Mimicking in-vivo exposures to drug combinations in-vitro: anti-tuberculosis drugs in lung lesions and the hollow fiber model of infection

Frank Kloprogge,Karin Kipper,Oscar Della Pasqua,Stephen H Gillespie,Robert Hammond

doi:10.1038/s41598-019-49556-5

Frank Kloprogge, Karin Kipper + Show 3 more

Open Access

https://doi.org/10.1038/s41598-019-49556-5

Copy DOI

Abstract

Here, we evaluate protocol requirements to mimic therapeutically relevant drug concentrations at the site of infection (i.e. lung lesion) in an in-vitro hollow fibre model of infection using pulmonary tuberculosis as a paradigm. Steady-state pharmacokinetic profiles in plasma, lung tissue and lung lesion homogenate were simulated for isoniazid, rifampicin and pyrazinamide and moxifloxacin. An R-shiny User Interface was developed to support conversion of in-vivo pharmacokinetic CMAX, TMAX and T1/2 estimates into pump settings. A monotherapy protocol mimicking isoniazid in lung lesion homogenate (isoniazid CMAX = 1,200 ng/ml, TMAX = 2.2 hr and T1/2 = 4.7 hr), and two combination therapy protocols including drugs with similar (isoniazid and rifampicin (CMAX = 400 ng/ml)) and different half-lives (isoniazid and pyrazinamide (CMAX = 28,900 ng/ml and T1/2 = 8.0 hr)) were implemented in a hollow-fiber system. Drug levels in the perfusate were analysed using ultra-high-performance liquid chromatographic-tandem mass spectrometric detection. Steady state pharmacokinetic profiles measured in the hollow fiber model were similar to the predicted in-vivo steady-state lung lesion homogenate pharmacokinetic profiles. The presented approach offers the possibility to use pharmacological data to study the effect of target tissue exposure for drug combinations. Integration with pharmacokinetics modelling principles through a web interface will provide access to a wider community interested in the evaluation of efficacy of anti-tubercular drugs.

Highlights

The predictive value of nonclinical research for antimicrobial drug combinations, such as for tuberculosis, depends on how realistic experimental protocols reflect in-vivo conditions
Total drug concentrations in the lung tissue and lung lesion homogenate are lower for isoniazid (9.3% [RSE: 23.5%] and 29.8% [RSE: 21.7%]), rifampicin (24.8% [RSE: 15%] and 52.2% [RSE: 13%]) and pyrazinamide (37.2% [RSE: 7%] and 36.8% [RSE: 8%]) as compared to plasma concentrations (Appendix Table 1)
Rifampicin and pyrazinamide, simulations of the profiles of moxifloxacin revealed a nearly doubled elimination half-life in lung tissue and lung lesion homogenate as compared to plasma

Summary

Introduction

The predictive value of nonclinical research for antimicrobial drug combinations, such as for tuberculosis, depends on how realistic experimental protocols reflect in-vivo conditions (e.g. nutrition, oxygen, drug exposure). Of note are the challenges associated with regimens including drugs with different pharmacokinetic characteristics (e.g. different distribution, peak concentrations and different elimination half-lives) which need to be mimicked simultaneously within the experimental infection protocol using a single in one hollow fiber system This oversight can have major implications for the dose rationale as well as the identification of suitable companion or partner compounds for combination therapy[1,2]. Serial sampling for pharmacokinetics and pharmacodynamics could be achieved using the in-vitro hollow fiber model of infection[7] Such protocols may enable the evaluation of a wide range of drug exposures[8], mimicking drug concentrations likely to occur at the site of infection. The method can be used to evaluate a variety of compounds in monotherapy or in differing combinations, allowing for a more comprehensive characterisation of bacterial killing effects based on factorial designs which remain rather complex, if not impossible, to conduct in-vivo[11]

Methods

Results

Conclusion