MANDEEP SINGH: Hi, everyone who have joined us virtually in this ongoing quarterly webinar series. I am your host and moderator, Mandeep Singh. I am a professor, and coming live to you from Mayo Clinic in Rochester, the Department of Cardiovascular Disease. And I have a distinct pleasure and honor to introduce-- actually reintroduce my teacher, my friend, and colleague Dr. Bernard Gersh.

People who have joined before had enjoyed his talk on pharmaco-invasive therapy. All these topics that we do on our webinar are archived and recorded, so you can enjoy them later. But today, he'll be speaking on coronary revascularization in patients with severe left ventricular dysfunction, and merely talking about the status of viability testing. Dr. Gersh is well known to all of us. The last I checked was 1,300 publications. His H-index is through the roof. He's very well known in the academic circle.

And he has been the editor, co-editor, reviewer in almost all the journals in the world. It is my distinct pleasure to introduce Professor Gersh, who will be talking on this topic. Dr. Gersh, welcome.

BERNARD GERSH: All right. Thank you very much, Mandeep, for that nice introduction. And as was pointed out, I'm going to talk about two aspects of revascularization. One is the results of coronary revascularization in patients with severe LV dysfunction. And then the second is really to take a look at the role of viability testing, and is it still viable? The prognosis of patients with LV dysfunction and coronary disease is really illustrated in this slide. The major determinant of mortality in patients with coronary disease is the severity of LV dysfunction.

And in patients with LV dysfunction, the major determinant of prognosis is the severity of coronary artery disease and ischemia. And the results of revascularization are in some ways a paradox. There is an increased periprocedural risk, but that is counterbalanced by the potential to reduce late mortality. So a little bit the greater the risk, the greater the reward. And to really understand that, one needs to realize that if you have severe LV dysfunction, clearly it will increase periprocedural risk.

If there are any complications from the procedure, they will be magnified. On the other hand, because we are addressing as a target the severity of ischemia and coronary artery disease, if one can correct that, that may have an effect on reducing late mortality. Somewhat greater effect than in patients, many patients, with normal LV function. And this has been documented in the original CASS study of coronary artery bypass surgery and medical therapy, published in the very early 1980s.

Overall trial results were neutral. This was a neutral trial. But if you look at the subset of patients within an ejection fraction between 35% and 50%, and those with multivessel disease and LV dysfunction, had a distinct survival advantage with coronary bypass surgery, and really I think another example of the potential benefits of correcting ischemia in patients with left ventricular dysfunction. It's been shown more recently in the ischemia trial that was published last year.

Now, in the ischemia trial, there was a small subset of patients, about 7.7%, who had heart failure with reduced ejection fraction, heart failure with preserved ejection fraction, and then a group with reduced ejection fraction but no heart failure. And if one looks at the results, again I should emphasize the ischemia trial was a neutral trial, for these endpoints-- CV death, nonfatal MI, hospitalization for unstable angina or heart failure, resuscitated cardiac arrest.

The yellow dots are the conservative arm, and the orange dots are the invasive arm in this trial. And these are patients who had heart failure and LV dysfunction. So it's this two groups of patients here. And amongst these, there is a distinct benefit for the revascularization strategy in these sicker patients with heart failure and LV dysfunction, in a trial which overall was neutral.

And then if you look at those who had no heart failure with LV dysfunction, shown in the solid lines, there's no difference at all in outcomes, and the interaction was an almost significant point of 0.055. And if one looks at the primary endpoint, which is this endpoint, all cause mortality and CV mortality in patients with prior heart failure, dotted arm is conservative-- the dotted line is conservative therapy, the orange is invasive.

And if one looks at an ejection fraction of somewhere around 55% to 60%, what you see is a benefit where the curves crossover, with a lower mortality in the invasive arm, and the lower CV mortality and a lower primary endpoint. The STICH trial was the first trial, however, to address this prospectively. And this was a trial confined to patients with an ejection fraction of 35% or less. Unlike what I've just shown you in the previous five, which was a subset analysis from a trial of patients who by and large had no heart-- by and large had normal LV function.

The median in ischemia-- I think the median ejection fraction was 60%. 1,200 patients, coronary artery disease, multivessel coronary disease, EF less than 35% or 35% or less. A very difficult trial to do, and a lot of credit goes to the investigators. And if one looks at death from any cause as the primary outcome, initially the trial was a trial of three to five years, but subsequently there have been-- the follow up has gone out to 10 years.

