Should CV Risk Equations Add Lp(a)?

Whether to measure Lp(a), a lipoprotein associated with increased cardiovascular risk, and how to use that information in risk assessment is a hot topic in cardiovascular medicine.

The American Heart Association recently introduced the Predicting Risk of Cardiovascular EVENT (PREVENT) equations — an update to the pooled cohort equations. Neither risk calculator includes values for Lp(a).

A recent study looked at whether adding Lp(a) to the PREVENT equations would improve risk prediction. It found that including the lipid parameter yielded a modest improvement at a population level but appeared to be more useful for personalized risk assessment, particularly among lower-risk individuals.

“Our results validate the PREVENT equations on a population level and show that they perform well, both in people with and without high levels of Lp(a),” lead author of the study, Harpreet Bhatia, MD, University of California San Diego, told Medscape Medical News.

“While I think our results would not support adding Lp(a) to the PREVENT equations, they confirm that on an individual-patient basis Lp(a) can add information,” he commented.

Bhatia explained that the PREVENT risk equations are going to be the future paradigm of risk stratification in the primary prevention setting in the US, eventually replacing the pooled cohort equations which have been used for many years.

The reason that Lp(a) wasn’t included in either risk score, he suspects is because the datasets on which the risk equations are based would not have these values available.

The PREVENT equations removed the consideration of race or ethnicity, as there is now acknowledgement that race is more of a social construct, but it is known that Lp(a) levels vary by ancestry, he said.

Bhatia believes that Lp(a) should be routinely tested at least once in all adults; “For those of us who practice preventive cardiology and lipidology, it can alter our clinical management.”

Current American Heart Association/ American College of Cardiology cholesterol guidelines from 2018 do not recommend universal testing of Lp(a) but include it as a risk enhancing factor; updated guidelines are expected within the next year. Guidelines from the European Society of Cardiology and Canada and the US National Lipid Association do recommend measuring Lp(a).

For the current study, Bhatia and colleagues examined data from the Multi-Ethnic Study of Atherosclerosis, a US study of 6670 people started in the year 2000, and the UK Biobank, a study of over 500,000 individuals from the UK started around 2006. Participants had no known cardiovascular disease at baseline and most had an Lp(a) measurement.

Bhatia noted that his study used risk thresholds established for the pooled cohort equations, but it has not yet been decided what the cut points will be for the PREVENT equations. Low risk was defined as < 5%; borderline as 5%-7.5%; intermediate as 7.5%-20%, and high as ≥ 20% predicted 10-year risk for cardiovascular disease.

‘PREVENT Equations Generally Do A Good Job’

“Essentially, what we saw was that the PREVENT equations generally do a good job of putting people into those 10-year risk categories across the board,” Bhatia said.

However, they also found that within each category and across the board, if Lp(a) was high, then the cardiovascular risk was increased compared to individuals with a lower Lp(a) and sometimes quite significantly increased.

The researchers also tried to establish whether Lp(a) could improve risk prediction on top of the PREVENT equations, using the Net Reclassification Index (NRI) which looks at the percentage of people who would be reclassified based on the new model (the proportion who move up minus the proportion who move down). They found that Lp(a) levels led to a modest improvement in risk prediction according to the NRI.

In terms of atherosclerotic cardiovascular disease (ASCVD), including elevated Lp(a) on top of the PREVENT equations appropriately reclassified about 6% of people. With regard to coronary artery disease, which Lp(a) is most strongly associated with, the NRI was about 8%.

Another measure of how well a new model predicts risk — the C index — found that the addition of Lp(a) did not significantly modify results.

“Our results suggest there does seem to be some improvement in risk prediction with Lp(a) for some individuals, particularly those at lower cardiovascular risk,” Bhattia said, an observation he described as “intriguing”.

He does not believe that new equations are needed that incorporate Lp(a).

He pointed out that statin therapy for prevention is more strongly recommended in intermediate/higher risk patients, with a weaker recommendation for those at lower risk when there’s the presence of an additional risk enhancing factor.

“It may be that someone at low risk in the equations with an elevated Lp(a) may be eligible for starting statins.”

