Changes in β-Cell Function and Insulin Sensitivity During Treatment With Dapagliflozin Alone or in Combination With Exenatide in Type 2 Diabetes

Abstract

OBJECTIVE

To examine the effects of sodium–glucose cotransporter 2 inhibitors (SGLT2is) alone or with glucagon-like peptide 1 receptor agonists (GLP-1RAs) on β-cell function (BCF) in type 2 diabetes. The hypothesis was that an SGLT2i combined with a GLP-1RA provides superior improvement in BCF than either agent alone.

RESEARCH DESIGN AND METHODS

Ninety patients underwent a 180-min oral glucose tolerance test (OGTT) 1) after one drug dose (acute study) (placebo [n = 15], dapagliflozin [n = 25], exenatide [n = 25], and dapagliflozin/exenatide [n = 25]) and 2) after 1 and 4 months of therapy. Corrected Matsuda index (cMI) for urinary glucose loss, insulin secretion, and BCF indices were calculated during OGTT.

RESULTS

In the acute study, mean ± SEM cMI in dapagliflozin (2.29 ± 0.33), exenatide (2.03 ± 0.12), and dapagliflozin/exenatide (2.36 ± 0.14) was higher (P < 0.05) than placebo (1.63 ± 0.36). After 1 and 4 months, cMI remained similarly elevated in exenatide and increased further (P < 0.001) in dapagliflozin and dapagliflozin/exenatide. In the acute study, insulin secretion in dapagliflozin was similar to placebo but higher (P < 0.001 vs. both) in exenatide and dapagliflozin/exenatide. After 1 and 4 months in exenatide and in dapagliflozin/exenatide, insulin secretion remained higher (P < 0.01 vs. both) than dapagliflozin. BCF index in the acute study was 0.40 ± 0.04 in placebo, 62% higher (P < 0.05) in dapagliflozin (0.65 ± 0.10), threefold higher in exenatide (1.17 ± 0.22), and fourfold higher in dapagliflozin/exenatide (1.69 ± 0.12) (all P < 0.001 vs. placebo). At 1 and 4 months, BCF rose further in dapagliflozin and exenatide but did not increase further in dapagliflozin/exenatide.

CONCLUSIONS

Dapagliflozin and exenatide monotherapy cause sustained improvements in BCF and insulin sensitivity. Combination therapy with dapagliflozin plus exenatide markedly augmented both BCF and insulin sensitivity above that with either agent alone.

Graphical Abstract

Introduction

Progressive β-cell failure, superimposed on insulin resistance, is the principal factor responsible for the continued decline in glucose control in patients with type 2 diabetes (T2D) (1–4). Multiple genetic and environmental factors contribute to the deterioration of β-cell function (BCF) (1–3). Sustained elevation of plasma glucose (PG) (glucotoxicity) (5,6) and free fatty acid (lipotoxicity) (7) concentrations contribute to the progressive decline in BCF in individuals at risk for developing T2D (8). Conversely, reducing the PG concentration with insulin therapy improves BCF in patients with T2D (9–11). However, in addition to lowering PG concentration, insulin exerts multiple metabolic effects, including reduction of plasma free fatty acid concentration, which independently can improve BCF.

Sodium–glucose cotransporter 2 inhibitors (SGLT2is) decrease PG concentration by blocking renal glucose absorption, thus producing glucosuria (12); the reduction in PG improves insulin-mediated glucose uptake in muscle (13–15). Because skeletal muscle does not express SGLT2, these results indicate that the improvement in peripheral tissue insulin sensitivity is not a direct effect of SGLT2is but is secondary to the attenuation of glucotoxicity (13–15). Studies in humans (16,17) and in experimental animal models of diabetes also have reported improved BCF with SGLT2is (18–21). Since no SGLT2 receptors are present in islet cells (22), the beneficial effects of SGLT2is on BCF most likely are related to reduced PG concentration. Results obtained with oral glucose tolerance test (OGTT) (16,23) and hyperglycemic clamp (24) have demonstrated that in patients with T2D, treatment with dapagliflozin and empagliflozin for 2–4 weeks was accompanied by improved BCF.

