Published ahead of print on February 27, 2004, doi:10.1164/rccm.200309-1219OC
© 2004 American Thoracic Society Effects of an Angiotensin-converting Enzyme Inhibitor–based Regimen on Pneumonia RiskTohoku University Graduate School of Pharmaceutical Science, Sendai; National Cardiovascular Center, Osaka, Japan; St. Mary's Hospital, London, United Kingdom; and Institute for International Health, University of Sydney, Sydney, Australia Correspondence and requests for reprints should be addressed to Takayoshi Ohkubo, M.D., Ph.D., c/o PROGRESS Collaborative Group, Department of Planning for Drug Development and Clinical Evaluation, Tohoku University Graduate School of Pharmaceutical Science, Clinical Pharmacology and Therapeutics, Tohoku University Hospital, Sendai, 980–8574, Japan. E-mail: tohkubo{at}mail.tains.tohoku.ac.jp
Observational studies conducted among Asian populations suggest that the risk of pneumonia is substantially reduced among users of angiotensin-converting enzyme (ACE) inhibitors but not other blood pressure–lowering agents. We conducted analyses of the effects of ACE inhibitor therapy on pneumonia in 6,105 patients with a history of stroke or transient ischemic attack enrolled in a randomized trial conducted in Australasia, Europe, and Asia. Patients were randomly assigned perindopril-based active treatment or placebo. The effects of ACE inhibitors on pneumonia (fatal or nonfatal) were determined from Cox models fitted according to the principle of intention to treat. During a median follow-up of 3.9 years, 261 patients developed pneumonia. Overall, active treatment was associated with a nonsignificant 19% lower risk of pneumonia (95% confidence interval, –3 to 37; p = 0.09) compared with placebo. Active treatment significantly reduced the risk of pneumonia among participants of Asian ethnicity (47%, 14–67%; p = 0.01), with no significant effect among non-Asian participants (5%, –27 to 29%; p = 0.7) (p for homogeneity = 0.04). These findings substantially add to the body of evidence about the effects of these drugs on pneumonia but do not provide the definitive information required to inform clinical decisions about the prevention of pneumonia with ACE inhibitors.
Key Words: pneumonia • angiotensin-converting enzyme inhibitor • angiotensin-converting enzyme insertion/deletion polymorphism • stroke • randomized controlled trial Pneumonia is a major cause of mortality and morbidity, particularly among those who are older and debilitated (1–4). In a number of observational studies conducted among older Japanese populations, the use of angiotensin-converting enzyme (ACE) inhibitors, but not other blood pressure–lowering drugs, has been associated with a reduced risk of pneumonia (5–11). Among populations who were older and debilitated, in whom silent aspiration of oropharyngeal pathogens (12) is believed to be a leading cause of pneumonia (1–4), it is possible that established effects of ACE inhibitors on cough (13) and swallowing (14) may provide protection against infection. Furthermore, any such benefit may be particularly marked among individuals of Asian ethnicity, in whom ACE inhibitor–related cough has been reported to be more prevalent (15, 16) and among whom ACE insertion/deletion (I/D) polymorphisms may influence the risk of pneumonia (17); however, the nonrandomized studies on which the observed effects of ACE inhibitors on pneumonia are based are subject to bias even after adjustment for known confounding factors, and the findings may thus be unreliable (18). The Perindopril Protection Against Recurrent Stroke Study (PROGRESS) was a large-scale randomized trial that was conducted among individuals with a history of stroke or transient ischemic attack, including participants from both Asian and non-Asian regions (19–23). The main objective of this secondary analysis of the trial was to determine the effects of the ACE inhibitor–based blood pressure–lowering regimen on the risk of pneumonia overall and among subgroups defined on the basis of Asian and non-Asian ethnicity.
