Effect of smoking status on the efficacy of the SMART regimen in high risk asthma

The optimal management of people with asthma with a significant smoking history is uncertain. The aim of this study was to determine whether the efficacy/safety profile of single combination inhaled corticosteroid (ICS)/long acting beta‐agonist (LABA) inhaler maintenance and reliever therapy is influenced by smoking status.


INTRODUCTION
The management of individuals with asthma who smoke is an important clinical priority. 1,2 Cigarette smoking is associated with greater morbidity from asthma and a higher risk of severe exacerbations. [3][4][5] Amongst individuals with asthma, heavy smokers are at greater risk of asthma mortality compared with non smokers. 6 The optimal management of asthmatics who smoke is uncertain. Large randomized controlled trials (RCTs) that inform asthma management guidelines generally exclude current smokers or ex smokers with at least a 10 pack year history to avoid recruitment of patients with concomitant chronic obstructive pulmonary disease (COPD). 7 Furthermore, it is known that individuals with asthma who smoke benefit less from inhaled corticosteroid (ICS) and oral corticosteroid therapy in terms of symptoms, lung function and risk of severe exacerbations. [8][9][10][11][12][13] In patients with severe asthma, the use of a single combination ICS/fast-acting long acting beta-agonist (LABA) inhaler as both maintenance and reliever therapy, (the SMART regimen), leads to reduced risk of severe exacerbations compared with combination ICS/LABA inhaler as maintenance and short acting beta-agonist (SABA) for reliever therapy. [14][15][16] This is based on RCTs that did not report treatment effects in relation to smoking status. [14][15][16] It is therefore uncertain if the favourable efficacy/safety profile of the single maintenance and reliever therapy (SMART) regimen can be generalised to patients with severe asthma who have important current or ex smoking histories. The SMART regimen could have a greater relative benefit for smokers with asthma because the increased ICS dose may partially reverse the reduced ICS responsiveness. 10 Alternatively, the SMART regimen may be less beneficial for smokers with asthma because of lesser efficacy from the increased use of ICS during worsening symptomatic asthma. Smokers with asthma may also have different responses to variable dosing of LABA and SABA therapy, compared with non smokers.
In recognition of the potential role for smoking status in response to treatment, this study reports a secondary analysis investigating whether smoking affects the efficacy of the SMART regimen in a RCT of high risk adults with asthma, of whom 51% were current or ex smokers. 17 Our hypothesis was that current and ex smokers would have worse clinical outcomes than non smokers and lesser efficacy from the SMART regimen compared with non smokers.

Design
This multicentre open-label study randomized 303 asthma patients to the SMART or the Standard therapy regimen. 17 The study was approved by the New Zealand Multi-Region Ethics Committee and has the Australian and New Zealand Clinical Trials Registry number ACTRN12610000515099. Full written informed consent was required prior to study participation.

Participants
Participants were aged 16 to 65 years and had a current prescription for ICS with at least one asthma exacerbation (presentation to an Emergency Department (ED) or general practice resulting in a prescription for oral corticosteroids or treatment with spacer-delivered or nebulised bronchodilator, or self-administration of prednisone for asthma for at least 3 days) in the previous year. 17 Exclusion criteria included a diagnosis of COPD or onset of respiratory symptoms after the age of 40 in current or ex smokers with a ≥10 pack year-smoking history. The study protocol is available at http://www. mrinz.ac.nz/uploads/mrinz/SMART_Protocol.pdf.

Interventions
Participants were randomized 1:1 to receive either the SMART regimen, which was 200/6-μg budesonide/ formoterol via metred dose inhaler (MDI) (Vannair; AstraZeneca Limited, Auckland, New Zealand; this is the MDI formulation of Symbicort Turbuhaler) for maintenance (two actuations twice daily) and one actuation as required for symptom relief, or the Standard regimen consisting 200/6-μg budesonide/formoterol via MDI for maintenance (two actuations twice daily) and 100-μg salbutamol via MDI (Ventolin; GlaxoSmithKline Limited, Auckland, New Zealand), 1-2 actuations as required for symptom relief. At the first visit, participants were given a written asthma self-management plan, and inhaler technique was checked. Subsequent visits took place at Weeks 3, 10, 17 and 24.
The Smartinhaler Tracker (Nexus6 Limited, Auckland, New Zealand) electronic monitor was incorporated into all MDIs and recorded the date and time of each actuation. Detailed trial quality control processes took place. 18,19 Data were uploaded from the inhalers at each visit.

