INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS REFERENCES

Inhaled corticosteroids are commonly used in the treatment of patients with moderately severe asthma because of their efficacy and perceived safety at low doses (1, 2). Many patients are reluctant to use inhaled corticosteroids, however, because of concerns over potential adverse effects such as alterations in bone metabolism when taken over the long term. Thus, there has been interest in alternative anti-inflammatory agents for asthma therapy. Indeed, agents used in other inflammatory conditions such as gold salts, methotrexate, and cyclosporin have been studied. Although salutary effects have been reported in asthma, none of these compounds has shown a risk-benefit profile that would justify widespread use as an alternative to inhaled corticosteroids (3).

Colchicine is an anti-inflammatory drug with proven efficacy in the treatment of familial Mediterranean fever, gout, Behçet's disease, and primary biliary cirrhosis (4). Interest in colchicine as a potential form of asthma therapy is based on its pharmacodynamic properties, lack of significant adverse effects with long-term use, and low cost. Relevant anti-inflammatory activities include interference with human leukocyte chemotaxis and lysosomal enzyme release (8); depression of 5-lipoxygenase product generation from activated human leukocytes (11); inhibition of histamine secretion from antigen-challenged human basophils and sensitized lung fragments (14); and potentiation of cyclic AMP responses after beta-adrenergic stimulation of leukocytes and macrophages (16, 17).

In a small clinical study of patients with mild allergic asthma, Schwarz and colleagues (18) reported improved symptoms with a reduction in beta-agonist use after colchicine therapy. Moreover, Kelly and colleagues (19) found that colchicine inhibited the late airway response after allergen inhalation in allergic asthmatics. In addition, it tended to block the early response as well as the increase in methacholine responsiveness that occurs after allergen inhalation.

Other investigators found that the prevalence of asthma in patients with familial Mediterranean fever was significantly lower than in the general population (1.1 versus 6.4%) and suggested that the difference was due to the use of colchicine prophylaxis in these patients (20).

On the basis of these observations, the Asthma Clinical Research Network (ACRN) of the National Heart, Lung, and Blood Institute conducted a study to test the hypothesis that in patients with moderate asthma who use inhaled corticosteroids for control of symptoms and lung function, colchicine offers therapeutic benefit as measured by maintenance of control when inhaled corticosteroids are discontinued. This report summarizes this 10-wk, double-blind, randomized, placebo-controlled trial, and it describes the ACRN experience with time to treatment failure as the primary outcome measure in an asthma clinical trial.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS REFERENCES

Subject Recruitment and Selection

Male and nonpregnant female subjects 18 to 60 yr of age with moderate asthma and baseline FEV₁ values 55 to 90% predicted were recruited from existing study populations and by advertising at each of five clinical centers. Moderate asthma was defined by symptoms poorly controlled with episodic administration of an inhaled beta₂- agonist. Eligible patients must have used inhaled corticosteroids at daily doses of 336 to 1,600 µg of triamcinolone acetonide, beclomethasone dipropionate, or flunisolide for control of symptoms for at least 30 d prior to entry. Eligible subjects were required to have documentation of  12% increase in FEV₁ as a short-term or longer-term response to any asthma therapy or a positive response to inhaled methacholine defined as a provocative concentration causing a 20% fall in FEV₁ (PC₂₀FEV₁)  8 mg/ml within 6 mo preceding study entry. The protocol was approved by an independent Protocol Review Committee and the institutional review boards of the participating centers; during the trial, patient safety was assessed by a Data Safety and Monitoring Board (see Appendix ). Written informed consent was obtained from each subject prior to participation.

Protocol Design

Qualifying subjects enrolled in a 4-wk run-in period. At the beginning of this run-in, all patients discontinued their usual asthma medications and were placed on open label triamcinolone acetonide metered-dose inhaler, 800 µg  /d (i.e., four puffs twice a day), and open label albuterol MDI to take as needed for symptom control. After 2 wk of stabilization on the uniform dose of triamcinolone, all subjects were treated with open label colchicine (0.6 mg twice a day) for the final 2 wk of the run-in. This 2-wk interval in which all subjects received colchicine and triamcinolone was designed to identify and exclude from further participation those subjects who were intolerant of colchicine. Only two subjects were excluded because of intolerable stomach cramps and diarrhea. Subjects whose prebronchodilator FEV₁ remained  55% of predicted and who demonstrated compliance with study procedures were eligible for randomization at Week 4. Lack of compliance with study procedures was defined as a failure to take scheduled doses of colchicine at least 80% of the time during the last 2 wk of the run-in period as determined by pill counts, or by a failure to comply with triamcinoline twice daily on 4 d or more during the last 2 wk of the run-in as determined by subject-reported diary entries.

