Risk Assessment: Primary Prevention LDL-C < 190 mg/dL

Introduction

In 2008, the National Heart, Lung and Blood Institute (NHLBI) convened the Risk Assessment Working Group (henceforth the Working Group) and charged it with:

• Developing a quantitative risk assessment approach that could be used to guide care
• Performing systematic reviews to address a small number of critical questions judged to be critical to refining and adopting risk assessment in clinical practice.

The Working Group was subsequently transitioned to the American College of Cardiology and American Heart Association (ACC/AHA) for implementation as the 2013 ACC/AHA risk assessment guideline.

The first critical question addressed by the Working Group was to evaluate the incremental information added by newer biomarkers and noninvasive imaging methods to traditional risk factors. The second question was to evaluate models for predicting longer term (≥15 years) cardiovascular (CV) risk. Their recommendations for risk assessment are summarized in Table 14-1.

The new Pooled Cohort Equations (PCE) for estimating 10-year atherosclerotic cardiovascular disease (ASCVD) developed by the Working Group were recommended in both the 2013 ACC/AHA and the 2018 multi-society cholesterol guidelines.

Note that ASCVD risk should be assessed only in individuals with intermediate density lipoprotein cholesterol (LDL-C) levels <190 mg/dL who are not receiving a statin and who do not have clinical ASCVD, advanced heart failure, or are receiving dialysis.

Clinical Highlight I

• Use the ACC/AHA risk calculator (Pooled Cohort Equations) to estimate ASCVD risk in the general patient population.
• Patients with high socioeconomic or educational status, such as health professionals, may be at lower than estimated ASCVD risk using the Pooled Cohort Equations.
• Asian American and Hispanic American patients may be at lower than estimated ASCVD risk using the Pooled Cohort Equations.
• Long-term risk factor control reduces ASCVD risk.
• Selected characteristics have been shown to indicate ASCVD risk may be higher than that predicted by the Pooled Cohort Equations.

Enlarge 

Lifetime Risk

After systematically reviewing a number of long-term epidemiologic studies, the 2013 ACC/AHA risk assessment guideline made recommendations to:

• Assess traditional ASCVD risk factors every 4-6 years in adults aged 20-79 years who are free of ASCVD: age, sex, total cholesterol, HDL-C, systolic blood pressure, use of antihypertensive therapy, diabetes and current smoking.
• Consider assessing 30-year or lifetime risk in adults 20-59 years of age who are free of clinical ASCVD and are not at high 10-year ASCVD risk (Class of Recommendation [COR] IIb, Level of Evidence [LOE] C; Expert).

Information on lifetime risk may be useful for motivating healthy lifestyle changes. Epidemiologic evidence and an understanding of the progression of atherosclerosis through the lifespan suggest that early interventions to improve lifestyle habits and control risk factors have major effects on atherosclerosis progression.

In individuals ≥60 years, 10-year and lifetime risk begin to converge so the calculator estimates only 10-year ASCVD risk in this group.

10-Year ASCVD Risk

The 2013 ACC/AHA risk assessment guideline systematically reviewed epidemiologic studies of near-term (10-year) risk prediction. They determined that new risk prediction equations were needed for two major reasons:

• To better predict the burden of ASCVD by including stroke as well as myocardial infarction (MI)
• To predict ASCVD risk in non-white populations.

Data from five community-based epidemiologic cohorts sponsored by the NHLBI were used to develop four Pooled Cohort Equations for non-Hispanic White and African American women and men. There were insufficient data to develop risk prediction equations for other racial/ethnic groups. They made these recommendations:

• It is reasonable to estimate 10-year ASCVD risk starting at age 40 years and every 4 to 6 years thereafter (COR IIa, LOE B). More frequent risk estimation may not reflect the full impact of changing risk factors.
• Use the race- and sex-specific Pooled Cohort Equations to predict 10-year risk of a first hard ASCVD event in non-Hispanic African American and non-Hispanic White patients aged 40-79 years (COR I, LOE B).
• In other populations, consider using the Pooled Cohort Equations (COR IIb, LOE C; Expert). ASCVD risk may be lower in Hispanic-American and Asian-American populations, and higher in American Indian, South Asian and Pacific Islander populations.

These equations are an important advance from the earlier Framingham equation since they more completely estimate the risk of cardiovascular disease (CVD) by including stroke, which constitutes a larger burden in younger African Americans and older White women. The Pooled Cohort Equations perform well in the general US population and have been validated in the REGARDS study, a contemporary randomly selected population-based epidemiologic study of African American and White men and women in the United States.

