Mehdi El Hassani 1, Daniel J.G. Thirion, Aysenur Yaliniz, Katherine Desforges, Josée Verdon, Emily G. McDonald, Amélie Marsot
1 University Of Montreal (, )
Objectives: Oral ciprofloxacin remains one of the few available oral agents with activity against Pseudomonas aeruginosa. However, exposure in older adults may be highly variable due to age-related renal impairment, comorbidities, and polypharmacy (1, 2). Despite frequent use, no dedicated population pharmacokinetic (popPK) model has been developed specifically for oral ciprofloxacin in this age group. Therefore, the objectives were (i) to develop a popPK model for oral ciprofloxacin in adults ≥75 years and (ii) to use this model to inform an optimized study design for a larger prospective study aimed at supporting ciprofloxacin precision dosing in this population.
Methods: Data were collected from a prospective pilot study in adults ≥75 years that were hospitalized in a tertiary care teaching center. The study included multiple antibiotics, including ciprofloxacin. To maximize feasibility, all antibiotics used the same sampling schedule (three samples per patient). PopPK modelling was performed in NONMEM v7.6 using stochastic approximation expectation-maximization and importance-sampling. One- and two-compartment models with first-order absorption and elimination were evaluated. Interindividual variability (IIV) was described using exponential models. Additive/proportional/combined residual errors were explored. Prior information ($PRIOR) was incorporated to improve model stability with limited data. Covariates (age, sex, weight, serum creatinine, creatinine clearance, C-reactive protein…) were explored using a stepwise approach. Because estimation incorporated prior information, objective function changes from added covariates were interpreted descriptively, and covariates were retained only if they meaningfully reduced IIV. Internal evaluation was performed using goodness-of-fit plots, nonparametric bootstrap and a prediction-corrected visual predictive check. To inform the design of a larger ciprofloxacin-focused study, we optimized the sampling strategy using the D-optimality criterion in $DESIGN. The optimizations were repeated while increasing GROUPSIZE. Expected precision was summarized using the relative standard error (RSE). To assess how sensitive the optimized sampling times were to uncertainty in fixed‐effects, we defined a log-normal distribution centered on each final THETA estimate with 20% CV. A total of 1000 SUBPROBLEMS were run, each using a different sampled THETA vector, while keeping OMEGA and SIGMA fixed (3).
Results: Fifteen patients (median age 82 years; median weight 75 kg; median Cockcroft-Gault creatinine clearance 46 mL/min, median Charlson Comorbidity index 7) contributed 42 ciprofloxacin plasma concentrations. A one-compartment model with first-order absorption and elimination best described the data. The absorption rate constant was not identifiable with the available sampling and was therefore fixed to 2.5 h-1 (4). Typical estimates of apparent clearance and apparent volume of distribution were 14.3 L/h and 247 L, respectively. IIV was large on clearance (75.2%) and moderate on volume (36.0%). Residual variability was best described by a proportional error model (16.1%). No covariates were retained in the final model. The internal evaluation showed that the model adequately described the observed data, with no evidence of systematic bias or model misspecification. In the design evaluation, using the typical sampling times from our study (1.5, 6.0, and 12h post-dose) and GROUPSIZE matching the study sample size (n=15), expected precision for CL/F and V/F was acceptable (RSE <30% for both), whereas Ka was poorly informed (RSE=180%). After time optimization, the proposed sampling times were 0.40, 2.05, and 12h. With these optimized sampling times, the expected precision of Ka markedly improved (RSE <30% at n=15, decreasing with increasing sample size), while RSEs for CL/F and V/F remained consistently <30% across all tested designs. For variability parameters, precision improved primarily with increasing sample size, regardless of sampling times. The IIV on CL/F was estimated with acceptable precision across all scenarios (RSE <50% at n=15 and decreasing with increasing sample size). In contrast, the RSE of IIV on V/F remained >50% until n=50. Expected precision for residual variability was acceptable across all scenarios (RSE <50%). In sensitivity analyses, the optimized sampling times remained consistent with the initial optimized design, with a median (2.5th-97.5th percentile) of 0.36h (0.21-0.55) for the first sample, 1.99h (1.46-2.63) for the second sample, and 12h (12-12) for the last sample. Conclusions: This is the first popPK model developed specifically for oral ciprofloxacin in adults ≥75 years and highlights substantial IIV in exposure in this population which may partly reflect unmodeled variability in oral bioavailability. An optimized sampling design with a minimum sample size of 50 was identified to improve parameter estimation in a future prospective study. References: 1. Mangoni AA, Jackson SH. Age-related changes in pharmacokinetics and pharmacodynamics: basic principles and practical applications. Br J Clin Pharmacol. 2004;57(1):6-14. 2. Pea F. Pharmacokinetics and drug metabolism of antibiotics in the elderly. Expert Opinion on Drug Metabolism & Toxicology. 2018;14(10):1087-100. 3. Bauer RJ, Hooker AC, Mentre F. Tutorial for $DESIGN in NONMEM: Clinical trial evaluation and optimization. CPT Pharmacometrics Syst Pharmacol. 2021;10(12):1452-65. 4. Junkert AM, Lazo REL, Deffert F, Carneiro J, Borba HHL, de Campos ML, et al. Pharmacokinetics of oral ciprofloxacin in adult patients: A scoping review. British Journal of Clinical Pharmacology. 2024;90(2):528-47.
Reference: PAGE 34 (2026) Abstr 12009 [www.page-meeting.org/?abstract=12009]
Poster: Methodology - Study Design