The Comparative Safety of Individual SGLT2 Inhibitors and the Risk of Diabetic Ketoacidosis

Background: SGLT2 inhibitors (SGLT2i) are increasingly prescribed to treat type 2 diabetes (T2D) given their cardiovascular benefits. However, SGLT2i are known to increase the risk of diabetic ketoacidosis (DKA), a rare yet serious outcome. It remains unclear whether individual SGLT2i are associated with a differential risk of DKA.

Objectives: The primary objective of this large population-based cohort study is to determine whether the use of dapagliflozin or empagliflozin is associated with the risk of DKA compared with the use of canagliflozin. The secondary objective was to determine molecule-specific effects compared to an external comparison of dulaglutide, a glucagon-like peptide-1 receptor agonist, or sitaglipin, a dipeptidyl peptidase-4 inhibitor.

Methods: Using 3 US healthcare claims databases (Optum Clinformatics Data Mart Database, IBM MarketScan and Medicare fee-for-service) from January 2014 to June 2021, we assembled 3 cohorts of adults with type 2 diabetes. For cohort 1, we considered patients initiating individual SGLT2i (canagliflozin, dapagliflozin, empagliflozin) for a within-class comparison of SGLT2i. For cohort 2, we considered patients initiating the first of either an SGLT2i or dulaglutide (out of class comparison #1). For cohort 3, we considered patients initiating the first of either an SGLT2i or sitagliptin (out of class comparison #2). Patients were followed using an on-treatment exposure definition, allowing for a 60-day grace period between prescriptions. Cohorts were reweighed using propensity score matching weights, accounting for >130 baseline patient characteristics. Hazard ratios (HRs) and 95% CIs were estimated for hospitalization for DKA in each database and pooled using fixed-effects models with inverse variance.

Results: We identified 665,691 patients newly treated with SGLT2i, 711,383 patients newly treated with SGLT2i or dulaglutide, and 897,922 patients newly treated with SGLT2i or sitagliptin. After propensity score weighting, the cohorts were balanced on all baseline confounders. Over a median follow-up of 6 months, canagliflozin use was not associated with a different risk of DKA compared with use of empagliflozin [HR (95% CI) = 1.11 (0.95-1.29)]. However, the use of dapagliflozin was associated with a slight reduction in the risk of DKA compared with empagliflozin [HR (95% CI) = 0.79 (0.65-0.96)]. All 3 SGLT2i were associated with a similar increase in DKA risk compared with dulaglutide, with HRs ranging from 1.55 (95% CI = 1.16-2.06) for dapagliflozin to 2.42 (95% CI = 1.91-3.05) for canagliflozin. All 3 SGLT2is were associated with an increased risk of DKA compared to sitagliptin: canagliflozin [HR (95% CI) = 1.71 (1.39-2.10)]; dapagliflozin: [HR (95% CI) = 1.39 (1.09-1.77)]; empagliflozin: [HR (95% CI) = 1.71 (1.42-2.05)].

Conclusions: In this large cohort study,
all three SGLT2is were associated with an increased risk of DKA compared with dulaglutide or sitagliptin. Dapagliflozin users appeared to have a smaller increase in risk of DKA compared to other SGLT2is. Replication of these findings in other populations is warranted.