Treatment with glucagon-like peptide 1 receptor agonists (GLP-1RAs) also has been shown to enhance insulin secretion in patients with T2D (25,26), and an acute infusion of GLP-1RA augments insulin secretion in dapagliflozin-treated patients with T2D (27). In the current study, we sought to determine whether chronic treatment with dapagliflozin, used alone or in combination with exenatide, improves insulin sensitivity, insulin secretion, and BCF. We hypothesized that SGLT2is combined with GLP-1RAs would provide superior improvement in BCF, insulin sensitivity, and glycemic control compared to treatment with either agent alone in patients with T2D.

Research Design and Methods

Participants

Ninety participants with T2D were included in the study. Background treatment included diet (n = 24), metformin alone (n = 37), and metformin plus sulfonylurea (n = 29) and was stable for at least 3 months. Except for having T2D, all participants were in good general health based on medical history, physical examination, screening blood tests, urinalysis, and electrocardiogram. Clinical, anthropometric, and laboratory data were similar in all groups and are shown in Table 1. Body weight was stable (±1.5 kg) for at least 3 months prior to study, and no participant participated in any excessively heavy exercise program. Participants taking drugs known to affect glucose metabolism (other than metformin and sulfonylurea) were excluded. The study was approved by the University of Texas Health San Antonio Institutional Review Board, and written informed consent was obtained from all participants.

Table 1.

Indices of insulin sensitivity, pancreatic insulin secretion, and BCF in all groups during the acute studies and after 1 and 4 months of therapy

	Acute study				1 Month			4 Months
	Placebo	Dapagliflozin	Exenatide	Dapagliflozin/exenatide	Dapagliflozin	Exenatide	Dapagliflozin/exenatide	Dapagliflozin	Exenatide	Dapagliflozin/exenatide
cMI	1.63 ± 0.16	2.29 ± 0.13*	2.03 ± 0.12*	2.36 ± 0.14*	2.85 ± 0.20@	1.94 ± 0.17##	3.10 ± 0.13@	2.87 ± 0.14	1.86 ± 0.26	2.98 ± 0.20@
Insulin secretion, µU/mL per mg/dL	0.24 ± 0.04	0.28 ± 0.06	0.57 ± 0.17**	0.72 ± 0.15**	0.21 ± 0.13	1.35 ± 0.15#	0.84 ± 0.18	0.21 ± 0.05	0.65 ± 0.22	0.53 ± 0.07
BCF	0.40 ± 0.04	0.64 ± 0.10*	1.17 ± 0.22**	1.69 ± 0.12**@	0.59 ± 0.06	2.62 ± 0.13#	2.60 01.9#	0.62 ± 0.09	1.21 ± 0.13	1.59 ± 0.13@

Data are mean ± SE.

*P < 0.01 vs. placebo.

**P < 0.001 vs. placebo.

#P < 0.001, 1 month vs. acute.

##P < 0.01, 1-month vs. acute.

@P < 0.05, dapagliflozin/exenatide vs. exenatide.

The results for glucose kinetics were reported in a prior publication (28) and are not repeated herein. In this report, we focused on the relationship between the improvements in glycemic control (HbA_1c and OGTT) and their physiologic determinants, including insulin sensitivity, insulin secretion, BCF, and urinary glucose excretion (UGE). Participants participated in three double-tracer OGTT studies (28) that were performed at baseline after a single dose (acute study) of medication (dapagliflozin [n = 25], exenatide [n = 25], dapagliflozin/exenatide [n = 25], or placebo [n = 15]) and after 1 month and 4 months of drug treatment. The placebo group participated only in the acute study with OGTT. In the placebo group, incomplete data were present in two participants; therefore, the presented results represent a mean of 13.

Experimental Design

All studies were performed in the Clinical Research Center of the Texas Diabetes Institute at 0700 h following an overnight fast. During the double-tracer OGTT, participants received an 8-h prime (40 µCi × fasting PG / 100) continuous (0.40 µCi/min) 3-³H-glucose infusion via an antecubital vein catheter, as previously described (29). After a 3-h tracer equilibration period (time 0), participants ingested 75 g of glucose (Trutol 75; Thermo Fisher Scientific, Middletown, VA) containing 100 µCi of 1-¹⁴C-glucose (PerkinElmer, Boston, MA). At time 0, participants also received 1) oral dapagliflozin 10 mg, 2) exenatide 5 µg subcutaneous, or 3) dapagliflozin 10 mg plus exenatide 5 µg in random order. Arterialized venous blood samples (hot box technique) were drawn from the hand vein at −30, −20, −10, −5, 0, and every 15–20 min thereafter for 3 h for PG, insulin, C-peptide, glucagon concentrations, and tritiated glucose and ¹⁴C-glucose radioactivity. Plasma insulin was determined with immunoradiometric assay (BioZ, Louvain, Belgium) and C-peptide and glucagon with radioimmunoassay (EMD Millipore, Billerica, MA). Urine was collected prior to and after drug administration for measurement of UGE.

