The Journal of Informed Pharmacotherapy 2004;15:202.
Reviewer: Glen J. Pearson, BScPhm, PharmD, FCSHP
Reviewer’s e-mail address: GPearson@cha.ab.ca
Reviewer's profession/specialty: Cardiology/Cardiac Transplantation/CV Risk Reduction
Sever PS, Dahlöf B, Poulter NR, Wedel H, Beevers G, Caulfield M, Collins R, Kjeldsen SE, Kritinsson A, McInnes GT, Mehlsen J, Nieminen M, O’Brien E, Östergren J, for the ASCOT Investigators. Prevention of coronary and stroke events with atorvastatin in hypertensive patients who have average or lower-than-average cholesterol concentrations, in the Anglo-Scandinavian Cardiac Outcomes Trial – Lipid Lowering Arm (ASCOT-LLA): a multicentre randomized controlled trial. Lancet 2003;361:1149-58. PubMed Citation
The Anglo-Scandinavian Cardiac Outcomes Trial (ASCOT) was a large multicentre, international trial, which involved two treatment comparisons in a two-by-two factorial design. (1) The first was a prospective, randomized, open, blinded endpoint (PROBE) design comparing two antihypertensive regimens (anticipated to be completed in early 2005).The second was a prospective, double-blind randomized comparison of the cardiovascular effects of the lipid-lowering agent atorvastatin 10 mg daily compared to placebo among patients who had non-fasting total cholesterol concentrations of ≤6.5 mmol/L. It is the lipid-lowering arm of this trial (ASCOT-LLA) that is the focus of this publication.
Men and women between the ages of 40 and 79 years at the time of randomization were eligible if they were hypertensive according to the study definitions and had at least three pre-specified cardiovascular risk factors. Subjects with untreated hypertension (not already receiving antihypertensive medication) were eligible if they had either a systolic blood pressure (SBP) ≥160 mmHg and/or a diastolic blood pressure (DBP) ≥100 mmHg, while those with treated hypertension (already receiving antihypertensive medication) were eligible if they had SBP ≥140 mmHg and/or DBP ≥90 mmHg. In addition, the study population was required to have at least three of the following risk factors for cardiovascular disease: non-insulin dependent diabetes mellitus (NIDDM), peripheral arterial disease (PAD), previous stroke or transient ischemic attack (TIA), left-ventricular hypertrophy (LVH), other specified abnormalities on electrocardiogram (ECG), microalbuminuria or proteinuria, plasma total cholesterol to HDL-cholesterol ratio ≥6, male gender, age ≥55 years, smoking, or premature history of coronary heart disease in a first degree relative. Patients were excluded if they had any of the following: previous myocardial infarction (MI), currently treated angina, a cerebrovascular event within the previous 3 months, fasting triglycerides >4.5 mmol/L, heart failure, uncontrolled arrhythmias, or any clinically important hematological or biochemical abnormality on routine screening.
Patients fulfilling the criteria for the blood pressure lowering arm of ASCOT, were subsequently eligible for the lipid-lowering arm if they had a non-fasting total cholesterol concentration ≤6.5 mmol/L and were not currently taking a statin or a fibrate. Following an initial study-screening visit, all patients completed a 4-week run-in period in order to allow sufficient time to confirm the patient’s eligibility and obtain informed consent prior to randomization. Also during this time, each patient’s family physician or local investigator had the opportunity to review the non-fasting screening lipid values and consider the need for lipid-lowering treatment. Only patients whose physicians did not intend to treat them with a statin or a fibrate were randomized to treatment with atorvastatin 10 mg daily or matching placebo. In this arm of the study, the investigators set out to determine whether lipid-lowering with a statin provided additional beneficial effects to antihypertensive treatment in hypertensive patients with average or below average levels of serum total cholesterol.
