Cardiovascular Outcomes Trials With Anti-Diabetes Medications

Dennis P. Scanlon, PhD: There are a number of cardiovascular outcome trials underway. How do those differ from the EMPA-REG OUTCOME trial? There’s a number of them—DECLARE, CANVAS, DEVOTE, and several others—and we could talk a little bit about what we hope to find from those trials. Let’s discuss what some of the unique features of them are in terms of study populations and, ultimately, what they might get us in terms of knowledge.

Zachary Bloomgarden, MD: What we hope to find is that every SGLT2 (sodium-glucose co-transporter 2) inhibitor is marvelous, wonderful, and magical for people with diabetes.

Robert Gabbay, MD, PhD, FACP: That would make life simple.

Zachary Bloomgarden, MD: We may not. CANVAS, with canagliflozin, is very similar to the empagliflozin trial, and we can sort of say we expect to see similar outcomes (although let’s not count our chickens before they’re hatched). DECLARE, the trial with dapagliflozin, is a much larger trial (I think about 17,000 versus 7000 people), and since this is a numbers game, it’s done in a population which has about a 30%, 40% as great a likelihood of cardiovascular outcomes per individual. So, that’s wonderful in the sense that it’s a trial which might allow us to extrapolate benefit to a larger population of people with diabetes, not those who already have cardiovascular disease. It also raises the possibility that it might be a so-called “negative study.” It might not show cardiovascular benefit because of, perhaps, the benefit in people who are kind of pre-heart failure. We’ll talk about heart failure later, but perhaps that benefit is not going to be seen in people who really are not anywhere near that stage. So, it’s opened up a whole set of questions now, as to how strictly must we insist that a trial be positive in order to allow one to explore subsets to look for benefits within subsets? And this is somewhat of a philosophical question for statisticians, but an extremely important one.

Silvio Inzucchi, MD: CANVAS is going to be presented in June of this year (so we have 2 months to go from the filming of this discussion). As Zachary pointed out (he’s spot on), the 2 main issues will be class effect, yes or no. If it’s yes, I think we’ll all breathe a sigh of relief. If it’s no, we’re really going to have to scratch our heads and say, “Was EMPA-REG OUTCOME a fluke, or are there differences between these compounds?” Canagliflozin has a little bit of SGLT1 (sodium-glucose co-transporter 1) effect, and that may or may not be beneficial. The other issue is this notion of expanding the effect to a primary prevention cohort. We may not understand that until DECLARE, frankly, in 2019, because DECLARE is a huge trial. It’s 17,000 patients, as Zachary pointed out, and I believe two-thirds of the patients have no overt cardiovascular disease. Remember, in EMPA-REG, everybody had cardiovascular disease. Whereas in CANVAS, I think a third do not. And in DECLARE, two-thirds do not. So, we’ll be able to answer that question. It’s a bit of a double-edge sword though, isn’t it?

One thing we didn’t talk about was this very early divergence of the event curves in EMPA-REG, and I think that may point us into the direction of why the drug has benefit. In most atherosclerosis trials, blood pressure trials, or lipid lowering studies, when the drug is effective, the strategy is effective. The divergence of the event curves (in other words, the patients on placebo versus patients on active therapy) tend to separate at about 12 months. And that, in our somewhat naive understanding of diabetic heart disease, suggests that there’s an effect on atherosclerosis. It takes a while for that effect to occur. In EMPA-REG, again, we were shocked to see that it was almost at month 2 or 3 that the cardiovascular mortality curves began to diverge. And, if you look at the heart failure hospitalization rates, they almost diverge in month 1. I’m not a vascular biologist, but that can’t be atherosclerosis. It’s either coagulation, which the drugs don’t have an effect on clotting, or some hemodynamic effect, as Bob alluded to, in terms of the diuretic capacity of these medications.

Getting back to this notion of primary versus secondary prevention, if you really have to have heart disease to benefit from an SGLT2 inhibitor (and that actually might be true), you have to have a diseased heart—either known ventricular dysfunction or about to develop ventricular dysfunction—to benefit from this drug class. The primary prevention cohort may be completely neutral, and that may eliminate the overall effect in these trials. In other words, a primary endpoint which is a pooled endpoint, both primary and secondary, may be negative. But when you analyze them separately, you might see an affect in the patients with overt cardiovascular disease, and a neutral effect in the patients without cardiovascular disease. It will be very interesting when these studies come out.

Robert Gabbay, MD, PhD, FACP: And the other thing that’s probably worth pointing out, as you described, is the divergence of the curves with the SGLT2 inhibitors. In the LEADER trial, with liraglutide, it followed the pattern that’s typical for atherosclerosis. The curves separated at 12 to 18 months (in that range), suggesting that the GLP-1 (glucagon-like peptide-1) agonists were acting more in what we would consider a traditional atherosclerotic mechanism. And therefore, that may have benefit in any one that is at risk. It’s just a matter of duration of time and how high that risk is. Whereas, these drugs, which may really work specifically on people that are diseased, so to speak, and need that volume depletion or whatever the mechanism is, may only work in people in secondary prevention.

Zachary Bloomgarden, MD: But I would bring up a caution, which is that in order for the curves to diverge at month 0 or month 1, we’re talking about a tiny number of individual events. And so, all notions of statistical reality are out the window. I would simply say that lots of people with diabetes do have heart disease. If this proves to be something which is applicable to our patients with diabetes and heart disease, not to the rest of those with diabetes, it still would be a wonderful benefit.

Dennis P. Scanlon, PhD: Dr Snow, I wonder if you could discuss the FDA’s decision to add a new empagliflozin indication to reduce the risk of cardiovascular death, and how payers responded to this change? What are the clinical decision making implications?

Kenneth Snow, MD, MBA: I think you’ve heard, already, some of the challenges that a payer now faces in trying to make a decision as to whether this is a drug effect in which case for the right population (those with proven coronary disease), this would be the drug to use. Or is it class effect? Is it all patients with diabetes, or only those with preexisting heart disease? These are research questions that are still in the process of being answered, and somehow, in the meantime, there needs to be a decision on coverage. I think that the decision is still pretty broad across the payer landscape. I think different companies tend to have approved or preferred therapies. And so, I think that’s still holding true at this point, but, obviously, as time goes by and more data comes in and we become smarter about what we know about these drugs and class versus individual, that will drive which direction coverage decisions go.

Zachary Bloomgarden, MD: But if CANVAS does not show benefit, and if DECLARE does not show benefit, it would be awfully difficult for any company to justify not providing empagliflozin to the patients.

Kenneth Snow, MD, MBA: Patients with known heart disease.

Zachary Bloomgarden, MD: With known heart disease of which there are so many.

Kenneth Snow, MD, MBA: Right, of which there’s so many.

Zachary Bloomgarden, MD: At the moment, we have to say that we have to learn more. But we’re never going to be able to disregard the EMPA-REG findings.

Kenneth Snow, MD, MBA: At the very least, there is now data out there about the population of folks who clearly got a benefit with a particular agent. And now, really the question is, is it unique and is it unique to that population? And we’ll know more. This is not a theoretical question of “in the next few years as opposed to within the next couple of months.” We will learn more. And depending on what we see in the next couple of years, we will be much better informed.