Commentary

New risk-prediction guideline: Headed in the right direction, but limitations persist

Atherosclerosis remains the leading cause of morbidity and mortality in the industrialized world. As such, all prevention guidelines emphasize that asymptomatic adults should undergo global risk assessment. However, limitations of the Adult Treatment Panel III version of the Framingham risk score for estimating hard 10-year CHD events have been well described in both underestimating and overestimating risk.

Erin D. Michos

Erin D. Michos

In the 2013 American College of Cardiology/American Heart Association Guideline on the Assessment of Cardiovascular Risk, global risk assessment remains endorsed as the first step for adults aged 40 to 79 years. A new 10-year risk score for estimating hard atherosclerotic CVD (ASCVD) events, including CHD and stroke, but not congestive HF or revascularization, was derived using pooled data from NHLBI-funded cohorts, including ARIC, the Cardiovascular Health Study, CARDIA and Framingham. Importantly, there are now separate risk equations for sex and non-white populations.

Limitations of the new risk score

Although I applaud this progress, the new risk score will still result in overtreatment or undertreatment. I outline some concerns regarding pitfalls of the new risk score.

Performance in diverse populations: One notable absence from the pooled cohorts was the Multi-Ethnic Study of Atherosclerosis (MESA), which had not reached 10-year follow-up. MESA is arguably the most racially/ethnically diverse of the cohorts, and important findings from MESA regarding risk prediction conferred by coronary artery calcium (CAC) was not considered for inclusion in the model. When the new risk score was applied to the MESA and REGARDS studies, it performed suboptimally, with C-statistics of approximately 0.6 to 0.7. For comparison, in MESA, the C-statistic for the prediction of CHD events was 0.76 (95% CI, 0.72-0.79) using a traditional risk factor model, but increased to 0.81 (95% CI, 0.78-0.84) with the addition of CAC. Thus, CAC may perform better than the ASCVD pooled risk score alone.

Exclusion of family history in the model: Family history reportedly did not improve model prediction. However, a family history of premature CHD has been shown to be a strong predictor of subsequent CVD events. In contrast, the Canadian Cardiovascular Society recommends that a person’s calculated risk should be doubled when a family history of premature CVD is positive.

Inclusion of stroke: I agree with the importance of preventing disability from total CVD. However, it is uncertain whether this risk threshold will discriminate who will most and least benefit from statin therapy. Only about 40% of strokes are from large-vessel atherosclerotic disease that would be predicted to be most amenable to risk reduction with statins, with the rest being cardioembolic, lacunar or hemorrhagic.

Lowering the high-risk threshold to >7.5%: Surprisingly, the guidelines recommend treating those with completely normal LDL cholesterol values if above this threshold. Nearly all men older than 65 years would crossover this 7.5% threshold, highlighting the fact that this risk assessment is more for populations rather than individual people. Furthermore, elderly individuals with no CAC actually have a lower mortality than younger individuals with high CAC scores.

Subclinical atherosclerosis: The new guideline gives a weak class IIb indication for optional screening tests when risk-based decisions to start pharmacologic therapy are “uncertain.” There is an option to upgrade to a higher risk class for family history of premature CVD, high-sensitivity C-reactive protein ≥2 mg/L, CAC scores ≥300 or ≥75th percentile, or ankle-brachial index <0.9. Carotid intima-medial thickness was not recommended.

Compared with traditional risk factors, CAC clearly outperforms. Individuals without risk factors but elevated CAC have substantially higher CHD and mortality event rates than those who have multiple risk factors but no CAC. In MESA, individuals with zero risk factors and CAC >300 had a CHD event rate 3.5 times higher than individuals with at least three risk factors and CAC score of 0. Individuals with zero CAC scores have very low event rates; thus, the number needed to treat to prevent one event with statin therapy among those with a CAC score of 0 would be prohibitively high. These findings challenge the exclusive use of traditional risk-assessment algorithms for guiding the intensity of primary prevention therapies.

It also is surprising that these screening tools were given equal weight in revising the risk score upward. When compared head to head in the same population, CAC by far is a much more potent risk discriminator than high-sensitivity CRP, with superior net reclassification.

I feel cost and radiation as reasons against a stronger endorsement for CAC have been overly emphasized. Radiation is now approximately 1 mSv, a dose comparable to the mammogram. The EISNER study did not show increased downstream costs with CAC screening; rather, subsequent testing and treatment costs were reduced for those with a CAC score of 0. The issue about incidental findings with CAC screening is valid and will need continued discussion.

A cumbersome risk score: It will be difficult to make pocket cards like the point-scoring system for ATP-III. There is a Web risk calculator, but without being able to do this quickly, many doctors may revert to “counting” individual risk factors, which is a suboptimal way to assess risk.

Ten-year risk vs. lifetime risk: The new guideline gives a modest IIb indication for assessing a 30-year or lifetime ASCVD risk for adults aged 20 to 59 years who are not at short-term risk. It is disappointing that lifetime risk assessment was not given a stronger endorsement. Traditional risk factors often manifest in middle age, when one’s estimated lifespan is at least another 4 more decades based on national averages. Estimating a 10-year risk may give younger individuals false reassurance that they are at low risk for CHD, when in reality this may not be true when considering their entire lifetime.