Study Design The design and principal results of PROGRESS have been reported previously (19–21). In brief, 6,105 patients with a history of stroke and/or transient ischemic attack were enrolled from 172 collaborating centers in 10 countries (Australia, New Zealand, France, Belgium, Italy, Japan, People's Republic of China, Sweden, United Kingdom, and Ireland). The ethics committee of each collaborating center approved the study, and all participants provided written, informed consent. Individuals with a history of cerebrovascular disease (stroke, transient ischemic attack, or amaurosis fugax) within the previous 5 years and no definite indication for or contraindication to treatment with an ACE inhibitor were eligible for the study. There were no blood pressure entry criteria, although it was recommended that individuals with uncontrolled hypertension receive antihypertensive therapy with agents other than ACE inhibitors before entry to the trial. It was also recommended that participants should be clinically stable for at least 2 weeks after their most recent vascular event before entry to the study. All potential participants entered a 4- to 6-week run-in period on 2–4 mg/day of perindopril to check for tolerability and adherence. Patients that successfully completed the run-in period were randomly assigned either to continued active blood pressure–lowering treatment or to matching placebo therapy in a double-blind manner. Study treatment allocation was provided by a central computer-based randomization service accessed by telephone or facsimile. Active treatment comprised a flexible regimen based on perindopril (4 mg daily) with the addition of indapamide (2.5 mg daily or 2 mg daily in Japan) in patients for whom the responsible physician judged there to be no specific indication for or contraindication to treatment with a diuretic. Participants were instructed to take study tablets daily. All other aspects of medical and surgical care were left to the discretion of the responsible physician. Randomized participants had visits scheduled for five occasions in the first year after randomization and for six monthly intervals thereafter. Wherever possible, clinic visits were continued for the entire scheduled duration of follow-up for all surviving randomized patients, including those who discontinued study drug treatment for any reason. At each visit, inquiry was made about the occurrence of major nonfatal events, including pneumonia. An endpoint adjudication committee reviewed source documentation about the cause of death for all participants that died during the scheduled period of follow-up. This committee assigned a proximate or an underlying cause of death, or both. In this analysis, pneumonia was taken as the cause of death whether it was proximate or underlying. Pneumonia was defined according to the ninth revision of the International Classification of Diseases (pneumonia = 480–487, 507, or 514) (24) and was deemed fatal if the individual died within 28 days of the diagnosis. Samples of venous blood were collected in ethylenediaminetetraacetic acid for extraction of DNA from buffy coats using a salting-out procedure. Genotyping of the ACE I/D polymorphism was performed after polymerase chain reaction amplification of the region encompassing the polymorphism with three primers by hybridization with allele-specific oligonucleotides (23).
Statistical Power
Statistical Analysis
Participant Enrollment and Baseline Characteristics A total of 7,121 potential participants were registered and 6,105 were randomized. Of the 1,016 (14%) registered participants that were not randomized, 192 (2.7%) withdrew during the 4-week active run-in period because of cough. There was no difference in the rate of withdrawal due to cough in the run-in period between Asian and non-Asian participants (63 [2.4%] vs. 129 [2.9%], respectively; p = 0.2). Of those randomized, 3,051 were assigned active treatment and 3,054 were assigned placebo. Overall, 58% were treated with combination therapy or double placebo and 42% with perindopril alone or single placebo. The characteristics of the participants have previously been published in detail (20, 21) and are only summarized here (Table 1) . Overall, 39% were Asian and 61% were non-Asian. There was good balance between active treatment and placebo groups for all recorded participant characteristics, including those that might possibly be expected to influence the incidence and outcome of pneumonia such as age, dependency (measured by the Barthel index), smoking status, diabetes, and ACE I/D polymorphisms. ACE I/D genotyping was successful among the 5,688 randomized participants: Of these, 2,828 were in the active treatment group, and 2,860 were in the placebo group (23).
Follow-up and Discontinuation of the Treatment The median follow-up was 3.9 years, and vital status at the scheduled end of follow-up was ascertained for all but three randomized participants (0.05%). By the end of scheduled follow-up, or prior death, 1,350 (22%) participants had prematurely discontinued all study treatments (active 23%, placebo 21%; p = 0.02) (21). In addition, by the end of scheduled follow-up, a further 134 participants had prematurely discontinued perindopril/placebo (active 87, placebo 47) but continued with indapamide/placebo. Among the 1,484 participants that prematurely discontinued both study treatments or perindopril/placebo alone, cough was the reason cited for discontinuation in 127 (4.2%) actively treated participants and in 25 (0.8%) placebo-treated participants (p < 0.001). Overall, treatment with the ACE inhibitor was associated with a 5.3 (95% confidence interval [CI], 3.4 to 8.1) increase in the odds of treatment discontinuation caused by cough. This increased risk was of similar magnitude in both Asian (odds ratio, 5.8; 95% CI, 2.4 to 13.9) and non-Asian participants (5.1; 95% CI, 3.1 to 8.4) (p for homogeneity = 0.8).
