Author + information
- Received December 5, 2001
- Revision received April 2, 2002
- Accepted April 8, 2002
- Published online July 3, 2002.
- ↵*Reprint requests and correspondence:
Dr. Karen M. Kuntz, Center for Risk Analysis, 718 Huntington Avenue, Boston, Massachusetts, USA.
Objectives We sought to estimate the potential gains in life expectancy achieved with statin therapy for individuals without overt hyperlipidemia but with elevated C-reactive protein (CRP) levels.
Background Persons with low-density lipoprotein (LDL) cholesterol levels below current treatment guidelines and elevated CRP levels are at increased risk of cardiovascular disease and may benefit from statin therapy.
Methods We constructed a decision-analytic model to estimate the gains in life expectancy with statin therapy for individuals without overt hyperlipidemia but with elevated CRP levels. The annual risks of myocardial infarction (MI) and stroke, as well as the efficacy of statin therapy, were based on evidence from randomized trials. Estimates of prognosis after MI or stroke were derived from population-based studies.
Results We estimated that 58-year-old men and women with CRP levels ≥0.16 mg/dl but LDL cholesterol <149 mg/dl would gain 6.6 months and 6.4 months of life expectancy, respectively, with statin therapy. These gains were similar to those for patients with LDL cholesterol ≥149 mg/dl (6.7 months for men and 6.6 months for women). In sensitivity analyses, we identified the baseline risk of MI and the efficacy of statin therapy for preventing MI as the most important factors in determining the magnitude of benefit with statin therapy.
Conclusions Our results suggest that individuals with elevated CRP levels, many of whom do not meet current National Cholesterol Education Program guidelines for drug treatment, may receive a substantial benefit from statin therapy. This analysis supports a crucial need for direct intervention trials aimed at subjects with elevated CRP levels.
Current use of statins in the primary prevention of cardiovascular disease is largely limited to those with low-density lipoprotein (LDL) cholesterol levels >160 mg/dl (1). However, half of all heart attacks and strokes occur among individuals classified as having low to moderate risk according to lipid screening alone.
One approach to improving global risk assessment and targeting certain patients for statin therapy includes evaluation of C-reactive protein (CRP) (2), an emerging inflammatory biomarker that is a strong independent predictor of future vascular risk among apparently healthy men and women (3–7). This approach also has pathophysiologic appeal, as the magnitude of benefit of statin therapy appears to be greater among individuals with elevated CRP levels (8,9). Furthermore, several studies indicate that statin therapy reduces CRP levels, independent of its LDL cholesterol-lowering effect (10–14).
The clinical impact of these data is likely to be greatest in primary prevention, where it has recently been demonstrated that statins are highly effective in reducing the risk of first coronary events among those with relatively low lipid levels but elevated CRP levels (13). However, despite the potential for CRP screening to identify individuals with LDL cholesterol levels below current treatment guidelines who might still benefit from statin therapy, the overall impact of treating this large segment of the population is unknown. To address this question, we used a decision-analytic model (15)to estimate the potential life-expectancy gains with statin treatment in the general population classified according to LDL cholesterol and CRP levels.
We constructed a Markov state-transition model to simulate hypothetical cohorts of men and women over their lifetime, where the annual events of interest were myocardial infarction (MI), stroke and death. The cohorts were defined according to median LDL cholesterol and CRP levels in the Air Force/Texas Coronary Atherosclerosis Prevention Study (AFCAPS/TexCAPS), to create three distinct groups: 1) LDL cholesterol <149 mg/dl and CRP <0.16 mg/dl (low LDL/low CRP); 2) LDL cholesterol <149 mg/dl and CRP ≥0.16 mg/dl (low LDL/high CRP); and 3) LDL cholesterol ≥149 mg/dl (high LDL; this group included subjects with both high and low CRP levels). For each group, we compared statin therapy with no treatment (assuming all groups were receiving dietary counseling).
