Dapagliflozin acutely improves endothelial dysfunction, reduces aortic stiffness and renal resistive index in type 2 diabetic patients: a pilot study

Sixteen patients with T2D, consecutively recruited among those attending our diabetes outpatient’s clinic between January and September 2016 received dapagliflozin. Inclusion criteria were age between 40 and 70 years, BMI < 40 kg/m², an adequate glucose control (HbA₁c < 64 mmol/mol) reached with any oral anti-hyperglycemic drug, clinic BP values < 140/90 mmHg, treatment-naïve status for high BP. Exclusion criteria were an established hypertensive status, a moderate to severe renal function impairment (estimated glomerular filtration rate—eGFR < 60 ml/min/1.73 m²) and presence of clinically relevant cardiovascular/pulmonary/haematologic/hepatic disease, insulin treatment. Given the relatively unexpected significant BP reduction after only 2 days of dapagliflozin treatment, ten T2D patients were selected after the completion of the sixteen cases on the basis of the same inclusion criteria; they followed the same protocol but receiving hydrochlorothiazide (HCT), and served as control group, to test whether the results obtained in cases could be specifically related to the mode of action of dapagliflozin or, more extensively, to its diuretic and BP-lowering effect. This was conceived as a pilot study aimed at testing the aforementioned working hypothesis and planning larger studies in the future.

The nature and purpose of the study were carefully explained to all subjects before they provided their written consent to participate. The protocol was approved by the local Ethics Committee and registered in the Registry of the Italian Drug Agency (Agenzia Italiana del Farmaco, AIFA, #772/2015).

When screened for eligibility, selected participants were advised to keep a daily Na intake of 92 mmol for the week preceding the study and until the end of the experimental protocol. Each subject was studied twice as outpatients: Visit 0, or baseline visit (day 0), and 2 days after a treatment with dapagliflozin 10 mg or HCT 12.5 mg QD (Visit 1, day 3); we chose 48 h on the basis of the pharmacokinetic properties of dapagliflozin [16]. In both visits, blood and urinary samples were collected for routine analyses, and determinations of plasma renin activity, aldosterone, norepinephrine, epinephrine and 24 h urinary electrolytes and 8-isoprostane; a complete non-invasive vascular evaluation, including flow-mediated dilation of the brachial artery (FMD), baseline and dynamic renal resistive index (RI), carotid-femoral pulse-wave velocity (PWV) and augmentation index (AIx) were also performed. All vascular tests were carried out by the same physician, with a long-lasting experience in this field (RMB).

Endothelium-dependent response was obtained by FMD, as previously described [17]. A longitudinal ultrasound scan of brachial artery was performed by using a high-resolution ultrasound machine equipped with 10 MHz linear explorer (MyLab 25, ESAOTE Florence, Italy). After 1-min baseline recording, the cuff was inflated for 5 min at 300 ± 30 mmHg around the forearm and then deflated to induce reactive hyperaemia. Endothelium-independent vasodilation was obtained by the sublingual administration of 25 μg glyceryl trinitrate (GTN). Brachial artery diameter and flow velocity were continuously monitored by computerized edge detection system (Cardiovascular Suite; Quipu srl, Italy). FMD and response to GTN were calculated as the maximal percentage increase in diameter above baseline. Hyperemic stimulus was quantified by baseline and hyperemic shear rate (SR = 8 × mean flow velocity/brachial artery diameter) and hyperemic SR area under the curve (AUC). The intra-session (1 h apart) and inter-session (30 days apart) coefficients of variation were 7.6% (0.3–10.9) and 11.6% (2.1–13.2) in a group of 20 healthy volunteers [18].

Three velocimetric measurements of the interlobar renal arteries (adjacent to the medullary pyramids) in the mesorenal area of each kidney were taken with a translumbar or anterior ultrasonographic approach. Renal RI was calculated in both kidneys according to the formula: (systolic peak velocity–end diastolic velocity)/systolic peak velocity and assessed both at baseline and 5 min after administration of GTN 25 μg. Dynamic Renal Resistive Index (DRIN) (%) was calculated as: (postGTN RI–baseline RI) × 100/baseline RI [19]. Intra-observer coefficient of variation was 2.1% for RI and 11.6% for DRIN measurements [19].

Arterial tonometry (SphygmoCor, AtCor Medical, Sidney, Australia) was performed according to the international recommendations [20]. Central BP was derived from radial pressure waveform by means of a validated transfer function and averaged on three measurements. Augmented pressure was calculated as the difference between the second and the first systolic peak, and AIx as the ratio between augmented pressure and pulse pressure (PP) and normalized at a heart rate of 75 bpm. Time to reflection was defined as the total travel time of the pulse-wave to the periphery and its return. Carotid-femoral PWV was calculated as the ratio of the surface distance between the two recording sites (direct distance * 0.8) and wave transit time. Three consecutive measurements were recorded and the median value was considered.