And if you look at death from any cause, which was the primary outcome, at three to five years there was a difference in favor of bypass surgery that was magnified when you get out to 10 years. And at 10 years, the hazard ratio of 0.84, a 16% reduction in mortality, P 0.025. And what I really wanted to point out again is this paradox.

Very early on there actually was a higher mortality in the bypass arm. Not significant, but trending that way. And then, after two years, the curves diverge in the other direction with a benefit for revascularization. Number needed to treat is surprisingly small. It's only 14 patients. Death from cardiovascular causes, hazard ratio 0.79. Same thing, slightly higher cardiovascular mortality early on in orange, and then a distinct benefit for bypass surgery. P equals 0.006.

And if you look at all cause death and cardiovascular hospitalization, there's a 28% reduction in favor of bypass surgery. Again, it took a period of time, but the benefit really only became clear one to two years after the initial operation. P less than 0.001. Now, this is a subset analysis from the STICH trial that makes, I think, an important clinical point.

And that is, if one looks at all cause mortality, cardiovascular mortality, and all cause mortality, and CV hospitalization-- and on this axis is age, and this is the hazard ratio-- you see a very distinct benefit, and quite a large benefit, all cause mortality early on in younger patients under the age of 50, both with CV mortality and all the endpoints.

Now, this doesn't mean that there's no reduction in mortality in older patients, but, basically, once you get over 65, 70, the risk is greater. There still is a benefit in cardiovascular mortality. There still is a benefit on mortality and CV hospitalization, but the benefits are smaller and really emphasizes that when one deals with revascularization in sick patients with LV dysfunction and heart failure, you have to look at the entire patient very carefully, because the benefits and the risks are obviously go up with age.

And that does not mean that older patients may not benefit, but patient selection needs to be very stringent and careful. Now, the mechanisms of improved prognosis following revascularization. Potential benefits, I mean, the obvious one is an improvement in resting ejection fraction, a recovery in regional function, and this is why people benefit. And this may apply in some patients, but there are other explanations.

We may-- one benefit from revascularization is to reduce inducible ischemia and repetitive stunning, which is a phenomenon people with LV dysfunction and coronary disease. That, in turn, could improve diastolic function. In turn, it could also improve ventricular remodeling and the symptoms of heart failure, and improve prognosis through those mechanisms. The other is one I've always believed, that ischemia is a cause of malignant ventricular arrhythmias.

That acute ischemia or even chronic ischemia can cause sudden cardiac death and prevention of ventricular arrhythmias may be another potential benefit from revascularization. And the one that is gaining attention now is the concept that the benefits of opening an artery may have effects that are independent of ischemia and viability. But what they do is prevent recurrent events, and, in particular, recurrent myocardial infarction from other vessels that are not obstructed, but will be obstructed in the future.

And that I think is a very plausible hypothesis that has recently been floated in the literature. And this is an interesting analysis from Dr. Julio Penza based on the 10-year follow-up of the STICH trial. If you look at the change in ejection fraction on this axis with medical treatment alone and bypass surgery plus medical treatment, now the numbers are small and I accept that, and this is in patients without viable myocardium. There's no effect really at all on ejection fraction.

If they had viable myocardium, both medical therapy and bypass surgery improve ejection fraction. And this has been well documented, that medical therapy with beta blockers and other drugs may really improve the ventricular function in patients with viable myocardium. But here's the catch, there was no association at all between these changes in ejection fraction and subsequent mortality.

So it's complex. There's no doubt that revascularization has been beneficial. I believe that the STICH trial showed that. But the mechanisms are multifactorial and less easily understood. Now, hibernating myocardium is a concept that was proposed by the late Shahbudin Rahimtoola, who died recently and was a friend of many of us at the Mayo Clinic. This was published in 1982.

And this was a patient with single vessel disease and occluded LAD, ejection fraction 37%, dilated left ventricle. And after nitroglycerin, the LV remodeled, or the volumes were reduced, and the ejection fraction went up to 51%. And after bypass surgery, post-operatively with a patent graft to the LAD, EF-- I'm not sure why it was 76%, which is actually hyperdynamic, but the LV end diastolic volume was much smaller than originally.

