cohort of Chinese PLHIV. Hair cortisol levels were, however, not associated with fatigue levels in the PLHIV studied. We thus show that Chinese PLHIV who have higher cortisone levels are associated with higher fatigue levels.

Keywords:

• Fatigue
• hypothalamic-pituitary-adrenal axis
• glucocorticoid
• hair cortisol
• hair cortisone
• people living with HIV

1. Introduction

Fatigue is one of the most common and disabling symptoms in people living with HIV (PLHIV) (Barroso et al., 2010). As indicated by a systematic review, fatigue prevalence varied from 33% to 88% among PLHIV (Jong et al., 2010). Fatigue is described as a lack of energy, sleepiness, tiredness, exhaustion, an inability to get enough rest, or weakness (Adinolfi, 2001). Previous research largely supported that fatigue had adverse impacts on PLHIV’s physical, psychological, social relationships, daily activities, and quality of life (Claborn et al., 2015; Perazzo et al., 2017). However, the etiology of fatigue in PLHIV is uncertain, with many HIV disease-mediated mechanisms (e.g. HIV infection) and secondary factors (e.g. psychological and socio-economic variables) potentially implicated (Barroso et al., 2003; Barroso et al., 2010; Lee et al., 2001; Leserman et al., 2008; Meynell & Barroso, 2005; Phillips et al., 2004). Therefore, a better understanding of the pathophysiology and the biological mechanisms contributing to fatigue is needed for preventing fatigue development and establish treatments against fatigue in PLHIV.

Of the putative mechanisms contributing to fatigue, glucocorticoids (GCs) abnormalities, indicators of dysfunction of the hypothalamic-pituitary-adrenal (HPA) axis, has been proposed as one of the crucial causes of fatigue in PLHIV (Barroso, 1999; Jong et al., 2010). Cortisol is the principal GC hormone in humans, plays an essential role in the energy balance, the 24 h circadian rhythm, and stress responses. Cortisol abnormalities are well documented in PLHIV (Zapanti et al., 2008), with some studies find elevated cortisol levels in PLHIV (Christeff et al., 1999; Collazos et al., 2003), whereas others contradict these findings with lower or normal cortisol levels in PLHIV (Langerak et al., 2015; Merenich et al., 1990; Odeniyi et al., 2013). In addition, cumulative evidence has shown an association between GCs abnormalities and fatigue in patients with chronic fatigue syndrome (CFS) (Papadopoulos & Cleare, 2011; Powell et al., 2013). However, data are limited on the relationship between GCs levels and fatigue in PLHIV (Jong et al., 2010). To the best of our knowledge, only one study reported data on the relationship between salivary cortisol and fatigue in PLHIV without empirically tested (n = 40) (Barroso et al., 2006).

As well known, cortisol is an active glucocorticoid, and cortisone is an inactive glucocorticoid originating from the local conversion of cortisol by the 11β hydroxysteroid dehydrogenase (11β-HSD) type 2 enzyme, and 11β-HSD type1 enzyme is responsible for the reversible conversion of cortisone to cortisol (Perogamvros et al., 2010; Zhang et al., 2017). Accordingly, cortisol and cortisone interaction regulates stress-induced psychological and physiological responses together (Vanaelst et al., 2013; Wang et al., 2015). Therefore, measurement of cortisone and cortisol levels may provide a thorough assessment of active and inactive GCs exposure (Zhang et al., 2018). In addition, cortisone levels are higher than those of cortisol levels in biological samples (e.g. saliva, urine, and hair) (Perogamvros et al., 2010; Zhang et al., 2017). Moreover, cortisone levels could more closely approximate unbound, biologically active cortisol levels than total cortisol levels (Perogamvros et al., 2010). Regardless of the relatively little is known about cortisone's physiological significance, previous studies examed the relationship of cortisol and cortisone levels and fatigue in CFS and found that CFS patients had lower salivary and urinary cortisol and cortisone levels than healthy controls (Jerjes et al., 2005; Jerjes et al., 2006). Thus, there are potential benefits to assess the association between GCs levels and fatigue by employing both cortisol and cortisone levels in PLHIV