In support of risk assessment

In summary, I wholeheartedly support the concept of risk assessment and tailoring therapy to those at highest risk. I agree that the new 10-year pooled ASCVD risk score with sex- and race-specific equations is an improvement from the ATP III/Framingham risk score version for hard CHD. However, the absence of MESA in the pooled cohorts and the weak endorsement for incorporating the potent discriminator of CAC in patients without dyslipidemia but having a predicted risk near the arbitrary 7.5% threshold are major limitations that still may be leading us off target. Hopefully, these limitations will be updated in the next version of the pooled ASCVD risk model.

Akosah KO. J Am Coll Cardiol. 2003;41:1475-1479.
Anderson TJ. Can J Cardiol. 2013;29:151-167.
Appelros P. Stroke. 2009;40:1082-1090. 
Blaha MJ. Circ Cardiovasc Qual Outcomes. 2011;4:253-256. 
Blaha MJ. Lancet. 2011;378:684-692. 
Budoff MJ. Am Heart J. 2009;158:554-561.
Cook NR. Circulation. 2012;125:1748-1756.
Goff DC. Circulation. 2013;doi:10.1161/01.cir.0000437741.48606.98.
Goff DC. J Am Coll Cardiol. 2013;doi:10.1016/j.jacc.2013.11.005.
Lakoski SG. Arch Intern Med. 2007;167:2437-2442.
Lloyd-Jones DM. JAMA. 2004;291:2204-2211.
Nasir K. Circ Cardiovasc Imaging. 2012;5:467-473. 
Nasir K. JACC Cardiovasc Imaging. 2012;5:111-118.
Polonsky TS. JAMA. 2010;303:1610-1616.
Silverman MG. Eur Heart J. 2013;doi:10.1093/eurheartj/eht508.
Third report of the National Cholesterol Education Program (NCEP) Expert Panel on detection, evaluation and treatment of high blood cholesterol in adults (Adult Treatment Panel III) final report. Circulation. 2002;106:3143-3421.
Tota-Maharaj R. Eur Heart J. 2012;33:2955-2962. 
Erin D. Michos, MD, MHS, FACC, is assistant professor of medicine in the division of cardiology and the Ciccarone Center for the Prevention of Cardiovascular Diseases at The Johns Hopkins University.

Disclosure: Michos reports no relevant financial disclosures.

Atherosclerosis remains the leading cause of morbidity and mortality in the industrialized world. As such, all prevention guidelines emphasize that asymptomatic adults should undergo global risk assessment. However, limitations of the Adult Treatment Panel III version of the Framingham risk score for estimating hard 10-year CHD events have been well described in both underestimating and overestimating risk.

Erin D. Michos

Erin D. Michos

In the 2013 American College of Cardiology/American Heart Association Guideline on the Assessment of Cardiovascular Risk, global risk assessment remains endorsed as the first step for adults aged 40 to 79 years. A new 10-year risk score for estimating hard atherosclerotic CVD (ASCVD) events, including CHD and stroke, but not congestive HF or revascularization, was derived using pooled data from NHLBI-funded cohorts, including ARIC, the Cardiovascular Health Study, CARDIA and Framingham. Importantly, there are now separate risk equations for sex and non-white populations.

Limitations of the new risk score

Although I applaud this progress, the new risk score will still result in overtreatment or undertreatment. I outline some concerns regarding pitfalls of the new risk score.

Performance in diverse populations: One notable absence from the pooled cohorts was the Multi-Ethnic Study of Atherosclerosis (MESA), which had not reached 10-year follow-up. MESA is arguably the most racially/ethnically diverse of the cohorts, and important findings from MESA regarding risk prediction conferred by coronary artery calcium (CAC) was not considered for inclusion in the model. When the new risk score was applied to the MESA and REGARDS studies, it performed suboptimally, with C-statistics of approximately 0.6 to 0.7. For comparison, in MESA, the C-statistic for the prediction of CHD events was 0.76 (95% CI, 0.72-0.79) using a traditional risk factor model, but increased to 0.81 (95% CI, 0.78-0.84) with the addition of CAC. Thus, CAC may perform better than the ASCVD pooled risk score alone.

Exclusion of family history in the model: Family history reportedly did not improve model prediction. However, a family history of premature CHD has been shown to be a strong predictor of subsequent CVD events. In contrast, the Canadian Cardiovascular Society recommends that a person’s calculated risk should be doubled when a family history of premature CVD is positive.

Inclusion of stroke: I agree with the importance of preventing disability from total CVD. However, it is uncertain whether this risk threshold will discriminate who will most and least benefit from statin therapy. Only about 40% of strokes are from large-vessel atherosclerotic disease that would be predicted to be most amenable to risk reduction with statins, with the rest being cardioembolic, lacunar or hemorrhagic.