So he proposed the concept that this was hibernating myocardium, and that the myocardium hibernated to protect against ischemia and prevent necrosis. And this was really his concept, which I think still holds true today. Now, the definition of viable myocardium, when you look at the literature, one has to be careful. It's myocardium that is dysfunctional at rest. It's not scarred, and it has the potential for functional recovery after revascularization.

And in fact, you can only diagnose hibernation after the fact. And that is, it's a term that should be used retrospectively only to describe those segments which actually improve following revascularization. And this is important because often one sees the term viability used incorrectly in a number of studies. Now, viability and prognosis in patients with LV dysfunction.

There are a number of different substrates, and some of these are still under investigation. Hibernation implies resting ischemia, and it may be a protective phenomenon to prevent the ischemia-causing necrosis. Another form of hibernation may be repetitive stunning, and that is that in an obstructive vessel there is recurrent ischemia, which is inducible when the pulse rate goes up, for instance. And repetitive stunning can cause LV dysfunction which is reversible.

One needs to take into account not just the viable myocardium, but the extent of scar, the amount of remodeling, and the duration of hibernation. Because over time, if you make a diagnosis of viable myocardium that may be hibernating, and you come back six, nine months later, that hibernation may have progressed to necrosis. And one of the questions that we really don't know is, how much is enough? It's not an all or none issue.

It's not enough just to say viability or hibernation. How much viability? And this is a question we don't know. And really the question is, how much viability is there that if we correct that, and revascularize, and improve the viable myocardium, it will have an effect on prognosis. And future trials I think will continue to address the clinical impact of multi-modal imaging strategies in guiding treatment in terms of outcomes.

And what I really mean is we now have a number of sophisticated imaging techniques to try and get at this issue. How much is enough? When is it sufficient that we can justify a fairly risky procedure of revascularization? Now, this is a meta-analysis that is now 11 years old, but tried to actually answer this question-- 29 studies, 4,000 patients meta-analysis.

And on this axis is the amount of viable myocardium as a percentage of dysfunctional myocardium. So 50% of the ventricle is not working. What proportion of that and dysfunctional myocardium do you need to have viability present to justify the improvement? Or let me put it differently. How much viability does there need to be there to justify the risk of surgery, or maybe PCI, with a view to improving prognosis?

With PET about 26%, stress echo 36%, SPECT almost 40%. So it's not enough to just report an imaging study and say, viable myocardium is present. It's how much viability is key. Now, the clinical indications. You don't have to look for viability in patients with normal LV function. These are patients with coronary disease and severe LV dysfunction. My cutoff would be an ejection fraction of 35% or less.

If they have severe coronary disease, they've never had a history of MI, that should make you raise some questions about viability. No Q waves on the EKG. If they have significant angina or stress induced ischemia, you should have viable myocardium to cause that. If at angiography you see subtotal occlusions with extensive collaterals. And then the entity of flash pulmonary edema with subsequent dramatic improvement.

Typically, this occurs in people with hypertension and heart failure with preserved ejection fraction. But occasionally, one will see a patient with severe pulmonary edema, EF 30%, 35%, and you come back the next day and the ejection fraction is normal or up in the 50s. And that should make you think of a large amount of ischemic, and, by definition, viable myocardium. And all of these should raise your clinical suspicions about viability testing.

But we've all seen this. This is a patient before surgery, LVEF 26%. The orange reflect scar. The yellow areas of hypokinesis, which are reversible presumably. After surgery. LVEF 45%, and much of the areas of so-called scar tissue have, in fact, improved. And we've all seen this clinically. And again, this is, I think, 14, 13 years old. A well-known meta-analysis of 14 nonrandomized studies, showing that if viability was present, revascularization had a very large beneficial impact on mortality.

If viability was absent, there was still a benefit from revascularization over medical therapy, but it wasn't nearly as impressive. But no randomized trials. And then in the STICH trial-- and what I want to emphasize this was a myocardial viability substudy, it was not a trial. It was a substudy from a randomized trial.

But in the viability testing study, which was about 50% of the patients in the trial who underwent SPECT or dobutamine stress echo, if you look at mortality without viability versus with viability, the mortality without viability, or at least the mortality in patients with viability who had revascularization, hazard ratio 0.64, 36% reduction in mortality in those with viability. P 0.003.

This is in people not just revascularized, this is in people revascularized and with medical therapy. So having viability is better than no viability. And that may be because medical therapy also works in patients with viability. This was the surprise. In this viability substudy, if you look at the results of bypass surgery versus medical therapy with viability, 14% reduction in mortality. Not significant. Without viability, 30% reduction in mortality. Also not significant.