Plasma renin activity (PRA) and aldosterone were assayed by radioimmunoassay (DiaSorin, Saluggia, Italy). Plasma norepinephrine and epinephrine were assayed by high performance liquid chromatography.

24-h urinary 8-isoprostane concentration was measured by a specific affinity sorbent. Urine samples were centrifuged to remove sediment (10 min at 800 g); aliquots of supernatants were incubated with 8-isoprostane affinity sorbent for 60 min according to the manufacturer’s instructions (Cayman Chemical, Ann Harbor, MI, USA).

Statistical analysis was performed using NCSS 8 (NCSS, Kaysville, Utah, USA). Data were given as mean ± standard deviation when normally distributed, as median and interquartile range when non-normally distributed, and as count and percentages when discrete. Differences between groups at baseline were analysed by 2-sided paired Student’s t test for normally distributed continuous variables, Wilcoxon rank-sum test for not normally distributed continuous variables, or χ² for categorical variables, as appropriate. Differences between Visit 0 and Visit 1 in the dapagliflozin and HCT treatment arms were analysed by repeated measures ANOVA, considering treatment as between factor and time as within factor. Bonferroni post hoc comparison test was also used. A post hoc sample size calculation was performed considering FMD as outcome variable and using the data collected in this study. A sample size of 24 patients was adequate to demonstrate the detected difference in FMD with a power of 0.82 and alpha 0.05.

We demonstrate here that dapagliflozin is able to induce a rapid improvement of conduit artery function, measured as FMD. These results are likely specific for SGLT2 inhibitors, since were not replicated in the HCT arm, as reported in previous studies [21]. Interestingly, this effect occurred in the presence of unchanged brachial artery diameter, hyperemic stimulus and response to GTN, indicating a selective positive effect of dapagliflozin on the endothelium, which was previously demonstrated only in animal models and after longer duration of treatments. Chronic therapy with SGLT2 inhibitors reverses vascular dysfunction by reducing oxidative stress in aortic rings from streptozotocin-treated rats [22] and from db/db rats [23], and chronically, but not acutely, enhances coronary vasodilation to NO donors in coronary arteries of diabetic mice [24]. Our study provides the first evidence in humans that dapagliflozin improves FMD and reduces oxidative stress; if confirmed in longer, placebo-controlled trials, the former effect might play a role in the reduction of cardiovascular risk induced by SGLT2 inhibitors [25, 26], while the latter could participate in the observed acute improvement in vascular function. Indeed, hyperglycemia-induced endothelial dysfunction is prevented by acute antioxidant administration [27]. Conversely, RAAS does not seem to mediate these acute vascular effects; the small increase in PRA we have observed with both treatments is conceivably a counter-regulatory response to the increased 24 h diuresis, as already reported [7]. Noteworthy, since both sympathetic activation [28] and hyperglycemia [29] may induce endothelial dysfunction, the presence of unchanged glucose and catecholamines levels after dapagliflozin treatment suggests a direct effect of dapagliflozin on the vascular endothelium. However, a sympatho-inhibitory effect of dapagliflozin cannot be definitely excluded since, after an acute reduction in BP, a baroreflex increase in catecholamines and heart rate might have been expected. The lack of HR increase in the dapagliflozin arm, at variance with HCT, reinforces the above-reported hypothesis. Dedicated studies, performed with more sophisticated and reproducible techniques [30], are needed to elucidate this crucial issue, though some neutral results were found in type 1 diabetic patients [31] and in animals [32].

In agreement with the reported restoration of endothelium-dependent vasodilation, the results regarding arterial stiffness are also of interest. The significant reduction in aortic PWV induced by dapagliflozin occurred after only 2 days of treatment and was independent of BP reduction, since was not observed in patients treated with HCT, in which a similar BP reduction was obtained. This finding suggests that dapagliflozin may act on the functional component of arterial stiffness, which is under the control of endothelium-mediated mechanisms especially in type 2 diabetes [12]. Cherney and coauthors demonstrated an effect of 8-week treatment with SGLT2 inhibitors on aortic stiffness in type 1 diabetic patients during hyperglicemic but not during euglycemic clamp [32]; however, these results are hardly comparable with those obtained in our study, for the different study population, treatment duration and experimental conditions.

In parallel with FMD findings, if a sustained reduction in PWV will be confirmed by longer trials, this may well explain a reduction in cardiovascular events [33]; in this view, studies with longer treatment duration are currently ongoing [34] to confirm whether a long-lasting effect on atherosclerosis and large artery stiffness does exist. The reduction in oxidative stress, that we explored as secondary outcome of our study, can be regarded as a unifying hypothesis of the positive effects on the vasculature.

