β-Cell Function, Incretin Effect, and Glucose Kinetics in Response to a Mixed Meal in Patients With Type 2 Diabetes Treated With Dapagliflozin Plus Saxagliptin Free

Type 2 diabetes (T2D) is a chronic disease driven by several pathophysiological mechanisms (1), requiring multifactorial behavioral and pharmacological treatments. The American Diabetes Association/European Association for the Study of Diabetes consensus report recommends considering early combination therapy to allow simultaneous targeting of multiple pathophysiological processes to increase the durability of glycemic control (2). For instance, dipeptidyl peptidase 4 (DPP-4) inhibitors increase the availability of endogenously secreted glucagon-like peptide 1 (GLP-1), eliciting glucose-dependent insulin secretion and suppression of glucagon release (3,4), whereas sodium–glucose cotransporter 2 (SGLT2) inhibitors lower plasma glucose concentrations independently of insulin secretion or action by increasing urinary glucose excretion (5). Their glucose-lowering capacity, however, is possibly counteracted by a paradoxical increase in endogenous glucose production (EGP) (5–7) partly attributable to changes in the glucagon-to-insulin ratio (5). Because DPP-4 inhibitors reduce glucagon levels, their combination with an SGLT2 inhibitor could result in a complementary mode of action and sustained glucose-lowering efficacy (8–10).

The mechanisms triggered by the DPP-4/SGLT2 inhibitor combination have not been fully investigated. Recently, Qin et al. (11) showed that adding saxagliptin to dapagliflozin did not prevent the paradoxical elevation of fasting EGP triggered by dapagliflozin alone, with no difference in the insulin-to-glucagon (I/G) ratio. Therefore, other mechanisms may account for the complementary mechanisms of action of this combination, such as an effect on glucose metabolic clearance rate, a modulation of the rate of glucose absorption after meal ingestion, or a potentiation, either direct or indirect through relief of glucose toxicity, of β-cell function and/or GLP-1 secretion.

To enhance our understanding of potential synergistic effects, we explored the acute and chronic impacts of adding a DPP-4 inhibitor and SGLT2 inhibitor combination to metformin on hormonal responses and glucose metabolism in individuals with T2D with unsatisfactory glycemic control while receiving metformin monotherapy.

This phase 3, randomized, double-blind, parallel-group study was performed at the Department of Clinical and Experimental Medicine of the University of Pisa (Italy) between December 2019 and July 2022. The study was approved by the local ethics committee (Pisa, Italy; EudraCT no. 2016-005140-41) and conducted in accordance with the ethical principles of the Declaration of Helsinki and consistent with Good Clinical Practice. All participants provided written informed consent before participation into the study. The trial was registered with ClinicalTrials.gov (NCT03714594).

We recruited 51 patients with T2D (age >18 years; BMI <40 kg/m²; HbA_1c ≥7.0% to ≤10.0% [53–86 mmol/mol]) who had been receiving metformin (≥1,500 mg daily) at a stable dose for ≥12 weeks before randomization. Patients with poor glycemic control (>240 mg/dL fasting glucose), those using glucose-lowering agents other than metformin, those with a Chronic Kidney Disease Epidemiology Collaboration estimated glomerular filtration rate <60 mL/min/1.73 m², and those who experienced a cardiovascular event within 3 months before study initiation were not included in this study.

