Literature Model for FEV1 in COPD Trials – Separating the Dynamic Components of Placebo Effect, Disease Progression and Interacting Drug Effects
Jakob Ribbing (1), Christine Falcoz (2), Itzela Correa (2, 3), Steven W Martin (4)
(1) Pfizer AB, Sollentuna, Sweden; (2) Pharsight Consulting Services, A division of Certara, St. Louis, MO, USA;(3) Currently employed by Metrum Research Group LLC (4) Pfizer Inc, Cambridge, MA, USA
Objectives: In chronic obstructive pulmonary disease (COPD) the forced expiratory volume in one second (FEV1) is the most important biomarker for lung function, and is used for dose selection.[1] The objective of this work was to develop a longitudinal model for FEV1 based on literature (summary level data) on COPD trials, to quantify placebo effect and disease progression, as well as treatment effects and their interaction in combination treatment.
Method: Criteria for inclusion were a) randomised, blinded COPD maintenance trial b) including treatments class: LABA, LAAC, ICS or PDE4i c) FEV1: troughs were used when available. Pre-study-drug measurements occurring after administration of a short-acting bronchodilator (post SABD) were otherwise used.
Background therapy was generally allowed and any interaction was handled by drug-drug interaction models. Estimation was performed in NONMEM.
Results: The database included 87 studies, totalling 59775 patients across 228 treatment arms (including 72 placebo arms). These trials reported 1080 FEV1 values, each representing the mean in an arm, at a certain time during the study. Study durations ranged: 1 week to 4 years.
The final structural model included components which described the baseline and the time course for: a) placebo response b) disease progression and c) drug effect. The drug-effect model included separate estimates for 13 compounds and described dose-response where possible. Drug interactions were estimated for the combination LABA+LAAC as well as for LABA or LAAC measured post SABD. An anti-inflammatory agent (ICS or PDE4i) in combination with a direct bronchodilator (LABA or LAAC) provided efficacy as the sum of the two mono components. Random inter-study variability (ISV) was included in all four structural components and in addition inter-arm variability in baseline. Important covariates were identified.
Conclusion: This exercise a) consolidated relevant information across compounds, in terms of efficacy, dose-response and time course for onset of drug-effect b) positions each published trial result into a broader evidence based context and c) illustrates the impact of PD interactions and other covariates.
Furthermore our FEV1 model will be developed to predict exacerbations, providing predicted trough FEV1 even for trials that only measured FEV1 post SABD. This is an important efficiency gain since the late phase exacerbation trials require thousands of patients and at least one year duration.
Reference:
[1] Outcomes for COPD pharmacological trials: from lung function to biomarkers. M. Cazzola et al. Eur Respir J 2008; 31: 416-468.
