INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Inhaled -adrenergic drugs play an extremely important role in the management of asthma (1). In the United States as well as in much of the rest of the world, inhaled albuterol is the most commonly prescribed drug in this class (2). Metered-dose inhaler (MDI) formulations of albuterol marketed in the United States all employ chlorofluorocarbon (CFC) compounds as propellants (Ventolin, Proventil, and at the time of this writing, four approved generic equivalents to Ventolin or Proventil sold under 17 brand names). With each actuation of the inhaler, expansion of the CFC propellant provides the energy needed to aerosolize the drug (3).

Release of CFCs into the atmosphere has been shown to cause degradation of stratospheric ozone concentrations which, in turn, allows increased penetration of ultraviolet radiation to the surface of the earth. Adverse effects on the environment and human health are associated with this increase (4). Consequently, by international agreement (5), CFC production and use are being phased out. While use of CFCs in MDIs is currently permitted under a temporary "essential use exemption," it is clear that these CFC-containing products will eventually be phased out as well.

This has led to diligent efforts by the pharmaceutical industry to develop alternative aerosol delivery systems that do not use CFCs (3, 6, 7). Proventil HFA, developed by 3M Health Care (St. Paul, MN) and marketed in the United States by Schering-Plough Pharmaceuticals (Madison, NJ), is the first non-CFC metered-dose inhaler approved in the United States (8, 9). This system uses a more environmentally friendly propellant, hydrofluoroalkane (HFA 134a), to deliver albuterol. Another strategy has been to develop sophisticated, multidose, dry powder inhaler (DPI) systems that derive the energy necessary to aerosolize drugs for inhalation from sources other than expansion of propellant (typically from the patient's inspiratory effort) (10, 11).

With any non-CFC inhaled albuterol formulation intended to serve as a replacement for CFC-containing inhalers, it is critically important that clinicians know whether the new inhaler delivers essentially the same, more, or less drug to the relevant -adrenergic receptors in the airways as do marketed CFC formulations. If a replacement inhaler delivers significantly less drug to the airways, patients may experience insufficient relief of symptoms when they use the inhaler. If the new formulation delivers significantly more drug, adverse events associated with excessive doses of inhaled -agonist use may be enhanced (12).

It has been usual practice to study ₂ agonist-induced bronchodilation to obtain this information. However, in the typical study of subjects with resting asthmatic airway obstruction, it is frequently not possible to distinguish between the effects of different doses of the reference product (i.e., a significant dose-response relationship is not present). This usually appears to be because one actuation of the reference product is near the top of the dose-response curve. Measurement of drug-induced changes in nonspecific airway responsiveness appears to have additional advantages over studies of bronchodilation (13). Measurement of airway responsiveness with bronchial provocation with histamine or methacholine is highly reproducible (16) and can reflect broad changes in inhaled ₂-agonist activity in the lung under conditions in which bronchodilation has "plateaued" near maximal achievable levels (13, 14). Information obtained from bronchoprovocation studies appears to be at least as clinically relevant as information obtained from studies of bronchodilation, because airway responsiveness to histamine and methacholine itself correlates with exercise-induced bronchospasm (17), cold air-induced bronchospasm (18), overnight drop in peak flows (19), medication requirements (19, 20), and severity of asthmatic symptoms. Loss of -agonist protective effects against methacholine-induced bronchospasm (21), which occurs with regular salmeterol administration, is paralleled by loss of protection against exercise- and allergen-induced bronchospasm (22, 23).

The Spiros inhaler is an effort-assisted, battery-powered, DPI delivery system that is breath actuated to minimize the need for patient coordination. It contains a removable, circular, multiple-dose cassette advanced by opening and closing a clear plastic lid. A twinblade impeller generates an aerosol cloud for delivery to the lungs with a slow, full breath. This study used a prototype version of the Spiros inhaler that contained identical powder path and aerosol delivery specifications, but that used a short lever for cassette advancement. We used a methacholine challenge-based clinical bioassay of pulmonary drug delivery to compare the relative quantity of albuterol delivered to the -receptors by the Spiros DPI and Ventolin MDI, the international reference product. It was our hypothesis that the two systems would deliver comparable amounts of albuterol to the airways.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Design