Participants were randomly assigned (1:1:1 by centralized blocked randomization schedule) to receive 5 mg saxagliptin (SAXA), 10 mg dapagliflozin (DAPA), or 5 mg saxagliptin plus 10 mg dapagliflozin (SAXA+DAPA) on top of metformin therapy. The metformin dose remained unchanged during the 4-week experimental period. At visit 1, patients were referred to the study center after a 10- to 12-h overnight fast for a screening visit and collection of baseline blood and urine samples for biochemical analyses. At visit 2, participants underwent, after overnight fasting, a mixed-meal tolerance test (MMTT). An antecubital vein was cannulated for baseline determination of glucose, insulin, C-peptide, glucagon, GLP-1, D-[6,6-²H₂]glucose, and [U-¹³C₆]‐glucose. A wrist vein also was cannulated, and the hand placed in a hot box (60°C) for intermittent arterialized blood sampling. A primed-constant D-[6,6-²H₂]glucose infusion (0.28 μmol ⋅ min⁻¹ ⋅ kg⁻¹; prime 28 μmol ⋅ kg⁻¹ ⋅ FPG/5) was initiated at time −180 min and maintained until study completion. At time 0, patients underwent (<10 min) an MMTT (75 g glucose dissolved in water spiked with 1.5 g [U-¹³C₆]‐glucose, 50 g cheese, one boiled egg, and 40 g white bread; 710 kcal; 58% carbohydrate; 24% fat; 18% protein). Blood samples were obtained at regular intervals (5, 15, 30, 60, 90, 120, 150, 180, and 240 min) until study end, and plasma glucose monitored every 15 min. Blood was sampled into fluoride tubes for plasma glucose analysis and into tubes containing heparin or EDTA plus aprotinin (250 KIU; BD, Plymouth, U.K.), as appropriate, for tracer and hormone determination. Specimens were promptly processed, and plasma stored at −80°C until analysis. A 24-h urine collection was performed the day before visit 2 to measure urinary glucose excretion (UGE). Urine was also collected throughout the entire duration of the MMTT to measure postprandial UGE. After visit 2, participants received study drugs according to randomization. All data obtained at visit 2 were considered baseline data (i.e., before drug exposure). The same procedures described were repeated after 3 (visit 3) and 28 days (visit 4) of treatment.

Plasma and urine glucose concentrations were measured by glucose oxidase. Enrichment of D-[6,6-²H₂]glucose and [U-¹³C₆]‐glucose was determined by gas chromatography/mass spectrometry (GC7890/MS5975; Agilent, Santa Clara, CA) using electron impact ionization and selective ion monitoring at mass-to-charge ratios of 202:200 and 205:200, respectively. Plasma insulin and C-peptide levels were determined by radioimmunoassay (Beckman Coulter, Brea, CA); the detection limits of the assay were 2 units/mL and 0.1 ng/mL, respectively, with intra- and interassay coefficients of variation <4%. Plasma glucagon was measured by radioimmunoassay (DRG Instruments, Marburg, Germany); the detection limit was 20 ng/mL, with intra- and interassay coefficients of variation of 4% and 9%, respectively. Plasma intact and total GLP-1 concentrations were measured using an ELISA kit (MilliporeSigma, Burlington, MA), with intra- and interassay coefficients of variation of 4%.

A calibration curve was first created providing a tracer-to-tracee ratio (TTR). From the TTR and plasma glucose concentration, the tracer concentration was obtained. The test meal included 75 g glucose plus 1.5 g [U-¹³C₆]‐glucose tracer. Therefore, a second TTR (TTR_meal) was determined, allowing estimation of plasma exogenous glucose concentration. We therefore used a circulatory model (12), as previously described (13), from which glucose clearance was calculated. The estimated clearance rate was used in the model, through a mathematical deconvolution-like operation, to quantify EGP. The rate of glucose disappearance (Rd) was calculated from endogenous glucose concentration and glucose clearance. UGE was calculated from urinary glucose concentration and urine volume obtained in the 24-h urine collection before the MMTT. Postprandial UGE was calculated from urinary glucose concentration and urine volume collected during the MMTT. Tissue glucose use (tissue Rd) was calculated as the difference between Rd and UGE.

After meal ingestion, the total glucose rate of appearance (RaO) was calculated using Steele’s equation, as previously described (14), which allows estimation of the metabolic clearance rate of glucose (MCR_gluc; i.e., volume of plasma irreversibly cleared of glucose as measured in unit time). Insulin resistance was assessed in fasting conditions by HOMA for insulin resistance (HOMA-IR) (15) and postprandial insulin sensitivity by PREDIM (16). β-Cell function was assessed from meal data using a previously described mathematical model to generate β-cell glucose sensitivity, rate sensitivity, potentiation factor, basal and total insulin secretion, and insulin secretion at selected glucose levels (17).

Data are presented as mean ± SEM, unless significant deviation from Gaussian distribution occurred or they were known from previous studies. In such cases, interquartile range is given. Primary (glucose, C-peptide, insulin, GLP-1, and glucagon levels) and derived data (glucose fluxes, parameters of β-cell function, and insulin resistance/sensitivity) were analyzed by general linear models with a statistical design of two-way ANOVA, with or without repeated measures as appropriate and with covariate(s) if necessary. Post hoc testing was performed with Bonferroni correction for multiple comparisons. Variables were log_e-transformed before analysis if deviation from Gaussian distribution was known or detected by Kolmogorov-Smirnov test.