So these were trends. But this is really what came as an incredible surprise. If you look at this visually, you would say that the group that did better, even though the numbers are fairly small, the group that did better were the group that had bypass surgery without viability. And this was very difficult to understand and still is. And did raise the question, the English physiologist, the whole concept of viability "might be another beautiful hypothesis slain by ugly facts."

Now, I have to tell you that almost overnight, at least in the United States, testing for viability dropped by about 25% to 30% on the basis of this substudy, Well, there are other studies from this myocardial viability substudy. And at the time, I said to the investigators, I'm pretty sure that if you look at those patients who had inducible myocardial ischemia and viability, those patients with inducible ischemia would clearly have a benefit from revascularization.

Well, if you look at mortality in patients with inducible myocardial ischemia-- this is no ischemia, and this is ischemia in the substudy. No difference at all from bypass surgery and medical therapy. Another huge surprise and difficult to understand. And then this study on the importance of angina in patients with coronary disease, heart failure, and LV systolic dysfunction.

And I again said to the investigators, I think if you look at the group who have significant angina in addition to heart failure, they'll do much better with bypass surgery. Well, the conclusions from this study were the presence of angina did not confer a worse prognosis, but neither did it demonstrate a benefit from reversing a greater benefit from revascularization by bypass surgery. CABG didn't-- it improved the symptoms of angina compared with medical therapy alone, but had no effect on prognosis.

So this, again, is very difficult to understand, because I've always been brought up to believe you revascularize for ischemia, you revascularize for angina, and that will improve prognosis. And my colleague Jeff Geske and I published this editorial for this paper, "Angina in Revascularization of ischemic Cardiomyopath, The Whole Quilt or Just a STICH?" So to try and understand this, because I don't know about you, I found this very confusing. And if you're not confused by this, I think you're just not thinking clearly.

And my colleague Ray Gibbons and Todd Miller at the Mayo Clinic, also raised this question. "Is ischemia dead after STICH?" And made a number points. The study was underpowered by today's standards. The use of the ICD was very low. There was single vessel disease and greater than 25%. And really, maybe we have better ways of looking at viability in ischemia using SPECT and DSE.

And their conclusions were that there's still a role to identify patients with moderate to severe and severe ischemia, significant coronary disease, who are at relatively low risk for revascularization. And made the statement that, "The absence of evidence of benefit must not be interpreted as evidence of lack of benefit." I think a good editorial. So I think that there is a way to understand this.

And I want to go back to this study from Cedar Sinai Medical Center. Not a trial, but a very large study of 14,000 patients who underwent adenosine or exercise SPECT imaging. On this axis is the total amount of myocardium that is ischemic. This is a log hazard ratio. The orange is revascularization, and the yellow is medical therapy. And if you look at this point here, once about 12.5% to 15% of the myocardium became ischemic, you then see a higher log hazard ratio with medical therapy, and a clear benefit from revascularization.

But in a subsequent study from the same database, what was shown was that the role of ischemia. This powerful effect of ischemia in patients who had a fixed myocardial defect or scar tissue greater than 10%, the role of ischemia, particularly the percent of ischemic myocardium, fell away. It lost its significance, and there was no interaction with medicine, and medical therapy, and revascularization.

And so, to translate that, what it means is that the role of ischemia on the benefit of revascularization was nullified by the presence of extensive infarction or scar. And the patients in the STICH trial had very extensive scar and prior infarcts. So is there a role for viability in ischemia testing? Is the concept still valid and rational? Well, I think if you go to the STICH trial, you have to say in STICH patients, no. There was no effect of viability, inducible ischemia, and angina on surgical outcomes.

There is improved remodeling. There is increased remodeling with nonviability, but no effect on surgical outcomes. And remember, this was a group of patients, ejection fraction 35% or less. But what about other patient subgroups? And I would argue that there is a very important role for assessing ischemia and viability in other subgroups with severe LV dysfunction. In patients with LV dysfunction and coronary disease, are the presence of viability inducible ischemia and angina still therapeutic targets?

Fair question after STICH, but I would say, yes. And the considerations are not just the presence and extent of viability in ischemia, but how much scar is there? How much remodeling? Because with extensive scar and remodeling, I showed you in the previous trial-- not trial, but study-- that the extent of scar and remodeling could overwhelm the effects of viability and ischemia. And when is the point of no return? And that's the question.