The conventional GCs measurement methods in serum, saliva, or urine are susceptible to reflect acute or short-term GCs levels. GCs levels in those biological samples are easily affected by diurnal circadian rhythm and a host of potentially confounding (Wosu et al., 2013). Relative to GCs in those biological samples, GCs levels in scalp hair can retrospectively reflect long-term GCs exposure from weeks to months, depending on the length of the collected sample (Stalder & Kirschbaum, 2012). Comparing between hair GCs levels and GCs levels in repeated saliva sampling (Chen et al., 2019; Short et al., 2016; Sugaya et al., 2020; Zhang et al., 2018), testing the reliability of the hair GCs levels across a period of several months to a year (Chen et al., 2019; Stalder et al., 2012; Zhang et al., 2017), and employing hair GCs levels in stress or chronic disease research (Davison et al., 2019; Feeney et al., 2018; Musana et al., 2020; Scharlau et al., 2018; van den Heuvel et al., 2020), all suggest that hair GCs levels are novel retrospective indicators of long-term GCs exposure. To our knowledge, only two studies employed hair cortisol levels in HIV-related research. One study validated hair cortisol as a biomarker of chronic stress (Qiao et al., 2017), and another study investigated the relationship between hair cortisol and metabolic syndrome (Langerak et al., 2015). Limited data are available on employing hair cortisone levels in HIV-related research.

Accordingly, we assessed fatigue levels and hair GCs levels in a large cohort of PLHIV in Guangxi, China, and aimed to examine the association between hair GCs levels and fatigue levels among Chinese PLHIV.

2. Methods

2.1. Participants

Participants were recruited from an HIV disclosure project, a longitudinal study that aims to explore the mechanism underlying the linkage between HIV disclosure and clinical outcomes in Guangxi, China (Yang et al., 2019; Zhang et al., 2020a, 2020b). PLHIV aged 18 years, willing to consent to retrieve clinical outcomes from their medical charts, and providing hair samples were eligible for participation. PLHIV with any of the following characteristics were excluded: (1) linguistic, mental or physical inability to respond to assessment questions; (2) opportunistic infections, coinfection (e.g. hepatitis and sexually transmitted disease), comorbidities (e.g. cancer, heart disease, diabetes, iron deficiency, anemia, depression, and sleep disorders), endocrine diseases (e.g. adrenal dysfunction, autoimmune thyroid diseases, and hyperamylasemia), or reported any other diseases; (3) currently take hormonal drugs (e.g. prednisolone); (4) known history of drug use; (5) chemical hair treated (e.g. dyed, permed, or bleached) or scalp hair in the posterior vertex was less than 1 cm.

Medical staff or HIV case managers at the study sites referred potential participants to the research term members. Research team members screened PLHIV for eligibility, discussed the study's benefits and risks, and invited them to participate. Research team members were local CDC staff or health care workers in the HIV clinics who had received intensive training on research ethics and interview skills with PLHIV before the field data and hair specimen collection. Finally, a total of 446 PLHIV participated in this study. The research protocol was approved by the Institutional Review Boards of the University of South Carolina in the United States and the Guangxi CDC in China. The written informed consent also has been obtained from the participants.

2.2. Data collection

Hair samples were cut from the posterior vertex region as close as possible to the scalp following a standard protocol (Cooper et al., 2012; Sachs, 1997) in private rooms of local CDC or HIV clinics. The hair thatch then was completely enclosed by a piece of foil, and a small label indicating the study ID number was placed over the distal end of the hair thatch. The interviewer-administered questionnaire was then used to collect fatigue levels, sociodemographics, lifestyle, and HIV-related information. Each participant received a gift with a value equal to the U.S. $5.00 (≈ 35 Chinese Yuan) after finished the survey.

2.3. Measures

2.3.1. Fatigue levels

Fatigue levels were assessed using the Chinese version of the Medical Outcomes Study HIV Health Survey (MOS-HIV) fatigue subscale (Lau et al., 2006; Wu et al., 1997). The fatigue subscale is consists of four items: (a), how often during the past four weeks, did you feel full of pep?; (b), how often during the past four weeks, did you feel worn out?; (c), how often during the past four weeks, did you feel tired?; (d), how often during the past four weeks, did you have enough energy to do the things you wanted to do?. Each of the four questions has a 6-point response option from 1 = “all of the time” to 6 = “none of the time.” The raw scale score was constructed and then transformed to a 0 to 100 scale following the instruction. A higher transformed score was indicative of less fatigue. Scores on this scale range from 0 to 100, with scores below 50 indicating limitations or disability related to fatigue in the general population and breast cancer survivors (Hays et al., 1993; Ware & Sherbourne, 1992).

