Lessons from the bedside: ketoacidosis and SGLT2 inhibitors

Sodium–glucose cotransporter type 2 (SGLT2) inhibitors are an emerging oral glucose-lowering therapy. In May 2015, the United States Food and Drug Administration issued a warning amid observation of ketoacidosis cases that may have been associated with SGLT2 inhibitor use in people with type 2 diabetes (T2DM). Potential mechanisms include inhibition of insulin secretion through glycosuria, increased glucagon secretion and hypovolaemia.1-3

We report two recent cases that highlight predisposing factors and characteristics of SGLT2-associated ketoacidosis.

Our first patient was a 51-year-old Sri Lankan man with T2DM of 5 years duration that had been managed with metformin. Dapagliflozin had been commenced 2 weeks before admission with a myocardial infarction and diabetic ketoacidosis (serum pH, 6.88 [reference interval (RI), 7.38–7.43]; capillary blood glucose, 22 mmol/L; capillary ketones, 6.0 mmol/L). His C-peptide level (1.6 nmol/L; RI, 0.3–2.3 nmol/L) was inappropriately normal for hyperglycaemia. However, given the patient’s body mass index of 30 and undetectable glutamic acid decarboxylase (GAD) and islet antigen 2 (IA2) antibodies, the C-peptide level was likely due to glucose toxicity of islet cells rather than indicating a misdiagnosis of type 1 diabetes. The patient was discharged on insulin therapy and metformin.

Our second patient was a 48-year-old woman of European descent electively admitted for cerebral aneurysm clipping. T2DM was diagnosed following gestational diabetes that did not resolve postpartum. GAD and IA2 antibodies were not detected. She had never experienced diabetic ketoacidosis and was overweight, with previous bariatric surgery. On admission, she was on a relatively significant amount of basal–bolus insulin (total daily dose of 52 units), metformin and canagliflozin (initiated 12 months earlier). Reduced insulin doses were given the night before surgery and glucose-lowering agents were withheld. Owing to poor oral intake from nausea, insulin was withheld but glucose-lowering agents were recommenced on Day 1. Investigations from Day 2 revealed euglycaemic ketoacidosis (serum pH, 7.14; capillary blood glucose, 7.1 mmol/L; capillary ketones, 3.4 mmol/L). Lactic acidosis was excluded as the patient’s lactate level was 0.7 mmol/L (RI, 0.6–2.2 mmol/L). Treatment with intravenous insulin infusion was required. Metformin and canagliflozin were ceased indefinitely and she was discharged on insulin.

A notable characteristic of ketoacidosis involving SGLT2 inhibitors is the serum glucose level, which is often modestly elevated and sometimes normal. This can delay diagnosis of serious metabolic derangements. Increasing recognition of drug-associated ketoacidosis should caution clinicians to withhold SGLT2 inhibitors in situations of intercurrent illness and prolonged fasting. Reduction of insulin doses, particularly in patients with higher insulin requirements and therefore low endogenous insulin production, should be monitored carefully and rapid reductions avoided.