Effect of Dapagliflozin on Renal and Hepatic Glucose Kinetics in T2D and NGT Subjects Free

Increased endogenous glucose production (EGP) in the postabsorptive state is a characteristic feature of type 2 diabetes (T2D) (1). The liver primarily is responsible for the increase in EGP, although some studies have suggested that the kidneys also contribute (2,3). As reviewed by Cherrington et al. (4), following an overnight fast, the kidneys may contribute 10–20% to total glucose production in normal glucose tolerant (NGT) subjects. In contrast, other studies have found that the kidney contribution to EGP in the postabsorptive state is minor and not clinically significant (5).

In healthy subjects, the sodium–glucose cotransporter-2 (SGLT2) is responsible for most of the of renal glucose reabsorption (6). In patients with diabetes, the filtered load of glucose is increased, and glucose, along with sodium, reabsorption in the proximal tubule is increased (7,8), resulting in decreased sodium delivery to the macula densa cells of the juxtaglomerular apparatus. This leads to the local release of angiotensin, efferent arteriolar vasoconstriction, increased intraglomerular pressure, and renal damage (8). However, animal (9,10) and human (11,12) studies have provided conflicting evidence concerning whether SGLT2 expression is increased in the diabetic state. SGLT2 inhibitors (SGLT2i) were initially developed for their antihyperglycemic effect (13,14), and their use in the treatment of T2D patients has rapidly increased because of their cardio- and renoprotective effects (15,16). SGLT2i reduce fasting plasma glucose (FPG) by ∼30 mg/dL and HbA_1c by ∼0.5–1.0% (17). We and others consistently have shown that this class of drugs increases EGP (18–22). The increase in EGP starts at 30 min after a single dose of dapagliflozin (DAPA) (19,21) and persists for at least 2–4 weeks (21,22). This stimulatory effect of SGLT2i on EGP significantly offsets the beneficial glucose-lowering effect of SGLT2i (17).

Several studies have examined potential mechanisms responsible for the increase in EGP with SGLT2i. Although glucagon initially was postulated to contribute to the SGLT2i-induced increase in EGP, subsequent studies have ruled out glucagon (23–25), as well as insulin and glucose, as contributory factors (23). A recent study showed that the increase in EGP following DAPA in T2D patients primarily was due to an increase in gluconeogenesis (26). However, the site—liver versus kidney—of the increase in EGP following SGLT2i has yet to be identified. Unlike the liver, the amount of glucose stored as glycogen in the kidney is small, and renal glucose production (RGP) principally is accounted for by gluconeogenesis (27). While the liver is sensitive to glucagon, kidneys do not respond to glucagon but do respond to insulin’s suppressive effect on glucose production (28–30).

SGLT2 primarily is located in the early S1 segment of the proximal tubule, and the key enzymes involved in gluconeogenesis are expressed in S1–S3 segments of the proximal tubule (29,31). In animal models, renal gluconeogenesis is increased in diabetes and insulin-resistant states (32,33). In diabetic mice, as well as in human proximal tubular (HK2) cell lines, DAPA increases renal gluconeogenesis by enhancing expression of phosphoenolpyruvate carboxykinase and glucose-6-phosphatse (34). Further, SGLT2i increases gluconeogenic gene expression in kidney in mice, while no effect was observed in liver (35). However, EGP was not measured in these studies, and thus, the relevance of these studies to the SGLT2i-induced increase in EGP in humans is unclear. Further, the basal rate of EGP in mice and rats is six- to eightfold higher than in humans. In the current study, we examined the role of the kidneys in the increase in EGP following acute administration of the SGLT2i DAPA in T2D and NGT subjects using renal vein catheterization in combination with tracer methodology.

The study was done at UT Health and Audie L. Murphy Memorial Veterans’ Hospital, San Antonio, TX, and was approved by UT Health Institutional Review Board. The study is registered at ClinicalTrials.gov (NCT02981966).