2.3.2. Hair cortisol and cortisone

The proximal 1-cm of hair segments (approximately reflect the last month of accumulative cortisol and cortisone levels) was cut finely, and 20 mg processed and analyzed using high-performance liquid chromatography-tandem mass spectrometry (LC-MS/MS) followed the protocol described by Gao and colleagues (Gao et al., 2013). The method showed the limit of detection (LOD) for both cortisol and cortisone at 0.1 pg/mg and the limit of quantitation (LOQ) for both cortisol and cortisone at 0.3 pg/mg. Intra-day and inter-day percentage coefficients of variation (CV) were less than 7% at standard concentrations of 1.25, 25, and 250 pg/mg, and recovery ranged between 95% and 107% for both cortisol and cortisone.

2.3.3. Sociodemographic, lifestyle, and clinical characteristics

Participants self-reported their sociodemographic and lifestyle characteristics that might have a potential influence on hair GCs levels (Staufenbiel et al., 2015; Wosu et al., 2013), including age (years), gender (male vs. female), ethnicity (Han vs. non-Han), marital status (married vs. other), education level (>9 years vs. ≤9 years), employment status (employed vs. unemployed), monthly household income level (≥3,000 Yuan vs. <3,000 Yuan), height (cm), weight (kg), frequency of hair washing (≥4 times/week vs. <4 times/week), using the hairdryer, curling iron or hair straightener (yes vs. no), using hair styling products (yes vs. no), smoking (yes vs. no), and drinking (yes vs. no). Based on weight and height, body mass index (BMI) was calculated.

Clinical characteristics were abstracted from the participants’ medical charts at the time of hair collection, including years of HIV diagnosis, combination antiretroviral therapy (cART) status, and CD4 count. CD4 count was dichotomized as >500 CD4 cells/mm³ vs. ≤500 CD4 cells/mm³ because the low limit of normal CD4 count is 500 CD4 cells/mm³ in adults. The cART status was category into untreated, first-line cART (non-nucleoside reverse transcriptase inhibitors based cART) and second-line ART (protease inhibitors based cART).

2.4. Statistical analysis

Twenty-one participants were excluded from the final analysis due to the insufficient weight of hair samples (less than 20 mg) for assaying cortisol and cortisone (n = 13) or unavailable data of fatigue measure (n = 8). This left 425 participants for analysis.

The Kolmogorov-Smirnov test indicated that hair GCs levels and hair cortisol to hair cortisone ratio were not normally distributed. Thus Winsorized (5th/95th percentile) (Dixon & Yuen, 1974) and Box–Cox transformed (Clark et al., 2016) were conducted to effectively reduced the skewness statistic. In preliminary analyses, simple linear regression analysis was used to examine the associations of hair GCs levels and hair cortisol to hair cortisone ratio with sociodemographic, lifestyle, and clinical characteristics.

Linear regression analyses with three models were used to assess the associations of hair GCs levels and hair cortisol to hair cortisone ratio with fatigue levels. In Model 1, fatigue levels were regressed with no adjustment. Model 2 was adjusted for sociodemographic and lifestyle characteristics. Model 3 included an additional adjustment for clinical characteristics. All analyses were performed using SPSS 26.0 (SPSS Inc, Chicago, IL).

3. Results

A description of the characteristics of all participants can be found in Table 1. Of the 425 PLHIV, the mean (SD and range) fatigue scores were 62 (19, 12.5–100), and 36.2% had fatigue scores below 50. The median (IQR) of hair cortisol levels was 7.60 (5.00 − 11.80)pg/mg, and hair cortisone levels was 43.92 (31.18 − 67.17) pg/mg. Hair cortisol levels and hair cortisone levels were positively correlated (r = 0.26, p < 0.001).

Table 1. Characteristics of the study sample (n = 425).

