A population pharmacokinetic analysis to evaluate the impact of body weight, renal function, and patient status on vancomycin pharmacokinetics
Tan Zhang (1), Elke H.J. Krekels (1), Cornelis Smit (2), Eric P.A.van Dongen (3), Roger J.M. Brüggemann (4,5), Catherijne A.J. Knibbe (1,6)
(1) Systems Biomedicine and Pharmacology, Leiden Academic Centre for Drug Research, Leiden University, Leiden, The Netherlands; (2) Pediatric Pharmacology and Pharmacometrics, University Children's Hospital Basel (UKBB), University of Basel, Basel, Switzerland (3) Department of Anesthesiology and Intensive Care, St. Antonius Hospital Nieuwegein, the Netherlands. (4) Department of Pharmacy, Radboud University Medical Centre, Radboud University, Nijmegen, The Netherlands; (5) Center of Expertise in Mycology Radboudumc/CWZ, Nijmegen, The Netherlands. (6) Department of Clinical Pharmacy, St. Antonius Hospital, Nieuwegein, The Netherlands
Objectives: Vancomycin is commonly used to treat severe Gram-positive bacterial infections. Due to its highly variable PK and narrow therapeutic window, this renally cleared glycopeptide antibiotic is particularly challenging to dose in patients with obesity, renal insufficiency, critical illness, or a combination of these characteristics. With increasing numbers of obese patients being admitted to hospitals for treatment, we here applied a population PK approach with covariate analysis to develop a PK model in obese patients with varying degrees of renal function admitted to the hospital ward or ICU. Model-based simulations are used to provide model-derived dosing recommendations.
Methods: A model was developed based on data extracted from electronic health records, including patients admitted to the ward or ICU with BMI ≥ 25 kg/m2 who received ≥ 1 dose of vancomycin and had ≥ 1 vancomycin plasma concentration and ≥ 1 serum creatinine measurement. Patients receiving renal replacement therapy were excluded. The data were pooled with previously published dense prospective data from 8 non-obese and 20 (morbidly) obese but otherwise healthy subjects (HS) [1] and analyzed with a non-linear mixed effect modeling approach using NONMEM 7.4. Covariates that were considered in the analysis were weight related covariates (total body weight (TBW), lean body weight (LBW), adjusted body weight), renal function estimates (modification of diet in renal disease (MDRD), chronic kidney disease epidemiology (CKD-EPI), Cockcroft-Gault with LBW (CG-LBW) or TBW (CG-TBW)), age, and patient status (HS, ward or ICU). Initially the covariate relationship between TBW and clearance (CL) was fixed to what was found in the HS [1] to assess the influence of other covariates, while in a final run all covariate relationships were estimated simultaneously. The final model was qualified with goodness-of-fit (GOF) diagnostics and ETA plots that were split for the main covariates (patient status, bodyweight and CKD-EPI), bootstrap and normalized prediction distribution errors (NPDE) analysis. Stochastic simulations were performed to design an optimal dosing guideline targeting similar exposure across all patients.
Results: A total of 1499 serum concentrations were available from 217 obese hospitalized patients (TBW: 59-143 kg, CKD-EPI: 5.99-176.99 ml/min/1.73m2 , ICU: 25.81%), in conjunction with 249 concentrations from 28 non-obese and (morbidly) obese HS (TBW: 60-235 kg, CKD-EPI: 70.28-127.96 ml/min/1.73m2). A three-compartment model with first-order elimination rate was selected as structural model. Patient status was recognized as a critical covariate due to the difference in TBW range and renal function between the patient group and HS group, since only after this covariate was included on both CL and volume of distribution, the impact of renal dysfunction (CKD-EPI) and bodyweight (TBW) could be accurately quantified. In the final model, CL was 3.46 L/h in a ward patient with a TBW of 100 kg and CKD-EPI of 75 ml/min/1.73m2. Compared to HS, CL was 41.2 % lower in ward patients and 51.3 % lower in ICU patients. The estimated exponents in the exponential relationships between TBW and CKD-EPI on clearance were 0.42 and 0.71 respectively. The peripheral volume of distribution was 1.33 L and increased with TBW with an exponent of 2.84. Additionally, this volume was 34.8 times higher in patients compared to HS. Split GOF diagnostics, ETA plots, bootstrap, and NPDE showed that the final model was robust and had good predictive performance. The proposed practical dosing guidelines showed that to achieve similar exposure due to covariates on CL, higher doses are needed with increasing TBW and increasing renal function and that compared to HS a 50% dose reduction in the ward patient and 60% dose reduction in the ICU patient should be applied.
Conclusions: This data-driven population PK analysis describes the PK of vancomycin in both ward and ICU patients with varying degrees of obesity and renal dysfunction, which covers a large and relevant clinical population. We found patient status to be important and that this together with bodyweight and renal function should be taken into account to get safe and effective exposure in all patients.
References:
[1] Smit C, et al. Br J Clin Pharmacol. 2020;86(2):303-317.