Sodium-glucose cotransporter 2 inhibitors reduce myocardial infarct size in preclinical animal models of myocardial ischaemia-reperfusion injury: a meta-analysis

Abstract

Aims/hypothesis: Large cardiovascular outcome trials demonstrated that the cardioprotective effects of sodium-glucose cotransporter 2 (SGLT2) inhibitors might reach beyond glucose-lowering action. In this meta-analysis, we sought to evaluate the potential infarct size-modulating effect of SGLT2 inhibitors in preclinical studies.

Methods: In this preregistered meta-analysis (PROSPERO: CRD42020189124), we included placebo-controlled, interventional studies of small and large animal models of myocardial ischaemia-reperfusion injury, testing the effect of SGLT2 inhibitor treatment on myocardial infarct size (percentage of area at risk or total area). Standardised mean differences (SMDs) were calculated and pooled using random-effects method. We evaluated heterogeneity by computing Τ2 and I2 values. Meta-regression was performed to explore prespecified subgroup differences according to experimental protocols and their contribution to heterogeneity was assessed (pseudo-R2 values).

Results: We identified ten eligible publications, reporting 16 independent controlled comparisons on a total of 224 animals. Treatment with SGLT2 inhibitor significantly reduced myocardial infarct size compared with placebo (SMD = -1.30 [95% CI -1.79, -0.81], p < 0.00001), referring to a 33% [95% CI 20%, 47%] difference. Heterogeneity was moderate (Τ2 = 0.58, I2 = 60%). SGLT2 inhibitors were only effective when administered to the intact organ system, but not to isolated hearts (p interaction <0.001, adjusted pseudo-R2 = 47%). While acute administration significantly reduced infarct size, chronic treatment was superior (p interaction <0.001, adjusted pseudo-R2 = 85%). The medications significantly reduced infarct size in both diabetic and non-diabetic animals, favouring the former (p interaction = 0.030, adjusted pseudo-R2 = 12%). Treatment was equally effective in rats and mice, as well as in a porcine model. Individual study quality scores were not related to effect estimates (p = 0.33). The overall effect estimate remained large even after adjusting for severe forms of publication bias.

Conclusions/interpretation: The glucose-lowering SGLT2 inhibitors reduce myocardial infarct size in animal models independent of diabetes. Future in vivo studies should focus on clinical translation by exploring whether SGLT2 inhibitors limit infarct size in animals with relevant comorbidities, on top of loading doses of antiplatelet agents. Mechanistic studies should elucidate the potential relationship between the infarct size-lowering effect of SGLT2 inhibitors and the intact organ system.

Keywords: Cardioprotection; Infarct size; Ischaemia–reperfusion injury; Meta-analysis; Sodium–glucose cotransporter 2 inhibitor; Systematic review.

Figures

Fig. 1

Flow chart of the study identification and selection process. A systematic review yielded 316 unique records as of 16 June 2020. After application of inclusion and exclusion criteria, a total of ten eligible studies were identified reporting 16 individual comparisons, which were included in the meta-analysis

Fig. 2

Exploration of within-study bias. The prespecified modified form of the CAMARADES validated checklist was used to assess study quality. The list consists of the depicted ten points, which were evaluated for each individual comparison

Fig. 3

Forest plot of the size of effect of SGLT2 inhibitor treatment on myocardial infarct size vs placebo. Myocardial infarct size (% area at risk or total area) is quantified as SMD (black circles) and 95% CIs. The pooled effect estimate is shown as a diamond and 95% CIs are depicted. The dashed line represents the total pooled estimate and the shading, bounded by dotted lines, depicts its 95% CI. RE, random-effects; SGLT2i, SGLT2 inhibitor

Fig. 4

Impact of experimental factors on the infarct size-lowering effect of SGLT2 inhibitors. SMDs according to prespecified subgroups are depicted. Significance of interactions is shown according to univariate meta-regression, without correction for multiple comparisons. Residual heterogeneities are reported according to each experimental variable. The dashed line represents the total pooled estimate and the shading, bounded by dotted lines, depicts its 95% CI, corresponding to data shown in the forest plot (Fig. 3). In line with the prespecified protocol, we only performed meta-analysis on at least three independent comparisons. Hence, dapagliflozin could not be included in the comparison regarding the efficacy of SGLT2 inhibitor subtypes. Furthermore, only one study tested SGLT2 inhibitor in swine, therefore it is excluded from the species subgroups. RE, random-effects

Fig. 5

Funnel plot depicting SMDs plotted against their measure of precision (SE). The vertical line represents the total pooled estimate corresponding to that on the forest plot (Fig. 3). Its pseudo-95% CIs are depicted with dashed lines. Note that two points appear to be optically fused because of similar SMDs (−1.01 and −1.03) and the same SE (0.67)

Fig. 6

Exploration and adjustment for publication bias. (a, b) Trim-and-fill analysis showing the modified forest plot and funnel plot with values according to the theoretically missing studies (process called ‘filling’; white circles). Note that on the funnel plot, two points appear to be optically fused because of similar SMDs (−1.01 and −1.03) and a same SE (0.67). (c) Summary of methods used to explore and adjust for publication bias. A priori weight functions were applied to simulate moderate and severe one-tailed selection (i.e. pattern of selection that tends to favour the publication of studies reporting significant positive effects). The adjusted summary estimate from the trim-and-fill analysis is also depicted. The unadjusted summary estimate is shown as a dashed line and the shading, bounded by dotted lines, depicts its 95% CI, corresponding to data shown in the forest plot (Fig. 3). RE, random-effects