We conducted an observational pilot study with inpatients recruited from the University Hospital Basel, Switzerland. The trial lasted from September 2018 to June 2019. The patient sample size was chosen to generate the first pilot data with a reasonable amount of resources and time frame. The Ethics Committee of Northwest and Central Switzerland (EKNZ Reg-ID 2018-01155) approved this study, and all patients provided written informed consent. The ClinicalTrials.gov registration number is NCT03678142.

The study included patients who had been treated with intravenous β-lactam ABs (cefepime, imipenem, and flucloxacillin) for >24 h, had an estimated glomerular filtration rate (eGFR) >50 mL·min⁻¹·1.73 m⁽²⁾⁻¹ and were >18 years old. Patients were excluded if they were allergic to pilocarpine or were receiving continuous oxygen supplementation.

Sweat samples were collected 4 h after starting the last imipenem and flucloxacillin infusions. For cefepime, the sweat and blood samples were collected at baseline (BL, before) and 2, 4, 6, and 8 h after administration. Sweat was induced and collected using a standardized method. Briefly, local eccrine sweat glands of the volar lower arm were stimulated for 5 min using pilocarpine iontophoresis. Sweat samples were collected using a CE certified Macroduct™ sweat collector for ~30 min and transferred to storage containers on dry ice (23). Sweat samples were stabilized by adding 1 μL Halt™ protease inhibitor cocktail (100×) and stored at −80°C.

During sweat gland stimulation in patients administered cefepime, we simultaneously collected reference blood samples, which were transported on dry ice and stored at −80°C. Standard laboratory parameters were included in the analysis (creatinine, eGFR, ALT, AST, gamma-glutamyltransferase, alkaline phosphatase, and bilirubin), and demographic data, such as age, weight, medication, and vital signs, were collected. Bio-fluid samples were transported on dry ice to Colorado State University, CO, USA. Bio-fluid analysis was conducted by the Proteomic and Metabolomic Facility of Colorado State University, CO, USA.

The mass spectrometer was operated in the multiple reaction monitoring (MRM) mode. In this mode, a parent ion is selected by the first quadrupole, fragmented in the collision cell, and then a fragment ion(s) is selected by the third quadrupole. Product ions, collision energies, and cone voltages were optimized for each analyte by direct injection of individual synthetic standards and their stable isotope-labeled internal standards (IS). Samples were injected into a reverse phase UPLC column, and target analytes and IS samples were eluted at specific retention times. A calibration curve was generated using authentic standards of each compound, and their corresponding stable isotope-labeled IS in 50% methanol:50% artificial sweat (https://www.pickeringtestsolutions.com/AP-eccrine/). The peak area values of target analytes were normalized to those of the appropriate IS and were plotted against expected concentrations ranging from 0.28 to 1,500 ng/mL.

To extract the ABs, blood samples were thawed on ice, and 100 μL of each was placed into a 2.0 mL polypropylene microfuge tube. Then, 900 μL ice-cold methanol (spiked with 275 ng/mL of cefepime-d3) was added to each sample and vortexed for 10 min at 4°C. Samples were stored at −80°C for a minimum of 1 h to facilitate protein precipitation. Precipitated proteins were collected by centrifugation at 15,000 × g for 20 min at 4°C, then 900 μL of the supernatant was collected into 2 mL autosampler vials.

Sweat samples (50 μL) were diluted 1:1 (v/v) with 100% methanol (spiked with 500 ng/mL meropenem-d6, cefepime d3, and flucloxacillin-13C4). Each sample was briefly vortexed, then 50 μL was transferred into autosampler vials.

Authentic standards of imipenem, flucloxacillin, and cefepime were diluted from 1 mg/mL library stocks to a 10 μg/mL master mix. Isotopically-labeled IS samples were diluted from 1 mg/mL library stocks to a 2.5 μg/mL IS-Master Mix, which was further diluted with artificial sweat (Pickering Laboratories, Mountain View, CA, USA) at a concentration of 500 ng/mL (IS-Artificial Sweat). To construct the standard curve, the standard master mix in 100% methanol was serially diluted 3.2 times from 25 ng/mL to 0.28 ng/mL in 100% methanol. Then, 0.1 mL of each methanol dilution was added to 0.1 mL of IS-Artificial Sweat, and the final concentration of the IS was 250 ng/mL.

LC-MS/MS was performed using a Waters Acquity UPLC coupled with a Waters Xevo TQ-S QqQ mass spectrometer. Chromatographic separations were conducted using a Waters UPLC T3 C18 stationary phase (1 × 50 mm, 1.7 μM) column. Mobile phases were 99.9% acetonitrile with 0.1% formic acid (B) and 99.9% water with 0.1% formic acid (A). The analytical gradient schedule was as follows: 0 min, 0.1% B; 1.0 min, 0.1% B; 2.5 min, 99% B; 3.5 min, 99% B; 3.55 min, 0.1% B; and 5 min, 0.1% B. The flow rate was 800 μL/min and injection volume was 5 μL. Samples were held at 6°C in the autosampler, and the column was operated at 45°C. The MS was operated in positive ionization mode with the capillary voltage set to 3 kV. The inter-channel delay was set to 3 ms, and the source and desolvation temperatures were 150 and 500°C, respectively. Desolvation and cone gas flow rates were 1,000 and 150 L/h, while the collision gas flow was 0.2 mL/min. The nebulizer pressure (nitrogen) was set to 7 bar, and argon was the collision gas. The MS acquisition functions were scheduled by retention time and provided 30 s windows. Autodwell feature was set for each function, and the dwell time was automatically calculated, while 12 points across a peak were specified as the minimum data points per peak, resulting in a minimum dwell time of 0.017 s.

