How SGLT2 Inhibitors Became Heart Failure Therapy

Sodium-glucose cotransporter 2 inhibitors began as glucose-lowering medicines for type 2 diabetes. Their heart failure role emerged when cardiovascular safety trials repeatedly showed fewer hospitalizations for heart failure. Dedicated trials then enrolled people because they had heart failure, included participants without diabetes, and demonstrated benefit across reduced, mildly reduced, and preserved left ventricular ejection fraction.

The progression matters. An unexpected secondary signal can generate a hypothesis. It does not create a new indication by itself. DAPA-HF, EMPEROR-Reduced, EMPEROR-Preserved, and DELIVER converted the signal into direct randomized evidence with prespecified heart failure outcomes.

The regulatory experiment came first#

After concerns about cardiovascular harm from some glucose-lowering therapies, FDA required new type 2 diabetes medicines to demonstrate acceptable cardiovascular safety. Sponsors conducted large outcome trials in people at high cardiovascular risk. The primary goal was generally to rule out unacceptable excess risk in major cardiovascular events.

These trials created a useful platform for discovery. They enrolled thousands of participants, adjudicated cardiovascular outcomes, followed people over time, and compared active drug with placebo on top of usual care. Heart failure hospitalization was not always the primary endpoint, but it was measured consistently enough to reveal a pattern.

That history also warns against rewriting the original question. A diabetes cardiovascular-outcomes trial does not automatically tell you how to treat a general heart failure population; participants may have established atherosclerotic disease, specific kidney thresholds, and background diabetes care that differ from a heart failure clinic.

EMPA-REG OUTCOME produced the signal#

EMPA-REG OUTCOME enrolled 7,020 people with type 2 diabetes and established cardiovascular disease and randomly assigned empagliflozin or placebo. The primary three-point major cardiovascular composite was lower with empagliflozin, and cardiovascular death was also lower.

Heart failure hospitalization occurred in 2.7% of the empagliflozin group and 4.1% of the placebo group, and in a later heart failure analysis, the combined outcome of heart failure hospitalization or cardiovascular death occurred in 5.7% versus 8.5%, with a hazard ratio of 0.66.

The early separation of event curves and modest glucose differences suggested that the result was not explained solely by long-term glycemic control. Possible mechanisms included natriuresis, osmotic diuresis, kidney hemodynamics, plasma-volume change, metabolic shifts, vascular effects, and altered cellular energetics. No one mechanism was established as the complete explanation. The result was striking but still exploratory for a broader heart failure indication. The next study needed to enroll people with heart failure, define ejection fraction and symptoms, and include people without diabetes.

DAPA-HF tested reduced ejection fraction directly#

DAPA-HF enrolled 4,744 participants with symptomatic heart failure and ejection fraction at or below 40%. Participants received dapagliflozin 10 mg or placebo in addition to recommended therapy. About 45% did not have type 2 diabetes.

The primary composite was worsening heart failure, defined as hospitalization or an urgent visit requiring intravenous therapy, or cardiovascular death. It occurred in 16.3% of the dapagliflozin group and 21.2% of the placebo group over a median 18.2 months, giving a hazard ratio of 0.74.

The effect was similar by diabetes status. That subgroup finding was prespecified and biologically important: the medicine was acting as heart failure therapy rather than only correcting high glucose. Symptoms measured by a heart-failure questionnaire also favored dapagliflozin on average.

Absolute benefit depends on baseline risk and follow-up, and the 4.9 percentage-point difference in the primary composite corresponds to about 20 people treated over the trial period to prevent one primary event, using a simple reciprocal calculation. That number is trial-specific, not a permanent property of the drug you are prescribing.

EMPEROR-Reduced tested reproducibility#

EMPEROR-Reduced enrolled 3,730 participants with symptomatic heart failure and ejection fraction at or below 40%, with selection that produced a population at relatively high risk. Empagliflozin reduced the composite of cardiovascular death or heart failure hospitalization, with a hazard ratio of 0.75.

The result was driven mainly by fewer heart failure hospitalizations. Kidney-function decline was slower in the empagliflozin group under the trial's analyses. Benefits again appeared regardless of diabetes status.

Replication with another drug and related outcome strengthened a class-level inference for SGLT2 inhibition in reduced-ejection-fraction heart failure. It did not make every molecule interchangeable. Dapagliflozin and empagliflozin had the direct dedicated evidence, specific doses, labels, kidney thresholds, and safety data.

Preserved ejection fraction was the harder test#

Heart failure with preserved ejection fraction is heterogeneous. Participants can have different combinations of obesity, hypertension, atrial fibrillation, kidney disease, pulmonary vascular disease, ischemia, and age-related change. Earlier drug trials often improved a physiologic measure without clearly reducing major outcomes across the full population.

EMPEROR-Preserved enrolled 5,988 participants with symptomatic heart failure and ejection fraction above 40%, and the primary composite of cardiovascular death or heart failure hospitalization occurred in 13.8% with empagliflozin and 17.1% with placebo, a hazard ratio of 0.79. The result was driven mostly by fewer hospitalizations.