The primary endpoint was the combination of nonfatal MI (symptomatic and the so-called silent MI) plus fatal CHD. The secondary endpoints of interest included each of the following: the primary outcome (non-fatal MI plus fatal CHD) without silent events, all-cause mortality, total CV mortality, fatal and non-fatal stroke, fatal and non-fatal heart failure, total coronary endpoints, and total CV events and procedures. The pre-specified tertiary endpoints included: silent MI, unstable angina, chronic stable angina, PAD, life-threatening arrhythmias (VF or sustained VT or complete heart block), the development of diabetes mellitus or renal impairment, and the primary endpoint among several subgroups. In addition, the investigators planned for an assessment of the effects of treatment on health care costs and on all the major study endpoints among 18 different subgroups of patients (e.g. diabetics, males/females, and previous vascular disease). (2,3)
The principal funding source for ASCOT was Pfizer (New York, NY; USA). In addition, the Servier Research Group (Paris, France) and Leo Laboratories (Copenhagen, Denmark) provided support in the form of study medications. However, the authors indicate that ASCOT was conceived, designed, and coordinated by an investigator-led independent steering committee with members representing all of the countries where the trial was being conducted. While the principal funding source had two members on the steering committee, both had non-voting privileges. In addition, the data analysis and manuscript preparation were conducted independent of the principal funding source. The publication also explicitly outlines the potential conflict of interest for the fourteen authors listed. (2)
1. Was assignment of patients randomized?
Yes. Patients were recruited from February 1998 to May 2000 out of 686 general practices in Denmark, Norway, and Sweden and 32 regional centres to which patients were referred by their family physicians in the United Kingdom and Ireland. From these practices 19,342 patients were randomized to one of the two antihypertensive regimens (10,244 from the Nordic general practices and 9,098 from the UK/Ireland regional centres) with 10,305 of these subjects eligible for ASCOT-LLA and further randomized to receive atorvastatin 10 mg daily or placebo. Randomization to one of the two groups was assigned by computer, with the use of minimization procedures at one of the two coordinating centres in London and Gothenberg. While this article and the previous publication2 describing the rationale and design of the trial do not specifically address the issues of the total number of patients screened for the study, more than 90% of the patients eligible for the lipid-lowering arm were randomized.
2. Were all patients who entered the trial properly accounted for and attributed at its conclusion?
Yes. A total of 10,305 subjects randomized in the lipid-lowering arm, 5,137 to placebo and 5,168 to atorvastatin 10 mg daily. At the close of follow-up for the lipid-lowering arm, complete information was available for 98.8% (10,186) of the patients randomized. For the remainder of subjects with incomplete data, vital status was obtained for all but 17 of them (10 placebo-assigned subjects (0.19%) and 7 atorvastatin-assigned subjects (0.14%) lost to follow-up). Analysis was conducted in accordance with the intention-to-treat principle. All analyses excluded endpoints deemed invalid by the endpoint committee and statistical censoring was enforced at the end of the study (October 1, 2002) or death before that date.
1. How large was the treatment effect?
The study was stopped prematurely upon the recommendation of the data safety monitoring board (DSMB), on the grounds that atorvastatin had resulted in a significant reduction in the primary endpoint and a significant reduction in the incidence of stroke. While the study was designed to follow-up patients for an average of 5 years, the DSMB recommended this arm of the study be discontinued after a median follow-up of only 3.3 years (33,041 patient-years). The intention-to-treat analysis revealed that the primary endpoint of non-fatal MI (including silent MI’s) plus fatal CHD occurred in 1.9% of the atorvastatin group and 3.0% of the placebo group. The unadjusted hazard ratio (HR) for the primary endpoint was 64% (95% CI 50% to 83%; p=0.0005), resulting in a relative risk reduction (RRR) of 36% (95% CI 17% to 50%), an absolute risk reduction (ARR) of 1.1% (95% CI 0.5% to 1.7%), and a number needed to treat (NNT) of 91 patients (95% CI 59 to 200). Statistically significant reductions also occurred in the atorvastatin-treated patients for the following secondary endpoints: fatal and non-fatal stroke (p=0.0236, ARR = 0.7%, NNT = 143), non-fatal MI’s (excluding silent MI) plus fatal CHD (p=0.0005, ARR =1.0%, NNT = 100), total coronary events (p=0.0005, ARR = 1.4%, NNT = 71), as well as total cardiovascular events and procedures (p=0.0005, ARR = 2.0%, NNT = 50). There was no difference in the incidence of all-cause mortality, cardiovascular mortality, or fatal and non-fatal heart failure. Only one of the seven tertiary endpoints was statistically significant between the groups; chronic stable angina occurred in 0.6% of the atorvastatin-treated patients and 1.1% of the placebo patients resulting in an unadjusted HR = 59% (95% CI 38% to 90%; p=0.0135), RRR = 41%, ARR = 0.5%, and a NNT = 200.
