Design and Methodology of the CAPRICORN Trial — A Randomised Double Blind Placebo Controlled Study of the Impact of Carvedilol on Morbidity and Mortality in Patients with Left Ventricular Dysfunction After Myocardial Infarction

Acute myocardial infarction is now recognised as an important stage in the ‘cardiovascular continuum’ that describes the progression of cardiovascular disease from risk factors to end stage heart failure. Despite the advent of reperfusion and secondary preventive strategies over the past 15 years, the 30-day mortality rate of acute myocardial infarction remains high at 15–25% 1–3. A further 5–10% of patients die in the 1st year after infarction and mortality continues at a rate of 5% per annum thereafter 4,5. The recognition of this serious acute major coronary event is, therefore, an important opportunity to intervene in order to prevent progression to and of left ventricular dysfunction with subsequent morbidity and mortality 6. However, despite 22 randomised, placebo controlled trials of beta blockade which collectively demonstrate a 23% relative reduction in mortality, recent studies from both Europe and the United States indicate that only 30–40% of post-MI patients can expect to receive a beta-blocker 4,7,8. There are a number of possible explanations as to why this should be the case. Firstly, elderly patients were largely excluded from the majority of these trials leaving clinicians in some doubt as to the applicability of their findings to this common patient group. Secondly, the trials were all performed in the ‘pre thrombolytic’ era, and their results 9 are not necessarily applicable to the majority of post-MI patients seen nowadays in everyday clinical practice. Thirdly, the population of patients enrolled in these trials were relatively ‘low-risk’ in terms of the presence of risk factors for further major cardiovascular events and early mortality. For example, the annual mortality of those who survive to reach hospital should, according to epidemiological studies, be approximately 20–35% for an unselected population of patients. In contrast, the annualised mortality rates of the first ‘landmark’ post-MI beta-blocker trial, the Norwegian timolol study 10 and the subsequent much larger BHAT trial of propranolol 11 were 10% and 6%, respectively. Moreover, although neither trial formally excluded patients with a history of heart failure during the acute event, even the heart failure subgroup in BHAT had a 2-year mortality of only 18%. This is in contrast to the 1-year mortality of 46% in a large contemporary study of consecutive patients admitted to a coronary care unit in London 1 who developed heart failure, suggesting that the patient population included in BHAT had rather mild impairment of ventricular function. Thus, although post hoc analyses have been suggestive, the hypothesis that it is in patients with heart failure or documented evidence of left ventricular dysfunction that beta-blockers might be most beneficial has never been tested in an appropriately designed randomised controlled trial. On the contrary, the role of ACE inhibitors in post-MI therapy has been quite clearly defined, and this has contributed to its widespread use in the fairly short time since the clinical trials were carried out 12–16. Those trials also demonstrated that the best way of identifying a true high risk post-MI population is not by symptoms or ECG criteria alone, but by cardiac imaging to demonstrate definite LV systolic dysfunction. The renewed interest in beta-blocker therapy post-MI, and in particular for patients with LV impairment, is largely due to the recent demonstration of substantial clinical benefit in terms of outcome in patients with chronic heart failure 17–20. Enthusiasm has been fuelled also by the growing consensus concerning the pathophysiological importance of neuroendocrine activation in the development and progression of heart failure. This hypothesis has been supported by the ACE inhibitor trials, the beta-blocker trials in heart failure and most recently, by the beneficial effects of aldosterone antagonism by spironolactone 21. There is considerable evidence for neuroendocrine activation following myocardial infarction 22–24. Over-activation of the renin angiotensin and sympathetic nervous systems individually or in concert can be detrimental in a number of ways following acute myocardial infarction. In haemodynamic terms an increased heart rate and increased systemic vascular resistance have significant and immediate deleterious effects on cardiac function. Probably more importantly, however, are the secondary, longer term effects on structure and function that promote the process of cardiac remodelling. This process, if unchecked, results in myocyte hypertrophy, necrosis and apoptosis, fibrosis and dilatation leading to systolic and diastolic dysfunction, arrhythmogenesis, progressive cardiac dysfunction and ultimately heart failure and death. This background was the rational for the CAPRICORN study (CArvedilol