Effect of sodium-glucose cotransporter 2 (SGLT2) inhibitors on left ventricular remodelling and longitudinal strain: a prospective observational study

Background Sodium-glucose cotransporter 2 inhibitors (SGLT2i) lower cardiovascular events in type 2 diabetes mellitus (T2DM) patients, although the mechanisms underlying these benefits are not clearly understood. Our aim was to study the effects of SGLT2i on left ventricular remodelling and longitudinal strain. Methods Between November 2019 and April 2020, we included 52 patients with T2DM ≥ 18 years old, with HbA1c between 6.5 and 10.0%, and estimated glomerular filtration ≥ 45 ml/min/1.73 m2. Patients were classified into SGLT2i group and control group, according to prescribed treatment by their referring physician. Conventional and speckle tracking echocardiography were performed by blinded sonographers, at baseline and after 6 months of treatment. Results Among the 52 included patients (44% females, mean age 66.8 ± 8.6 years, mean HbA1c was 7.40 ± 0.7%), 30 patients were prescribed SGLT2i and 22 patients were classified as control group. Mean change in indexed left ventricular mass (LVM) was − 0.85 ± 3.31 g/m2 (p = 0.003) in the SGLT2i group, and + 2.34 ± 4.13 g/m2 (p = 0.58) in the control group. Absolute value of Global Longitudinal Strain (GLS) increased by a mean of 1.29 ± 0.47 (p = 0.011) in the SGLT2i group, and 0.40 ± 0.62 (p = 0.34) in the control group. We did not find correlations between changes in LVM and GLS, and other variables like change in HbA1c. Conclusions Among patients with T2DM, SGLT2i were associated with a significant reduction in indexed LVM and a significant increment in longitudinal strain measured by speckle tracking echocardiography, which may explain in part the clinical benefits found in clinical trials.


Background
Sodium-glucose cotransporter 2 (SGLT2) inhibitors are a recent and fast growing class of oral anti-hyperglycaemic agents available to treat patients with type 2 diabetes (T2DM) [1]. They function through a novel mechanism by reducing renal tubular glucose reabsorption, and produce a reduction in blood glucose without stimulating insulin release. When compared with other oral antihyperglycaemic agents, SGLT2 inhibitors have demonstrated non-inferiority along with additional metabolic benefits of weight loss and blood pressure lowering [2]. In addition, SGLT2 inhibitors have been shown to improve cardiovascular outcomes in type 2 diabetes mellitus (T2DM) trials [3][4][5][6]. The mechanisms underlying the clinical cardiovascular beneficial effects, especially on heart failure, are not fully understood and have been the subject of various studies and publications [7,8].
Reverse ventricular remodelling refers to a "more-normal" chamber geometry restoration [9]. Several pharmacological treatments such as angiotensin-converting enzyme (ACE) inhibitors [10], beta-blockers [11,12] and mineralcorticoid receptor antagonists [13], have been shown to promote reverse ventricular remodelling, with reductions in left ventricular mass (LVM) and volume and improved left ventricular systolic function. These changes are consistently associated with reductions in morbidity and mortality. As a result, some authors advocate that reverse remodelling can serve as a valid surrogate endpoint for clinical outcomes in studies of new therapies [14].
Furthermore, Global Longitudinal Strain (GLS) determined by Speckle Tracking technique is a surrogate of left ventricular systolic function [15]. Clinical studies of the effects of SGLT2 inhibitors on myocardial deformation parameters are scarce [16,17]. Although LV longitudinal strain was previously measured by cardiac magnetic resonance [18], to our knowledge, there are no studies estimating GLS by speckle tracking echocardiography in patients treated with SGLT2 inhibitors.
We hypothesised that the SGLT2 inhibitors effects on left ventricular remodelling may play a role in the underlying mechanisms through which SGLT2 inhibitors reduce the risk of heart failure in people with diabetes. Our aim was to study the effects of SGLT2 inhibitors on left ventricular remodelling and function in patients with T2DM.