The sample size determination was not straightforward. Based on previous studies reporting SGLT2 inhibitor and DPP-4 inhibitor effects on EGP (3,6,7), we calculated a minimum of 12 participants per group in a paired design to detect a difference between two means of 37%, with an SD of 30%, for α of 0.05 and power of 0.8. Statistical significance was declared at two-sided P < 0.05.

Patients’ disposition is shown in Supplementary Fig. 1. Of 62 screened patients, 51 entered the study and were randomly assigned to SAXA (n = 18), DAPA (n = 16), or SAXA+DAPA (n = 17). Six patients did not complete the study: two withdrew consent for personal reasons, one withrew for mild gastrointestinal intolerance, and three withdrew after visit 3 because of the SARS-CoV-2 pandemic. Therefore, 15 participants per group completed the study. Participants had an average age of 58 ± 8 years, mean diabetes duration of 7.7 ± 1.0 years, and baseline HbA_1c of 58 ± 6 mmol/mol. There were no differences across treatment groups in the explored parameters (Supplementary Table 1).

After 3-day treatment, glucose levels were reduced to a greater extent with SAXA+DAPA compared with individual agents (−31.1 ± 1.6 vs. −7.1 ± 2.1 vs. −17.0 ± 1.1 mg/dL, respectively; P < 0.01) (Fig. 1A–C). Basal insulin secretion rate did not change with SAXA, but it was reduced with DAPA and SAXA+DAPA (−11.3 ± 8.5 and −9.9 ± 3.7 pmol ⋅ min⁻¹ ⋅ m⁻², respectively; P < 0.05) (Fig. 2A). HOMA-IR decreased similarly in all groups. Intact GLP-1 levels increased with SAXA and SAXA+DAPA (+4.00 ± 1.04 and +5.70 ± 1.96 pmol/L, respectively; P = 0.001) and to a lesser extent, although significantly, with DAPA (+1.70 ± 0.60 pmol/L; P = 0.01) (Fig. 3A–C). Total GLP-1 levels did not change in the three groups (Supplementary Fig. 2). Glucagon concentrations increased with DAPA and SAXA+DAPA (+1.5 ± 3.4 and +2.7 ± 3.4 pmol/L, respectively; P = 0.04 vs. baseline) but did not change with SAXA (Supplementary Fig. 3). The I/G ratio decreased with SAXA+DAPA and DAPA (−1.86 ± 0.21 and −1.90 ± 0.20 molar ratio, respectively; P < 0.001 vs. baseline) (Supplementary Fig. 4). Similar results were obtained when the ratio of C-peptide to glucagon (C-pep/G) was considered (Supplementary Fig. 5).

EGP increased with SAXA+DAPA and DAPA, but it did not change with SAXA (from 2.09 ± 0.16 to 2.69 ± 0.27 and from 1.94 ± 0.15 to 2.56 ± 0.19 mg ⋅ kg⁻¹ ⋅ min⁻¹, respectively; P = 0.01 vs. baseline; P = 0.3 SAXA+DAPA vs. DAPA) (Table 1). Rd increased more with SAXA+DAPA than with SAXA (+0.67 ± 0.08 vs. +0.14 ± 0.08 mg ⋅ kg⁻¹ ⋅ min⁻¹; P = 0.01) and similarly between SAXA+DAPA and DAPA (+0.52 ± 0.11 mg ⋅ kg⁻¹ ⋅ min⁻¹; P = 0.12) (Table 1). Consistently, MCR_glu increased more with SAXA+DAPA and DAPA than with SAXA (+1.06 ± 0.19 and +0.73 ± 0.09, respectively vs. +0.44 ± 0.14 mL ⋅ kg⁻¹ ⋅ min⁻¹; P = 0.005; P = 0.41 SAXA+DAPA vs. DAPA) (Table 1). There was no change in 24-h UGE with SAXA, but it increased more with DAPA than with SAXA+DAPA (−2.10 ± 1.00 vs. +75.49 ± 7.48 and +51.96 ± 7.35 g, respectively; P < 0.001; P = 0.02 DAPA vs. SAXA+DAPA), consistent with lower fasting glucose with SAXA+DAPA (Table 1).

