In this study, certified reference materials were used. Linezolid (LIN, purity 98%), caspofungin (CAS, purity >90%), and teicoplanin (TEI, purity 98%) were purchased from TRC (Toronto, Canada). Daptomycin (DAP, purity 95%) and the internal standard (IS) daptomycin-d5 trifluoroacetic (DAP-d5, purity 95%) were obtained from Hengrui Pharmaceutical Co., Ltd. (Jiangsu, China). Piperacillin (PIP, purity 95.2%), [²H₅]-piperacillin sodium salt (PIP-IS, purity 95.2%), cefoperazone (CEF, purity 93.09%), and tazobactam (TAZ, purity 98%) were obtained from LGC Standards Ltd. (Molsheim, France). Ethylparaben (ETH, purity 99%, used as IS) was procured from Sigma Aldrich (Missouri, United States). Meropenem (MER, purity 98%) was purchased from Sumitomo Dainippon Pharma Co., Ltd. (Osaka, Japan). Fluconazole (FLU, purity 99.8%) and sulbactam (SUL, purity 99.5%) were purchased from ANPEL-TRACE Standard Technical Services Co. Ltd. (Shanghai, China). Vancomycin (VAN, purity 98%) was purchased from Vianex S.A. (Plant C) (Athens, Greece). Tigecycline (TIG, purity 96%) was purchased from Hisun Pharm Co. Ltd. (Zhejiang, China). Voriconazole (VOR, purity >99.5%) and [¹³C₂,²H₃]- voriconazole (VOR-IS, purity >99.5%) were purchased from Sichuan Meidakang Huakang Pharmaceutical Co. Ltd. (Sichuan, China). Posaconazole (POS, purity 99.9%) was obtained from CATO (Eugene, Oregon, United States). HPLC-grade acetonitrile (ACN), methanol (MeOH), and formic acid (FA) were purchased from Thermo Fisher Scientific (Waltham, MA, United States). Ultrapure water was generated using a Clever-S ultrapure water machine (Shanghai, China). Chromatography-grade methanol and acetonitrile were obtained from Tedia company (Ohio, United States). Dimethyl sulfoxide (DMSO) was purchased from Meryer Co., Ltd. (Shanghai, China).

Blood samples were obtained from healthy volunteers and ICU patients at Zhongnan Hospital of Wuhan University under the guidance of the ethical review committee [2022238K]. All participants signed an informed consent form. Serum samples were collected in a yellow separate glue coagulant tube and centrifuged at 4500 rpm (1940g) for 10 min; the supernatant was stored at −80°C until analysis.

The serum samples were thawed at room temperature, and 100 μL of these samples was added to a polypropylene tube. Next, 20 μL of 40 μg/mL mixed IS solution was added to this solution, following 480 μL of methanol (0.1% FA) was added to induce precipitation. After vortexing for 1 min, the samples were centrifuged at 12,000 rpm (13780 g) for 8 min. Finally, 500 μL of the supernatant was transferred to a vial for analysis.

LC–MS/MS analyses were performed using an LCMS-8050 triple-quadrupole mass spectrometer (Shimadzu, Japan). Chromatographic analyses were conducted on a Waters Acquity UPLC C₈ column (1.7 μm, 2.1 mm × 50 mm). The column temperature was maintained at 40°C and 2 μL of the sample was injected into the LC-30A UPLC system. Mobile phase A was 0.1% FA in water, and mobile phase B was 0.1% FA in acetonitrile. The flow rate was set to 0.4 mL/min. The gradient was set as follows: B, 5% (0 min) → 5% (0.5 min) → 50% (3 min) → 100% (4 min) → 100% (7 min) → 5% (9 min) → 5% (10 min). The total run time was 10 min. The gradient was shown in Supplementary Table S1.

The following interface settings were employed for sample analysis: ion transfer tube temperature, 300°C; sheath gas flow, 3.0 L/min; auxiliary gas flow, 10.0 L/min; and vaporizer temperature, 400°C. The desolvation temperature was maintained at 250°C. Quantification was performed by multiple reaction monitoring (MRM).

