Can changes in blood eosinophils (white blood cells) following treatment with inhaled corticosteroids predict clinical response in chronic obstructive pulmonary disease? A post-hoc analysis of the FLAME trial.

Proposal
11385

Title of Proposed Research

Lead Researcher

Prof. Jørgen Vestbo DMSc, FRCP, FERS, FMedSci.

Affiliation

Division of Infection, Immunity and Respiratory Medicine, School of Biological Sciences, The University of Manchester, Manchester, UK. and North West Lung Centre, Wythenshawe Hospital, Manchester University NHS Foundation Trust, Manchester, UK.

Funding Source

Potential Conflicts of Interest

Data Sharing Agreement Date

16 August 2021

Lay Summary

Chronic obstructive pulmonary disease (COPD) is a frequent and burdensome long-term lung disease causing persistent, progressive respiratory symptoms and loss of quality of life. Inhaled corticosteroids (ICS) are medications that are administered by inhalation, directly to the lungs. By reducing the ongoing inflammation, they decrease the frequency of exacerbations (respiratory symptom flare-ups), improve the quality of life and the function of the lungs. However, they are also associated with side effects, that include an increased risk of pneumonia, osteoporosis and diabetes. While all patients are at risk of the steroids' side effects, not everyone with COPD might gain benefit from ICS. Therefore, it is important to identify accurate blood tests that will point out patients who may gain benefit, in order to personalize their administration and to avoid putting people at risk of side effects for no benefit.It appears that only patients with raised blood eosinophils, which are cells of the immune system, respond to the administration of ICS. As a result, blood eosinophils are increasingly used to guide the administration of ICS. However, it has been suggested that the administration of ICS may influence the levels of eosinophils. As a result, using blood eosinophils without taking into consideration whether ICS were used at the time of the measurement may not be optimal. In a recent study, we tested if we can accurately predict how patients with COPD respond to treatment with ICS using different test results, all based on blood eosinophils: (i) blood eosinophils measured while patients were not receiving any steroids, (ii) blood eosinophils measured while patients were receiving ICS treatment, and (iii) the change in eosinophil levels after the ICS therapy, compared to baseline. We found that change in eosinophils following treatment with ICS may be more accurate in predicting whether ICS are beneficial in COPD. Moreover, we found for the first time that in 20% of participants whose blood eosinophils rose after ICS therapy, treatment with ICS was harmful. More specifically, it was associated with more exacerbations and a faster decline in the lung function. Higher eosinophil levels measured while patients are not receiving any steroids could also predict response to treatment with ICS (but less accurately). Finally, blood eosinophils measured while patients were treated with ICS did not appear accurate in predicting response to treatment with ICS.These findings have important clinical implications, since at present we use indistinctively eosinophils measured while patients are not receiving steroids, or while they are receiving ICS, possibly giving suboptimal clinical advice, especially in the second group of patients. Eosinophils change might be a much more accurate biomarker. However, while our study had strengths, it also had some limitations and our findings will need to be confirmed in a second retrospective study before considering testing them in prospective controlled clinical trials or introducing them in clinical practice.We would like to re-test our findings in the FLAME study population, a large, controlled clinical trial comparing a combination of an ICS with a bronchodilator (a medication that helps open-up the airways), versus the combination of two bronchodilators. Our hypothesis is that change in blood eosinophils after ICS therapy can more accurately predict if patients would respond better to the ICS containing treatment or the other treatment. The primary question of this study will be whether change in blood eosinophils could predict treatment response to ICS with regards to the rate of exacerbations. Secondary outcomes will include whether blood eosinophils can predict clinical response to ICS with regards to (i) different types of exacerbations, (ii) lung function decline, (iii) time-to-first exacerbation, (iv) health status, (v) risk of pneumonia.

Study Data Provided

[{ "PostingID": 4595, "Title": "NOVARTIS-CQVA149A2318", "Description": "A 52-week Treatment, Multi-center, Randomized, Double-blind, Double Dummy, Parallel-group, Active Controlled Study to Compare the Effect of QVA149 (Indacaterol Maleate / Glycopyrronium Bromide) With Salmeterol/Fluticasone on the Rate of Exacerbations in Subjects With Moderate to Very Severe COPD. (FLAME)." }]

Statistical Analysis Plan

In our main analysis, we will include patients who (i) were receiving ICS for at least for 4 weeks prior to recruitment, (ii) who did not receive any systemic corticosteroids during the run-in period, and (iii) who had their blood eosinophil counts measured before and after the run in period. Therefore, for each of the participants we will have one measurement of EOS off steroids for at least 4 weeks (measured after the run-in period) and one measurement of EOS on ICS for at least 4 weeks (measured before the run-in period). Using these two values we will also be able to evaluate EOS change (=EOS on ICS - EOS off steroids).For evaluating whether the previously mentioned eosinophil biomarkers (EOS off steroids, EOS on ICS and EOS change) can predict clinical response to ICS/LABA versus LABA/LAMA we will use the following statistical tests/models:1. Generalised linear model assuming negative binomial distribution for assessing the rate of exacerbations and rate of pneumonias.2. Cox proportional hazards model (using the Kaplan-Meier curves and Schoenfeld residuals for assessing the proportional hazard assumption) for evaluating time-to-first exacerbation, time-to-first episode of pneumonia and mortality.3. Mixed effects model repeated measures (MMRM) for evaluating change over time in lung function (FEV1, FVC, FEV1/FVC) and in health status (SGRQ).In all analyses we will account for the following confounding factors: age, sex, exacerbations history at baseline, smoking status, FEV1 at baseline.For each outcome, we will repeat the analyses using (i) EOS off steroids, (ii) EOS on ICS and (iii) EOS change. We will use the Akaike Information Criterion to evaluate the quality of each model and to identify the most predictive biomarker.We plan to conduct the following sensitivity analyses:1. We will exclude all lung function values measured during the first two months of treatment, to account for the increase that the initiation of inhaled treatment confers to the mean FEV1, as our aim is to evaluate the impact of ICS on lung function decline over time, and not the acute effects of ICS on FEV1.2. Extended study population: When EOS off steroids and/or EOS on ICS measurements are not available during the visits after and/or before the run-in period, we will accept any EOS values measured at any study timepoint, while patients were not receiving any steroids, or were receiving ICS, respectively, to maximize the study population (e.g. for patients who were not receiving ICS at baseline, EOS on ICS might be captured during study treatment, if they were allocated to the LABA/ICS group).Statistical analyses will be performed using R statistical software, version 3.4.3 or newer (R Foundation for Statistical Computing, Vienna, Austria).

Publication Citation