The impact of COVID-19 pandemic and lockdown on chronic stress assessed by hair cortisol analysis via LC-MS/MS

Chemicals

The reference standard cortisol and its deuterated internal standard, cortisol-d₄ (IS), were procured from Cayman Chemical (Ann Arbor, MI, USA) at respective quantities of 1 g and 0.5 mg. Ultrapure water (H₂O), acetonitrile (ACN), 98 % formic acid (HCOOH) and dichloromethane (DCM), were obtained as analytical grade reagents from Carlo Erba (Milan, Italy). Methanol (MeOH) of HPLC grade (99.9 %) was purchased from Sigma-Aldrich (Milan, Italy). LC/MS grade MeOH was obtained from Merck KGaA (Darmstadt, Germany).

Equipment

Analyses were performed on a 1290 Infinity UHPLC system (Agilent Technologies, Palo Alto, CA, USA) coupled to a QTRAP 5500 triple quadrupole linear ion trap MS (Sciex, Darmstadt, Germany), equipped with an ESI source. Compounds were separated on a Kinetex LC XB-C18 column (100 mm length × 2.1-mm i.d, 2.6 µm particle size) (Phenomenex, Torrance, CA, USA) at 30 °C using 0.1 % formic acid in H₂O (A) and 0.1 % formic acid in ACN (B) as mobile phases, with the following 10-min linear gradient: 0.0–0.50 min (85 % A, 15 % B), 0.50–2.0 min (75 % A, 25 % B), 2.0–5.0 min (65 % A, 35 % B), 5.0–6.5 min (2.0 % A, 98 % B), 6.5–8.0 min (2.0 % A, 98 % B), 8.0–8.1 min (85 % A, 15 % B), 8.1–10 (85 % A, 15 % B). The flow rate was 0.5 mL/min and the injection volume was 5 µL. The MS parameters were optimized by direct infusion of a standard mix solution (cortisol and cortisol-d₄, 100 ng/mL) as follows: curtain gas, ion gas 1, and ion source gas 2 were set at 25, 50, and 55 psi (pound per square inch) respectively, the source temperature was 500 °C, the ionization voltage was 5,500 eV, the entrance potential was 6 and 8 eV for cortisol and its deuterated, respectively. MS acquisitions were performed by multiple reaction monitoring (MRM) mode. The MRM conditions and parameters including ion transitions, de-clustering potential (DP), and collision energy (CE) are provided in Table 1. Data acquisition and processing were performed using Analyst^®1.7.1 and Sciex OS-Q^®1.6 software (Sciex, Darmstadt, Germany), respectively.

Cortisol validation procedure

Method validation was conducted in accordance with international recommendations for the validation of analytical methods [22]. Selectivity was assessed using cortisol-free distal hair samples (25 mg) defined as hair longer than 30 cm, with and without the addition of IS (blank and zero samples). Linearity was assessed by analysis of eight calibration levels (0, 5, 10, 20, 50, 100 and 200 pg/mg) with seven replicates (n=7) for each level. Linearity was determined by plotting the peak area ratios of the analytes to the IS against the nominal analyte concentration, employing a weighted 1/x2 linear regression. Sensitivity was expressed in terms of limit of detection (LOD) and limit of quantification (LOQ). The LOQ was defined as the lowest concentration at which precision and accuracy values were within ± 20 % and a signal-to-noise (S/N) ratio of the peak areas was ≥10. The LOD was defined as the lowest concentration with an S/N ratio of the peak areas ≥3. Precision and accuracy were determined at all calibration levels through the analysis of seven independent replicates of quality control (QC) samples, calculating the coefficient of variation (CV %) and the bias (BIAS %). CV and BIAS values below 15 % were deemed acceptable. The matrix effect (%) was determined by fortifying cortisol-free distal hair extracts and MeOH with increasing cortisol concentrations (5, 10, 20, 50, 100, 200 pg/mg, n=7 for each level) and comparing the mean peaks areas. The stability of cortisol in QCs extracts was assessed by comparing the response factor after 24 h, 48 h, 72 h and 7 days at room temperature, after 7 days at 4 °C and after one freeze and thaw cycle with the original vial at T0.