Characteristics	n (%), mean ± SD, or median (IQR)
Age, years	41.29 ± 8.38
Gender, n (%)
Male	288 (67.8%)
Female	137 (32.2%)
Ethnicity, n (%)
Han	261 (61.4%)
No-Han	164 (38.6%)
Marital status, n (%)
Married	335 (78.8%)
others	90 (21.2%)
Education level, n (%)
≤9 years	380 (89.4%)
>9 years	45 (10.6%)
Employment status, n (%)
Employed	331(77.9%)
Unemployed	94 (22.1%)
Monthly income level, n (%)
≥3000 Chinese Yuan	101(23.8%)
<3000 Chinese Yuan	324 (76.2%)
BMI, kg/m^2	21.79 ± 2.65
Smoking
Yes	165 (38.8%)
No	260 (61.2%)
Drinking
Yes	214 (50.4%)
No	211 (49.6%)
Frequency of hair washing, n (%)
≥4 times/week	117 (27.5%)
<4 times/week	308 (72.5%)
Using hair dryer, curling iron, or hair straightener
Yes	89 (20.9%)
No	336 (79.1%)
Using hair styling products
Yes	21 (4.9%)
No	404 (95.1%)
Years since HIV diagnosis	1.57 ± 0.80
cART status
Second-line cART	40 (9.4%)
First-line cART	348 (81.9%)
Untreated	37 (8.7%)
CD4 count
>500 cells/mm^3	120 (28.2%)
≤500 cells/mm^3	305 (71.8%)
MOS-HIV fatigue scores	62.23 ± 16.19
Hair cortisol, pg/mg	7.60 (5.00–11.80)
Hair cortisone, pg/mg	43.92 (31.18–67.17)
The ratio of hair cortisol to hair cortisone	0.16 (0.09–0.27)

n: number; SD: standardized deviation; IQR: interquartile range; BMI: body mass index; cART: combination antiretroviral therapy; MOS-HIV: Medical Outcomes Study HIV Health Survey; Han Ethnicity is the majority Ethnicity; Monthly income less than 3000 Chinese Yuan is the level of poverty, 1 United States dollar ≈ 7 Chinese Yuan.

Table 2 shows the univariate regressions of hair GCs levels and hair cortisol to hair cortisone ratio with sociodemographic, lifestyle, and clinical characteristics. Female PLHIV was associated with both higher hair cortisol levels and hair cortisone levels. Aging, no-Han ethnicity, be married, and had a higher BMI were associated with higher hair cortisol levels. Higher education and monthly income levels and using hair styling products were associated with lower hair cortisol levels. Smoking and more frequency of hair washing were associated with lower hair cortisone levels. Higher education and monthly income levels were associated with higher hair cortisol to hair cortisone ratio. Smoking and more frequency of hair washing were associated with higher hair cortisol to hair cortisone ratio. Years since HIV diagnosis, cART status or CD4 count was not associated with hair GCs levels or hair cortisol to hair cortisone ratio.

Table 2. Univariate regressions of hair cortisol, hair cortisone and the ratio of hair cortisol to hair cortisone.

	Hair cortisol			Hair cortisone			Hair cortisol/hair cortisol
	β	t	p	β	t	p	β	t	p
Age, years	.100	2.074	.039	.064	1.309	.191	.038	.772	.440
Gender (female vs. male)	.081	1.666	.096	.210	4.410	.000	−.078	−1.619	.106
Ethnicity (No-Han vs. Han)	.096	1.977	.049	.022	.462	.645	.067	1.389	.166
Marital status (Married vs. other)	.091	1.885	.060	.002	.032	.974	.057	1.181	.238
Education level (>9 years vs. ≤9 years)	−.104	−2.159	.031	.013	.262	.794	−.090	−1.860	.064
Employment status (employed vs. unemployed)	−.020	−.419	.675	−.040	−.825	.410	−.018	−.379	.705
Monthly income level (≥3,000 Yuan vs. <3,000 Yuan)	−.188	−3.941	.000	−.012	−.252	.801	−.135	−2.795	.005
BMI, kg/m^2	.095	1.956	.051	.016	.323	.747	.094	1.940	.053
Smoking (yes vs. no)	.028	.568	.570	−.154	−3.199	.001	.124	2.572	.010
Drinking (yes vs. no)	.075	1.546	.123	−.013	−.274	.784	.0.062	1.286	.199
Frequency of hair washing (≥4 times/week vs. <4 times/week)	−.007	−.147	.883	−.167	−3.491	.001	.100	2.065	.040
Using hair dryer, curling iron, or hair straightener (yes vs. no)	.026	.541	.589	−.036	−.733	.464	.056	1.155	.249
Using hair styling products (yes vs. no)	−.096	−1.983	.048	−.069	−1.416	.158	−.069	−.014	.775
Years since HIV diagnosis	.005	.106	.916	.001	.024	.981	−.002	−.031	.975
cART status
First-line cART vs. untreated	−.057	−.859	.391	.006	.086	.931	−.023	.351	.726
Second-line cART vs. untreated	.023	.344	.731	−.016	−.232	.816	.027	.412	.680
CD4 count (>500 cells/mm^3 vs.≤500 cells/mm^3)	.019	.392	.696	.080	1.649	.100	−.043	−.894	.372

β: standardized coefficient; BMI: body mass index; cART: combination antiretroviral therapy; Han Ethnicity is the majority Ethnicity; Monthly income less than 3,000 Chinese Yuan is the level of poverty.

