Author + information
- Received May 8, 2004
- Revision received July 9, 2004
- Accepted July 13, 2004
- Published online November 2, 2004.
- ↵*Reprint requests and correspondence:
Dr. Michael J. Koren, Jacksonville Center for Clinical Research, 4085 University Boulevard South, Suite 1, Jacksonville, Florida 32216
Objectives This study sought to determine if an aggressive, focused low-density lipoprotein cholesterol (LDL-C)-lowering strategy was superior to usual care for coronary heart disease (CHD) patients enrolled in health maintenance organization or Veterans Administration settings.
Background Statin therapy benefits are well established. No prospective, randomized studies have tested strategies to optimize these benefits in a “real-world” setting.
Methods A total of 2,442 CHD patients with hyperlipidemia were randomized to either an aggressive treatment arm using atorvastatin or usual care and followed for 51.5 months on average. Atorvastatin-group patients were titrated to LDL-C goals of <80 mg/dl (2.1 mmol/l) or a maximum atorvastatin dose of 80 mg/day. Usual-care patients received any treatment deemed appropriate by their regular physicians. End point assessments were complete in 958 atorvastatin-group and 941 usual-care patients. Partial assessments occurred in 259 patients in the atorvastatin group and 284 patients in the usual care group who did not complete four years of study participation because of adverse events, withdrawn consent, or follow-up loss. The primary efficacy parameter was time to first cardiovascular event.
Results A total of 289 (23.7%) patients in the atorvastatin group compared with 333 (27.7%) patients in the usual care group experienced a primary outcome (hazard ratio, 0.83; 95% confidence interval 0.71 to 0.97, p = 0.02). This reduction in morbidity was largely due to fewer non-fatal myocardial infarctions (4.3% vs. 7.7%, p = 0.0002). Levels of LDL-C were reduced more (34.3% vs. 23.3%, p < 0.0001) and National Cholesterol Education Program goals (LDL-C <100 mg/dl) more likely met at end-of-study visits (72.4% vs. 40.0%) in patients receiving atorvastatin compared with those receiving usual care.
Conclusions An aggressive, focused statin therapy management strategy outperformed usual care in health maintenance organization and Veterans Administration clinic patients with CHD.
Recent epidemiological data suggest that the downward slope of the 40-year trend in U.S. cardiovascular disease mortality that began in the 1950s has flattened over the last several years (1,2). Ironically, this change in trend may be occurring during a period when clinical trials have definitively demonstrated significant morbidity and mortality reductions in both primary and secondary prevention populations with drugs such as statins (3–9) and angiotensin-converting enzyme inhibitors (10–12). This disparity raises questions about whether benefits demonstrated in controlled treatment trials are being achieved in clinical practice (13).
Recommendations of the second and third reports of the National Cholesterol Educational Program Adult Treatment Panel (NCEP ATP) (14,15) for low-density lipoprotein cholesterol (LDL-C) lowering have been endorsed by most clinicians. These guidelines have become a quality benchmark for managed care organizations, Health Plan Employer Data and Information Set measurements, and heart disease risk factor disease management programs. Although these recommendations are widely accepted, we are unaware of studies that have prospectively compared different treatment strategies for compliance with these guidelines and consequent clinical outcomes.
The Aggressive Lipid-Lowering Initiation Abates New Cardiac Events (ALLIANCE) study is the first prospective, randomized trial, to our knowledge, to evaluate outcomes attributable to different lipid treatment strategies in a predefined coronary heart disease (CHD) patient group. During the trial, patients were randomized at study centers to receive either an aggressive lipid-lowering regimen using atorvastatin or “usual care” as deemed appropriate by the patient's regular physicians. The study was designed and executed as a “real-world” trial to determine if an aggressive, focused approach to hyperlipidemia would outperform current physician practice in CHD patients.
The design of the ALLIANCE study has been described in detail previously (16). The study was conducted at 16 centers throughout the U.S. These centers could be a staff model health maintenance organization, a community physician open-provider health maintenance organization, or a Veterans Affairs system. The study was open only to patients identified by diagnosis codes related to CHD from relevant managed health care organizations or Veterans Administration facilities databases. Letters were sent to these patients inviting them to be screened for the study at research centers. After screening, eligible patients were randomized to either the atorvastatin group or the usual-care group by the central laboratory between July 1995 and June 1998.
