EMPA-REG OUTCOME: CV and Renal Benefits in People With eCVD and T2D

Dr Javed Butler (Baylor Scott and White Research Institute): This whole sort of thing, as you know, 2008 the FDA came up with a mandate that the new therapies for management of diabetes has to be assessed in large outcomes trial to assess the safety of these therapies in large outcomes trial.

At that time, these were drugs for diabetes and their management in heart failure was not even that somebody was even thinking about at any rate. The first trial that was done was EMPA-REG OUTCOME trial with large outcomes trial, with new diabetes medications. Whether you look at SGLT2 inhibitors, whether you look at DPP-4 inhibitors, whether you look at GLP-1 receptor agonist, first trial comes out and the results were very, very pleasantly surprising.

I mean, if you just think about it, we forget that the number one cause of death in patients with diabetes is cardiovascular death, and up until a decade ago, there was not a single agent or a drug that was known definitively to reduce the risk of CV mortality till this trial came out.

But then there was this incredibly powerful benefit in heart failure that was seen. The benefit was that not only was there a reduction in heart failure, which was both clinically meaningful and highly statistically significant, the curves that started separating early, but there were some really peculiarities. It was just really easy to just say this is related to better glycemic control. Except that before 2015, we had tens of trials with glycemic control and none of them showed reductions in heart failure. If anything, in some of the trials there was a question about a potential increase in the risk.

Then in this particular trial, the beginning hemoglobin A1c, the ending hemoglobin A1c, and the changes in hemoglobin A1c had no correlation with heart failure. Clearly, there was something beyond glycemic control that was modulating that risk.

[00:35:03]

Clinical Evidence for SGLT2i in HFrEF

A lot of the discussion started around this thing as to why this is.

[00:35:09]

DAPA-HF and EMPEROR-Reduced: SGLT2i in Ambulatory Patients With HFrEF

Obviously, science continues to move on. In the meantime, there were more basic science data, animal data, and human data coming out in patients with and without diabetes that these drugs were associated with improvement in cardiac structure and function, in vascular structure and function, in renal structure and function, in a whole host of systemic effects like inflammation, oxidative stress, adiposity, endothelial function, insulin resistance, et cetera.

The question came up is that, one, you cannot ignore that heart failure signal. We need to understand in heart failure trials what is going on with these patients, but more importantly, if a drug can have a pharmacodynamic profile that I just very high level mentioned. Again, happy to discuss in the Q&A things related to mechanism of action. If a drug can do all of these things, what does this have to do with diabetes?

Now, we sometimes tend to unnecessarily binarize our discussion that it is this or that, but obviously in this particular case, it was looking like both. That it is a good drug for diabetes management. Of course, there is a reduction in CV mortality with empagliflozin, but the issue is that when it comes to cardiorenal risk protection, is it another cardiorenal risk protection agent irrespective of diabetes?

Again, now in 2024, 2025, it is easy to stand up and talk about this, but I tell you in 2016 when these trials were being designed, there was a very concentrated push to do the heart failure studies only in patients with diabetes because there is this diabetic cardiomyopathy and these are the diabetes drugs.

It took really some convincing to say, well, let us just do the trial in heart failure irrespective of diabetes. Two sets of programs that started, one with empagliflozin, one with dapagliflozin, both give or take, 10,000 patients each, both of them had 2 components, HFrEF and HFpEF, and the similar endpoints that were tested in these trials.

[00:37:04]

SGLT2i in Ambulatory Patients With HFrEF

Here is the bottom line summary. I mean, talk about replication of the results. First of all, both trials had 90%-plus RAS inhibitor, 90%-plus beta blocker, and about 70%-plus MRA use. Whatever we are seeing is on top of very, very good baseline medical therapy, more so than we will ever see, or we have at least so far ever seen in a real world registry. These are incremental benefit on top of already very good medical therapy.

The other thing is that there is just no heterogeneity. Now, I am using the term heterogeneity really loosely because you have to have head to head trials. But if you just do side by side trials, these 2 curves look completely the same.

[00:37:48]

DAPA-HF and EMPEROR-Reduced: Reduction in CV Death/HF Hospitalization With or Without T2D

If you were to take out all the letters, numbers, and words from this slide, you would not know which is DAPA-HF, which is EMPEROR-Reduced. You really get the sense that, yes, these drugs improve heart failure outcomes at least with the 2 therapies that were tested. Then the question is, did our experiment work out whether there was diabetes or no diabetes difference?

