Background: Patients on vancomycin often develop acute kidney injury (AKI), altering pharmacokinetics (PK). Due to vancomycin’s narrow therapeutic index, model-informed precision dosing (MIPD) is widely used, however model predictive performance during AKI remains uncertain.

Aims: We evaluated the performance of select vancomycin PK models in patients with AKI.

Methods: Patient data entered into the InsightRX Nova MIPD app between January 1 2023 and August 1 2024 were de-identified and analyzed retrospectively. For each age group (adult, pediatric, neonatal) 3-4 models validated for use in that patient population were selected. Measured drug concentrations were iteratively predicted using MAP Bayesian estimation with regular or flattened model priors. AKI rates were determined according to KDIGO criteria. Prediction error was assessed based on if the patient was not in AKI, or was in AKI for one (“rising”) or more (“sustained”) consecutive serum creatinine levels, quantified by root mean square error (RMSE).

Results: The data set included 454,671 adults, 6198 children and 2117 neonates, who experienced AKI rates of 5.4%, 3.4% and 5.2%, respectively. Relative to model performance in patients without AKI, models performed similarly or slightly worse in patients with sustained AKI (RMSE = -1-10%, +8-39%, +2-64% respectively) but performed considerably worse in patients with rising AKI (RMSE: +13%-27%, +23-73%, +45-105%). Using flattened priors improved predictions overall, and reduced the discrepancy in predictive performance between patients with or without AKI.

Conclusions: Predictive performance in patients with AKI varies between models. Flattened priors improves model prediction, particularly for patients in worsening condition.

Background: Therapeutic Drug Monitoring (TDM) of antibiotics is based on pharmacokinetic-pharmacodynamic (PK-PD) indices that compare antibiotic exposure to the minimum inhibitory concentration (MIC) of the targeted pathogen. This latter information is often not available, and worst-case scenarios are then assumed.

Aims: Creation of a webapp (www.antibiomic.com) to improve TDM interpretation of antibiotic/microbial combinations.

Methods: Developed with a Vue.js front-end and Node.js back-end (using Express for endpoints), this site pulls MIC data for various pathogens/anti-infectives from EUCAST (European Committee on Antimicrobial Susceptibility Testing) and computes free fractions via DruMAP (Drug Metabolism and pharmacokinetics (DMPK) Analysis Platform). Therapeutic targets come from literature or consensus.

Results: First, the time-dependent class of beta-lactam antibiotics is interpreted. Once the patient form is completed, a 100% ft>4 x MIC target was selected for results interpretation. In terms of efficacy, the free active concentration (fu.p) is relevant, and was used to compare the antibiotic exposition to the ECOFFs (highest MICs in a wild bacterial population). The prediction site https://adme.nibiohn.go.jp was used to calculate fu.p of each antibiotic. The MIC histogram distribution and ECOFF are daily uploaded from the website (https://mic.eucast.org/search/). Finally, the tab “interpretation” provides three essential indications: (i)efficacy via the calculated free concentration versus ECOFF, (ii)total concentration versus the usual pharmacological targets and (iii)notion of toxicity.

Conclusion: For TDM, this app allows results individualization and supports clinical/biological diagnosis. Antibiomic.com is successfully used for time-dependent antibiotics. Work is in progress on AUC-dependent antibiotics, aminoglycosides, and a patient history.

Keywords : Antibiotics, MIC, TDM, Web_App, personalized biology.

Background
Tofacitinib (JAK inhibitor) and voriconazole (triazole antifungal) are frequently co-administered in immunocompromised patients with autoimmune disorders. Voriconazole and its metabolite N-oxide inhibit CYP enzymes, while tofacitinib is metabolized by CYP3A4. However, the pharmacokinetic relationship between them remains poorly characterized, especially in populations with organ impairment or CYP2C19 polymorphisms.

Aims
To (1) establish physiologically based pharmacokinetic (PBPK) models for voriconazole/N-oxide and tofacitinib; (2) elucidate the CYP-mediated DDI mechanisms; (3) quantify exposure risks in clinical subpopulations.

Methods
PBPK models were developed in Simcyp® using physicochemical/ADME parameters from literature and optimized to match observed plasma profiles. Voriconazole/N-oxide models incorporated CYP2C19/3A4 metabolism and renal elimination; tofacitinib models included CYP3A4/2C19 clearance. DDI predictability was validated using midazolam (victim) and fluconazole/ketoconazole (perpetrators). Virtual trials (n=1,000) simulated interactions in healthy, hepatic/renal impairment, and CYP2C19 phenotypes (extensive [EM], intermediate [IM], poor metabolizers [PM]).

Results
Predicted/observed AUC and Cmax ratios fell within twofold error. Voriconazole/N-oxide increased tofacitinib exposure most markedly in severe renal impairment (AUCR 4.55, 95% CI: 4.21–4.93) and CYP2C19 PM (AUCR 4.18 vs. EM 3.58), driven by N-oxide accumulation. Rheumatoid arthritis patients showed reduced AUCR (2.14). Dose-normalized AUC inversely correlated with CYP2C19 activity (R²=0.89, p<0.001).

Conclusions
This study demonstrates metabolite-amplified CYP2C19 inhibition by voriconazole, elevating tofacitinib toxicity risks in renal impairment and CYP2C19 PM populations. Proactive dose reductions (≥50%) are recommended for these subgroups.

