Relationship Between Angina Pectoris and Outcomes in Patients With Heart Failure and Reduced Ejection Fraction

An Analysis of the Controlled Rosuvastatin Multinational Trial in Heart Failure (CORONA)

Athar A. Badar; Ana Cristina Perez-Moreno; Pardeep S. Jhund; Chih M. Wong; Nathaniel M. Hawkins; John G.F. Cleland; Dirk J. van Veldhuisen; John Wikstrand; John Kjekshus; Hans Wedel; Stuart Watkins; Roy S. Gardner; Mark C. Petrie; John J.V. McMurray


Eur Heart J. 2015;35(48):3426-3433. 

In This Article


Local ethics committees from participating sites approved the trial and all patients provided informed consent. A total of 5011 patients aged ≥60 years with symptomatic [New York Heart Association (NYHA) class II–IV], systolic (LVEF ≤40%, but ≤35% if NYHA class II) HF of ischaemic aetiology were enrolled. Patients were randomized to receive 10 mg of rosuvastatin or matching placebo daily. The median follow-up was 32.8 months and rosuvastatin did not reduce the risk of the primary composite outcome of death from cardiovascular (CV) causes, non-fatal MI, or non-fatal stroke.[6,7]

Investigators recorded a history of 'past or current angina' in the case report form at the time of enrolment using a checkbox system. Chest pain 'during the past few days' was also recorded and measured using a five-point exertion scale (0 no pain, 1 pain on heavy exertion, 2 on moderate exertion, 3 on slight exertion, and 4 on rest). While not specifically validated for this purpose, in our analysis, current chest pain was considered to reflect ongoing angina. Patients were classified into three mutually exclusive groups: those with no history of angina and no current chest pain at baseline (Group A), those with a past history of angina but no current chest pain at baseline (Group B) and those with a history of angina and current chest pain at baseline (Group C). Patients without a history of 'past or current angina' but who described chest pain at baseline, were excluded from this analysis (n = 133, 2.7%).

Clinical Outcomes

The outcomes analysed included the pre-defined secondary 'coronary event' endpoint in CORONA—the composite of sudden death, fatal or non-fatal MI, percutaneous coronary intervention (PCI), coronary artery bypass graft surgery (CABG), ventricular defibrillation by implantable cardioverter defibrillator (ICD), resuscitation after cardiac arrest or hospitalization for unstable angina (UA). We also examined the more restricted composite of non-fatal MI, UA, PCI, or CABG so as to exclude sudden death and ventricular defibrillation by an ICD which may not, in many cases, reflect myocardial ischaemia or infarction. In addition, we analysed the composite of only non-fatal MI and UA as coronary revascularization may reflect physician preference and practice as much as disease activity. We also examined the composite of CV death or HF hospitalization, as the most commonly used measure of HF-related mortality and morbidity, and all-cause mortality.

Statistical Analysis

Baseline characteristics for each group were presented as means and standard deviations for continuous variables (or medians for data not normally distributed) and percentages for categorical variables. One-way analysis of variance was used to compare continuous variables across groups and the χ2 test was used for categorical variables.

The relationship between history of 'past or current angina', with or without chest pain, at baseline and the endpoints described above was evaluated using Cox proportional-hazard models time-to-first event regression analyses, i.e. outcomes in Groups A and B were compared, as were those in Groups A and C. Both unadjusted analyses and adjusted analyses were carried out. The adjusted analyses used a previously published CORONA risk-model; the prognostically significant variables in that model were: age, sex, NYHA class, LVEF, body mass index (BMI), systolic blood pressure, heart rate, smoking, MI, CABG or PCI, aortic aneurysm, hypertension, diabetes, baseline atrial fibrillation/flutter (AF/F), stroke, intermittent claudication, pacemaker and ICD implantations, ApoA-1, ApoB, creatinine, alanine aminotransferase, creatine kinase, thyroid stimulating hormone, high-sensitivity C-reactive protein, and log NT-proBNP.[8] Data on these predictor variables were missing in 1% or less of patients except for history of CABG (9%) and NT proBNP (27%); patients with missing variables were omitted from analyses. Adjustment was also made for differences in baseline medications, namely loop or thiazide diuretics, aldosterone antagonists, ACE inhibitors, or angiotensin receptor blockers (ARBs), beta-blockers, digitalis glycoside, anti-arrhythmic therapy, antiplatelet or anticoagulant drugs, as well as for randomization to rosuvastatin or placebo. Linearity was assessed visually by plotting the data graphically and more formally by performing a Wald test. Proportionality assumptions were verified using the Schoenfeld residuals method. Kaplan–Meier cumulative event curves were presented by symptom category for fatal outcomes and composites with fatal and non-fatal components. The cumulative incidence function was used to analyse non-fatal outcomes to account for the competing risk of death. Finally, we examined the association between chest pain and outcomes with chest pain entered as a time-updated covariate in a Cox model. We adjusted for a history of angina and then all of the covariates noted above. A Bonferonni correction was used to correct for the multiple comparisons in relation to the coronary outcomes. A two-tailed P-value of <0.05 was considered statistically significant.