I can tell you this is probably the only trial in my career where I was maybe even more interested in the secondary analysis and the primary analysis is like, did we mess up? Here is the result. I mean, there is just no difference. If you look at the combined EMPEROR-Reduced, DAPA-HF, the hazard ratio is 0.74, 0.75.

When we look at this dichotomize, if you trichotomize it, pre-diabetes, no glycemic abnormalities, or we just look at hemoglobin A1c as a continuous variable. There was no hint that there was a modulation with glycemic status, or glycemic control, or baseline glycemic therapy. We at least proved that these therapies work in patients with heart failure irrespective.

[00:38:53]

DAPA-HF and EMPEROR-Reduced: Adding SGLT2i to ARNI

Whether or not the baseline patients were getting ARNI. I think at this point, it is a moot point. I just put the slide on because it was heavily debated at this point because ARNI use was slowly going up, but was not as uniform as it is right now. The question was, well, what if you better treated with ARNI rather than ACE inhibitor? Does it make a difference? Again, absolutely no difference per se.

[00:39:12]

Left Ventricular Remodeling Trials

There are 3 remodeling studies that have also come out showing improvement in LV volumes and improvement in ejection fractions.

[00:39:19]

Left Ventricular Remodeling Trials

It is consistent with the rest of the results. Again, some of the more detailed data for LV remodeling. Again, the LV remodeling studies have come out both in patients with diabetes or without diabetes.

[00:39:32]

EMPEROR-Preserved: Empagliflozin + SoC for HFpEF, Regardless of DM Status

Then we move on to heart failure with preserved ejection fraction. One, EMPA-REG OUTCOME trial had already made history in medicine once by showing reduction in CV mortality in type 2 diabetes, and then made history of medicine again in EMPEROR-Preserved trial. Again, we can have a real robust academic discussion how to evaluate subgroup analyses.

But prior to this, none of the trials at HFpEF have ever been positive. There were subgroup analysis that look really attractive, and as you know, the guidelines have actually embraced recommendations on the basis of some of those subgroup analyses. They are weaker recommendations, but overall, in terms of the trial, no trials thus far had been positive in patients with HFpEF till this trial came out, EMPEROR-Preserved.

[00:40:16]

EMPEROR-Preserved: Effects of Empagliflozin on Composite of CV Death or HHF in HFpEF

Again, very much the same, large outcomes trial, 6,000 patients, EMPA vs placebo, looking at the same outcomes. Again, a highly statistically significant and a clinically relevant 21% relative risk reduction in cardiovascular death, heart failure hospitalization in patients with HFpEF.

These were the 2 aha moments in your medical career. First one was when the EMPA-REG OUTCOME outcomes Kaplan-Meier was shown that you can actually modulate CV mortality in type 2 diabetes, and then when this Kaplan-Meier curve was shown that you can actually improve outcomes with EMPEROR-Preserved.

Now, down the road we have data with FINEARTS with finerenone and the data with GLP-1 is looking pretty good that we are now getting into this comfortable zone that we could change the natural history of HFpEF. I tell you, for 2 decades we were almost giving up that nothing will change.

[00:41:12]

DELIVER: Dapagliflozin for HFpEF

Here in that sense, it was a pretty landmark trial. Then the DELIVER trial came out with dapagliflozin. Again, the same structure, some minor nuances in terms of the definition of the outcomes, et cetera, but again, exactly the same.

[00:41:27]

DELIVER: Composite of CV Death or Worsening HF in Mildly Reduced or Preserved EF

There is a 2 by 2 validation of everything that we were talking about. One, I said that if we were to take out all the letters, numbers, and words from the slide and just show you the Kaplan-Meier curves, you would not be able to tell whether this is DAPA-HF or EMPEROR-Reduced. I think we have the same dynamic here that you will not be able to tell whether it is DELIVER or it is EMPEROR-Preserved.

Really good internal validation that for these 2 drugs that were tested, you are seeing the results that are very consistent.