The study participants were 32 men, 20 with T2D and 12 with NGT. Their clinical, anthropometric, and laboratory data are summarized in Table 1. Except for diabetes, all subjects were in general good health based on medical history, physical examination, screening chemistry/hematology tests, urinalysis, and electrocardiogram. Weight was constant (± 3 pounds) in all subjects for at least 3 months before the study, and no subject participated in an excessively heavy exercise program. T2D subjects with evidence of more than background retinopathy, urinary albumin-to-creatinine ratio >30 mg/g, or estimated glomerular filtration rate <90 mL/min ⋅ 1.73 m² were excluded. Of the subjects with diabetes, 16 were on metformin, 2 were on metformin/glipizide, 1 was on metformin/sitagliptin, and 4 were treated with diet. NGT subjects received a 2-h, 75-g oral glucose tolerance test to confirm NGT according to American Diabetes Association criteria. T2D and NGT subjects were randomized 2:1 to receive DAPA or placebo.

Within 7 days of the screening visit, subjects reported to Bartter Research Unit (BRU) at 6 a.m. after a 10-h overnight fast. At 7 a.m. (time = −180 min), a catheter was placed into an antecubital vein and a prime (40 μCi × FPG/100)-continuous (0.4 μCi/min) [3-³H]glucose infusion was started and continued until 2 p.m. (time = +240 min). At 8 a.m. (time = −120 min) subjects were transferred from BRU to the cardiac catheterization laboratory where a catheter was inserted into the renal vein via jugular vein under fluoroscopy. A second catheter was inserted into the radial artery for a blood draw. After completion of these procedures, subjects were transferred back to BRU. At 9 a.m. (time = −60 min), a prime (8 mg/kg)-continuous (12 mg/min) infusion of para-aminohippuric acid (PAH) was started for determination of renal blood flow (RBF). After 2.5 h (9:30 a.m.) (time = −30 min) of tracer equilibration in T2D, blood samples were drawn from the renal vein and arterial catheters at −30, −20, −10, −5, and 0 min for measurement of plasma glucose concentration, tritiated glucose radioactivity, and PAH concentration. At time 0 (10 a.m.), subjects ingested DAPA (10 mg) or placebo, and blood samples were obtained from radial arterial and renal vein catheters every 20 min from 10 a.m. to 2 p.m. (time = +240 min). Plasma glucose, insulin, C-peptide, glucagon, cortisol, growth hormone, catecholamine, and PAH concentrations, hematocrit, and [3-³H]glucose radioactivity were measured in blood samples drawn from the arterial catheter. Urine was collected from 10 a.m. to 2 p.m. (time = 0 to +240 min). Urine volume and urinary glucose and PAH concentrations were measured to quantitate urinary glucose excretion and renal plasma flow (RPF; PAH clearance).

Plasma glucose was measured in duplicate at the bedside by Analox Glucose Analyzer (Analox Instruments Ltd, Amblecote, U.K.). Plasma PAH concentration was determined by colorimetric method (36). Plasma insulin and C-peptide were measured by radioimmunoassay (Linco Research, St. Louis, MO). Plasma glucagon was measured by radioimmunoassay (MilliporeSigma, Burlington, MA) with inter- and intraassay coefficient of variation of 4.85% and 11.7%, respectively. Samples for plasma insulin, C-peptide, and glucagon assay were collected in separate aliquots to prevent freeze/thawing. Plasma [3-³H]glucose radioactivity was measured in duplicate and dehydrated Somogyi precipitates. Plasma epinephrine was measured by ELISA (ALPCO, Salem, NH).

Data are presented as mean ± SEM. Variables before and after DAPA were compared with the paired t test. Between-group comparisons were performed with two-way ANOVA with Bonferroni correction for multiple comparisons. P value <0.05 was considered significant.

