A Prediction Scoring Model for the Effect of Withdrawal or Addition of Inhaled Corticosteroids in Patients with Chronic Obstructive Pulmonary Disease

Purpose: The aims of this study were to develop a scoring model that predicts the effects of withdrawing inhaled corticosteroids (ICSs) from triple therapy and to examine its adaptability when applied to assess the effect of adding ICSs to dual bronchodilators patients with chronic obstructive pulmonary disease (COPD).Patients and Methods: A scoring model was developed using the IMPACT study dataset, consisting of 2389 COPD patients treated with triple therapy before enrollment (ICS withdrawal dataset). The developed model consisted of COPD duration, Acute exacerbation history, Sex, Pulmonary function tests, blood Eosinophil count, and Race (CASPER) and was used to predict composite events of moderate-to-severe exacerbation, all-cause mortality, and pneumonia. Treatment heterogeneity was assessed using Cox interaction analyses. The CASPER model was applied to 540 COPD patients treated with dual bronchodilator before enrollment (ICS addition dataset). Validity was assessed using Harrell's C-index, time-dependent receiver operating characteristic curves, and calibration plots.Results: The cumulative incidence of the composite event was 60.1% over 12 months in the ICS withdrawal dataset. Cox interaction analyses revealed that ICS was different according to race and blood eosinophil counts. The hazard ratios (HRs) for dual bronchodilator compared with triple therapy were 1.318 (95% confidence interval (CI)=1.170-1.485; P-value <0.001) in whites and 0.922 (95% CI = 0.712-1.195; P-value=0.541) in other races. The treatment effect was different in the eosinophil count >= 0.3 group (HR = 1.586; 95% CI = 1.274-1.975) and in the eosinophil count = 0.1-0.3 group (HR = 1.211; 95% CI = 1.041-1.408), but it was same in the eosinophil count <0.1 (HR = 1.009; P-value=0.940). The CASPER model performed well with good discrimination and calibration, which were superior to the prediction based on exacerbation history and blood eosinophil count. Conclusion: The presented CASPER model might be able to predict and compare the risk of composite events when dual bronchodilator or triple therapy is administered to COPD patients.