No drug has been shown to prevent acute kidney injury after cardiac surgery, and the incidence is high enough that any candidate is worth attention. This double-blind placebo-controlled trial at two academic and five non-academic Dutch hospitals randomised 784 adults undergoing elective cardiac surgery to dapagliflozin 10 mg or placebo, starting the day before surgery and continuing to the second postoperative day — four doses in total. Median age was 68, three-quarters were male, and median eGFR was 80 mL/min/1.73 m2.
AKI by KDIGO criteria over seven postoperative days occurred in 28% on dapagliflozin against 52% on placebo (relative risk 0.54, 95% CI 0.45 to 0.65, p<0.001). Atrial fibrillation was identical at 45% in both arms, and reoperation was 11% against 10%. Follow-up was 99% complete.
A relative risk of 0.54 from four tablets is a large effect for a cheap, familiar drug, and the mechanism is plausible given what SGLT2 inhibitors do to renal haemodynamics. Two things temper it. The endpoint is creatinine and urine output rather than dialysis or death, and a drug that alters glomerular haemodynamics can move creatinine without changing what happens to the patient. And the trial says nothing about ketoacidosis risk in this setting, which is precisely where perioperative SGLT2 inhibitor use has caused trouble before — current advice is to stop these drugs before surgery, not start them.
So treat this as an important finding that is not yet an instruction. It justifies a hard-endpoint trial and a serious look at perioperative ketone monitoring. It does not justify adding dapagliflozin to a pre-operative order set on Monday.
- Do not change pre-operative practice yet; the endpoint is biochemical AKI, not dialysis or death.
- Note the tension with existing advice to withhold SGLT2 inhibitors before surgery because of ketoacidosis risk.
- If this is trialled locally, monitor perioperative ketones, not just creatinine.
- The population was largely male, White and with preserved eGFR — generalisation to Indian surgical cohorts is untested.
- Atrial fibrillation and reoperation rates were unchanged, so no signal of procedural harm.
The statistics, in plain English
A relative risk of 0.54 with an interval of 0.45 to 0.65 is a large, precisely measured effect — and because the placebo rate was 52%, the absolute difference is 24 percentage points, roughly four patients treated to prevent one episode. The catch is the endpoint. KDIGO acute kidney injury is defined by a rise in creatinine or a fall in urine output, both of which a drug acting on renal blood flow can shift directly. Until a trial shows fewer patients needing dialysis or dying, a change in the marker is not proof of a change in the disease.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for cardiology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free