Deepak L. Bhatt, MD: Hello, I am Deepak Bhatt from Boston and it is my pleasure to moderate this session on "Ischemic Heart Disease: Mitigating the Risk of Sudden Cardiac Death." With me today to discuss this topic are Roxana Mehran from Mount Sinai Medical Center in New York and Bill Weintraub from Christiana Care Health System in Newark, Delaware.

Each year, over 1 million patients have a myocardial infarction (MI) with many more patients hospitalized for unstable angina (UA). It is estimated that there are close to 400,000 deaths each year in the United States that occur as a result of ischemic heart disease. And there are over half a million stent procedures performed on patients, many of whom have acute coronary syndromes (ACS), such as an acute myocardial infarction (AMI), but who remain at risk for recurrent events, even if they have no or limited symptoms after revascularization.^[1] Just how to manage these patients entails educated decisions based upon an assessment of risk, and that is part of our task today.

We plan to discuss a few different ways of risk-stratifying patients, something that remains a particular challenge in terms of trying to predict mortality. As they say, prediction is difficult, especially of the future, and that is especially true for mortality. Accurate risk stratification after primary percutaneous coronary intervention (PCI) for ST-segment elevation myocardial infarction (STEMI) is quite important in guiding future patient management. There are a number of different risk scores and models, but we are going to focus on 2 particular ones: the CADILLAC risk score and the more recent National Cardiovascular Data Registry model. I cannot think of 2 better people than our discussants today to get into both of these different models and consider on how to best risk-stratify patients and then treat those at high risk.

Perhaps, Roxana, we can start with you and the CADILLAC trial. If you could briefly remind everyone about the trial design and what the variables were, and then discuss how the risk score was calculated.

Roxana Mehran, MD: Thank you, Deepak, for allowing me to discuss this important topic. As you stated, prediction of mortality after an acute MI is an important one. If we could accurately risk-stratify patients, perhaps we can better impact outcomes. In CADILLAC, 2082 patients with acute MI were randomly assigned to stent vs balloon angioplasty.^[2] This is an older study and the stents were all bare metal stents, with or without abciximab. This was a great study, actually, to use as a basis for a developmental database for a risk-stratification algorithm for patients with AMI.

What we call the CADILLAC risk score was developed from the variables using the database from the CADILLAC trial and it was then validated using the Stent-PAMI trial.^[3] This is an older study with 900 patients undergoing primary PCI comparing stenting vs balloon angioplasty.

The variables were based first on univariate predictors and then in the final model for both 1-year and 30-day mortality, left ventricular ejection fraction (LVEF) less than 40%; presence of renal insufficiency at the time of STEMI; the Killip class presentation of 2 or 3; final Thrombolysis in Myocardial Infarction (TIMI) flow, which was suboptimal 0 to 2; age greater than 65; presence of anemia; presence of 3-vessel disease; and the burden of coronary disease at the time of presentation. All of these were important in predicting the risk of mortality. These were added together and given integer scores. The variable that was most important was LVEF less than 40% and this was given a score of 4.

This really had a great predictability, both in the 30-day and 1-year mortality with C statistics of about 0.8 -- 0.82 -- so it provides a very good predictive model that can be used at the time of STEMI presentation. It is very simple. You do not need important baseline angiographic predictors, but certainly you can use some of these parameters to predict morality.

Dr Bhatt: That is really great. How about glomerular filtration rate (GFR)? Was that in the score?

Dr Mehran: Presence of renal insufficiency was not included at the time. Renal insufficiency was any creatinine greater than 1.5, so it was not an estimated GFR (eGFR). Now as you know, we are asking to look at eGFR and a creatinine of 1.5 equated to patients with eGFR of 45 and below.

Dr Bhatt: How about other things like the location of the MI, an anterior-wall MI vs an inferior-wall MI?

Dr Mehran: MI location was included in the univariate predictors, but it did not pan out, in fact, because LVEF was more predictive; an LVEF of less than 40% was much more predictive than the location of the MI. In the multivariate model, that is the one that gained the highest integer score of 4.

Dr Bhatt: Maybe the LVEF is incorporating what prior scores had split out as the heart rate and the blood pressure had been shown to be powerful predictors in other models. Maybe the ejection fraction (EF) sort of incorporated the patients whose left ventricle (LV) had taken a big hit.

Dr Mehran: It points out the importance of performing some modality to assess LV function during that first hospitalization of a patient with a STEMI. I think LV function is a very important predictor of how these patients do down the line. We know that recurrent rehospitalizations and the first 30-day readmissions are very related to the baseline or the discharge EF.

Dr Bhatt: That is a great point.

Dr Mehran: So, it is an important variable to assess at the time of STEMI.