At the end of the run-in, subjects were randomized to either continue receiving colchicine or to receive an identically appearing placebo over a 6-wk double-blind treatment phase. At the same time, all subjects discontinued their use of inhaled triamcinolone. During this treatment phase, asthma was monitored on a daily basis by recordings of peak expiratory flow (PEF), symptoms, and rescue albuterol use. In addition, subjects were assessed clinically, and spirometry was performed at 2-wk intervals; telephone contacts were made in the intervening weeks. Methacholine responsiveness was measured at Weeks 2 and 4 of the run-in period, and a disease-specific asthma quality of life questionnaire (21) was administered at Weeks 2 and 4 of the run-in period and also at study termination.

Because we anticipated that subjects randomized to placebo would experience deterioration in asthma control during the double-blind treatment phase, we defined criteria for treatment failure (Table 1) and chose time to treatment failure in both treatment groups as the primary outcome measure. Other outcome measures included morning PEF, PEF variability, defined as [(PM PEF minus AM PEF)  PM PEF], symptom scores, quality of life scores, and rescue albuterol use. Asthma symptoms were rated on a 0 to 3 point scale (0 representing no symptoms; 3 representing severe symptoms). Patients refrained from using rescue albuterol for at least 6 h before clinic visits. To measure PC₂₀FEV₁, methacholine aerosols were generated with a Model 646 nebulizer (DeVilbiss Healthcare, Somerset, PA) and calibrated dosimeter (S&M Instruments, Doylestown, PA). The methacholine PC₂₀-FEV₁ was determined by standard procedures (22). Spirometric testing was performed using a Collins Eagle II spirometer (Warren E. Collins, Quincy, MA). All outside-of-center PEF measurements were obtained using the AirWatch^® Monitoring System (Enact Health Management Systems, Inc., Mountain View, CA), which furnishes a digital display of recorded measurements. This device is capable of storing in memory PEF values as well as the date and time of their measurement for later downloading by computer modem for analysis. In this study, however, PEF values recorded by subjects in their diaries at the time of measurement were used for analysis. Values stored by the AirWatch^® were used to verify compliance and will be analyzed separately as part of a companion study to test the reliability of this device.

For safety, subjects were given a card indicating the 65% threshold for PEF as well as a treatment algorithm employing rescue beta-agonists should their PEF fall below the threshold value. Cards also indicated the maximum number of albuterol puffs that could be taken in a 24-h period without exceeding threshold values defined by treatment failure criteria. In addition, subjects were given an 8-d supply of oral prednisone with instructions for rescue treatment as well as a phone number to contact a study physician at any time.

Patients whose albuterol use or PEF values indicated that they met treatment failure criteria between clinic visits and those who declared a lack of satisfaction with the treatment regimen were advised to contact their study site immediately and visit the clinic within 24-h for evaluation and confirmation of treatment failure status. Subjects meeting treatment failure criteria were removed from double-blind treatment and were placed on a regimen to restore asthma control. Those removed from the trial because of treatment failure were continued on their regular medications and monitored by spirometry, PEF measurements, and daily diaries until they would have completed the study. Shortly after the study was initiated, we recognized the importance of attempting to document whether the treatment failure criteria chosen for this study reflected a clinically meaningful level of deterioration. Accordingly, a questionnaire administered to both subjects and study coordinators was filled out as closely as possible to the time the subject was removed from the study because of treatment failure. The questionnaire asked both the subject and study coordinator to indicate whether the subject was removed from the study too early, too late, or at the right time.

Standardization of Procedures and Quality Assurance

Procedures for measuring lung function, methacholine responsiveness, and quality of life were standardized across participating centers. Study personnel were tested to ensure proficiency and uniformity in all network-related skills and had to pass certification examinations before the data they gathered could be used for analysis. Standardization of equipment and procedures and all quality assurance techniques were identical to those used in a previous trial comparing regular versus as-needed beta agonists (23). AirWatch^® devices were tested against a Jones flow-volume syringe and compared with spirometer readings at each clinic visit. If the device failed to meet established performance standards, it was replaced. Study drug compliance was monitored by counting unused colchicine medication returned in blister packs. In addition, compliance with visit schedules and diary card entries was also monitored.