Following completion of baseline double-tracer OGTT (acute study), participants were randomized to dapagliflozin 10 mg/day, exenatide 5 μg twice daily, or dapagliflozin 10 mg/day plus exenatide 5 μg twice daily and instructed to consume a weight-maintaining diet for 4 months. Participants continued to take their background dose of metformin or metformin/sulfonylurea throughout the 4-month study period. Participants were contacted weekly by phone and returned to the Clinical Research Center every 2–4 weeks. At 1 month and 4 months, the double-tracer OGTT study was repeated. After the OGTT study at 1 month, participants receiving exenatide 5 μg twice daily and those receiving dapagliflozin 10 mg/day plus exenatide 5 μg twice daily were switched to exenatide 2 mg/week subcutaneous injection. At 4 months, patients switched to exenatide 2 mg/week received the OGTT 4 days after the last dose of weekly exenatide.

Calculations

The Matsuda index of insulin sensitivity was calculated as follows (30):

$$\mathrm{\text{Matsuda index}}=\begin{array}{c}\mathrm{10,000}/\sqrt{\left(Glu[b]\ast \text{Ins}[\text{b}]\right)\ast \left(G\text{lu}[0,180]\right)\ast (\text{Ins}[0,180]),}\end{array}$$

where Glu[b] is fasting PG concentration, Ins[b] is fasting plasma insulin concentration, and Glu[0,180] and Ins[0,180] represent the mean PG and insulin concentrations during 0–180 min of the OGTT.

In participants who received dapagliflozin treatment, either alone or in combination with exenatide, a corrected Matsuda index (cMI) was calculated. This was achieved by determining the total amount of glucose excreted in the urine during 180 min of the OGTT, which was subtracted from the 75 g of the oral glucose intake, according to validated data (31).

Insulin secretion was calculated as follows:

$$\mathrm{\text{Insulin \hspace{0.17em}secretion}}=\text{Ins}\left[0,180\right]–\text{ Ins}\left[\text{b}\right]/\text{Glu}\left[0,180\right]–\text{ Glu}\left[\text{b}\right].$$

BCF was calculated as follows:

$$\mathrm{\text{BCF}}=\left[\text{IS}\right]\times \left[\text{cMI}\right],$$

where IS represents the mean increment plasma insulin concentration from 0 to 180 min.

Statistical Analysis

The primary end point was the change in BCF in the group receiving combined dapagliflozin/exenatide therapy after 1 and 4 months compared with the change in the group receiving dapagliflozin monotherapy or exenatide monotherapy after 1 and 4 months using ANOVA. Changes in BCF following the administration of each individual drug (or placebo) during the acute and chronic studies (i.e., from baseline to 1 month and from baseline to 4 months) were compared using paired t tests. Post hoc testing was done using Bonferroni correction. Pearson correlation coefficient was used to determine the relationship between 1) the decrease in HbA_1c and 2) glucose excursion during the 180-min OGTT versus cMI, insulin secretion, BCF index, and UGE. Values are presented as mean ± SEM. P < 0.05 was considered significant.

Data and Resource Availability

The data sets generated during current study are available from the corresponding author upon reasonable request.

Results

PG Concentration

During the acute study, PG (in mg/dL) during the 180-min OGTT in placebo increased from a mean of 161 ± 7 to 294 ± 8 (Δ 133 ± 6), which was significantly greater (P < 0.01) than with dapagliflozin (152 ± 9 to 264 ± 12, Δ 112 ± 3), exenatide (152 ± 7 to 214 ± 11, Δ 62 ± 4) and dapagliflozin/exenatide (182 ± 10 to 212 ± 17, Δ 30 ± 7); the increase in PG with dapagliflozin/exenatide also was lower (P < 0.01) than with dapagliflozin and exenatide alone. At 1 month, fasting PG in dapagliflozin (126 ± 4), exenatide (129 ± 8), and dapagliflozin/exenatide (128 ± 7) was similar and significantly reduced (P < 0.01) versus the pretreatment value. During the 180-min OGTT, the increment in PG in dapagliflozin/exenatide increased from 127 ± 7 to 158 ± 8 (Δ 31 ± 1), which was comparable to that in exenatide from 129 ± 8 to 169 ± 11 (Δ 40 ± 3), and both were significantly (P < 0.01) lower than the increase in dapagliflozin from 126 ± 4 to 244 ± 7 (Δ 118 ± 3). At 4 months, fasting PG was lower (P < 0.01) in dapagliflozin/exenatide (109 ± 5) versus dapagliflozin (129 ± 5) and exenatide (125 ± 4). During the 180-min OGTT, the PG increment in dapagliflozin (129 ± 5 to 240 ± 6, Δ 111 ± 1) was higher (P < 0.05) than the increment in exenatide (125 ± 8 to 214 ± 7, Δ 89 ± 1), which was higher (P < 0.05) than the increase in dapagliflozin/exenatide (109 ± 5 to 187 ± 9, Δ 78 ± 4).