Subjects  12 yr of age with mild to moderate asthma were recruited for this study (1). At an initial screening visit, FEV₁ and PC₂₀ (provocative concentration of methacholine causing a 20% fall in FEV₁) were required to be  65% of predicted and  4 mg/ml, respectively, and PC₂₀ was required to increase by at least eightfold in response to administration of two actuations of Ventolin. At subsequent visits, FEV₁ had to be  65% of predicted, and PC₂₀ had to be within twofold of the screening value. Subjects were excluded if they used more than an average of one -agonist inhaler per month (determined on the basis of prescription refills), had had a respiratory tract infection within 30 d, used oral corticosteroids within 3 mo of screening, had a history of life-threatening asthma, or had significant illness in addition to asthma. Stable doses of theophylline, inhaled corticosteroid, oral or inhaled -agonist, oral antihistamines (excluding astemizole), and intranasal corticosteroids were permitted. None of the subjects were taking ipratropium, cromolyn, or nedocromil before or during the study.

Thirty-one nonsmoking subjects were enrolled in this double-blind, randomized, crossover study. Twenty-four subjects completed all four treatment periods (15 males, 9 females; mean age, 26.2 yr; age range, 12 to 46 yr; Table 1). Two subjects were discontinued before study completion secondary to use of a drug prohibited in the protocol (prednisone) in response to an adverse event (asthma exacerbation). Four subjects were discontinued because of failure to meet protocol-specified criteria for between-visit reproducibility of baseline FEV₁ or baseline PC₂₀ and one subject was discontinued for failure to meet protocol-specified criteria for minimal responsiveness of PC₂₀ to albuterol administration. Data from the 24 subjects completing the trial were used in the efficacy analyses. Data from all 31 subjects randomized into the study were used in safety assessments.

Four treatments were studied: one and three actuations of Ventolin MDI (90 µg of albuterol base per actuation) and one and three actuations of Albuterol Spiros DPI (108 µg of albuterol sulfate, equivalent to 90 µg of albuterol base per actuation). One of these treatments was administered on each of four separate study days. Blinding to both dose and device was accomplished by use of a double-dummy technique. The protocol received institutional review board approval, and written informed consent was obtained from subjects before participation.

Methacholine challenges were initiated 3 h before (predose) and 15 min after (postdose) albuterol dosing. The protocol required that short-acting -agonists (e.g., albuterol) be withheld for 6 h, salmeterol for 48 h, theophylline for 48 h, hydroxyzine for 96 h, astemizole for 3 mo, all other antihistamines for 48 h, and  blockers and nonsteroidal antiinflammatory drugs for 7 d before each methacholine challenge. Safety of each treatment was assessed by monitoring the nature and frequency of adverse events; by measurement of pulse rate, blood pressure, blood glucose, and serum potassium before and after each treatment; and by performing a 12-lead electrocardiogram before and after treatment.

Methacholine challenges were performed according to the method of Cockcroft and coworkers (19, 20). Spirometry was performed with a KOKO pneumotachograph and software (Pulmonary Data Systems Research, Louisville, CO). The software was specifically adapted to accommodate the tidal breathing (19) method of methacholine challenge. Results were expressed as provocative concentration of methacholine (mg/ml) required to produce a 20% decrease in FEV₁ (PC₂₀). For purposes of data presentation, results are also expressed as the activity ratio: Activity ratio = postdose PC₂₀/geometric mean predose PC₂₀ for that subject. This indicates the number of times by which the PC₂₀ was increased by albuterol treatment. Results of statistical analysis of data expressed in these two ways (PC₂₀ and activity ratio) are mathematically identical. Thus, only the results of statistical analysis for PC₂₀ are presented.

