Eva Germovsek (1), Joanna Lewis (2), Rollo L Hoare (1,2), Robin Callard (1,2), Nigel Klein (1), Joseph F Standing (1)
(1) UCL Institute of Child Heath, London, UK; (2) UCL CoMPLEX, London, UK
Objectives: Paediatric T-cell population dynamics are influenced by higher thymic output and peripheral T-cell division rates than in adults. CD4 T-cell numbers decline in untreated HIV infection; this can be prevented with antiretroviral therapy (ART), but paediatric patients sometimes interrupt ART. To describe CD4 cell dynamics different complex mechanistic models have previously been tested [1]. However, since only 2% of CD4 cells are in the blood, it is difficult to measure and define all the proposed model parameters, which can limit mechanistic interpretation. The aim is to develop a simpler, mechanistic CD4 reconstitution model.
Methods: Naïve CD4 concentrations from the PENTA 11 trial [2] were converted to total body CD4 counts and modelled using NONMEM 7.3 with FOCE-I. The data included 29 and 31 HIV patients on continuous and interrupted treatment, respectively; mean (range) age 10.2 (2.2-19.1) years. 762 naïve CD4 levels were measured. A mechanistic model with 3 estimated parameters (thymic output, loss of naïve CD4 cells due to death and proliferation, and initial naïve CD4 cell count) was applied to the data. Effects of age [3] on both input and loss of naïve CD4 cells from the naïve T-cell reservoir were included, as well as a competition function [4] that describes the change in loss with changing cell number. A drug parameter was introduced as a time-varying binary covariate.
Results: Compared to a simple 1-compartment model, a model that included age-related effects and competition function provided a 19 unit drop in OFV. One of the parameters describing the competition function was fixed. Other parameters (input as a multiple of that in a healthy child [5], loss, initial cell number and effect of ART) were estimated. Final estimates (relative standard error, %) for these parameters were 0.119 (19.7), 0.659 day-1 (11.8), 5.18 x1010 cells (7.22) and -0.781 (4.56), respectively. Goodness-of-fit plots and VPC suggest that the model fits the data well and there are no major model misspecifications.
Conclusions: A mechanistic model was used to describe CD4 dynamics, with fewer parameters and ODEs than previous models, and might provide more reliable results. Provisional results imply that ART reduced the net loss of naïve CD4 cells to 21.9% of the value when off therapy. Future work will consider viral load and memory CD4 cells in the model, to further understanding of the effects of treatment interruption on long-term CD4 reconstitution.
References:
[1] Lavielle et al., Biometrics 2011; 67(1): 250-9
[2] PENTA, AIDS 1996; 24(2): 231-41
[3] Bains et al., J Immun. 2009; 183: 4329-36
[4] Hoare et al., PAGE 22 (2013) Abstract 2676
[5] Huenecke et al., Eur J Haematol. 2008; 80(6): 532-39
Reference: PAGE 23 () Abstr 3227 [www.page-meeting.org/?abstract=3227]
Poster: Drug/Disease modeling - Paediatrics