Glucose specific activity was determined using the following procedure: Plasma was deproteinized with Ba(OH)₂ and ZnSO₄, supernatant was evaporated, and the dried glucose residue was resuspended and counted in a liquid scintillation counter. The [3-³H]glucose-specific activity (disintegrations per minute [dpm]/mg) was calculated by dividing the plasma dpm by plasma glucose concentration (mg/dL). The infusate was diluted 1:100 and 1:1,000 to assess the tracer infusion rate. Plasma samples for [3-³H]glucose radioactivity were determined in duplicate. Under steady-state postabsorptive conditions, the basal rate of endogenous R_a equals the R_d and is calculated by dividing the [3-³H]glucose infusion rate (dpm/min) by steady-state plasma-tritiated glucose-specific activity (dpm/mg). Following DAPA, nonsteady-state conditions for [3-³H]glucose-specific activity prevail, and R_a was calculated with the Steele equation, assuming a pool fraction 0.65 for glucose and distribution volume of 200 mL/kg for glucose.

RGP was calculated as previously described (28). RPF in mL/min was calculated as the clearance of PAH, and RBF was calculated as RPF/(1−hematocrit). Net renal glucose balance = RBF × (artery [Art]_glu − renal vein [RV]_glu). Renal glucose fractional extraction (FE) was calculated from tritiated glucose radioactivity as ([Art_dpm/mL − RV_dpm/mL]/Art_dpm/mL) and renal glucose uptake = RBF × FE × [Glu]_art. As such, renal glucose uptake represents the amount of glucose taken up by the kidney and excreted in the urine plus the amount of glucose taken up and metabolized by renal tissues. The net renal glucose balance = renal glucose uptake – RGP, from which RGP was calculated as (RBF × FE × [Glu]_Art) – (RBF × [Glu_{Art − RV}]). Hepatic glucose production (HGP) was calculated as R_a(total) − RGP. The effect of DAPA versus placebo on RGP and HGP was calculated as the difference in each parameter (RGP and HGP) at each time point from 0 to 240 min minus the mean value before drug administration. Renal glucose excretion was subtracted from the total renal glucose uptake to calculate renal tissue glucose uptake.

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

There was no significant difference between arterial versus renal vein plasma glucose concentration at baseline in T2D or NGT subjects (Fig. 1). Arterial [3-³H]glucose-specific activity was constant over the 30-min baseline period (Supplementary Fig. 1E), indicating that steady state was achieved in all subjects.

Following both DAPA and placebo (Fig. 1), plasma glucose concentration declined significantly over the 4-h study (reflecting the prolonged overnight fast), but the decline was greater in subjects receiving DAPA. After 4 h, plasma glucose in T2D declined by 23% with DAPA versus 13% with placebo (P < 0.05). In NGT subjects, plasma glucose declined slightly but not significantly by 8.0% with DAPA versus 6.9% with placebo. Following DAPA, renal vein glucose concentration was significantly lower than the arterial glucose concentration in both T2D and NGT (Fig. 1).

The basal rate of EGP was similar between DAPA and placebo groups in both T2D (2.01 ± 0.12 vs. 2.05 ± 0.09 mg/kg ⋅ min) and NGT (1.73 ± 0.08 vs. 2.01 ± 0.1 mg/kg ⋅ min) subjects. EGP declined progressively following placebo in T2D (2.05 ± 0.09 to 1.29 ± 0.09 mg/kg ⋅ min, P < 0.05) and NGT (2.01 ± 0.1 to 1.95 ± 0.15 mg/kg ⋅ min, P = not significant [NS]). However, following DAPA, EGP during the last 40 min of the study (+200 to +240 min) was higher than baseline in both NGT (2.13 ± 0.09 vs. 1.73 ± 0.08 mg/kg ⋅ min, P < 0.05) and T2D (2.27 ± 0.1 vs. 2.01 ± 0.1 mg/kg ⋅ min, P < 0.05) (Fig. 2). The increment in EGP following DAPA versus placebo in both T2D and NGT was similar to the amount excreted in urine (Fig. 3).