Hair samples collection

Cortisol-free human hair samples were collected from volunteers and analyzed to exclude any source of chromatographic interferences. These samples were then mixed to obtain homogeneous pools of blank hair for cortisol calibrators and QCs. Hair was collected from 87 subjects tested in our laboratory during 2020. Each subject provided one proximal segment (3–6 cm from the scalp). When sufficient hair length was available, the sample was further divided into three consecutive segments, resulting in a total of 240 hair samples. Segments were classified according to the period of hair growth they represented, as follows: the most proximal segment reflecting the period following lockdown (2–3 months post-lockdown, post-L), the intermediate segment corresponding to the Italian first lockdown period (March 9th to May 18th, L), and the distal segment representing the pre-lockdown period (before March 9th, pre-lockdown, pre-L) (Figure 1). In instances of insufficient hair length, samples were divided into only two segments, corresponding to the Italian lockdown and the post-lockdown periods. All subjects included in this study provided written informed consent for the anonymous use of their samples for research purposes.

Figure 1:

Schematic workflow of hair sample collection.

Sample preparation and extraction

All collected hair samples, obtained as described in the previous paragraph, were initially washed with 4 mL of DCM and 2 mL of MeOH (10 min) in ultrasonic bath subsequently dried at room temperature, and cut into small pieces (2–4 mm) using scissors. An aliquot (20–120 mg) from each sample was accurately weighed into 1.5 mL Eppendorf tubes and used for cortisol extraction.

Statistics

Statistical analyses were performed using GraphPad Prism 7 (Graph- Pad Software, CA, USA). Hair samples (n=87) were categorized into three distinct groups: Pre-L, L, and Post-L. Data for each group and increments observed between groups are presented as box and whisker plots. In these plots, the box delineates the 25^th to 75^th interquartile range, with the horizontal line within the box representing the median value. The whiskers extend to the extreme values, excluding outliers, which are represented by individual dots. All reported information on median and standard deviation values is expressed in ng/mg. The cortisol content among the three groups was statistically analyzed using a completely randomized ANOVA, following a logarithmic transformation of the data to achieve normality. This was subsequently followed by orthogonal comparisons. For all statistical tests conducted, the level of significance was set at p<0.05.

Instrument conditions optimization

Optimization of the MS instrumental conditions was achieved by infusing a standard solution of cortisol and its deuterated analogue in acetonitrile, as described above. Although methods described in the literature often utilize both positive and negative ionization modes, we obtained higher intensity and superior signal symmetry using the [M+H]⁺ precursor ion for both cortisol (m/z 363.3) and cortisol-d₄ (m/z 367.3). Consequently, the positive ionization mode was selected for our analysis. Following the optimization of the Declustering Potential (DP) and Entrance Potential (EP) parameters related to the aforementioned precursor ions, fragmentation analysis was performed to identify the most suitable product ions. Among the fragmented ions, those exhibiting the highest intensity and optimal peak shape were selected, alongside setting the source temperature and gas pressure parameters (curtain, 1 and 2). The chromatographic separation was performed using a gradient elution, with H₂O containing 0.1 % HCOOH as mobile phase A and ACN containing 0.1 % HCOOH as mobile phase B, as detailed in paragraph “Equipment”. The gradient was specifically designed to ensure that cortisol and its deuterated analogue eluted approximately midway through the chromatographic run. This strategic elution time helps to prevent the signal from being affected by potential interferences that typically emerge during the initial minutes of the analysis.

Cortisol method validation

A complete validation was performed as described in paragraph “Cortisol validation procedure”. Given that cortisol is an endogenous compound, obtaining a cortisol-free hair matrix is a critical aspect for the validation of the analytical method. Therefore, the cortisol calibration curve was constructed using a pool of distal hair samples (from hair strands longer than 30 cm). This choice was based on the assumption that cortisol concentration is expected to be very low or absent in hair tips due to potential wash-out or damage effects. Initially, the collected distal hair samples were analyzed to confirm the absence of endogenous compounds, and an example chromatogram is provided in chromatogram A of Figure S1 (Supplementary Material S1). Moreover, a previous study by Binz et al. [23] suggests good linearity for both cortisol and (13)C₃-labeled cortisol, as well as minor differences in the limit of quantification (LOQ) (cortisol: 1 pg/mg vs. (13)C₃-cortisol 0.5 pg/mg). In accordance with the authors, we achieved a LOQ of 1.03 pg/mg. The calibration curves showed excellent linearity (R²>0.998) over the entire investigated range when using linear correlation (Table 2). Extracted ion chromatograms of the quantifier ions (cortisol: m/z 363.3>121.5 and cortisol-d₄: m/z 367.3>121.2) for QCs at low, intermediate, and high concentration levels are presented in Figure S1 (Supplementary Material S1). The LOD and LOQ values were deemed adequate for the objectives of the present study and are detailed in Table 2. Inter-day precision (expressed as coefficient of variation) and accuracy (expressed as percentage error) were consistently below 15 % (Table 3). Repeatability was evaluated by measuring relative standard deviations (RSDs) at each concentration levels for n=7 curves and results are presented in Table 3. Sample stability tests were performed as previously described, and the cortisol concentration was found not to be significantly reduced. Specifically, samples analyzed after 24 h, 48 h, 72 h, and 7 days at room temperature, after 7 days at 4 °C, and after 7 days at −20 °C showed a mean variation from the nominal concentration of less than 15 % (Table 4). These data confirm that samples remained stable under the storage conditions employed. The matrix effect (%) ranged from −8.0 to +9.0 %, indicating a negligible impact of the hair matrix on the ionization process.

