Clinical Trial of Low-Dose Theophylline and Montelukast in Patients with Poorly Controlled Asthma

Asthma treatment guidelines recommend the addition of controller medication in patients with poorly controlled asthma (1, 2). Usually, inhaled corticosteroids (ICS) are the mainstay of asthma controller therapy, but some patients require additional treatment or prefer not to use ICS. In such patients, oral once-daily medication is an attractive alternative, with a choice of either theophylline or a leukotriene antagonist. Theophylline, once widely used for asthma symptom control, has fallen into disfavor in recent years because of concerns about side effects and the expense and inconvenience of monitoring blood levels (3). Recent studies, however, suggest that theophylline has significant antiinflammatory and immunomodulatory effects at lower serum concentrations (< 10 mg/L), and can be given as once-daily oral medication, which does not require monitoring of blood levels (4). Antileukotriene agents, such as the leukotriene receptor antagonist montelukast, are also recommended as add-on therapy for inadequately controlled asthma (1, 2). Montelukast is effective in improving lung function and asthma symptoms and also is given as a once-daily oral preparation not requiring titration and monitoring of blood concentrations, but it is more expensive than generic theophylline.

The use of low-dose theophylline as add-on therapy to ICS has shown benefit in some studies (5, 6) but not others (7, 8). Recent meta-analyses have questioned the efficacy of antileukotrienes as asthma therapy when added to ICS (9, 10). Moreover, the comparative effectiveness of low-dose theophylline and montelukast for patients with poorly controlled asthma is unknown, and no previous studies have directly compared the effectiveness of these two once-daily oral agents for control of asthma symptoms. Accordingly, we conducted a double-masked, randomized, placebo-controlled clinical trial to compare the effectiveness of these two drugs in controlling asthma. Because the goal of asthma therapy is to minimize episodic deterioration in terms of peak flow reduction, increased use of bronchodilators, and heath care use, the present study focused on asthma control as a primary outcome rather than intermediate endpoints such as lung function or inflammatory markers (11). The composite outcome measure of episodes of poor asthma control (EPACs) was used to reflect the several dimensions of good asthma control, including physiology, symptoms, and health care use. Moreover, because the interaction of low-dose theophylline with ICS is controversial, we enrolled patients with poorly controlled asthma who were and were not using ICS.

METHODS

Protocol

This study was a randomized, double-masked, placebo-controlled trial of the effectiveness of low-dose theophylline or montelukast for patients with poor asthma control. Participants were recruited from 19 centers in the American Lung Association Asthma Clinical Research Centers. Participants were age 15 yr or older, had physician-diagnosed asthma, had been prescribed daily asthma medications for 1 yr or more, had an FEV^sub 1^ of 50% or more of predicted values (12), and had poor asthma control defined by a score of 1.5 or greater on the Asthma Control Questionnaire (ACQ) (13, 14). Volunteers were ineligible if they used oral corticosteroids, leukotriene antagonists, or theophylline within 4 wk preceding enrollment, were current or former smokers with 20 pack-years or more smoking history, or had other significant illness. The protocol was approved by the institutional review boards at each institution.

Participants completed five study visits, including a run-in period of 7 to 14 d and a treatment period of 24 wk (Figure E1 of the online supplement). At Visit 1, participants were assessed for eligibility and gave consent. After 7 to 14 d, participants returned for review of diaries, spirometry, and completion of the Asthma Symptom Utility Index (ASUI) (15), ACQ (13), and Asthma Quality-of-Life Questionnaire (AQLQ) (16).

Participants were randomly assigned in equal allocation ratio to theophylline 300 mg/d (Theochron; Inwood Laboratories, New York, NY), montelukast 10 mg/d (Singulair; Merck, Whitehouse Station, NJ), or placebo using permuted blocks stratified by clinic. Treatments were masked by opaque capsules and taken after the evening meal.

Participants were telephoned 2 wk after randomization to assess compliance, side effects, and asthma control. Participants returned for visits at 4, 12, and 24 wk.

Medication adherence was assessed by diary, and plasma montelukast or theophylline concentrations at 1 and 6 mo. Theophylline concentration was measured by particle-enhanced turbidimetric inhibition immunoassay (analytic range, 2.0-40 mg/L) (17), and montelukast concentration by reversed-phase liquid chromatography (detection limit, 5 ng/ml) (18).