It is important to note that blood pressure was similar in both groups (mean values of ~138/80 mmHg) and that an average blood pressure reduction of 25/14 mmHg was achieved in this trial. At the end of the study follow-up, patients treated with atorvastatin experienced relative reductions of 23% in total cholesterol (TC), 33% in LDL-cholesterol, and 22% in triglycerides (TG), compared to the placebo-treated patients who experienced relative reductions of only 5% in TC, 5% in LDL, and 10% in TG. No significant change in HDL-cholesterol occurred in either group.
The proportional effect of atorvastatin on the combined primary endpoint was not significantly different in any of the 18 prespecified subgroups analyzed. However, there was no significant benefit observed in seven of the subgroups, including patients with diabetes and women.
2. How precise was the estimate of the treatment effect?
The confidence interval for the unadjusted HR for combined primary endpoint of non-fatal MI (including silent MI’s) plus fatal CHD [HR=64%; 95% CI 50% to 83%; p=0.0005], as well as the secondary endpoint of fatal and non-fatal stroke [HR=73%; 95% CI 56% to 96%; p=0.0236] are narrow, indicating precision of point estimates of the treatment effect.
2. Were all clinically important outcomes considered?
Yes. All of the outcomes were clearly defined and are clinically important.
Yes. The reductions observed in the in the primary combined endpoint (non-fatal MI plus fatal CHD), as well as the secondary endpoints are both statistically and clinically significant. Atorvastatin was well tolerated in this population, with 87% of patients originally assigned to the active treatment group still taking a statin after 3 years of follow-up. The number of serious adverse events and rates of liver enzyme abnormalities did not differ between the atorvastatin and placebo treated patients. While there was one non-fatal case of rhabdomyolysis in a patient receiving atorvastatin, it was noted that his concurrent high alcohol intake and recent febrile illness might have been a confounding contributor.
Clearly long-term lipid-lowering treatment with a statin is expensive, but the cost savings over time in terms of reduced cardiovascular morbidity and mortality have already been proven. Nevertheless, the cost-effectiveness of long-term use of atorvastatin 10 mg daily for primary prevention among hypertensive patients with a cardiovascular risk profile as low as most of those included in this study remains to be clarified.
The current study assessed the impact of atorvastatin specifically in hypertensive patients with reasonably controlled blood pressures and normal or mild-to-moderately elevated levels of serum cholesterol. While patients with hypertension have been included in the previous lipid-lowering trials, this study evaluated the benefits of lipid lowering with a statin in this specific population of patients with a lower range of lipids levels (total cholesterol ≤6.5 mmol/L) than the prior trials. Consequently, the results of the ASCOT-LLA both confirm and extend the results of the two previously published primary prevention trials (WOSCOPS (4) and AFCAPS/TexCAPS (5)) documenting the benefits of a statin for the prevention of major fatal and non-fatal events coronary and cerebrovascular events. While the magnitude of the benefits observed is consistent with those seen in the earlier primary prevention trials, the present trial had a significantly shorter duration of follow-up (mean 3.3 years) due to its early termination. If the study had been continued for the 5 years of follow-up that was originally planned, we might expect to have observed a further reduction in events. It should also be noted that the study evaluated a fixed starting dose of atorvastatin (10 mg daily). Therefore, whether or not titration of the dose to achieve greater reductions in serum cholesterol may have resulted in any further reduction of events is matter of speculation.
