Dose rationale for antibiotic combination therapy in infection diseases

Doctors need to use a complex range of drug combinations to cure many infectious diseases. A good example is tuberculosis (TB). Globally, 9 million people are living with TB. It causes about 1.5 million deaths per year - more than any other infectious disease.

Standard treatment involves four different anti-TB drugs, given in combination over six months - this is a long time to take drugs that cause a range of side effects. Many people do not complete the course of treatment, meaning that their TB may not be cured. They may also develop resistance to TB drugs.

Up to now, recommendations about the amount of drugs and length of treatment that should be offered to people with TB have been based mainly on observational experience, supported by expert opinions.

If we could shorten the length of time that people have to take anti-TB drugs, they are more likely to complete the course of treatment and respond to the drugs effectively. However, a number of multi million pound multi centre trials that have attempted to do this (for example by substituting the existing drugs with a drug called moxifloxacin) have failed.

Across a range of health conditions, the use of mathematical models (i.e. nonlinear mixed-effects models) is becoming increasingly common. These models help experts to make recommendations about the dose of drugs that should be offered, and how long this treatment should be offered for. However, at the moment there is no such model available for the treatment of TB.

This research proposal aims to address this problem. It will bring together two different types of data to develop a mathematical modelling framework:
1. drug concentrations in the blood and lungs of human beings
2. data from innovative in-vitro experiments, which can mimic the antibacterial drug effect in a human lung.

The combination of these two different types of data in a modelling framework is new. Because of this, a variety of nonlinear mixed-effects models and statistical methods will need to be tested to develop the most reliable framework.

This modelling framework can then be used to make recommendations about the level of anti-TB drugs which should be offered, and how long these should be offered for, personalised for patient populations (i.e. children and elderly). Eventually this will translate into personalised TB treatments, thus improving the quality of life for patients and saving healthcare time and resources.

If this project is successful, the modelling framework will also serve as the base for dose selection and personalisation of other drugs and in other infectious diseases. For example, the framework could inform the design of new clinical trials and the selection of doses of new anti-TB drug combinations for the treatment of multi-drug resistant TB.

The project will be based at University College London, and will be undertaken by Dr Frank Kloprogge. It will be hosted by Professor Oscar Della Pasqua, and involves collaborations with other researchers from University College London, St Andrews University and Université Paris Diderot.

Parameterisation of pharmacokinetic and pharmacokinetic-pharmacodynamic (PKPD) data into nonlinear mixed-effects models is required to better understand the time course of bacterial growth and bactericidal activity of drugs. To date, little attention has been paid to how these models should be parameterised to quantify drug effects during combination therapy or explain variability associated with clinical, genetic and demographic covariate factors. In this proposal, in-vivo clinical data will be integrated with in-vitro data from innovative hollow fibre experiments using nonlinear mixed-effects modelling techniques to improve dose selection for drug combinations and to personalise the treatment of anti-infective agents in different patient groups. Moreover, the approach will be used to support the design of prospective clinical studies in tuberculosis, an area which has gained scientifically and regulatory relevance over the last decade.

In previous investigations, lung tissue/lesion concentrations and antibacterial activity has been described using empirical or more mechanistic and single-state or multi-state nonlinear mixed-effects models, respectively. Here, nonlinear mixed-effects modelling approaches will be identified that enable further characterisation of bacterial growth over time in experimental conditions which mimic disease in humans. The ultimate goal is to identify new parameterisations supported with prior distributions in a Bayesian framework which facilitate the translation of nonclinical data, enabling the prediction of treatment response in patients using the software package NONMEM (aim 1).

The availability of such a framework will allow the evaluation of simulation scenarios to optimise, personalise and recommend standard drug combination therapy for pulmonary tuberculosis (aim 2). In combination with optimality concepts, the proposed approach can be used to support the design of innovative and efficient prospective PKPD studies (aim 3).

The four beneficiaries of the proposed research will be:
- Patients requiring a combination of antibiotic compounds to treat an infectious disease
- Clinicians and nurses
- For profit and not for profit organisations conducting pharmaceutical and clinical research and development for drug combinations to treat infectious diseases
- Medicine regulatory agencies and policy makers

The combination of isoniazid, rifampicin, pyrazinamide and ethambutol or moxifloxacin to treat pulmonary tuberculosis is selected as paradigm. Pulmonary tuberculosis patient, clinicians and nurses will eventually benefit from the research in that optimised and personalised treatment guidelines will be developed for the selected paradigm drugs. The developed modelling framework will enable to optimise the dosages of each active compound in order to maximise the total antibacterial activity of the treatment combination whilst minimising the course of duration. This will contribute to a lower risk of relapses, lower risk of resistance development and a longer therapeutic life time of the drug combination. This results into a reduction in the burden of expensive long follow-up programmes. Optimised and personalised treatments using the modelling framework will be communicated in the form of scientific papers in the third year of the fellowship.

Any for profit and not for profit organisation involved in updating dosing guidelines of existing drug combinations, by shortening the treatment duration and personalising dosing recommendations (e.g. for paediatric and elderly patients), or selection of dosages for novel drug combinations will be beneficiaries of this research. The proposed modelling framework, combines different types of data, and enriches the pharmacokinetic-pharmacodynamic information content. This enables rational dose selection for drug combination to treat diseases such as pulmonary tuberculosis what has been till date not possible. Due to the nature of the proposed research, the modelling framework provides the base for rational dose selection of novel antibiotic drug combinations to treat multi drug resistant tuberculosis or in other therapeutic areas.

At last, medicines regulatory agencies and policy makers such as, WHO, Public Health England and the British Thoracic Society, will benefit from this research through their role in both guiding the types of studies required, and in assessing submissions for product licensing.