Kaplan-Meier survival curves. Curves were analyzed according to prescription of an angiotensin-converting enzyme inhibitor (ACEI) alone (group 1), an ACEI combined with aspirin at a low dose (≤160 mg) (group 2), or an ACEI combined with aspirin at a high dose (≥325 mg) (group 3). Differences between groups 2 and 1 were not significant (P = .19). On the contrary, differences between group 3 and groups 1 and 2 were significant (log-rank χ22 = 8.47, P = .009).
Guazzi M, Brambilla R, Rèina G, Tumminello G, Guazzi MD. Aspirin–Angiotensin-Converting Enzyme Inhibitor Coadministration and Mortality in Patients With Heart FailureA Dose-Related Adverse Effect of Aspirin. Arch Intern Med. 2003;163(13):1574-1579. doi:10.1001/archinte.163.13.1574
It is debated whether in patients with chronic heart failure (CHF), aspirin may contrast the clinical benefits of angiotensin-converting enzyme inhibitors (ACEIs). Two major unresolved issues in patients with CHF are whether these agents together can affect mortality and whether the interaction is related with the dose of aspirin. We aimed at exploring these possibilities.
We evaluated more than 4000 hospitalizations with a principal discharge diagnosis of CHF from January 10, 1990, to December 31, 1999. The final analysis was restricted to 344 patients taking ACEIs who satisfied the selection criteria, in whom reliable information was available concerning drug therapy during follow-up. In these patients, treatment included no aspirin in 235 (group 1), a low dose (≤160 mg) in 45 (group 2), and a high dose (≥325 mg) in 64 (group 3).
During a mean follow-up of 37.6 months, there were 84 (36%) deaths in group 1, 15 (33%) in group 2, and 35 (55%) in group 3. By the Kaplan-Meier approach, survival was similar in groups 1 and 2, and significantly (P = .009) worse in group 3 compared with groups 1 and 2. After adjusting for potential confounding factors (including treatment, cause of heart disease, age, smoking, and diabetes mellitus), a time-dependent multivariate Cox proportional hazards regression analysis showed that the combination of high-dose aspirin with an ACEI was independently associated with the risk of death (hazard ratio, 1.03; P = .01) and that the combination of low-dose aspirin with an ACEI was not (hazard ratio, 1.02; P = .18).
These results support the possibility that in some patients with CHF who are taking an ACEI, a dose-related effect of aspirin may adversely affect survival.
IT IS BELIEVED that inhibition of prostaglandin synthesis is the primary mechanism by which aspirin may contrast the clinical benefits of angiotensin-converting enzyme inhibitors (ACEIs),1 which in part rely on these eicosanoids to mediate various pharmacological effects.2 The clinical potential for this detrimental interplay is significant in patients with 1 or more cardiovascular diseases, such as heart failure caused by ischemic heart disease.1,3 The influences of the combination of aspirin and an ACEI on hemodynamics, renal function, pulmonary dynamics, and ventricular remodeling in patients with chronic heart failure (CHF) have been examined in several clinical studies,4 but the most crucial question involves the effect that these agents together have on mortality rate.4 In patients with coronary artery disease taking an ACEI, several studies5- 8 have documented an adverse effect of aspirin, whereas others9- 12 did not. A difference in the dosage used might be a reason for the discrepancy, even if in most trials the aspirin dosage is not reported. A pertinent question is whether the clinical problem may be circumvented with an aspirin-based compromise using low doses that preferentially inhibit platelet thromboxane A2synthesis without appreciable effects on prostaglandin production.13 Studies10,14- 16 evaluating low-dose aspirin generally showed a low propensity to produce adverse effects, but not all17,18 are in agreement; and it is basically unproved that the interaction is dose related.
According to these premises, the main purposes of the study were (1) to evaluate the effect and interaction of aspirin and an ACEI on mortality among patients with CHF of any origin, in whom treatment with both medications was considered appropriate, and (2) to explore the hypothesis that the effect of aspirin may be related to the dose used.