When applied to more selected, often lower-risk populations, the Pooled Cohort Equations may overestimate risk. This phenomenon was observed in evaluations of health professionals in an aspirin trial by Ridker and Cook, and in the MESA study, which had a large proportion of low-risk Chinese and Hispanic participants in addition to improved risk factor control following noninvasive imaging testing during the study.

Refining 10-Year ASCVD Risk Estimation

The 2013 ACC/AHA risk assessment guideline also systematically reviewed newer risk markers that might contribute to refinement of the 10-year ASCVD risk estimate from the Pooled Cohort Equations. When assessing biomarkers and tests, the 2013 ACC/AHA risk assessment guideline panel considered availability, cost, assay reliability and the risks of the test or downstream testing. Statistical evaluations of the incremental information provided by the new test beyond that of traditional risk factors were considered. These recommendations (summarized in Table 14-2) are as follows:

• If, after quantitative risk assessment, a risk-based decision is uncertain, assessment of one or more of the following indicators of increased ASCVD risk may be considered to inform decision making (COR IIb LOE B; this was a weaker recommendation because the papers reviewed did not assess performance using the 2013 ACC/AHA cholesterol guideline risk cut-points):
• Family history of premature ASCVD (first-degree relative: male <55 years or female <65 years)
• High sensitivity CRP ≥2.0 mg/L
• Coronary artery calcium (CAC) score ≥300 Agatston units or 75th percentile for age/sex/ethnicity
• Ankle-brachial index <0.9
• LDL-C ≥160 mg/dL was added by the 2013 ACC/AHA cholesterol guideline panel (and maintained in the 2018 multi-society cholesterol guideline) as an indicator of increased ASCVD risk due to genetic contribution and likely long-term exposure to elevated cholesterol levels.
• Do not routinely measure carotid intimal medial thickness (CIMT) due to concerns about measurement quality and minimal provision of incremental risk prediction information (COR III: No benefit, LOE B).
• Not recommended due to insufficient information to determine the incremental contribution to risk assessment: Apolipoprotein B, chronic kidney disease, albuminuria and cardiorespiratory fitness (no recommendation).

Enlarge 

Performance of 2013 ACC/AHA Risk Indicators

Due to the lack of consistently collected data in the five cohorts, the 2013 ACC/AHA risk assessment panel was unable to derive quantitative estimates of the incremental information provided by any of these biomarkers or tests to the Pooled Cohort Equations.

However, an analysis of nondiabetic participants in the Multi Ethnic Study of Atherosclerosis (MESA) cohort found that 57% of ASCVD events occurred in individuals with <7.5% 10-year ASCVD risk. Among those with <7.5% 10-year ASCVD risk, coronary artery calcium (CAC) reclassified the greatest proportion to higher risk (6.8%), with an average 10-year ASCVD risk of 15%. Family history of premature ASCVD reclassified 4.6% with an average 10-year ASCVD risk of 15%. High-sensitivity C-reactive protein (hs-CRP) ≥2 mg/L reclassified 2.6% with an average 10-year ASCVD risk of 10%. Few individuals were reclassified by LDL-C ≥160 mg/dL (0.5%, 5% 10-year ASCVD risk) or ankle-brachial index >0.9 (0.6%, 9% 10-year ASCVD risk).

Evaluating New Biomarkers for Clinical Practice

On a population basis, the major “traditional” risk factors (e.g., age, sex, total cholesterol, high-density lipoprotein cholesterol (HDL-C), smoking, systolic blood pressure and antihypertensive treatment and diabetes) predict about 80% of near-term and lifetime ASCVD. However, for the individual patient, the positive predictive value is low. The hope for new biomarkers and noninvasive imaging of atherosclerotic burden is that they can better individualize risk prediction so that those most likely to benefit are treated, and those unlikely to benefit are not treated. The test should be safe and relatively inexpensive, including both the test itself and any downstream testing that arises. Additional requirements for a clinically useful new biomarker/test are summarized in Table 14-3.

Three types of statistical tests are used to evaluate the incremental information provided by a new biomarker: discrimination, calibration and reclassification. All three are required before considering a change in clinical practice. The highest standards for the usefulness of a new biomarker or test are whether it changes clinician and/or patient behavior and is cost-effective.

