Mathematical and Statistical Modelling of CCR5 Inhibitor Effects in Adults and Children with HIV-1 Infection
Lead Research Organisation:
University College London
Department Name: Immunology and Molecular Pathology
Abstract
This study will use mathematical models to find out how medicines for HIV act in children and adults. Measurements of blood concentrations of white cells, viral load and drugs levels in patients receiving different medicine combinations will be used. Mathematical models are descriptions of these measurements using a system of equations. The main outcome is to find a model that describes the measurements in a way that relates to biological processes. Investigating which parts of the model differ between patient groups and different treatment regimes will answer questions such as: What happens to white cell counts if different drug combinations or doses are used? Can we predict paediatric drug effects from adult data and expected biological differences? What happens to viral load and white cell count if doses are missed? By creating a model of the system including drug levels, viral load and white cell counts, a better understanding of how these relate to each other is gained. This understanding will make the development of new medicines more efficient, and help optimise the use of current medicines. What makes this study special is the mathematical modelling techniques proposed have not been widely applied to HIV infection.
Technical Summary
CCR5 receptor inhibitors are a new class of antiretroviral that inhibit viral entry into host immune cells. This work aims to use mathematical/statistical modelling to understand the long-term effects of CCR5 inhibitors in adults and children. A major difference between the method proposed and a purely statistical approach is that ordinary differential equations will be applied to longitudinal data. This novel method allows fitting of mechanistic models to sparse, unbalanced observations, giving the potential to go beyond merely describing the data, to making predictions under varying hypothetical conditions. Exploration of developmental differences between adults and children, determining the place of CCR5 inhibitors in HIV-1 therapy, and mathematical investigation of non-adherence to antiretroviral medications are key objectives.
Longitudinal data on viral load and CD4 counts in 1934 HIV-1-infected adults treated with antiretroviral regimens including the CCR5 inhibitor maraviroc will be used in the first part, which aims to develop a mechanistic model describing the changes in CD4 count and viral load over 48 weeks of treatment. The studies available provide data on various drug combinations, along with data in patients with non-CCR5 tropic HIV-1 infection, which will be useful in investigating immune system effects in the absence of observed antiretroviral activity. Pharmacokinetic data will also be integrated into the model.The second part adds paediatric data from around 100 subjects to be analysed using the adult mechanistic model. Developmental differences in both immunology and drug effects will be investigated through covariate model building guided by prior knowledge of ontogeny. Using the mechanistic insights gained, optimal design of treatment strategies in both clinical HIV-1 treatment and antiretroviral study design will be possible. Using the data and model, the final study is an investigation on the impact of non-adherence on drug concentrations, CD4 count and viral load. Novel methods for identifying patterns of non-adherence and for appropriate handling of data suspected to arise from non-adherence will be developed, which will be used to detect and predict its impact on clinical effect. Scientific opportunities presented in this work include access to large clinical datasets, and the chance to employ sophisticated mathematical and statistical modelling techniques that have not been widely applied to this field. The medical opportunities lie in using mathematical models to investigate antiretroviral mechanisms. Simulation from model parameters will guide the design of future studies and provide treatment recommendations for clinical practice, in particular on the place of CCR5 inhibitors.
Longitudinal data on viral load and CD4 counts in 1934 HIV-1-infected adults treated with antiretroviral regimens including the CCR5 inhibitor maraviroc will be used in the first part, which aims to develop a mechanistic model describing the changes in CD4 count and viral load over 48 weeks of treatment. The studies available provide data on various drug combinations, along with data in patients with non-CCR5 tropic HIV-1 infection, which will be useful in investigating immune system effects in the absence of observed antiretroviral activity. Pharmacokinetic data will also be integrated into the model.The second part adds paediatric data from around 100 subjects to be analysed using the adult mechanistic model. Developmental differences in both immunology and drug effects will be investigated through covariate model building guided by prior knowledge of ontogeny. Using the mechanistic insights gained, optimal design of treatment strategies in both clinical HIV-1 treatment and antiretroviral study design will be possible. Using the data and model, the final study is an investigation on the impact of non-adherence on drug concentrations, CD4 count and viral load. Novel methods for identifying patterns of non-adherence and for appropriate handling of data suspected to arise from non-adherence will be developed, which will be used to detect and predict its impact on clinical effect. Scientific opportunities presented in this work include access to large clinical datasets, and the chance to employ sophisticated mathematical and statistical modelling techniques that have not been widely applied to this field. The medical opportunities lie in using mathematical models to investigate antiretroviral mechanisms. Simulation from model parameters will guide the design of future studies and provide treatment recommendations for clinical practice, in particular on the place of CCR5 inhibitors.
