To optimize extraction yield, hair was completely dissolved through enzymatic digestion. Once the samples were prepared, the final extracts were then analyzed using liquid chromatography-tandem mass spectrometry, as this is the current best practice for cortisol analysis. The workflow is depicted in Fig. 1. We used a two-dimensional liquid chromatography-tandem mass spectrometry (2D LC–MS/MS) method for quantifying cortisol in the digested hair samples. Experiments were then conducted to compare the extraction yield of this new approach with methanol extraction of minced hair. After optimization, the protocol was then validated.

Fig. 1

Schematic depiction of sample processing steps

Reagents

Cortisol (Cerilliant, Round Rock, TX, USA), 13C3-cortisol (99% 13C), 1,4-dithioerythritol (DTE) (Sigma-Aldrich, Darmstadt, Germany), Tris base (Roche, Basel, Switzerland), potassium chloride, and proteinase K (Merck, Darmstadt, Germany), along with isopropanol (ULC/MS grade), tert-Butyl methyl ether (MTBE, HPLC grade), methanol (ULC/MS grade), and formic acid (ULC/MS grade) (Biosolve, Valkenswaard, The Netherlands), were all used in the experiments. Furthermore, ammonium acetate (HPLC grade, Honeywell Fluka, Charlotte, NC, USA) and sodium dodecyl sulfate (SDS, Fisher Scientific, Waltham, MA, USA) were also used. For each experiment, the digestion buffer was freshly prepared, containing 2 mg/mL of proteinase K, 6 mg/mL of DTE, and 20 mg/mL of SDS in a Tris buffer (0.1 M, pH 7.2).

Sample collection

For the validation experiments, hair samples were acquired, following informed consent, from adult female volunteers (n = 26) and drop-outs of a clinical study (n = 43). In this study, healthy lactating females from the Netherlands, all aged over 18 years, were included. A lock of hair (consisting of 100–200 strands) was snipped with scissors from the posterior vortex as close to the scalp as possible. Each hair sample was carefully attached to a piece of paper with tape to indicate the scalp end. The samples were stored in envelopes at room temperature in a dark closet until analysis [8]. The set of samples included a variety of hair types with varying thickness, texture (ranging from straight and fine to curly and coarse), color (from blond, brown, black, red, to gray), and also some dyed hair samples. For analysis, an average of 10.6 mg of hair was prepared from each sample.

Standard preparation

Stock solutions of cortisol and 13C3-cortisol (internal standard) were prepared in ethanol and stored at 4 °C. A calibration curve was created by adding dilutions of the cortisol stock to a fixed amount of the internal standard, to cover a relevant range (6–574 pg). Calibration standards were dried under a gentle stream of nitrogen and reconstituted in 60 µL of 20% methanol in water.

Hair results are initially computed as an absolute amount in picograms (pg), which is subsequently corrected for the precise weight used in the preparation and reported as picograms per milligram of hair (pg/mg hair). For easier calculation and correction of the initial results, the calibration curve covers a range from 6 to 574 pg. This represents the quantity of cortisol used in the preparation, without any correction for the hair weight.

Control samples

A significant quantity (approximately 10 g) of hair was gathered, using only the first 3 cm measured from the scalp. It was then washed with isopropanol. After stirring it for 1 min, the washing fluid was poured off, and the hair was left to dry for at least 4 days at room temperature in a fume hood to make sure of complete evaporation. The hair was finely cut and mixed to ensure a homogeneous pool. Note that this cutting process was only to guarantee the uniformity of a large volume of hair used as a control pool for quality control purposes and deviates from the standard preparation protocol for hair samples, where intact locks are taken into preparation. The finely cut, well-mixed pool of hair was stored in room temperature conditions without light exposure. Aliquots were weighed separately on analysis days to serve as control samples.

Sample preparation protocol

The first 3 cm of hair lock, measured from the scalp, was cut and transferred into a glass tube. Isopropanol (2 mL) was added to each sample, which was then vortexed for 1 min. After centrifugation for 1 min at 3000 × g, the fluid was decanted, and as much liquid as possible was removed from the tubes. The outer walls of the tubes were then covered with aluminum foil; this reduced exposure to light to prevent potential cortisol degradation while keeping the opening free to allow any leftover isopropanol to evaporate. The samples were placed at room temperature in a fume hood for 4 days.

The washed hair locks were accurately weighed into silanized vials using an analytical balance (ME204 T, Mettler Toledo, Netherlands). An internal standard (10 µL of 23.5 pg/µL 13C3-cortisol in ethanol = 235 pg absolute) and 1 mL of digestion buffer were added to each vial. The samples were then incubated at 40 °C for 16–40 h. After incubation, the digested hair samples were transferred into borosilicate tubes. A small amount of potassium chloride (about 0.15–0.2 g) was added and thoroughly mixed by vortexing. The samples were then incubated at 4 °C for 1 h.

To extract cortisol from the digest, 2 mL of MTBE was added. The samples were then vortexed for 1 min, incubated at − 20 °C for 5 min, and centrifuged for 5 min at 4 °C with 3000 × g. The top layer was transferred into a clean silanized vial, dried under a gentle nitrogen stream at 50 °C, and reconstituted in 60 µL of 20% methanol in water. This was then transferred into a total recovery vial and analyzed on 2D LC–MS/MS.