Data Analysis

Results were analyzed using repeated-measures analysis of variance (ANOVA) with log_e(PC₂₀) as the dependent variable. Log_e transformation of PC₂₀ was necessary in order to meet the normality and equal variance assumption of ANOVA. Treatment, study day, and subject were factors in the ANOVA. The relative potency of the Spiros and Ventolin inhalers was calculated according to the 2-by-2 method of Finney (24). Study validity was evaluated by established criteria that test for a significant dose-response relationship (regression contrast), for the presence of a significant deviation from parallelism of the Spiros and Ventolin dose-response curves (parallelism contrast), and for the presence of a significant difference in mean responses to Spiros and Ventolin (preparations contrast). A significant preparations contrast would indicate that the Spiros and Ventolin dose- response curves did not overlap sufficiently for a valid potency estimate to be obtained. A significant parallelism contrast would indicate that curves for Spiros and Ventolin were not parallel. When this is true, relative potency is not constant, and changes as dose changes. Relative potency is expressed as the number of actuations of Ventolin required to produce an increase in PC₂₀ equal to that produced by one actuation of Spiros DPI. For the purposes of this study, we prospectively specified that the two inhalers would be considered "comparable" if one actuation of the Spiros inhaler was at least half as potent and no more than twice as potent as one actuation of Ventolin. This requires that the 90% confidence interval for relative potency of the Spiros inhaler be contained within an interval of 0.5 to 2.0 (25). Conceptually, this is equivalent to demonstrating that the DPI is more than half as potent (p < 0.05) and less than twice as potent (p < 0.05) as Ventolin. This statistical approach, known as two, one-sided hypothesis testing, is accepted and widely used by the Food and Drug Administration (FDA) in tests of bioequivalence.

An interim analysis was carried out after the first 11 subjects had completed the protocol. The intent of the interim analysis was to permit early detection of a marked difference in potency of the Spiros and Ventolin inhalers. If the per-actuation potency of the Spiros inhaler had been markedly different from that of Ventolin, the study sponsor planned to stop this study and to modify the amount of albuterol contained in one actuation of the DPI before proceeding with further clinical evaluation. On the basis of the results of a preestablished decision algorithm, the conclusion from the interim analysis was that the study should be extended to completion (24 evaluable subjects). Final analyses were adjusted for performance of the interim analysis, using the method of Lan and Demets and the O'Brien-Fleming spending function (26, 27).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Geometric mean predose PC₂₀ values were not significantly different (0.65, 0.65, 0.59, and 0.64 mg/ml for one and three actuations of Spiros, and one and three actuations of Ventolin, respectively; p = 0.798). The geometric mean postdose PC₂₀ values for the four treatments were 7.50, 10.15, 5.55, and 12.27 mg/ml, respectively. The corresponding geometric mean activity ratios for these four treatments (which indicate the change in PC₂₀ induced by the albuterol inhaler treatment) were 11.47, 15.62, 9.48, and 19.18, respectively. A significant dose-response relationship was present (p = 0.0002) and the parallelism and preparations contrasts were not significant (p = 0.079, 0.686; Figure 1). The estimated relative potency based on these data indicated that one actuation of the Spiros inhaler is equivalent to 1.12 actuations of Ventolin (90% confidence interval 0.68 to 1.94). Adverse event profiles were similar for the two inhalers (Table 2) and no serious events were reported. Safety assessment (pulse rate, blood pressure, serum potassium, blood glucose, and electrocardiogram) showed no clinically meaningful differences between the Spiros and Ventolin inhalers.

View larger version (13K): [in this window] [in a new window]   Figure 1.   Dose-response curves for Ventolin MDI and Spiros DPI. (A) Response is plotted as PC20 measured by a methacholine challenge initiated 15 min after Ventolin or Spiros dosing. (B) The same results expressed as an activity ratio (number of times by which the PC20 was increased by albuterol treatment). These data indicate that 1 actuation of Spiros is equivalent to 1.12 actuations of Ventolin (90% confidence interval, 0.68 to 1.94).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Results presented here provide evidence that the dose of albuterol delivered to the site of relevant -receptors in the lung by Spiros DPI is "comparable" to that delivered by Ventolin. The relative potency estimate indicates that each actuation of Spiros inhaler delivers approximately the same quantity of albuterol to the vicinity of these -receptors, commonly referred to as the "biophase," as 1.12 actuations of Ventolin. The fact that the 90% confidence interval for this estimate, 0.68 to 1.94, is completely contained within the 0.5-to-2.0 interval indicates that the Spiros inhaler meets the definition of comparability we set out prospectively for the purposes of this study. Markers of systemic effects (blood pressure, pulse rate, serum potassium, blood glucose, and electrocardiogram) indicate no important difference in systemic absorption of albuterol for the two delivery systems.