Eligible subjects were men or women >18 years of age with known CHD defined as a history of acute myocardial infarction (MI) (>3 months before screening), percutaneous transluminal coronary angioplasty (>6 months before screening), coronary artery bypass graft surgery (>3 months before screening), or unstable angina (>3 months before screening). Inclusion LDL-C levels were between 110 mg/dl (2.8 mmol/l) and 200 mg/dl (5.2 mmol/l) for patients receiving lipid-lowering medication and between 130 mg/dl (3.4 mmol/l) and 250 mg/dl (6.5 mmol/l) for patients receiving no lipid-regulating therapy. Lipid therapy was not removed for the patients' baseline laboratory measurements. Approximately two-thirds of patients were on lipid-regulating agents at baseline.
The patients in atorvastatin group received study medication and underwent visits at the research center where only hyperlipidemia was treated. Primary-care physicians managed other cardiovascular risk factors. These patients had their current lipid-lowering therapy discontinued after baseline laboratory measurements and were started on atorvastatin at 10 mg/day. The atorvastatin dosage was doubled every four weeks until an LDL-C level of <80 mg/dl (2.1 mmol/l) or a maximum dose of 80 mg/day was achieved. Combination lipid therapy was not allowed. After dose titration was completed, subjects were followed every six months.
The patients in the usual care group were maintained on the lipid-lowering program already prescribed. Adjustments in lipid therapy were made entirely at the discretion of regular physicians. Lipid-regulating therapy in patients in the usual care group could include atorvastatin after its approval in 1997. The patients in the usual care group were invited back to the research centers every six months for collection of adverse-event data. In many cases, these follow-up evaluations were performed by telephone because subjects in the usual-care arm received no medication, counseling, or interim laboratory tests from lipid study centers.
Efficacy and safety evaluations
Primary cardiovascular events included cardiac death, non-fatal MI, resuscitated cardiac arrest, cardiac revascularization, and unstable angina requiring hospitalization. The primary efficacy end point was the time from randomization to the first occurrence of a primary cardiovascular event. Secondary outcomes included non-cardiac death, peripheral revascularization, hospitalization for congestive heart failure, and stroke.
Laboratory analyses for the patients in the atorvastatin group at baseline, at interim visits, and at study completion were performed by Pacific Biometrics (Seattle, Washington). For the patients in the usual care group, laboratory values were measured by the central laboratory only at baseline and at the end of the study. All other laboratory testing for the patients in the usual care group was directed by treating physicians.
The study was conducted in accordance with the Declaration of Helsinki and approved by the local ethics committees or institutional review boards. An independent outcomes committee of five cardiologists reviewed and adjudicated all study outcomes. Interim monitoring of study events and serious adverse events was performed by an independent data safety monitoring committee.
A sample size of 900 patients in each treatment group (1,800 patients total) was expected to provide 90% power to detect a 29% reduction in the four-year cumulative primary cardiovascular event rate (20% for usual care and 14.2% for atorvastatin therapy). Allowing for a 25% overall reduction in sample size due to patients not completing the study for personal or medical reasons and managed health care organizations attrition,accrual of 1,200 patients per group (2,400 patients total) were to be randomized to receive either atorvastatin therapy or usual care.
Intention-to-treat analysis was performed on all randomized patients (n = 2,442). For the primary end point analysis, Kaplan-Meier survival curves for the two groups were compared using the log-rank statistic. In addition, each primary outcome was analyzed separately to determine if differences in time to event were present. The primary outcomes were also examined by a Cox proportional hazards model to assess treatment differences adjusted for center effect and any additional factors that might affect survival. Secondary outcomes were examined in a similar manner.
Of 3,451 patients screened at 16 centers, 2,442 were randomized, with 1,217 assigned to the atorvastatin group and 1,225 to the usual care group (Fig. 1).Patients were seen in the research centers from July 6, 1995, to July 17, 2002. Demographic and clinical characteristics of patients randomized to the atorvastatin and usual care groups were similar at baseline (Table 1).The patients were followed for a median period of 54.3 months (mean 51.5 months, standard error = 0.37).