Effects of Treatment on Pneumonia in All Participants
Effects of Treatment on Pneumonia in Participant Subgroups There was a significant reduction in the risk of pneumonia in the subgroup of participants of Asian origin (relative risk reduction 47%; 95% CI, 14 to 67%; p = 0.009) equating to 45 individuals treated for 5 years to avert one such event (95% CI, 27 to 146), but no clear reduction in participants recruited outside of Asia (relative risk reduction, 5%; 95% CI, –27 to 29%, p = 0.7) (p for homogeneity between Asians and the other origins = 0.04) (Figure 3) . There was no independently significant reduction in the risk of pneumonia in either of the participant subgroups defined on the basis of intended study treatment regimen (single drug therapy or combination therapy), with no statistical evidence of heterogeneity between the subgroups (p for homogeneity = 0.9) (Figure 3). Neither was there any evidence of heterogeneity between the effects of treatment in subgroups defined by sex (p for homogeneity = 0.2) or age (p for homogeneity = 0.9).
Effects of ACE I/D Polymorphism on Pneumonia The distribution of the ACE I/D polymorphisms in the PROGRESS population is shown (Table 1). There was a greater prevalence of the II polymorphisms in the Asian (II = 41%, ID = 44%, DD = 15%) compared with non-Asian populations (II = 21%, ID = 47%, DD = 32%) (p < 0.001). There was no clear effect of ACE I/D polymorphisms on the risk of pneumonia (Table 2) . Neither did the ACE I/D polymorphism interact with the effect of randomized treatment on pneumonia (Table 3) .
This randomized, controlled trial conducted among patients with cerebrovascular disease substantially adds to the body of evidence about the effects of ACE inhibitors on pneumonia. The study results suggest that ACE inhibitors might confer protection against pneumonia but do not provide the conclusive information required to consider these agents a new therapeutic option for pneumonia prevention. In particular, the post hoc nature of the analyses and their rather limited statistical power provide for persisting uncertainty about this possible new role of ACE inhibitors. The main reason that PROGRESS does not fully resolve the uncertainty surrounding the effects of ACE inhibitors on pneumonia is that the trial was not specifically designed to address this question. The rather few pneumonia events recorded in PROGRESS (n = 261) limit the statistical power and the multiple comparisons, and the post hoc nature of the analyses increases the chance of spurious positive or negative findings (18, 25). Thus, the apparent difference in protection against pneumonia afforded to Asian participants and non-Asian participants could easily be a consequence chance, as could the failure to detect anticipated differences in rates of cough between these population subgroups (15, 16). However, although statistical power may not have been high, the randomized design of PROGRESS did greatly reduce the likelihood of confounding of the analyses (18, 25) and provided an excellent opportunity to explore the validity of the associations reported in observational studies (5–11). In PROGRESS, the overall direction of effect of treatment was similar to that reported in the observational studies, and the size of the protective effect in the Asian population subgroup was broadly similar to the size of benefit reported previously (5–11). Although the comparability of the findings could still be coincidental, the PROGRESS results somewhat decrease the likelihood that the previously reported protective effects of ACE inhibitors on pneumonia were entirely the result of biased analyses. The means by which ACE inhibitors affect the respiratory system is thought to be through elevation of substance P, a neurotransmitter for primary sensory afferent nerves that is normally degraded by ACE. Elevated levels of substance P are associated with ACE inhibitor cough (5, 26) and may also produce the enhanced swallowing reflex observed among patients treated with ACE inhibitors (14, 27). In combination, an increased cough reflex cough and improved swallowing provide a reasonable basis for a decreased risk of aspiration pneumonia among ACE inhibitor–treated patients. A recent report in which the ACE D allele (of the ACE I/D polymorphism) was identified as an independent risk factor for pneumonia (17) provided a possible explanation for differences in the response of Asian and non-Asian population groups to ACE inhibitor therapy with regard to respiratory disease. However, although there was the anticipated differential distribution of the ACE I/D polymorphisms between Asian and non-Asian participants in PROGRESS (23), there was no effect of the D allele on pneumonia risk, and there was no interaction of the ACE I/D polymorphism with the effects of treatment on pneumonia. Furthermore, this was true not only for pneumonia but also for many other serious outcomes recorded in the study (21, 22). The absence of differences between the rates of cough in Asian and non-Asian participants in PROGRESS during either the run-in phase or the randomized treatment phase contrasts with previous reports documenting higher rates in Asian populations (15, 16). It is difficult to be certain whether the comparability in the rates of cough in these two population groups in PROGRESS is real, raising uncertainty about the previous reports of differences, or a consequence of a study design that was not specifically intended to address such questions (19). In particular, the PROGRESS study only