Current FDA policies draw a clear distinction between inhaled albuterol preparations being developed as generic equivalents to Ventolin or Proventil and new inhaled albuterol delivery systems being developed to replace these CFC-containing inhalers as they are phased out of use. Ventolin and Proventil deliver bioequivalent quantities of albuterol to the lung (28) and all generic brands currently marketed in the United States have been rigorously tested for in vivo bioequivalence to one or both of these "innovator" albuterol brands.* This requires demonstration (with 90% confidence) that the generic inhaler delivers between 0.67 and 1.5 times as much drug to the biophase as Ventolin or Proventil. In contrast, current policies do not require that rigorous standards of "bioequivalence" to Ventolin and/or Proventil be met by new inhalers, but that they be shown to be "comparable" to the CFC formulations. Standards for demonstration of comparability do not appear to be clearly defined. This regulatory distinction between generic formulations and new non-CFC formulations of currently marketed drugs is not clearly made in other countries (29). Nor is it likely to be important to clinicians and patients who simply desire that "new" inhalers not be less effective or more risky than those they currently use.

We believe that the prospective definition of comparability used in this study is appropriate from a clinical perspective. We chose a lower relative potency limit of 0.5, which indicates that one actuation of Spiros (with 90% confidence) should be more than half as potent as Ventolin. In many situations in which patients use Ventolin or Proventil MDI, one or two inhalations are sufficient to provide relief of or protect from asthma symptoms (e.g., exercise-induced bronchospasm). In other circumstances (e.g., when the patient awakens with nocturnal asthma), as many as five actuations may be required to normalize lung function (30). If a new albuterol preparation were, in truth, half as potent as Ventolin, the number of actuations required to produce clinical benefit would double. This is not likely to present a large practical problem for patients. In fact, two other commonly used -agonist MDIs (metaproterenol and terbutaline) are no more than half as potent per actuation as Ventolin (15, 31) (Figure 2). On the other hand, if the new preparation, in truth, delivered one-quarter or one-third as much drug to the biophase as Ventolin, then three to four or more times as many actuations would be needed. This situation is likely to be impractical for many patients.

View larger version (34K): [in this window] [in a new window]   Figure 2.   Comparison of selected studies used to evaluate comparability or bioequivalence of new albuterol inhalers to Ventolin MDI. (A) The x-axis represents the variability of responses measured (s) divided by the slope of the observed dose-response relationship (b; see text). The y-axis represents the difference in response between test inhaler and Ventolin (with 90% confidence interval) expressed as effect size (number of standard deviations by which the two inhalers differ [32]). Curved reference lines represent the expected difference between the two inhalers as a function of the true s/b ratio and the true potency of the test inhaler relative to Ventolin. (B) The y-axis represents the potency of test inhaler relative to Ventolin (with 90% confidence interval). Plotted bars represent individual studies and are derived from results displayed in (A) using bioassay statistical procedures. Studies marked with an asterisk violate one or more study validity criteria, rendering study results of uncertain reliability.

For the present study, we chose a relative potency of 2.0 as an upper acceptability limit, indicating that the new inhaler must deliver no more than twice as much albuterol to the airway biophase as Ventolin. This is based on concern about potential adverse effects, should too much albuterol be delivered. Concern over acute toxic effects is minimal because inhaled albuterol is a drug with a wide margin of safety (except perhaps in sensitive patients such as the elderly). On the other hand, there is long-standing concern over whether continued excessive use of albuterol (or other inhaled  agonists) can degrade rather than enhance control of asthma symptoms, or worse, increase the risk of life-threatening symptoms or death from asthma. Existing data indicate that this risk is greater for fenoterol than for albuterol (12). Because the marketed fenoterol preparation is approximately twice as potent as Ventolin, we reasoned that it would be undesirable for the potency of Spiros to be twice as great or more than that of Ventolin.

