Effects of the SGLT2 inhibitor dapagliflozin on cardiac function evaluated by impedance cardiography in patients with type 2 diabetes. Secondary analysis of a randomized placebo-controlled trial

Patients with type 2 diabetes (T2D) have a two-to-fivefold increased risk of developing heart failure (HF) compare to non-diabetic patients [1]. This risk remains high despite control of known major cardiovascular risk factors [2]. HF is one of the first manifestations of T2D-related cardiovascular disease [3] and is one of the leading causes of hospitalizations in this population [4].

The presence of T2D negatively impacts HF outcomes. Patients with T2D are more likely to be hospitalized for HF [5], to be re-hospitalized for HF [6], and to experience a longer hospital stay [7] compared to patients without diabetes. Moreover, patients with T2D and HF have a higher risk of cardiovascular and all-cause mortality than HF patients without T2D [5].

Over the past decade, in patients with T2D, ischemic heart disease, and stroke mortality has declined, especially in men, but heart failure deaths did not change [8]. Therefore, therapeutic strategies able to reduce the HF risk are particularly appealing.

Sodium glucose co-transporter 2 inhibitors (SGLT2i) induce glycosuria, thereby reducing plasma glucose, blood pressure, and body weight [9]. The recent results of the four cardiovascular outcome trials (CVOTs) with SGLT2i have shown remarkable benefits in reducing HF hospitalization by about 30% in patients with T2D with or without established cardiovascular disease and also in patient with or without history of heart failure [10‐13]. In addition, large retrospective studies confirmed that treatment with SGLT2i, when compared with other glucose-lowering medications (GLMs), was associated with lower rates of hospitalization for HF among patients with and without history of cardiovascular disease [14‐16].

Several possible pathophysiological mechanisms for the protection exerted by SGLT2i on HF risk have been proposed [17], but not yet fully elucidated and mostly not confirmed in human studies.

In this study, we aimed to evaluate whether treatment with SGLT2i affected hemodynamic variables and cardiac function using impedance cardiography (ICG) in a randomized placebo-controlled trial.

ICG has been proposed as a non-invasive method of hemodynamic monitoring especially in critically ill and surgical patients [18]. ICG measures continuously the resistance (impedance) to the transmission of a low intensity electrical current through the chest. The changes in the resistance over time are proportional to the dynamic changes during each cardiac cycle. This allow the calculation of stroke volume (SV), cardiac output (CO) and other hemodynamic variables [19].

Thirty-three patients were enrolled in the study. Seventeen were assigned to dapagliflozin 10 mg, 16 to matching placebo. Two patients in the dapagliflozin group were excluded because they withdraw consent or were lost to follow-up, one patient in the placebo group was excluded because ICG data were missing, leaving 30 completers for the analysis, 15 in each group (Fig. 1).

The baseline characteristics of the study population are summarized in Table 1. Patients were on average 63.7-year-old, with a known diabetes duration of 14.1 years and a baseline HbA1c of 8.2% (66 mmol/mol). The vast majority of patients (93.5%) were on metformin, 32.2% were on other oral GLMs, and 51.6% were concomitantly treated with insulin. 63.3% of patients had established cardiovascular disease (symptomatic or asymptomatic) and 35.4% had microangiopathy. No patient included in the study had a history of HF or a reduction of left ventricular ejection fraction. Almost half of the patients had LV hypertrophy (46.7%) and/or diastolic LV dysfunction (43.3%). After 12 weeks, dapagliflozin significantly reduced HbA1c by 0.9% and body weight by 3.1 kg. In the placebo group, HbA1c increased non-significantly by 0.4% and body weight remained stable (+ 0.2 kg).

Compared with baseline, SBP declined by 4.7 ± 1.3 mmHg in the dapagliflozin group and by 1.0 ± 2.3 mmHg in the placebo group (p = 0.035). DBP declined by 1.3 ± 0.6 mmHg in the dapagliflozin group and by 0.4 ± 1.6 mmHg in the placebo group (p = 0.317) (Table 2).

