Model Based Network Meta-Analysis for Pharmacometrics and Drug-Development: a 3 year Research Collaboration between Pfizer and the University of Bristol.
David Mawdsley (1), Meg Bennetts (2), Sofia Dias (1), Martin Boucher (2), Tony Ades (1), and Nicky J. Welton (1)
(1) School of Social and Community Medicine, University of Bristol (2) Pharmacometrics, Global Clinical Pharmacology, Pfizer Ltd
Objectives: Meta-analysis is a well-established methodology for combining the results of randomised controlled trials (RCTs) that compare the same treatments and outcomes. However, in drug-development, early-phase studies explore the action of compounds over dose and time. Model-based meta-analysis (MBMA) has been developed to use non-linear pharmacokinetic-pharmacodynamic type models that allow dose-response and time-course effects within meta-analysis [1-3]. Network Meta-Analysis (NMA) allows simultaneous comparison of multiple treatments [4,5], and provides a framework for model comparison and assessment of evidence consistency [6]. Although multiple doses can be compared, this is done by either “lumping” similar doses, or by regarding them as separate treatments. “Lumping” introduces the potential for inconsistency between comparisons, which can invalidate the results of NMA. Regarding them as separate treatments can lead to sparse or unconnected networks. MBMA has recently been used to assess multiple treatment comparisons, however little attention has been given to assessment of model fit and consistency. This project aims to integrate the two approaches in a Model Based Network Meta-Analysis (MBNMA), to model multiple treatments, dose and time-course that incorporates assessment of model fit [6] and consistency [7].
Methods: We illustrate the importance of assessing model fit using a MBMA comparing Naproxen vs placebo for treating pain. The fit and model predictions are compared for different time-course models. We indicate how the methods can be extended to multiple treatments, using a network of trials of multiple treatments, doses and time-points for osteoarthritis.
Results: We show that parameter estimates are sensitive to choice of model and that “lumping” in a NMA can lead to inconsistent treatment effects, motivating a model-based analysis.
Conclusions: Results can be sensitive to model choice, and ignoring dose can lead to inconsistency. It is therefore essential to assess fit and consistency. In January 2015 Pfizer Ltd & the University of Bristol started work on a 3 year Medical Research Council (MRC) Industry Collaboration Agreement project to develop MBNMA methods. This offers the potential to combine all the available dose-response and time-course evidence in a MBNMA to compare the relative efficacy of multiple treatments, while allowing model fit [6] and evidence consistency [7] of the whole network to be assessed.
References:
[1] Mould, D. R. (2012). Model-based meta-analysis: an important tool for making quantitative decisions during drug development. Clinical Pharmacology and Therapeutics, 92(3), 283–6. doi:10.1038/clpt.2012.122
[2] Mandema, J. W., Cox, E., & Alderman, J. (2005). Therapeutic benefit of eletriptan compared to sumatriptan for the acute relief of migraine pain--results of a model-based meta-analysis that accounts for encapsulation. Cephalalgia : An International Journal of Headache, 25(9), 715–25. doi:10.1111/j.1468-2982.2004.00939.x
[3] Mandema, J. W., Salinger, D. H., Baumgartner, S. W., & Gibbs, M. A. (2011). A dose-response meta-analysis for quantifying relative efficacy of biologics in rheumatoid arthritis. Clinical Pharmacology and Therapeutics, 90(6), 828–35. doi:10.1038/clpt.2011.256
[4] Dias, S., Sutton, A. J., Ades, A. E., & Welton, N. J. (2013). Evidence synthesis for decision making 2: a generalized linear modeling framework for pairwise and network meta-analysis of randomized controlled trials. Medical Decision Making: An International Journal of the Society for Medical Decision Making, 33(5), 607–17. doi:10.1177/0272989X12458724
[5]Salanti, G., Higgins, J. P. T., Ades, A. E., & Ioannidis, J. P. A. (2008). Evaluation of networks of randomized trials. Statistical Methods in Medical Research, 17(3), 279–301. doi:10.1177/0962280207080643
[6] Welton, N. J., Sutton, A. J., Cooper, N. J., Abrams, K. R., & Ades, A. E. (2012). Evidence Synthesis for Decision Making in Healthcare. Wiley-Blackwell.
[7] Dias, S., Welton, N. J., Caldwell, D. M., & Ades, A. E. (2010). Checking consistency in mixed treatment comparison meta-analysis. Statistics in Medicine, 29(7-8), 932–44. doi:10.1002/sim.3767