Patients with a principal discharge diagnosis of CHF at the Institute of Cardiology, University of Milan, from January 10, 1990, to December 31, 1999, were selected for review. To perform this study, we evaluated more than 4000 hospitalizations. We obtained demographic and clinical information relating to these hospitalizations from the combined use of administrative and clinical databases and medical record reviews. Sampling was conducted during a 6-month period. We restricted the sample to patients whose antifailure therapy at discharge (all discharge medications were reviewed) included an ACEI (mostly enalapril maleate [mean ± SD daily dose, 17.5 ± 5.5 mg]; and captopril [mean ± SD daily dose, 62.5 ± 12.0 mg]), associated or not with aspirin, who were 70 years or younger and had cardiac dysfunction causing heart failure confirmed at admission by symptoms, radiographic findings, and a left ventricular ejection fraction of 35% or less. We excluded patients who were taking warfarin sodium and who were considered to have a terminal illness or metastatic cancer because the goals of their treatment may not have focused on a survival benefit. To decrease the heterogeneity of the sample, we excluded patients with severe aortic and mitral valve stenosis or heart failure secondary to an acute illness. To examine the effects of an ACEI and aspirin in a cohort of patients who would be expected to benefit from these therapies, we further restricted the sample of patients who had documented indications or no documented contraindications for an ACEI or aspirin. This was a group of patients for whom the guidelines would recommend these compounds, and who would be eligible for enrollment in a clinical trial of these medications. Thus, we also excluded subjects with ACEI intolerance, a systolic blood pressure lower than 105 mm Hg at discharge, current or past episodes of bleeding, creatinine levels greater than 2 mg/dL (177 µmol/L), and documented intolerance to aspirin. To discern whether the effects of aspirin were related to the dose used, we only included patients taking aspirin at a daily dose of 160 mg or less or 325 mg or more. The exclusion criteria resulted in a final study sample of 412 patients. The final analysis was further restricted to 344 patients, because of missing reliable information for 47 (concerning drug therapy during follow-up) or because of changes in medication for 21. Data reliability was tested by random reabstraction, with overall variable agreement averaging 93%.
Potential confounding factors were examined, including sex, age, body surface area, smoking, comorbidity (hypertension, diabetes mellitus, previous myocardial infarction, coronary bypass surgery, or coronary angioplasty), biochemistry findings (serum creatinine, urea nitrogen, sodium, and glucose concentrations), atrial fibrillation, ejection fraction, preadmission use of aspirin and/or an ACEI, and discharge prescriptions (β-blockers, digoxin, furosemide, amiodarone, nitrates, or spironolactone [Aldactone], in addition to an ACEI alone or with aspirin).
The outcome variable of the study was death from any cause during a maximum follow-up of 10 years (the average follow-up was 3.1 years). To assess vital status, we resorted to the combined use of administrative and clinical databases; records during readmission and outpatient records during follow-up (most patients attended our outpatient clinic) were reviewed (for patients who were hospitalized more than once in the examined period, only the first admission was included). When vital status could not be determined from these methods, patients or their families were interviewed by telephone.
The evaluation of ejection fraction and left ventricular volume was based on 1 of these 2 methods: contrast ventriculography or ultrasonography (2-dimensional echocardiography by using the Simpson rule). Priority was in this order for measurements performed with more than 1 method.
The effect of 2 dosages (low [≤160 mg/d] and high [≥325 mg/d]) of aspirin and of an ACEI on mortality after discharge was considered. Patients were classified, according to their medication prescription at discharge, into 3 groups: an ACEI only (group 1), an ACEI combined with aspirin at a low dose (group 2), and an ACEI combined with aspirin at a high dose (group 3). Differences between groups were assessed by using χ2 (for discrete variables) and analysis of variance. Variables possibly related to survival were analyzed by the Kaplan-Meier approach with log-rank testing for equality of survivor functions and by univariate Cox proportional hazards regression analysis. A multivariate time-dependent Cox proportional hazards regression model was used to assess the joint relevance of treatment and of other possible predictors (sex, age, smoking, comorbidity, biochemistry findings, atrial fibrillation, ejection fraction, preadmission use of aspirin and/or an ACEI, and discharge prescriptions). Only the variables significant in the univariate approach were included in the model, namely, treatment and age. Age and treatment were entered into the model as dummy variables to test their effect with respect to their reference category, as follows. For age, when 0 was entered for the low and high doses, the age was younger than 58 years (33rd percentile) (reference); when 0 was entered for the low dose and 1 for the high dose, the age was 58 to 68 years (33rd-66th percentile); and when 1 was entered for the low dose and 0 for the high dose, the age was older than 68 years (66th percentile). For treatment, when 0 was entered for the low and high doses, the treatment was an ACEI only (reference); when 0 was entered for the low dose and 1 for the high dose, the treatment was an ACEI and an aspirin dose of 160 mg or less; and when 1 was entered for the low dose and 0 for the high dose, the treatment was an ACEI and an aspirin dose of 325 mg or more.
The 2 treatment dummy variables were also entered into the Cox proportional hazards regression model as time-dependent variables, as interactions with time. The proportional hazard assumption and the goodness of fit were assessed by the Schoenfeld residuals and the likelihood ratio test, respectively.
All P values reported were 2-sided, and P<.05 was considered significant. Statistical analyses were performed by computer software (Stata 7.0 package).
