Cigarette smoking and response to inhaled corticosteroids in COPD

To the Editor:

Inhaled corticosteroids (ICS) and long-acting β-agonists (LABA) are frequently used in patients with chronic obstructive pulmonary disease (COPD). Combination treatment with ICS/LABA improves lung function and quality of life, and reduces exacerbation frequency compared to treatment with either ICS or LABA alone. Although it is presumed that continued cigarette smoking impairs acute responses to these medications in patients with COPD, there is little direct evidence to support this view. Studies of ICS use in asthma have shown less short-term improvement in lung function and reduced anti-inflammatory effects in active smokers compared to non-smokers [1, 2]. Although similar effects are plausible in COPD, they have not been definitively demonstrated. We hypothesised that former smokers with COPD would have greater short- and long-term changes in lung function, respiratory-related quality of life, and exacerbation risk, in response to ICS than continuing and intermittent smokers.

We used data from a large double-blind, randomised, controlled trial, comparing the effects of combination ICS (fluticasone furoate, FF) and LABA (vilanterol, VI), with ICS (FF) alone, LABA (VI) alone, and a placebo (the Study to Understand Mortality and Morbidity in COPD (SUMMIT) trial). We performed a post hoc analysis to assess the relationship between smoking status and respiratory outcomes. The details of SUMMIT have been previously published [3]. Briefly, participants were 40–80 years of age, with at least a 10 pack-year smoking history, and post-bronchodilator FEV₁/FVC≤0.70 with a post-bronchodilator FEV₁ 50%–70% predicted, and enhanced cardiovascular risk. Spirometry assessments were made at enrolment and every 3 months thereafter, and respiratory-related health status assessments (St George's Respiratory Questionnaire (SGRQ)) were measured in a subset, at enrolment and at 3, 6 and every 6 months thereafter. Exacerbations were defined as the worsening of respiratory status, necessitating therapy with antibiotics and/or systemic corticosteroids, and those that required hospitalisation were deemed severe exacerbations. Participants were classified as either current or former smokers, based on smoking status at enrolment. The primary outcome was change in post-bronchodilator FEV₁ at each visit, compared to baseline, up to 1 year after enrolment, and secondary outcomes were change in SGRQ up to 1 year after enrolment and exacerbation risk over the entire study. All analyses for outcomes between current and former smokers were performed in comparison to changes in the placebo group. The FEV₁ and SGRQ were analysed using mixed model repeated measures. The models included terms for treatment, age, gender, baseline measurement, visit, smoking status, baseline measurement by visit, treatment by visit, smoking status by visit, smoking status by treatment and smoking status by treatment by visit. Exacerbation rates were analysed using negative binomial regression. The model included terms for treatment, age, gender, number of previous COPD exacerbations in the 12 months prior to screening, smoking status and smoking status by treatment. No adjustments were made for multiple comparisons.

Of the 16 485 participants, 7678 (47%) were current smokers at enrolment, and 8807 (53%) were former smokers. Current smokers included 27% females, 63±8 (mean ±SD) years of age, and body mass index (BMI) of 27±6 kg·m⁻², whereas former smokers included 25% females, 67±7 years of age and BMI 29±6 kg·m⁻². During the course of the study, 1859 (11%) participants quit smoking, 81 (<1%) switched from non-smoker to active smoking status, 778 (5%) had multiple changes in their smoking status, and 13 767 (84%) had no change in smoking status.

