CV Risk Assessment: A Process, Not a Calculation

An Interview With Donald Lloyd-Jones

Interviewer: Michael J. Blaha, MD, MPH; Interviewee: Donald M. Lloyd-Jones, MD, ScM, FACC


November 16, 2018

Editorial Collaboration

Medscape &

Michael J. Blaha, MD, MPH: Hi. I'm Mike Blaha from Johns Hopkins Ciccarone Center for the Prevention of Heart Disease. I'm here in Chicago at the 2018 American Heart Association (AHA) Scientific Sessions, where new guidelines [for cholesterol management][1] have been rolled out.

I'm joined by Dr Donald Lloyd-Jones from Northwestern, who was lead author on the risk-assessment portions of the new guidelines.[2] Today we are going to talk about risk assessment in the setting of the new guidelines—some of the things that have changed and some of the things that have stayed the same. Let's start with a general overview of risk assessment and where we stand now in the new guidelines.

Approach to the 2018 Guidelines

Donald M. Lloyd-Jones, MD, ScM, FACC: Building on the changes that were introduced in the 2013 guidelines, we really tried to flesh out the details of how risk assessment should proceed. I would encapsulate that very simply as: Risk assessment is a process; it's not a calculation. There are really three steps: Estimate, personalize the risk, and reclassify the patient if you need to.

Blaha: That is a great point. Personalization is a major feature of the discussion in the new guidelines. Let's start with the atherosclerotic cardiovascular disease (ASCVD) risk calculator and the pooled cohort equations. I understand that we are using the same pooled cohort equations as before. Would you like to talk a little about that?

Lloyd-Jones: Sure. As a part of the process of the guidelines, we did a number of systematic reviews. We were interested in finding out how the risk equations from 2013, the pooled cohort equations, performed in different types of populations. We identified 23 papers that were published validating those equations in different settings. To summarize, they are pretty well calibrated in the broad US clinical population, especially at the decision threshold of around 7.5%. In some subpopulations, though, they clearly do overestimate, and I would characterize those groups as people with higher socioeconomic status and those who are more routinely engaged with the health system. We also found that the risk equations underestimate the risk in groups with lower socioeconomic status and in people who have chronic inflammatory conditions like HIV or rheumatologic conditions.

Blaha: In my reading of the guidelines, there are some new risk groups that you are talking about, right? There is low risk, borderline risk, intermediate risk, and high risk. Would you want to share with us those cut-points and the thinking behind that?

Lloyd-Jones: We are talking about that first step of estimating risk. Once you have done the calculation with the tools that are available from American College of Cardiology or AHA online or as apps, you put the patient in a bucket. The low-risk group is < 5%, borderline is 5%-7.5%, intermediate-risk (the group we are really focusing on) is 7.5%-19.9%, and high-risk is ≥ 20%. Again, we are only talking about primary prevention.

Personalization of the Risk

Blaha: I'm a big fan of the intermediate-risk concept. We encounter this group clinically very commonly and are a little uncertain about risk. There are some features of the new guidelines that I really like; one is the addition of things called "risk-enhancing factors." These are additional things that you could consider when you are thinking about where that person's risk really is. Could you tell us about risk-enhancing factors?

Lloyd-Jones: Now we are kind of transitioning to phase 2: personalization of the risk. We know that risk scores are developed in populations, but you and I are doctors and we have individual patients sitting in front of us. How do I take that risk score and make it live in context of this patient sitting in front of me? Personalization first involves engaging the patient in understanding what it is about them, specifically, that would be of interest that might shade our estimate of their risk, particularly if it would shade it upwards. Are they of South Asian ancestry? Do they have a strong family history of premature coronary disease? Do they have chronic kidney disease? Is it a woman who may have undetected risk because she had a history of gestational diabetes or hypertensive disorders of pregnancy? While we didn't recommend measurement of specific blood biomarkers, we gave guidance for what kind of thresholds would be used to determine that the patient might be a little bit higher risk if those are measured.

Biomarkers and Coronary Artery Calcium Scoring

Blaha: The guideline clearly states not to measure all of these things, but certainly to consider the things you mentioned. Plus, if there are biomarkers available that might shift someone's risk, you can personalize your approach using those. Lipoprotein(a) [LP(a)] was mentioned in the guidelines. Some people are measuring that nowadays in people with a family history. What are the cut-points for LP(a) in the guidelines?

Lloyd-Jones: For the first time in these guidelines, we have a thoughtful approach to how to interpret both apolipoprotein B (apo B) and LP(a). There have been a lot of calls to incorporate those. We are now in a place where the assays are standardized enough that we understand better how to use them. Specifically, if the apo B level is > 130 mg/dL or if the LP(a) level is > 50 mg/dL for men—and for women if it's above 50 mg/dL—and they are hypercholesterolemic, those are clearly indicators that this patient is towards the higher end of the spectrum.