Keywords: physiologically based pharmacokinetics, drug-drug interaction, CYP2C19 polymorphism, voriconazole, tofacitinib, organ impairment

Background: Osteoarticular infections (OAI) frequently require long-term oral levofloxacin treatment. This antibiotic is associated with numerous adverse drug reactions (ADR), the mechanism of which remains poorly understood. The literature identifies the levofloxacin area under the curve of concentrations (AUC0-24h) /MIC ratio as the best PK-PD biomarker for efficacy. However, no study has investigated the link between levofloxacin exposure and the occurrence of ADR.

Aims: This monocentric retrospective study aims at evaluating the relationship between exposure parameters of levofloxacin (AUC0-24h, Cmax, Cmin) and the risk of ADR in the context of OAI.

Methods: Exposure data were collected from our TDM activity between July 2018 and July 2024. ADR occurring during the treatment were extracted from patients’ records and levofloxacin imputability was assessed by the regional pharmacovigilance center.

Results: Preliminary results of the first 79 patients of the study are presented. The most frequently dosing schedule were 500, 750 and 1000 mg/d (n=24, 30, 20, respectively) . The average duration of treatment was 83 days. Eleven patients presented at least one ADR attributed to levofloxacin (tendon and joint pain, neurological, hepatic or cardiological disorders). Median AUC0-24h, Cmax and Cmin in case of ADR were 161.6µg.h/mL [123.9;182.7], 10.5µg/mL [7.9;12.8] and 4.4µg/mL [2.4;5.4], respectively, vs 109.0µg.h/mL [83.9;163.6], 9.8µg/mL [7.1;12.1] and 1.9µg/mL [1.1;4.7] in the group without ADR.

Conclusions: These preliminary results show a significantly higher Cmin (p=0.01) and a trend to higher AUC0-24h (p=0.07) in patients developing ADR . These results are promising for refining the therapeutic range of levofloxacin.

Keywords: Levofloxacin, safety, exposure

Background:
Cefoperazone/Sulbactam (CPZ/SUL) has distinct protein-binding properties, affecting free drug concentrations and probability of target attainment (PTA).

Aim:
This study reports real-world free fractions (F) of CPZ/SUL and investigates physiological factors influencing free drug concentrations.

Methods:
An observational study was conducted at a tertiary hospital in Taiwan (June–December 2024). Patients receiving CPZ/SUL (1:1) were enrolled to measure free (fCPZ/fSUL) and total concentrations via ultra-high-performance liquid chromatography. Measured fCPZ/fSUL was compared to estimated values (EF: 0.2 for CPZ, 0.8 for SUL). Linear regression analyzed factors affecting free drug levels.

Results:
Thirty-three patients (48.5% male, mean age 75.2 ± 13.8) provided 66 CPZ/SUL concentrations. Median F was 0.07 (0.05,0.09) for CPZ and 0.95 (0.87,1.01) for SUL. fCPZ correlated with total-bilirubin (p < 0.01), and fSUL with creatinine (p < 0.01). PTA for 50% fT>MIC (MIC = 8) was lower using measured vs. estimated fCPZ/fSUL (21.2% vs. 52.5%, p = 0.01).

Conclusions:
CPZ/SUL free fractions deviated from estimates, reducing PTA. Direct fCPZ/fSUL measurement is crucial for patients with hyperbilirubinemia or renal impairment to optimize dosing and improve treatment outcomes.

Key word: unbound fraction, free drug concentration, probability of target attainment (PTA)

Background: Beta-lactamase-inhibitors (BLI) are co-formulated with beta-lactams (BL). However, changes in
pharmacokinetics and drug exposures may not remain proportional for both components.

Aims: We evaluated percentage of subtherapeutic BLI exposure and compared plasma BL:BLI ratio against
original product formulation ratio (8 for piperacillin-tazobactam, 4 for ceftazidime-avibactam and 2
for ceftolozane-tazobactam).

Methods: We retrospectively reviewed patients (≥16 years old) on BL/BLI TDM service in a quaternary hospital
from Oct-2019 to Jan-2025. Blood samples (maximum 4 timepoints within a dosing interval, including peaks and troughs) were collected and assayed using liquid-chromatography-tandem-mass-spectrometry. Unbound BLI troughs>4mg/L were considered subtherapeutic.

Results: Seventy-eight patients were included: 57 males, mean age 60.1 ± 15.0 years; mean body weight 69.8
± 24.6 kg. They received 44 piperacillin-tazobactam, 32 ceftazidime-avibactam, 6 ceftolozane-tazobactam courses, which yielded 395 plasma concentrations (208 piperacillin-tazobactam, 154 ceftazidime-avibactam, 33 ceftolozane-tazobactam). Median (IQR) total drug concentrations were:
piperacillin 125.4 (58.1-217.1) mg/L, ceftazidime 74.2 (49.0-108.1) mg/L, ceftolozane 54.8 (44.7-103.2) mg/L, tazobactam 17.7 (8.0-32.0) mg/L, avibactam 14.6 (9.3-21.3) mg/L. Median (IQR) BL:BLI ratios were 7.1 (6.0-8.1), 5.3 (4.1-7.0), and 4.2 (2.8-7.5) for piperacillin-tazobactam, ceftazidime-avibactam and ceftolozane-tazobactam respectively. Unbound BLI were subtherapeutic for 20 of 62 (32%) tazobactam troughs and 12 of 37 (32%) avibactam troughs.

Conclusion: BLI concentrations were low especially for ceftazidime-avibactam and ceftolozane-tazobactam.
Original BL:BLI ratios were not achieved in plasma. BL TDM alone does not guarantee adequate BLI exposure. TDM should be performed for both BL and BLI. BLIs should be formulated as standalone products for easier dose adjustments.

Keywords: beta-lactam, beta-lactamase inhibitor, tazobactam, avibactam