After the mixed meal, glucose levels improved more with SAXA+DAPA than with SAXA or DAPA (Fig. 1A–C); area under the curve was reduced by 36% compared with 15% and 13% with individual drugs (P ≤ 0.05) (Fig. 1D).

Insulin and C-peptide levels were similar across groups (Supplementary Figs. 6 and 7). Based on mathematical modeling assessment of β-cell function, a greater enhancement of insulin secretion rate occurred with SAXA+DAPA (+75%) than with SAXA or DAPA (+11% and +3%, respectively) (Fig. 2B).

Glucose sensitivity increased in all groups (Fig. 2C), with no changes in rate sensitivity or potentiation factor. Insulin sensitivity (PREDIM) increased to a greater extent with SAXA+DAPA than with SAXA or DAPA (+2.2 ± 1.7 vs. +0.4 ± 0.7 vs. +0.4 ± 0.4 mg ⋅ kg⁻¹ ⋅ min⁻¹; P = 0.007; P = nonsignificant [NS] SAXA vs. DAPA) (Fig. 2D).

Intact GLP-1 levels were higher with SAXA+DAPA than with SAXA or DAPA (+14.7 ± 3.0 vs. +7.8 ± 1.1 vs. +1.9 ± 1.0 pmol/L; P < 0.01) (Fig. 3D). Total GLP-1 levels decreased with SAXA (−9.5 ± 1.2 pmol/L; P = 0.004 vs. DAPA and SAXA+DAPA) and slightly increased with DAPA and SAXA+DAPA (Supplementary Fig. 2). Mean glucagon levels increased with DAPA and SAXA+DAPA (+1.0 ± 2.6 and +2.3 ± 4.0 pmol/L; P < 0.01 vs. baseline) but were lower with SAXA (−1.7 ± 3.4 pmol/L; P = 0.008 vs. baseline) (Supplementary Fig. 3). I/G and C-pep/G ratios rose and fell in a fashion parallel to glucose levels (Supplementary Figs. 4 and 5).

Meal ingestion was followed by mean EGP inhibition during the 240 min of the experimental procedure (Table 1), with greater suppression (i.e., difference between fasting and mean EGP during MMTT) seen with SAXA+DAPA and DAPA than with SAXA (−48% vs. −44% vs. −34%, respectively; P = 0.02 SAXA+DAPA vs. SAXA; P = 0.2 SAXA+DAPA vs. DAPA) (Supplementary Fig. 8). Total Rd increased similarly in all groups, although during the first 60 min after MMTT, it increased more with SAXA+DAPA than with SAXA (P < 0.01) and similarly between SAXA+DAPA and DAPA (P = 0.26). Mean MCR_glu showed a similar pattern; it increased in all groups (P = 0.8), but the increase was greater during the first 60 min after MMTT with SAXA+DAPA than with SAXA or DAPA (+0.99 ± 0.12 vs. +0.37 ± 0.11 and +0.58 ± 0.08 mL ⋅ kg⁻¹ ⋅ min⁻¹, respectively; P < 0.01 vs. SAXA; P = 0.03 vs. DAPA) (Supplementary Fig. 9). Tissue Rd (Table 1) and RaO of oral glucose were similar across groups (Supplementary Fig. 10).

After 4-week treatment, body weight, BMI, and blood pressure decreased in all groups (Supplementary Table 1).

Glucose levels and basal insulin secretion rate declined similarly in the three groups compared with baseline (Figs. 1A–C and 2 A), and HOMA-IR remained lower.

With SAXA+DAPA and SAXA, intact GLP-1 levels were similar to those at day 3, although the difference from baseline was greater with SAXA+DAPA (+6.4 ± 2.2 and +2.9 ± 1.1 pmol/L, respectively; P < 0.05) (Fig. 2A–C). With DAPA, intact GLP-1 levels returned to baseline levels. Total GLP-1 levels did not change in the three groups. Glucagon concentration declined with DAPA and SAXA+DAPA (Supplementary Fig. 3); the C-pep/G ratio returned to baseline level with SAXA+DAPA (0.24 ± 0.04 vs. 0.25 ± 0.03 molar ratio; P = 0.07) and remained lower with DAPA but did not change in the SAXA group (Supplementary Fig. 5).