A total of 958 patients in the atorvastatin group and 941 in the usual care group completed end point assessments. These numbers included patients who returned for the final visit, those who died during the course of the trial, and those who reported an end point during the study but were either lost to follow-up at last patient visit or discontinued from the trial before study end. Partial end point assessments occurred in 259 patients in the atorvastatin group and 284 patients in the usual care group who did not complete the study for various reasons. These reasons included discontinuations for adverse events and administrative issues such as withdrawn consent, protocol violations, and subjects lost to follow-up. Patient disposition by study arm is shown in Figure 1.
In the atorvastatin group, the median dose of atorvastatin received by patients over the course of the study was 40.5 mg/day; 545 patients (45%) received 80 mg/day.
Serum lipid levels
The mean LDL-C levels at baseline (on pre-randomization lipid-lowering treatment) were 147 mg/dl (3.8 mmol/l) in the atorvastatin group and 146 mg/dl (3.8 mmol/l) in the usual care group. At the end of the study, LDL-C levels had decreased by 34.3% to 95 mg/dl (2.5 mmol/l) in the atorvastatin group and by 23.3% to 111 mg/dl (2.9 mmol/l) in the usual care group (p < 0.0001) (Table 2).At study completion, 72.4% of patients in the focused group had achieved their ATP III goal (LDL-C <100 mg/dl [2.6 mmol/l]) compared with 40.0% of those receiving usual care (p < 0.001).
The patients in the atorvastatin group also demonstrated significantly greater reductions in total cholesterol than those in the usual care group (p < 0.0001). Reductions in triglycerides and increases in high-density lipoprotein cholesterol were moderate in both groups (Table 2).
During the study, 289 patients (23.7%) receiving aggressive treatment with atorvastatin experienced a primary outcome, compared with 333 patients (27.7%) receiving usual care (p = 0.026) (Table 3).The event curves for the atorvastatin and usual care groups began to diverge at approximately one year and continued to separate until the end of the study (Fig. 2).The Cox regression model indicated a 17.1% reduction in the risk of a primary cardiovascular outcome for the patients in the atorvastatin group compared with patients in the usual care group (hazard ratio = 0.829; p = 0.020). This benefit of focused treatment compared with usual care was due largely to a significant reduction in non-fatal MI (4.3% vs. 7.7%; p = 0.0002). For this cardiovascular event, the Kaplan-Meier survival curves for the atorvastatin and usual care groups separated within months and continued to diverge up to the study's end (Fig. 3).The Cox regression model indicated a 47.4% non-fatal MI risk reduction for atorvastatin therapy versus usual care (hazard ratio = 0.526; p = 0.0002). Trends favoring the aggressively treated group that did not reach statistical significance were seen for all other components of the primary composite end point (Fig. 4).The Cox regression analysis did not show statistically significant interactions between treatment and center (p = 0.17). Subgroup analyses by gender, age, and race showed efficacy results consistent with the overall population. No significant interactions between treatment and gender, age, or race were observed.
The occurrence of serious adverse events was similar in the atorvastatin (40%) and usual care groups (42%). In both treatment arms, the most commonly reported serious adverse events were chest pain, atrial fibrillation, pneumonia, cellulitis, and gastrointestinal hemorrhage. No unexpected safety concerns were identified in the patients aggressively treated with atorvastatin.
Protocol-stipulated routine laboratory testing, performed only in the atorvastatin group, showed abnormal aspartate aminotransferase or alanine aminotransferase levels (>3 times the upper limit of normal) in 8 (0.7%) and 16 patients (1.3%), respectively. The mean changes in aspartate aminotransferase (2.7 U/l) and alanine aminotransferase (3.1 U/l) were not clinically significant. There were no documented cases of creatine phosphokinase levels >10 times upper limit of normal in the atorvastatin group and no documented cases of rhabdomyolysis or myopathy in either group. There were 144 cases of cancer, with 67 in the atorvastatin group and 77 in the usual care group.
Multiple landmark studies over the past decade have established statin therapy as a cornerstone of preventative cardiovascular care. These placebo-controlled trials have demonstrated the benefit of several agents in various primary and secondary prevention populations. The ALLIANCE study breaks ground by showing the incremental benefit attributable to an aggressive, focused clinical strategy compared with usual care in predefined groups of CHD patients with equal access to health care resources.