Dr Bhatt: How about diabetes?

Dr Mehran: Diabetes did not pan out, believe it or not, in this model in the multivariate predictors. It was there in the univariate predictors. We did not have that many diabetic patients. There were about 22% or so in the CADILLAC study.

Dr Bhatt: It is interesting that diabetes was not significant. Bill, what about the CathPCI Registry? That has given us a lot of insight into a number of different issues, but specific to predicting risk and predicting survival, what can you tell us from that important registry?

William S. Weintraub, MD: Deepak, thank you for asking me represent the National Cardiovascular Data Registry (NCDR) in discussing our prediction models.

The ACC CathPCI Registry is largely an in-hospital registry. We needed to get long-term follow-up data. We did that by linking the registry to the Center for Medicare& Medicaid Services (CMS) database to get long-term survival data. We were able to do that after 3 years.

Our patients are from the years 2004 through 2007, and we have over 340,000 patients in the model.^[4] We divided it into a derivation model and a validation cohort. That allows us to do the modeling in the most appropriate way. Then we divided the patients into those without STEMI and those with STEMI, allowing us to have 2 different models. Then we looked at risk factors for less than a month, 1 month, 12 months, and greater than 12 months. We had 24 different variables in the models.

What we found was that severity or acuteness at the time of presentation were the strongest predictors, especially in the first month. If we look in our model with STEMI, patients in cardiogenic shock had a hazard ratio of over 4 and it was similar in the patients without STEMI. How sick the patients were initially was the most important thing. However, I agree with Roxana that it is very important also to look at LV function. We found that ejection fraction was a strong predictor throughout, but was most powerful in the first month.

Dr Bhatt: That is really interesting. In terms of predicting mortality, then, it sounds like this is a good model. Are you able to specifically predict whether the mortality is from heart failure or from arrhythmia?

Dr Weintraub: No, we cannot do that in this model, although we know, especially early on, that the patients have a high probability of dying from arrhythmic death.

Dr Bhatt: With these 2 ways of risk-stratifying patients, it actually is quite useful to determine that a patient is at particularly high risk post PCI -- or I suppose, for that matter, particularly low risk -- but from a physician perspective, we tend to focus on those who are at high risk. What do you think we can do to further improve their risk beyond just angiotensin-converting enzyme (ACE) inhibitors, angiotensin receptor blockers (ARBs), β-blockers, dual antiplatelet therapy, and blood pressure and lipid control? What are the things we can do in terms of drugs or devices that you think are novel or worth considering more than we currently use?

Dr Mehran: There are guidelines for treatment of these patients, especially when the ejection fraction has been disturbed or is very low or less than 40%, or 35%.^[5] Clearly, these patients can benefit from ARBs and β-blockers. Beyond that, the question is whether or not the use of implantable defibrillators is going to be important in these patients. We all know that post STEMI, there is a change in the EF associated with the total remodeling process that needs to happen. The big question that comes back is in the guidelines: does there have to be a waiting period before you actually assess the LVEF before you determine whether an implantable defibrillator should be placed in the patient for prevention of sudden cardiac death? I always wonder how many patients we lose in that interim, and how can we better assess those patients?

The data are not very good, for how many patients actually have sudden cardiac death at home during that waiting period? I believe it is 90 days before we can make that assessment to actually place an implantable defibrillator.^[5]

Dr Bhatt: Yes, that is really a great point, and that is a tough situation, especially for a high- risk patient who has a big MI where the EF appears to have taken a big hit. Of course, even out to 3 months post PCI, as you know, there can be recovery of LV function. Nice work has documented that and in part that is why guidelines and even coverage decisions have been based on waiting a period of time before putting in an implantable defibrillator.^[7,8] However, a proportion of those patients will continue to have low EF and will be at risk during that waiting period. That is a potential vulnerable period. Bill, what do you think? What can we do for those sorts of patients who are at high risk? Certainly, they ought to be on good pharmacotherapy. One pharmacotherapy I did not mention previously is aldosterone antagonists, which I think tends to be underutilized in patients who are post MI and with heart failure.

In terms of defibrillators and optimizing their timing and use, do you have any thoughts?

Dr Weintraub: I do. First let me say, I agree with the thoughts about the importance of pharmacotherapy and it is not just enough to know that a patient should be getting these. We have to follow through and make sure patients actually do get the pharmacotherapy they need, because we know that many patients actually do not. We do not really live up to the guidelines by applying them in practice. There is a gap there.

In terms of risk, it is clear that patients with repressed EF post-MI will benefit from defibrillators. The clinical trial data are clear on that.^[8,9]

It is also clear that left ventricular function can return That is why the national coverage determination for ICDs recommends this waiting period.^[10]

That is a period of great danger, that first month or so. The models that we have from the NCDR clearly point to this, because the patients have greater risk in these high-risk subsets in the first month, so that cardiogenic shock in the STEMI model in the first month has a hazard ratio of 4.

