Sodium glucose cotransporter 2 inhibitors and risk of serious adverse events: nationwide register based cohort study

Sodium glucose cotransporter 2 (SGLT2) inhibitors have been associated with cardiovascular (CV) benefits in randomized clinical trials (RCTs) [1]. However, specific safety issues remain unclear. Ueda et al. observed an increased risk of lower limb amputation and of diabetic ketoacidosis with the use of SGLT2 inhibitors compared to glucagon-like peptide 1 (GLP1) receptor agonists (incidence rate 2.7 vs. 1.1 events per 1000 person years, HR 2.32 [95% CI 1.37; 3.91], and 1.3 vs. 0.6, HR = 2.14 [1.01; 4.52], respectively) [2].

Ueda et al. reported an original analysis of nationwide registers from Sweden and Denmark, to assess the risk of seven serious adverse events when using SGLT2 inhibitors. They used a propensity score matched cohort of 34 426 new users of SGLT2 inhibitors or GLP1 receptor agonists. Their results suggested potential harmful effects of SGLT2 inhibitors. We examined these results at the light of those of 3 powerful RCTs [3-5] assessing SGLT2 inhibitors in patients with type 2 diabetes, to discuss the agreement between observational and experimental data, taking into account the benefit/risk balance estimations.

First, the three CV outcomes RCTs assessing SGLT2 inhibitors [3-5] showed conflicting results regarding the risk of amputation. In the CANVAS Program [3], patients randomized to canagliflozin displayed higher rates of lower limb amputation (HR 1.97 [95% CI 1.41; 2.75]). In the EMPAREG OUTCOME trial [4, 6] and in the DECLARE TIMI 58 trial [5], the risk of amputation was not increased in the SGLT2 inhibitor group compared to placebo (HR=1.00 [95% CI 0.70; 1.44] and 1.09 [95% CI 0.84; 1.40], respectively); these confidence intervals excluding the HR estimated in CANVAS. As expected, the meta-analysis of the risk increase of amputation in those 3 RCTs displayed a significant heterogeneity (I² = 77%, supplementary figure available on request). The meta-analysis of the risk increase of diabetic ketoacidosis was consistent across the 3 RCTs (RR = 2.14 [95% IC 1.24; 3.71], I² = 0%, supplementary figure available on request).

Second, the treatment effect estimation through those three randomized, double-blinded, placebo controlled trials has a high level of evidence and a lower risk of bias than from pharmaco-epidemiological study. Diabetic ketoacidosis was a pre-specified safety outcome in all 3 RCTs; amputations were adverse events of interest in the CANVAS Program trial [3], collected retro- and prospectively in the DECLARE TIMI 58 trial [5], and searched for in post hoc analysis in the EMPAREG OUTCOME trial [6].

Third, register-based cohort are usually helpful as they allow big sample size and then to overcome the limit of statistical power for safety issue in RCT regarding rare events. However, the study of Ueda et al. included 17 213 patients treated with SGLT2 inhibitors, 61% with dapagliflozin, 38% with empagliflozin and only 1% with canagliflozin. The three CV RCTs pooled results on 19 056 patients treated with SGLT2 inhibitors, i.e. a similar sample size. However, the power of the study of Ueda et al. was considerably lower than the meta-analysis of the 3 RCTs, with only 62 amputations and 30 diabetic ketoacidosis against 552 and 62 in the meta-analysis, respectively (supplementary figure available on request).

Fourth, the meta-analysis of Zelniker et al. showed a decreased risk of major adverse cardiovascular events (MACE) and of CV death or hospitalization for heart failure with SGLT2 inhibitors (HR = 0.89 [95% CI 0.83; 0.96] and HR = 0.77 [95% CI 0.71; 0.84], respectively) [1]. The absolute risk difference for amputation, diabetic ketoacidosis, MACE and hospitalization for heart failure, estimated from the 3 RCTs, suggested a favorable benefit/risk balance for the SGLT2 inhibitors, even when taking into account the risk of amputation and of diabetic ketoacidosis (Table available on request).

To conclude, Ueda et al. suggested a risk of amputations with SGLT2 inhibitors, but the meta-analysis of more powerful RCTS excluded a class effect. The risk of diabetic ketoacidosis seemed to be related to the class, but did not offset the CV benefits. Pharmaco-epidemiology approach could help to identify safety signal, but its potentially biased estimation should be interpreted with caution. When both approaches disagree, as long as they are powerful enough, RCTs results dominate due to their higher level of evidence. Here, they advocate not to drop out drugs with positive benefit-risk balances.

Supplementary figure: Forest plot showing the effect of SGLT2 Inhibitors on risk of amputation and of diabetic ketoacidosis, through meta-analysis of the CANVAS Program, DECLARE TIMI 58 and EMPAREG OUTCOME trials (available on request)

References
1. Zelniker, T.A., et al., SGLT2 inhibitors for primary and secondary prevention of cardiovascular and renal outcomes in type 2 diabetes: a systematic review and meta-analysis of cardiovascular outcome trials. Lancet, 2018
2. Ueda, P., et al., Sodium glucose cotransporter 2 inhibitors and risk of serious adverse events: nationwide register based cohort study. BMJ 2018;363:k4365
3. Neal, B., et al., Canagliflozin and Cardiovascular and Renal Events in Type 2 Diabetes. N Engl J Med, 2017. 377(7): p. 644-657
4. Zinman, B., et al., Empagliflozin, Cardiovascular Outcomes, and Mortality in Type 2 Diabetes. N Engl J Med, 2015. 373(22): p. 2117-28
5. Wiviott, S.D., et al., Dapagliflozin and Cardiovascular Outcomes in Type 2 Diabetes. N Engl J Med, 2018
6. Inzucchi, S., et al., Empagliflozin and Assessment of Lower-Limb Amputations in the EMPA-REG OUTCOME Trial. Diabetes Care 2018 Jan; 41(1): e4-e5.