ASCVD risk algorithm including known ASCVD

Clinical background

The decision on statin intensity in primary prevention requires a quantified risk estimate, not merely a qualitative assessment of risk factors. The 2018 AHA/ACC guidelines summarise this in an algorithm in which the Pooled Cohort Equations (PCE) estimate the 10-year risk of atherosclerotic cardiovascular disease (ASCVD) in patients without known ASCVD, and in which thresholds for low, borderline, intermediate and high risk govern the choice of statin intensity [1]. For patients who already have known ASCVD the decision is simpler: secondary prevention with high-intensity statin therapy is indicated whatever the risk score. The value of the algorithm therefore lies in primary prevention, where it replaces ad hoc assessment with a reproducible, validated risk calculation.

Applying the ASCVD risk algorithm including known ASCVD

The algorithm has two arms. If known clinical ASCVD is present (previous myocardial infarction, stroke, TIA, peripheral arterial disease or revascularisation), the statin intensity for secondary prevention is assigned directly, without a risk calculation. Otherwise the PCE estimates the 10-year risk of ASCVD, defined as non-fatal myocardial infarction, fatal or non-fatal stroke, or death from coronary disease [1].

The PCE rests on a Cox proportional hazards model with separate coefficients for men and women and for Black and White/other individuals. The general structure is:

$$\text{10-year ASCVD risk} = 1 - S_{10}^{\exp\left(\sum_{i} \beta_i \cdot X_i - \sum_{i} \beta_i \cdot \bar{X}_i\right)}$$

where $S_{10}$ is the group-specific baseline survival at 10 years, $\beta_i$ are the coefficients for each risk factor $X_i$ (age, total cholesterol, HDL cholesterol, systolic blood pressure, treatment for hypertension, smoking, diabetes), and $\bar{X}_i$ is the mean of each risk factor in the derivation cohort. The six variables are coded in a sex-specific way, and the coefficients differ between ethnic groups. The PCE was derived from four contemporaneous American prospective cohorts (ARIC, the Cardiovascular Health Study, the Framingham Offspring Study and the Jackson Heart Study), together comprising just over 25,000 individuals recruited from the 1990s onwards [1].

The result is mapped to a statin intensity according to the thresholds of the 2018 guidelines:

Risk-enhancing factors that may justify a statin at 5 to $<7{.}5$ include a family history of premature ASCVD, chronic kidney disease, the metabolic syndrome, a persistently raised LDL-C $\geq 4{.}1$ mmol/L, chronic inflammatory diseases and a raised lipoprotein(a) [1].

Interpretation in practice

The algorithm separates three different situations:

Known ASCVD: Secondary prevention is always indicated. A high-dose statin (atorvastatin 40 to 80 mg or rosuvastatin 20 to 40 mg) or the highest tolerated dose. At very high risk, defined as multiple ASCVD events or one event combined with several risk-enhancing conditions, an LDL-C target of $<1{.}8$ mmol/L (70 mg/dL) is warranted, and adding ezetimibe may be considered if the target is not reached with a statin alone [1].

Diabetes mellitus, age 40 to 75 years: A moderate-intensity statin is indicated without a risk calculation. At high risk (multiple risk factors or age 50 to 75 years) a high-intensity statin is warranted [1].

Primary prevention without diabetes: The PCE estimates the 10-year risk, and the thresholds in the table above govern. At borderline risk (5 to $<7{.}5$), risk-enhancing factors should be evaluated, and if the decision remains uncertain the coronary artery calcium (CAC) score can be used. A CAC of $\geq 100$ Agatston units justifies a statin whatever the PCE risk [1].

Validation and performance

The PCE has been validated in several external cohorts, and the results consistently point to systematic overestimation of risk.

In a large calibration study from Ontario, Canada, comprising 84,617 individuals without known ASCVD (mean age 56.3 years, 56.9% women), a c-statistic of 0.73 was measured. The PCE overestimated the observed 5-year risk by 115%, from a predicted 3.51% to an observed 1.4%. The overestimation varied with age and ethnicity [2].

In an even larger validation from Kaiser Permanente Northern California comprising 307,591 adults without diabetes, aged 40 to 75 years, 2,061 ASCVD events were observed over 1,515,142 person-years. The PCE overestimated the actual 5-year risk in all risk categories. In the highest predicted category ($\geq 5$ 5-year risk), the observed risk was 1.85% against a predicted $\geq 5$. The c-statistic was 0.74 for the whole cohort, 0.72 for women and 0.68 for men. Among 4,242 patients with diabetes, discrimination was poorer (c-statistic 0.64) but calibration somewhat better [3].

The overestimation was consistent across ethnic subgroups (non-Hispanic White, non-Hispanic Black, Asian/Pacific Islander and Hispanic), which is significant because the PCE groups Asian and Hispanic individuals with the White group in its coefficient calculation [3].

Limitations

The algorithm applies only to adults aged 40 to 79 years. Patients with an LDL-C $\geq 4{.}9$ mmol/L (190 mg/dL) should under the guidelines receive a high-intensity statin without a risk calculation, and the PCE must not be used to justify withholding treatment in this group [1]. The same applies to patients with familial hypercholesterolaemia.

The ethnicity coefficients are a recognised weakness. The PCE has only two ethnic categories, Black and White/other, and lacks specific coefficients for Asian, Latin American or Native American populations. This can lead to inaccurate risk estimates in these groups [3]. In 2023 the AHA published the PREVENT equations, which are sex-specific but race-free and which additionally incorporate renal and metabolic markers. PREVENT has not yet been implemented in the ACC/AHA cholesterol guidelines but is a likely future replacement for the PCE [4].

Systematic overestimation of risk may lead to more patients being offered a statin than the actual event rate in modern cohorts warrants. This should be weighed into the clinical discussion, particularly at borderline values around 7.5%, where a decision on a statin can turn on whether the risk is overestimated by a third [2, 3].

References

1. Grundy SM, Stone NJ, Bailey AL, et al. 2018 AHA/ACC/AACVPR/AAPA/ABC/ACPM/ADA/AGS/APhA/ASPC/NLA/PCNA Guideline on the Management of Blood Cholesterol. Circulation. 2019;139(25):e1082-e1143. PMID: 30586774
2. Ko DT, Sivaswamy A, Sud M, et al. Calibration and discrimination of the Framingham Risk Score and the Pooled Cohort Equations. CMAJ. 2020;192(17):E442-E449. PMID: 32392491
3. Rana JS, Tabada GH, Solomon MD, et al. Accuracy of the Atherosclerotic Cardiovascular Risk Equation in a Large Contemporary, Multiethnic Real-World Population. J Am Coll Cardiol. 2016;67(18):2118-2130. PMID: 27151343
4. Khan SS, Coresh J, Pencina MJ, et al. Novel Prediction Equations for Absolute Risk Assessment of Total Cardiovascular Disease Incorporating Cardiovascular-Kidney-Metabolic Health: A Scientific Statement From the American Heart Association. Circulation. 2023;148(24):1982-2004. PMID: 37947094