Outcomes

The primary outcome measure was the annualized rate of EPACs defined by any of the following events occurring within a 1-wk window recorded by diary: a decrease of peak expiratory flow more than 30% of personal best for 2 or more consecutive days; use of bronchodilator rescue medication over baseline by more than four metered-dose inhalations (or two nebulizer treatments) in 1 d; oral corticosteroid treatment for asthma; or an unscheduled asthma health care visit to a physician, emergency department, or hospital (19). Secondary outcomes included the ASUI, AQLQ, ACQ scores, and pre- and post-bronchodilator spirometry.

Analysis

Analyses were performed by intention-to-treat. Poisson regression models with Huber-White variance estimates were used to evaluate event rates of EPACs. Linear and logistic regression models with generalized estimating equation variance estimates were used to evaluate differences among treatment groups for continuous or dichotomous outcomes, respectively (20, 21). Analyses presented are not adjusted for baseline covariates. Results from models adjusted for baseline characteristics (age, sex, race, FEV^sub 1^) were similar to the unadjusted results. Data were analyzed using SAS version 8 (SAS Institute, Cary, NC) and Stata version 9 (StataCorp, College Station, TX) software. Assuming that 50% of the placebo group experienced an EPAC, the sample size of 489 had 80% power to detect a 15% difference (50 vs. 35%) in the proportion of patients with one or more episodes (Figure E2) (22).

RESULTS

Baseline Characteristics

A total of 489 participants were randomized, with 95% completing diary cards and 94% completing follow-up spirometry. The enrollment and follow-up flow diagram is available in the online supplement (Figure E3). Baseline characteristics for the participants are shown in Table 1 and Table E1. The participants were on average middle-aged, predominantly female, and with adult-onset asthma. The treatment groups were well matched for baseline asthma characteristics, with approximately three-fourths of all subjects using ICS. About 9% of participants were prescribed daily asthma medication, but did not use it. Six percent of participants had been prescribed a leukotriene antagonist and less than 1% of participants had been prescribed theophylline at some time in the past year, which they had discontinued at least 4 wk before enrollment. Post-bronchodilator lung function was mildly reduced on average compared with a normal population. By design, the participants had poorly controlled asthma, with an ACQ score of 2.3 to 2.4 (with a score greater than 1.5 indicating poor asthma control) (14). The group assigned to theophylline tended to have a slightly lower ASUI score, indicating slightly greater asthma symptoms than the other groups (0.66 for theophylline vs. 0.68 for montelukast and 0.70 for placebo, p = 0.16 for overall comparison).

Adherence to Therapy

Self-reported adherence to the study drugs was 84% for theophylline and 88% for both montelukast and placebo. We also measured adherence using biological measures, defined as a plasma drug concentration that exceeded the lower limit of detection (> 2 mg/L for theophylline and > 5 ng/ml for montelukast). Using biological measures, adherence was lower and fell over time. Adherence at 4 wk was 79% (n = 136) for theophylline and 71% (n = 136) for montelukast. Adherence fell to 60% (theophylline, n = 120; montelukast, n = 127) in both groups at 24 wk. In participants with detectable drug concentrations, the mean

Adverse Events

Symptoms that were considered potential adverse effects of theophylline were acquired by direct questioning, which lead to relatively high rates of reporting in all treatment groups (Table E2). Reports of nausea at 4 wk were more frequent in the theophylline group (33%) compared with placebo (22%) and montelukast (13%). Similarly, at 4 wk, nervousness was reported in 33% of the theophylline group compared with 21 and 22% of the montelukast and placebo groups, respectively. The proportion of patients reporting nervousness declined from baseline in the placebo and montelukast groups, but remained essentially unchanged in the theophylline group. At 12 and 24 wk, the prevalence of nausea and nervousness were similar among the three groups. Poor appetite was less common in the montelukast group than the other two treatment groups at 4 wk, but was similar at 12 and 24 wk. There were no differences between treatment groups at any time point for symptoms of tremor, heart palpitation, vomiting, or insomnia.

EPACs

Asthma control rates were analyzed using EPAC composite events as well as each component individually (Figure E4; Table 2). Overall unadjusted event rates were similar among the groups. The overall EPAC rates and 95% confidence intervals (CI) were as follows: 4.9 (episodes/person-year; CI, 3.8-6.4) for placebo, 4.0 (CI, 3.0-5.4) for montelukast, and 4.9 (CI, 3.6-6.1) for theophylline. The proportion of participants who experienced one or more episodes of poor asthma control was close to our projected figure of 50%: theophylline, 53%; montelukast, 49%; and placebo, 52%. Thus, it seems implausible that increasing the size of the study would have increased the power of the study enough to demonstrate a clinically meaningful significant treatment effect. The incidence of peak flow drops greater than 30% was significantly less with montelukast compared with placebo (p = 0.008). There were no other components that were different between theophylline and montelukast. Adjustment for baseline covariates (age, sex, race, obesity, and baseline FEV^sub 1^% predicted) did not alter this outcome. Subgroup analysis of adherent patients only, defined by detectable 24-wk drug concentrations, also did not show a significant improvement in EPACs for theophylline or montelukast compared with the aggregate placebo group.