So my conclusions are that viability testing may predict the response to revascularization in selected patients with coronary disease and LV dysfunction. It's a marker of prognosis. It may influence the response to medical therapy. I think this gets underplayed. Viable myocardium responds to drugs. The impact of viability and residual ischemia may be overwhelmed by extensive scar and remodeling.

If you have a ventricle that is already six centimeters in diameter, perhaps that's too far, point of no return. Ejection fraction less than 20%, perhaps. It should not be a routine determinant of the decision to revascularize. So in my practice, if a patient with LV dysfunction, EF 30%, 35%, if they have significant angina, good distal vessels, no Q waves, preserved voltage on the ECG, important clinical sign, and they're a reasonable surgical risk-- and I'll close with one or two comments about PCI-- in those patients I don't think it's necessary to get viability testing.

But our patient population is sicker, and older, and has more comorbidities. In patients with severe LV dysfunction, extensive remodeling, LV and diastolic volume over six centimeters, multiple co-morbidities, incomplete revascularization is likely, angina is less severe. And under multiple comorbidities I will put advanced age. In this situation, I think I need all the help I can get in order to make a clinical decision, as does the surgeon or the interventional cardiologist.

And I think viability testing, particularly with PET and MRI, may be extremely helpful. Now, again, the negative, I think, impact of the STICH substudy on viability testing has had a number of, I think, unfortunate side effects, unfortunate sequelae. So this illustrates the utilization of testing for significant obstructive coronary disease among patients hospitalized with new onset heart failure.

And I think that this, a large study from Boston, was a missed opportunity. Now it was an administrative database, 67,000 patients, some of whom had no known coronary disease, some had known baseline coronary disease. They presented with new onset heart failure. The only imaging tool really used was during the index hospitalization, and within 90 days was the echocardiogram.

Hardly anybody underwent a nuclear stress or a stress echo. If you look at the use of invasive testing during initial hospitalization, very few underwent diagnostic coronary angiography, 11%. And within 90 days, only 16.5% underwent diagnostic angiography. Plus another 8.4% that also had the right heart catheterization. So very under-utilized, given the fact that coronary disease is still probably the leading cause of heart failure.

Now, if you just a few closing words about PCI. This is a trial from the UK, by Dr. Pereira. This is a British Cardiac Interventional Society trial, 700 patients, "PCI for Ischemic Cardiomyopathy." This slide is out of date. They have completed enrollment. EF less than 35%, extensive coronary disease using an interventional score. They had a viability study. They had to have at least four segments that could be-- four dysfunctional segments that could be revascularized by PCI.

And if in that case, they were randomized to OMT, optimal medical therapy, or PCI and optimal medical therapy. Endpoint, all cause death or hospitalization for heart failure. I think we will have the results of this trial within the next six to 12 months. Very important trial. Not much other data on PCI in patients with severe LV dysfunction.

This is a retrospective study from Ontario in Canada. EF less than 35%, either left vein LAD or multivessel disease. Emergency revascularization excluded. Almost 5,000 patients, propensity matched. If you look at death due to cardiovascular disease, slightly higher with PCI. P 0.001. But the magnitude of the difference in favor of bypass surgery is small.

If you look at myocardial infarction, a high rate of MI with PCI did not perform as well as bypass surgery. But I suspect there still is a role for PCI, and that's what Dr. Pereira's trial will objectively tell us one way or another. Now, this is the text of a cable sent by Mark Twain from London, when he was having his breakfast at the London club. And he read his obituary in the United States papers, which had been mistakenly published.

And he sent back this telegram, "The reports of my death are greatly exaggerated." And I believe there's a great deal more we need to know about the quantification of viability, and specifically how much is enough to demonstrate a benefit. But I really believe that the role of viability testing is still alive and well, and the subject of further investigation, and continued investigation. Thank you for your attention.

MANDEEP SINGH: Thank you, Dr. Gersh. I have a couple of questions for you, just to open it up following this excellent presentation. First, I think we've made a lot of inroads in terms of medical management with ACE, beta blockers, mineral receptor antagonists in patients with heart failure. We have made a lot of inroads with device therapies. But somehow, this revascularization component has lagged behind.