The other interesting thing is, is that actually the magnitude of benefit, obviously numerically it is different, but the magnitude of benefit in HFrEF and HFpEF is also the same, about give or take, 21%, 25%. For the first time, we are also thinking about that we have a therapy that works across the spectrum of ejection fraction, and this obviously has major implications in terms of the management of the patient.

[00:42:22]

Meta-analysis of DELIVER and EMPEROR-Preserved

Those investigators did this pool analysis. Again, you are seeing very consistent results, no heterogeneity. I am a Bayesian by orientation, so p value of 0.052 vs 0.049, it really does not matter. It is a matter of power and how long the patients have been followed overall. Again, for the first time, it looks like there is a modest reduction in CV mortality as well.

[00:42:48]

Meta-analysis of DELIVER and EMPEROR-Preserved

This at one point was also heavily debated based on the data with MRAs, and data on RAS inhibitors and ARNI, that even if things work, they work in EF less than 60%. Nothing works in EF more than 60%. Again, as you can see, no heterogeneity overall. If you look at the hazard ratio across the 3 categories of EF 40 to 50, 50 to 60, greater than 60, all 3 of them were benefiting equally with the use of SGLT2 inhibitors.

[00:43:17]

Posttest 1

Let us take this question again and let us see if you can have a quick discussion. In a meta-analysis of DELIVER and EMPEROR-Preserved, use of an SGLT2 inhibitors in patients with HFpEF show significant reduction in those with EF greater than 40, 40 to 50, 50 to 60, or greater than 50 as an isolated group.

This is a little bit of a trick question, because there is not a wrong answer. There is a preferred right answer. In all of those 3 groups, it is clearly shown to improve the outcome for these patients. If you were to choose the most appropriate answer, as you can see in the posttest, as most people said, LVEF greater than 41%, and that is true. The trick in this is that it is not isolated within any EF bucket of 40 to 50, 50 to 60, greater than 50 across the spectrum of ejection fraction the benefit is seen. Yes.

[00:44:35]

KCCQ Summary Scores and Domains

Speaker: HFpEF by definition is over 50%. [inaudible].

Dr Butler: Yeah. This is heavily debated when the data were being done. Let me just repeat the question for folks on the audience. The question is, well, technically speaking, your STEM said HFpEF and HFpEF by definition is EF more than 50%. How can you say it is positive in HFpEF when you have the mildly reduced? 100% correct. I cannot argue with you on that.

This is more about the history of medicine as opposed to the science of medicine. All the heart failure with reduced ejection fraction trials, that the way they were designed in from 1980s was EF less than 40%. Then we realized that normal ejection fraction is not 40% to 50%, it is greater than 50%. The universal definition of heart failure now clearly defines HFpEF as greater than 50%.

The problem is that the clinical trialers have not embraced doing trials in HFpEF limiting to only greater than 50, because then you will have this island of people, about 15% of heart failure patients, between 40% and 50%, and then they will be always left without evidence.

Because the historical trend is to do rEF trial in less than 40%, the HFpEF trials have technically included mildly reduced and HFpEF both. You are 100% correct, but in terms of the spirit that it works across the spectrum is the point. Point well taken.

Now, our patients are wanting to not only live longer without hospitalization, but they also want to feel better. The quality of life has been recognized as a good clinical endpoint to follow. For our colleagues and friends online and in the room who do not think about the quality of life all the time, we measure it using a scale called KCCQ, Kansas City Cardiomyopathy Questionnaire. It has 3 domains. One domain is strictly limited to symptoms, how you feel, what is the total burden, et cetera.

If on top of the symptoms you add physical limitation, how much is your ability to do something that is called CSS. On top of that, if you were to go ahead and add the overall quality of life metric, your social integration, sexual function, how good you feel, how you integrate, how you socialize, that is called OSS or overall summary score.

[00:46:53]

EMPEROR-Reduced: Effect of Empagliflozin on KCCQ-CSS Over Time

Now, in EMPEROR-Reduced, it was measured at 3 timepoints. First after baseline, early at 3 months, and then eight months, and then at 12 months. Then as you can see, the first time when it was measured, there was already a statistically significant improvement in quality of life, and that that was maintained all through the trial, and the same results were obtained for all 3 domains.