All raw data files were imported into the Skyline open-source software package (24). Each target analyte was visually inspected for retention time and peak area integration. Peak areas were extracted for target compounds detected in biological samples and normalized to the peak area of the appropriate IS or surrogate in each sample. Normalized peak areas were exported to Excel, and absolute quantitation was obtained using the linear regression equation generated for each compound from the calibration curve. Limits of detection (LOD) and limits of quantification (LOQ) were calculated as 3 or 10 times the standard deviation (SD) of the blank divided by the slope of the calibration curve, respectively (25, 26).

Study-specific quality control (QC) samples were generated to represent the pool of all samples. They were injected into the system after every seven samples were analyzed.

We recruited 13 patients, from which four were treated with flucloxacillin and cefepime, and three were administered imipenem. One patient was treated with a combination of imipenem and cefepime and was included for both groups, but a protocol violation occurred for imipenem (AB treatment <24 h). However, the samples were still included because the study was observational (Figure 1). One cefepime sweat sample was lost because of sample mishandling; one pair of cefepime sweat and blood samples could not be collected because the patient did not attend the follow-up visit.

Patient characteristics are shown in Table 1 with a detailed summary available as Tables S1–S3. Patients were aged 34 to 90 years with kidney function (eGFR) values ranging from 58 to 124 mL·min⁻¹·1.73 m⁽²⁾⁻¹, and their indications for AB therapy varied from osteomyelitis to pneumonia.

In this study, cefepime, flucloxacillin, and imipenem were detected for the first time in all of the respective human eccrine sweat (Tables 2–4). Sweat sample volumes ranged from 25 to 50 μl.

The highest interpersonal AB sweat concentration variation was detected between patients treated with cefepime at 2 h with values of 0.041 mg/L (Patient C2) compared to 3.422 mg/L (Patient C1), corresponding to a variation factor of 84 (Table 2). The lowest interpersonal AB concentration variation was found between patients treated with imipenem with a variation factor of 3.6 (Table 4).

Cefepime AB concentration changes over time showed similar dynamics in blood and sweat (Figures 2A–F). The mean AB concentration of cefepime in sweat and blood ranged from 0.036 to 1.15 mg/L and 13.648 to 76·342 mg/L, respectively (Table 2). Sweat concentrations showed the highest variation 2 h after BL with a factor of 84 (Patients C1 and C2, Table 2) with comparable dynamics in concentration changes (Figure 2A). Patient C2 showed similar trends in samples analyzed at 2, 4, 6, and 8 h after BL (Figure 2B). In patient C3, after an earlier increase in sweat cefepime concentrations compared to blood concentrations, samples analyzed 4, 6, and 8 h after BL showed simultaneous changes in both sweat and blood. Importantly, patient C3 was administered the lowest cumulative daily AB dose (single dose of 1,000 mg at BL), which resulted in the lowest mean AB concentration in sweat and a prolonged concentration peak in the blood (Figure 2C). Patient C5 showed the highest total mean concentration of cefepime in the blood (76.342 mg/L), but concentration changes in blood and sweat of patient C5 showed a delay of 2 h compared to those of patients C1, C2, and C3 (Figure 2E).

In all patients treated with cefepime, the interpersonal factor of variability derived from the maximum peak concentration variability was higher in sweat (3.422 [maximum]/0.041 mg/L [minimum]) = 84 [factor]) than in blood (159.948/3.947 mg/L = 41). Analysis of the highest variation of trough levels detected in patients showed a slightly higher interpersonal concentration variability with sweat (0.074/0.007 mg/L = 11) than with blood (38.415/3.947 mg/L = 10). By dividing the patients into two groups based on age (>60 years) and kidney function (eGFR < 85 mL·min⁻¹·1.75 m⁽²⁾⁻¹), the concentration variability factor reduced substantially for maximum and trough levels. The maximum concentration variability factor of sweat decreased from an initial value of 84 to 2 (0.096/0.045 mg/L) in the younger patients and 5 (3.422/0.768 mg/L) in older patients. In blood samples, the maximum concentration variability factor decreased from 40 to 1 (34.667/33.630 mg/L) in young patients and 2 (159.948/67.016 mg/L) in older patients. A comparison of patient trough levels between the two age groups showed a sweat concentration variability factor of 2 (0.011/0.007 mg/L) in young patients and 1 (0.094/0.074 mg/L) in older patients. Serum trough levels showed a cefepime concentration variability factor of 1 (3.958/3.947 mg/L) in younger patients and 2 (38.415/20.285 mg/L) in older patients. Further, the gap between mean blood and sweat AB concentrations was substantially smaller in the older patients (patient C1: 41.210/1.150 mg/L = 36, patient C5: 76.342/0.292 mg/L = 261) than in younger patients (patient C2: 16.754/0.036 mg/L = 465, patient C3: 15.666/0.025 mg/L = Factor 627).

The elimination rate constant (K_e) showed comparable values in sweat and blood in younger patients with adequate GFR (patient 2: K_e was 0.16/h in sweat and 0.17/h in blood, patient 3: K_e was 0.14/h in sweat and 0.13/h in blood, patient 4: K_e was not calculated due to missing values). In older patients, K_e diverged widely compared to younger patients, and showed the widest divergence in the oldest patient (patient 1: K_e was 0.27/h in sweat and 0.09/h in blood, patient 5: K_e is 0.22/h in sweat and 0.14/h in blood).