Annabelle Lemenuel-Diot (1), Nicolas Frey (1)
(1) Hoffmann-La Roche Ltd, Basel, Switzerland
Objectives: In clinical drug development of Chronic Hepatitis C treatments, early VL information between week 2 and week 8 (e.g.: magnitude of VL drop or VL below the detection limit), are generally used to predict SVR, the primary clinical endpoint defined as undetectable VL 24 weeks after end of treatment. The objective of this work is to investigate if an integration of the early VL profile using a viral kinetic (VK) model would allow improving the SVR prediction. The impact of using VL information from the relapse that may occurred following treatment interruption was also investigated.
Methods: Using the Hepatitis C viral kinetic model previously developed [1], the analysis plan was defined as follow:
- Simulation of different treatment durations: 2, 4, 8 weeks, with or without treatment interruption, for a set of 1000 patients using the individual VK parameters from the original model.
- Estimation of the population/individual VK parameters with Monolix 3.2 using the available VL information for each simulated treatment.
- Simulation of full treatment duration (48 weeks) with the estimated individual parameters to predict SVR.
- Assessment of the SVR predictive performance for the different predictors: Early Viral Response, Early VL drop, Early VL time course w/wo treatment interruption.
Results: The main results are summarized in the table below with: Predicted SVR (the true being 55%), Sensitivity (True SVR among predicted SVR) and Specificity (True noSVR among predicted noSVR).
|
|
Predicted SVR (%) |
Sensitivity (%) |
Specificity (%) |
|
3 log VL drop w2 |
22 |
89 |
55 |
|
2 log VL drop w8 |
81 |
66 |
89 |
|
VR w4 |
15 |
97 |
52 |
|
VR w8 |
41 |
89 |
68 |
|
VL Time Course 4w |
74 |
68 |
81 |
|
VL Time Course 4w + relapse |
55 |
90 |
86 |
Among the different predictors, the full integration of the VL information with a treatment interruption, to use the relapse information, is the best way to accurately predict SVR combining both high sensitivity and high specificity.
Conclusions: Different strategies using early VL drop criteria are generally implemented in clinical drug development to predict SVR. From the simulation, it can be shown the limitation of such approaches. Actually, the best predictive performances of SVR are obtained using full information of VL time course from a short treatment including a treatment interruption. However, the issue related to a treatment interruption would be the potential higher risk of development of resistance.
References:
[1] Snoeck E, Chanu P, Lavielle M, Jacqmin P, Jonsson EN, Jorga K, Goggin T, Grippo J, Jumbe NS and Frey N. A Comprehensive Hepatitis C Viral Kinetic Model Explaining Cure. Clin Pharmacol Ther (2010).
Reference: PAGE 22 () Abstr 2851 [www.page-meeting.org/?abstract=2851]
Poster: Other Drug/Disease Modelling