Two-day dapagliflozin treatment, but not HCT, induced a significant decrease in renal RI. This index has been related to renal arteriolosclerosis and proposed as an integrated index of arterial compliance, pulsatility and downstream microvascular impedance [35]. However, there is an ongoing debate about the clinical significance and the physio-pathological meaning of this simple ultrasound parameter, since some evidence suggest that renal RI is influenced by systemic hemodynamics, rather than by renal vascular resistance [36]. Furthermore, recent data from our group suggest that in morbid obesity, a condition characterized by renal hyperfiltration and increased renal plasma flow, RI is directly correlated with renal plasma flow and inversely correlated with renal vascular resistance, and is reduced after bariatric surgery proportionally to renal plasma flow, suggesting that in these conditions RI is an index of hyperperfusion rather than of arteriolosclerosis [37, 38]. The effect of 8-week treatment with SGLT2 inhibitors on renal hemodynamics (measured by classical techniques, i.e. inulin and para-aminohippurate clearance) has been so far explored only in type 1 diabetic patients [6, 39], where empagliflozin was able to significantly reduce hyperfiltration and hyperperfusion, possibly restoring tubuloglomerular feedback and increasing afferent arteriolar resistance. Thus, it is conceivable that the significant, acute RI reduction induced by dapagliflozin is explained by the reduction in renal plasma flow exerted by SGLT2 inhibition, blocking proximal tubule glucose and sodium reabsorption, which leads to increased sodium delivery to the macula densa and restoration of tubuloglomerular feedback. The fast time course of the effect and the broadly unchanged central pulse pressure exclude an effect of renal arteriosclerosis, as well as reduced systemic pulsatility, as possible mechanisms of acute RI reduction.

The assessment of renal vascular function by static or dynamic ultrasound imaging has recently emerged as an appealing target for identifying subtle vascular alterations responsible for the development of diabetic nephropathy [40]. High RI has a negative prognostic value in diabetic patients in terms of progression of renal disease [41] and it is associated with higher all-cause mortality in patients with chronic kidney disease [42]; treatment with RAS blockers, drugs able to modify the clinical course of diabetic nephropathy, reduces RI by reducing renal plasma flow through vasodilation of the efferent arteriole [43]. The correction of renal hyperperfusion, which is adjunctive and complementary to RAAS-blockade, concurs to renal protection exerted by dapagliflozin [7, 44]. In the present study DRIN, calculated as the reduction in RI rapidly induced by sublingual administration of low-dose GTN, was increased, indicating a reduced vasodilation to nitrates. DRIN might represent a marker of renal vasodilating capacity and a predictor of microalbuminuria onset in type 2 diabetic individuals [19, 45]; in the present study, dapagliflozin effects on DRIN are probably masked by the above-described effects on renal plasma flow.

A novel observation coming from our data is to characterize for the first time as dapagliflozin acutely acts as a purely “glucoretic” compound; in other words, beside glycosuria, it increases free water excretion without modifying the urinary electrolyte profile.

A common, and likely incorrect concept, is that SGLT2 inhibitors exert a beneficial natriuretic effect, even though few studies have so far measured natriuresis during SGLT2 inhibition in type 2 diabetes [7]. Recently, a transient increase in urinary sodium and chloride excretion has been reported with empagliflozin [7, 44]; we could not confirm this observation with dapagliflozin. The lack of variation in natriuresis, even after a 2-day treatment, supports the hypothesis of an increased expression and/or functional activity of other sodium transporters that might account for an immediately increased distal Na reabsorption. It is unlikely that the lack of change in natriuresis is attributable to the baseline small difference in serum osmolarity, more likely influenced by baseline difference in haematocrit.

Furthermore, acute variations in serum potassium levels were not observed in the present study. In a similar experimental setting, empagliflozin induces a modest, albeit significant increase in serum potassium levels [7, 44]. We cannot exclude that this different behaviour might be due to the different SGLT2 selectivity of the two drugs; however, this selective action on the free water pool can be regarded as a clinically advantageous effect respect to diuresis induced by classic thiazide compounds.

Another small mechanistic contribution of the present study is to show here that an increased serum magnesium level following SGLT2 inhibition, recently reported by an ample meta-analysis [45], is an immediate effect. Interestingly, magnesium in the high-normality range, is associated with a lower cardiovascular risk [46]. Furthermore, magnesium supplementation exerts favorable effects on endothelial function in individuals at risk for diabetes [47, 48], suggesting another possible mechanism through which SGLT2 inhibitors achieve vascular protection.

Some limitations of the present study should be acknowledged. First, this is a pilot study; though performed in a carefully selected population, the promising beneficial effects of dapagliflozin on endothelial function should be confirmed in larger cohorts and by randomized clinical trials. This is particularly relevant, since adjustment for known determinants of vascular function (i.e. brachial artery diameter for FMD), crucial for the interpretation of our findings [49], was not possible in the present study due to the small sample size. Second, before the described acute vascular effects can translate to an effective reduction of the cardiovascular risk, a demonstration of a long-lasting improvement in vascular function is necessary.