Kirill Zhudenkov (1), Robert Palmér (2), Alexandra Jauhiainen (3), Gabriel Helmlinger (4), Oleg Stepanov (1), Kirill Peskov (1,5), Ulf G Eriksson (2), Ulrika Wählby Hamrén (2)
(1) M&S Decisions LLC, Moscow, Russian Federation, (2) Quantitative Clinical Pharmacology, Early Clinical Development, IMED Biotech Unit, AstraZeneca, Gothenburg, Sweden, (3) Biometrics, Early Clinical Development, IMED Biotech Unit, AstraZeneca, Gothenburg, Sweden, (4) Quantitative Clinical Pharmacology, Early Clinical Development, IMED Biotech Unit, AstraZeneca, Boston, USA, (5) I.M. Sechenov First Moscow State Medical University of the Russian Ministry of Health, Moscow, Russian Federation
Introduction: Lung function, measured as forced expiratory volume in one second (FEV1), and exacerbations are two important endpoints commonly evaluated in chronic obstructive pulmonary disease (COPD) clinical trials. The correlation between FEV1 and exacerbation risk provides an opportunity to apply joint modelling of the endpoints, to potentially increase statistical power and enable assessment of efficacy in shorter and smaller clinical trials.
Objectives: To evaluate the potential usefulness of joint modelling in respiratory disease by quantifying the association between longitudinal FEV1 and the risk of exacerbation in COPD.
Methods: A joint model consisting of two sub-models, a Cox proportional hazards model for time-to-first exacerbation and a linear mixed-effects (LME) model for longitudinal pre-dose FEV1, was developed to evaluate the association between FEV1 and the risk of exacerbation. The sub-models were linked with an association parameter which describes how the estimated individual changes in FEV1 affects the exacerbation hazard. The effects of baseline covariates were tested on the exacerbation hazard. Patient-level data from a 12-month phase 3 clinical study in moderate-to-severe COPD [1], evaluating efficacy of fixed-dose combinations of a long-acting beta agonist (LABA) bronchodilator, formoterol, and an anti-inflammatory inhaled corticosteroid (ICS), budesonide, were used for model development. To evaluate the consistency of the association across studies and mechanisms of action, model parameters were subsequently re-estimated based on two additional phase 3 studies [2,3] and per treatment arm. A comparison to a standard Cox proportional hazards model was also performed. The JM-package [4] in R was used for modelling.
Results: A significant (p<0.0001) association between FEV1 and exacerbation risk was estimated, with an approximate 8-9% reduction in exacerbation risk for a 100 mL improvement in FEV1. This estimate was consistent across the three trials and across treatment arms. When considering the limited treatment effects on FEV1 in the COPD studies used in our analysis, on average around 50-90 mL, the average exacerbation risk reduction related to FEV1 improvements is ~4-7%. The instantaneous risk reduction related to the LABA-ICS combination treatment was approximately 30-35% in these studies. Thus, only a minor part of the exacerbation risk is accounted for by longitudinal changes in FEV1.
Breathlessness score and exacerbation history in the previous year were found to be important baseline predictors of exacerbation risk.
Conclusions: The consistent association between longitudinal FEV1 and exacerbation risk across studies supports the usefulness of applying joint modelling in analysis of COPD clinical trials. Due to the relatively modest contribution of FEV1 effects to exacerbation risk reduction, however, multivariate joint models including additional endpoints/biomarkers (e.g. breathlessness score measured longitudinally) should be considered to optimize statistical inference and predictions.
References:
[1] Rennard et al., Drugs 2009;69(5):549-65
[2] Tashkin et al., Drugs 2008;68(14):1975-2000
[3] Sharafkhaneh et al., Respir Med 2012;106(2):257-68
[4] Rizopolous, J Stat Software 2010;35(9):1-33
Reference: PAGE 28 (2019) Abstr 8957 [www.page-meeting.org/?abstract=8957]
Poster: Drug/Disease Modelling - Other Topics