Enlarge 

Discrimination

Discrimination measures how well an equation predicts events and is typically statistically assessed using the c-statistic or receiver-operator curve (ROC). The c-statistics for the Pooled Cohort Equations are provided in Table 14-4. Performance is excellent (>80%) in White and African American women and still very good at >70% in White and African American men. There was insufficient information from the five cohort studies to evaluate the additive value of new biomarkers to the Pooled Cohort Equations. However, it does not appear that a panel of biomarkers improved coronary artery disease (CAD) risk prediction with the Framingham Score. The level of B-type natriuretic peptide, CRP, homocysteine, renin and urinary albumin-to-creatinine ratio were independently associated with increased CAD risk. Adding all five biomarkers did not improve discrimination (Figure 14-1).

Most risk factor associations are in the range of a relative risk or odds ratio of 1.5-2.5. For a test to substantially improve discrimination, the magnitude of independent association should be an odds ratio or relative risk of ≥9 (Figure 14-2). The reason for this is that most of the biomarkers measured to date are closely intercorrelated—they are essentially measuring the same causal pathways already represented by the traditional risk factors. Therefore, to be of value, biomarkers should represent a new causal pathway for ASCVD. Lipoprotein (a) (Lp(a)) may be one such factor, as described below. Ongoing investigations are using proteomics, lipidomics and metabolomics to identify new causal pathways that could be used as drug targets or risk markers.

Enlarge 

Enlarge 

Enlarge 

Calibration

Calibration usually divides a new cohort into deciles of event rates and then compares the observed event rate to the predicted event rate using a risk prediction equation derived from another cohort. Figure 14-3 shows the calibration of the Pooled Cohort Equations for the REGARDS study participants. REGARDS is a contemporary population-based cohort of 30,000 randomly selected White and African American women and men. When applied to nondiabetic participants who would meet the criteria for primary prevention risk assessment (LDL-C 70-189 mg/dL and not taking a statin), the observed and predicted ASCVD event rates are very similar in the group with <10% 10-year ASCVD risk for whom the decision to use statins is being made.

Enlarge 

Reclassification

Reclassification is used to determine if a biomarker/test improves the accuracy of risk prediction for patients on the margin. Few studies to date have re-evaluated biomarkers using the new 2013 ACC/AHA cut-points of 7.5% or 5% 10-year ASCVD.

A study evaluating Lp(a) has found that Lp(a) improved the correct up- and down-classification of intermediate-risk persons based on the Reynolds Risk Score in a European cohort (Figure 14-4). Although the statistics are somewhat difficult to interpret, the central illustration in the figure nicely shows that few cases are reclassified as low risk and more non-cases are correctly up- and down-classified with the addition of Lp(a) (see Lp(a) below for further discussion).

Enlarge 

Coronary Artery Calcium

Coronary Artery Calcium (CAC) is currently the best tool now available for refining Pooled Cohort Equation risk estimates (see above) in middle-aged and older individuals. CAC has been extensively studied in selected clinical populations, with fewer data from prospective cohort studies. A CAC score of 0 identifies a patient group at lower than expected 10-year ASCVD risk and low 15-year mortality.

The CAC score adds information to the risk prediction using risk factors at the time of CAC scoring. However, long-term prospective studies have also shown that CAC reflects risk factors through the lifespan. The 2018 multi-society cholesterol guideline considers CAC scoring a reasonable (COR IIa) method of risk reclassification in patients with intermediate and selected patients with borderline 10-year ASCVD risk, particularly in deciding whether to initiate statin therapy. With a CAC score of 0 and in the absence of higher risk conditions, it is reasonable (COR IIa) to not initiate statin therapy. In patients 55 years of age or older with a CAC score of 1-99, it is reasonable (COR IIa) to start statin therapy. Finally, statin therapy is reasonable (COR IIa) in all patients with a CAC score of 100 or higher (or in the 75^th percentile or higher), regardless of age. The 2022 ACC Expert Consensus Decision Pathway on nonstatin use broadly agrees with the 2018 multi-society cholesterol guideline, providing the following recommendations:

• CAC score of 0: Consider deferring statin therapy and re-measuring CAC in 3-5 years
• CAC score of 1-99 and <75^th percentile for age, sex, or race/ethnicity: Consider moderate-intensity statin therapy; if <30% LDL-C reduction and LDL-C level ≥100 mg/dL, increase to high-intensity statin therapy
• CAC score of ≥100 or ≥75^th percentile for age, sex, or race/ethnicity: Consider moderate- to high-intensity statin therapy; if LDL-C reduction is below that expected for the given statin regimen and LDL-C level ≥70 mg/dL, may be reasonable to consider ezetimibe
• CAC score of ≥1000: Consider high-intensity statin therapy; if <50% LDL-C reduction and LDL-C level ≥70 mg/dL, may be reasonable to consider ezetimibe and – if LDL-C goals are still not met – may consider an anti-PCSK9 monoclonal antibody.

