Author + information
- Received May 7, 2009
- Revision received September 30, 2009
- Accepted October 12, 2009
- Published online December 15, 2009.
- Sabina A. Murphy, MPH,
- Christopher P. Cannon, MD* (, )
- Stephen D. Wiviott, MD,
- Carolyn H. McCabe, BS and
- Eugene Braunwald, MD
- ↵*Reprint requests and correspondence:
Dr. Christopher P. Cannon, TIMI Study Group, 350 Longwood Avenue, First Floor, Boston Massachusetts 02115
Objectives In addition to reducing first events in patients after an acute coronary syndrome (ACS), we hypothesized that high-dose atorvastatin 80 mg would also reduce recurrent cardiovascular events, and therefore total events, compared with pravastatin 40 mg during the 2-year follow-up.
Background In the PROVE IT–TIMI 22 (Pravastatin or Atorvastatin Evaluation and Infection Therapy–Thrombolysis In Myocardial Infarction 22) trial, more intensive lipid lowering with high-dose atorvastatin reduced the first occurrence of the primary end point (death, myocardial infarction, unstable angina requiring rehospitalization, stroke, or revascularization ≥30 days) compared with moderate lipid lowering with pravastatin.
Methods Poisson regression analysis was performed to compare the number of occurrences of the primary end point between high-dose atorvastatin and pravastatin in the PROVE IT–TIMI 22 trial.
Results As previously reported, first primary end point events were reduced by 16% with atorvastatin 80 mg versus pravastatin 40 mg (n = 464 vs. n = 537, respectively; p = 0.005). Additional events were also reduced by 19% with atorvastatin 80 mg (n = 275 vs. n = 340, respectively; p = 0.009). Overall, there were 138 fewer primary efficacy events with atorvastatin 80 mg versus pravastatin 40 mg (n = 739 vs. n = 877, respectively; rate ratio: 0.85, 95% confidence interval: 0.77 to 0.94, p = 0.001).
Conclusions Although analytic techniques commonly used in clinical outcomes trials censor patients who experience a component of the primary composite end point, total cardiovascular events are important to patients, clinicians, and health care payers. Maintaining low levels of low-density lipoprotein cholesterol is central to preventing additional atherosclerotic development and subsequent cardiovascular events. Atorvastatin 80 mg, a more intensive low-density lipoprotein cholesterol lowering agent, reduced both first and subsequent primary end point events compared with pravastatin 40 mg after ACS.
With customary statistical analysis of clinical outcomes trial data using survival methodology, patients who experience a component of a primary composite end point are censored from the analysis after the initial event, despite remaining at risk for additional events during the course of the trial follow-up. Reporting of a composite event only may be less sensitive to the variation of the treatment effect across the individual events, particularly if a treatment reduces the incidence of 1 type of event and increases the incidence of another (1). Additionally, if the events have a different impact on either quality of life or health care costs, it may be important to examine the frequency and treatment effects of the different events (1). Although data on all events occurring during the study are often collected, reporting of such events is generally examined by individual end points rather than by total number of cardiovascular events that occurred. From a clinical perspective, patients and physicians are concerned not only with the initial event a patient may experience but also with subsequent events. We recently examined the impact of more potent antiplatelet therapy on total cardiovascular events (2), but the impact of more intensive lipid-lowering therapy on total cardiovascular events has not been explored.
Compared with moderate lipid lowering using standard-dose pravastatin therapy, intensive lipid lowering with high-dose atorvastatin therapy after acute coronary syndrome (ACS) significantly reduced the first occurrence of the primary end point of death, myocardial infarction (MI), stroke, unstable angina (UA) requiring rehospitalization, or revascularization occurring >30 days after the index ACS event in the PROVE IT–TIMI 22 (Pravastatin or Atorvastatin Evaluation and Infection Therapy–Thrombolysis In Myocardial Infarction 22) trial (3,4). We hypothesized that atorvastatin 80 mg would also reduce recurrent cardiovascular events and therefore total events compared with pravastatin 40 mg during the 2-year follow-up.