Fasting EGP remained stable with SAXA+DAPA and DAPA (2.9 ± 0.3 and 2.6 ± 0.3 mg ⋅ kg⁻¹ ⋅ min⁻¹; P = 0.17) and did not change with SAXA (Table 1). Similarly, Rd and MCR_glu remained higher and similar to those at day 3 with all treatments (Table 1). No difference in tissue Rd could be detected (Table 1).

After meal ingestion, a greater glucose level reduction occurred with SAXA+DAPA compared with SAXA or DAPA (−106 ± 14 vs. −58 ± 15 vs. −57 ± 10 mg/dL; P < 0.01; P = NS SAXA vs. DAPA) (Fig. 1A–D). Supplementary Figs. 6 and 7 show insulin and C-peptide time courses. Insulin secretion rate slightly further increased in all groups, but it was higher with SAXA+DAPA compared with SAXA or DAPA (+57.3 ± 50.1 vs. +16.4 ± 35.4 vs. +16.0 ± 39.1 pmol ⋅ min⁻¹ ⋅ m⁻²; P = 0.003 vs. baseline; P = NS SAXA vs. DAPA) (Fig. 2B). Glucose sensitivity further increased with SAXA+DAPA, but it was slightly reduced or unchanged in the other groups (+18.3 ± 5.7 and −3.7 ± 7.8 and +1.8 ± 5.7 pmol ⋅ min⁻¹ ⋅ m⁻² ⋅ mmol/L⁻¹ vs. day 3; P = 0.02) (Fig. 2C).

Compared with day 3, insulin sensitivity (PREDIM) improved with SAXA+DAPA (+2.2 ± 1.7 vs. +0.3 ± 0.7 vs. −0.2 ± 0.1 mg ⋅ kg⁻¹ ⋅ min⁻¹; P = 0.007; P = 0.2 SAXA vs. DAPA) (Fig. 2D) but did not change with SAXA or DAPA.

There were no additional changes in intact GLP-1 levels, although they remained higher with SAXA+DAPA than with SAXA (25.8 ± 5.7 vs. 17.3 ± 3.2 pmol/L; P = 0.009) (Fig. 3D) and did not change with DAPA. With SAXA, total GLP-1 levels were similar to those at day 3, whereas with DAPA and SAXA+DAPA, they returned to baseline levels. Glucagon levels (Supplementary Fig. 3) were lower compared with those at day 3 with SAXA+DAPA and DAPA and similar to those at day 3 with SAXA (10.7 ± 6.5 vs. 10.6 ± 3.9 vs. 10.9 ± 5.5 pmol/L, respectively; P = 0.3). Similar changes occurred for I/G and C-pep/G ratios (Supplementary Figs. 4 and 5).

EGP suppression was superimposable in all groups with that of day 3 (SAXA −36%, DAPA −42%, and SAXA+DAPA −50%; P = 0.02 SAXA+DAPA vs. SAXA) (Table 1 and Supplementary Fig. 8). The changes in Rd, tissue Rd, and MCR_glu (Supplementary Fig. 9) observed at day 3 persisted after 4-week treatment in all groups (Table 1). UGE during MMTT did not change with SAXA, but with DAPA and SAXA+DAPA, it was similar to that at day 3 (+1.40 ± 0.28 and +1.12 ± 0.25 mg ⋅ kg⁻¹ ⋅ min⁻¹; P < 0.01 vs. SAXA; P = 0.53 DAPA vs. SAXA+DAPA) (Table 1). RaO did not change in the three groups (Supplementary Fig. 10).

No serious adverse events (hypoglycemia, urinary tract infections, or any other adverse effect) were observed throughout the study.

Concomitant use of dapagliflozin and saxagliptin added to metformin in individuals with T2D with unsatisfactory glycemic control with metformin monotherapy elicited a rapid and greater improvement in fasting and post-MMTT glucose profiles that persisted after 4-week therapy, in line with prior longer clinical trials (8,18). Our study provides insights into the complementary effect of drugs with different mechanisms of action by exploring changes in the hormonal milieu and glucose metabolism elicited by the concomitant use of a DPP-4 inhibitor and an SGLT2 inhibitor.

