(051) Incidence of Diabetic Ketoacidosis in Dapagliflozin Users vs. SGLT2i Non-users with Type 1 Diabetes Mellitus using a Japanese Claims Database

Background: Dapagliflozin, a sodium-glucose cotransporter 2 inhibitor (SGLT2i), was approved in Japan for treating type 1 diabetes mellitus (T1DM) in March 2019. Diabetic ketoacidosis (DKA) is a potentially life-threatening event among diabetes, especially for T1DM patients, and is an important identified risk of dapagliflozin. Additional pharmacovigilance for DKA in the dapagliflozin Risk Management Plan was requested by the Japanese regulatory authority. The Japan Medical Data Center (JMDC) claims database, an anonymized database of insurance claims, was used as it is reliable and in accordance with Good Post-Marketing Study Practice.

Objectives: To estimate the annual incidence of DKA in patients with T1DM treated with insulin and dapagliflozin vs. those treated with insulin without SGLT2i use (control group) between 26 March 2019 and 31 March 2021.

Methods: This cohort, post-marketing study utilized JMDC database data over 1 January 2005–31 March 2022. The primary objective was to estimate the annual incidence rate (IR) of DKA (including DKA hospitalization and DKA death) in dapagliflozin users vs. SGLT2i non-users; time-stratified sequential propensity score (PS)-matched cohorts were used in the primary analysis. Secondary outcomes included comparison of IRs between the dapagliflozin vs. control (matched cohort) using the Cox proportional model, the annual IRs and confidence intervals (CIs) of DKA events in the unmatched cohort, and the IR of repeated DKA events. The IR was calculated by dividing the number of events by the total follow-up period. CIs for IRs were derived based on a Poisson distribution.

Results: The overall cohort included 5,886 patients with T1DM; 327 were treated with dapagliflozin between March 2019 and March 2022. Following 1:3 PS matching, 327 dapagliflozin-treated patients and 980 matched controls were analyzed. Four patients had DKA events in the dapagliflozin group (IR [95% CI]: 1.54/100 person-years [PY] [0.42–3.95]) and 10 in the matched control group (IR [95% CI]: 1.14/100 PY [0.55–2.10]). The hazard ratio was 1.28 (95% CI: 0.40–4.09). In the unmatched overall cohort, the IRs (95% CIs) in the dapagliflozin group and the control group were 1.54/100 PY (0.42–3.95) and 0.97/100 PY (0.72–1.27), respectively.

Conclusions: The results did not suggest any unexpectedly higher risk of DKA than previously reported and hence did not support a need for any new measures to be taken regarding dapagliflozin prescription for patients with T1DM. A strength of this study is the use of a real-world Japanese database, which revealed a slightly numerically higher DKA incidence in the dapagliflozin group than the control group. However, the limited number of events prevented definitive conclusions.