In T2D subjects, the FE of [3-³H]glucose at baseline was 0.012 ± 0.007 and 0.020 ± 0.008% (P = NS) in placebo and DAPA groups, respectively. In NGT subjects, the FE at baseline was 0.020 ± 0.007 and 0.023 ± 0.007% (P = NS) in placebo and DAPA groups, respectively. The FE increased following DAPA in T2D (3.02 ± 0.68 vs. 0.02 ± 0.008, P = 0.005) and in NGT (1.8 ± 1.2 vs. 0.02 ± 0.007, P < 0.01; P = NS, T2D vs. NGT). Following placebo in T2D (1.81 ± 0.33 vs. 0.012 ± 0.007%, P = NS) and NGT (0.02 ± 0.007 vs. 1.61 ± 1.5%, P = NS) subjects, the FE did not change significantly. RPF in T2D did not change following DAPA (591 ± 62 vs. 558 ± 61 mL/min) or placebo (606 ± 130 vs. 535 ± 111 mL/min). Similarly, no change in RPF was observed in NGT subjects following DAPA (631 ± 75 vs. 654 ± 90 mL/min) or placebo (603 ± 101 vs. 682 ± 115 mL/min). Renal glucose uptake increased following DAPA in T2D (0.051 ± 0.02 to 0.320 ± 0.06 mg/kg ⋅ min, P = 0.001) and NGT (0.121 ± 0.04 to 0.267 ± 0.06 mg/kg ⋅ min, P = 0.03) following DAPA. No change in renal glucose uptake was observed following placebo in T2D or NGT.

Following DAPA, glucose excretion increased in both T2D (0.004 ± 0.001 to 0.264 ± 0.04 mg/kg/min) and NGT (0.003 ± 0.001 to 0.238 ± 0.07 mg/kg ⋅ min) compared with placebo (Fig. 3). Renal tissue glucose uptake slightly but not significantly increased following DAPA in T2D (0.05 ± 0.02 to 0.118 ± 0.04, mg/kg ⋅ min, P = NS) and NGT (0.121 ± 0.04 to 0.148 ± 0.06 mg/kg ⋅ min, P = NS). There was no change in renal tissue glucose uptake following placebo in T2D or NGT subjects.

RGP at baseline was 0.09 ± 0.03 mg/kg ⋅ min in T2D and 0.03 ± 0.09 mg/kg ⋅ min in NGT subjects (P = NS, T2D vs. NGT), and this was <5% of basal EGP in both groups (Fig. 2). RGP increased slightly but not significantly following DAPA in both T2D (0.09 ± 0.04 vs. 0.17 ± 0.05 mg/kg ⋅ min, P = NS) and NGT (0.03 ± 0.02 vs. 0.16 ± 0.02 mg/kg ⋅ min, P = NS). Following placebo, RGP did not change in T2D (0.01 ± 0.03 vs. 0.003 ± 0.05 mg/kg ⋅ min, P = NS) or NGT (0.09 ± 0.1 vs. 0.14 ± 0.12 mg/kg ⋅ min, P = NS). Following DAPA, hepatic glucose production (EGP minus RGP) increased significantly in T2D by 0.22 ± 0.09 mg/kg ⋅ min (P < 0.01) and in NGT by 0.34 ± 0.08 mg/kg ⋅ min (P < 0.05).

There was no difference in fasting insulin concentration between the subjects with T2D and NGT (17 ± 2 vs. 15 ± 4 μU/mL, P = NS). Plasma insulin (Supplementary Fig. 1A and B) and C-peptide concentrations (Supplementary Fig. 1C) did not change following placebo (Supplementary Fig. 1C); however, both declined significantly following DAPA (P < 0.05) in T2D subjects (Supplementary Fig. 1A and C).

At baseline, there was no difference in fasting plasma glucagon concentration between T2D and NGT (38 ± 5 vs. 39 ± 5 pg/mL, P = NS). After DAPA or placebo, plasma glucagon concentration did not change in T2D (Supplementary Fig. 1D) and did not change significantly in NGT. There were no significant changes in plasma epinephrine, norepinephrine, or cortisol concentrations following DAPA or placebo (data not shown).

The main finding of the current study is that the DAPA-induced increase in EGP results from an increase in hepatic glucose production. Since renal gluconeogenic enzymes are expressed in the cortex close to the proximal convoluted tubules where SGLT2 is expressed, there is a possibility that SGLT2i might influence RGP. However, we failed to observe a significant increase in RGP following acute DAPA in T2D or NGT subjects. Although DAPA increased net renal glucose uptake, this was entirely accounted for by an increase in urinary glucose excretion; renal tissue glucose uptake was not changed following DAPA.