The reduction in glucose levels induced by the three treatments was associated with improvement in β-cell function, as highlighted by the data generated by a validated mathematical model (17). Insulin secretory rate, adjusted for basal potentiation, increased to a greater extent with SAXA+DAPA than with SAXA or DAPA only 3 days after treatment (+75%, +11%, and +4%, respectively) and further increased after 4 weeks (+90%, +11%, and +21% from baseline, respectively). SAXA elicited an 80% increase in glucose sensitivity (i.e., the sensitivity of the β-cell to changes in glucose levels), in keeping with earlier studies exploring the effect of DPP-4 inhibitors using a similar mathematical modeling approach (14,19). Of interest, a similar early improvement in glucose sensitivity occurred with SAXA+DAPA and SAXA alone, most likely reflecting enhanced GLP-1 availability by SAXA. However, after 4-week treatment with SAXA+DAPA, glucose sensitivity increased further (+146% vs. +65% and +34% from baseline, respectively), a finding that may be attributable to the chronic glucose control attained with combination therapy and the subsequent greater relief in glucose toxicity to the β-cell.

Impaired insulin secretion, together with insulin resistance, is the main pathogenetic defect in T2D. In our study, all treatments were associated with a reduction in HOMA-IR, reflecting an improvement in fasting insulin sensitivity. Under insulin-stimulated conditions (i.e., after mixed-meal ingestion), the PREDIM index increased in all treatment groups and to a greater extent with SAXA+DAPA (Fig. 2D). In fact, after 3-day treatment, PREDIM increased by 87% with SAXA+DAPA and only by 19% and 15% with SAXA and DAPA, respectively. After 4-week treatment, PREDIM was still unchanged and higher with SAXA+DAPA, with a modest further increase with SAXA and DAPA (+6% and +13%, respectively).

Improved β-cell function may reflect increased intact GLP-1 availability elicited by DPP-4 inhibition. Not surprisingly, DAPA alone was not associated with any increase in GLP-1 levels besides the physiological response to the meal ingestion. Nonetheless, with SAXA+DAPA, a modest but detectable increase in fasting intact GLP-1 levels became apparent and was fully appreciated in the postprandial phase. Moreover, total GLP-1 during MMTT showed a significant reduction with SAXA but did not change with DAPA or SAXA+DAPA in early and late responses to treatment.

These findings are consistent with previous results. In a study in individuals with T2D receiving single-dose (acute) or 4-week (chronic) empagliflozin treatment, GLP-1 levels were found to increase acutely but not at study completion (7). In contrast, increased GLP-1 levels have been reported in patients pretreated with metformin (20). One study in healthy Japanese participants found increased GLP-1 levels with administration of canagliflozin and teneligliptin compared with teneligliptin alone (21). This effect has been hypothesized to result from a vagal nerve activation mediated by increased SGLT2 stimulation (22,23). We speculate that in our study SGLT2 inhibition may have interfered with the feedback mechanism in L cells, resulting in greater availability of active GLP-1 after nutrient intake. We therefore hypothesize that the enhanced intact GLP-1 availability elicited by SAXA+DAPA may have exerted a direct β-cell effect, because it was already apparent after 3-day treatment, and the chronic amelioration of glucose levels could have contributed to a further improvement in β-cell response (i.e., reduced glucose toxicity). The percentage of women was numerically (but not statistically different) higher in the SAXA+DAPA group, and a potential effect of sex was reported in a recent study by Xie et al. (24). However, this seems unlikely to account for our observation in individuals with T2D, whereas young lean individuals without diabetes undergoing intraduodenal glucose infusion were included in the study by Xie et al. (24).

Earlier studies demonstrated that SGLT2 inhibitors could elicit a paradoxical increase in glucagon secretion that in concomitance with lower insulin levels could lower the portal I/G ratio and sustain liver gluconeogenesis (11,25). DAPA was also associated with increased fasting EGP, in line with the original hypothesis of DeFronzo et al. (6) and Ferrannini et al. (7). However, as these authors had already suggested, glucagon increase was relatively short lived. In our study, after 4-week therapy, fasting glucagon levels and the C-pep/G ratio had returned baseline values, despite elevated fasting EGP persisting for the duration of the study, as reported by others (11). Altogether, these findings suggest that mechanisms other than increased glucagon levels must account for the persistence of fasting EGP with SGLT2 inhibitors, including but not limited to changes in other circulating substrates/hormones, catecholamines, and central nervous system mediators (26).