Statistical Considerations and Analysis

The primary outcome variable was the time to treatment failure, defined as the time from randomization to either treatment failure (defined by criteria in Table 1) or end of the study (last scheduled visit for those who did not fail treatment). A sample size of 45 patients was calculated to provide 80% power to detect a difference in the proportions failing of 60% in the placebo arm versus 20% in the colchicine arm, allowing for a 10% dropout rate. The secondary outcome variable was change in FEV₁ from the time of randomization to the study end point, defined as the time of treatment failure or the last scheduled visit for those who did not fail treatment. A sample size of 70 provided 80% power to detect a difference in the change in FEV₁ of 0.3 L and allowed for a 10% dropout rate. Therefore, a randomization goal of 70 patients was established.

The Kaplan-Meier method was used to estimate survival curves for the placebo and colchicine groups, and the log-rank test was used to test for a significant difference between the curves. Group differences in the secondary response variable (change in FEV₁) were assessed using the t test. The other secondary responses, morning PEF, PEF variability, asthma symptoms, quality of life measures, and use of rescue albuterol, were compared using the t test or Wilcoxon's test as appropriate. For variables recorded in diaries, we used the average of the values recorded over the final 3 d before study termination or time of treatment failure compared with the average of the final 3 d prior to the randomization visit.

In a secondary analysis, we compared various baseline measures between subjects who failed treatment and those who did not fail. The t test was used for continuous measures such as FEV₁, morning PEF, quality of life, and airway responsiveness. The chi-square test was used for categorical data such as sex and ethnicity.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS REFERENCES

Patient Enrollment and Demographics

Of the subjects recruited, 71 subjects were randomized, 36 to placebo and 35 to colchicine There were no significant differences between treatment groups in terms of baseline characteristics, including age, sex, ethnicity, and FEV₁ (Table 2) Moreover, the treatment groups did not differ with regard to reported dosages of inhaled corticosteroids used prior to entry. Two subjects, one from each treatment group, withdrew from the study after randomization for reasons unrelated to the study question or asthma.

Compliance

Compliance with the use of colchicine was greater than 94% as measured by counts on returned medications during study visits. In addition, of 42,197 possible protocol-specified diary entries, only 1,893 (4.5%) were missing; of 424 scheduled study visits, only 1 was missed, for a compliance rate of 99.8%.

Efficacy

Sixty percent of patients receiving colchicine and 56% of those receiving placebo failed to complete 6 study weeks of treatment because of treatment failure. The survival curves did not differ between treatment groups (log-rank p value = 0.38) (Figure 1). In addition, changes in FEV₁, morning PEF, PEF variability, symptom scores, rescue albuterol use, and quality of life scores between baseline and the study end point were not significantly different between treatment groups (Table 3).

View larger version (17K): [in this window] [in a new window]   Figure 1.   Estimated probability of remaining in the study after randomization to colchicine (solid line) or placebo (broken line). Inhaled corticosteroids were discontinued at Day 0. No significant difference between treatment groups was found.

During the final 2 wk of the run-in (between Weeks 2 and 4), when all subjects were treated with open-label colchicine and inhaled triamcinolone, we observed statistically significant improvement in overall quality of life and in each of the four component domains tested (Table 4). This change in quality of life, however, was not accompanied by a significant change in any of the other measures of asthma control.

Treatment Failures

Of 41 subjects failing to complete the 6-wk double-blind treatment period because of treatment failure, 31 met treatment failure criteria by experiencing a fall of more than 20% in FEV₁ from baseline. Fewer subjects met criteria by having an FEV₁ < 40% predicted (n = 4); PEFR < 65% of baseline (n = 7); increased use of albuterol (n = 4); and lack of satisfaction with the treatment (n = 8). Twelve subjects failed treatment by more than one criterion, and one of these subjects met three treatment failure criteria. Of eight subjects reported as treatment failures because of lack of satisfaction with the treatment regimen, two failed treatment according to objective criteria as well.