SGLT-2i consistently have been shown to increase EGP, and this increase in EGP offsets by ∼50% the increase in urinary glucose excretion (37). The mechanism(s) by which SGLT2i increase EGP is not clear. A recent study by Wolf et al. (26) showed that the increase in EGP following DAPA was secondary to an increase in gluconeogenesis, not glycogenolysis. The tissue responsible for the increase in gluconeogenesis/EGP was not examined in their study. Gluconeogenesis was not measured in the current study, but the DAPA-induced increase in EGP was most likely due to an increase in hepatic gluconeogenesis, based on the results by Wolf et al. (26). Several studies in animal models have suggested that SGLT2i increase renal, but not hepatic, gluconeogenesis (34,35). However, these studies did not measure the total-body EGP. In healthy subjects, hypoglycemia induced by hyperinsulinemia has been shown to increase in renal gluconeogenesis (36). However, the plasma glucose concentration in the current study changed minimally in NGT subjects and always was >90 mg/dL, while in T2D subjects it remained >100 mg/dL throughout the 4-h study period. Although the renal medulla does contain glycogen, quantitatively the amount is small, and those portions of the kidney where glycogen is present lack sufficient renal glucose-6-phosphatase activity (38). Further, the liver, not the kidney, was shown to be the site of DAPA-induced stimulation of EGP.

In studies using the pancreatic clamp technique, we have shown that changes in plasma glucagon, insulin, and glucagon cannot explain the SGLT2i-induced increase in EGP (23,25). Therefore, we do not believe that the modest decline in plasma insulin concentration could play role in the increase in EGP following DAPA. Since plasma glucagon, epinephrine, norepinephrine, and cortisol did not change significantly, these hormones also are unlikely to have contributed to the rise in EGP.

Another important observation in our study is that during the basal state, glucose production by the kidney could not be demonstrated and that there was no difference in RGP between T2D and NGT subjects. The contribution of the kidneys to hyperglycemia in T2D is controversial. While some studies suggest that increased RGP contributes to postprandial as well to postabsorptive hyperglycemia (28,39–41), other studies have demonstrated that the contribution of the kidneys is minor (5,42). In the current study during the basal state, RGP explained <5% of total EGP. One possible explanation for these discrepant observations (5,28,39–42) is that the T2D patients in the current study were in reasonably good glycemic control (mean HbA_1c ∼6.6% and mean FPG = 138 mg/dL), precluding the inclusion of patients with diabetes with an elevated basal rate of RGP.

In summary, the present results demonstrate that a single dose of DAPA acutely increases EGP and that the increase in EGP is due to an increase in HGP.

Funding. This work was supported by National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases grant DK-24092-36 (to R.A.D.).

Duality of Interest. This study received support from AstraZeneca. D.T.’s salary is supported by the South Texas Veterans Affairs Health Care System, Audie L. Murphy Memorial Veterans’ Hospital. At the time that the study was completed, X.C. was a faculty member in the Division of Diabetes at University of Texas Health Science Center at San Antonio, and currently is an employee of Eli Lilly. C.S.-H. is a member of the speaker bureau of Novo Nordisk and member of the advisory board for Bayer. E.C. received research funds from AstraZeneca. R.A.D. receives grant support from AstraZeneca, is a member of the advisory boards of AstraZeneca, Intarcia, and Novo Nordisk, and is a member of the speakers bureau of AstraZeneca. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. X.C., D.T., and R.C. performed the study. X.C., D.T., E.C., and R.A.D. analyzed the data and wrote the manuscript. A.H.-D., M.S., and C.S.-H. reviewed and edited the manuscript. R.A.D. designed the study. All authors read and approved the manuscript for submission. R.A.D. is the guarantor of this work and, as such, had full access to all the data in study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 82nd Scientific Session of the American Diabetes Association, New Orleans, LA, 3–7 June 2022.