Simple linear regression analysis revealed that hair cortisone levels, but not hair cortisol levels or hair cortisol to hair cortisone ratio, were associated with fatigue levels (Model 1: β = − 0.12, p = 0.013; β = −0.02, p = 0.73; β = 0.06, p = 0.22, respectively). As listed in Table 3, the association between hair GCs levels and fatigue levels was persisted in Model 2 and Model 3. However, the association between hair cortisol to hair cortisone ratio and fatigue levels became marginal in Model 2 and Model 3.

Table 3. Multivariate regressions of fatigue.

	Model 2			Model 3
	β	t	p	β	t	p
Hair cortisol	.023	.467	.641	−.014	−.259	.796
Age, years	−.009	−.171	.865	.011	.144	.886
Gender (female vs. male)	.015	.206	.837	.002	.034	.973
Ethnicity (No-Han vs. Han)	.003	.055	.956	−.029	−.573	.567
Marital status (married vs. other)	−.021	−.415	.679	−.018	−.340	.734
Education level (>9 years vs. ≤9 years)	−.010	−.193	.847	.176	3.508	.001
Employment status (employed vs. unemployed)	.166	3.357	.001	.103	1.950	.052
Monthly income level (≥3,000 Yuan vs. <3,000 Yuan)	.100	1.900	.058	.094	1.891	.059
BMI, kg/m^2	.088	1.794	.074	.029	.480	.632
Smoking (yes vs. no)	.023	.397	.691	−.013	−.229	.819
Drinking (yes vs. no)	−.013	−.231	.818	−.023	−.373	.709
Frequency of hair washing (≥4 times/week vs. <4 times/week )	−.025	−.397	.691	−.020	−.375	.708
Using hair dryer, curling iron or hair straightener (yes vs. no)	−.008	−.156	.876	.090	1.818	.070
Using hair styling products (yes vs. no)	.092	1.872	.062	−.035	−.710	.478
Years since HIV diagnosis	–	–	–	.052	.772	.440
cART status
First-line cART vs. untreated	–	–	–	.115	1.685	.093
Second-line cART vs. untreated	–	–	–	−.026	−.505	.614
CD4 count (>500 cells/mm^3 vs.≤500 cells/mm^3)	–	–	–	−.014	−.259	.796
Hair cortisone	−.121	–2.439	.015	−.115	−2.320	.021
Age, years	.001	.024	.981	−.004	−.077	.939
Gender (female vs. male)	.046	.645	.520	.040	.540	.589
Ethnicity (No-Han vs. Han)	.005	.094	.925	.004	.087	.931
Marital status (married vs. other)	−.021	−.420	.674	−.029	−.574	.567
Education level (>9 years vs. ≤9 years)	−.006	−.120	.905	−.016	−.291	.771
Employment status (employed vs. unemployed)	.165	3.365	.001	.174	3.489	.001
Monthly income level (≥3,000 Yuan vs. <3,000 Yuan)	.099	1.928	.054	.103	1.976	.049
BMI, kg/m^2	.089	1.829	.068	.095	1.920	.056
Smoking (yes vs. no)	.015	.256	.798	.019	.321	.749
Drinking (yes vs. no)	.001	.026	.980	.001	.022	.982
Frequency of hair washing (≥4 times/week vs. <4 times/week )	−.027	−.446	.656	−.026	−.419	.675
Using hair dryer, curling iron or hair straightener (yes vs. no)	−.018	−.336	.737	−.028	−.518	.605
Using hair styling products (yes vs. no)	.083	1.688	.092	.081	1.641	.102
Years since HIV diagnosis	–	–	–	−.036	−.737	.461
cART status
First-line cART vs. untreated	–	–	–	.051	.773	.440
Second-line cART vs. untreated	–	–	–	.108	1.587	.113
CD4 count (>500 cells/mm^3 vs.≤500 cells/mm^3)	–	–	–	−.018	−.361	.718
Hair cortisol/hair cortisone	.087	1.763	.079	.084	1.683	.093
Age, years	−.010	−.193	.847	−.015	−.282	.778
Gender (female vs. male)	.021	.301	.764	.017	.232	.817
Ethnicity (No-Han vs. Han)	−.001	−.021	.983	−.002	−.031	.975
Marital status (married vs. other)	−.025	−.503	.615	−.033	−.651	.515
Education level (>9 years vs. ≤9 years)	−.006	−.110	.913	−.014	−.266	.790
Employment status (employed vs. unemployed)	.165	3.348	.001	.174	3.483	.001
Monthly income level (≥3,000 Yuan vs. <3,000 Yuan)	.107	2.061	.040	.110	2.099	.036
BMI, kg/m^2	.094	1.906	.057	.099	1.999	.046
Smoking (yes vs. no)	.016	.269	.788	.021	.350	.727
Drinking (yes vs. no)	−.011	−.202	.840	−.011	−.200	.841
Frequency of hair washing (≥4 times/week vs. <4 times/week )	−.031	−.497	.620	−.029	−.468	.640
Using hair dryer, curling iron or hair straightener (yes vs. no)	−.018	−.330	.741	−.029	−.531	.596
Using hair styling products (yes vs. no)	.092	1.873	.062	.090	1.821	.069
Years since HIV diagnosis	–	–	–	−.035	−.713	.476
cART status
First-line cART vs. untreated	–	–	–	.054	.805	.421
Second-line cART vs. untreated	–	–	–	.112	1.650	.100
CD4 count (>500 cells/mm^3 vs. ≤500 cells/mm^3)	–	–	–	−.023	−.458	.647