Data analysis and study outcomes
Data analysis was by intention to treat the primary outcome variable was the number of participants with at least one severe exacerbation, according to the American Thoracic Society/European Respiratory Society Taskforce criteria: the use of systemic corticosteroids for at least 3 days, or admission to hospital or visit to the ED because of asthma that required systemic corticosteroids. 20 High beta-agonist use was defined as >16 actuations of salbutamol in the Standard regimen and >12 actuations of budesonide/formoterol for the SMART regimen (i.e. > eight actuations of budesonide/formoterol, additional to the four maintenance doses), in 24 h. These definitions were based on self-management plan recommendations for beta-agonist use requiring medical review, 21,22 and supported by the bronchodilator equivalence of 6-μg formoterol to 200-μg salbutamol with repeat dosing in acute asthma. 23,24 For the Standard regimen, marked and extreme beta-agonist overuse was defined as >24 and >32 salbutamol actuations in 24 h, respectively. For the SMART regimen, marked and extreme overuse were defined as >16 and >20 budesonide/formoterol actuations in 24 h, respectively (i.e. >12 and >16 actuations of budesonide/formoterol, additional to the four maintenance doses, respectively).
Odds ratios for the risk of at least one severe exacerbation by randomized group, the primary outcome variable for this analysis, were estimated by logistic regression. Secondary outcome variables analysed by logistic regression were at least one hospital or ED attendance, at least 1 day of beta-agonist overuse, marked overuse or extreme overuse, at least 1 day of one or less budesonide/formoterol inhaler actuations per day, and at least 1 day of no budesonide/formoterol inhaler actions per day over the study period. Relative rates by Poisson regression, with an offset for the observation time, were used for the count variables including number of severe exacerbations, number of courses of oral corticosteroids and number of days of high use, marked overuse or extreme overuse, one or less budesonide/formoterol inhaler actuations per day, and no budesonide/formoterol inhaler actuations per day. Analysis of covariance was used for differences on the logarithm transformed scale, where exponentiation is interpreted as mean ratios, for daily equivalent ICS use. Survival analysis for day to first exacerbation used Cox proportional hazards. Contingency table analysis Smoking status and the SMART regimen was used for oral corticosteroid dose (prednisone equivalent per year) by creating four bands of use. For this outcome, the pre-specified main RCT analysis plan was to seek an appropriate transformation for the dose, such as the logarithm transformation, or use a nonparametric method (the Mann-Whitney test), to compare the groups. After the data were collected, many participants were found to have no oral prednisone use, so neither of these strategies was able to be used. The other continuous variables, which had appropriate data distributions, were analysed by analysis of covariance: forced expiratory volume in 1 s, forced expiratory volume in 1-s percentage predicted and asthma control questionnaire-7 (ACQ-7).
The general analysis strategy for the secondary analysis of the effect of smoking reported here was to test a smoking-treatment interaction term for each outcome variable. Smoking was classified as 'current smoker', 'ex smoker' and 'never smoker'. Participants reported which category they belonged to at the first study visit. Our analysis plan was to report the difference in outcome variables between SMART and Standard for current smokers compared with never smokers, and ex smokers compared with never smokers, if there was evidence of statistical significance, P < 0.05, for an interaction between smoking and randomized treatment. Otherwise, we planned to report the difference in outcome for current smokers versus never smokers, and ex smokers versus never smokers, adjusted for randomized treatment. In this case, the lack of statistical evidence of an interaction would be consistent with the same relative effect of treatment for all smoking categories.

RESULTS
Three hundred and three participants were enrolled between June 2010 and September 2011. Fifty nine (19%) were current smokers, 97 (32%) ex smokers, and 147 (49%) were never smokers ( Table 1, Fig. 1). Current smokers had between 0.3 and 44 pack years of smoking, and ex smokers between 0.2 and 60 pack years of smoking. Current smokers had higher ACQ-7 scores (i.e. worse asthma control) compared with ex and never smokers.

