Effect of Sodium–Glucose Cotransporter 2 Inhibitors on Diabetic Ketoacidosis Among Patients With Type 2 Diabetes: A Meta-analysis of Randomized Controlled Trials

Sodium–glucose cotransporter 2 (SGLT2) inhibitors are a novel class of antidiabetes drugs for the treatment of type 2 diabetes (T2D) (1). In addition to their hypoglycemic effect, SGLT2 inhibitors also offer several beneficial effects, such as weight loss and blood pressure reduction (1). However, the overall health benefits of these drugs needed to outweigh their possible side effects. Recently, cumulative evidence suggests that SGLT2 inhibitors may lead to diabetic ketoacidosis (DKA), which is a serious acute complication of diabetes (2,3). In May 2015, the U.S. Food and Drug Administration issued an updated drug safety communication warning about SGLT2 inhibitors potentially increasing the risk of DKA (4). As DKA is a rare adverse effect, the evidence from individual studies or simply pooling the numbers from multiple reports is generally weak. Therefore, we conducted a meta-analysis of randomized controlled trials (RCTs) to examine whether SGLT2 inhibitors affect the risk of DKA in patients with T2D.

We searched PubMed, Embase, Cochrane Central Register of Controlled Trials (CENTRAL) and ClinicalTrials.gov from inception to 27 January 2016 to identify the published and unpublished RCTs of SGLT2 inhibitors that reported DKA events in patients with T2D. Two reviewers (H.T. and D.L.) independently performed the study selection, data extraction, and quality assessment. A Peto odds ratio (OR) with 95% CI was used due to a very low event rate. The I² statistic was used to detect the possible between-study heterogeneity. All statistical analyses were performed with STATA version 14.

A total of 10 eligible RCTs involving 13,134 patients and 14 DKA events were identified from 1,268 citations (5–14). Overall, the event rates were 0.1% in the group of SGLT2 inhibitor users versus 0.06% in the control groups. The meta-analysis results are shown in Fig. 1. Overall, SGLT2 inhibitor groups were not associated with a significantly higher risk of DKA compared with the control groups (OR 1.71 [95% CI 0.56, 5.20]). Furthermore, our subgroup analyses showed that SGLT2 inhibitors were not significantly associated with an increased risk of DKA when compared with placebo (1.98 [0.56, 6.94]) or dipeptidyl peptidase 4 (DPP-4) inhibitors (1.00 [0.09, 11.01]). No statistical heterogeneity was observed in the analyses, except for the subgroup analysis of SGLT2 inhibitors versus DPP-4 inhibitors (I² = 66.7%).

Previous trials reported increased DKA cases with the use of SGLT2 inhibitors, especially when they were used off-label in patients with type 1 diabetes (2,3). Some plausible mechanisms are already proposed by which SGLT2 inhibitors might trigger DKA (3,15). Given the current evidence from RCT data, we found that SGLT2 inhibitors were not significantly associated with an increased risk of DKA among patients with T2D. Consistent with a previous report (15), the frequency of reported DKA events related to SGLT2 inhibitor treatment in T2D patients is less than 0.1%. By synthesizing cumulative evidence from RCTs, our study did not support the adverse effect on DKA among T2D patients. Although the null results presented the highest strength of evidence from available RCT data, we cannot rule out the possibilities of a modest effect on DKA by SGLT2 inhibitors, an effect on a specific clinical phenotype of DKA (e.g., euglycemic DKA), or a nonclass effect. There is some evidence indicating that the risk of euglycemic DKA related to SGLT2 inhibitors may be increased among long-standing T2D patients with marked β-cell insufficiency, in latent autoimmune diabetes in adults, or under other severe medical conditions (15). In this regard, further safety monitoring based on a larger number of cases and detailed clinical information on related DKA cases is warranted to resolve the uncertainty about this specific drug safety issue.