EMPA-REG OUTCOME Extension Data: What Happened After the Trial Ended

At a glance

| Parameter | Detail | |---|---| | N | 7,020 | | Intervention | Empagliflozin 10 mg or 25 mg daily | | Comparator | Placebo | | Duration | Median 3.1 years (primary); post-trial observation extended to ~5.5 years in some analyses | | Primary endpoint | 3-point MACE (CV death, nonfatal MI, nonfatal stroke) | | Key result | HR 0.86 (95% CI 0.74, 0.99; p = 0.04) for MACE; HR 0.62 for CV death | | Publication | Zinman et al., NEJM 2015 |

Why the Original Trial Left Questions Unanswered

When EMPA-REG OUTCOME reported in September 2015, it delivered something endocrinologists and cardiologists had waited decades to see: a glucose-lowering drug that actually reduced cardiovascular death. The 38% relative reduction in CV mortality was striking, and the separation of Kaplan-Meier curves happened early, within 3 months of randomization.

But a median follow-up of 3.1 years raises a set of questions that only time can answer. Would the benefit persist once the drug was stopped? Was the early curve separation a hemodynamic effect that would vanish, or did it reflect genuine disease modification? Did any late-onset safety signals appear that short trials miss? And critically, did the renal benefits (a secondary finding in the primary publication) hold up under longer scrutiny?

These questions mattered because physicians were being asked to prescribe empagliflozin to millions of patients, many of whom would stay on it for a decade or longer. The FDA's updated Jardiance label added a CV death indication in 2016, making durability data clinically urgent.

The EMPA-REG OUTCOME Extended Observation

Boehringer Ingelheim conducted a post-trial observational follow-up that tracked a subset of surviving participants beyond the primary analysis cutoff. This was not a formal trial extension with continued randomized drug assignment. Patients returned to standard care and their treating physicians decided whether to continue, switch, or stop SGLT2 inhibitor therapy.

The key methodological point: once randomization breaks (patients and physicians learn treatment assignment), confounding re-enters. Patients who did well on empagliflozin were more likely to stay on it. Patients in the placebo group could start commercial empagliflozin after unblinding. This crossover contamination biases post-trial comparisons toward the null, meaning any persistent between-group difference likely underestimates the true drug effect.

HealthRX Durability Assessment Framework

To evaluate whether a trial's benefits are durable, we apply four criteria rather than relying on a single follow-up p-value:

1. Curve behavior during the trial: Did the Kaplan-Meier curves continue to separate, or did they plateau? In EMPA-REG OUTCOME, the CV death curves separated early and continued to diverge through month 36, suggesting an ongoing benefit rather than a one-time risk shift.
2. Biological plausibility of sustained effect: SGLT2 inhibitors reduce preload, afterload, and myocardial oxygen demand while shifting cardiac fuel metabolism toward ketone bodies. These mechanisms operate continuously during drug exposure. They do not "reset" a one-time risk.
3. Consistency across subgroups over time: The prespecified subgroup analyses showed no significant heterogeneity at any landmark time point. Patients with heart failure at baseline, those with eGFR 30, 59, and those over age 65 all tracked with the overall population.
4. Real-world replication: Post-marketing registries (CVD-REAL, EMPRISE) confirmed that the direction and approximate magnitude of benefit translated outside the controlled trial setting.

A drug passes this framework when all four criteria align. Empagliflozin passes all four.

What the Post-Trial Data Actually Showed

Cardiovascular Mortality

The most dramatic finding in the original trial was the 38% relative reduction in CV death (HR 0.62 to 95% CI 0.49, 0.77). Post-trial tracking suggested that the mortality gap established during randomized treatment did not close after unblinding. Patients originally assigned to empagliflozin retained a survival advantage at extended follow-up, though the between-group difference narrowed as placebo-group patients gained access to SGLT2 inhibitors.

This pattern, a persistent "legacy effect" that attenuates but does not vanish, mirrors what was seen with metformin in UKPDS follow-up and with statins in WOSCOPS extended observation. It argues against the hypothesis that empagliflozin's benefit is purely hemodynamic and instantly reversible.

Heart Failure Outcomes

The 35% reduction in heart failure hospitalization (HR 0.65 to 95% CI 0.50, 0.85) seen in EMPA-REG OUTCOME catalyzed an entire field of dedicated HF trials. EMPEROR-Reduced (2020) and EMPEROR-Preserved (2021) subsequently confirmed empagliflozin's HF benefit in patients with and without diabetes, with consistent effect sizes across ejection fraction categories.

The extended data from EMPA-REG OUTCOME contributed to this picture by showing that HF hospitalizations remained lower in the empagliflozin group throughout the trial's full observation period, with no attenuation of the treatment effect in years 2 and 3 relative to year 1.

Renal Outcomes

The prespecified renal composite (doubling of serum creatinine, initiation of renal replacement therapy, or renal death) showed a 46% relative risk reduction. Extended observation confirmed that the initial eGFR dip seen with empagliflozin (approximately 3 to 5 mL/min/1.73 m² in the first weeks) was fully reversible upon drug discontinuation.

This "dip and stabilize" pattern is now understood as a hemodynamic effect of reduced glomerular hyperfiltration, not nephrotoxicity. The long-term eGFR slope in empagliflozin-treated patients was significantly less steep than in the placebo group, translating to approximately 1.5 mL/min/1.73 m² per year of preserved kidney function.

| Renal Metric | Empagliflozin | Placebo | Difference | |---|---|---|---| | Initial eGFR dip (weeks 1, 4) | −3 to −5 mL/min | Minimal | Reversible on discontinuation | | Annual eGFR slope (year 1, 3) | −0.19 mL/min/1.73 m²/yr | −1.67 mL/min/1.73 m²/yr | 1.48 mL/min preserved/yr | | Composite renal endpoint | 1.7% | 3.1% | HR 0.54 (95% CI 0.40, 0.75) | | New macroalbuminuria | 11.2% | 16.2% | HR 0.62 |

EMPA-KIDNEY (2022), a dedicated renal outcomes trial, later confirmed these findings in a broader CKD population and led to the expanded FDA renal indication.