Study design and participants
This was a prospective observational study conducted in a single centre in Jerez de la Frontera (Spain). Patients were recruited from the endocrinology outpatient department. Fifty-two consecutive diabetic patients were included prospectively. The inclusion criteria were: patients with T2DM with at least 18 years of age attending our clinic between November 2019 and April 2020; glycated haemoglobin levels between 6.5 and 10.0%; 3) Clinical stability. The exclusion criteria were: history of type 1 diabetes mellitus; current SGLT2 inhibitor or glucagon-like peptide receptor agonist use; an estimated glomerular filtration rate < 45 ml/min/1.73 m 2 ; acute coronary syndrome the last 2 months; previous cardiac surgery; pregnant women; New York Heart Association IV symptoms of heart failure; greater than moderate valvular disease; or suboptimal echocardiographic acoustic window.
Clinical decisions on medical management were made by the referring physician based on clinical data and comorbidities at baseline visit, according to current recommendations [19].

Data collection and follow-up
Clinical, anthropometric, analytical and echocardiographic assessments were performed at baseline and after 6 month follow-up. Arterial blood pressure was also estimated during the initial visit. According to the prescribed treatment at this point, patients were classified into the SGLT2 inhibitors group or the control group. The same sonographers, who were blinded to clinical data, baseline echocardiographic data and prescribed treatment, performed both echocardiographic examinations.

Variables
The primary outcome endpoint was the change in ventricular remodelling and function between initial and follow-up echocardiographic assessment.

Standard echocardiographic examination
Two-dimensional trans-thoracic echocardiographic and Doppler studies were obtained with clinical ultrasound machines equipped with 2.5-3.5 MHz transducers (iE33 Phillips Medical Systems, The Best, The Netherlands). All tests were conducted by two experienced sonographers, who were blinded to the clinical data and prescribed treatment. Baseline echocardiographic examination was performed during the first 7 days after inclusion in the study.
Left ventricular chamber dimensions and wall thicknesses were measured, and left ventricular mass was calculated according to the American Society of Echocardiography guidelines [20]. Left ventricular hypertrophy (LVH) was defined as indexed left ventricular mass of 95 g/m 2 or greater for women and 115 g/m 2 or greater for men [20]. The relative wall thickness (RWT) was calculated as the ratio of posterior wall thickness/left ventricular diastolic radius, independently of the presence of LVH. A ratio of 0.42 or greater indicated concentric left ventricular geometry. End-diastolic and end-systolic left ventricular volumes were estimated and left ventricular ejection fraction (EF) was assessed by the modified Simpson's Biplane Method. To assess diastolic function, the following mitral Doppler pulse and tissue Doppler variables were measured: early (E) and late (A) diastolic filling velocity, E/A ratio, septal (septal e') and lateral (lateral e') early mitral annular tissue velocity. We also calculated the E/e' ratio.

Strain analyses
Myocardial strain was measured using Speckle Tracking echocardiography. To assess LV, longitudinal strain the endocardial and epicardial borders were traced in the apical two-, three-and four-chamber echocardiographic views on an end-diastolic frame. The software then automatically divided the myocardium into 17 segments. Peak systolic strain was estimated for each segment, and then GLS was calculated by averaging the 17 segments values. All images were stored electronically and LV strain was analyzed off-line with 2D Speckle Tracking software (QLab 10).

Statistical analysis
Data were expressed as mean ± standard deviation for continuous variables, and were compared by using the unpaired t-test. Categorical variables were expressed as percentages and were compared using chi-square analysis or the Fisher exact test. Comparison of variables between baseline and 6 months after treatment was made using the paired test or Wilcoxon signed-rank test. Baseline and 6-month variables were also assessed by ANOVA for repeated-measures with time and group-time interaction effects. Comparisons between changes in indexed LVM, GLS and other continuous variables were calculated by Pearson correlation.
Analyses followed an intention-to-treat approach, where all patients were included in their corresponding group according to the initial prescribed treatment.
Differences were considered significant at p values < 0.05. For data analysis, the statistical program SPSS version 20.0 (SPSS Inc., Chicago, Illinois) was used.