Asthma Symptoms

The overall mean changes in asthma symptoms assessed by the ASUI, quality of life assessed by the AQLQ and control of asthma by the ACQ scores were not statistically different in either treatment group compared with placebo (Table 3). At 4 wk, the montelukast group was significantly better than placebo, but not at the later time points.

Lung Function

Overall, prebronchodilator FEV^sub 1^ was improved in both the theophylline (p = 0.006) and the montelukast (p = 0.003) groups (Table 4; Figure 1); however, the overall change in prebronchodilator FVC was not improved in either group. The post-bronchodilator FEV^sub 1^ was improved only by theophylline at 4 (p = 0.006) and 12 wk (p = 0.04), but not at 24 wk (p = 0.15). Overall, the improvement in post-bronchodilator FEV^sub 1^ by theophylline, including all measurements, was significant (p = 0.005 vs. placebo), and the trend for montelukast was similar but smaller (p = 0.13 vs. placebo). The post-bronchodilator FVC was not significantly improved by either of the active treatments.

Influence of ICS Use

It has been hypothesized that there is a beneficial antiinflammatory interaction between corticosteroids and theophylline based on additive molecular effects on histone deacetylase (4, 23). However, a previous clinical study in 24 patients with asthma receiving ICS showed no antiinflammatory effect of low-dose theophylline (8). Therefore, we analyzed the data stratified by use of ICS. There were significant statistical interactions between ICS use and outcome measures, so we conducted separate analyses for those taking and not taking ICS. The p values for an interaction of ICS use and outcomes were as follows: 0.008 for EPAC rate, 0.08 for the ACQ, 0.12 for the AQLQ, and 0.04 for the ASUI. Contrary to expectation, we found that there was a significant (p = 0.002) beneficial effect of theophylline on EPAC rate in participants not using ICS, with a substantial reduction of events to 1.8 events/person-year (CI, 1.1-3.0) compared with the event rate on placebo of 5.7 (CI, 3.3-9.9; Figure E4). Theophylline reduced both the fall in peak expiratory flow (p = 0.02) and ß-agonist inhaler/nebulizer use (p = 0.01) components of the event rate, but did not alter the use of health care, an infrequent event. Participants not taking ICS who were assigned to montelukast had a borderline reduction in asthma event rate (p = 0.08), with the rate of peak flow declines significantly lower than placebo (p = 0.02). Treatment with low-dose theophylline in participants not using ICS also showed significant improvement in symptoms as assessed by the ASUI (p = 0.03) and the ACQ (p = 0.02). There were no significant differences in plasma drug levels or adherence rates for either add-on treatment between users and nonusers of ICS, suggesting treatment adherence did not account for the observed differences (Tables E3-E5).

DISCUSSION

The purpose of this study was to compare the effectiveness of adding either low-dose theophylline or montelukast to the treatment of patients with poorly controlled asthma. The primary outcome was selected to be the rate of EPACs, as this is a measure of asthma control that is relevant to quality of life and costs of medical care, and is the goal of asthma care under current practice guidelines. Secondary outcomes were lung function, asthma symptom scores, and asthma-related quality of life. We studied both low-dose theophylline and montelukast because both are taken as once-daily oral formulations, which are convenient for patients and therefore should optimize adherence. We purposely designed the study to reproduce the clinically realistic situation where therapies are added to existing treatment regimens as prescribed by their usual asthma care provider. Our aim was to evaluate the effectiveness of these treatments in a clinically realistic situation with clinically relevant outcomes rather than to evaluate treatment efficacy in an ideal setting using intermediate physiologic and biological outcomes.

The main finding of this trial was that neither theophylline nor montelukast had additional benefit in reducing the EPACs, reducing asthma symptoms, or improving quality of life compared with placebo. Both theophylline and montelukast improved prebronchodilator spirometry, but only theophylline improved FEV^sub 1^ after bronchodilator. Thus, theophylline appeared to augment the bronchodilator effect of the albuterol inhalation, thus overcoming residual bronchoconstriction in that group of patients. However, the magnitudes of the spirometric changes were small (0.08-0.09 L) and of uncertain clinical importance. We found that low-dose theophylline was relatively well tolerated, but did cause gastrointestinal symptoms during the first 4 wk that abated with continuing treatment. Secondary subgroup analysis showed that theophylline substantially reduced both event rates and symptoms in those patients with asthma who were not using ICS.