In the simulation, subjects free of MI and stroke were assigned to either statin therapy or no therapy. Each year, individuals could experience no event, have a fatal or nonfatal MI or have a fatal or nonfatal stroke, or they could die of other causes. The base-case analysis was for 58-year-old men (a representative patient in AFCAPS/TexCAPS) (16). All analyses were performed using the decision-analytic program DATA (Treeage Software, Inc., Williamstown, Massachusetts).
Data and assumptions
The probabilities used in our model are shown in Table 1.
Risk of MI
The overall annual incidence of MI observed in AFCAPS/TexCAPS were 5.6 per 1,000 person-years (16). We assumed that this rate was applicable to men aged 55 to 64 years. In a recent analysis of the AFCAPS/TexCAPS population, the risk of MI in subjects with either high LDL cholesterol or high CRP was approximately twice that of individuals with low LDL cholesterol and low CRP (13). Based on the reported relative risks among the LDL/CRP groups, we split the overall MI rate into group-specific rates. For example, we calculated that the annual incidence of MI for individuals with low LDL cholesterol and low CRP was 2.9 per 1,000 person-years. We adjusted these rates by age and gender according to data from national population-based studies (17,18). The efficacy of statin therapy in AFCAPS/TexCAPS was shown to be similar in subjects with either high LDL cholesterol or high CRP. However, no benefit was shown for individuals with low LDL cholesterol and low CRP in that trial (relative risk of 1.08 for statin therapy versus dietary counseling; 95% confidence interval 0.56 to 2.08) (13).
Risk of stroke
We calculated an overall annual incidence of stroke of 2.0 per 1,000 person-years from a weighted average of the rates observed in the control groups of nine randomized trials in the primary prevention of stroke (0.002) (19–27). We assumed that this rate was applicable to men age 55 to 64 years, and we split this rate into group-specific rates based on the relative risks of stroke by CRP levels estimated from the Physicians’ Health Study (3). We further adjusted the stroke rates on the basis of age- and gender-stratified population data (28). We assumed that LDL cholesterol levels did not predict the risk of stroke (29,30). The efficacy of statin therapy for stroke prevention, on the basis of the CRP levels, is not known. Thus, we assumed the benefit observed in the West Of Scotland Coronary Prevention Study (WOSCOPS) for stroke prevention (10%) was applied to all subgroups (19).
Stroke after MI
The overall risk of stroke after MI was increased according to the rates observed in the Cholesterol And Recurrent Events (CARE) trial (31). We assumed that all patients would receive statin therapy after MI, regardless of their initial treatment strategy, and that statins would reduce the risk of stroke after MI by 22% (31,32).
MI after stroke
The overall risk of MI after stroke was increased according to the rate observed for this subgroup of patients in Clopidogrel vs. Aspirin in Patients at Risk of Ischemic Events (CAPRIE) trial (33). We assumed that all patients with LDL cholesterol ≥149 mg/dl would receive statin therapy after their stroke, and that the relative risks and efficacy estimates for MI by LDL cholesterol and CRP levels were equivalent to those of stroke-free individuals.
We used the Coronary Heart Disease Policy Model to estimate the rate of fatal MI (within one year), adjusted for age and gender, and the subsequent annual coronary heart disease-specific mortality rates, adjusted for age and gender (34). The Coronary Heart Disease Policy Model is a deterministic, population-based, computer stimulation model that incorporates available epidemiologic and clinical data (35).
Stroke-specific mortality, adjusted for age, was calculated based primarily on data from WOSCOPS (19,36). In the absence of published data, we assumed that the stroke fatality rates for patients age 35 to 54 years were the same as for those age 55 years.
Mortality rates due to other causes, specific for age and gender, were based on U.S. life tables (37).
Based on our model, we projected that 58-year-old men with LDL cholesterol <149 mg/dl and CRP ≥0.16 mg/dl would live an average of 19.7 years without treatment and 20.3 years with statin therapy, yielding a life-expectancy gain of 6.6 months. We projected that the life expectancy for 58-year-old women with LDL cholesterol <149 mg/dl and CRP ≥0.16 mg/dl was 24.1 years without treatment and 24.7 years with statin therapy, yielding a life-expectancy gain of 6.4 months. These life-expectancy gains were similar to those observed for 58-year-old men and women with LDL cholesterol ≥149 mg/dl (6.7 months for men and 6.6 months for women). In contrast, statin therapy for men and women in the low LDL/low CRP group resulted in only a modest gain in life expectancy (0.6 months for men and 0.6 months for women).
