SGLT2i in HFmrEF and HFpEF across the spectrum of glycemia

Hello. I'll present the primary results from DELIVER or to provide an update based on glycemic subgroups in the trial.

For background, the SGLT2 inhibitors reduce morbidity and mortality in patients with heart failure and reduced ejection fraction, typically defined as a left ventricular ejection fraction under 40%. Their use is now strongly recommended in current clinical practice guidelines. Few pharmacological treatment options are available, however, for patients with heart failure and either mildly reduced or preserved ejection fraction or HFpEF, and this represents about half of our patients with heart failure. The EMPEROR-Preserved trial demonstrated a reduction in CV death or heart failure hospitalizations with the SGLT2 inhibitor empagliflozin in this population. What is the effect of dapagliflozin in this setting, and specifically, how does dapagliflozin affect outcomes across glycemic subgroups?

This is a study design of DELIVER. It was a multinational randomized clinical trial. To be enrolled in DELIVER, participants had to be over 40 years old and have New York Heart Association class either II, III, or IV with left ventricular ejection fraction greater than 40%, as well as evidence of structural heart disease and increased natriuretic peptides. Patients were recruited from ambulatory clinics but also from hospitals. Participants were randomized to either 10 milligrams of dapagliflozin, an SGLT2 inhibitor, or placebo, and the primary endpoint was time to first cardiovascular death or worsening heart failure.

This is the primary endpoint from DELIVER. The placebo event rate was 9.6 per 100 person-years, reduced by dapagliflozin to 7.8 per 100 person-years. This translated to a hazard ratio of 0.82, so a relative risk reduction of 18% with a highly significant p-value. These now are the components of the primary endpoint on the left worsening heart failure defined as heart failure hospitalization or urgent heart failure visits. This was reduced relatively by 21%. For cardiovascular death, the results were not statistically significant, but there was a numerical reduction with a hazard ratio of 0.88.

These are now pre-specified subgroups from DELIVER. First, let's look at ejection fraction divided to patients with 40% to 49% ejection fraction, 50% to 59%, and greater than 60%. One can see consistency of effect and no statistical heterogeneity whatsoever. For those enrolled within 30 days, most of these were enrolled during the course of their hospitalization or soon thereafter. Again, no heterogeneity compared to the group of patients recruited after 30 days. An interesting subgroup are those patients who used to have HFrEF and now have improved ejection fraction. Again, consistency of effect.

In this, a meta-analysis involving patients from an earlier dapagliflozin trial known as DAPA-HF. These would be patients with reduced ejection fraction or HFrEF combined with those patients from DELIVER, looking at effectiveness for the primary outcome across the entire spectrum of left ventricular ejection fraction. One can see entire consistency across these two trials. What about other subgroups of interest? One is whether patients with or without diabetes had similar benefits, and the answer is absolutely yes. Finally, baseline renal function, so GFRs above or below 60. No heterogeneity in that effect as well.

For new data, we asked the following questions, what is the distribution of normoglycemia versus pre-diabetes versus type 2 diabetes in DELIVER, and how did these individual subgroups behave in terms of their event rates, irrespective of treatment assignment? Then what were the dapagliflozin treatment effects in each of these subgroups? Did they differ? Also, any trends in the subgroups based on left ventricular ejection fraction? Now focusing in on the type 2 diabetes subgroup, did the effect differ based on the duration of diabetes, the baseline hemoglobin A1C, or by baseline antihyperglycemic therapy? This is the breakdown based on glycemic status. One can see that about 45% of patients had type 2 diabetes at baseline, and we added another 5% based on the enrollment hemoglobin A1C of 6.5% or higher, so newly diagnosed patients, so there were about 50% of the DELIVER participants with diabetes. For those without diabetes, it's about a 2:1 ratio between those with prediabetes. This is defined by the American Diabetes Association criteria of an A1C of 5.7% to 6.4%. The balance, those would be patients without a history of diabetes and with an A1C under 5.7%, were considered normoglycemic. It's interesting to note that less than 20%, so less than one in five of the DELIVER patients actually, were considered normoglycemic.

Here now are the key outcomes by glycemic status, and I'd focus on the upper left panel, the primary outcome. One can see that as the dysglycemic status worsened, so did the event rates. The highest event rates were in the type 2 diabetes subgroup intermediate for prediabetes and the best outcomes in those patients that were considered normoglycemic. This is the hemoglobin A1C distribution at baseline in DELIVER. The mean hemoglobin A1C was 6.6%. Focus on the primary outcome and the worsening heart failure event outcome. One can see a linear relationship between the baseline hemoglobin A1C and worsening event rates. Let's look now at the primary outcome by the glycemic status of the patient, the three glycemic subgroups and treatment. One can see that whether we're talking about the normoglycemic patients, those with prediabetes, or those with type 2 diabetes, consistently, the lower event rate was with dapagliflozin. Here again, an 18% overall relative risk reduction. One can see on this forest plot there is consistency of effectiveness across each of the three glycemic subgroups without any statistical interaction. This analysis looks at the hemoglobin A1C as a continuous variable. Again, for the primary outcome and for worsening heart failure events, there is absolute consistency, no significant interaction.

Now, what about the type 2 diabetes subgroup itself? Diabetes duration and hemoglobin A1C at baseline did not influence the benefit of dapagliflozin, also baseline treatment with metformin and baseline treatment with insulin consistency of effect. We did see a modest interaction based on baseline sulfonylurea use. That would be patients who took sulfonylureas appear to benefit more. This may be a statistical fluke. We don't see this in other SGLT2 inhibitor trials, and certainly deserving of further evaluation.

Finally, this is the treatment effect of dapagliflozin versus placebo in the individual three glycemic subcategories. Normoglycemia in green, prediabetes in blue, and type 2 diabetes in red. Based on the range of ejection fraction studied in DELIVER, again, consistency of effectiveness.

We can conclude that dapagliflozin reduced the risk of worsening heart failure or CV death by 18% in patients with heart failure and mildly reduced or preserved ejection fraction who were enrolled in the DELIVER trial. Dapagliflozin's benefits are consistent across the range of left ventricular ejection fraction. The DELIVER population consisted of about 50% patients with type 2 diabetes. About 30% with prediabetes and less than 20% were considered normoglycemic. In general, the event rates increased with worsening glycemic status, both categorically as well as by hemoglobin A1C as a continuous variable. There was no statistical interaction between the categorical glycemic subgroupings nor by A1C as a continuous variable and dapagliflozin's benefits. Also, no significant trends based on diabetes duration, hemoglobin A1C at baseline, or baseline antihyperglycemic therapy with either metformin or insulin or based on the left ventricular ejection fraction. We can conclude that dapagliflozin improves heart failure outcomes in heart failure with mildly reduced or preserved ejection fraction across the spectra of both ejection fraction and glycemia.