In the past, theophylline was widely used as maintenance bronchodilator therapy for asthma. However, the introduction of inhaled long-acting ß-agonists (LABAs) and ICS has largely supplanted theophylline. Like montelukast, low-dose oral theophylline remains an attractive alternative for asthma control in patients who will not or cannot take ICS insofar as it can be given as a once-daily oral formulation, does not require blood level monitoring, is inexpensive, and is reasonably well tolerated. Adherence was reasonably good in all treatment groups when measured by diary self-report at 4 wk and 24 wk. However, by 24 wk, detectable blood concentrations were absent in 40% of patients in both the montelukast and theophylline groups, a fall-off in adherence that is comparable to biological or electronic monitoring in other clinical trials and in clinical practice (24). Although this limited adherence certainly would have reduced any drug treatment effect on asthma symptom control, the intention-to-treat results still reflect the utility of the treatments in clinical practice and did not prevent us from observing significant bronchodilating effects of the active treatments. Moreover, secondary analysis of efficacy, excluding those patients without 24-wk detectable blood concentrations did not show a beneficial impact of either theophylline or montelukast.

Development of tolerance to the adverse effects of theophylline has long been observed with high-dose theophylline, as we observed in this study. Usually, this has been attributed to induction of metabolic pathways or central nervous system adaptation. Because of the discrepancy between self-reported adherence and serum concentrations, it is also possible that the adaptation may be due, in part, to nonadherence in patients who experienced more side effects.

A growing body of evidence, mostly in the setting of asthma and allergic inflammation, suggests that theophylline has antiinflammatory properties. These include the following: inhibition of neutrophil migration; inhibition of neutrophil, lymphocyte, and monocyte activation; production of the antiinflammatory cytokine interleukin-10; and inhibition of inflammatory mediators and the proinflammatory gene regulator nuclear factor-?B (4, 25-30). Low-dose theophylline has been shown to reduce airway eosinophilia in patients with asthma, even in the absence of bronchodilation response or reduction in expired nitric oxide concentrations (31). Ito and colleagues have postulated that the likely mechanism for these antiinflammatory effects is the activation of histone deacetylase (23, 32). Histone deacetylase is a component of the pathway by which corticosteroids are believed to inhibit proinflammatory gene expression, and its activity is reduced in some patients with asthma, especially cigarette smokers (33). The restoration of this enzyme by theophylline permits both endogenous and corticosteroid-mediated down-regulation of inflammatory genes and occurs at concentrations (4.3 ± 0.8 mg/L) below those typically used in the past for bronchodilator activity (10-20 mg/L) (32, 33). Accordingly, we hypothesized that patients using inhaled steroids may benefit by the addition of low-dose theophylline more than those patients not using inhaled steroids. Surprisingly, we did not find this to be the case. When we stratified our patient population by use of ICS, participants assigned to theophylline who were not using ICS had both statistically and clinically significant improvement in their asthma control and symptoms. The reason for this is not clear, but presumably ICS are such effective antiinflammatory agents that the addition of other drugs is of minimal additional benefit.

ICS are generally considered to be the mainstay of antiinflammatory controller treatment in asthma. Some patients or asthma care providers may find ICS to be too expensive, difficult, or inconvenient to use, or have concerns about long-term adverse effects. Thus, the current study suggests that monotherapy low-dose theophylline holds promise as an alternative to ICS in the treatment of selected patients with asthma with inadequate symptom control. Because the subgroup on ICS was small and was not the primary focus of the study, more research investigating this group is necessary to confirm this result. Theophylline, although a mild bronchodilator, is only of marginal benefit, if any, in terms of asthma symptom control for patients with asthma already treated with ICS.

We compared add-on therapy with low-dose theophylline with montelukast, a widely prescribed, oral, once-daily asthma control therapy. Like theophylline, montelukast has been shown in some studies to improve lung function or symptoms when used together with ICS (8, 34, 35). As monotherapy, however, leukotriene antagonists are generally not as effective as ICS (36, 37). The current study demonstrated that, aside from improvements in lung function, montelukast, like theophylline, had no additional beneficial effect on asthma control as measured by a composite measure of lung function variability, ß-agonist use, and health care use. In the subgroup of participants not taking ICS at baseline, the use of montelukast did not improve asthma control, although the rate reduction approached statistical significance, an outcome that is consistent with previous reports (37).