The study design and primary results papers for the PROVE IT–TIMI 22 trial have been published previously (3,5). The trial enrolled 4,162 patients with an ACS within the prior 10 days. Patients were randomly assigned to either 80 mg atorvastatin (intensive lipid lowering) or 40 mg pravastatin (moderate lipid lowering). Prior statin use was not an exclusion criterion in the PROVE IT–TIMI 22 trial. Baseline and on-treatment lipid levels and high-sensitivity C-reactive protein (CRP) were measured in central laboratories. Patients were followed up for clinical events for an average of 2 years.
All end points used in the analyses were adjudicated by members of an independent clinical events committee who were blinded to the treatment assignment (5). Fatal events were counted as a single event, not as 2 separate events. For example, if a patient experienced an MI and then had cardiovascular death with the cause of death adjudicated as being due to the MI, the event was considered 1 fatal MI event and was both an MI and cardiovascular death. Patients were to remain on the study drug even if the subject experienced 1 of the nonfatal efficacy end points of the study.
The present analysis was not pre-specified in the primary analytic plan and should be considered exploratory. Baseline clinical characteristics are presented as frequencies for categorical variables and as medians and interquartile ranges for continuous variables. Comparisons between baseline characteristics for patients with no events, a single event, or multiple events were made using the chi-square test for categorical variables and Kruskal-Wallis for continuous variables. Poisson regression analysis was performed to compare the total number of occurrences of the primary end point between all patients in the atorvastatin and pravastatin groups. Analyses were performed using Stata/SE version 9.2 (Stata Corp., College Station, Texas).
Patients who had multiple events had more frequent comorbidities at study entry, including a history of diabetes mellitus, hypertension, and hypercholesterolemia (Table 1).Interestingly, patients who had multiple events were more likely to present with UA as the index event. There was no difference in baseline low-density lipoprotein cholesterol (LDL-C) or total cholesterol among patients who had multiple events compared with patients who had no event or a single event, but triglycerides were higher among patients who had multiple events. Patients having multiple events were more likely to have been statin users before the qualifying event (31.6% for patients with multiple events, 28.8% for patients with single event, 23.7% for patients with no events; p < 0.001).
The primary end point of first occurrence of death, MI, stroke, UA requiring rehospitalization, or revascularization occurring >30 days after the index ACS event was significantly reduced among patients randomly assigned to intensive lipid-lowering therapy with atorvastatin 80 mg as compared with moderate lipid-lowering therapy with pravastatin 40 mg (22.4% [n = 464] vs. 26.3% [n = 537], hazard ratio: 0.84, 95% confidence interval: 0.74 to 0.95, p = 0.005), as previously reported (3). Thus, by the usual analysis that takes account of only first events, the difference between the 2 treatment arms was 73 events. In addition to the reduction in first events, subsequent events were also reduced in the atorvastatin 80 mg group (n = 275 in the atorvastatin 80 mg group vs. n = 340 in the pravastatin 40 mg group, p = 0.009) (Fig. 1),resulting in 138 fewer total primary events during follow-up (total events n = 739 vs. n = 877, rate ratio: 0.85, 95% confidence interval: 0.77 to 0.94, p = 0.001). Using a marginal Cox proportional hazards model rather than a Poisson regression analysis produced a hazard ratio of 0.82 (p = 0.008) in favor of the atorvastatin 80 mg group. Among the additional events, there were 177 second events in the atorvastatin 80 mg group (8 fatal and 169 nonfatal) versus 203 second events in the pravastatin 40 mg group (12 fatal and 191 nonfatal), and 60 third events in the atorvastatin 80 mg group (3 fatal and 57 nonfatal) versus 74 third events in the pravastatin 40 mg group (3 fatal and 71 nonfatal).
When examining the individual components of the primary composite end point, the majority of additional events reduced in the atorvastatin 80 mg group were UA events requiring rehospitalization and revascularization, with a similar number of deaths, MI, and strokes (Table 2).
Among patients with a single event or multiple events, cholesterol and CRP levels before the event were compared to those of the final study sample for patients with no events (Table 3).Total cholesterol, LDL-C, and high-density lipoprotein cholesterol were lower and CRP was higher among patients with single and recurrent end point events than among patients without any end point events.