recorded information about cough when serious enough to cause treatment discontinuation and did not gather detailed information about cough severity. Furthermore, the rather few discontinuations of therapy because of cough limited statistical power for comparisons between patient subgroups. Although the trial design makes it is very unlikely that other factors influencing decisions about discontinuation would have varied between treatment groups, it is possible that factors such as cultural attitudes could have resulted in systematic differences between population subgroups with regard to the termination of ACE inhibitor therapy because of side effects. For example, it is possible that a real difference in the effect of the ACE inhibitor on cough between Asian and non-Asian populations could have been obscured by a greater determination to continue with study treatment among one group compared with another. Misclassification of pneumonia as heart failure, the other leading cause of acute respiratory symptoms in this patient population, does not appear to have been an important problem in PROGRESS. Fatal cases of pneumonia were centrally verified, and diagnosis was likely very reliable. Nonfatal cases were diagnosed by clinicians at the many collaborating centers, and it is possible that some errors in diagnosis may have occurred. However, there was no evidence of any difference in the effects of randomized treatment on pneumonia between the patients treated with the combination (ACE inhibitor and diuretic vs. placebos) compared with those treated with the single-drug therapy (ACE inhibitor vs. placebo). If a large proportion of the cases of pneumonia were actually misdiagnosed cases of heart failure, a greater beneficial effect of combination treatment compared with single-drug treatment would have been expected. In PROGRESS, combination treatment resulted in twice the proportional benefit of single-drug therapy for heart failure (22) but only a 2% additional benefit for pneumonia. Similarly, the absence of heterogeneity of treatment effects by the presence or absence during follow-up of prior recurrent stroke suggests that prevention of stroke-related pneumonia was not a strong determinant of the treatment effects observed. In conclusion, these data from PROGRESS provide some additional support for beneficial effects of ACE inhibitors on the risk of pneumonia. However, this report does not fully resolve all of the outstanding uncertainties about the effects of ACE inhibitors on this outcome for either Asian or non-Asian individuals. Furthermore, these findings introduce important new questions about the putative role of the ACE I/D polymorphism in any such effects. Overviews of these results in conjunction with other completed and ongoing large-scale randomized trials of ACE inhibitors would provide one means of further elucidating the effects of ACE inhibitors on respiratory outcomes. In the meantime, ACE inhibitor–based blood pressure–lowering therapy remains a highly effective treatment for the prevention of serious cardiac and cerebral complications among patients with a history of vascular disease.
A complete list of the PROGRESS Collaborative group has been previously published (21). The Management Committee is as follows: J. Chalmers, S. MacMahon, C. Anderson, M. G. Bousser, J. Cutler, S. Davis, G. Donnan, L. Hansson (deceased), S. Harrap, K. Lees, L. Liu, G. Mancia, B. Neal, T. Omae, J. Reid, A. Rodgers, R. Sega, A. Terent, C. Tzourio, and M. Woodward. The Endpoint Adjudication Committee is as follows: G. Donnan (chair), N. Anderson, C. Bladin, B. Chambers, G. Gordon, and N. Sharpe. Statistical Analysis is as follows: S. Colman, A. Lee, and M. Woodward. The study was designed, conducted, analyzed, and interpreted by the investigators independently of all sponsors.
Supported by grants from Servier, the Health Research Council of New Zealand, the National Health and Medical Research Council of Australia, and an overseas research fellowship from the Japanese Heart Foundation for the current analysis (T.O.). Conflict of Interest Statement: T.O. has no declared conflict of interest; N.C. has no declared conflict of interest; B.N. has been paid honoraria for lectures delivered at symposia arranged by Bayer, Pfizer, and Servier (total $33,255) during the past three years and over this time has received support for a meeting of the Blood Pressure Lowering Treatment Trialists Collaboration from Bayer, Boehringer Ingelheim, Merck, Pfizer, Servier, and Solvay (total US $180,000) and support from Pfizer for a new trial of an Internet-based strategy for cholesterol management (A$125,000); M.W. received $2,000 for an invited lecture financed by Servier; T.O. has no declared conflict of interest; J.C. has received honoraria for $11,000 in 2001, $15,000 in 2002, and $10,000 in 2003 for speaking at conferences sponsored by Servier and research grants were provided by Servier for two trials for which he is Co-Principal Investigator, for the PROGRESS trial, directly to the University of Auckland (Uniservices), with annual grants in excess of $1 million from 1995 to 2000, and for the ADVANCE study, directly to the University of Sydney, with annual grants in excess of $3 million from 2000 to 2003. Received in original form September 3, 2003; accepted in final form February 20, 2004
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
65 years of age). Homogeneity of treatment effects between subgroups was tested by adding an interaction term to the relevant statistical model. 
