The really big medical news today is the publication of the highly anticipated Empa-Reg Outcome study in the New England Journal of Medicine. It is the first large clinical trial with a diabetes drug (empagliflozin, Jardiance) to demonstrate a significant reduction in cardiovascular outcomes. The trial even resulted in a surprising, statistically significant reduction in mortality.
I asked Sanjay Kaul for his perspective on the trial. Here’s his detailed analysis:
This is really a big deal! The most impressive findings are early and profound reductions in all-cause and CV mortality and hospitalization for heart failure. Such a large and early treatment effect is unlikely to be related to effect on atherothrombotic events, glycemic control or blood pressure or weight lowering effect.
The p value for the primary endpoint is not persuasive enough to support superiority claim on the basis of a single trial. However, mortality clearly meets the FDA evidence guidance on this – prespecification, replication and preservation of type 1 error. One is tempted to view the mortality findings as implausibly large (too good to be true). However, given the large number of events (nearly 450 for all cause mortality and over 300 for CV mortality) and highly persuasive p values (<0.001), this is highly unlikely to be due to a play of chance.
So it would appear that the holy grail of diabetes disease management, i.e., reducing CV risk including mortality, has been finally achieved by empagliflozin in this trial.
An obvious question would be is this a class effect? Given the fact that the treatment benefit appears to be unrelated to desirable attributes (BP, blood sugar and weight lowering effects) that are common to SGLT2 inhibitors, that question has now become more difficult to answer. We will have to await the results of CVOT (cardiovascular outcomes trials) with other agents within the class. I would be interested in knowing if there are any unique effects of empagliflozin that separate it from others in the class.
Kaul shared a few additional thoughts:
It would be a mistake to view this trial as validation of the hypothesis that improved glycemic control results in improved CV outcomes.
I hope the data monitoring committees of ongoing CVOTs with other SGLT2 inhibitors will not act in haste and that they will allow these trials to continue. I am interested in knowing if the stroke signal seen with Empagliflozin is replicated in other trials, and in also knowing the clinical impact of these strokes, whether they were disabling or not. If these drugs increase disabling strokes, then the mortality benefit would be a wash in my opinion. Many physicians and patients would rather die than have a disabling stroke. Having said that, while total strokes showed a HR of 1.18, fatal strokes (a proxy for disabling strokes) were going in the right direction. I don’t think the dysutility index of strokes (Modified Rankin’s score, etc.) were formally assessed in EMPA REG OUTCOME trial.
While we are all quick to speculate the potential mechanisms, the reality is that we do not know the exact mechanism. This trial was designed to address CV outcomes, and not the underlying mechanism of benefit. All we can say is that it is unlikely to be mediated via blood pressure lowering (because strokes are not favorably impacted), weight loss (too large and too early), or glycemic effect (too large and early and the point estimate going in the wrong direction in those with baseline HbA1c >8.5).
Like all good RCTs, this trial raises more questions than provides answers.
Kaul disclosed that he has served as a consultant for Boehringer Ingelheim.
Update:
Adding this great perspective from James Stein in the comment section:
This is very exciting. The main reason we treat diabetes mellitus is to prevent its complications, most notably cardiovascular disease. We don’t treat blood sugar just for the sake of lowering it – we want to prevent the sequelae of high blood sugar. To date, there has been a dearth of evidence suggesting that even our best drugs prevent heart disease though they do prevent other complications of diabetes. I welcome a medication that treats diabetes mellitus and reduces risk of cardiovascular death risk. Nevertheless, it is one study and diabetes drugs have a checkered past when it comes to heart disease risk, we need replication and careful analysis. It is a little strange that some components of the primary endpoint did not even show trends, but that likely is do to small numbers (chance) or perhaps misclassification. No subgroup did worse, though some signals of older patients and those with less out of control diabetes doing better – they are hypotheses generating. It is interesting that the improvement in death, CV death, and HF is much great er than expected from a small reduction in HgbA1C. Unclear what the mechanism is, but I suspect it is not just the lower blood sugar levels. Since waist circumference and BP went down and HDL-C went up, it appears to be treating the Metabolic Syndrome, perhaps by reducing insulin resistance. Those salutary effects trumped a small increase in LDL-C. Of course the add-in therapy to the placebo arm (sulfonylureas mostly, also insulin) could have caused some harm so some of the apparent benefit maybe from the harm of how we treat diabetic patients. It also is safe with fewer AEs than placebo
except for more urinary infections and urosepsis. This is an advance to be applauded, explored and hopefully replicated.
This is very exciting. The main reason we treat diabetes mellitus is to prevent its complications, most notably cardiovascular disease. We don’t treat blood sugar just for the sake of lowering it – we want to prevent the sequelae of high blood sugar. To date, there has been a dearth of evidence suggesting that even our best drugs prevent heart disease though they do prevent other complications of diabetes. I welcome a medication that treats diabetes mellitus and reduces risk of cardiovascular death risk. Nevertheless, it is one study and diabetes drugs have a checkered past when it comes to heart disease risk, we need replication and careful analysis. It is a little strange that some components of the primary endpoint did not even show trends, but that likely is do to small numbers (chance) or perhaps misclassification. No subgroup did worse, though some signals of older patients and those with less out of control diabetes doing better – they are hypotheses generating. It is interesting that the improvement in death, CV death, and HF is much great er than expected from a small reduction in HgbA1C. Unclear what the mechanism is, but I suspect it is not just the lower blood sugar levels. Since waist circumference and BP went down and HDL-C went up, it appears to be treating the Metabolic Syndrome, perhaps by reducing insulin resistance. Those salutary effects trumped a small increase in LDL-C. Of course the add-in therapy to the placebo arm (sulfonylureas mostly, also insulin) could have caused some harm so some of the apparent benefit maybe from the harm of how we treat diabetic patients. It also is safe with fewer AEs than placebo
except for more urinary infections and urosepsis. This is an advance to be applauded, explored and hopefully replicated.