Baseline characteristics
A total of 52 patients (29 males and 23 females) were included in the study after exclusion of 2 patients because of suboptimal acoustic window. Mean age of the patients was 66.8 ± 8.6 years, mean duration of diabetes was 104 ± 101 months, mean glycated haemoglobin was 7.40 ± 0.7%. Of the participants, 65% had arterial hypertension, 13% were current smokers, 54% had dyslipidaemia, and 4 patients (8%) had coronary artery disease. At baseline, 29% were on DPP-4 inhibitors, 27% on insulin, 79% on metformin, 75% on RAAS inhibitors (angiotensin-converting enzyme inhibitors or angiotensin receptor blockers) and 9% on beta-blockers.
Of the total sample, 30 patients were prescribed SGLT2 inhibitors (67% empagliflozin, 17% dapagliflozin, 10% canagliflozin, 7% ertugliflozin), whilst the remaining 22 patients were included in the "control" group. Basal clinical characteristics of both groups are summarized in Table 1, and echocardiographic characteristics in Table 2. Patients prescribed SGLT2 inhibitors had significantly higher glycated haemoglobin and worse GLS; however, we did not find any other differences in basal characteristics between both groups.

Outcome
At 6-month visit, 3 patients in the SGLT2 group had stopped this treatment during the follow-up (one patient one month after the first visit and two patients 5 months after the baseline examination) because of minor side effects. Three patients in control group initiated treatment with SGLT2 inhibitors, as indicated by their referring physician during follow-up (one after  1 month, one after 3 months and one other 5 months after the initial assessment). Mean change in the indexed LVM from baseline to the 6-month visit was − 10.85 ± 3.31 g/m 2 (p = 0.003) in the SGLT2 group, and + 2.34 ± 4.13 g/m 2 (p = 0.58) in the control group (Fig. 2). Absolute value of GLS increased by a mean of 1.29 ± 0.47 (p = 0.011) from baseline to the 6-month examination in the SGLT2 group, and 0.40 ± 0.62 (p = 0.34) in the control group (Fig. 3). Tables 3, 4 and 5 summarize the changes from baseline to 6-month visit.
Ventricular remodelling classification did not change significantly in the control group after 6 months of follow-up. Nevertheless, 7 patients in the SGLT2 group changed from concentric hypertrophy to concentric remodelling due to a significant reduction in the indexed left ventricular mass (Figs. 4,5), indicating that the concentric hypertrophy decreased during the follow-up (from 33 to 10%, p = 0.006).
We failed to find any correlations between changes in LVM, GLS, septal e' and other variables (Table 6).