At baseline, there was no significant difference in ICG parameters between the two groups. In both groups, comparing data recorded at end of treatment with dapagliflozin or placebo to those recorded at baseline, we detected no significant change in parameters of blood flow (stroke volume, stroke index, cardiac output, cardiac index), systolic function (ejection fraction, acceleration and velocity indexes, systolic time ratio, left ventricular ejection time), circulatory function (systemic vascular resistance and systemic vascular resistance index, total arterial compliance and total arterial compliance index), and fluid status (thoracic fluid content and thoracic fluid content index) (Table 2). ICG-derived STR has been recently validated as a marker of LV diastolic dysfunction defined using echocardiography, with an accuracy of 97% [30]. At baseline, diastolic dysfunction based on the STR cut-off of 0.31 was present in 63.3% of patients (66.7% in the dapagliflozin group and 60% in the placebo group). At follow-up, the prevalence of diastolic dysfunction based on STR declined by 6.7% in the dapagliflozin group and remained stable in the placebo group (p = 0.325).

No significant correlation was detected between change in HbA1c or body weight from baseline to the end of follow-up and change in ICG parameters (not shown). In the whole study cohort, TFC correlated directly with total body water evaluated by bioelectrical impedance analysis (BIA) (r = 0.49; p < 0.001), but the change in TFC was not correlated to the change in total body water (r = 0.14; p = 0.456).

In this study, we found no significant effect of the SGLT2i dapagliflozin on measures of cardiac function evaluated by ICG. Since publication of the groundbreaking results of the EMPA-REG Outcome trial, the scientific community is struggling to understand the biological basis and the physiological mechanisms responsible for the cardiovascular benefits of SGLT2i, especially on HF.

Several possible pathophysiological mechanisms for the cardiovascular protection exerted by SGLT2i have been suggested, yet not formally proved. Briefly, these include reduction in preload, secondary to natriuresis and osmotic diuresis [31] and in afterload, secondary to reduction in blood pressure [32], myocardial Na⁺/H⁺ exchange inhibition [33, 34], reduction of necrosis and cardiac fibrosis [35, 36] and consequently of the development of HF. Reduction of inflammation and oxidative stress [37], restoration of the balance between pro- and anti-inflammatory adipokines and cytokine [38], reduction of epicardial adipose tissue mass [39] have been proposed as additional mechanisms. Moreover, the use of SGLT2i through an increased metabolism of free fatty acid and an increased production of ketone bodies may provide a more efficient source of energy for the myocardium [40].

Other possible mechanisms for the pleiotropic effects of SGLT2i on cardiovascular benefit involving the reduction of serum uric acid levels [41], haemoconcentration [42], improving endothelial function and aortic stiffness [43, 44] and may induce vasodilatation through activation of protein kinase G and the voltage-dependent K⁺ channel [45].

Some of these hypotheses result from studies in experimental models [33‐37, 45] and should be confirmed in clinical studies, while other ones are just speculations [40] and require further research.

Despite all these mechanisms could contribute to protection against HF, information with respect to the direct effects of SGLT2i on myocardial function in humans is very limited. A small observational study showed that in patients with T2D and established cardiovascular disease, a short-term empagliflozin treatment was associated with a significant reduction in LV mass index and improved diastolic function measured by 2D echocardiography. No differences were found in LV systolic function, LV end diastolic volume, and LV end systolic volume [46]. Another observational study conducted in 37 patients with T2D with or without cardiovascular disease, showed that a treatment with canagliflozin improved LV diastolic function (E/e′ ratio) [47]. Similar results were found in a prospective observational study involving T2D with stable HF treated for 6 months with dapagliflozin [48]. However, these observational findings need to be confirmed in randomized controlled trials (RCTs). RCTs are currently being conducted to investigate the effects of SGLT2i on LV remodelling in patients with T2D and HF [49, 50]. Furthermore, large randomized-controlled trials investigating SGLT2i as a treatment for HF are ongoing and will enroll HF patients either with or without T2D, with reduced (DEFINE-HF, ClinicalTrial.gov registration no. NCT02653482) or preserved EF (PRESERVED-HF, ClinicalTrial.gov registration no. NCT03030235).