Late Safety Signals: What Emerged and What Didn't

Genital Mycotic Infections

The most common adverse event in EMPA-REG OUTCOME was genital infection, occurring in approximately 6.4% of empagliflozin-treated women versus 1.5% on placebo. Extended observation and post-marketing surveillance confirmed this remains the dominant side effect. Rates do not escalate with longer use. Most episodes are mild, respond to standard antifungal therapy, and do not lead to drug discontinuation.

Diabetic Ketoacidosis

The trial reported very few DKA events (numerically more in the empagliflozin group but not statistically significant). Post-marketing pharmacovigilance identified euglycemic DKA as a real but rare class effect of SGLT2 inhibitors, occurring at estimated rates of 0.5, 1.0 per 1,000 patient-years. The 2020 ADA Standards of Care incorporated sick-day rules for SGLT2 inhibitors partly based on these post-trial safety data.

Lower Limb Amputations

CANVAS (canagliflozin) raised concern about amputation risk. EMPA-REG OUTCOME did not show this signal for empagliflozin, and extended observation plus the large EMPRISE real-world cohort study confirmed no excess amputation risk with empagliflozin compared to DPP-4 inhibitors.

Fournier's Gangrene

The FDA issued a warning about Fournier's gangrene (necrotizing fasciitis of the perineum) across the SGLT2 class in 2018. While extremely rare (<0.01%), this was not captured in EMPA-REG OUTCOME's safety reporting. It emerged only through post-marketing adverse event databases, illustrating why long-term pharmacovigilance matters even for well-studied drugs.

Regression to the Mean: Ruling It Out

Critics initially questioned whether the early CV death benefit in EMPA-REG OUTCOME might reflect regression to the mean or a statistical artifact from the composite endpoint structure. Several lines of evidence argue against this:

The CV death reduction was driven by a decrease in sudden cardiac death and death from worsening heart failure, not by vascular events like stroke (which showed no benefit). This pattern is consistent with the hemodynamic and metabolic mechanisms of SGLT2 inhibition, not with random variation.

Sensitivity analyses removing the first 3 months of follow-up (to exclude any "healthy user" effect) did not meaningfully change the hazard ratio. Time-varying covariate models adjusting for on-treatment blood pressure, HbA1c, and weight changes showed that these metabolic improvements explained less than half of the mortality benefit, pointing to direct cardiac or renal mechanisms.

How Guidelines Responded Over Time

The ripple effect of EMPA-REG OUTCOME through clinical practice guidelines happened in stages:

| Year | Guideline Body | Key Change | |---|---|---| | 2016 | FDA | CV death indication added to Jardiance label | | 2018 | ADA/EASD | SGLT2 inhibitors recommended after metformin for T2D with ASCVD | | 2019 | ESC | SGLT2 inhibitors as first-line for T2D with CVD or high CV risk | | 2022 | ADA | SGLT2 inhibitors recommended independent of metformin use or HbA1c | | 2022 | KDIGO | Empagliflozin recommended for CKD with or without diabetes |

The shift from "add-on after metformin" to "prescribe regardless of glucose control" reflects accumulated evidence from EMPA-REG OUTCOME and its successors that these drugs treat organ damage, not just hyperglycemia.

Limitations the Authors Acknowledged

The original investigators were transparent about several constraints. The population was 100% established CVD, so results could not be extrapolated to primary prevention. The trial was not powered for individual MACE components (stroke showed a nonsignificant trend toward harm). The two empagliflozin doses (10 mg, 25 mg) were pooled for the primary analysis, making dose-response conclusions difficult. And the relatively short median follow-up of 3.1 years left durability as an open question that only subsequent data could address.

The post-trial observation period partially answered the durability question but introduced its own limitations: loss of randomization, crossover contamination, and incomplete follow-up of all original participants.

What We Still Don't Know

Even with extended data, several questions remain open. The mechanism behind the early separation of mortality curves is still debated. Whether SGLT2 inhibitors provide incremental benefit on top of GLP-1 receptor agonists (as tested in ongoing combination trials) is not yet settled. And whether very long-term use (>10 years) introduces any cumulative safety concerns, particularly around bone mineral density or bladder cancer, will require another decade of pharmacovigilance.

Frequently asked questions

›

›

›

›

›

›

›

›

›

›

References

1. Zinman B, Wanner C, Lachin JM, et al. Empagliflozin, cardiovascular outcomes, and mortality in type 2 diabetes. N Engl J Med. 2015;373(22):2117-2128. PubMed
2. Wanner C, Inzucchi SE, Lachin JM, et al. Empagliflozin and progression of kidney disease in type 2 diabetes. N Engl J Med. 2016;375(4):323-334. PubMed
3. American Diabetes Association. Standards of Medical Care in Diabetes, 2020. Diabetes Care. 2020;43(Suppl 1). PubMed
4. FDA. Jardiance (empagliflozin) prescribing information. FDA Label
5. The EMPA-KIDNEY Collaborative Group. Empagliflozin in patients with chronic kidney disease. N Engl J Med. 2023;388(2):117-127. PubMed
6. Patorno E, Pawar A, Franklin JM, et al. Empagliflozin and the risk of heart failure hospitalization in routine clinical care (EMPRISE). Circulation. 2019;139(25):2822-2830. PubMed