Bhatia already considers Lp(a) levels in this way in his clinic. He said that Lp(a) testing is simple and widely available, and the majority of people will only need to be tested once in their lifetime.

He explained that most people who have low or high levels would stay in those categories long term, while people who have intermediate or borderline levels (30-50 mg/dL or 75-125 nmol/L) may need repeat testing if something changes that can affect Lp(a) levels longer term such as going through menopause, or the development of kidney or thyroid disease.

Lp(a) Testing Worthwhile

In an editorial accompanying the publication, Donald M. Lloyd-Jones, MD, Framingham Center for Population and Prevention Science, Boston University, Framingham, Massachusetts, and Amit Khera, MD, University of Texas Southwestern Medical School, Dallas, said the results are a useful validation of the PREVENT risk equations in contemporary broad populations and large real-world clinical samples.

In terms of whether Lp(a) should be incorporated directly as a variable in the PREVENT equations, the editorialists have a similar view to Bhatia. “That appears unnecessary,” they wrote.

But, like Bhatia, they believe that Lp(a) should be measured once in everyone to help understand and individualize risk.

“Lp(a) is indeed a risk-enhancing, likely causal, factor for ASCVD. Its absence does not exonerate traditional risk factors, but its presence can amplify and personalize that risk, and help guide clinicians and patients regarding use and intensity of preventive therapies,” they concluded.

Also commenting on this latest study, was Nathan Wong, PhD, from the University of California Irvine School of Medicine.

He told Medscape Medical News that the analysis shows that the PREVENT risk score predicts ASCVD outcomes similarly in those with and without elevated Lp(a) levels. The stronger prediction of Lp(a) in lower risk people “argues for the need to promote increased screening in the broader population, including those at lower risk” he said noting that most recommendations in the past have focused on people at higher risk such as those with a personal history of ASCVD.

Wong agreed with Bhatia that the value of Lp(a) is more at the individual level than at the population wide level. But he does believe a risk score incorporating Lp(a) could be helpful for personalizing treatment strategies in certain individual patients, particularly for those with elevated Lp(a) levels who may not already be identified as high risk.

Indeed, Wong and colleagues recently published such a score and showed that incorporating Lp(a) into ASCVD risk prediction models developed using a real-world clinic population moderately improves performance over 10 years, with good generalizability when applied to other US population cohorts.

In that paper, a 25 mg/dL increment in Lp(a) was associated with a 23% increased risk for incident ASCVD. Levels ≥ 75 mg/dL conferred a near two-fold greater risk for ASCVD, including a 2.5-fold greater risk for stroke compared with Lp(a) levels < 25 mg/dL.

They also demonstrated that adding Lp(a) to the pooled cohort equations ASCVD risk calculator correctly reclassified 45% of borderline-intermediate risk patients who experienced incident ASCVD as high-risk. However, about 24% who did not experience events were incorrectly reclassified high-risk (for an NRI of 21%). He cited a use case of a Black man aged 65 years with an Lp(a) of 80 mg/dL and a 10-year ASCVD risk of 18% without considering Lp(a) who would be up-stratified to 24% after factoring in Lp(a).

“Based on current guidelines, this person would now be clearly recommended for a statin to lower his ASCVD risk which may not have been as certain based on the risk not incorporating Lp(a),” he said.

Recent studies have also shown identification of elevated Lp(a) levels can result in greater use of lipid-lowering therapy, he added.

Wong also agreed with Bhatia on the distinction between risk prediction in populations vs individuals.

“We don’t practice medicine on populations. We practice it on individuals and for certain individuals a risk score that incorporates Lp(a) can reclassify their risk category significantly dependent on how high their Lp(a) is,” he stated.

Bhatia reported receiving consulting fees from Abbott, Arrowhead, Kaneka, and Novartis outside the submitted work. Wong reported receiving research support through his institution from Amgen, Novartis, and Regeneron, is on an advisory board for Amgen, and is a consultant and speaker for Novartis and a consultant for Ionis. Lloyd-Jones and Khera reported no disclosures.