©1997 by Steve Duff. All rights reserved
The heated debate over a possible causal relationship between drugs used to treat asthma and increased asthma morbidity and mortality has raged since the late 1940s. The controversy has predominantly focused on a class of drugs known as the short-acting ß-agonists that are used for bronchodilation. However, in the last 6 years a new type of ß-agonist has emerged on the market, the long-acting ß-agonist, and one drug in particular in this class, salmeterol xinafoate (Serevent®), has come under close scrutiny for investigation of this causal relationship. This paper will review the historical published evidence that appears to indict Serevent and that which appears to exonerate it. Hopefully, after thoughtful review, some conclusion can be drawn as to the probability that Serevent is responsible for increased asthma morbidity and mortality.
Serevent is a long acting, ß2-adrenergic receptor agonist administered by oral inhalation to treat the symptoms of asthma, mainly through bronchodilation.1 It is marketed by Allen and Hanburys, a division of Glaxo, and is used on a twice a day dosing regimen as opposed to a four times a day dosing regimen necessary with the other ß2-agonists. Because Serevent was an innovator in its class and is relatively new to the market (approved in the UK in 1990 and in the United States in 1994), it is not surprising that it has come under closer scrutiny than many other drugs. But, before presenting the evidence regarding Serevent, some background information on the asthma mortality epidemics and the proposed ß2-agonist association must be presented.
Asthma mortality epidemics seem to occur in waves with the first serious study of increasing asthma mortality taking place in the 1960s due to a dramatic increase in asthma deaths in England and Wales.2 FE Speizer was one of the initial investigators of the possible association of drugs used to treat asthma with sudden death from asthma and he attempted to use case reports3 to pinpoint which drugs might be the cause. Corticosteroids and pressurized aerosol bronchodilators were used by most of the cases he studied but no conclusive proof could be offered for their causing the deaths. WH Inman and PD Stolley were the next to pursue the cause and using vital statistics data4-6 as circumstantial evidence, such as a correlation between increased drug sales and an increase in mortality, postulated that isoprenaline forte may be associated with the deaths.
The next major asthma mortality epidemic occurred in the late 1970s in New Zealand and was studied much more extensively and with greater sophistication than the UK epidemic. Three major case-control studies,7-9 were carried out and all posited that fenoterol, a fairly non-selective ß2-agonist, was associated with an increased odds ratio of death of 1.99 - 3.00. The studies were soundly criticized for study design, control selection, sources of drug information, and severity stratification. However, the seed had been planted and rightly or wrongly, fenoterol was removed from the New Zealand drug tariff on August 1, 1990. This also set the stage for the entrance of Serevent in the UK and the continuation of the controversy.
Serevent has been the focus of many papers published in medical journals since the first clinical trials in 1984 and its introduction in the UK market in 1990. However, it is meaningless to present the seminal pieces of work on Serevent without also presenting some of the studies of other ß-agonists that were being published in the same period. The history and the controversy of Serevent has been intertwined and greatly impacted by this other body of literature.
Early clinical trials had been conducted with Serevent since January 198410 and the drug had been shown to have long-acting bronchodilating effect in patients with moderate asthma,11 fewer asthma exacerbations, and a decreased need for rescue albuterol used on demand. However, in the December 1, 1990 edition of the Lancet, two papers appeared that would drag Serevent into the middle of the ß-agonist debate. The first paper, by Malcolm Sears and colleagues,12 was a double-blind, placebo-controlled, randomized crossover study of the effects of regular versus on-demand inhaled bronchodilator therapy. Although Serevent was not used in the study, the authors’ conclusions surely impacted Serevent use. They concluded that regular inhalation of ß-agonists was associated with deterioration of asthma control and recommended that they not be used regularly, but rather on demand only. They also stated in their Discussion that it was possible that the increasing use of betasympathomimetic drugs was contributing to the world wide increase in morbidity and mortality from asthma. These conclusions directly challenged the wisdom of using a drug such as Serevent in the treatment of asthma as it is only useful for regular use, not on an as needed basis.
The second article, by Orion P. Twentyman and colleagues,13 investigated the effects of salmeterol on early and late phase airway events provoked by inhaled allergens. They studied 10 asthmatic patients and concluded that the effects of salmeterol were shown to be unrelated to prolonged bronchodilation, the usual mechanism ascribed to Serevent, but rather to some "novel actions for topically active long-acting ß2-agonists in asthma that extend beyond their protective action on airway smooth muscle."
