Predicting reductions in chronic obstructive pulmonary disease (COPD) exacerbations from FEV1 – A model-based meta-analysis of literature data from controlled randomized clinical trials

Jakob Ribbing (1,2,3), Julia Korell (2,4), Frank Cerasoli (1,5), Peter A. Milligan (1), Steven W. Martin (1), Mats O. Karlsson (2)

(1) Pfizer LTD; (2) Uppsala University. Current employment: (3) Pharmetheus AB; (4) Model Answers Pty Ltd; (5) Medical Dynamics

Objectives: To describe the relationship between forced expiratory volume in one second (FEV₁) and annual rate of moderate-severe[1] exacerbations (ER) utilizing summary-level, literature data.

Shorter duration Phase 2 studies assess FEV₁ whereas Phase 3 chronic maintenance studies assess the registerable endpoint (prevention of COPD exacerbations).

Methods: Data was extracted from 29 randomized trials (80 treatment arms), of 43 472 patients. As predictors of ER, model-predicted trough FEV₁[2] at baseline and Week 12, as well as covariates, were investigated using NONMEM. Placebo ER was a function of covariates and interstudy variability. The ER ratio (treatment vs. placebo) was described by separate functions for FEV₁ efficacy (ΔΔFEV₁) from direct bronchodilators (long-acting; LABD) and anti-inflammatory (AI) agents. Outcomes were derived as point estimate [95%-Confidence interval] versus placebo/reference arm.

Results: The final model predicted that placebo ER increased with a) disease severity (FEV₁%Predicted), b) fraction of (ICS experienced) patients required to wash out from ICS (ICS_washout), and c) inclusion criteria requiring a history of exacerbations.

The log(ER-ratio) (treated vs untreated), was described by separate linear-slopes for LABD and AI ΔΔFEV₁, and in addition for %ICS_washout; by a ΔΔFEV_1AI-E_max model. The model predicted that for log(ER-ratio) < -0.2 (>18% ER reduction), LABDs must achieve at least a ΔΔFEV₁ 122 mL [114mL−132mL] improvement (over placebo/reference). For a scenario with 62% ICS_washout, an AI treatment (ICS or PDE4i) must achieve at least a ΔΔFEV₁ 45 mL [17mL−79mL] improvement, for log(ER-ratio) < -0.2.

Conclusions: The investigated AIs have modest efficacy on FEV₁, but if patients are washed out from ICS, these treatments achieve reductions in ER comparable to the new-generation LABD. The outcomes from this analysis may be applied while designing Phase 3 efficacy studies, pharmaco-economic outcomes studies[3,4], and quantifying comparative effectiveness of available treatments.

References:
[1] Mario Cazzola, et al. Outcomes for COPD pharmacological trials: from lung function to biomarkers. Eur Respir J, 2008. 31(2): p. 416-69.
[2] Julia Korell, Steven W. Martin, Mats O. Karlsson and Jakob Ribbing. A model-based longitudinal meta-analysis of FEV₁ in randomized COPD trials. Clin Pharmacol Ther, 2015 Aug 14 [Epub ahead of print].
[3] Julia F. Slejko, Richard J. Willke, Jakob Ribbing and Peter Milligan. Translating Pharmacometrics to a Pharmacoeconomic Model of COPD.
[4] Richard J. Willke, Julia F. Slejko, Jakob Ribbing and Peter Milligan. Calibration of a Health Economic Microsimulation Model to Pharmacometric Model-Based Meta-Analysis Predictions: A Chronic Obstructive Pulmonary Disease Example.

Reference: PAGE 25 () Abstr 3699 [www.page-meeting.org/?abstract=3699]

Poster: Drug/Disease modeling - Other topics

PDF poster / presentation (click to open)