Post-infaRct survIval COntRol in LV dysfunctioN). Carvedilol is a third generation non-selective beta-adrenergic receptor antagonist with α1-receptor blocking and anti-oxidant activity. It has been licensed for angina and hypertension for several years and more recently was the first beta-blocker to receive regulatory approval for heart failure therapy 25–32. The main evidence for efficacy in heart failure has come from the US Carvedilol Trials 17, which suggest a substantial reduction in mortality and morbidity when carvedilol is added to standard therapy in patients with stable chronic heart failure. The Australian and New Zealand Group trial and other trials 33–39, have also demonstrated a marked improvement in LV dimensions and function, and a corresponding improvement in symptoms and reduction in hospital admissions. Most recently the COPERNICUS trial of carvedilol in patients with severe heart failure 40 (i.e. patients with symptoms on rest or minimal exertion) was terminated prematurely due to a marked and highly significant reduction in mortality in patients on carvedilol. Thus carvedilol improves both morbidity and/or mortality in patients with all grades of stable heart failure. Carvedilol is therefore a suitable agent to test the hypothesis that comprehensive adrenergic blockade via beta-1 and -2 and alpha-adrenoceptor antagonism, in higher risk patients with left ventricular systolic dysfunction following acute myocardial infarction will result in a significant improvement in outcome. The objective of the CAPRICORN trial was to investigate the effects of carvedilol on morbidity and mortality in patients with documented evidence of left ventricular dysfunction with or without heart failure after an acute myocardial infarction. All cause mortality or cardiovascular hospitalisation. All cause mortality. Sudden death. Hospitalisation for heart failure. Exploratory analyses included cardiovascular mortality, all-cause hospitalisations, cardiovascular hospitalisations, unstable ischaemic events, non-fatal cardiovascular events. Confirmed MI occurring within the previous 21 days. Left ventricular ejection fraction ≤40% (directly by 2D echocardiography, radionuclide or contrast ventriculography) or indirectly by wall motion score index ≤1.3. Concurrent treatment with ACE inhibitor for > 48 h with the dose being stable for >24 h, unless proven intolerance of ACE inhibitors. Continued requirement for IV inotropic therapy or uncontrolled heart failure. Ongoing or expected need for beta-blockade. Complicating clinical conditions including unstable angina, uncorrected significant valve disease, hypotension 50 b.p.m., BP>80 mmHg systolic, the dose was increased to the next level. Trial medication was administered in the out patient department and the patient asked to remain for 2 h to ensure no significant side effects occurred. When the maximum tolerated dose was achieved they entered the maintenance phase. During the up titration phase, which lasted approximately 4–6 weeks, the dose of ACE inhibitor was not altered although other medication, for example diuretic therapy, could be adjusted. Patients were reviewed at 3-month intervals in the 1st year and 4-month intervals thereafter. At specified visits the ECG was recorded, NYHA class ascertained and venous blood for routine biochemistry and haematology tests taken. Investigators were encouraged to review the dose of study medication at each visit in order to ensure that each patient was receiving the maximum tolerated dose. Thus the dose of study medication was adjusted during the maintenance phase, either up or down as indicated. Sudden cessation of therapy was avoided to reduce the likelihood of a rebound phenomenon. Importantly, the doses of other drugs, especially ACE inhibitors, were also adjusted during this phase to ensure optimal dose levels. The maintenance phase continued until 633 validated primary end points occurred (all cause mortality or cardiovascular hospitalisation), at which point the down-titration phase began. It was planned that each patient should be in the trial for at least 3 months following randomisation before down titration commenced. Interim history and adverse experiences. Cardiopulmonary examination. Recording of all endpoints. Recording of all changes in cardiovascular medication. Resource utilisation. Sitting blood pressure, heart rate and body weight. Biochemistry and haematology — if abnormal at previous visit. At the end of the maintenance phase study medication was withdrawn in a step-wise manner over a period of 1–2 weeks by decreasing one dose level at a time every 3–4 days until the patient was off study medication completely. Subsequent use of open label beta-blockade was at the discretion of each investigator. The target sample size was based on the first primary end point ‘all cause mortality or cardiovascular hospitalisations’ with an alpha level of 0.045 a power of 90% and assuming a hazard ratio of 0.77. The second co primary end point ‘all cause mortality’ will be tested with an alpha level of 0.005. Using the assumptions detailed in the protocol it was calculated that recruitment of 1850 patients (925 patients per treatment arm) with 633 primary end points, deaths or cardiovascular hospitalisations, would be required. Following the intention to treat principle all analyses will be performed on all randomised patients. The primary efficacy parameters will be evaluated using the log-rank test with a significance level of 0.045 for the first primary parameter (all-cause mortality or cardiovascular hospitalisation) and with a significance level of 0.005 for the second primary parameter (all-cause mortality). All other statistical tests will be performed at the 5% level of significance and regarded as exploratory. For all time to event analyses hazard ratio and associated 95% confidence limits will be presented. The safety parameters include ECG vital signs (HR, BP), laboratory parameters and adverse experiences. Planned interim analyses. The Data and Safety Monitoring Board (DSMB) received detailed reports on key safety variables, from an independent statistical centre, on a regular basis during the trial. The DSMB carried out one formal interim analysis on the outcome of all cause mortality after 125 deaths had accrued. Following a recommendation from the DSMB based on emergent results from trials that were ongoing at the time of the start of CAPRICORN the steering committee adopted a new policy for the use of open label beta blockers. Patients who continued to have symptoms of heart failure despite optimisation of background medication with appropriate doses of ACE inhibitors and diuretics or digoxin and who were clinically stable could be started on open label beta-blocker treatment at the discretion of the investigator. Study medication had to be down titrated prior to open label beta-blocker use and the study blind not broken for these patients. According to the intent to treat analyses all such patients were followed according to the study procedures until the official end of the study. Professor Henry Dargie, Chairman and Principal Investigator (Europe), Professor Dr Wilson Colucci, Principal Investigator (North America), Professor Norman Sharpe, Principal Investigator (Australasia), Professor Ian Ford, Professor Dr Jose Luis Lopez-Sendon, Professor Willem Remme. The Executive Steering Committee (ESC) was responsible for the development of the protocol in collaboration with the sponsors. Initially the sole sponsor was SmithKline Beecham but as a result of new commercial arrangements Roche assumed responsibility for all centres worldwide excluding the US which remained the responsibility of SmithKline Beecham. Thereafter protocol amendments were discussed jointly with both sponsors. The ESC were charged with assessing and approving substudy proposals, reviewing the conduct of the study, helping to identify problems with recruitment and assisting in formulating solutions. The ESC also had sanction on the endpoint definitions and process for validation produced by the Endpoints Committee. In collaboration with the sponsors the ESC will plan and implement presentation and publication of study details and results. These functions of the ESC were the subject of a written charter agreed with the sponsors. Mortality endpoints: The EC classified all deaths as cardiovascular or non-cardiovascular, and within the cardiovascular group, as being sudden, due to MI, heart failure, stroke, procedural or other cause. Morbidity endpoints: All hospitalisations will be classified as non-cardiovascular or cardiovascular. Further sub-classifications will occur for hospitalisations for angina, chest pain, myocardial infarction, heart failure, stroke and TIA. The end point committee produced working definitions for the purposes of classification which were approved by ESC. Membership was Professor Desmond Julian, (Chairman), Professor Barrie Massie, Professor Simon Thompson and Professor Lars Wilhelmson. The Board were charged with supervising all safety aspects of the study and had the authority to recommend termination of the study. The sole criterion to be used for these analyses was the endpoint of all cause mortality. Statistical significance at P<0.001 (two-sided log-rank test) was required (Peto-type rule) to show evidence of benefit before early stopping would have been considered. Particular caution was used at the first interim analysis because of the small number of events that had accumulated. Statistical significance at P<0.016 (two-sided log rank test) was required to show evidence of harm before early stopping should have been considered. This maintained an overall type I error rate of 5% (Pocock-type rule). Should a value of P<0.016 have occurred on any occasion, other aspects of the data would have been considered prior to any recommendation. This was at the Robertson Centre for Biostatistics at the University of Glasgow, Glasgow and comprised: Professor Ian Ford, Michele Robertson, Jan Love and Adele Trainer.