Patients in both the atorvastatin and usual care groups experienced significant reductions in LDL-C levels from baseline measurements obtained on pre-study lipid therapy. About two-thirds of patients in each group were on lipid-lowering medication at the time of baseline measurements. Considering the significant improvement of lipid levels from baseline in patients in the usual care group, the incremental clinical benefit of the aggressively treated cohort in the ALLIANCE study is particularly impressive. The patients in the atorvastatin group had, on average, an additional 16 mg/dl reduction in LDL-C beyond that of patients in the usual care group at end-of-study visits and experienced a 17% overall and 47% non-fatal MI risk reduction.
The absolute risk reduction attributable to aggressive lipid therapy compared to usual care was 4% or 1 patient spared from a primary end point for each 25 patients treated. If this incremental benefit is added to the expected risk reduction accrued by patients in the usual care group because of improvements in their lipid levels during the study, the overall benefit of aggressive treatment is even greater. Further, the health benefits of aggressive lipid treatment were particularly favorable in that they were weighted toward reduction of “hard” end points such as non-fatal MIs and cardiac death (Fig. 4).
Although not defined a priori as a study end point, the rate of non-fatal MI and cardiac death combined was reduced by 42.6% in the atorvastatin group compared with the usual care group (hazard ratio = 0.574, 95% confidence interval 0.442 to 0.746; p = 0.0001).
The ALLIANCE study adds important safety information about the use of high-dose statin therapy. It is reassuring that patients treated aggressively achieved their goals with a low risk of adverse drug effects. Nearly one-half of the patients in the atorvastatin group received the maximum 80-mg dose, and the incremental clinical benefits observed in the atorvastatin group compared with the usual care group occurred without any detectable safety differences. The long-term ALLIANCE study also supports studies of shorter duration (17–19) showing that high-dose atorvastatin is well tolerated.
The design of the ALLIANCE study as a “real-world” study was both a strength and weakness. Study center follow-up of patients was limited to every six months once the initial dose titration of patients in the atorvastatin group was completed. This infrequent follow-up at study centers was designed to minimize the disruption of standard care and to limit crossover of patients in the usual care group to the atorvastatin group treatment patterns. In the initial study design, we envisioned that pharmacy, laboratory, and health care resource utilization data would be obtained from managed-care databases.
Unfortunately, due to the vicissitudes of the health care marketplace, managed-care databases for most subjects could not be accessed as anticipated. This limitation occurred because of the financial failure or merger of most of the managed-care organizations during the study. We also found that patients treated in one managed care often switched to another and that most hospitalizations and much other health care resource utilization occurred outside of systems controlled by managed-care entities. Because of these shortcomings, we do not have data on medication use or interim laboratory values of patients in the usual care group. Additionally, we have fewer verifiable follow-up lipid values in patients in the usual care group compared with the atorvastatin group (725 vs. 1,146) because patients in the usual care group received interim lipid profiles at the discretion of their treating physicians. The patients in the usual care group who discontinued study participation early because of events or dropouts often did not have follow-up lipids drawn at study sites.
Lack of access to managed-care data also limited our efforts to tally the reasons for patient study dropout. For example, we determined that 68 patients in the atorvastatin group stopped study participation because of non-serious side effects. The patients in the usual care groups who dropped out could not be similarly classified because of insufficient details on drug tolerance in this group. Therefore, the patients in the usual care group who dropped out were classified as unwilling to continue or lost to follow-up, although we believe that some unknown number of these patients discontinued study participation because of statin intolerance. Fortunately, due to the persistence of study sites, we obtained records for all patients known to have been hospitalized. This effort allowed us to calculate a serious adverse event rate for each study group and to conclude that there were no serious cases of rhabdomyolysis in either group.
Privacy issues also contributed to difficulties that would not have occurred had the study been performed entirely in a research setting. Because they were recruited in the ALLIANCE study through managed-care databases rather than through relationships with physician investigators, patients may have been less willing to share follow-up information after changing health plans andmay have subsequently withdrawn consent at a higher rate in the ALLIANCE study compared with more traditional clinical trials. Additionally, several investigators cited privacy issues when asked if they would expand their efforts to retrieve information from patients who had changed health plans, clinical providers, or their residence. The loss to follow-up rate among Veterans Administration study subjects was lower than that for health maintenance organization participants (6.3% vs. 9.8%), suggesting that future studies such as the ALLIANCE study may be most successful in single-payer health care environments.