Despite the persistent basal EGP increase, the maximal effect of its suppression after meal ingestion was attained after only 3 days of treatment. Of interest, with SAXA+DAPA, EGP suppression during MMTT (day 3 −48% and week 4 −50%) tended to be greater compared with the two drugs alone, and it was maintained after 4 weeks (SAXA day 3 −34% and week 4 −36%; DAPA day 3 −44% and week 4 −42%), suggesting enhanced EGP suppression. This result differs from that observed in another study (27) conducted with single-dose DAPA, reporting no EGP suppression after an oral glucose load. In our study, longer treatment and a different challenge (MMTT vs. oral glucose tolerance test) may have reduced glucotoxicity more effectively and enhanced insulin secretion, with greater impact on EGP.

Although greater EGP inhibition may have contributed to the overall postprandial glucose improvement, it was mainly the SGLT2 inhibitor glucosuric effect that accounted for the better glycemic outcome in our study. This is fully apparent if glucose kinetic data are considered. In fasting and postprandial states, Rd increase was greater with SAXA+DAPA, with a concomitant increase in MCR_glu. Rd is the total rate of disappearance of glucose from the plasma glucose pool. With SGLT2 inhibitors, Rd comprises glucose use by body tissues plus glucose loss through urine. Therefore, when urinary glucose excretion was subtracted from total Rd, no major difference was apparent in tissue Rd. This may sound at variance with increased PREDIM, but the short duration of our study must be considered, because improved insulin action may require a longer time to be documented by the tracer technique.

With DAPA treatment, 24-h UGE increased both acutely and chronically. However, it is of interest that with SAXA+DAPA, UGE increased to a lesser extent than with DAPA alone. Whether this may contribute to reduced urinary tract and genital infections in patients treated with a combination of an SGLT2 inhibitor and DPP-4 inhibitor (9) remains an intriguing hypothesis (28). In our own participants, although limited in number, no episodes of infection were reported.

The main limitation of our study is its short duration, which makes it difficult to extrapolate all reported changes in hormonal and metabolic responses to a more generalized population with diabetes; also it remains to be determined whether the benefit provided by SAXA+DAPA is sustained beyond the 4-week treatment period of the study. Another limitation is the lack of a placebo group, which may have supported the net effects of SAXA+DAPA treatment. Finally, we cannot exclude an interaction with existing background metformin therapy, although metformin was used in all participants, and no change in dose occurred throughout the study duration.

The main advantage of the study is represented by a detailed assessment of the hormonal and metabolic effects of treatments using a state-of-the art technique for the assessment of in vivo glucose metabolism and β-cell function.

In summary, SAXA+DAPA as an add-on therapy to existing metformin treatment leads to a greater improvement in fasting and postprandial glucose than that seen with the addition of each monocomponent, both acutely (3 days) and chronically (4 weeks). These improvements were mainly based on improved β-cell function and, to some extent, insulin sensitivity and GLP-1 availability. These changes do not seem sufficient to reduce the compensatory increase in fasting EGP elicited by SGLT2 inhibition, but SAXA+DAPA may exert greater EGP suppression after meal ingestion. In conclusion, the final effect of SAXA+DAPA on glucose metabolism is additive and explained by modifications in the kinetics of glucose caused by the modulation of synergistic mechanisms.

Acknowledgments. The authors thank the nursing staff and, in particular, Angela Augugliaro, Jancy J. Kurumthodathu, and Simone Goretti for their valuable support in conducting the study.

Duality of Interest. The study was supported by AstraZeneca with unrestricted grant ESR-15-11272. S.D.P. consulted for Abbott, Amarin Corporation, Applied Therapeutics, AstraZeneca, Boehringer Ingelheim, Eli Lilly, Menarini International, Novo Nordisk, Sanofi, and Sun Pharmaceuticals and received funding for these consulting services; received grant support from AstraZeneca and Boehringer Ingelheim; and received speaker fees from Abbott, AstraZeneca, Berlin-Chemie, Boehringer Ingelheim, Eli Lilly, Laboratori Guidotti, Menarini International, Merck Sharp & Dohme, Novartis, Novo Nordisk, and Sanofi. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. G.D., A.T., and S.D.P. researched data and wrote the manuscript. A.B. and A.D. researched data, contributed to discussion, and reviewed/edited the manuscript. A.S., B.C., and V.S.-B. researched data. G.D. and S.D.P. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Handling Editors. The journal editors responsible for overseeing the review of the manuscript were Cheryl A.M. Anderson and Cuilin Zhang.