The overall rate of permanent study drug discontinuation within 30 days after the first event was low (3.0%) and did not differ by randomized group (3.3% in the atorvastatin 80 mg group vs. 2.7% in the pravastatin 40 mg group, p = 0.56). There were no cases of recurrent myalgia or myositis in the trial in either treatment group. Among patients with no events, compliance was classified as good in 85% of patients at the last contact with available compliance data. Among patients with 1 event, compliance was classified as good in 91% of patients at the contact before the event, whereas 88% of patients with >1 event had compliance classified as good.
This analysis from the PROVE IT–TIMI 22 trial demonstrates that intensive lipid-lowering therapy with high-dose atorvastatin therapy after an ACS prevented not only the first occurrence of the primary end point of death, MI, stroke, UA requiring rehospitalization, or revascularization but also reduced subsequent, and thus the total number of, primary end point events among patients with an ACS. These findings suggest that continued therapy with a regimen that maintains low LDL-C is central to preventing additional atherosclerotic development and cardiovascular events, including recurrent events. Based on the primary end point results from the PROVE IT–TIMI 22 trial, the number of patients needed to treat to prevent the first occurrence of the primary end point is 26. However, when considering total events, the needed-to-treat number to prevent 1 event is much lower, at 14.
Several studies have shown that when compared with placebo, the risk of death and cardiovascular events is lowered by reducing LDL-C levels with statin therapy for both primary and secondary prevention (6). More recent trials, such as the PROVE IT–TIMI 22 study and the TNT (Treating to New Targets) trial, demonstrated that further reductions in LDL-C levels with more intensive statin therapy conferred additional clinical benefit (3,7,8). The JUPITER (Justification for the Use of Statins in Primary Prevention: an Intervention Trial Evaluating Rosuvastatin) trial extended these findings, demonstrating improved outcomes for patients with lipid levels considered optimal under current guidelines but with elevated levels of high-sensitivity CRP (9,10). While these studies have shown improved outcomes with statin therapy to lower LDL-C levels, the primary end point for each trial was time to occurrence of a first major cardiovascular event, the exact definitions of which varied, but all of which were composite end points. The reduction in first event has been consistent across the studies, but total number of events prevented has not been fully reported.
Unexpectedly, the median LDL-C measured before the first event for patients having an event and measured at study end for patients without an event was lower in patients with single and recurrent end point events than in patients without any end point events. While the exact cause of this finding is not known, it is likely confounded by the observation that patients with single or multiple events more frequently presented as statin users before the qualifying event, contributing to the lower LDL-C; thus, the lower LDL-C levels in patients with events may be a spurious finding. It is also possible that the difference may be due to the timing of the LDL-C measurements, which were taken at a median of 2 years, at the time of the final study visit of patients without an event, and at a median of 6 months for patients with an event. Compliance with study drug medication is likely greatest early in the trial when the LDL-C measurement was used for the cohort with a single or multiple event and lower at trial end when the LDL-C measurement was used for the cohort without any events.
Expanded analysis of total events during the course of therapy may warrant more consideration in future trials, given the attention focused on health care resource utilization. Despite the frequent practice of censoring patients who experience a component of the primary composite end point in standard statistical analysis of clinical outcomes trial data, all events occurring during the course of the entire trial are important to patients, clinicians, and payers (11). Multiple ischemic events result in higher mortality and a reduced quality of life (12). Multiple events can also increase costs by requiring more hospitalizations, diagnostic tests, treatments, and physician visits. Another method for analyzing multiple events is to perform a landmark analysis to explore the treatment effect beyond a pre-determined timeframe to evaluate for both early and late benefit. While this type of analysis can be informative, it should be interpreted as a post-randomization analysis because there may be differences among patients who were censored because of the occurrence of a fatal event or differences in concomitant treatments among patients who experienced a nonfatal event during the early part of the trial. Other methods for analyzing multiple events in the presence of a terminal event have also been described by Chen and Cook (1), who propose use of marginal cumulative mean functions between treatment groups and reporting of a global test statistic.
