LAUSR

To the Editor: The results of the EMPA-REG Outcome trial heralded the cardiovascular (CV) relevance of sodium-glucose co-transporter-2 inhibitors (SGLT-2is), showing an unprecedented reduction of CV death by 38% over standard of care. Then, results from the CANVAS programme, DECLARE-TIMI 58, CREDENCE and the latest VERTIS-CV did not confirm this finding and challenged the beliefs of the scientific community that SGLT-2is may reduce CV death. Specifically, the VERTIS-CV trial failed to show the superiority of ertugliflozin over placebo with respect to both endpoints of major adverse cardiovascular events (MACE) and death from CV causes or hospitalization for heart failure (HF). However, a recent meta-analysis of all the available CV outcome trials (CVOTs) with both SGLT-2is and sotagliflozin, a dual SGLT-1/ SGLT-2i, displayed an overall significant reduction in the risk of CV death (HR 0.85, 95% CI 0.77-0.94, P = .001) with a moderate degree of heterogeneity (I = 36%, P = .13). In the SGLT-2i CVOTs, most CV deaths (82.7%) occurred in patients with established CV disease, although no significant interaction was found for those in primary versus secondary prevention (P = .41). Yu et al. observed that the EMPA-REG Outcome trial had the highest proportion of deaths as a result of CV events (66.7% of all-cause deaths) and hypothesized that this could have contributed to the heterogeneity of the effects of SGLT-2is on CV mortality. However, in VERTIS-CV, which was excluded from that analysis, an even higher proportion of CV deaths was counted (72.2% of all-cause deaths), yet the effect of ertugliflozin on this particular outcome was neutral. Moreover, both EMPA-REG Outcome and VERTIS-CV enrolled only patients with ascertained CV disease, and subgroup analyses aiming to investigate the potential impact of baseline HbA1c, estimated glomerular filtration rate, albuminuria and history of HF on risk reduction of CV death failed to show any significant interaction. In interpreting the different results of EMPA-REG Outcome and VERTIS-CV, a drug-related benefit on the risk of CV death seems unlikely: empagliflozin and ertugliflozin share a similar selectivity for SGLT-2, higher compared with dapagliflozin, canagliflozin and sotagliflozin; moreover, in patients with HF and reduced ejection fraction of the left ventricle, the risk of CV death was reduced by dapagliflozin and not by empagliflozin. Of note, sotagliflozin also reduced CV death in diabetic patients with recent worsening of HF. The definition of CV death comprises several categories that slightly differ across trials but are mainly represented by fatal acute myocardial infarction (MI), sudden death, worsening of HF or cardiogenic shock, fatal stroke and death attributable to CV events other than those listed above or because of undetermined causes. We conducted a prespecified meta-analysis (Appendix S1) of the causes of CV death in the SGLT-2i CVOTs for which these data were available. Interestingly, treatment with SGLT-2is significantly reduced overall CV death and CV death attributable to worsening of HF or cardiogenic shock (RR 0.70, 95% CI 0.52-0.94, P = .02) with moderate heterogeneity in the meta-analysis (I = 65%) (Figure 1A,B). By contrast, SGLT-2is did not reduce the risk of death as a result of MI, stroke, sudden cardiac death and death attributable to other or undetermined causes (Figure 1C-F). Accordingly, the meta-analysis by Yu et al., not including data from VERTIS-CV and DECLARE-TIMI 58 but including data from CREDENCE, showed a significant effect of SGLT-2is on death attributable to HF or cardiogenic shock, and this was largely driven by the beneficial effect of canagliflozin compared with placebo in CREDENCE (1.2 vs. 4.5 events per 1000 patient-years). Altogether, these results on CV death are consistent with the protection from hospitalization for HF observed in almost all SGLT-2i CVOTs. A major limitation of this work is represented by the lack of pre-emptive registration of this meta-analysis on any public database. Another relevant limitation of this analysis and of the interpretation of the effect of SGLT-2is on CV death is that the above CVOTs were not designed to detect differences in the single components of three-point MACE and, consequently, in CV death subgroups; indeed, the number of events for each cause appears to be exiguous. For instance, deaths attributable to HF or cardiogenic shock were only 7.6%-11.6% of the total number of CV deaths. Nevertheless, the highest proportion of CV death as a result of HF or cardiogenic shock occurred in EMPA-REG Outcome, while the lowest was in VERTIS-CV. Thus, in assessing the impact of SGLT-2is on CV death, the role of chance should not be dismissed, especially if a different number of deaths attributable to HF randomly occur in a given CVOT. Despite its limitations, our analysis suggests that the effect of SGLT-2is in reducing CV death may represent an extension of their ability to prevent hospitalization for HF. The beneficial effects of these drugs on electrolyte balance/fluxes, cardiac haemodynamics Received: 21 March 2021 Revised: 16 May 2021 Accepted: 17 May 2021