A total of 344 patients were included in this analysis. They were classified into 3 groups, based on the use of an ACEI alone (group 1 [n = 235]), an ACEI with aspirin at a daily dose of 160 mg or less (group 2 [n = 45]), or an ACEI with aspirin at a daily dose of 325 mg or more (group 3 [n = 64]). Associations between demographic and clinical characteristics and patients in the 3 medication groups are shown in Table 1. About 31% of the patients had had a previous myocardial infarction, 25% were current smokers, and hypertension and atrial fibrillation were comorbidities in 52% and 28%, respectively; serum sodium level and renal function were within normal limits; and the average fasting blood glucose level was somewhat increased because of the presence of type 2 diabetes mellitus in 13% of the population. Compared with patients who were prescribed an ACEI only, those prescribed the aspirin combination were similar in age, somatic characteristics, and ejection fraction; received similar doses of ACEI; and were more likely to have experienced a myocardial infarction or to have undergone percutaneous transluminal coronary angioplasty and to have had or to have atrial fibrillation. They were also more likely to be prescribed amiodarone and were similarly likely to be prescribed a β-blocker at discharge. Differences between the group receiving high-dose aspirin and the group receiving low-dose aspirin were not significant (P = .15). Patients receiving an ACEI and aspirin also had a definitely higher preadmission prescription of aspirin.
After an average follow-up of 37.6 months, there were 134 (39%) deaths, of which 84 (36%) were in group 1, 15 (33%) were in group 2, and 35 (55%) were in group 3 (Table 2). For those prescribed low-dose aspirin and an ACEI together, compared with an ACEI alone, the survival differences were not statistically significant (P = .19). In contrast, aspirin at a high dose coadministered with an ACEI was associated with a significant (P = .009) decline of survival compared with the administration of an ACEI alone or combined with a low dose of aspirin. Figure 1 shows the Kaplan-Meier survival curves of groups 1, 2, and 3.
Table 2 shows that in the 3 groups, there were no significant (P = .13) associations of clinical and demographic data and drug prescription (apart from aspirin and ACEIs) with outcome, with the 2 exceptions of a more likely occurrence of smoking in nonsurvivors in group 3 and of diabetes mellitus in nonsurvivors in group 1.
After adjusting for the potential confounders reported in the "Statistical Analysis" subsection of the "Methods" section, a time-dependent multivariate Cox proportional hazards regression analysis showed that the combination of an ACEI and aspirin at a high dose was independently associated with the risk of death (hazard ratio, 1.03; P = .01). On the contrary, the combination of an ACEI and aspirin at a low dose was not related to mortality (hazard ratio, 1.02; P = .18). These results are shown in Table 3.
The main purpose of the study was to probe whether long-term counteraction of aspirin to an ACEI can result in enhancement in mortality from any cause in patients with CHF, and whether this effect may be dose related. Compared with an ACEI alone, coadministration of these agents was associated with reduced survival in patients in whom high doses of aspirin were used. It remains elusive whether the benefits of inhibition of the platelet function by aspirin19 detracted from the effects of a possible counteraction to ACEIs.
Before this information is accepted, sampling modalities and reliability of the data collected should be critically reviewed. All records came from the same institution and were analyzed by the same trained individuals (M.G., R.B., and G.T.), eligibility was determined in a specific appropriate reference time, and strategies to decrease abstraction errors and variability included training sessions and detailed data definitions. Random reabstractions were performed to monitor data reliability; patients in whom we were unable to determine discharge medications or medications during follow-up were excluded. Most patients had preadmission prescription of the test drugs, what presumably contributed to improve compliance. We restricted the sample to patients who had documented indications and no documented contraindications for both agents, in whom guidelines would recommend them and who would, thus, be eligible for enrollment in a clinical trial of these medications, if one were to be performed.
Why, compared with group 1, was mortality increased in group 3 and not in group 2, despite an aspirin prescription in both? Sex, age, biochemistry findings, left ventricular ejection fraction, and end diastolic volume were comparable in the 3 groups. In group 2, despite a greater incidence of myocardial infarction, percutaneous transluminal coronary angioplasty, and diabetes mellitus, the outcome was similar to that in group 1. A significantly higher percentage of these patients had atrial fibrillation and were taking amiodarone. Thus, one may wonder whether aspirin in group 2 looks nonharmful because of a decreased incidence of sudden death because of the administration of amiodarone. Our information on the mode of death was not detailed enough to clarify this point. Such a possibility, however, is not supported by findings in patients in group 3. Despite the facts that these patients had an incidence of myocardial infarction, percutaneous transluminal coronary angioplasty, and atrial fibrillation and a prescription of amiodarone similar to those in group 2, they had a worse outcome. Therefore, the cause of heart failure did not offer a preferential substrate for the results in this study.20
The incidence of smoking was higher in nonsurvivors in group 3 than in the other groups, which might imply that smoking may facilitate an adverse aspirin × ACEI interaction, oppose the antiplatelet effect of aspirin, or affect prognosis per se. Some considerations are in order. The percentage of current smokers among nonsurvivors in group 2 was comparable, despite the prescription of aspirin, to that in nonsurvivors in group 1 (Table 2) (thus, denying that smoking facilitated an aspirin counteraction to ACEIs or opposed the antiplatelet effect of aspirin); the incidence of smoking in group 3 was comparable to that in the other 2 groups (Table 1) (because of this, the hypothesis that smoking might be the main determinant of a higher mortality in this group would imply the unlikely possibility that for some unknown reason the harm of smoking was greater in group 3 than in groups 1 and 2); and finally, smoking was accounted for as a potential confounding factor in this study.
