Reply to Suissa: Mortality in IMPACT: Confounded by Asthma?

From the Authors:

We read with interest the comments by Dr. Suissa on our manuscript (1) but disagree with his premise that the IMPACT (Informing the Pathway of COPD Treatment) results were driven by the withdrawal of steroids from patients with asthma.

The population enrolled in IMPACT was a typical chronic obstructive pulmonary disease (COPD) population with symptomatic disease and a history of exacerbation. IMPACT was carefully designed (2) on the basis of standard clinical parameters that are not only endorsed by major scientific societies but also routinely used in worldwide clinical practice. All patients within the trial met American Thoracic Society and European Respiratory Society criteria for COPD; current asthma was an exclusion, and investigators enrolled patients only if their symptoms were due to COPD. Clearly, patients with a previous history of asthma can still develop COPD (3). The average age of the population was 65 years and exhibited fixed airflow obstruction with an average FEV1% predicted of 45.5. All patients were active or former smokers with an average of almost 47 pack-years of cigarette exposure (4). There were no clinically relevant differences from other large COPD trials; in fact, levels of reversibility to albuterol were actually lower in IMPACT (only 18% of patients were reversible to albuterol in IMPACT compared with FLAME [45%], KRONOS [43%], and ETHOS [30.6%]) (5–7) with similar blood eosinophil levels. The IMPACT population is typical of a population with COPD that is clearly recognizable to any clinician caring for such patients.

There were differences in disease severity in the population of patients who came into the trial on inhaled corticosteroids (ICS) compared with those who did not. For example, rates of moderate or severe exacerbations were higher on all arms in the trial for those who came in on a triple regimen compared with those receiving long-acting muscarinic antagonist monotherapy. This would be expected as international recommendations, such as Global Initiative for Chronic Obstructive Lung Disease, suggest the use of ICS in more severe patients (8). Thus, we might expect to see a survival benefit in those who came into the trial on ICS because they have the greatest risk for exacerbations and thus the greatest risk for death (9). We believe that the most likely cause of the observed survival benefit was the reduction in recurrent exacerbations, especially those leading to hospitalization, demonstrating the benefit of ICS in this patient population.

It is important to recognize that IMPACT was not an ICS withdrawal study. Although ∼77% of patients entered the trial on ICS, because of the 2:2:1 randomization, only approximately 15% of the overall population underwent withdrawal of ICS. The vast majority of the population (85%) did not experience ICS removal. In addition, deaths occurred in all arms throughout the trial, indicating no “surge” in deaths caused by abrupt withdrawal of ICS. Overall mortality on the long-acting muscarinic antagonist–long-acting β2 agonist arm was actually lower than what has been previously observed in similar patients with advanced COPD (10, 11), also strongly suggesting that abrupt ICS withdrawal was not the cause of the findings.

Even if we were to believe Dr. Suissa’s argument that ICS withdrawal was harmful, we would then have to conclude that the addition of ICS was beneficial for these patients in the first place.

What is clear is that most patients who met the IMPACT inclusion criteria benefited from triple therapy compared with dual therapy. Patients with symptomatic COPD and a history of exacerbation who received triple therapy with fluticasone furoate/umeclidinium/vilanterol experienced clinically relevant improvements in lung function and health-related quality of life, reduction in exacerbations and hospitalizations, and now a confirmed additional benefit of improved survival compared with patients randomized to umeclidinium/vilanterol.