In patients who have known LV dysfunction, before we go for viability testing, do you think all these patients should have their coronary anatomy defined? As you pointed out in one of those studies where the testing and/or angiogram is dismally low in patients who present with acute heart failure.

BERNARD GERSH: Mandeep, thank you. You make two points. The first one is, I enjoy talking about clinical trials, and I've always been very involved with the design and some of the statistical aspects of clinical trials. And one of the points that I realized or make, but I think we all realized a long time ago, all of these trials have revascularization, which take 5 to 10 years to complete, because you need at least five years of follow up and three years of enrollment, because it's not easy to get the patients, they are all obsolete by the time they're finished.

I think in one of my slides I showed ICDs were only used in 20% of the STICH population. CRT was used in probably 15% to 20%. So in many ways, this trial is obsolete. It's actually 12 years old already. And when you sit on a guideline committee, there's nothing wrong with the trial. I mean, trials get obsolete because technology changes, and pharmacology changes, and techniques change. So when you sit in a guidelines committee, you have to take this into account.

And you don't dismiss the trial, but you have to understand that maybe what was a Class I indication, based on this trial of 10 years ago, may be a II-A indication 10 years later. So that's the first point. The second more important point that you raise, should you establish the presence or absence of coronary disease? I think in a patient with heart failure who could be a revascularization candidate in the future, I think you should establish the coronary anatomy even before you reach the stage where you're going to revascularize the patient.

I think you need to know whether you're dealing with a nonischemic cardiomyopathy or an ischemic cardiomyopathy. And I think there are other issues in that regard too, because depending on the ejection fraction, whether it's ischemic or nonischemic, the indications for an ICD may differ. The potential benefits from CRT may differ between people with ischemic and nonischemic etiology. I think, actually, the new ICD guidelines from the European Society will probably not distinguish ischemic from nonischemic for the ICD, but in this country we do.

I do think that what has changed our practice in people with chronic coronary disease is the use of the CT angiogram. And if one looks at the 2019 EEOC guidelines, that I was involved with, we made a big point that in people with a low to intermediate likelihood of coronary artery disease, a CT angiogram is probably the diagnostic modality of choice.

If someone comes to you with heart failure, and you feel there's a very high likelihood of coronary disease or they have coronary disease, there you may go to a stress test to try and assess the degree of ischemia and various other factors. But I'd like to know whether I'm dealing with coronary disease as a cause of the patient's heart failure, or some other etiology. And I think the CTA really has made our life much easier.

Now, I realize that CTA is not helpful in assessing lesion complexity and severity in people who are undergoing revascularization, but it's negative predictive value is extremely good.

MANDEEP SINGH: I agree. We have a question-- and I think I also wanted to ask you in terms of whatever studies you quoted, plus I read about, there is distinct underutilization of PET. Maybe they do not want to compromise specificity or sensitivity. Maybe it's not so much a perfusion metabolism, issue it is more about inducible ischemia issue. The stress echo or SPECT is more prevalent in determining viability rather than PET. What is your take on which tests should be done to document and demonstrate viability in this context?

BERNARD GERSH: I think the reason for the underutilization of PET in the trials is very simple. Most people didn't have expertise or didn't even have it. And so, they chose imaging modalities that were widely available in many parts of the world and in most hospitals in the United States. So I've always said, when people say, what test do you think we should do? I say, do whatever you do well in your institution. So if you feel you have a very good nuclear program, or echo program, or both, then you choose between stress echo or SPECT.

I think the ideal, and we have that at Mayo, is PET MRI. In really difficult patients, I think using MRI to look at volumes and scar tissue, and PET to look for areas of mismatch may be the best combination available. But I say this with the full knowledge that not everybody has access to this. And even if they do, you have to have expertise. And we have, as you know, at Mayo, a number of people who are particularly interested in this. So I think you've got to do and rely on what you do at your institution well.

MANDEEP SINGH: So two other questions I have. One is, now our department has done yeoman's work in terms of using artificial intelligence, not only predicting LV dysfunction, but also future LV dysfunction based on just simple EEG. And you've been a big proponent of if the ECGs normal, your ejection fraction is normal. I was your student when you used to teach that. Based on your knowledge, and based on what developments we have in artificial intelligence, do you think that can be factored into the management of these patients?

