Primary Prevention of Atherosclerotic Cardiovascular Disease: Title and subTitle BreakThe High Public Burden of Low Individual Risk

Physicians and many laypeople are well aware of the substantial risks posed by asymptomatic atherosclerosis.¹ Among 50-year-old adults enrolled in the Framingham Heart Study, the lifetime risk for developing symptomatic disease was 52% in men and 39% in women.² For many patients the first clinical manifestation is a potentially catastrophic event, such as stroke, myocardial infarction, or sudden death.¹ Despite numerous advances in the understanding of the epidemiology and prevention of clinical atherosclerosis, the public disease burden remains high and may, with the current obesity epidemic, be increasing.³ Primary prevention, that is preventing or delaying clinical disease among asymptomatic individuals, therefore remains an issue of major public health interest.

Guidelines for clinicians caring for asymptomatic adults focus on assessment of individual risk, lifestyle interventions, and, when appropriate, pharmacological strategies, including aspirin⁴ and lipid-lowering drugs.⁵ One recommended approach^{4 ,6} starts with estimating risk of clinical events using the Framingham Risk Score. For individuals with an estimated 10-year risk greater than 20%, aggressive interventions are appropriate. For those at intermediate risk (10%-20%), clinicians should consider testing for high-risk but asymptomatic atherosclerosis with high-sensitivity C-reactive protein (CRP),⁷ stress testing, electron-beam computed tomography, measurement of ankle-brachial index, or ultrasound to measure carotid intima-media thickness (CIMT).^{6 ,8} Persons deemed to be at low risk (<10%) should receive conservative management focusing on lifestyle interventions.

In this issue of JAMA, the METEOR trial investigators⁹ pose a serious challenge to this risk-based paradigm. The investigators screened 5751 asymptomatic adults to identify 984 who had substantially increased CIMT, but who otherwise appeared to be at low risk of atherosclerotic disease based on Framingham Risk Score of less than 10% or age (mean, 57 years) as the sole risk factor. The study participants were randomly assigned to receive either 40 mg/d of rosuvastatin or placebo and were followed up for up to 2 years with serial carotid imaging. CIMT increased by 0.013 mm/y among placebo patients, whereas there was essentially no change among patients assigned to rosuvastatin. Even though this difference was statistically significant, the investigators were not able to demonstrate regression of disease with rosuvastatin.

At first glance, the METEOR findings suggest that there may be a role for routine arterial imaging even among adults deemed to be at low risk for atherosclerosis by the Framingham Risk Score and that for those with increased CIMT rosuvastatin therapy may be indicated to reduce the rate of disease progression. This is a radically different approach to primary prevention than what is recommended by current guidelines.^{5 ,10} Some cardiovascular researchers advocate routine imaging of nearly all adults followed by statin therapy in select patients.¹¹ Thus, the METEOR report raises a number of fundamental questions for the medical and public health communities.

First, are medical strategies appropriate for primary prevention in low-risk individuals? Fifteen years ago, Rose¹² described a 2-pronged approach to primary prevention that envisioned distinct “high-risk” and “population” strategies. The individualized high-risk strategy, familiar to clinicians, calls for risk assessment with diagnostic testing followed by medical interventions in select patients. Affected patients realize substantial benefits and because of their increased baseline risk are willing to consider prolonged medical therapy with attendant adverse risks.

The major, yet highly underappreciated, problem with the high-risk strategy is that it has a relatively minor effect on overall population incidence of disease. “Low-risk” does not mean “no risk.” Some low-risk individuals, like those enrolled in METEOR, have subclinical and potentially dangerous atherosclerosis. Because there are so many low-risk individuals in the population, they generate the bulk of public clinical risk. This is particularly true for complex, multifactorial diseases such as atherosclerosis, for which continuous variables including blood pressure and cholesterol levels conspire to increase risk across a wide spectrum.¹³ Population-based studies have demonstrated that most of the public burden of disease can be attributed to low-risk individuals with relatively ”normal” levels of cholesterol or blood pressure.¹³ According to recent National Health and Nutrition Examination Survey data, among healthy adults aged 20 to 79 years, 85% had low-risk Framingham scores while only 2% had high-risk scores. Low-risk individuals were responsible for approximately two thirds of the overall population risk.¹⁴

Rose proposed that the low-risk population would be best served by population-based strategies outside the realm of traditional medical testing and therapy.¹² Examples include public education programs,¹⁵ mass screening, bans on public smoking¹⁶ or trans-fat,¹⁷ and adolescent physical activity initiatives.¹⁸ Even minor effects of population-based strategies, Rose argued, would result in a substantial decrease in clinical disease incidence because of the huge number of low-risk individuals.¹² Unfortunately, the literature on the effects of population-based strategies has mixed outcomes. Some programs, such as smoking bans,¹⁶ do appear to substantially reduce clinical events, but other community or lifestyle-based interventions have not worked well.¹⁵ This then raises the question as to whether a medical strategy might be applicable to the low-risk population.