Most asthma guidelines recommend a bronchodilator such as an LABA as the first choice for add-on therapy when ICS do not provide adequate asthma control. Studies comparing montelukast to LABA therapy as add-on treatment to ICS have shown that LABAs produce greater bronchodilation, and more reduction in symptoms and exacerbations (38, 39). No such studies have compared LABAs with low-dose theophylline. However, recent studies have raised questions about the safety of LABAs (40, 41). Because of this, some have suggested that low-dose theophylline may be a useful alternative to LABAs when ICS alone do not adequately control asthma (42). Although we did not compare them directly, this study does not support the use of add-on therapy with montelukast or low-dose theophylline as an efficacious alternative to LABA add-on therapy for asthma control.

To our knowledge, only one previous trial has directly compared the efficacy of theophylline with leukotriene receptor antagonists. Dempsey and colleagues (8) conducted a 2-wk crossover trial in 24 subjects comparing low-dose theophylline (400-600 mg/d) with zafirlukast, a leukotriene receptor antagonist, in patients who were taking either low-dose (100 µg/d beclomethasone) or medium-dose (400 µg/d beclomethasone) ICS. They found that the added antiinflammatory effects of zafirlukast and theophylline, measured with either exhaled nitric oxide or methacholine reactivity were only present with low-dose but not with medium-dose ICS. This study, therefore, supports our finding that the clinical benefit of these agents is diminished in the background of ICS use. The authors did not, however, include participants who were not using ICS, and did not specifically focus on measures of asthma control.

In summary, add-on therapy with either low-dose theophylline or montelukast in patients with poorly controlled asthma did not reduce EPACs, although both agents improved lung function. However, in the subgroup of patients with poorly controlled asthma who were not taking ICS, both low-dose theophylline and montelukast had similarly beneficial effects on asthma control, but only the theophylline-treated group reached statistical significance. Although low-dose theophylline and montelukast had similar adherence, efficacy, and tolerability, theophylline is considerably less expensive. Therefore, for patients with poorly controlled asthma who cannot or will not use ICS because of side effects, preference, lack of efficacy, or cost, low-dose theophylline is an inexpensive alternative asthma controller therapy and should be more widely applied.

Conflict of Interest Statement: C.G.I. has received an investigator-initiated grant ($150,000) from GlaxoSmithKline (GSK), has received honoraria from several organizations (Sepracor, Biogen, Isis), and has been a Merck-sponsored visiting professor in 2004 ($2,000) and 2005 ($2,000). D.A.K. has no financial relationship with a commercial entity that has an interest in the subject of this manuscript. N.R.A. has served on advisory boards for GSK and ALTANA, receiving approximately $2,000/yr for 5 yr, and has given two to three talks over the past 5 yr for GSK, with an honorarium of $2,500 for each. M.C. has no financial relationship with a commercial entity that has an interest in the subject of this manuscript. N.A.H. has served on the advisory board of GSK and DEY Pharmaceutical and has received honoraria for speaking at scientific meetings for GSK ($8,000 over the last 2 yr); he also has received grant support from GSK, Boehringer-Ingelheim, Sepracor, AstraZeneca, and Novartis. J.T.H. has no financial relationship with a commercial entity that has an interest in the subject of this manuscript. J.J.L. has no financial relationship with a commercial entity that has an interest in the subject of this manuscript. R.A.W. received consulting fees from GSK, Pfizer, sanofiaventis, Emphasys, Spiration, and Forest in the past 3 yr for research oversight and review committees; he has served on advisory boards for Boehringer-Ingelheim, Pfizer, GSK, Hill-Rom, Otsuka, Ortho, and Amgen; he has also received research grants from Boehringer-Ingelheim, Otsuka, and Pfizer; conflict of interest regarding human research is managed by the Johns Hopkins University.

American Lung Association Asthma Clinical Research Centers Writing Committee: Members are as follows: Charles G. Irvin, Ph.D., and David A. Kaminsky, M.D., University of Vermont School of Medicine; Nicholas R. Anthonisen, M.D., University of Manitoba School of Medicine; Mario Castro, M.D., M.P.H., Washington University School of Medicine; Nicola A. Hanania, M.D., Baylor College of Medicine; Janet T. Holbrook, Ph.D., M.P.M., The Johns Hopkins University School of Public Health; John J. Lima, Pharm.D., Nemours Children's Clinic; and Robert A. Wise, M.D., The Johns Hopkins University School of Medicine.