SGLT2 inhibitor initiation in acute decompensated heart failure: evidence update?

68F admitted with acute decompensated HFrEF (EF 25%), NYHA class IV symptoms. She is stabilized on IV furosemide 40mg q8h, currently making good urine (2.5 L/day). Creatinine 2.1, eGFR 32.

The EMPULSE trial showed benefit for empagliflozin initiation during AHF hospitalization. The SOLOIST-WHF trial had similar signals with sotagliflozin. But I remain cautious about starting SGLT2i with this degree of renal impairment.

Specific questions:

At what eGFR threshold do you hold SGLT2i initiation in AHF?
Do you monitor renal function differently when starting in the acute setting vs outpatient?
Has anyone seen significant AKI attributable to inpatient SGLT2i initiation?
Empagliflozin vs dapagliflozin preference in this setting?

As a nephrologist who sees these patients frequently, here is my practical approach:

1. eGFR threshold: I no longer have a hard cutoff. EMPA-KIDNEY enrolled patients down to eGFR 20, and the DAPA-CKD trial went to eGFR 25. In AHF, I initiate at eGFR ≥ 20 as long as the patient is hemodynamically stable and euvolemic or nearing euvolemia.

2. Monitoring: In the acute setting, I check creatinine at 48h and 7 days post-initiation. The expected initial eGFR dip is 3 to 5 mL/min. If the dip exceeds 10 mL/min or if urine output drops below 0.5 mL/kg/h, I hold and reassess volume status before blaming the SGLT2i.

3. AKI risk: In my experience, the "AKI" attributed to SGLT2i is almost always a hemodynamic dip from volume contraction in patients who are over-diuresed. The SGLT2i itself causes osmotic diuresis of roughly 300 to 500 mL/day, which compounds loop diuretic effect. I routinely reduce loop diuretic dose by 20 to 30% when adding SGLT2i.

4. Agent selection: No meaningful clinical difference between empagliflozin and dapagliflozin in this setting. I use whichever is on formulary. Both are dosed once daily, both are well tolerated.

The real risk is not starting the SGLT2i. The EMPULSE data showed a number needed to treat of 14 for the composite of death, HF events, time to first HF event, and change in KCCQ score at 90 days. That is a strong signal.

Agree with Dr. Ghosh. I want to add the timing angle. In EMPULSE, patients were enrolled once stabilized (no IV vasopressors, not in ICU). I do not start SGLT2i while the patient is still on IV inotropes or if they required vasopressors in the last 24 hours.

My clinical rule: once IV diuretic dose is stable (not requiring uptitration) for 24 hours and the patient is making adequate urine, that is my window for initiation.

One practical tip: start at the full dose (empagliflozin 10mg or dapagliflozin 10mg). There is no evidence supporting a "start low" approach, and subtherapeutic dosing delays the hemodynamic benefits.

From the general medicine ward perspective, I see many of these patients downstream after the cardiology team has initiated SGLT2i. The one complication I have encountered more than expected is euglycemic DKA in patients with borderline nutritional status. We had two cases in the last year where diabetic patients on insulin who were also NPO or on calorie-restricted diets developed DKA with blood glucose under 200.

Not a reason to withhold the drug, but worth flagging for ward teams who may not be as familiar with this entity.

3 Answers