β: standardized coefficient; BMI: body mass index; cART: combination antiretroviral therapy; Han Ethnicity is the majority Ethnicity; Monthly income less than 3,000 Chinese Yuan is the level of poverty.

4. Discussion

The current study examined the association between hair GCs levels and fatigue levels among a large cohort of Chinese PLHIV. We found that PLHIV in Guangxi, China showed similar fatigue levels (MOS-HIV fatigue scores M ± SD: 62.23 ± 16.19) with PLHIV in Hongkong (57.40 ± 18.56) (Lau et al., 2006), but lower fatigue levels than PLHIV in Zimbabwe (47.78 ± 25.11, 41.87 ± 23.89) (Taylor et al., 2008) and United States (51.84.78 ± 22.47) (Vosvick et al., 2003). In addition, given that the fatigue scores below 50 indicate limitations or disability related to fatigue among the general population and breast cancer survivors (Hays et al., 1993; Ware & Sherbourne, 1992), our finding with 36.2% of PLHIV had fatigue scores below 50 is consistent with a recent systematic review that reported HIV-related fatigue prevalence varied from 33 to 88% in PLHIV (Jong et al., 2010).

We found a nonsignificant association between hair cortisol levels and fatigue levels, which is inconsistent with Barroso et al. finding that reported an association of an upward trend in the salivary cortisol levels with the highest fatigue severity index in PLHIV (Barroso et al., 2006). Interestingly, we found a positive association between hair cortisone levels and fatigue levels. Previous studies indicated that hair cortisone levels were significantly correlated with salivary cortisone levels (Zhang et al., 2018), and cortisone levels were more closely approximate unbound, biologically active cortisol levels than total cortisol levels (Perogamvros et al., 2010). Therefore, hair cortisone levels may provide a useful surrogate for assessing long-term free cortisol levels and representing the accumulative activity of the HPA axis. In this context, regardless of the differences between our study and Barroso et al. study in sample size (n = 40 vs. n = 425), the time window of the GCs exposure (short-term vs. long-term), and fatigue measures (MOS-HIV fatigue scale vs. HIV-related fatigue Scale), both studies implied a positive relationship between GCs levers and fatigue levels in PLHIV. Furthermore, we found a marginal association between hair cortisol to hair cortisone ratio and fatigue levels, which is inconsistent with previous findings in CFS patients that reported no significant difference in salivary or urinary cortisol to cortisone ratio between CFS patients and healthy controls. While the hair cortisol to hair cortisone ratio’s physiological significance is still needed to address, our finding indicates that cortisol metabolic, one of the critical pathways to regulate the HPA axis activity, may also involve the pathophysiology and the biological mechanisms of fatigue in PLHIV.