[00:47:16]

Clinically Meaningful Improvements and Deteriorations in KCCQ-CSS, -TSS, and -OSS Over Time

Now, this might look like a complicated slide, but the interpretation of this slide is incredibly simple. There is a general sense, provided obviously backed by data, that a 5-point improvement in KCCQ is clinically meaningful improvement. A 10-point improvement is a lot improvement and a 15-point improvement is a very clinically meaningful improvement.

The summary of this slide is that does not matter whether you are looking at CSS, TSS, or OSS, does not matter whether you are looking at a 5-point, 10-point, or a 15-point threshold for improvement, and does not matter whether you are looking at 3-month, 8-month, or 12-month, no matter how you slice and dice the data on SGLT2 inhibitor, the symptoms got better and on placebo designation, things were getting worse despite the high triple therapy that were given to the patient.

[00:48:13]

Acute HF: The Missing Link

In the interest of time, I have shown you data with EMPA. It is the same with DAPA, and it is the same in both heart failure with reduced and heart failure with preserved ejection fraction

That is all in the chronic heart failure. In the acute heart failure setting it is the same disease. A lot of the time people said, well, the question is not whether it will be efficacious. I mean, it is the same disease just because you got hospitalized. It is not a different disease. The real question is safety. Is it safe to start in the hospital setting when there are all of these other hemodynamic changes going on?

Here, it is different with SGLT2 inhibitor because there is a hypothesis of not only a safety but a potential efficacy. We all know that when patients come into the hospital, Day 2, Day 3, Day 4, they start becoming diuretic resistance, that there is a breaking phenomenon. They are still congested, clinically better than admission, but still congested clinically. But now the creatinine is bumping and there is this degree of unease, and maybe they were diuresing 2 liters, and now they are diuresing like 500 cc, and there is pressure for length of stay and all that stuff that goes on.

The question is, if you were to give this therapy early, not as a chronic modulating therapy at the time of discharge, but early during the hospitalization can because of its osmotic diuretic and natriuretic properties have a synergy with loop diuretic, better decongestion, and better outcomes. In this particular case, the question was whether you can give the therapy early?

[00:49:37]

SOLOIST-WHF: Primary Outcome

The first therapy that did was more so in the mindset of worsening heart failure and chronic therapy that was the SOLOIST trial with sotagliflozin. That is both an SGLT-1/2 combined inhibitor. As you can see, that if you can give peri-discharge, it was associated with remarkable improvement in the risk of morbidity hospitalization.

Again, we have moved one step closer to acute heart failure, but not necessarily acute heart failure per se. This was the peridischarge trial.

[00:50:09]

EMPULSE: Primary Endpoint

The acute heart failure trial was the EMPULSE trial, where people, if they were not on inotropes, IV vasodilators, not in emergency room, but they could be day 2 in the hospital, still receiving IV diuretic, but were otherwise stable, were randomized. Even in the short-term trial, just 3 months, there was a significant improvement in the risk of mortality, recurrent hospitalization, and quality of life scores.

What I am about to say is mathematically incorrect, but conceptually, directionally it is correct because win ratio is a new endpoint, but you can conceptualize it that it is the reciprocal of hazard ratio. When we showed you a hazard ratio of 0.75, that means there is a 25% chance that bad things will not happen to you.

When I show you a win ratio of 1.36, that basically says that there is a 36% chance that good things will happen to you. That is one way to conceptualize.

Now, this was a short-term acute heart failure trial.

[00:51:11]

EMPULSE: Reduced Time to All-Cause Death or HFE

This was not a 5000-patient long-term trial. You see this composite endpoint, but even in this short-term trial if you do a secondary analysis and granted its secondary analysis that mortality heart failure hospitalization, even in the short-term trial, actually was a statistically significant, a 35% relative risk reduction, P value of less than 0.05.

[00:51:33]

EMPULSE: Change in Body Weight

This whole thing about the synergy between the 2 really worked out. There was more weight loss, not at the expense of worsening kidney function, and that more weight loss was maintained post discharge as well.

[00:51:45]

Meta-Analysis of 5 Large Placebo-Controlled Trials

We have the data across the spectrum. If you start thinking about heart failure across the spectrum as one large entity, a significant reduction. That is not the most impressive part for me. What is most impressive is the consistency of finding across EF and across the drugs as opposed to a lot of heterogeneity, and one drug is pulling the results in one direction or another. Really, a consistency of the results.