I believe that these two articles were the critical point at which Serevent now came under suspicion. This belief is supported by the numerous letters that were written in response to the articles.14-24 Letters responding to the Sears article were predominantly negative. The overall feeling was that extending the results for the ß-agonists studied to those that were not was inappropriate. Several correspondents referenced the large bulk of clinical trial data that had been accumulated and showed no deterioration of asthma control, but rather improvements in control. However, one respondent did feel that the Sears article was final "proof" that ß-agonists should not be used as a regular treatment. The bulk of the letters reacting to the Twentyman article did not believe that the evidence supported an anti-inflammatory claim, while others felt that much more research was needed in this area before making this claim. I sensed that the overall theme of the letters was a feeling that the manufacturers were trying to advance a claim that would give Serevent an edge over other broncholdilators, but as yet was unproven.
The stage had now been set for the flurry of papers that would seek to illuminate the "truth" about Serevent and other ß-agonists. In April and June of 1991, further evidence25,26 was presented in the Lancet in answer to Sears’ assertions of deteriorating asthma control. Glaxo researchers presented evidence from clinical trials that was in contrast to that of the Sears publication. These trials were initiated to monitor exacerbations of asthma symptoms when patients were treated with Serevent over a 12 month period. The results presented showed that after an increase in the number of patients with exacerbations in the first 3 months (21%), the number of patients with exacerbations leveled off at about 13% over the remaining 9 months. The number of exacerbations for patients taking low or high doses of concomitant inhaled corticosteroids was similar to those not taking corticosteroids, 13% versus 14%, respectively. In another study, the percentage of patients experiencing exacerbations was lowest in the patients randomized to the Serevent group (16%), and highest (32%) in the placebo treated patients who received salbutamol (albuterol) as required--the regimen promoted by Sears and colleagues.
A Swedish report that was published as an editorial in the European Respiratory Journal27 was also critical of the generalizations that Sears and colleagues had made, citing the more thorough prospective evaluation that salmeterol had undergone as compared to the older ß-agonists. The editorial concluded that there was insufficient evidence to support a causal relationship between long-acting ß-agonists and any deleterious effects. But, lest anyone believe that the debate had died down, a review article in Clinical and Experimental Allergy,28 reiterated the evidence against isoprenaline forte and fenoterol and concluded that the lessons learned should not be ignored and further study with ß-agonists was warranted. This was probably prudent, however the authors concluded by stating that "those who do not learn the lessons of history are condemned to repeat it." I’m sure some took this to be a veiled reference to the use of new, long-acting drugs causing another asthma epidemic. It just as easily could be interpreted as a warning that confounding by severity would continue to plague observational studies of newer medications for asthma, just as it had with earlier drugs.
In December 1991, the Pulmonary-Allergy Drugs Advisory Committee of the Food and Drug Administration deemed the debate surrounding adverse effects and inhaled ß-agonists important enough to convene a meeting to discuss the issues. The two day meeting involved the Committee, academic and industry researchers, FDA personnel, physicians, and representatives from groups such as the American Lung Association.29 The majority of the discussion at the meeting was focused on short-acting ß-agonists. There were numerous presentations and discussions of issues such as overuse of ß-agonists, the concomitant use of anti-inflammatory therapies, the epidemiological evidence of increased asthma morbidity and mortality as well as evidence for and against a causal relationship with ß-agonists. Although long-acting ß-agonists were infrequently discussed, issues regarding their use did arise. Some attending the meeting expressed their concern that these agents could mask a loss of asthma control, while a presentation by Dr. Harold Nelson noted the advantages of long-acting medications for nocturnal symptoms.
Despite the numerous opinions on the subject, the consensus of the Committee was that "beta-2 agonists were an important and necessary form of treatment for asthma and should remain available for use..." They also felt that the Sears and New Zealand studies presented insufficient evidence to suggest that ß-agonists as a class were harmful. However, the Committee also made many comments and recommendations; that long-term safety and effectiveness should continue to be studied, that overuse and unnecessary use were concerning, and that concomitant anti-inflammatory/ß-agonist therapy should also continue to be studied.
The 1992 publication of a case-control study30 that had previously been presented at the 1991 FDA meeting, continued to fan the flames of controversy. Using subjects drawn from a large cohort in Saskatchewan’s linked health insurance databases, Walter Spitzer and colleagues matched 129 patients who had fatal or near-fatal asthma events with 655 controls from the period 1978 to 1987. Their analysis showed that ß-agonists administered by a metered-dose inhaler (MDI) were associated with an increased risk of death or near death from asthma, especially for fenoterol. The odds ratio calculated for fenoterol and albuterol were 5.4 and 2.4, respectively, or 2.3 and 2.4, respectively, when adjusted to be on a microgram-equivalent basis. Although the authors did not state that the association was clearly causal, they did assert that high ß-agonist use was a marker of deteriorating asthma and recommended that physicians be wary when using ß-agonists. Again, Serevent was not mentioned in this study because it was not yet approved for use during the period studied, yet this report must still have impacted many physicians’ view of the wisdom of using any type of ß-agonist.