SMART versus Standard regimen, with adjustment for smoking
There was no evidence of interaction between smoking status and randomized treatment interaction term for any outcome measure. This means that the relative effect of randomized treatment was the same for participants with different smoking status (Tables 2 and 3).  4 (1 to 60) 0 (0 to 0) 0 (0 to 0) ICS dose conversion: 500-μg fluticasone = 800-μg budesonide, 1000-μg beclomethasone = 800-μg budesonide. ACQ, asthma control questionnaire; FEV 1 , forced expiratory volume in 1 s; ICS, inhaled corticosteroid; LABA, long acting beta-agonist; no, number; SD, standard deviation; SMART, single ICS/LABA maintenance and reliever therapy.
The outcomes by randomized treatment, adjusted for smoking status, are shown in Tables 2 and 3. The proportion of participants with at least one severe exacerbation was lower in those randomized to the SMART regimen, with an odds ratio of 0.45 (95% CI: 0.26 to 0.77), P = 0.004; P value for interaction between smoking status and treatment 0.29 (Table 2).
After adjustment for smoking status, the number of severe exacerbations was lower in participants randomized to SMART (Table 2). There was no significant difference in the composite systemic corticosteroid exposure between the two regimens following adjustment for smoking status (Table 2). In addition, the ACQ-7 scores at visit 5 were lower in the SMART group, compared with Standard (Table 2).
After adjustment for smoking status, the SMART regimen was not associated with a significantly different proportion of participants with at least one episode of high, marked or extreme beta-agonist overuse when compared with the Standard regimen (Table 3). However, the SMART group had significantly fewer numbers of days of high use, marked overuse, extreme overuse and number of days of high use without medical review within 48 h ( Table 3). The number of days of non-adherence to maintenance therapy was also lower in the SMART group ( Table 3).

Outcomes of smokers, ex smokers and never smokers, adjusted for treatment
After adjustment for treatment regimen, smoking status was associated with differences in the risk of at least one severe exacerbation, an increased number of severe exacerbations, an increased risk of at least one hospital admission or ED attendance, an increased number of courses of oral corticosteroids and increased composite systemic corticosteroid exposure, compared with never smokers. For each of these outcomes, there was a significant difference between ex smokers and never smokers, but no significant difference between current and never smokers (Table 4).
After adjustment for treatment regimen, smoking status was associated with high, marked and extreme beta-agonist overuse, with current smokers and ex smokers having higher rates compared with never smokers (Table 5). After adjustment for treatment regimen, smoking status was associated with overuse days without medical review and the number of days of no budesonide/formoterol actuations, with more days occurring in the current smokers compared with never smokers (Table 5).

DISCUSSION
This study shows that the favourable efficacy/safety profile for the SMART regimen in high risk adults with asthma was similar regardless of smoking status. This suggests that the SMART regimen can be recommended in current and ex smokers, who represent a particularly high risk group.
The broad inclusion criteria of this RCT ensured the findings are widely generalisable to patients with high risk asthma. Fifty one percent of participants were current smokers or ex smokers, so we could robustly assess if the efficacy of the SMART regimen was influenced by smoking status. Our study complements the findings of an analysis of the influence of   Corticosteroid conversion: 100-mg intravenous hydrocortisone = 25-mg oral prednisone. Budesonide dose was converted to prednisone equivalent dose, based on a bioequivalence conversion calculated in a prior study (5000-μg inhaled budesonide = 10-mg oral prednisone). 25 The sum of the prednisone equivalent dose and systemic corticosteroid dose was annualized. The logarithm of the annualized steroid use was the response variable in a weighted normal linear model, with the randomized treatment as a predictor and the treatment exposure time as a weight (individuals with longer periods of treatment exposure were given more weight and those with shorter periods of treatment exposure less weight in the analysis). ACQ, asthma control questionnaire; ED, Emergency Department; no, number; SD: standard deviation; SMART, single maintenance and reliever therapy.  Smoking status and the SMART regimen smoking status on response to two different maintenance dosing regimens for SMART with 200/6 μg of budesonide/formoterol, one versus two actuations twice daily. 26 In that study, there was a significantly greater reduction in severe exacerbations by use of two maintenance budesonide/formoterol inhalations twice daily versus one inhalation twice daily in smokers, but not in non smokers. These findings suggested that the budesonide/formoterol maintenance dose 200/6 two puffs twice a day is the preferred maintenance dose for smokers. Our primary outcome was the number of participants with at least one severe exacerbation, defined in accordance with the American Thoracic Society/European Respiratory Society Task Force criteria. 20 The SMART regimen reduced the odds of a severe exacerbation by about 50%, with adjustment for smoking status, and with no significant interaction between smoking status and randomized treatment. When adjusted for treatment regimen, smoking status had a significant effect on the number of severe exacerbations with ex smokers having a higher risk of severe exacerbations compared with never smokers. This suggests that the absolute reduction in the number of severe exacerbations with the SMART regimen is greater in ex smokers because of their higher risk of this outcome compared with non smokers.

