Title - Euglycemic Diabetic Ketoacidosis Associated with SGLT2 Inhibitor Use in Non-Type 1 Diabetes Mellitus Admitted for Inpatient Hospital Care

Presentation Number: SUN 625
Date of Presentation: April 2nd, 2017

Lakshmi Kannan*1, Michael A Ruggero2, Ilona S Lorincz3 and Michael R. Rickels4
1University of Pennsylvania, PHILADELPHIA, PA, 2Hospital of The University of Pennsylvania, PHILADELPHIA, PA, 3University of Pennsylvania Hospital, Philadelphia, PA, 4University of Pennsylvania Perelman School of Medicine, Philadelphia, PA


Sodium – glucose cotransporter 2 (SGLT2) inhibitors have been associated with the development of euglycemic diabetic ketoacidosis (DKA), mainly when used off-label in patients with type 1 diabetes. Here we report on our hospital’s experience with SGLT2 inhibitor-associated euglycemic DKA in patients with non-type 1 diabetes and provide evidence for a relationship between early case detection and rapid resolution of acidosis. Euglycemic DKA was defined by the presence of an anion gap metabolic acidosis attributable to ketosis (serum beta hydroxybutyrate [B-OHB] ≥3 mmol/L) without marked hyperglycemia (blood glucose < 250 mg/dL). Five patients with non-type 1 diabetes were identified as developing SGLT2 inhibitor associated euglycemic DKA during inpatient hospitalization over an 8 month period between October 2015 and June 2016. Two patients had type 2 diabetes and three had type 2/pancreatogenous diabetes with one presenting post-Whipple’s resection of a pancreatic cyst and two others with pancreatitis. The median age of patients was 64 years (range 56 - 65), median duration of type 2 diabetes was 15 years, and median duration of SGLT2 inhibitor use was 6 months (range 1 – 18 mos). Four patients were receiving canagliflozin and one on empagliflozin. Median time to DKA from last dose of SGLT2 inhibitor was 2 days (range 0.75 - 6). Median time to DKA from a precipitating event of pancreatic insult and/or fasting restriction (n.p.o.) was 2 days (range 1 - 4). Median B-OHB level at diagnosis of euglycemic DKA was 5.3 mmol/L (range 3.3 - 5.9), and median blood glucose was 112 mg/dL (range 73 – 208). Glucosuria (>1000 mg/dL) was present at diagnosis in all patients and persisted for up to 12 days following the pre-hospital discontinuation of SGLT2 inhibitor use. Management with insulin and dextrose infusion led to resolution of DKA after 0.5 - 4 days. The time to diagnosis of euglycemic DKA from the identified precipitant was correlated to the time to resolution of DKA (r = 0.89; P < 0.05).

Conclusion: Hospitalization for pancreatic disease may represent an additional risk factor beyond fasting (n.p.o) for euglycemic DKA in patients on SGLT2 inhibitor therapy, and this risk can be minimized by earlier discontinuation of SGLT2 inhibitor therapy prior to planned surgery. Prompt recognition of euglycemic DKA can lower morbidity, and should involve monitoring urinary glucose (and ketones) in fasted patients exposed to SGLT2 inhibitors in whom on-going catabolism can be addressed by providing insulin and dextrose therapy even when blood glucose is “on-target” for a hospitalized patient.


Nothing to Disclose: LK, MAR, ISL, MRR