Results of questionnaires examining the timing of treatment failure are shown in Table 5. Subjects and study coordinators agreed in their subjective assessment of the timing of treatment failure in 29 of 41 cases. We estimated the level of agreement using a weighted Kappa statistic and found evidence to suggest that the agreement was significantly greater than expected by chance (weighted Kappa statistic = 0.53; 95% confidence interval, 0.30 to 0.76). In 10 cases, either the subject or the coordinator indicated that withdrawal from the study occurred too soon. In these cases, treatment failure occurred in seven subjects on the basis of  20% fall in FEV₁, whereas in three cases subjects failed because of lack of satisfaction with the treatment. In 12 cases either the subject or the coordinator indicated that withdrawal from the study occurred too late. In nine of these cases failure was determined by  20% fall in FEV₁, and in one case by greater than the threshold use of rescue beta-agonists. Two subjects who thought they were removed from the study too late did not meet objective failure criteria and were removed because of lack of satisfaction with treatment. In both of these cases, subjects noted an increase in symptoms and were using twice as much rescue albuterol medication than they used at baseline.

An analysis of baseline characteristics of subjects who failed treatment versus those who did not fail revealed a significantly lower PC₂₀ (0.81 ± 1.38 versus 2.11 ± 2.74; p = 0.01) in those who failed (Table 6). In addition, those who failed treatment tended to be older (p = 0.08). On the other hand, factors such as sex, minority status, duration of asthma, and FEV₁ were not associated with treatment outcome.

Safety

The most common adverse events were gastrointestinal symptoms; these included diarrhea/loose stools, stomach cramps or upset, heartburn, and flatulence. These, events were reported by nine of 35 subjects receiving colchicine and six of 36 subjects receiving placebo (p = 0.40).

No serious adverse events were observed because of withdrawal of inhaled corticosteroids. In particular, there were no deaths, hospitalizations or emergency room visits. The lowest FEV₁ measured after glucocorticoid withdrawal was 1.0 L (28.1% predicted). The lowest peak flow recorded was 101 L/ min, and the highest amount of albuterol use over a 24-h period was 20 puffs. One patient receiving active study drug experienced severe abdominal pains and was hospitalized. He had experienced similar pains in the past, prior to study participation. The patient underwent surgery when adhesions were determined to be the cause of his symptoms. The patient was followed to the end of the study, but taken off study medications.

	DISCUSSIONS

Colchicine has a broad array of relevant anti-inflammatory properties that seemed to make it a suitable candidate for investigation of its possible effects as an asthma treatment. In this study, however, we found that colchicine was not significantly better than placebo in preventing protocol-defined treatment failure in asthmatic subjects when inhaled corticosteroids were withdrawn. Additionally, there were no significant treatment group differences with respect to changes in spirometry, peak flow, peak flow variability, symptom scores, rescue albuterol use, or quality of life measures between baseline and the end of the study. In this protocol, it was important in the secondary analyses to define the end of the study as the time of the last scheduled clinic visit or the time of treatment failure. An intent-to-treat analysis (i.e., use of end of study as the final study visit for all subjects) would be less informative since subjects failing treatment prior to the last scheduled visit would have received steroid rescue therapy and would have been placed on their usual treatment regimen, thus obscuring any differences between placebo and colchicine.

Although we observed a significant improvement in quality of life during the final 2 wk of the run-in when all subjects were treated with colchicine and inhaled triamcinolone, we are reluctant to interpret this observation as either clinically important or colchicine-related for several reasons. First, there was no placebo control during this interval; thus, it is inappropriate to ascribe observed changes to a colchicine effect when they could simply be due to better compliance with inhaled corticosteroid therapy or participation in the trial itself. Further, not only were changes in quality of life unaccompanied by improvement in other variables, they were also quite small. The degree of change for overall quality of life and for each of the four domains tested was less than the 0.5 level of change noted by Juniper and coworkers (21) to represent a minimally important difference in outcome.

The primary measure of efficacy in this trial was the time to treatment failure after withdrawal of inhaled corticosteroids. Similar "survival" type clinical trials have been performed to examine the effects of inhaled corticosteroid withdrawal in patients requiring these agents for symptom control (24, 25). Such a trial design has several advantages, including possibly smaller sample size requirements and the freedom to remove subjects from the study should they experience clinical deterioration, thus enhancing patient safety. Moreover, because deterioration in asthma symptom control is the specified endpoint, data are not "lost" if patients are prematurely terminated from the study because of asthma exacerbations.