2D LC–MS/MS

A Waters LC–MS system (Waters Acquity UPLC equipped with a binary solvent manager, a quaternary solvent manager, a sample manager, a column heater, and a Xevo TQ-S, Waters, Milford, USA) was utilized for all analyses. This system was operated in 2D LC mode, featuring a Protein BEH C4 column (1.7 µm, 2.1 × 50 mm, Waters) as the first column, and an Acquity HSS T3 column (1.8 µm, 2.1 × 50 mm, Waters) as the second column. After injecting a 10 µL sample, separation was achieved by a gradient with two mobile phases: mobile phase A, composed of 2 mM ammonium acetate and 0.1% formic acid in water, and mobile phase B, composed of 2 mM ammonium acetate and 0.1% formic acid in methanol. The LC gradients can be found in the supplementary material. Between 4 and 5 min, the effluent of the first column was directed to the second column, with cortisol eluting after 7.4 min. The total duration per process was 10.5 min.

Cortisol and 13C3-cortisol were recorded by the Xevo TQ-S, which operated in MRM mode (transitions were 363 > 121 and 366 > 124, respectively). Data acquisition was carried out using MassLynx V4.2 SCN1001. Additional details on the instrument settings can be found in the supplementary material.

Method validationPrecision

The intra-day precision was determined by analyzing a homogenized control pool that was prepared repeatedly on a single day (n = 16). The inter-day precision was assessed by analyzing this same pool, which was prepared repeatedly on different days (5 days, n per day ≥ 2). From these data, we calculated the coefficient of variation (CV%).

Additionally, hair locks from volunteers with various hair types were gathered. After washing, the hair was divided into two aliquots for duplicate analysis. Based on these results, the CV% was calculated.

Recovery, matrix effect

We cannot determine the true extraction yield, as there is no means to verify the exact endogenous concentration in hair samples. Current techniques may be biased, and spiking does not accurately represent endogenous hair cortisol. By completely digesting the hair and eliminating any solid structure, we are confident that we have maximized the extraction yield, if not achieved 100%.

The recovery of sample preparation was assessed by adding IS before and after preparation of a hair sample. The matrix effect was evaluated through the analysis of a prepared hair sample, spiked with IS after sample preparation, and the analysis of the same amount of IS added to the solvent.

Additionally, a spiking experiment was performed, in which several locks of hair were divided into aliquots. For each lock of hair, one unspiked aliquot was analyzed to determine the concentration in the respective hair sample, while two or three aliquots were spiked with cortisol before digestion.

The equations required for the calculation of these parameters are presented as supplementary material.

Linearity

The standard curve ranged from 6 to 550 pg, an R2 of > 0.99 is acceptable. Sample weights ranging from 5.8 to 38.6 mg of hair from the pool were analyzed to determine whether the sample size affected the outcome. The measurement results were then plotted against the theoretical outcomes based on the starting material’s weight. A slope between 0.9 and 1.1 was deemed acceptable, along with an intercept of less than 25% of the lowest observed result.

To determine the accuracy of the final extract’s dilution, a sample with a high cortisol level was diluted in a series: 1:1, 1:3, 1:7, 1:15, and 1:31 (sample:20% methanol in water).

Limit of quantification

The limit of quantification was determined from a tenfold analysis of both a sample and a standard with a low cortisol concentration. A CV% of less than 20% is acceptable for the lower limit of quantification.

Stability

The stability of the samples was assessed over a relevant time frame after being prepared. Multiple samples were prepared and pooled to ensure ample material for repeated injections on days 0, 2, 7, 9, and 14 post-preparation. In between runs, these samples were stored at 4 °C. The CV% of these results should be less than 10%, consistent with the acceptable limit for inter-day precision.

Carryover

Carryover was assessed using the statistical module specifically designed for carryover on the EP Evaluator software (Data Innovations LLC, Colchester, USA, Build 12.3.0.2), based on the Clinical & Laboratory Standards Institute (CLSI) EP10 guideline. Two samples, one with a high and one with a low concentration of cortisol, were repeatedly analyzed in a prescribed order according to the module’s instructions.

Comparison to methanol extraction

To compare the results of digestion to those obtained through the commonly utilized methanol extraction technique, eight locks of hair were divided into two aliquots after the isopropanol washing step. One aliquot from each lock underwent digestion following the procedure detailed in the “Sample preparation protocol” section, while the other was treated using the following method:

The hair was finely cut with scissors and then incubated in 1 ml of methanol for 16 h at 52 °C. Afterward, the samples were cooled and centrifuged for 5 min at 3000 × g. The supernatant was decanted and dried under a gentle stream of nitrogen. This methanol extraction was based on the publication by Noppe et al. [22]. The remaining steps were based on a novel sample preparation procedure to ensure result comparability: The residue was dissolved in 0.5 mL water, onto which 2 mL of MTBE was added. The samples were then vortexed for 1 min and centrifuged for 5 min at 4 °C 3000 × g. The top layer was transferred into a clean silanized vial, dried under a gentle stream of nitrogen at 50 °C, and reconstituted in 60 µL of 20% methanol in water. This was then transferred into a total recovery vial and analyzed with 2D LC–MS/MS.

Data analysis

The peak area was integrated using TargetLynx software version V4.2 SCN1001 (Waters, USA). Further calculations were carried out on Excel 2016 (Microsoft Corporation, USA). The slope and intercept of the standard curves were derived from linear regression analysis. For the comparison of methods and evaluation of volunteers’ results, MedCalc® Statistical Software version 22.016 (MedCalc Software Ltd, Ostend, Belgium; https://www.medcalc.org; 2023) was used.