Although cost-effectiveness analyses incorporate total costs when available, including repeat hospitalizations and multiple procedures, multiple events are often not incorporated into the effectiveness component of the analysis (13). From a cost economics perspective, the occurrence of multiple events would have a negative impact on long-term outcomes of both life-years and quality-adjusted life-years. Incorporating the total number of events into cost-effectiveness analyses may provide a more comprehensive estimate of the total cost effectiveness of a given therapy when the agent has been shown to reduce events beyond the first occurrence of the end point.
In a similar analysis of total events in the TRITON–TIMI 38 (Trial to Assess Improvement in Therapeutic Outcome by Optimizing Platelet Inhibition With Prasugrel–Thrombolysis In Myocardial Infarction 38) study, more potent antiplatelet therapy with the novel thienopyridine prasugrel was shown to reduce first, subsequent, and total occurrences of the primary end point of cardiovascular death, MI, or stroke compared with clopidogrel (2). As with the current analysis, which suggests that continued therapy with a more intensive lipid-lowering agent confers added benefit beyond the first event, the TRITON–TIMI 38 study analysis demonstrated the need for maintaining treatment with continued high levels of platelet inhibition after an ACS.
The National Cholesterol Education Program currently recommends an optional target LDL-C of <70 mg/dl for patients at high risk of cardiovascular events (14–16), including those with an ACS event. Baseline characteristics of patients in the present study were analyzed to identify high-risk characteristics associated with the occurrence of multiple events. Although baseline LDL-C levels did not differ among patients without an event, with a single event, or with multiple events, baseline triglyceride levels were slightly higher in patients who had multiple events. Other established risk factors such as diabetes mellitus and hypertension were most common among patients having multiple events. Presentation with an index event of UA was also more frequent among patients who experienced multiple events.
The TIMI Study Group received research grants from Bristol-Myers Squibb to conduct the PROVE IT–TIMI 22 trial. Dr. Cannon receives research grants/support from Accumetrics, AstraZeneca, Bristol-Myers Squibb/Sanofi Partnership, GlaxoSmithKline, Intekrin Therapeutics, Merck, Merck/Schering-Plough Partnership, Novartis, and Takeda; and is a clinical advisor with equity in Automedics Medical Systems. Dr. Wiviott has received speaking honoraria from Pfizer, Merck, AstraZeneca, and Bristol-Myers Squibb; and consulting fees from Bristol-Myers Squibb. Dr. McCabe has received grant support from Bristol-Myers Squibb (for the PROVE IT trial) and Schering-Plough. Dr. Braunwald has received grant support from Bristol-Myers Squibb (for the PROVE IT–TIMI 22 trial), Schering-Plough, and Merck; and has received honoraria and is an advisory board member of Schering-Plough and Merck.
- Abbreviations and Acronyms
- acute coronary syndrome
- C-reactive protein
- low-density lipoprotein cholesterol
- myocardial infarction
- unstable angina
- Received May 7, 2009.
- Revision received September 30, 2009.
- Accepted October 12, 2009.
- American College of Cardiology Foundation
- Chen B.E.,
- Cook R.J.
- Murphy S.A.,
- Antman E.M.,
- Wiviott S.D.,
- et al.
- Wiviott S.D.,
- de Lemos J.A.,
- Cannon C.P.,
- et al.
- Cannon C.P.,
- Steinberg B.A.,
- Murphy S.A.,
- Mega J.L.,
- Braunwald E.
- Ridker P.M.,
- Fonseca F.A.,
- Genest J.,
- et al.
- Gold M.,
- Siegel J.,
- Russel L.,
- Weinstein M.
- Spertus J.A.,
- Radford M.J.,
- Every N.R.,
- Ellerbeck E.F.,
- Peterson E.D.,
- Krumholz H.M.
- Anderson J.L.,
- Adams C.D.,
- Antman E.M.,
- et al.
- Grundy S.M.,
- Cleeman J.I.,
- Merz C.N.,
- et al.