To mitigate this issue of incomplete follow-up, we performed analyses on both the cohort for which we had complete information and the cohort whose survival status was known from death index data and partial records. There were no significant differences between these groups. We believe that the higher-than-expected dropout rate is unlikely to have a material impact on the efficacy results of the trial, because losing patients to follow up was mostly related to nonmedical issues (changing health plans and privacy concerns).
Future lipid guidelines may consider the results of the ALLIANCE study and other recent trials such as the Reversing Atherosclerosis with Aggressive Lipid Lowering (REVERSAL) (19) and Pravastatin or Atorvastatin Evaluation and Infection Therapy (PROVE-IT) (20)that have demonstrated the benefit of lipid-lowering below current guidelines in high-risk patients. Some of the pitfalls of treating CHD patients in the “real world” should also be considered. For example, in the ALLIANCE study, the atorvastatin group mean LDL-C of 95 mg/dl at study completion was above a target of 80 mg/dl (2.1 mmol/l). The mean end study LDL-C of 111 mg/dl in the usual care group fell short of the NCEP goal of 100 mg/dl. In light of these results, perhaps the treatment paradigm for cholesterol management should shift from a target LDL-C goal to a maximumacceptable level for a given patient risk type. Future guidelines may also recommend expanded use of the focused disease management techniques that helped patients in the atorvastatin group meet current NCEP goals nearly twice as often as patients in the usual care group. Focused disease management may achieve results through better compliance with aggressive pharmacological therapy, support for non-pharmacological dietary measures, and encouragement of lifestyle changes such as increased physical activity.
In summary, an aggressive, focused approach to LDL-C lowering using atorvastatin improved clinical outcomes compared with usual care for patients with CHD. This incremental benefit was accomplished without additional risk. The ALLIANCE study results further support a growing body of evidence to suggest that LDL-C treatment goals below current guidelines are desirable for appropriate patients (7–9,17–21).
The authors acknowledge Barbara LePetri, Donald Black, David Reed, Robert Singer, Sheryl Grant, Jonathan Ma, Joan Barton, and Noah Rosenberg for their invaluable assistance with the conduct of this study and the manuscript preparation.
For a list of the ALLIANCE Study Investigators, please see the November 2, 2004, issue of JACCat www.onlinejacc.org.
Parke-Davis and Pfizer Pharmaceuticals funded the study. Drs. Koren and Hunninghake were clinical investigators in the trial who were consulted on study design, collected subject information, assisted in all aspects of data analysis, and drafted the manuscript. The sponsors initiated the study design and also participated in all aspects of logistical support during the trial, statistical and data analyses, and manuscript preparation. Pfizer Pharmaceuticals reviewed the final manuscript and assisted with editing and revisions, but the authors retained final decisions on content.
- Abbreviations and acronyms
- Aggressive Lipid-Lowering Initiation Abates New Cardiac Events study
- coronary heart disease
- low-density lipoprotein cholesterol
- myocardial infarction
- Received May 8, 2004.
- Revision received July 9, 2004.
- Accepted July 13, 2004.
- American College of Cardiology Foundation
- Cooper R.,
- Cutler J.,
- Desvigne-Nickens P,
- et al.
- American Heart Association. Heart Disease and Stroke Statistics: 2003 Update. Dallas, TX: American Heart Association. Available at: www.americanheart.org/downloadable/heart. Accessed May 2004.
- Sever P.S.,
- Dahlof B.,
- Poulter N.R,
- et al.
- The European Trial on Reduction of Cardiac Events With Perindopril in Stable Coronary Artery Disease Investigators
- The National Cholesterol Education Program
- The Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults
- Isaacsohn J.L.,
- Davidson M.H.,
- Hunninghake D.,
- Singer R.,
- McLain R.,
- Black D.M.
- Schwartz G.G.,
- Olsson A.G.,
- Ezekowitz M.D,
- et al.
- Cannon C.P.,
- Braunwald E.,
- McCabe C.H,
- et al.