Projected life-expectancy gains for men and women in the low LDL/high CRP group are shown by age in Table 2. Life-expectancy gains decline with increasing age because of the decline in the number of years of remaining life expectancy. Compared with women of the same age, younger men have a greater gain in life expectancy with statin therapy, reflecting the greater differences in the risk of MI and stroke between men and women at younger ages. However, with increasing age, the benefits in life expectancy observed with statin treatment are slightly greater in women than in men, reflecting the increased rates of cardiovascular disease in postmenopausal women and the longer life expectancy for women overall.
We performed sensitivity analyses to assess whether our results were stable when clinically plausible variations were introduced in our base-case baseline probabilities and assumptions. Our results were most sensitive to the rate of MI and the efficacy of statin therapy in reducing the risk of MI. Figure 1shows a two-way sensitivity analysis of the baseline rate of MI (varied from 0.5 to 3 times the base-case values) and the efficacy of statin therapy for reducing MI (varied from 30% to 60%) in subjects with low LDL cholesterol and high CRP. As the rate of MI increases and the efficacy of statin therapy for reducing MI increases, the life-expectancy gains with statin therapy increase. Based on these two variables, the life-expectancy gains with statin therapy ranged from 2.5 months (least favorable assumptions) to 18 months (most favorable assumptions). The magnitude of this effect was similar for men and women.
Our results were moderately sensitive to the annual mortality rate after MI, the efficacy of statin therapy for prevention of stroke and the annual risk of stroke (Fig. 2). Modifying the other variables in our model within clinically plausible ranges resulted in variations in life-expectancy gains of <2 weeks.
The relative risk of the groups with either high LDL cholesterol or high CRP, compared with the low LDL/low CRP group, was comparable to that in AFCAPS/TexCAPS (13). To create a bias against a benefit for subjects in the low LDL/high CRP group, compared with the high LDL cholesterol group, we performed sensitivity analyses in which the relative risks of MI and stroke for the low LDL/high CRP group were reduced to 1.3 and 1.1, respectively, and the relative risks of MI and stroke for the high LDL cholesterol group were increased to 4 and 2, respectively. This resulted in projected life-expectancy gains of 3.5 months and 3.4 months, respectively, for men and women in the low LDL/high CRP group and gains of 8.2 months and 8.2 months, respectively, for men and women in the high LDL cholesterol group.
Although the analysis from AFCAPS/TexCAPS showed no benefit of statin therapy in the low LDL/low CRP group, if the efficacy of statin therapy for preventing MI in this group was increased to 20%, the life-expectancy gains for subjects with low LDL cholesterol and low CRP would be 1.9 months for both 58-year-old men and women.
We sought to estimate the potential gains in life expectancy achieved with statin therapy for subjects without overt hyperlipidemia but with elevated CRP levels. Our analysis suggests that the projected life-expectancy gain with statin therapy for 58-year-old men and women with LDL cholesterol <149 mg/dl and elevated CRP levels is approximately 6.5 months, which is similar to that for treating patients with LDL cholesterol ≥149 mg/dl.
Our results can be placed in the context of the projected life-expectancy gains demonstrated with other primary preventive strategies (38). As shown in Table 3, the gains predicted with statin therapy for 35-year-old men and women in the low LDL/high CRP group are similar to those previously reported for fundamental interventions in the primary prevention of cardiovascular disease, such as smoking reduction, treatment of hypertension and cholesterol reduction (39), and are greater than those reported for cancer screening programs for the general population (38).