DELIVER enrolled 6,263 participants with symptomatic heart failure and ejection fraction above 40%, including people whose ejection fraction had previously been 40% or lower and later improved; the composite of worsening heart failure or cardiovascular death occurred in 16.4% with dapagliflozin and 19.5% with placebo, a hazard ratio of 0.82. DELIVER's effect was consistent across prespecified ejection-fraction groups and glycemic categories. Together with EMPEROR-Preserved, it filled an evidence gap above 40% and supported guideline expansion.

Composite endpoints need to be opened#

All four dedicated trials used composites. A composite increases event count and can capture several ways heart failure worsens, but its components may differ in clinical importance and treatment effect.

In the preserved-ejection-fraction trials, benefit was driven mainly by fewer heart failure hospitalizations or urgent worsening events. Cardiovascular death alone was not clearly reduced in each trial. Saying the drugs “reduced cardiovascular death and hospitalization” without opening the composite can imply both components were separately significant.

Recurrent hospitalizations also matter because one person can be admitted several times. First-event analyses count only the first. Supporting analyses that include total events can better represent the burden you are trying to reduce, though they use different statistical assumptions.

Patient-reported symptoms provide another dimension. Average questionnaire improvements were modest, and response distributions overlap. Trials should spell out both what happened clinically and how people felt and functioned.

Guidelines followed the dedicated evidence#

The 2022 AHA, American College of Cardiology, and Heart Failure Society of America guideline made SGLT2 inhibition one of four foundational medication classes for symptomatic chronic heart failure with reduced ejection fraction, regardless of diabetes status. It gave a weaker but favorable recommendation for mildly reduced and preserved ejection fraction based on evidence available at the time.

The 2023 European Society of Cardiology focused update incorporated EMPEROR-Preserved and DELIVER and gave dapagliflozin or empagliflozin a Class I, level A recommendation for symptomatic heart failure with mildly reduced or preserved ejection fraction to reduce hospitalization or cardiovascular death. Guideline class is not a substitute for eligibility. Heart failure diagnosis, ejection fraction, congestion, blood pressure, kidney function, type of diabetes, frailty, interactions, cost, and current product labeling still shape treatment.

Safety and implementation are part of the evidence#

SGLT2 inhibitors increase urinary glucose and can increase genital mycotic infections. Volume depletion and symptomatic hypotension can occur, especially with diuretics or poor intake, and kidney filtration may show a small early dip before the longer-term trajectory, a pattern that needs interpretation rather than automatic discontinuation.

Euglycemic diabetic ketoacidosis is uncommon but serious. Risk rises with type 1 diabetes, acute illness, fasting, surgery, marked insulin reduction, or very low carbohydrate intake. These medicines are not routine heart failure therapy for type 1 diabetes under current labeling and guidance.

Fournier gangrene is rare, and product labels include warnings. Canagliflozin's earlier amputation signal in a diabetes trial illustrates why safety should not be assumed identical across every agent and population. Heart failure trials have their own event definitions and follow-up. Implementation needs a sick-day and perioperative plan, review of volume status and diuretics, kidney-function monitoring in context, and clear education about ketoacidosis symptoms, and these details are clinical care, not secondary paperwork.

What remains uncertain#

Trials underrepresent some people with very low blood pressure, advanced kidney disease, recent instability, severe frailty, or multiple comorbidities. Results may not transport unchanged to every excluded group, and the person you are treating may belong to one. Very advanced heart failure and acute shock require separate evidence.

Mechanism remains plural. Natriuresis and osmotic effects occur, but the event benefit cannot be reduced to one diuretic mechanism. Kidney protection, cardiac loading conditions, inflammation, metabolism, and cellular signaling may contribute differently over time.

Comparative effectiveness between dapagliflozin and empagliflozin has not been established by a large head-to-head heart failure outcomes trial. Cross-trial comparisons are confounded by different eligibility, event rates, follow-up, and endpoints. Long-term population effects also depend on uptake, adherence, affordability, and whether other foundational therapy is optimized: a highly efficacious trial result can have little public-health effect if access or implementation fails.

A reusable evidence sequence#

The SGLT2 story offers a disciplined pattern for repurposing. First, identify a reproducible signal in trials designed for another primary question. Second, test the new hypothesis in the target disease population. Third, include participants who separate the new effect from the original indication, here people without diabetes. Fourth, replicate with another agent and broaden the phenotype carefully. Fifth, update guidelines only after direct outcomes evidence.

At each stage, preserve the actual endpoint. A safety trial signal is not a dedicated efficacy result. A composite is not all its components. A subgroup is not a new population until it has adequate support. A guideline is not a personalized prescription.

Related guides explain heart failure with preserved versus reduced ejection fraction and why newer diabetes drugs undergo cardiovascular-outcomes trials. The site's research overview connects trial appraisal to chronic-disease reasoning.

References#

1. EMPA-REG OUTCOME
2. DAPA-HF
3. EMPEROR-Reduced
4. EMPEROR-Preserved
5. DELIVER
6. 2022 US heart failure guideline
7. 2023 ESC focused heart failure update