Primary stock solutions were obtained by dissolving the powders in different solutions, depending on their solubility. CEF, TAZ, POS, VOR, and LIN were dissolved in DMSO; SUL, TEI, CAS, TIG, and VAN were dissolved in water; PIP, MER, DAP, and FLU were dissolved in methanol. A stock solution of 10 mg/mL was prepared for each compound. Following this, 10 μL of the stock solution of each compound was transferred to a new tube, and 860 μL of blank serum was added to prepare a 100 μg/mL mixed stock solution. The samples were serially diluted with human serum to obtain concentrations of 0.1, 0.2, 0.3, 0.5, 1, 2, 5, 10, 20, 50, and 100 μg/mL according to their calibration curve. All four IS concentrations were 1 mg/mL. They were stored at −80°C until use.

Six concentration points were selected to construct the standard curve. Quality control (QC) samples of different concentrations were obtained by spiking human serum samples with an appropriate amount of the mixed stock solutions. The concentrations of standard curves and QCs were shown in Supplementary Table S2. The standards and QC solutions were divided into aliquots and stored in polypropylene tubes at −80°C until use.

Data analysis was performed using Microsoft Excel 2016. Statistical analyses were performed using the SPSS 21.0 software.

All the parent ions and product ions are listed in Table 1. Every analyte exhibits two transitions, one for quantitation, and the other for confirmation. In the positive ionization mode, the intensities of the protonated molecular peaks of all compounds were higher than the peak intensities of the corresponding molecular ions produced by deprotonation in the negative ionization mode, except for SUL, TAZ. Ideally, the molecular weight of sodium salt and crystalline water should be subtracted from the molecular weight of their parent ion. For example, these can include the sodium salt of CEF and SUL and crystalline water from PIP. Three different columns (Agilent SB C₁₈: 2.7 μm, 2.1 mm × 30 mm; Waters Acquity UPLC C₈: 1.7 μm, 2.1 mm × 50 mm; and SHIMADZU C_8: 2 μm, 2.1 mm × 100 mm) were tested, and the column affording the optimal separation was adopted. The Waters Acquity UPLC C₈ column was found to be superior for retaining aqueous compounds, with better chromatographic separation. The mobile phases initially used were water with 0.1% FA and acetonitrile with 0.1% FA (Phase B). Different concentrations of ammonium acetate (2 and 10 mM) were added to the mobile phases to identify compounds that were sensitive to pH variation and to obtain better peak shapes. However, no remarkable improvement was observed compared to the peaks obtained using the mobile phase without ammonium derivatives. In the initial stages of this study, the retention times of MER and VAN using the first gradient tested (initial mobile phase B concentration 10%; held constant for 1 min; increased to 90% in 4 min; held constant for 4 min; decreased to 10% in 1 min) were very short. After examining different gradients, the gradient was finally set as follows: 5% (0 min) −50% (3 min)–100% (4 min)–100% (7 min)– 5% (9 min)–5% (10 min). Under these conditions, the compounds could be identified and quantified based on well-resolved peaks. The chromatogram is shown in Figure 1.

Subsequently, six different organic solvents (methanol, methanol/acetonitrile (50:50), acetonitrile, methanol (0.1% FA), methanol/acetonitrile (50:50, 0.1% FA), and acetonitrile (0.1% FA) were used for protein precipitation to optimize sample pretreatment. The extraction recovery of all the analytes was calculated as the sample response ratio of the drug added before protein precipitation to that added after protein precipitation in the same solution. Pretreatment with methanol containing 0.1% FA was selected, as a superior recovery value and consistent results for all the compounds could be achieved compared to that using other solvents.

The matrix effect is important because it may affect the quantification precision, particularly with electrospray ionization (ESI). When possible, the utilization of 14 isotope-labeled ISs is an ideal choice. Generally, the selection of an IS is based on its ability to correct and reproduce the analytical behavior of each antibacterial agent. In this study, PIP-D5, DAP-D5, VOR-D5, and ETH were chosen as the ISs to balance cost and accuracy. Furthermore, we found that these four ISs were stably detected in all measurements and exhibited good performance for standard curve correction.

The validated method was used to examine 255 serum samples for TDM, and samples were obtained from 139 ICU patients treated with antibacterial agents at the Zhongnan Hospital of Wuhan University. The patients received the drug intravenously (except for POS, which was administered orally), and serum samples were collected just before the next administration post five to seven administrations (i.e., once the steady-state concentrations were achieved). The results are listed in Table 4 . The determined drug concentrations were correlated with the minimum inhibitory concentrations for optimizing the treatment in each patient according to the specific PK/PD indexes. For example, the mean value of VAN was 20.57 μg/mL, which is above the recommended range (10–20 μg/mL). This is probably because of the limited renal function in most ICU patients. These results support the need for the TDM of antibacterial agents to promote their appropriate administration.