Implicit in the FDA criteria for demonstration of in vivo bioequivalence for generic albuterol preparations, and in the foregoing discussion of the definition of comparability for new albuterol preparations, is the presumption that comparisons between a new albuterol preparation and Ventolin should be made along the dose axis rather than the response axis of the dose-response curve. In other words, results should be expressed as a ratio of doses producing equal effects (synonymous with potency ratio or relative potency) rather than a simple comparison of the magnitude of responses observed at equal nominal doses of the two preparations. Comparisons of magnitude of response will yield different results depending on the nature and units of measurement of the specific response evaluated, characteristics of the asthmatic subjects studied, and the study design used. Consequently, albuterol preparations that appear comparable in one study may not be in another. It will be difficult or impossible to reconcile different answers concerning comparability that are obtained from different studies comparing response. In contrast, studies that estimate relative potency will all produce results that are expressed in the same units and that can be directly compared with one another even if study design and subject characteristics differ. Furthermore, statistical procedures for computing relative potency have well-established validity criteria and methods for determining confidence limits on potency ratio. These serve as an indicator of study quality (i.e., the ability to accurately distinguish differences in dose delivered). Studies comparing magnitude of response do not.

To provide relative potency information, studies must be specifically designed as clinical bioassays of the quantity of albuterol delivered, rather than as clinical trials of efficacy. This concept has been discussed in more detail elsewhere (33). In brief, each study must use a clinically relevant response measure, must give more than one dose of each of the two preparations being compared, must meet internal validity criteria (as laid out in METHODS), and finally, must have sufficient statistical power to provide clinically useful information.

Published studies provide instructive examples of both successful clinical bioassay experiments and problems that can arise when these principles are not followed. Dockhorn and coworkers (8), in a pivotal study assessing comparability of Proventil HFA to Ventolin, found that the bronchodilation responses to one and two actuations of each of these inhalers were both significantly different from responses to placebo but not from each other (hence "comparable"). Yet, in reality, this study lacks the power to clearly establish relative potencies of the two preparations. Statistical power of a bioassay experiment is a function of both the variability of responses measured (e.g., standard deviation or equivalent, commonly symbolized as "s") and the slope of the dose-response curve (commonly symbolized by statisticians as "b"). In fact, it is the ratio of these two factors, s/b, that is used in sample size computations for design of bioassay experiments (the lower the value of s/b, the fewer subjects required and the greater the power of the study) (24). For this reason, s/b is used on the x-axis for Figure 2A. When results of the study by Dockhorn and coworkers are reanalyzed as a bioassay experiment, a relatively large s/b value is obtained (Figure 2A) and the 90% confidence interval on the relative potency estimate is so wide that it is not clinically useful (Figure 2B). Results of this study violate one of the standard bioassay validity criteria in that a significant dose-response relationship is not present (regression contrast). Therefore, the potency estimate is of uncertain reliability. A study published by Kleerup and colleagues (9) also assessed comparability of Proventil HFA and Ventolin. This study has a low s/b value when reanalyzed using bioassay statistical procedures (Figure 2A). Consequently, there is a narrow confidence interval about the relative potency estimate (Figure 2B). Yet, this study violates two validity criteria: the dose-response curves for the two preparations differ significantly from being parallel (parallelism contrast), and the curves do not overlap sufficiently (preparations contrast). This makes the potency estimate from the study of uncertain reliability. Fortunately, another study of Proventil HFA, specifically designed as a bioassay experiment, has been done that meets the standard validity criteria. This study uses methacholine bronchoprovocation to demonstrate that the potency of Proventil HFA relative to Ventolin is 1.05. This estimate has a 90% confidence interval (0.78 to 1.42) that even falls within the FDA 0.67-1.50 bioequivalence range set for generic inhaled albuterol MDIs (not shown in Figure 2) (34). Similarly, the current study, and a similar evaluation of CFC-containing Proventil, have also successfully used a clinical bioassay approach to establish bioequivalence to Ventolin (28).

In summary, we used a methacholine bioassay to demonstrate that Albuterol Spiros, a new DPI, delivers an estimated 1.12 times as much drug to the lung as Ventolin (90% confidence interval, 0.68 to 1.94). We propose that for this and other new albuterol inhaler systems to be considered "comparable" to Ventolin, they should demonstrate, in a clinical bioassay, the ability to deliver between 0.5 and 2.0 times as much albuterol to the -receptor biophase in the lung.