Our results were most sensitive to changes in the baseline rate of MI and the efficacy of statin therapy for preventing MI. Our baseline rate of MI was taken from AFCAPS/TexCAPS (16). The AFCAPS/TexCAPS cohort had low high-density lipoprotein (HDL) cholesterol levels, which represents a potential limitation of our study. Nevertheless, the AFCAPS/TexCAPS cohort was a relatively low-risk group. For example, only 12.5% were current smokers, 2.5% were taking oral treatment for diabetes, 22% were hypertensive and 17% were taking prophylactic aspirin. In comparison, the overall rate of MI was over twofold higher in WOSCOPS (19). As shown in Figure 1the predicted life-expectancy gains would be even larger if a higher risk group, as reflected by a higher baseline rate of MI, was considered.
Our model involves other assumptions. Because the benefit of statin therapy in AFCAPS/TexCAPS were similar in the low LDL/high CRP and high LDL cholesterol groups (13), we assumed an average effect across these groups. However, our sensitivity analyses showed that even if the benefit in the low LDL/high CRP group was reduced to 30%, 58-year-old men and women still gain 4.6 months of life expectancy with statin therapy.
We assumed a class effect for statin therapy in our analysis (10–14). However, if future clinical studies show that some statins have more potent anti-inflammatory effects, or that more aggressive LDL cholesterol reduction leads to increased risk reduction, then our projected life-expectancy estimates of a benefit from therapy would be conservative.
The clinical benefits anticipated with statin therapy for subjects with LDL cholesterol levels <149 mg/dl with elevated CRP levels may apply to a large proportion of adults in the U.S. The target population and projected gains in life expectancy would be similar or even larger if statin treatment was given to those with elevated CRP levels and LDL cholesterol <160 mg/dl, many of whom do not currently meet National Cholesterol Education Program guidelines (1)for statin therapy for the primary prevention of cardiovascular disease. Thus, our analysis suggests that there may be substantial value in screening for CRP in subjects with LDL cholesterol <160 mg/dl, supporting the need for randomized trials to directly address the benefits of statin therapy in individuals without overt hyperlipidemia.
↵* Dr. Ridker is named as co-inventor on a pending patent application, filed by the Brigham and Women’s Hospital, on the use of markers of inflammation of coronary disease.
- Air Force/Texas Coronary Atherosclerosis Prevention Study
- Clopidogrel vs. Aspirin in Patients at Risk of Ischemic Events
- Cholesterol and Recurrent Events
- C-reactive protein
- high-density lipoprotein
- low-density lipoprotein
- myocardial infarction
- West of Scotland Coronary Prevention Study
- Received December 5, 2001.
- Revision received April 2, 2002.
- Accepted April 8, 2002.
- American College of Cardiology Foundation
- Ridker P.M.
- Koenig W.,
- Sund M.,
- Frohlich M.,
- et al.
- Kuller L.H.,
- Tracy R.P.,
- Shaten J.,
- Meilahn E.N.,
- Multiple Risk Factor Intervention Trial
- Danesh J.,
- Whincup P.,
- Walker M.,
- et al.
- Ridker P.M.,
- Rifai N.,
- Pfeffer M.A.,
- et al.
- Horne B.D.,
- Muhlestein J.B.,
- Carlquist J.F.,
- et al.
- Ridker P.M.,
- Rifai N.,
- Pfeffer M.A.,
- Sacks F.,
- Braunwald E.
- Ridker P.M.,
- Rifai N.,
- Lowenthal S.P.
- Sonnenberg F.A.,
- Beck J.R.
- U.S. Department of Health and Human Services
- Wilhelmsen L.,
- Berglund G.,
- Elmfeldt D.,
- et al.
- Frantz I.D. Jr.,
- Dawson E.A.,
- Ashman P.L.,
- et al.
- Stinnett A.A.,
- Mittleman M.A.,
- Weinstein M.C.,
- et al.
- Oster G.,
- Huse D.M.,
- Lacey M.J.,
- Epstein A.M.
- Anderson R.N.
- Tsevat J.,
- Weinstein M.C.,
- Williams L.W.,
- Tosteson A.N.,
- Goldman L.