UGE

In the acute study, the increase in UGE (in mg/kg/min) during the 180-min OGTT in dapagliflozin increased from a mean of 0.03 ± 0.01 at baseline to 0.98 ± 0.15; this was similar to the increase with dapagliflozin/exenatide (0.07 ± 0.01 to 1.02 ± 0.10), and both were significantly higher (P < 0.001) than the increase in UGE with placebo (0.04 ± 0.02 vs. 0.20 ± 0.06) and exenatide (0.03 ± 0.01 vs. 0.17 ± 0.05). During the 180-min OGTT at 1 month, UGE increased similarly in dapagliflozin (0.51 ± 0.08 to 1.11 ± 0.07) and dapagliflozin/exenatide (0.30 ± 0.08 to 0.82 ± 0.12) and did not change with exenatide (0.03 ± 0.02 to 0.05 ± 0.02) (P < 0.001 vs. dapagliflozin and dapagliflozin/exenatide). At 4 months, changes in UGE paralleled those at 1 month in all three groups.

Plasma Hormones

During the acute study, the increment in plasma insulin (in μU/mL) during the 180-min OGTT in dapagliflozin (mean ± SEM 11 ± 1 vs. 42 ± 4) was similar to placebo (13 ± 2 vs. 45 ± 7), slightly higher in exenatide versus placebo (15 ± 2 vs. 50 ± 5), and significantly lower (P < 0.01) in dapagliflozin/exenatide versus placebo (13 ± 1 vs. 34 ± 4). After 1 month, the increase in plasma insulin in dapagliflozin (11 ± 1 to 35 ± 6) and dapagliflozin/exenatide (13 ± 2 to 39 ± 7) was comparable, and the increase with exenatide (17 ± 2 vs. 71 ± 5) was significantly higher (P < 0.01). After 4 months, the increase in plasma insulin was higher in dapagliflozin/exenatide (11 ± 1 vs. 50 ± 7) than in dapagliflozin (11 ± 1 vs. 35 ± 5) (P < 0.05) and highest with exenatide (15 ± 2 vs. 72 ± 7) (P < 0.01 vs. dapagliflozin and dapagliflozin/exenatide).

In the acute study, the mean increase in plasma C-peptide concentration (in ng/mL) during the 180-min OGTT was similar in placebo (3.3 ± 0.2 to 7.0 ± 0.5), dapagliflozin (3.1 ± 0.2 to 6.4 ± 0.7), and dapagliflozin/exenatide (3.1 ± 0.3 to 6.3 ± 0.8), and all were significantly lower than with exenatide (3.6 ± 0.3 to 8.4 ± 0.8) (P < 0.05). At 1 month, the increase in plasma C-peptide during the 180-min OGTT in dapagliflozin (3.2 ± 0.2 to 7.6 ± 0.8) and dapagliflozin/exenatide (3.3 ± 0.3 to 8.1 ± 0.6) was similar and less than the increase with exenatide (3.9 ± 0.3 to 11.1 ± 1.3) (P < 0.05). Likewise, after 4 months, the increase in plasma C-peptide during the 180-min OGTT in dapagliflozin (3.1 ± 0.3 to 8.1 ± 0.9) was similar to the increase with dapagliflozin/exenatide (3.0 ± 0.2 to 8.8 ± 0.7) and less than the increment with exenatide (4.0 ± 0.4 vs. 11.5 ± 1.4) (P < 0.05).