To the best of our knowledge, this is the first study exploring the cardiac effects of SGLT2i using ICG in T2D. This study suggests that, in T2D patients without a history of HF, a 12-week treatment with dapagliflozin exerted no overt effect on haemodynamic and cardiac function parameters compared to placebo. Power calculated a posteriori revealed that we could rule out an effect of dapagliflozin greater than 80 ms for LVET, 18 mL for SV (measures of systolic function) and 0.07 for STR.

Our data do not deny the strong protection provided by SGLT2i against the risk of HF, as demonstrated in solid and large CVOTs [10‐13]. Rather, these negative findings may be attributed to several reasons that are at least in part related to study limitations. First, the observation time may be too short to detect changes in ICG parameters. However, in the 4 CVOTs with SGLT2i, the onset of the benefit in terms of reduction in HF hospitalization was very rapid. The curves started to separate in the first weeks after randomization, suggesting that diuretic or haemodynamic actions of SGLT2i could be at least in part responsible for the observed benefit.

Second, sample size may be too small to detect differences in ICG measures between the two groups of treatment. In fact, our study was designed to identify a significant change in the primary end-point, the effects of dapagliflozin compared to placebo on cholesterol efflux capacity [20] and power calculation was not performed based on ICG measurement variability. Thus, this exploratory re-analysis may be underpowered to detect significant changes in ICG parameters. Yet, in the same study, the analysis of body composition based on BIA clearly showed different effects of dapagliflozin versus placebo [20]. Therefore, we expected that the same sample size was sufficient to reveal overt effects of dapagliflozin on ICG, which is also based on impedance analysis. We detected a direct correlation between total body water and TFC, suggesting concordance of the two methods. Interestingly, while dapagliflozin reduced total body water by about 2.5 L [20], we herein found no change in TFC, suggesting that volume reduction occurred mainly in extra-thoracic compartments.

Furthermore, we recorded no parameter directly related to diastolic function. Although generally considered a measure of contractility, STR (the ratio between electric and mechanical systole) has a high accuracy for the diagnosis of diastolic dysfunction [30]. We found no effects of dapagliflozin on STR, either continuous or categorized according to the recommended cut-off [30], but we acknowledge that other parameters of diastolic function should be evaluated. Previous observational studies documented effects of SGLT2i on diastolic cardiac function in T2D, but a recent re-analysis of the DECLARE-TIMI58 study reported that dapagliflozin prevented HF in patients with a normal or reduced EF and improved survival especially in patients with reduced EF [51]. Thus, a selective effect of SGLT2i on diastolic function is unlikely to explain the observed benefit on HF rates.

Finally, the use of ICG for evaluation of cardiac variables has provided conflicting results in terms of reliability: studies comparing ICG with other methods to measure cardiac and hemodynamic parameters showed an agreement ranging from good [52‐54] to poor [55, 56]. In addition, it should be mentioned that ICG is based on thoracic impedance triggering on cardiac cycle, from which all other measures are derived. Since SGLT2i typically reduce body fluid [57], it is also possible that the resulting change in impedance, that we have already demonstrated with BIA [20], masked the effects on cardiac function and hemodynamic parameters.

Despite strong evidence that SGLT2i prevent HF in T2D with or without prior HF episodes, we found no effects of dapagliflozin on parameters of cardiac function measured by ICG. Short duration of treatment, small sample size, lack or specific diastolic function measures, limited correlation of ICG parameters with gold standard examination of cardiac function, and the masking diuretic effects of SGLT2i are all possible explanations of these negative results. In the end, we also speculate that SGLT2i exert no effects on cardiac function parameters in patients with normal cardiac function but may play a major role at the very early onset of cardiac disfunction, preventing hospitalization and mortality. Further studies are needed to address this hypothesis.