BERNARD GERSH: Oh, what a great question. And as you know, I've got a huge interest in the AI program. And we're focused at the moment at Mayo, the focus is really been on the EKG to assess LV function, and the ECG to assess atrial fibrillation. You're right. I have always said if you have a normal ECG, you're going to have normal ejection fraction. The actual paper was written by myself, and it was written by our colleague, mutual colleague, Jay Reiffel, and myself.

And from a registry, we looked at 20,000 patients. And if they had a normal ECG, no history of infarction, and a normal cardiac silhouette on a chest X-ray, the likelihood that they would have an ejection fraction of greater than 45%, I think, was about 98%. And when I found that very useful is when you're asked to assess someone in the middle of the night, they want to go ahead and do an operation or a procedure on a patient, and you don't have time to do an echo. And I've always found that clinically very useful.

We now have a very good algorithm, developed by Paul Friedman at our institution and others, that shows that using AI on an electrocardiogram, you can predict the reduced ejection fraction with great accuracy. And in fact, in our clinical practice, we have an AI dashboard. When I'm doing a consult, particularly if it's a remote consult when I'm outside the clinic, I use that dashboard to see what the ejection fraction is. And if it comes back there's a 98% likelihood it is normal, well, then you don't have to do an echo.

Now, the next step is, how are we going to use it in these patients with severe LV dysfunction and coronary disease? And this is a topic that I think would be very exciting to investigate. And the one thing I've learnt with our AI experience-- and I think this is a word of caution for everyone, just because you show in one group of patients that the algorithm will predict atrial fibrillation in people who are in sinus rhythm, who had an episode of atrial fibrillation a month ago, that it will apply to other patients with atrial fibrillation.

You have to retest it. It looks like it does work but, for instance, we recently looked at our algorithm to see whether we could predict perioperative atrial fibrillation. And we couldn't, because it's a different mechanism. So when you think of the patient with coronary disease, a very abnormal ECG, severe LV dysfunction, the question we would like to ask is, can you use AI to predict a good prognosis or salvage after revascularization? That's a wonderful study to do.

I'm not sure how we would go about doing it, but that's one to look at. That would be very instructive and helpful. And it may not be-- it may not be AI of the ECG. It may be AI of an echo or AI of some other imaging technique plus the ECG. I don't know. I don't think we know the answer.

MANDEEP SINGH: So the only other question I have is, we have all traditionally labeled severity of LV dysfunction either in a binary or polytomous terms. Less than 35 is the number to go to. It's a go-to number to say if it is less than 35, it is severe LV dysfunction. It connotes a different prognostic value and meaning, not only to the provider, but to the patient as well.

My question is, now the benefit, as you pointed out, it's a high risk surgery, or now maybe PCI. The lower your ejection fraction, more the reward in terms of improving your outcomes. But one, can we just look at this paradigm in a continuous fashion?

BERNARD GERSH: Of course.

MANDEEP SINGH: And two, what happens if my ejection fraction is 36%, or I'm in the mid-range as the ischemia trial post hoc analysis demonstrated? So in other words, when can we just be a little more optimistic or pessimistic, notwithstanding the nuances of scar, and the remodeling, and other issues.

BERNARD GERSH: Well, I think the reason that we use under 35, 35 to 49, and then greater than 50, really goes back again to the original trials. And that goes back to the very early 1980s. And that-- and those were the cutoffs that we used. And 35, at that time, 1982, this sort of era, people thought that-- I think the general thinking was, if your ejection fraction was under 35%, you were far too high risk to put into a trial. And, therefore, the first trial of revascularization within a year for 35% or less was the STICH trial 12 years ago.

But the first trials of coronary bypass surgery were over 30 years ago. And in fact-- I beg your pardon, over 40 years ago. So it's an artificial categorization, but it's been clinically useful and helpful. I mean, I think this comes up-- should we look at it as a continuous variable? Yes. But to give you an idea of how artificial it is, if you look at the guidelines for defibrillators, it's 35 or below. Well, what about 36?

Now, I know, and you know Mandeep, that we can't tell the difference between 35% and 36%. Not with an angiogram, not with a mega, and not with an echo. So if I really feel that the patient needs a defibrillator, and I see a report of 36%, I actually call the lab and say, don't you think-- to me, it looks more like 34%.

[SHARED LAUGHTER]

BERNARD GERSH: Or 33 or 32. Are you really that confident that it's 36%? And that's where the guidelines, I've always said-- I mean, they're doing the best they can. I mean, guidelines are not easy to write, but it's very arbitrary. And so, another example is the CHA2DS2-VASc score. And I was on another Zoom conference a week ago, on risk scores for atrial fibrillation.