In the acute study, during the 180-min OGTT, the plasma glucagon concentration (in pg/mL) in placebo (mean ± SEM 48 ± 4 vs. 45 ± 6) and dapagliflozin/exenatide (44 ± 4 vs. 45 ± 5) did not change. In exenatide, plasma glucagon decreased from 55 ± 5 to 47 ± 5, and in dapagliflozin, it increased from 50 ± 5 to 56 ± 5 (P < 0.05 exenatide vs. dapagliflozin). At 1 month, plasma glucagon during the 180-min OGTT increased with dapagliflozin (43 ± 4 to 49 ± 5), decreased with exenatide (57 ± 5 to 42 ± 4) and remained unchanged with dapagliflozin/exenatide (41 ± 3 vs. 38 ± 2) (P < 0.05 exenatide vs. dapagliflozin). After 4 months, plasma glucagon during the 180-min OGTT did not change with dapagliflozin (53 ± 6 vs. 44 ± 4) and dapagliflozin/exenatide (42 ± 3 vs. 36 ± 4) and decreased significantly with exenatide (43 ± 3 vs. 30 ± 2) (P < 0.01).

Indices of Insulin Sensitivity (cMI), Insulin Secretion, and BCF

During the OGTT in the acute study, mean cMI in dapagliflozin (2.29 ± 0.33), exenatide (2.03 ± 0.23), and dapagliflozin/exenatide (2.36 ± 0.14) was higher than placebo (1.63 ± 0.36) (P < 0.01 to P < 0.001) (Table 1). After 1 and 4 months, cMI remained the same in exenatide (2.60 ± 0.31 and 2.82 ± 0.26, respectively) and increased in dapagliflozin (2.85 ± 0.20 and 2.87 ± 0.14, respectively) and in dapagliflozin/exenatide (3.10 ± 0.13 and 2.98 ± 0.20, respectively). The increase in cMI in dapagliflozin/exenatide was significantly greater than in dapagliflozin alone and exenatide alone (P < 0.05) (Table 1).

In the acute study, mean insulin secretion (in µU/mL per mg/dL) in dapagliflozin (0.28 ± 0.06) was similar to placebo (0.24 ± 0.04) and markedly lower (P < 0.001 vs. both) than in exenatide (0.57 ± 0.17) and dapagliflozin/exenatide (0.72 ± 0.15) (Table 1). After 1 and 4 months, insulin secretion in exenatide (1.35 ± 0.15 and 0.65 ± 0.22, respectively) and dapagliflozin/exenatide (0.84 ± 0.18 and 0.53 ± 0.07, respectively) also was higher (P < 0.001 vs. both) than dapagliflozin (0.21 ± 0.13 and 0.21 ± 0.05, respectively).

The mean BCF index in the acute study was 0.40 ± 0.04 in placebo and increased by 62% in dapagliflozin to 0.65 ± 0.10 (P < 0.01), increased by threefold in exenatide (1.17 ± 0.22), and increased by fourfold in dapagliflozin/exenatide (1.69 ± 0.12) compared with placebo (P < 0.001 vs. placebo and dapagliflozin). At 1 month, BCF remained elevated in dapagliflozin (0.59 ± 0.06) and increased further in exenatide (2.62 ± 0.13) and dapagliflozin/exenatide (2.60 ± 0.19) (both P < 0.001 vs. acute). At 4 months, BCF remained elevated in dapagliflozin (0.62 ± 0.09), exenatide (1.21 ± 0.13), and dapagliflozin/exenatide (1.59 ± 0.12); the BCF index in dapagliflozin/exenatide was 25% higher (P < 005) than in exenatide (Table 1 and Fig. 1).

Figure 1.

Change in BCF indices in the placebo (PCB), dapagliflozin (DAPA), exenatide (EXE), and dapagliflozin plus exenatide combination (DAPA + EXE) groups in the acute study and following 1 and 4 months of therapy. In the acute study, the BCF index after a single dose of DAPA was 62% higher than in PCB and threefold higher in EXE and fourfold higher in DAPA + EXE than in PCB. After 1 month of therapy, the increase in BCF index in DAPA remained the same as in the acute study, more than doubled in EXE, and increased further by ∼65% in DAPA + EXE. After 4 months of therapy, the increases in the BCF index in all three groups remained similar to the acute study; the BCF index in DAPA + EXE was 25% higher than in EXE. *P < 0.01 vs. PCB, **P < 0.001 vs. PCB, ***P < 0.01, DAPA + EXE vs. EXE; #P < 0.001, 1 month vs. acute; @P < 0.05, DAPA + EXE vs. EXE. AU, arbitrary unit.