Compounding the insinuations that Serevent should be considered as "dangerous" as the short-acting ß-agonists, was a trickle of case reports of adverse events that began in October 1992. James Wilkinson reported six patients that had suffered acute bronchospasm after inhaling salmeterol.31 However, after further study he concluded that the observed bronchoconstriction was not due to salmeterol but to a chemical in the diskhaler, and suggested that affected patients be transferred to a dry powder formulation. Shortly after Wilkinson’s report came CE Clark’s case reports of respiratory arrests in young asthmatics who were taking salmeterol.32 Clark reported case histories of three young asthmatics, none of whom had suffered a respiratory arrest until 1-4 weeks after starting salmeterol therapy. Clark also referenced 26 cases of non-fatal deterioration closely temporally associated with the inhalation of salmeterol which were all reversible on stopping the drug.33 Based on this evidence, his group concluded that they would not prescribe salmeterol in young asthmatics until more evidence as to its safety was presented. Soon after the publication of the case reports, an observant physician noted34 that Clark’s three cases may have occurred due to the fact that anti-inflammatory therapy had been discontinued, an action not recommended by The British Thoracic Society guidelines for management of chronic asthma.35
In keeping with the pattern of published evidence showing an association followed closely by published evidence showing no association, results from a double blind randomized clinical trial involving over 25,000 patients in the UK were presented shortly after the aforementioned case reports. The study,36 conducted by Win Castle and colleagues, was at the time the largest clinical trial ever conducted in the UK.
It compared the safety of Serevent used two times a day with salbutamol used four times a day over a 16 week period. The results of the study showed that there were fewer medical withdrawals by patients taking salmeterol than by those taking salbutamol. However, there was an excess mortality risk of 3.0 associated with salmeterol, although it did not reach statistical significance. The authors concluded that the control of asthma was better in patients taking salmeterol and that serious adverse events occurred in patients most at risk on entry and were probably due to the disease rather than the treatment. Of course, even though the authors felt this vindicated salmeterol, a rash of letters were received criticizing the study.37-40
William Inman and the Drug Safety Research Unit did not believe that Castle’s study should be accepted as any form of postmarketing surveillance and referred readers to preliminary results from their ongoing prescription event monitoring study--a "true" surveillance study--that had so far shown a much higher mortality rate than that reported by Castle. They cautioned readers not to use Castle’s results to estimate the mortality from asthma. Robert Bunney’s criticism of the Castle article was well taken. He stated that given the number of subjects, the study would not have had sufficient power to detect a small increase in death rate at the 0.05 level of significance. In fact, the death rate due to salmeterol would have to be 4.33 times greater than salbutamol to show significance. Castle and colleagues replied41 to this criticism by stating that the timescale involved prohibited conducting a larger study and that the study was designed to examine the overall profile of serious adverse events with salmeterol compared to salbutamol. Graham Crompton believed that the study design, comparing salmeterol to salbutamol, prohibited answering whether an apparent deterioration in asthma during prolonged regular treatment existed with salmeterol. Castle replied that other studies had shown that there was a reduction in exacerbations and admissions with salmeterol treatment.42,43 Sears and Taylor still felt that salmeterol may mask or increase the severity of the underlying disease and referenced Clark’s respiratory arrest case reports. They conclude that Castle’s study and mortality data did not alleviate the concerns that regular use of ß-agonists may be contributing to morbidity due to and mortality from asthma.
The next piece of published evidence was a large meta-analysis44 of case-control studies that had attempted to measure the association between ß-agonists and death from asthma. MaryLou Mullen tested a total of six case-control studies with 364 cases and 1388 controls. She concluded that there may be a relationship between ß-agonist use and death from asthma, but that it was extremely small and only evident with drugs administered by nebulizer and not by MDI. She described the results from the Spitzer/Saskatchewan study as "outliers" and that the conclusions they had drawn were not borne out by the accumulated weight of evidence.
The publication of the results from the full cohort that had spawned Spitzer’s nested case-control study was reported in March 1994 by Samy Suissa and colleagues.45 When all 12,301 users of asthma drugs were studied, a strong association only existed with use of the drugs in excess of recommended limits. This appeared to be the first retreat from the strong stance that no ß-agonist therapy should be used.
