One More Time: The Impact of Inhaled Corticosteroid Withdrawal on IMPACT

We thank the editors for the opportunity to respond to the editorial written by Dr. Samy Suissa on our original article, “The Effect of Inhaled Corticosteroid Withdrawal and Baseline Inhaled Treatment on Exacerbations in the IMPACT Study: A Randomized, Double-Blind, Multicenter Clinical Trial,” in this issue of the Journal (pp. 1237–1243) (1). We thank Dr. Suissa for his contribution to the ongoing scientific dialogue on this important topic. We would, however, like to take this opportunity to clarify several points.

The first point is that Dr. Suissa notes that the IMPACT (The Informing the Pathway of COPD Treatment) study included patients with a history of asthma. Although this is true, it is important to note that patients with a current diagnosis of asthma were excluded. All patients met American Thoracic Society/European Respiratory Society and Global Initiative for Chronic Obstructive Lung Disease criteria for chronic obstructive pulmonary disease (COPD), had a mean age of approximately 65 years, tobacco exposure of nearly 47 pack-years, and fixed airflow obstruction with an FEV₁% predicted of 45.5. Furthermore, investigators also excluded patients whose symptoms were not believed to be due to COPD. As a prior diagnosis of asthma in current COPD is also relatively common, we believe these inclusion criteria better represent the patient population physicians might actually encounter in practice and underscores the generalizability of the IMPACT trial data.

Dr. Suissa also incorrectly notes in his editorial that we examined a “large number of patients (7,360) who had ICS abruptly withdrawn.” Because of the 2:2:1 randomization schema (inhaled corticosteroid [ICS]/long-acting muscarinic antagonist [LAMA]/long-acting β₂-agonist [LABA]:ICS/LABA:LAMA/LABA), only 20% of these patients (n = 1,481) would have had ICS withdrawn (not 7,360 patients), which is 14% of the total IMPACT patient population. The remaining 86% of patients continued seamlessly on an ICS or had not been on an ICS because IMPACT did not employ an artificial washout period, again mimicking clinical practice.

A further concern raised by the editorial was a semantic one, noting that we should have reversed the estimates to compare LAMA/LABA therapy versus triple therapy, termed by Dr. Suissa “effect of ICS withdrawal” as opposed to triple therapy versus LAMA/LABA therapy “exacerbation reduction.” Regardless, we believe it is important to point out that we clearly see a statistically significant 35% decrease in severe exacerbations (risk ratio, 0.65) comparing triple therapy with LAMA/LABA therapy regardless of prior ICS use. Hence, the effect of triple therapy on severe exacerbations cannot be attributed to ICS withdrawal.

Dr. Suissa also raises the point that our analysis excludes “early exacerbations” but fails to exclude “early exacerbators,” suggesting that the majority of ICS effect is being driven by a small group of patients who are in fact harmed by ICS withdrawal as opposed to a patient population that experiences longer-term benefits. Again, this concept fails to account for the benefit of triple therapy versus LAMA/LABA therapy on severe exacerbations, irrespective of ICS use. Dr. Suissa also states that “by pooling rather than splitting, this analysis fails to identify the key patient groups who could benefit from ICS withdrawal or from continuation.” We would like to clarify that we have not pooled data, but we present the entire intention-to-treat population. Furthermore, it would have been statistically irresponsible to “split” off a subgroup of patients from an analysis after randomization, particularly on the basis of events that occurred after randomization. Furthermore, our graphs of cumulative exacerbation data demonstrate that events continue to increase throughout the trial.

Finally, we must underscore that when considering the risk:benefit profile of triple therapy, in prespecified secondary analyses, we see a reduction in all-cause mortality among patients treated with triple therapy compared with those treated with LAMA/LABA therapy.