Since some participants were treated with sulfonylureas, we also calculated the insulin sensitivity and BCF indices in the subgroup of patients who received metformin plus sulfonylurea (Supplementary Table 1). The results in this group were very similar to those obtained in the entire group of participants, which included patients treated with diet alone, metformin alone, and metformin plus sulfonylurea (Table 1).

After 4 months of treatment with dapagliflozin, there was a strong positive correlation between the BCF index and the decrease in HbA_1c (r = 0.45, P < 0.01) and a negative correlation between the BCF index and the increment in PG during 180-min OGTT (r = −0.29, P < 0.05). There also was a negative correlation between the amount of glucose excreted in the urine and the increment in PG during 180-min OGTT (r = −0.30, P < 0.05) (Fig. 2A–C).

Figure 2.

A: Correlation between the increment in PG concentration during the 180-min OGTT and the calculated BCF index in patients treated with dapagliflozin for 4 months. The increment in PG was calculated as the difference between the mean PG value during 180-min OGTT and the baseline PG concentration. The negative correlation indicates that higher calculated BCF indices were associated with lower increments in PG during the 180-min OGTT. B: Correlation between the change in HbA_1c after 4 months of therapy and the calculated BCF index in patients treated with dapagliflozin. The change in HbA_1c was calculated as the difference between the mean value at the end and at the beginning of the study. The positive correlation indicates that higher calculated BCF indices were associated with higher declines in HbA_1c values after 4 months of treatment. C: Correlation between the increment in PG concentration and the amount of glucose excreted in the urine during the 180-min OGTT in patients treated with dapagliflozin for 4 months. The increment in PG was calculated as the difference between the mean value during the 180-min OGTT and the baseline PG concentration and the UGE rate. The negative correlation indicates that higher UGE rates were associated with lower increments in PG during the 180-min OGTT. AU, arbitrary unit.

In the group treated with exenatide for 4 months, there was a positive correlation between the decrease in HbA_1c and the insulin secretion index (r = 0.42, P < 0.05) and BCF index (r = 0.43, P < 0.05). There also was a negative correlation between the increment in PG during the 180-min OGTT and the BCF index (r = −0.34, P < 0.05) (Supplementary Fig. 1A–C).

In patients treated with dapagliflozin/exenatide for 4 months, there was a strong negative correlation between the increment in PG during 180-min OGTT and the cMI (r = −0.69, P < 0.01) and the insulin secretion index (r = −0.65, P < 0.01). There also was a positive correlation between the decrease in HbA_1c and the BCF index (r = 0.53, P < 0.01). In the dapagliflozin/exenatide group, there was a strong negative correlation between the amount of glucose excreted in the urine and the decrease in HbA_1c (r = −0.53, P < 0.01) and a very strong positive correlation between the amount of glucose excreted in the urine and the increment in PG during the 180-min OGTT (r = 0.82, P < 0.001) (Fig. 3A–E).

Figure 3.

A: Correlation between the increment in PG concentration during the 180-min OGTT and the cMI in patients treated with the combination dapagliflozin plus exenatide for 4 months. The increment in PG was calculated as the difference between the mean value during the 180-min OGTT and the baseline PG concentration in the study conducted after 4 months of therapy. The strong negative correlation indicates that higher calculated cMIs were associated with lower increments in PG during the 180-min OGTT. B: Correlation between the increment in PG concentration during the 180-min OGTT and the calculated insulin secretion index in patients treated with dapagliflozin plus exenatide for 4 months. The increment in PG was calculated as the difference between the mean value during the 180-min OGTT and the baseline PG concentration. The strong negative correlation indicates that higher calculated insulin secretion indices were associated with lower increments in PG during the 180-min OGTT. C: Correlation between the decrement in HbA_1c after 4 months of therapy and the calculated BCF index in patients treated with dapagliflozin plus exenatide. The change in HbA_1c was calculated as the difference between the mean value at the end and at the beginning of the study. The positive correlation indicates that higher calculated BCF indices were associated with greater declines in HbA_1c after 4 months of treatment. D: Correlation between the increment in PG concentration and the amount of glucose excreted in the urine during the 180-min OGTT in patients treated with dapagliflozin/exenatide for 4 months. The increment in PG was calculated as the difference between the mean value during the 180-min OGTT and the baseline PG concentration and the UGE rate during the 180-min period. The negative correlation indicates that higher UGE rates were associated with lower increments in PG during the 180-min OGTT. E: Correlation between the amount of glucose excreted in the urine during the 180-min OGTT and the decrease in HbA_1c in patients treated with dapagliflozin/exenatide for 4 months. The change in HbA_1c was calculated as the difference between the mean value at the end and at the beginning of the study and the UGE rate during the 180-min period. The strong positive correlation indicates that higher UGE rates were associated with greater declines in HbA_1c after 4 months of therapy.