Compared with the placebo, the adjusted mean change in FEV₁ at 3, 6, 9 and 12 months was greater in former smokers than in current smokers (figure 1), with both FF (mean difference 30, 95% CI 9–51 mL; p=0.005) and FF/VI (mean difference 22, 95% CI 1–43 mL; p=0.038), but not with VI alone (not shown, mean difference −6, 95% CI −27 to 15 mL; p=0.572). For the SGRQ, the treatment differences associated with FF versus the placebo were similar, regardless of smoking status (mean difference −1.0, 95% CI −2.8 to 0.8 units; p=0.267), and for FF/VI versus the placebo (mean difference −1.1, 95% CI −2.9 to 0.7 units; p=0.226). However, for VI versus the placebo, there was greater benefit for former smokers than for current smokers (mean difference −2.5, 95% CI −4.3 to −0.8 units; p=0.005). In addition, former smokers had a greater percentage reduction in rates of moderate or severe exacerbations when on FF/VI compared to the placebo than did current smokers (36%, 95% CI 27%–43% versus 19%, 95% CI 7%–29%; p=0.013). No difference in exacerbation rate reduction was observed between former versus current smokers taking FF versus the placebo (17%, 95% CI 6%–26% versus 6%, 95% CI −8% to 17%; p=0.176), or taking VI versus the placebo (14%, 95% CI 4%–24% versus 4%, 95% CI −9% to 16%; p=0.227). No differences in hospitalised exacerbation rate reduction were observed among former versus current smokers taking FF/VI versus the placebo (14%, 95% CI −22% to 40%; p=0.396), or FF versus the placebo (−16%, 95% CI −64% to 18%; p=0.413), or VI versus the placebo (4%, 95% CI −35% to 32%; p=0.806). There was no evidence to suggest that smoking status influenced the risk of pneumonia associated with treatment (the overall p-value was 0.960 for smoking by treatment for all groups, and 0.775 for FF versus the placebo, 0.584 for FF/VI versus the placebo and 0.802 for VI versus the placebo). After adjustment for age, sex, smoking status and treatment by smoking status interaction, the hazard ratios for pneumonia in current and former smokers, comparing FF/VI versus the placebo, were 1.10, 95% CI 0.81–1.50 and 0.99, 95% CI 0.79–1.25, respectively; in FF compared with the placebo, 1.07 95% CI 0.79–1.45 and 1.01, 95% CI 0.80–1.28, respectively; and in VI compared with the placebo, 0.75, 95% CI 0.53–1.04 and 0.71, 95% CI 0.55–0.91, respectively.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1

Comparison of change in forced expiratory volume in 1 s (FEV₁) at 90, 180, 270 and 360 days. Compared to current smokers, former smokers showed a greater response to active treatment with both fluticasone furoate (FF) (mean difference 30, 95% CI 9–51 mL; p=0.005) and FF/vilanterol (FF/VI) (mean difference 22, 95% CI 1–43 mL; p=0.038), adjusted for age, sex, baseline FEV₁, visit, smoking status, FEV₁ by visit, smoking status by visit, smoking status by treatment and interaction for smoking status by treatment by visit.

We demonstrated that compared to former smokers, current smokers have a blunted FEV₁ response with ICS and a smaller reduction in exacerbation frequency with ICS/LABA. Data on continued smoking and response to therapy are mostly derived from studies in asthma. In a small study of patients with mild-to-moderate asthma, continuing smokers had less improvement in FEV₁/FVC at 6 months with budesonide compared to non-smokers [4]. Chalmers et al. [1] also found that in mild asthma, non-smokers had a significantly greater increase in morning peak expiratory flow (PEF) after 3 weeks of fluticasone therapy than did continuing smokers. Two randomised controlled trials have confirmed that continued smoking is associated with a lower improvement in PEF [2, 5]. To the best of our knowledge, these assessments have not been made for ICS or ICS/LABA in COPD, and studies of the impact of smoking on responses to other COPD medications have shown varied results. Moita et al. [6] found no difference in FEV₁ response by smoking status, but analyses of the UPLIFT study showed that short- and long-term improvement in FEV₁ with tiotropium was greater in continuing smokers than in ex-smokers or intermittent smokers [7]. Post hoc analyses suggest that the exacerbation reduction effect of azithromycin might not be seen in those who actively smoke [8].

The mechanisms underlying the blunted response to these medications in active smokers are unknown, although the pro-inflammatory effects of cigarette smoke have been well described and could blunt the anti-inflammatory effects of ICS [9]. This is supported by our finding of no difference in response for FEV₁ and exacerbations by smoking status with VI alone. Alternatively, the poor response could be due to relative steroid resistance mediated by a reduction in the enzyme histone deacetylase 2, which is essential for glucocorticoid binding to their receptors [9]. We found no similar differences in change in the SGRQ, likely because it reflects outcomes with more heterogeneous mechanisms.

One limitation of the present study is our combination of intermittent and continued smokers into one group, based on smoking status at enrolment. However, previous studies on lung function and exacerbations suggest that continuing and intermittent smokers behave similarly in terms of lung function decline, as well as exacerbation frequency [10, 11]. Active smoking status was self-reported. In addition, smoking status changed in a small proportion of patients over the course of the study. Some exacerbations might have gone unreported, but these are likely to occur across all groups of the study. Our findings have considerable public health importance, as 39% of all patients with COPD continue to smoke, and smoking cessation could result in better overall outcomes [12].

In summary, similar to the effects observed in patients with asthma, continued smoking is associated with an impaired response to ICS and thereby affects the attainment of important clinical outcomes in COPD patients.

Disclosures