Another interesting result that necessitated retreat from their prior position was published in October 1994 by Ronald Mann at the Drug Safety Research Unit.46,47 This publication presented the final results that Inman had alluded to in criticizing Castle’s results. In 1993 Inman had reported that the death rate would be much higher than that reported by Castle, and that Castle’s results should be viewed with caution. However, upon conclusion of the study it was found that the death rate due to asthma of those patients taking salmeterol was much less than previously predicted. Although the death rate was higher in this study than in Castle’s, it was due to the longer study duration, and an older and more severely ill population. Mann concluded that the study "elicited no evidence that salmeterol contributed to the deaths from asthma."
If these two studies are viewed as some vindication for Serevent, then further proof of its relative safety came in September 1995 with the publication of yet another clinical trial48 that compared the safety of salmeterol and albuterol in the maintenance treatment of 556 asthmatics over a 12 week period. It presented evidence that asthma exacerbations were lower in the salmeterol group than in the albuterol group, that there appeared to be no evidence of any increased risk of cardiovascular events or arrhythmogenic effects, and that no deaths had occurred over the study period. A review49 and an editorial50 also published in 1995 took the stance that Serevent was safe and effective for the treatment of asthma.
There is no doubt that the speculated association between drugs such as isoprenaline forte and fenoterol, and increased mortality and morbidity from asthma has impacted Serevent since its introduction in the UK six years ago. Although, it is always wise to be fairly cautious when a new drug is introduced, the wisdom of generalizing results from studies of short acting ß-agonists is definitely questionable. A review of the pharmacology of many of the older ß-agonists shows that they are much less receptor specific than the newer generation of ß-agonists and certainly the older drugs did not have to undergo as extensive of an approval process in the 1960s as is common today. Numerous clinical trials have shown Serevent to be quite safe, even when compared to existing standard therapies such as albuterol. However, due to the fact that clinical trials are usually of fairly short duration and may not involve populations large enough to elucidate rare occurrences, and that many have postulated that albuterol may also be harmful, they must not be taken as final proof. Therefore, it is necessary to observe what other proof exists that Serevent may be associated with increased morbidity and mortality.
Because it seems obvious that simply extending the conclusions drawn from older, less specific ß-agonists is not sufficient to prove a causal relationship for Serevent, it is necessary to view only the evidence directly attributable to Serevent. The main pieces of evidence indicating that Serevent may be causing an increase in morbidity and mortality are case reports of non-fatal respiratory arrests and several deaths where the patients were found clutching their inhalers.51 The non-fatal respiratory arrest case reports were later shown to probably have been caused by discontinuation of anti-inflammatory medication rather than Serevent. As for the two elderly patients with moderately severe asthma that died clutching their inhalers, it can probably never be shown that these isolated cases were truly indicative of a causal relationship. Certainly in this population, education is integral to the proper use of drugs, and compliance can be very poor. There are many other explanations that can be posited for the deaths, including death from asthma unrelated to drug therapy. The "Webber effect"--an increased number of adverse event reports soon after drug introduction to the market followed by a decrease over time--must surely be kept in mind when reviewing case reports.
Unlike the older ß-agonists, there have not been numerous, large retrospective or prospective studies studying Serevent and mortality and morbidity. The prescription event monitoring study of 15,407 patients over two years was touted in 1993 in the British Medical Journal as showing a much larger risk of death with Serevent than the large Castle clinical trial. However after completion, it eventually showed a risk only slightly larger than Castle’s and the authors explained that the increase was due to a study population that was older and sicker than that in Castle’s trial. The Castle nationwide surveillance study of 25,180 patients was the only other study of that magnitude, and it did not show a higher risk of mortality that was statistically significant. It should be kept in mind that it may not have had enough enrolled patients to see statistically higher risks.
When viewed apart from the data on short acting ß-agonists, the evidence attempting to link Serevent with increased morbidity and mortality from asthma does not appear convincing. There is much more of an abundance of clinical trial evidence that shows that Serevent is much safer than its older brethren than there is that shows it to be a causal factor in death or exacerbation of asthma symptoms. Past work on the short-acting ß-agonists should not be generalized to indict a drug that is much more receptor specific and that has been subjected to much more initial pre-approval study. However, this does not mean that Serevent is entirely innocuous. More study should be conducted with large cohorts and surveillance should be continued. What is probably most important is that patients who use Serevent are well educated on its proper use and that they receive careful and frequent medical attention.
Return to the DDIL Homepage
This
document has been provided through the Drugs and Devices Information Line,
http://www.hsph.harvard.edu/Organizations/DDIL/ddil.html