Analysis of authentic hair samples

A total of 240 hair segments from 87 individuals were analyzed. Demographic data of the cohort, including sex distribution and age, are reported in Table 6. When possible, hair was divided into three segments corresponding to Pre-L, lL, and Post-L periods. In cases of insufficient length, only L and Post-L segments were available. Cortisol was reliably quantified above the LOQ in 62 subjects (71 %), while 25 samples were below LOQ (Table 6). Median cortisol concentrations were 2.28 pg/mg in Pre-L, 3.06 pg/mg in L, and 3.68 pg/mg in Post-L segments (Table 5). ANOVA revealed statistically significant differences between Post-L and both Pre-L and L (Figure 2, left). Pairwise comparisons confirmed that individual cortisol concentrations were significantly higher in Post-L compared to both Pre-L and L segments (Figure 2, right). In 45 individuals with complete three-segment profiles (Pre-L, L and Post-L), 76 % exhibited higher cortisol in Post-L than in Pre-L.

Figure 2:

Box-and-whisker plots of cortisol concentrations: Left, distribution across pre-lockdown (Pre-L), lockdown (L), and post-lockdown (Post-L) segments; right, within-individual differences across the three periods (ANOVA with Tukey’s post hoc test).

Subgroup analyses revealed pronounced differences by sex and age, as reported in Table 7. During lockdown, cortisol increased in 93 % of females compared to 50 % of males. In the 17–24-year age group, 89 % showed an increase, confirming young adults as the most affected subgroup. Post-L, cortisol remained elevated in most females (71 %), while males were more evenly distributed between increases and decreases. Age-related analysis indicated that although increases were most marked in younger adults, elevated cortisol was observed across all age groups.

Additional testing of hair samples for chronic alcohol and cannabis use (ethylglucuronide and Δ⁹-tetrahydrocannabinol, EtG and THC) [24], 25] yielded few positives (9 for EtG, 11 for THC). Among positives, EtG concentrations increased in 75 % (Pre-L vs. L) and 67 % (L vs. Post-L), while THC decreased in 86 and 82 % of cases across the same comparisons. Due to the low number of positives, no correlations with cortisol were performed.

Limitations

Several limitations of this study should be acknowledged. First, cortisone, an important metabolite of cortisol and frequently included in hair steroid analyses, was not measured. Future studies should consider both cortisol and cortisone to provide a more comprehensive evaluation of hypothalamic–pituitary–adrenal axis activity. Second, no additional covariates such as employment status, psychological disorders, or self-reported stress measures were collected, limiting the ability to directly associate cortisol variations with specific psychosocial factors. Third, distal hair segments may be affected by partial wash-out of endogenous cortisol, which could lead to underestimation of concentrations in older time windows [26], [27], [28], [29]. Fourth, the number of subjects positive for EtG or THC was too small to allow meaningful correlation analyses with cortisol levels. Finally, although our method was fully validated and robust, the exploratory nature of this cohort study calls for replication in larger, independent populations.

Strength

A key strength of this study lies in the combination of segmental hair analysis with a fully validated LC-MS/MS method for cortisol quantification. This approach enables a non-invasive, retrospective, and time-resolved assessment of chronic stress from a single biological sample. By analyzing hair segments corresponding to different pandemic phases, the method provides a unique temporal profile of stress exposure, offering high analytical sensitivity and specificity. This makes it particularly valuable for both clinical and population-level studies investigating the long-term psychological impact of external stressors such as the COVID-19 pandemic.