Body Weight and HbA_1c

At 4 months, dapagliflozin reduced mean body weight by –2.3 ± 0.5 kg (from 85.2 ± 3.1 to 82.9 ± 3.2 kg), while exenatide decreased body weight by 3.3 ± 1.4 kg (from 93.2 ± 3.7 to 89.9 ± 3.5 kg). The decrease in body weight in patients treated with dapagliflozin/exenatide (Δ −7.6 kg, from 92.1 ± 2.5 to 84.5 ± 2.3 kg) was greater than with each drug alone (P < 0.01). At 4 months, HbA_1c decreased similarly with dapagliflozin (from 66.3 ± 2.4 to 59.1 ± 1.8 mmol/mol [8.3 ± 0.2 to 7.4 ± 0.1%]) and exenatide (from 63.9 ± 2.4 to 56.7 ± 2.2 mmol/mol [8.0 ± 0.2 to 7.1 ± 0.3%]). Combination therapy with dapagliflozin/exenatide produced an additive and greater HbA_1c reduction (from 67.1 ± 2.6 to 52.7 ± 1.5 mmol/mol [8.4 ± 0.3 to 6.6 ± 0.1%]) than exenatide alone and dapagliflozin alone (both P < 0.01).

Adverse Effects

Among patients treated with dapagliflozin, six experienced a genital mycotic infection, which responded to local therapy. Upon initiation of therapy, two patients had hypoglycemia, which resolved spontaneously; both were taking sulfonylurea. Four patients receiving exenatide reported nausea, and one had vomiting. In patients treated with dapagliflozin/exenatide, five experienced a genital mycotic infection, which responded to local therapy, and three had mild hypoglycemia; all were taking sulfonylurea, and two had nausea and vomiting.

Conclusions

In this study, we demonstrate that 4 months of combination treatment with dapagliflozin plus exenatide in patients with T2D led to a marked and sustained improvement in BCF and twofold greater decrement in HbA_1c (1.8% vs. 0.9%) compared with therapy with exenatide alone or dapagliflozin alone. After 4 months, the BCF index was 25% higher in patients with T2D who received dapagliflozin/exenatide therapy compared with those treated with exenatide alone and nearly threefold higher than in those treated with dapagliflozin alone. In all three treatment groups, BCF was markedly greater than in the placebo group at baseline (acute study) (Table 1). These are novel findings that have important clinical and pathophysiologic significance and make a strong argument for initiating combination therapy in patients with poorly controlled T2D. The combined effects on both tissue insulin sensitization (Matsuda index) and stimulation of insulin secretion (insulin secretion index) explain the robust 1.8% decrease in HbA_1c. In a previous study, we demonstrated that the BCF index increased by 73% and 112% after 24 h and 14 days of treatment with empagliflozin, respectively (24). In this same study, β-cell glucose sensitivity, determined with the hyperglycemic clamp, increased by 42% on day 1 and by 54% on day 14 of treatment. Thus, the current observations are consistent with these previous results and provide further evidence that longer-term therapy with SGLT2is, whether used alone or in combination with GLP-1RAs, enhance BCF in patients with T2D. Furthermore, the current results demonstrate that the improvement in tissue sensitivity to insulin that was observed with acute dapagliflozin/exenatide therapy is maintained for at least 4 months.