We found hair cortisone levels were higher than hair cortisol levels in PLHIV, and two hair GCs levels were positively correlated, which is similar to the findings in previous studies (Kuehl et al., 2015; Stalder et al., 2013; van den Heuvel et al., 2020; Zhang et al., 2018). In addition, both our and previous studies showed that hair cortisone levels have some benefits to hair cortisol levels, which indicated that hair cortisone levels might be a useful biomarker for assessing the relationship between the HPA axis dysregulation and fatigue in PLHIV. For example, both our study (see Table 2) and previous studies found that hair cortisol was influenced by additional factors (Rippe et al., 2016; Staufenbiel et al., 2015). Moreover, some studies also have found stronger associations between hair cortisone levels and variables studied than hair cortisol levels did, including Parkinson’s disease (van den Heuvel et al., 2020), Cushing’s syndrome (Savas et al., 2019), cardiometabolic variables (Feeney et al., 2018; Stalder et al., 2013), and stress-related variables (Davison et al., 2019; van den Heuvel et al., 2020). All this may not only partly explain why hair cortisone levels, but not hair cortisol levels, were associated with fatigue levels in our study but also provides implications for future research to consider both hair cortisol and cortisone levels to represent the long-term activity of the HPA axis.

We found that clinical characteristics were nonsignificantly associated with hair GCs levels, which is in line with previous studies that reported a nonsignificant association of cART status (Langerak et al., 2015; Qiao et al., 2017) or CD4 counts (Langerak et al., 2015) with hair cortisol levels in PLHIV. In addition, our results not only contribute to the HIV research by exploring the association of hair GCs levels with sociodemographic and lifestyle variables in a large cohort study of PLHIV but also add to the other field by confirming the association of hair GCs levels with those variables previously observed in other populations (Braig et al., 2015; Feller et al., 2014; Fischer et al., 2017; Staufenbiel et al., 2015).

Our study includes one of the largest investigations of the association between hair GCs levels and fatigue levels to data. Several limitations of this study may constrain the generalizability of its findings. First, while hair GCs levels have been considered as novel biomarkers of chronic stress, our study lacks a measure of chronic stress (e.g. perceived stress) as the gold standard for hair GCs in this large sample. Second, our study cannot make causal inferences due to the cross-sectional design. Future research should benefit from using longitudinal designs to investigate whether the cortisone change is consistent in PLHIV who experienced fatigue and whether the possible cause of fatigue in PLHIV is abnormalities of the GCs. Third, we employed the MOS-HIV fatigue scale to assess overall fatigue in the past four weeks. Future research should benefit from using validity fatigue scales (e.g. HIV-related fatigue scale), including several essential components of fatigue (e.g. intensity, chronicity, and subtypes of fatigue) (Barroso & Lynn, 2002; Bormann et al., 2001). Fourth, while hair GCs levels have advantages in reflecting the long-term activity of the HPA axis, diurnal rhythm patterns and levels of GCs in other biological samples (e.g. plasma, saliva, and urine) and a dexamethasone suppression test as a further measure of HPA function should be considered in the future study (Jiang et al., 2019). For example, the dexamethasone suppression test and the measurement of cortisol awakening response and diurnal cortisol slope would have been very informative to see whether PLHIV had dysregulated HPA axis function. To investigate the associations between those measurements of GCs levels and fatigue levels will give us a full picture of the relationship of HPA axis function with fatigue in PLHIV. Fifth, because all participants were from Guangxi, China, our findings may not be generalizable to PLHIV in other settings. Finally, some other potential factors (e.g. physical activity) that might influence hair GCs levels and fatigue levels should be included in the future study. Previous studies indicated that physical activity is associated with increased hair cortisol in young adults (Gerber et al., 2013) but reduced fatigue in PLHIV (Webel et al., 2016). Therefore, further study should include physical activity and investigate the relationship between physical activity, cortisol levels, and fatigue levels among PLHIV.

5. Conclusion

In summary, we found higher hair cortisone levels were associated with higher fatigue levels in a large cohort of PLHIV. Future work will focus on the longitudinal relationship between multiple HPA function measurements (e.g. diurnal rhythm patterns, long-term GCs exposure) and fatigue levels assessed by validity scales with more essential fatigue components.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the National Institutes of Health (NIH) Research Grant [Grant numbers R01HD074221, R21AI122919].