EBM Tools for Practice: Statins for Primary Prevention in Women: How an Individual-Benefit Model Could Expand Treatment to Hypercholesterolemic Women Under Age 65 with 10-Year ASCVD Risk Less Than 7.5%

Evidence in support of statins for the primary prevention of atherosclerotic cardiovascular disease (ASCVD) in women was sparse until recently.¹  However, meta-analyses have documented that statins significantly reduce ASCVD event rates in women with elevated baseline risk.^2,3,4

Yet a large subset of women under age 70 with hypercholesterolemia but no other risk factors does not reach the current threshold for statin eligibility, i.e., 7.5% absolute ASCVD risk in 10 years, even when total cholesterol elevation is severe.⁵ The reason is that 10-year risk prediction as currently recommended by the American College of Cardiology/American Heart Association (ACC/AHA) for primary prevention patients ages 40 to 75 years without diabetes or LDL-C > 190 mg/dl,⁶ and optionally by the National Lipid Association for those with 2 risk factors,⁷ is weighted heavily by age and less so by the absolute level of LDL-C. Thus, older women (and men) with relatively low LDL-C usually will be statin-eligible because 10-year ASCVD risk almost always exceeds 7.5%, while those under age 65 with LDL-C levels up to 189 mg/dl but 10-year risk under 7.5% will be ineligible. The fact that almost one-third of ASCVD events in women (and one-half in men) occur before the age of 65 highlights the fundamental limitation of this approach.⁸

As the optimal timing of statin initiation for primary prevention is being debated, a new strategy that accounts for the expected individual benefit from statin therapy based on clinical trial data has emerged that would expand treatment to a significant number of non-elderly, primary prevention women with hypercholesterolemia.

The individual-benefit model for statin allocation first was illustrated by Soran, et al., in 2014.^9,10 The authors showed mathematically that the number of events prevented by statin therapy is a function of both 10-year ASCVD risk and baseline LDL-C. As such, those with a higher baseline LDL-C but lower 10-year risk can obtain an equivalent or greater benefit from statin therapy, i.e., a higher absolute risk reduction (ARR) and a lower number needed to treat (NNT), as those with a higher 10-year risk but lower LDL-C. That is because the magnitude of LDL-C lowering is a function of both statin potency and baseline LDL-C. At a given dose of a statin that lowers LDL-C by 43% (e.g., atorvastatin 20 mg), those with a higher LDL-C will experience a greater magnitude of LDL-C lowering than those with a lower baseline LDL-C. This will translate into a greater relative risk reduction (RRR) because the hazard ratio (HR) based on Cholesterol Treatment Trialist Collaboration meta-analytic data is 0.78 for every 1 mmol (39 mg/dl) of LDL-C lowering that occurs.¹¹

Therefore, those with a higher baseline LDL-C who achieve a 2 mmol or 78 mg/dl reduction in LDL-C will have an HR of 0.78 x 0.78, or 0.60, or a 40% relative RR, while those with an even higher LDL-C who achieve a 3 mmol or 117 mg/dl reduction  in LDL-C will have an HR of 0.78³ or 0.47, or a 53% relative RR. This then can be multiplied by absolute risk in 10 years to calculate the number of events prevented per 100 patients over 10 years, as follows: # of events prevented/100 = [1 - 0.78ⁿ] x absolute 10-year risk, where n = LDL-C decrease in mmol/L.

For example, if ASCVD risk is 10% over 10 years and baseline LDL-C is relatively low and drops by 1 mmol/L or 39 mg/dl with statin therapy, the number of CVD events prevented per 100 people treated for 10 years is (1 - 0.78³) × 10 = 2.2. ARR in 10 years as a percentage then is 2.2% or 0.022, and the NNT to prevent 1 event is 1/0.022, or 45. (Alternatively, the NNT = 100/2.2). On the other hand, if 10-year ASCVD risk is just 5% but LDL-C is high at baseline and lowered by 3 mmol/L, the number of events prevented per 100 people = (1 - 0.783 ) x 5, or 2.6, and the ARR over 10 years will be 2.6% and NNT 38, an even more cost-effective undertaking.

As seen in Table 1, it then follows that, at a given level of absolute risk, NNT  by pre-treatment LDL-C, and guideline algorithms should consider adoption of the NNT in statin decision-making. Moreover, abolishing LDL-C therapeutic goals puts at a disadvantage those people with higher LDL-C levels who derive greater benefit from treatment aimed at specific targets.⁵

In 2016, Thanassoulis, et al., modeled statin eligibility based on individual benefit vs. risk-based strategies using a nationally representative sample of 2,134 primary prevention individuals age 40-75 from the National Health and Nutrition Examination Survey (NHANES) for the years 2005 to 2010.¹²

An individualized ARR of >2.3% over 10 years (corresponding to a NNT of 44 or less) was used as an acceptable cut-off for statin eligibility. The authors plotted the 10-year predicted risk and baseline LDL-C levels required to produce a constant individual benefit at pre-specified ARR levels (See Figure 1). As shown, a minimum ARR of 2.3% in 10 years (blue or middle curve in Figure 1) was met across a range of baseline LDL-C and 10-year risk levels, from an LDL-C of 189 mg/dl with a 10-year risk of <5% to an LDL-C of 70 mg/dl with a 10-year risk of just under 10%. On this basis the authors showed that the benefit model applied to their NHANES sample would extend statin treatment to 24.6 million Americans, 9.5 million more than the current ACC/AHA risk-based approach, and 60% of those newly statin-eligible would meet criteria for at least 1 statin trial (Figure 2). For the entire 24.6 million identified via the benefit-based model, the average 10-year ASCVD risk, 10-year ARR, and NNT were 10.9%, 4% and 25 (range 9-44), respectively, compared to 13.9%, 4.8% and 21 (range 9-44), respectively, for the 15 million targeted via the risk-based approach. As expected, the 9.5 million additional statin-eligible Americans were significantly younger (mean age 55.2 versus 62.5 years) and had a significantly higher baseline LDL-C (140 vs. 133 mg/dL). Moreover, a significant proportion (about one-third) were women, and their proportional increase in statin eligibility was at least as high as for men.¹³