The success of a trial design using treatment failure as the major outcome depends on how treatment failure is defined and whether failures observed by predefined criteria actually reflect clinically meaningful deterioration. Although several studies have used "treatment failure" as an outcome variable, there is little information concerning which measurement variables best predict or define treatment failure. Accordingly, we felt that data obtained in this trial could provide insight about treatment failure as an outcome measure for asthma, and, more specifically, address the following questions: (1) How do subjects fail? (2) Can treatment failure be predicted? (3) Are treatment failure criteria appropriate? (4) Is treatment failure a safe outcome?

The majority of treatment failures were identified by a >  20% fall in FEV₁. Only four failures were identified by diary information alone (i.e., PEF and beta-agonist use). If a  20% fall in FEV₁ is accepted as a reasonable criterion of treatment failure, this would suggest that criteria based on PEF or beta-agonist use were too rigid and required subjects to be "sicker" in order to fail. Chervinsky and colleagues (25) used a less severe PEF threshold (i.e., 20% decline) to define treatment failure, but they did not report how sensitive this measure was in comparison with other criteria such as spirometry, albuterol use, or symptoms. Although PEF changes are reportedly more sensitive than changes in PEF variability in predicting asthma deterioration when inhaled corticosteroids are withdrawn (26), others have shown that symptoms usually precede PEF changes both during spontaneous exacerbations and after inhaled corticosteroid withdrawal (27, 28). Whereas patient-recorded daily measurements may allow for more timely recognition of treatment failure between clinic visits, further studies are needed to evaluate the degree of changes in PEF, beta-agonist use, symptoms, or a combination of these variables that best reflects a safe but meaningful deterioration in asthma control.

On the basis of a previous study of steroid withdrawal in subjects with similar baseline characteristics (25), we projected the proportion failing to be approximately 60%; this was quite close to the observed proportion failing (56%) in the placebo group. Conversely, just under half of the subjects who were receiving inhaled corticosteroids prior to enrollment were able to discontinue such therapy for 6 wk without a significant deterioration in their asthma. Subjects who were able to complete 6-wk of therapy in this study after corticosteroid withdrawal had a significantly higher methacholine PC₂₀ at baseline than did those who failed treatment. Further prospective studies would be useful to determine the value of methacholine PC₂₀ as a predictor of the need for inhaled corticosteroids. Another potentially important implication of this analysis is the ability to enhance the statistical power of protocols using survival analysis by identifying, on the basis of methacholine PC₂₀, those patients who have a greater probability of failing after corticosteroid withdrawal.

Because the decision to administer questionnaires assessing the timing of treatment failure was made after the initiation of the protocol, a number of questionnaires were completed after the reported time of treatment failure. The median time between treatment failure and administration of the questionnaire was 13 d (interquartile range, 28 d). We recognize the limitations imposed by a retrospective analysis of treatment failure timing in some subjects. However, since published experience with treatment failure as an outcome in asthma clinical trials is lacking, we felt that these questionnaires could provide valuable preliminary information for designing future studies. We found only 10 cases in which either the subject, the coordinator, or both, felt that removal from the study was too early. Thus, in 31 of 41 cases (76%), removal from the double-blind treatment phase was felt to occur at either the right time or too late. In 12 cases, either the subject, the coordinator, or both felt that removal from the study occurred too late. Nevertheless, there were no mortalities, emergency room visits, or hospitalizations related to withdrawal of inhaled corticosteroids. The lowest FEV₁ and PEF recorded were 1.00 L (28% predicted) and 101 L/min, respectively. In addition, the highest beta-agonist use reported was 20 puffs/24 h. Thus, we believe that the criteria used in this study were a reasonable reflection of a level of deterioration that is clinically meaningful and that, with careful monitoring and rescue algorithms, also appears to be safe.

In summary, on the basis of an analysis of the time to treatment failure after discontinuation of inhaled corticosteroids, we found that there was no evidence to suggest that colchicine was better than placebo as an alternative treatment to inhaled glucocorticoids in patients with moderate asthma. This study did not address the potential effects of colchicine as added therapy or its effects in patients with more severe asthma. We also found that the use of the time to treatment failure as the primary outcome measure in an asthma clinical trial where treatment is withdrawn is both feasible and safe. Despite their feasibility and apparent safety, protocols employing treatment withdrawal and treatment failure analysis have predictable risks and should be undertaken only under carefully monitored conditions with rigorous safeguards to identify and manage asthma exacerbations.