ASCOT is not the only lipid-lowering trial that has been conducted specifically among patients with hypertension. The recently published antihypertensive drug trial ALLHAT also included a randomized lipid-lowering arm. Among a subgroup of 10,355 patients in ALLHAT, the effects of pravastatin 40 mg daily were compared to usual care. (6) Unlike the present trial, there were no significant benefits in terms of cardiovascular and cerebrovascular events were apparent with statin use in ALLHAT. There are numerous reasons why the ASCOT and ALLHAT results are so different. First of all, the baseline demographics of the patients included in the two trials differ substantially; the ALLHAT patient population was slightly older and included patients with a history of CHD (14%), as well as a greater proportion of women and non-white patients. More importantly, there was a substantial use of statins among the patients in the usual care group of ALLHAT, which resulted in a LDL-cholesterol difference of only 17% between the two groups (compared to 29% in ASCOT). Consequently, the most conceivable explanation to account for this discrepancy between the results of ASCOT and ALLHAT is the dose-response effect on cardiovascular events that we expect to be associated with the degree of LDL-cholesterol reduction achieved.
The observation in ASCOT that there was no significant benefit of atorvastatin among women and patients with diabetes contradicts the findings of sub-group analyses of prior lipid-lowering outcome studies. While this observation is surprising, it is likely just a reflection of the fact that there were relatively few events experienced among these patient subgroups. A primary endpoint occurred in only 36 women and 84 diabetics during the study follow-up. Since the study was discontinued prematurely, the apparent lack of benefit of atorvastatin is probably a chance observation due to lack of power to detect a true difference among in these subgroups.
While the current study assessed the benefits of cholesterol lowering in the primary prevention of CHD in hypertensive patients who might not conventionally be considered dyslipidemic, the average number of cardiovascular risk factors (in addition to hypertension) of the population included in the trial was 3.7. The authors report that the placebo group experienced the equivalent of a 9.4% 10-year coronary event rate (non-fatal MI and fatal CHD), which represents a low risk population. However without the average blood pressure reduction of 25/14 mmHg that was achieved in this trial over the follow-up period, the true 10-year coronary risk in the placebo population would have been closer to 15%, indicating that this was at least moderate risk population. Consequently, the data from ASCOT provide support for the adoption of lower lipid treatment thresholds and targets in the clinical practice guidelines, which are currently being revised in Canada. The post-hoc analysis that revealed similar hazard ratios (HR) for the primary endpoint across a range of baseline total cholesterol values (<5.0 mmol/L: HR = 0.63; p=0.098; 5.0-5.99 mmol/L: HR = 0.62; p=0.011; ≥6.0 mmol/L: HR = 0.69; p=0.084) also lends support for lowering the recommended lipid targets.
One important methodological issue that should be highlighted is that nearly 40% of the patients enrolled in this trial would have been classified as being at very high risk for CAD (10-year risk of CAD >30%), based on North American guidelines. (7,8) From their medical histories, many of these patients had what is considered to be a coronary heart disease (CHD) equivalent: ~25% had diabetes, ~10% had a prior stroke or TIA, and the ~5% of had peripheral vascular disease. In accordance with the current Canadian guidelines7 and usual contemporary practice, these patients would be usually be treated with a statin to achieve a target LDL-cholesterol of <2.5 mmol/L.
ASCOT is the first study to prospectively demonstrate the significant benefits of a statin in patients with hypertension for the primary prevention of both coronary and cerebrovascular events. However, these results confirm the expected benefits of statin therapy observed in subgroup analyses from earlier trials. In addition, by enrolling patients with total cholesterol ≤6.5 mmol/L, this trial extends the range of serum cholesterol across which statins have a proven benefit. Consequently, the importance of this trial will not come from its influence on individual patient pharmacotherapeutic decisions but from its potential to impact the treatment thresholds and lipid targets recommended in future clinical practice guidelines. With the new Canadian guidelines for the management and treatment of dyslipidemia looming on the horizon, we shall soon see whether or not ASCOT has had an impact.
Dr. Pearson has served as a consultant to and received payment for speaking at meetings or funding for clinical research from pharmaceutical companies marketing lipid-lowering medications, including Astra-Zeneca Canada Inc., Bristol-Myers Squibb Canada Inc., Merck Frosst Canada Inc, Merck-Frosst/Schering Pharmaceuticals and Pfizer Canada Inc.
Copyright © 2003 by the Journal of Informed Pharmacotherapy. All rights reserved.