So you have a patient who doesn't want to go on anticoagulants, and that patient is 74 years old. He's a CHA2DS2-VASc of 1. No hypertension, no other problem. CHA2DS2-VASc of 1. Well, if it's CHA2DS2-VASc of 1, the guidelines say you should consider anticoagulation, but it's not a Class I indication. CHA2DS2-VASc of 2 everybody agrees should be anticoagulated if possible. Now, let's say the day you saw him it was the day before his birthday. So he's a CHA2DS2-VASc 1, he says, I don't want to be on anticoagulants.

So you say, OK. Well, you don't have to be until tomorrow, and then you're a CHA2DS2-VASc of 2. We have these arbitrary cut offs, but we also have a brain, and we have to make clinical decisions. And this is why medicine is still fun, because it's not down-- it's not just down to an algorithm yet. I think it was Voltaire, the French philosopher, who said, Common sense is not that common." And we have to use common sense.

MANDEEP SINGH: So what is your take on the remodeling, in terms of how long does it take? So when I look at the STICH trial, they had to ask for additional funds to extend the observation from five to 10. Now we have these new trials coming up, which then randomized patient PCI versus optimal medical therapy, and the duration is two years. Do you think those will not be able to capture the endpoints so soon after?

BERNARD GERSH: Well, yes and no. I think it depends on the patient population. If it's a group with a high incidence of recurrent ischemia, I think we will be able to probably come to a conclusion after two to three years. In a situation like the STICH trial, the difference is, at five years-- I think it was-- I can't remember whether it was three, or four, or five years, but the original trial, I think it was five years. But the majority of patients had only been followed for three years, somewhere around there.

And the differences, which were becoming-- they were either significant but still quite small, and I was one of many that argued for additional funding to go out to 10 years, and really prove the case beyond all doubt. So that's in this particular patient population. I think in others, if you think of the trials of left main disease, at three years there was a trend in favor of surgery that was not significant. And originally, it was decided to follow the patients out to five years. The funding was there.

And that trend became more obvious after five years, but still not statistically significant. Well, depending on which endpoint you look at, and that's still controversial-- the point being that when you start to see curves crossover, that's a trend and a sign that you need a longer follow-up. The same with the ischemia trial. I mean, the ischemia trial is neutral at five years. I suspect in 10 years we may see a difference in favor of revascularization, or PCI in terms of myocardial infarction.

Now, just because it reaches a P value is not the end of the story, because you then have to start looking at the magnitude of the difference. So it's not just P less than 0.05. How much of a difference is there? And I think we also need to realize that when we do these trials, all the patients in the surgical arm and all the patients in the PCI arm were suitable candidates for either procedure. Highly selected patients.

And just because we do show a trend as we continue to follow up, it still only provides you half the answer to the question. But I do think that, at least with the ischemia trial, I have lobbied as much as I can for funding from NHLBI to continue for out to 10 years. So I think when you-- I don't mean a two-year trial is useless. Not at all. It depends upon the endpoints that you're looking at.

And it depends upon the event rate. If you've got a very high event rate, maybe you'll get an answer in two years. But these trials do take time. They're difficult to do, and they do take time.

MANDEEP SINGH: Very true. Very true. Now, I thank you, Dr. Gersh, for, again, taking your time off your busy schedule. And the topic was very nice. And I had to honestly read myself, refresh myself in what the literature says. Because it's such a topic where most of us don't even know about how to approach a patient who has these issues of severe LV dysfunction, when and how to offer revascularization to them. So thank you again.

BERNARD GERSH: It's my great pleasure, Mandeep. I think the reason I got interested in this was exactly that. I wasn't sure what to do with these people. Then along came the trial, and then-- I always used to get viability testing, and then along came the trial. And it certainly took two or three years of quite serious thought to try and understand the trial, and the viability substudy, and the implications.

And I now find I think I feel fairly comfortable in making a decision in these patients. But the sicker they are, the more difficult it is, the more you need the heart team, and the input of the surgeon, and interventionist, and the input from the patient.

MANDEEP SINGH: Yeah. I thank you for your time, Dr. Gersh. Have a good afternoon.

BERNARD GERSH: Pleasure, thank you.