Smoking status and the SMART regimen
An important and novel feature of the RCT was the use of electronic monitors in all MDIs to capture all actuations self-administered by participants in the study. 18,19 This builds on previous studies, [14][15][16] by enabling an in-depth assessment of the relative safety of the SMART regimen including measures of betaagonist overuse, delay in seeking medical assistance in worsening asthma and systemic corticosteroid exposure. There were similar proportions with at least one episode of high, marked and extreme overuse in the two treatment groups; however, the SMART regimen led to a 40-50% reduction in the number of days of high, marked and extreme overuse episodes, with no significant interaction with smoking status. We observed that current and ex smokers had a twofold to sixfold greater rate of these overuse episodes than never smokers, suggesting that there will be greater absolute benefit for this outcome in smokers. The SMART regimen also reduced the risk of delay in seeking medical review during worsening asthma, with no significant interaction with smoking.
The pattern of ICS use is of interest due to the reduced sensitivity to the effects of ICS therapy in smokers. [8][9][10][11][12][13] Interestingly, with electronic monitoring, we observed that current smokers were less likely to take their ICS than never smokers during the period of the study. The number of days of no ICS use was reduced with the SMART regimen, with no significant interaction between smoking status and treatment.
Overall the findings were similar to those in Māori, a disadvantaged high risk group in New Zealand with substantive morbidity from asthma, 27 in whom the relative benefits of the SMART regimen were comparable to non-Māori. 28 Together the findings suggest that the SMART regimen may not only be preferred in a general high risk population, as recruited in this study, but also specific very high risk groups within this population such as Māori and smokers. Smoking status was significantly associated with all outcome measures above, except daily budesonide dose.
Odds ratios, relative rates and ratio of means are reported with 95% confidence intervals. The weighted mean rate per year is the total number of events in the study group/total person follow-up time in years for the study group. Relative rates were calculated by Poisson regression with an offset of the logarithm of the period of observation (for the analyses of severe exacerbation, hospital admission or ED attendance and number of courses of systemic corticosteroid), unless otherwise stated. † Adjusted by treatment regimen (SMART vs Standard). ‡ Odds ratios. § Relative rates. ¶

Ratio of means. † †
Corticosteroid conversion: 100-mg intravenous hydrocortisone = 25-mg oral prednisone. Budesonide dose was converted to prednisone equivalent dose, based on a bioequivalence conversion calculated in a prior study (5000-μg inhaled budesonide = 10-mg oral prednisone). 25 The sum of the prednisone equivalent dose and systemic corticosteroid dose was annualized. The logarithm of the annualized steroid use was the response variable in a weighted normal linear model, with the randomized treatment as a predictor and the treatment exposure time as a weight (individuals with longer periods of treatment exposure were given more weight and those with shorter periods of treatment exposure less weight in the analysis). ED; Emergency Department; μg; micrograms.
This analysis addresses secondary hypotheses from a RCT and thus, despite our finding of no interaction between smoking and randomized treatment, will have been at risk of Type I error rate inflation. As for most interaction analyses in RCTs, the study was designed with statistical power to detect a difference in the whole group of participants. For some of the outcome variables, there were low numbers of events in the smoking categories, limiting the ability to detect a moderate or weak effect of smoking status. We are confident that participants with COPD did not enter the RCT because we excluded those with an active diagnosis of COPD, and current or ex smokers who had the onset of respiratory symptoms after the age of 40 and a >10 pack year-smoking history.
In conclusion, the favourable efficacy/safety profile of the SMART regimen, when compared with the standard maintenance ICS/LABA and SABA reliever therapy regimen in this high risk population, also applies to current and ex smokers. Because of the higher baseline risk of morbidity and at risk behaviour in current and ex smokers, the absolute reduction in risk with the SMART regimen is greater in these patients. We recommend the use of the SMART regimen in current and ex smokers with asthma.