Blaha: Another part of personalization was the recognition that the coronary artery calcium (CAC) score was perhaps the best single test to add on to initial risk assessment, both when trying to figure out whether it's truly low risk or high risk. Of course, the calcium score is a noncontrast, cardiac-gated CT that can be done in 10 or 15 minutes, with only about 1 mSv of radiation. It has an accessible cost for most people, usually around $75-$200. This is a new part of the guideline; it's a Class IIa recommendation. Tell us some thinking about including calcium scoring and how to use it once you get that test.

Lloyd-Jones: Now we are transitioning from that second phase (personalization) to a third phase, which might be necessary for some patients. If you think that some of those risk enhancers apply to your patients who are intermediate-risk or borderline-risk (5%-7.4%), we recommend (Class IIa) that if there is clinical uncertainty or if the patient is still undecided about whether they want to commit to lifelong therapy, a CAC score is clearly the best test to reclassify that risk and decide whether they are in a benefit group or are not yet in a benefit group. Kudos to you and your group at Hopkins for leading the way in helping us understand within the context of the pooled cohort equations how CAC can classify people.[3] Maybe the most important data for people to hear is that among that intermediate-risk group (7.5%-19.9% risk), 50% of those people have no coronary calcium and a very low event rate over the ensuing 10 years. If you choose to reclassify with coronary calcium and you get that CAC score of zero, we recommend deferring the initiation of statin therapy because those are such low-risk patients and they are not as likely to benefit.

If the CAC score is > 100 Hounsfield units (HU), we say that is clearly a benefit group and we recommend initiation of a statin. Between 1 and 99 HU is still a bit of a gray zone; you haven't really reclassified the risk in a meaningful way. There is atherosclerosis. If the patient is older than 55 years, you probably lean towards starting a statin, but it's really a discussion at that point. If you decide not to do a statin today, it's reasonable to repeat the CT scan perhaps in 5 years and see if there has been a significant change.

Blaha: You don't have to stick with this one measurement. Five years down the line, if the measurement converts to non-zero, you can reenter that risk discussion.

Lloyd-Jones: One caveat to that: If you do start a statin, we specifically recommend against remeasuring the coronary calcium because you have used it in the way it should be used to make the decision. We know that statins can accelerate calcification even though they are stabilizing the underlying disease.

Blaha: Statin therapy actually increases the density of that calcium, and the Agatston score could go up. So it's not the right application for the test.

Case Study: Patient With Borderline Risk and Risk Enhancer

Blaha: Let's close out with a simple case to see if we can apply some of these principles. Let's start with a South Asian man with borderline risk. The pooled cohort equations come out at 5%-6%. The fact that he is South Asian is a risk enhancer. Where do we start with that patient in that risk discussion?

Lloyd-Jones: It's a really interesting case in a particularly interesting population. We know that, as a group, they tend to be a little bit more insulin resistant, they tend to have higher LP(a). So this might be one of those situations where my yield of a test like an LP(a), especially if there is a family history, might be particularly high. I would work with him to find out what is driving his risk. Is it his cholesterol or is it his blood pressure? What about his lifestyle? Always check lifestyle first. I think we need to reiterate that as often as possible.

Blaha: Even if the calcium score is zero, you still have to push lifestyle.

Lloyd-Jones: First, we identify the factors that seem to be driving his risk. Second, are there other features about him that would push us closer to that 7.5% range? I would be looking for those. And as part of this discussion, you have to assess the patient's desire to prevent a heart attack or stroke. If it's low and they are very medication averse, okay, [maybe don't recommend a statin]. If it's high and they want to do anything, I'm close to a level with this patient where I would probably recommend that we start a moderate-intensity statin. See if he tolerates it; see how he does. Try to get a good result and go from there.

Blaha: Let's say that he's what I call a "statin-reluctant" patient. He's not statin intolerant, he's reluctant—"I'm not so sure about my risk. And I'm not so sure that you are sure about my risk." Let's say that one of these risk enhancers, like LP(a), comes back borderline. Would you consider a calcium score in that case?

Lloyd-Jones: Absolutely. I think this is most useful when the patient is statin reluctant or if the clinician is uncertain. It pushes us off the fence. We learn so much and we make smarter decisions. I would walk him through the discussion and say, "I think we would both be more comfortable in our decision if we knew whether you have atherosclerosis or a significant burden." If he's willing, I would talk about the low radiation exposure, which is about the same as a mammogram. I would talk about the out-of-pocket costs. (We hope the guidelines may push a change in that.) Then I would say, "With those results in our hands, we are going to be able to make smarter decisions for you, whether we are deferring for today or whether we do think you need this. You can be more comfortable that we are committing you to something that is going to work."

Blaha: I love these guidelines. I'm very supportive of all of the changes. I love the intermediate-risk group and the idea of personalization. I love the calcium scoring in appropriate populations and considering lifestyle with everybody.

Thanks for joining us. I hope you have learned a lot about the risk-assessment approach in the new guidelines.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.