SGLT2 inhibitors: a therapy for everybody but not for anything?

This editorial refers to ‘SGLT2 inhibitors for the prevention of atrial fibrillation: a systemic review and meta-analysis’, by H.-D. Zhang et al. https://doi.org/10.1093/eurjpc/zwad356. Sodium-glucose co-transporter 2 inhibitors (SGLT2-i) were originally introduced to treat type 2 diabetes mellitus (DM); by inhibiting the SGLT2 cotransporter, mainly located on the luminal membrane of the cells of the renal proximal tubule, SGLT2-i increase glycosuria and decrease glycaemia. In addition, SGLT2-i promote weight loss, reduce blood pressure, and improve cardiovascular (CV) outcomes, especially in patients with heart failure (HF).1 Despite the mechanisms leading to a reduction of HF hospitalizations and CV death are still debated,2 the impact of this class of drugs in the management of patients with HF is impressive: the number of patients needed to be treated (NNT) to prevent one CV death or HF hospitalization in the EMPEROR-Reduced and DAPA-HF trial was 19 and 21, respectively.3 The relevance of these numbers is magnified if we consider the NNT to prevent major adverse cardiovascular events (MACE) with other drugs widely prescribed in daily clinical practice or procedures considered to be life-saving. For instance, the NNT to prevent a MACE with aspirin in patients with known coronary artery disease or stroke is ∼50,4 while the NNT to prevent one in-hospital death with primary percutaneous coronary intervention in patients presenting with ST-elevation myocardial infarction is 24.5 Being atrial fibrillation (AF) frequently present and a potent predictor of adverse outcomes in HF patients, independently from functional class or left ventricular ejection fraction,6 studies examining the potential impact of SGLT2-i on the prevention of AF have been conducted. To date, no clear association between the use of SGLT2-i and AF prevention has been established. A pooled analysis of 31 randomized trials including more than 75 000 patients found that SGLT2-i are associated with a reduced incidence and recurrence of AF as well as of CV events.7 Conversely, similar findings could not be retraced in a previous meta-analysis based on reported trial data until October 2019,8 but were confirmed by a post hoc analysis of the DECLARE-TIMI 58 study,9 which demonstrated a reduction in the occurrence of atrial arrhythmias by 19% with dapagliflozin compared with placebo in 17 160 patients with DM. Consistent findings were observed in a recent analysis of the large Food and Drug Administration adverse event reporting system, which revealed a lower incidence of AF in diabetic patients treated with SGLT2-i if compared with other glucose-lowering drugs, suggesting a protective role against arrhythmic occurrence.10 In the current issue of the Journal, Zang et al.11 reported the results of an updated systematic review and meta-analysis evaluating such relation in patients with DM, HF, or chronic kidney disease. The authors included 46 randomized, placebo-controlled trials enrolling more than 100 000 patients, but, despite the large sample size, the study failed to show a reduction in AF occurrence (relative risk and 95% confidence intervals: 0.90, 0.80–1.01). Consistent results were found in multiple sensitivity and subgroup analyses, therefore yielding an overall negative result, regardless of SGLT2-i type and dosage. From a mechanistic perspective, these might be seen as unexpected findings, as the haemodynamic effects induced by SGLT2-i, such as the reduction in HF and in turn of atrial stretch, as well as metabolic modulations and indirect effects on cardiomyocytes, could potentially reduce the risk of AF. However, difficulties in providing a definitive result on such an interesting topic might be related to some methodologic flaws of the present metanalysis that limit the accuracy of the findings, as well as to the research subject itself. First, the meta-analysis included studies characterized by important differences regarding study design, endpoint definition, and length of follow-up, and, most importantly, none of them analysed AF occurrence as the primary endpoint. In addition, none of the 46 trials included in the meta-analysis excluded patients with history of AF and the authors were not able to clarify whether AF detected during follow-up was a recurrence of a pre-existing arrhythmia. Whether SGLT2-i could be more effective in preventing new-onset AF or its recurrence remains unknown, as this study does not provide any further insights on this issue. Eventually, despite a higher sample size than previous metanalysis, the event rate might still be too small to detect a significant difference. As a matter of fact, a relevant aspect arising from these studies is the low overall rate of events, ranging approximately from 1% to 3% of the study population.7–9 This consideration is of paramount importance: despite enrolling more than 100 000 patients, the current metanalysis fails to demonstrate a significant risk reduction of AF occurrence, but, even in case a benefit had been observed, would such reduction be clinically relevant? A reduced risk of AF occurrence provided by SGLT2-i has not been reported in other trials than the DECLARE-TIMI 589 and, as the absolute reduction of the risk would be minimal, the clinical importance of this finding is probably negligible. Indeed, most of the included studies reported AF occurrence in <1% of patients, therefore the computed 10% relative risk reduction in such outcome represents a small treatment effect in a surprisingly rare condition among the high-risk patients enrolled in the trials evaluating SGLT2-i. Despite being common practice to analyse risk ratios and reduction in relative risk, such approach might not properly convey the real benefits of a specific treatment, as the number of patients who need to receive the intervention could be enormous. To help with data interpretation and favour the transmission of research findings to daily clinical practice, presenting easily interpretable measures, such as the NNT, could clarify the real impact of an intervention and its relevance in a real word setting: as an example, based on the present analysis, the NNT to avoid one event is >600 patients. As a result, the study published in this issue of the Journal does not help in clarifying the real role of SGLT2-i in the prevention of AF occurrence, despite a trend towards a decrease in relative risk, but, more importantly, questions the need for further research on the subject, which is probably the greatest strength of the study. Of course, considering the still ongoing re-evaluation of the molecular mechanisms behind the CV benefit of SGLT2-i, an antiarrhythmic effect might be identified. Once, and whether, such effect will be defined, only high-quality research, as randomized clinical trials in high-risk population, could elucidate whether a role of SGLT2-i in modulating atrial arrhythmias exists; however, based on current data, clinical relevance of a study evaluating this aspect would still be limited and the rationale for those trials could be missing. No new data were generated or analysed in support of this research.