Lowering the high-risk threshold to >7.5%: Surprisingly, the guidelines recommend treating those with completely normal LDL cholesterol values if above this threshold. Nearly all men older than 65 years would crossover this 7.5% threshold, highlighting the fact that this risk assessment is more for populations rather than individual people. Furthermore, elderly individuals with no CAC actually have a lower mortality than younger individuals with high CAC scores.

Subclinical atherosclerosis: The new guideline gives a weak class IIb indication for optional screening tests when risk-based decisions to start pharmacologic therapy are “uncertain.” There is an option to upgrade to a higher risk class for family history of premature CVD, high-sensitivity C-reactive protein ≥2 mg/L, CAC scores ≥300 or ≥75th percentile, or ankle-brachial index <0.9. Carotid intima-medial thickness was not recommended.

Compared with traditional risk factors, CAC clearly outperforms. Individuals without risk factors but elevated CAC have substantially higher CHD and mortality event rates than those who have multiple risk factors but no CAC. In MESA, individuals with zero risk factors and CAC >300 had a CHD event rate 3.5 times higher than individuals with at least three risk factors and CAC score of 0. Individuals with zero CAC scores have very low event rates; thus, the number needed to treat to prevent one event with statin therapy among those with a CAC score of 0 would be prohibitively high. These findings challenge the exclusive use of traditional risk-assessment algorithms for guiding the intensity of primary prevention therapies.

PAGE BREAK

It also is surprising that these screening tools were given equal weight in revising the risk score upward. When compared head to head in the same population, CAC by far is a much more potent risk discriminator than high-sensitivity CRP, with superior net reclassification.

I feel cost and radiation as reasons against a stronger endorsement for CAC have been overly emphasized. Radiation is now approximately 1 mSv, a dose comparable to the mammogram. The EISNER study did not show increased downstream costs with CAC screening; rather, subsequent testing and treatment costs were reduced for those with a CAC score of 0. The issue about incidental findings with CAC screening is valid and will need continued discussion.

A cumbersome risk score: It will be difficult to make pocket cards like the point-scoring system for ATP-III. There is a Web risk calculator, but without being able to do this quickly, many doctors may revert to “counting” individual risk factors, which is a suboptimal way to assess risk.

Ten-year risk vs. lifetime risk: The new guideline gives a modest IIb indication for assessing a 30-year or lifetime ASCVD risk for adults aged 20 to 59 years who are not at short-term risk. It is disappointing that lifetime risk assessment was not given a stronger endorsement. Traditional risk factors often manifest in middle age, when one’s estimated lifespan is at least another 4 more decades based on national averages. Estimating a 10-year risk may give younger individuals false reassurance that they are at low risk for CHD, when in reality this may not be true when considering their entire lifetime.

In support of risk assessment

In summary, I wholeheartedly support the concept of risk assessment and tailoring therapy to those at highest risk. I agree that the new 10-year pooled ASCVD risk score with sex- and race-specific equations is an improvement from the ATP III/Framingham risk score version for hard CHD. However, the absence of MESA in the pooled cohorts and the weak endorsement for incorporating the potent discriminator of CAC in patients without dyslipidemia but having a predicted risk near the arbitrary 7.5% threshold are major limitations that still may be leading us off target. Hopefully, these limitations will be updated in the next version of the pooled ASCVD risk model.

Akosah KO. J Am Coll Cardiol. 2003;41:1475-1479.
Anderson TJ. Can J Cardiol. 2013;29:151-167.
Appelros P. Stroke. 2009;40:1082-1090. 
Blaha MJ. Circ Cardiovasc Qual Outcomes. 2011;4:253-256. 
Blaha MJ. Lancet. 2011;378:684-692. 
Budoff MJ. Am Heart J. 2009;158:554-561.
Cook NR. Circulation. 2012;125:1748-1756.
Goff DC. Circulation. 2013;doi:10.1161/01.cir.0000437741.48606.98.
Goff DC. J Am Coll Cardiol. 2013;doi:10.1016/j.jacc.2013.11.005.
Lakoski SG. Arch Intern Med. 2007;167:2437-2442.
Lloyd-Jones DM. JAMA. 2004;291:2204-2211.
Nasir K. Circ Cardiovasc Imaging. 2012;5:467-473. 
Nasir K. JACC Cardiovasc Imaging. 2012;5:111-118.
Polonsky TS. JAMA. 2010;303:1610-1616.
Silverman MG. Eur Heart J. 2013;doi:10.1093/eurheartj/eht508.
Third report of the National Cholesterol Education Program (NCEP) Expert Panel on detection, evaluation and treatment of high blood cholesterol in adults (Adult Treatment Panel III) final report. Circulation. 2002;106:3143-3421.
Tota-Maharaj R. Eur Heart J. 2012;33:2955-2962. 
Erin D. Michos, MD, MHS, FACC, is assistant professor of medicine in the division of cardiology and the Ciccarone Center for the Prevention of Cardiovascular Diseases at The Johns Hopkins University.

Disclosure: Michos reports no relevant financial disclosures.