CAC should be used with caution for down-classification of ASCVD risk before age 50 years, since a younger person with multiple or severe cardiovascular risk factors, such a familial hypercholesterolemia, may have an excessive burden of uncalcified plaque yet still be at high lifetime ASCVD risk. This “low” risk (at least in the next 10 years) person may experience twice the relative risk reduction from statin therapy as higher risk persons, and statins have been shown to be cost-effective even in populations at very low ASCVD risk.

Race/Ethnicity

Although fewer data are available, CAC is associated with increased ASCVD risk in South Asian Americans (India, Pakistan, Bangladesh), similar to the increased risk observed in White patients. In Multi Ethnic Study of Atherosclerosis (MESA), African Americans had less CAC than White patients, which does not reflect the increased risk of ASCVD events in African Americans (Figure 14-5). Rates of both CAC and ASCVD events are lower in Hispanics and Chinese Americans. Therefore, race/ethnicity calibrated risk estimation for CAC should be used.

Enlarge 

Sex

Fewer data are available for women regarding the value of CAC for risk ASCVD prediction or reclassification. CAC is a stronger predictor of CAD than stroke events, and therefore likely predicts ASCVD risk better in White men than in women and African Americans, who are more likely to experience stroke than White men. Although the lifetime ASCVD risk is high for most women, ASCVD events occur later in life in women than men, and women with similar risk factor profiles to a similarly aged man have a lower atherosclerotic burden.

Clinical Highlight II

• Consider obtaining a CAC score in patients 40-75 y with LDL-C levels of 70-189 mg/dL when the decision about initiating statin therapy for primary prevention is uncertain.
• A CAC score of 0 favors withholding statin therapy.
• A CAC score of 1-99 in patients ≥55 y favors initiating moderate-intensity statin therapy.
• A CAC score of 100 or above (or in the 75^th percentile and above) favors initiating moderate- or high-intensity statin therapy.
• A CAC score of 100 or above (or in the 75^th percentile and above) favors initiating moderate- or high-intensity statin therapy and consideration of ezetimibe if LDL-C reduction goal is not met.
• A CAC score of 1000 or above favors initiating high-intensity statin therapy and consideration of ezetimibe followed by a PCSK9 monoclonal antibody if LDL-C reduction goal is not met.
• Do not obtain a CAC score in patients who are already on statin therapy.

Lipoprotein (a)

Lp(a) is a lipoprotein with both atherogenic and thrombotic potential (see Lipid and Lipoprotein Basics). Data from a contemporary US cohort, the MESA study, suggest that Lp(a) cut-offs of 50 mg/dL for White and Hispanic adults and 30 mg/dL for African Americans indicate increased ASCVD risk. Lp(a) was not associated with increased ASCVD risk in Chinese Americans.

Lp(a) is not normally distributed in the population (Figure 14-6). Relatively few White and Hispanic Americans have Lp(a) levels ≥50 mg/dL, while almost half of African Americans have Lp(a) ≥30 mg/dL. Lp(a) ≥30 mg/dL improved net reclassification of CAD cases in African Americans but not in the other racial/ethnic groups, perhaps because of inadequate sample size. Unfortunately, small sample size also precluded evaluation of reclassification for the 2018 multi-society cholesterol guideline (Table 14-5). The 2018 cholesterol guideline considers elevated Lp(a) levels (≥50 mg/dL) an ASCVD risk enhancer and is thus potentially useful in the consideration of whether to initiate statin therapy in primary prevention patients 40-75 years of age with LDL-C levels 70-189 mg/dL and an intermediate or borderline 10-year ASCVD risk.

The Lp(a) assay used in MESA measured Lp(a) mass concentration with a latex-enhanced turbidimetric immunoassay (Denka Seiken, Tokyo, Japan). This assay uses an analytical approach that circumvents the problems in measuring Lp(a) by using multiple calibrators that control for varying apo(a) sizes (187 to >662 kDa) among individuals and isoform-insensitive antibodies that are not directed to the repeating element within apo(a), Kringle 4 type 2 (Figure 2-5).

Enlarge 

Enlarge 

Enlarge 

Clinical Highlight III

• Might consider measuring Lp(a) levels in selected patients for whom a treatment decision is uncertain.
• If measured, a Lp(a) level of 50 mg/dL or higher is considered an ASCVD risk enhancer and may inform the decision to initiate statin therapy in patients with intermediate or borderline 10-y ASCVD risk.