Discussion
The main findings of this study were that the addition of SGLT2 inhibitors to standard anti-hyperglycaemic treatment in people with T2DM was associated with: (1) a significant decrease in indexed LVM; (2) an improvement in left ventricular GLS assessed by speckle tracking echocardiography.
Although SGLT2 inhibitors have demonstrated a reduction in heart failure outcomes [3][4][5][6], even in non-diabetic patients [22] mechanisms to explain the cardiovascular benefits of these drugs are not clearly understood. Our data support the theory that the benefits of SGLT2 inhibitors are, at least in part, mediated via a mechanism independent of its glucose-lowering activity.
LV hypertrophy is a strong determinant of cardiovascular outcomes and mortality in the general population  [23] and also in people with T2DM [24]. Several studies showed previously significant reductions in LVM in mice with [25] and without T2DM [26,27]. In clinical research, two small-sized-sample studies found that empagliflozin [28] and canagliflozin [29] reduced LVM, although these studies were not controlled by placebo. Verma et al. [30] showed that mean LVM regression assessed by cardiac magnetic resonance after 6 months of treatment with empagliflozin in patients with coronary artery disease was 2.6 g/m 2 . Similarly, treatment with dapagliflozin reduced LVM measured by cardiac resonance [31]. In our study, we found a higher reduction in LV hypertrophy, probably due to different inclusion criteria, higher baseline LVM and overestimation of LV hypertrophy by echocardiography [32].  LVM reduction supposed a change in ventricular remodelling classification in SGLT2 inhibitors patients that could have an impact on cardiovascular outcomes [33].
One of the strengths of our study was that, to our knowledge, this is the first clinical study to show an improvement in absolute value of GLS estimated by speckle tracking echocardiography in patients treated with SGLT2 inhibitors. Speckle tracking echocardiography is a relatively new method used to measure systolic myocardial function, with higher prognostic value than LV ejection fraction [15].
Several studies observed that diabetic patients have lower absolute GLS values despite normal LV ejection fraction [34,35]. Other authors suggested that GLS by speckle tracking echocardiography might detect changes in systolic function earlier than conventional methods [36], which could explain why other studies did not find differences in LV ejection fraction in T2DM patients [28][29][30].
Garcia-Ropero et al. [37] found that empagliflozin improved myocardial deformation estimated by speckle tracking echocardiography in an ischemic non-diabetic porcine model. However, clinical studies of the effects of SGLT2 inhibitors on myocardial deformation parameters are lacking. Tanaka et al. [38] showed a GLS enhancement in patients treated with dapagliflozin, with similar results to our study, although this study was not a placebo-controlled one.
The mechanisms of the beneficial effects of SGLT2 inhibitors on cardiac remodelling and function are not completelyunderstood. Improved glycaemic control and hypotensive effects seem unlikely, given that their benefits would have taken years. Other hypotheses like intravascular volume reduction, inhibition in the Na/H exchanger, and tissue oxygenation improvement via  increased haematocrit have been proposed [39]. It has also been postulated that SGLT2 inhibitors may increase myocardial energy supply and metabolic efficiency, thereby, improving myocardial performance. Santos-Gallego et al. [40] showed that empagliflozin switched myocardial fuel utilization away from glucose towards other molecules like ketone bodies, free fatty acids and branched-chain amino acid that improved myocardial energetics.
Although other authors demonstrated that SGLT2 inhibitors improve diastolic function in T2DM patients [29,41,42] we achieved a significant improvement only on septal e' values (Table 4), probably due to our reduced sample size also without signification when using ANOVA analysis (Table 5).
In our opinion, our main limitation was the nonrandomized design of our study that hampered the establishment of a cause-effect relationship between SGLT2 inhibitors and positive effects on LV mass and function. However, there is a biologic plausibility for a relation between our results and the positive clinical impact of SGLT2 inhibition on patients with heart failure and reduced ejection fraction both with and without T2DM [43]. Other limitations of our study were the short duration of the follow-up and the reduced number of patients. These limitations made it difficult to obtain statistically significant differences in other variables. In addition, unfortunately, the small sample size hampered the comparisons between different iSGLT2. However, despite the limited number of patients and relatively short follow-up, it seems that there are large differences in significant variables between the groups taking into account SGLT inhibitors as a whole. Finally, although cardiac magnetic resonance is the gold standard for cardiac chambers volume and mass assessment, we preferred the use of echocardiography due to a more widespread use. On the other hand, one strength of the study was that it evaluated the effects of SGLT2 inhibitors on LV mass and function in real-world settings. It included a patient population that may be more  representative of the non-selective population normally used in randomised controlled trials and provided evidence that the treatment may exert positive effects in the every day practice.

Conclusions
The present study showed that T2DM patients treated with SGLT2 inhibitors displayed positive effects on left ventricular remodelling due to a reduction in LVM, and LV longitudinal function assessed by speckle tracking echocardiography. These findings could explain the beneficial effects on cardiovascular outcomes seen in clinical trials.