Second, are surrogate end points appropriate in primary prevention trials? The problem with assessing the effectiveness of a medical approach for primary prevention in low-risk populations is that these populations, by definition, have a low rate of events. Investigators need enormous sample sizes to demonstrate an effect of any intervention, whether medical or community-based. The temptation is therefore strong to use surrogate end points such as CIMT. Surrogate end points have been criticized, however, after numerous bad experiences whereby agents that improved surrogate end points yielded no benefit or were even found to cause harm when tested for their ability to prevent clinical events. Classic examples of this include vitamin E¹⁹ and postmenopausal hormone therapy.²⁰

Are surrogate end points ever appropriate? According to Prentice²¹ a surrogate end point may be useful if there is a clear association between treatment and the surrogate end point, treatment and the clinical end points, the surrogate end point and the clinical end point, and most importantly, a demonstration of a clear link between intervention-induced changes in the surrogate end point and changes in the clinical end point. This last requirement is the most difficult to demonstrate.²²

Third, are CIMT measurements clinically meaningful? A number of trials have demonstrated that statins prevent cardiovascular events²³ and reduce progression of carotid atherosclerosis²² ; these studies have primarily involved intermediate or high-risk populations. There is only limited evidence that the change in CIMT induced by statin therapy accounts for the reduction in atherosclerotic events.²² The magnitude of change in CIMT required to translate into a reduction in clinical events (difference of 0.012 mm/y between groups) is similar to the results observed by the METEOR investigators.²² Nonetheless, the evidence is not yet sufficient to comfortably conclude that a reduction in progression of atherosclerosis by carotid imaging in low-risk individuals will necessarily translate into a reduction in clinical events.

The METEOR trial was not powered to assess the effect of rosuvastatin therapy on clinical events. Among 984 patients, there were only 6 ischemic events, which curiously all occurred in the rosuvastatin group. Although the ASTEROID trial demonstrated that rosuvastatin induced regression of coronary atherosclerosis,²⁴ to date there have been no randomized trials showing this particular statin to reduce risk of clinical events. One large trial involving 15 000 patients with elevated high-sensitivity CRP levels but normal LDL-cholesterol levels is in progress.²⁵

Two other concerns with the METEOR study deserve mention. Because the investigators had 2 distinct objectives—demonstrating disease regression and showing superiority to placebo—they randomized study participants using an unusual 5:2 ratio. The objective of demonstrating atherosclerosis regression was not met. The better outcome in terms of carotid intima thickness for rosuvastatin compared with placebo may have been partially due to the necessarily increased variance in the relatively small placebo sample. Another concern is that a fair number of enrolled patients failed to complete the protocol and were lost to follow-up, a problem not unique to this trial.²⁶ To the investigators' credit, sensitivity analyses were performed to assess the effects of missing data, but a higher rate of follow-up clearly would have increased the credibility of the findings.

And finally, where to go from here? Should low-risk individuals undergo routine arterial imaging followed by statin therapy when evidence of asymptomatic disease is discovered? On the basis of current evidence, including the METEOR trial, the answer is clearly no. However, like a shooting star in the night, the METEOR findings reflect a warning of problems lurking “out there” in the vast person-time space of the low-risk population. Epidemiological data have provided overwhelming evidence that low-risk populations are the source of most clinical disease.¹³ The METEOR investigators have now provided biological evidence for a plausible but unproved strategy.

The METEOR trial should be just the beginning, a stimulus for the medical and public health communities to work together. Ambitious event-based randomized trials involving large numbers of patients and communities must be done. While these trials will be difficult and expensive, even greater and less desirable challenges will occur by choosing to ignore the enormous public burden of clinical atherosclerosis arising from the population of low-risk individuals.