The improvement in whole-body tissue insulin sensitivity (as measured with the cMI) during treatment with dapagliflozin or exenatide, used alone or in combination, at month 4 most likely is explained by reduced glucotoxicity and weight loss. The Matsuda index reflects whole-body insulin sensitivity. Thus, to the extent that dapagliflozin stimulates glucose production (15,16,23), any improvement in muscle insulin sensitivity would be underestimated. The increment in the insulin secretion index reflects a direct insulin stimulatory effect of exenatide, a well-known therapeutic action of the GLP-1RA drug class (26). With regard to SGLT2is, however, the beneficial effects on insulin secretion most certainly results from attenuation of glucotoxicity (9–11,16). Since no SGLT2 receptors are present in islet cells (22), our observations suggest that maintenance of lower, near-normal PG levels are responsible for the improvement in BCF, in line with reports obtained in numerous experiments using animal models of diabetes (18–20). Therefore, the data obtained in this study further support the notion that the attenuation of glucotoxicity during dapagliflozin therapy is the most important mechanism behind the improvements in tissue insulin sensitivity and insulin secretion in patients with T2D.

The strong correlations observed between the BCF indices and the decrease in HbA_1c after 4 months of dapagliflozin monotherapy (Fig. 2B) and between the cMI (tissue insulin sensitivity) and lower increments in PG during the 180-min OGTT after 4 months of the combination therapy (Fig. 3A) support the role of reduced glucotoxicity. The strong negative correlation between the insulin secretion index and increment in PG during the 180-min OGTT (Fig. 3B) and the strong positive correlation between the BCF index and change in HbA_1c (Fig. 3C) also support the notion that treatment with dapagliflozin attenuates glucotoxicity and further enhances BCF when combined with exenatide. The strong correlation between the UGE rate and increments in PG during the 180-min OGTT (Fig. 3D) and with the decreases in HbA_1c (Fig. 3E) in patients with T2D treated with the combination therapy provides additional evidence for a continued and sustained glucose-lowering effect of SGLT2i agents.

There are some limitations to this study. Although participants were randomly assigned to the four different groups at baseline, during the 4-month follow-up period both participants and investigators were aware of the treatment. In an effort to reduce any potential bias, however, investigators were blinded and unaware of the subgroup distribution during the data analysis. The absence of a placebo control group for comparisons at 1 and 4 months represents a potential weakness, although results at 1 and 4 months were internally consistent and robust. Since we did not measure gastric emptying, we cannot evaluate the contribution of delayed gastric emptying to the improvement in OGTT during the acute or 1- and 4-month studies. However, this would not affect the measurement of insulin sensitivity or BCF. Approximately 30–40% of patients with T2D in each group were receiving long-term therapy with a sulfonylurea, and the sulfonylurea dose was not altered in any participant. Therefore, we do not believe that the sulfonylurea could have had any effect on the improvements in BCF observed with dapagliflozin or exenatide or the combination thereof. Furthermore, the effect of dapagliflozin, exenatide, and dapagliflozin/exenatide on insulin sensitivity and BCF in the subgroup of patients treated with metformin plus sulfonylurea (Supplementary Table 1) was similar to that in the entire group of patients, which included those treated with diet alone, metformin alone, and metformin plus sulfonylurea (Table 1). Another limitation relates to use of the cMI of tissue sensitivity to insulin. However, this index has been shown to be strongly correlated with the gold standard insulin clamp measure (30). Since the loss of glucose in the urine following an oral glucose load is substantial and varied considerably during dapagliflozin therapy in each patient, the degree of tissue insulin sensitivity may have been overestimated. To minimize this, we used a mathematical model designed to correct the disappearance rate of glucose from the extracellular space that is excreted in the urine and which is entirely independent of insulin (31). Finally, the OGTT did not reflect the normal physiological postprandial conditions, but the test is used for the diagnosis of diabetes and represents the test that is most commonly used in metabolic studies to evaluate overall glucose tolerance.

In summary, the current results demonstrate that dapagliflozin monotherapy for 4 months improved insulin sensitivity and BCF in patients with T2D. As expected, exenatide monotherapy increased insulin secretion and improved BCF, as well as augmented insulin sensitivity. Combination therapy with dapagliflozin plus exenatide was accompanied by an additional and prolonged increase in insulin secretion and BCF, and this prolonged improvement is a novel finding. Attenuation of glucotoxicity, associated with concomitant amelioration of insulin resistance, most likely is responsible for the increase in BCF index. These results demonstrate that combination therapy with a GLP-1RA plus SGLT2i provides superior glycemic control compared with either drug alone and leads to sustained and long-term preservation of BCF in patients with T2D.

This article contains supplementary material online at https://doi.org/10.2337/figshare.29310200.

Funding Statement

This study was supported by AstraZeneca Pharmaceuticals and by National Institutes of Health/National Institute of Diabetes and Digestive and Kidney Diseases grant R01-DK107680-05.