That drops after a year all the way down to 1.2 with an ICD and is no longer even statistically significant. Our patients are really at great risk in the first month even though many of them will have return of LV function. There is a treatment gap in which I think there is opportunity to save lives.^[4]

Dr Bhatt: Yes, you raise a good point. I had mentioned the risk in those patients who have a low EF, whose EF remains low, but there are also those patients, even though there is recovery of EF that might be substantial. For a period of time, the EF is, in fact, diminished and they are at risk then. For both those types of patients, there can be that window of vulnerability. Do you think wearable defibrillators are a potential option there?

Dr Weintraub: That is one of the few options we have to try and prevent sudden cardiac death in those patients. It would be great to have more data and there is an ongoing trial to look at that right now, but it is a very difficult trial, very difficult to enroll patients. Right now that is our treatment option and we do want to see more data.

Dr Mehran: What do you think about home monitoring? I think one of the issues about these patients is not that they get lost to follow up, but adherence to the medication and guideline therapies. There is still room for improvement on that front. Some of the home monitoring devices are smart phones, and they could have some potential impact on reducing readmission rates, and also surveying those patients who are at high risk. I know that there are ways to even do some electrocardiographic 2- or 3-lead home monitoring with the devices that are out there. What do you think about those in the future?

Dr Bhatt: There is a lot of promise to that idea of getting patients more involved with their healthcare and using technology to augment that. One of the problems with sudden cardiac death is it is hard to predict it until it happens, other than by relatively crude measures like low EF. That sort of technology is for someone who might have transient atrial fibrillation (AF). It could be potentially useful to determine if there is a recurrence of AF after an AF ablation. In terms of further risk-stratifying for sudden cardiac death, I am not sure there are that many additional electrocardiographic markers that a type of app-based approach would really help with.

Dr Weintraub: Roxana raises a very good point. I do not think we have explored that as fully as we should. There are companies that are using radiotransmittable waveforms that would allow us to investigate that in much greater detail. There is tremendous potential to find people at greater risk.

Dr Bhatt: Anything that further enhances our ability to risk-stratify and target therapies to high-risk patients would be cost-effective. Bill, you have done a ton of research on cost-effectiveness.

Dr Weintraub: Oh, absolutely. Right now, the way we approach cost-effectiveness in clinical trials most of the time is look at everybody, so everybody gets everything. That is something that we cannot afford to do anymore. People have thought about personalized medicine, trying to risk-stratify making sure people get what they need is somehow at odds with evidence-based medicine in clinical trials. That is not the case. In fact, we need to make these things work together.

Dr Mehran: Something that has bothered me a lot, and I would love to hear both of your views if time allows us, is when a patient that comes in with an EF that is quite low, say even 20%. Why do we have to wait that 90 days? Can remodeling actually bring this patient to a place where they could be protected beyond the 90 days? Those are the cases that really bother me a lot, and those are the patients in whom based on the guidelines, we cannot just go in and throw in a defibrillator.

Dr Weintraub: It is not just a matter of guidelines. It is also national coverage determination. This is coming from CMS. If you were to put in an ICD without a national coverage determination and bill for it, there can be difficulty. The Justice Department is actually weighing in on this right now, so it is fairly uncertain how this will turn out. I think where CMS is going to back off are with people who have known low EF and then come in with a recurrent MI. To wait to put in an ICD in those patients would be irresponsible. I think we will see some changes in the coverage determination.

Dr Bhatt: Yes, that is a very subtle and important point you raise, Bill. Those are really different patients because their EF after their MI is certainly not going to get better than what it was before, if it was already low.

Dr Weintraub: We see a lot of patients like that. This is not an unusual scenario. We have patients with an EF of 30% or 25%, and the fact that they have a relatively mild STEMI hardly changes anything; they have really been at risk all along.

Dr Bhatt: But certainly the data do not support putting in an implantable defibrillator very early on after a de novo MI, at least in terms of that being superior to waiting. There are patients who do recover substantial amounts of myocardial function post primary PCI, even out a couple of months after the procedure. That is one of the great benefits of primary PCI.

Well, it has been a terrific discussion, really insightful comments from both of you. This is a topic that clinicians have to weigh everyday: how to best risk stratify patients coming in with a STEMI getting primary PCI and trying to decide what is best to do to further decrease their risk. I would like to thank the audience members for participating in this program. Hopefully, it was interesting and educational.

You may now take the CME posttest by clicking on the Earn CME Credit Link. Please also take a moment to complete the program evaluation. Thank you very much for joining us.