Tapping into the potential of pharmacometrics in preclinical infection models: a low-hanging fruit or joined efforts needed?
Linda B.S. Aulin1*, Robin Michelet1*, Roger Le Grand2, Mustapha Si-Tahar3, Virginie Hervé3, Mara Baldry4, Jean Claude Sirard4, Charlotte Kloft1
1. Dept. of Clinical Pharmacy and Biochemistry, Institute of Pharmacy, Freie Universitaet Berlin, Berlin, Germany 2. Center for Immunology of Viral, Auto-immune, Hematological and Bacterial Diseases, CEA, Université Paris-Saclay, Inserm UMR 1184, 92265 Fontenay-aux-Roses, France. 3, The Research Center for Respiratory Diseases, CEPR, Inserm UMR 1100, 37000 Tours, France. 4. Institut Pasteur de Lille, Université Lille, CNRS, Inserm, Centre for Infection & Immunity of Lille, 59000 Lille, France * Shared first authors
Introduction: In the development of alternative anti-infective combination therapies, preclinical in vivo infection studies are performed to characterise the PKPD of lead compounds alone, and in combination with antibiotics. Such studies provide core building blocks that aid in moving compounds through the development pipeline, in which translation and clinical trial design are supported by well-established modelling and simulation workflows. However, the design of the underlying preclinical studies often lacks the same rigorous pharmacometric efforts. Here, we aim to showcase how pharmacometrics can be applied to optimise the design of preclinical studies in which available antibiotics are combined with an immunomodulatory lead compound, and outline some of the current challenges with such efforts.
Methods: Within the context of the FAIR project [1], we supported the design of preclinical (super-) infection models in mice and non-human primates (NHP) [2]. In these studies, an immunomodulatory protein, FLAMOD, was given together with amikacin (AMK; 1.) or amoxicillin (AMX; 2.-3.). The studies aimed to characterise the PKPD of monotherapies (FLAMOD or antibiotic) as well as combinations (FLAMOD + antibiotic). Initially, we identified relevant models and/or data from literature or in-house. Depending on availability, we either used pre-existing NLME models or developed new models based on digitised data [3]. When needed, we applied inter-species scaling according to allometric principles with fixed exponents. The models were used for simulations supporting selection of antibiotic dose and/or sampling design. The target doses were based on literature-derived PK/PD indices and the MIC of the study pathogen.
Results:
- Intraperitoneal AMK in Pseudomonas aeruginosa infected mice
We identified three human AMK PK models [4-6] and two publications containing murine PK data [7,8]. When applying allometric scaling principles, these models underpredicted the murine clearance. This is in line with previous studies reporting murine clearance of aminoglycosides as higher than expected based on kidney function alone. Subsequently, based on the murine data (digitised mean values), a one-compartment model was developed and used for dose selection. We recommended 80 mg/kg as an efficacious dose for the study pathogen (MIC 8 µg/mL) and one to two plasma samples during the first two hours to characterise the absorption, complemented by one late sample (2-3 h) to cover the elimination processes. - Peroral (PO) AMX in Streptococcus pneumoniae superinfected mice
A previously developed two-compartment model with linear absorption of 10 or 350 µg PO AMX in mice [9] was used to recommend a sampling schedule identifying possible impact of FLAMOD on AMX PK in mice given 150 or 1000 µg of AMX. We used internal AMX data (10, 50, 150, 350 µg PO) to qualify the predictions, which revealed nonlinear PK of AMX and poor prediction of the 50 and 150 µg doses. To gain insights into possible saturable processes, we are currently leveraging digitised IV and PO rat plasma and urine data, covering a dose range of 1.1 to 8.8 mg AMX [10,11]. - Intramuscular (IM) AMX in S. pneumoniae superinfected NHP
No models nor data describing the PK of AMX in NHP were identified. As no human AMX IM models were available, we used a PK model of AMX after PO administration that included nonlinear absorption as a starting point [12]. We allometrically scaled the model, adapted, and qualified the prediction of unbound plasma AMX in NHP after IM using data from obese patients after PO AMX administration [13] and a cross-over study in which AMX was administered IV, PO, or IM to healthy volunteers [14]. The adapted model was used for simulating subtherapeutic doses (assumed MIC 1 µg/mL), from which we recommended 10 mg/kg of AMX IM every six hours.
Conclusions: Here, we have showcased several preclinical study design challenges, each requiring their own model-based strategy. Preferably, a generic workflow should be established to allow for standardised analysis. However, such a workflow is currently not feasible, and the approach must be tailored to each specific task. The main reason is the lack of publicly available preclinical data. To overcome this, we invite the pharmaceutical industry to share preclinical data for these often out-of-patent compounds. The data would greatly support the establishment of workflows for model-based design of preclinical infection studies.
References:
[1] The ‘Flagellin Aerosol therapy as an Immunomodulatory adjunct to the antibiotic treatment of drug-Resistant bacterial pneumonia’ (FAIR) project; https://cordis.europa.eu/project/id/847786
[2] Michelet et al., Pharmaceutics, 13(5), 601 (2021).
[3] Rohatgi A. WebPlotDigitizer. Austin, Texas, USA: 3.10 (2016).
[4] Carrié et al., AAC, 64 (2020).
[5] Alhadab et al. AAC, 62 (2018).
[6] Kato et al., Drugs in R and D, 17, 177–187 (2017).
[7] Andes and Craig, Int. J. Antimicrob., 19, 261–268 (2002).
[8] Du et al., J Mol Med., 84, 573–582 (2002).
[9] Franck et al., Pharmaceutics, Mar 30;13(4):469 (2021).
[10] Torres-Molina et al., Biopharm. Drug Dispos., 13, 23-38 (1992).
[11] Torres-Molina et al., Biopharm. Drug Dispos., 13, 39-52 (1992).
[12] De Velde et al., JAC., Oct 1 (2016)
[13] Mellon et al., JAC., Dec 1 (2020)
[14] Spyker et al., AAC., 11(1):132 (1977).