More consistent with our results is the interpretation that an increase in mortality may be in some way related with an adverse influence of high doses of aspirin.
What are the mechanisms of the aspirin activity and what are the differences between low and high doses? The use of aspirin in patients with CHF has been controversial.1,4,8,10,18,21,22 In fact, the evidence in patients with CHF of enhanced von Willebrand factor–mediated platelet thrombus formation and other alterations of coagulation4 suggests that the action of aspirin could be of particular benefit in these patients, mainly in the presence of ventricular enlargement, atrial enlargement, and fibrillation. On the other hand, physiology-based studies have provided a rationale for a potential harm of prescribing prostaglandin inhibitors, like aspirin, in patients with CHF. Recent heart failure guidelines23 warn about the combined use of antiplatelet agents and diuretics, because the effects on renal function of the former can blunt the natriuretic effects of the latter. A mechanism like this as a cause of the adverse outcome with aspirin in this study is contradicted by the fact that low doses of aspirin did not affect survival even if low doses have been shown to disturb renal function,11 particularly in those with advanced heart failure.18 In the setting of an aspirin counteraction to ACEIs, a more likely possibility is that aspirin inhibits the compounded eicosanoid-mediated activities of the ACEIs, such as the regulation of vascular tone (an important counterregulatory pathway in these patients) in systemic and renal circulation1; the prevention of hyponatremia,1 left ventricular dysfunction,10 and remodeling4; the modulation of the neuroendocrine activation24; and the facilitation of respiratory gas transfer across the alveolar-capillary membrane.25 It is reasonably presumable that aspirin and ACEIs interact at multiple levels, with agonism and antagonism of variable intensity, depending on their concentrations and the variable preponderance of the various types of prostaglandins. The mechanisms implicated in survival are unknown. Previous studies26,27 have demonstrated a direct relationship between aspirin and ACEI interaction and prognostic indicators. Short-term high aspirin doses, when given to enalapril-treated patients and not when given alone, significantly adversely affect peak exercise oxygen consumption and the slope of increase of minute ventilation with respect to carbon dioxide production,27 2 variables that powerfully predict prognosis in patients with CHF.28- 30 On the other hand, in a study by Baur et al,31 low doses of aspirin administered to patients receiving enalapril failed to reverse the enalapril-mediated improvement in maximal oxygen uptake. The reasons for a lower propensity of low-dose aspirin to interact with ACEIs,20 and to affect prognostic indicators and mortality, remain elusive. Yet, the important issue could not be explored of whether prescription of higher doses of ACEIs32 can prevent the occurrence of an interaction.
An age cutoff of 70 years was used as an inclusion criterion. This would, on one side, reduce the influence of aging on the outcome, but, on the other side, it would make findings in this study only applicable to a relatively young cohort of patients: many patients with CHF are older than 70 years. Because we aimed at reviewing patients without contraindications to both agents,18 we excluded patients with excessively high creatinine levels. This, again, is not typically what one observes in patients with advanced CHF who are taking high doses of diuretics. This study also has the limitations common to cohort studies, including its retrospective nature and lack of randomization. We also did not have the opportunity to assess a wider range of outcomes. However, it has the strengths of a long follow-up, prospective definitions of end point and collection of data, and rigorous population selection.
The results are in favor of a dose-dependent counteraction of aspirin to ACEIs that may affect survival in some patients with CHF. The exact mechanisms linking interaction with mortality remain unclear.
Corresponding author and reprints: Maurizio D. Guazzi, MD, PhD, Istituto di Cardiologia, Università degli Studi di Milano, Via Parea 4, 20138 Milano, Italy (e-mail: firstname.lastname@example.org).
Accepted for publication September 19, 2002.
This study was supported by the Ministry of Health, Rome, Italy; and the Luigi Berlusconi Foundation, Milan, Italy.
This study was presented in part at the American Heart Association Scientific Sessions, November 13, 2001, Anaheim, Calif.