Sodium-Glucose Cotransport Inhibition With Dapagliflozin in Type 2 Diabetes

From December 2005 to September 2006, drug-naive type 2 diabetic patients, aged 18–79 years, with A1C ≥7% and ≤10%, were recruited at 98 clinical centers in the U.S., 24 in Canada, 8 in Mexico, and 3 in Puerto Rico. Inclusion criteria included fasting C-peptide >1.0 ng/ml, BMI ≤40 kg/m², and renal status as follows: glomerular filtration rate >60 ml/min per 1.73 m², serum creatinine <1.5 mg/dl (men)/<1.4 mg/dl (women), and urine microalbumin/creatinine ratio ≤300 mg/g.

This was a prospective, 12-week, randomized, parallel-group, double-blind, placebo-controlled study, with a 2-week diet/exercise placebo lead-in and 4-week follow-up (Fig. 1). Patients were randomly assigned equally to once-daily dapagliflozin (2.5, 5, 10, 20, or 50 mg), metformin XR (750 mg force-titrated at week 2 to 1,500 mg) (therapeutic benchmark), or placebo. Safety and efficacy were assessed at all study visits. Patients with fasting plasma glucose (FPG) >240 mg/dl at weeks 4 and 6, >220 mg/dl at week 8, or >200 mg/dl at week 10 were discontinued from the study and were eligible to receive additional antidiabetic agents. The study was conducted pursuant to the Declaration of Helsinki and was approved by institutional review boards/independent ethics committees at participating sites. Patients provided written informed consent before enrollment.

The primary objective was to compare mean A1C change from baseline for each dapagliflozin group versus placebo after 12 weeks. Secondary objectives were comparisons of dapagliflozin versus placebo for FPG change from baseline, dose-dependent trends in glycemic efficacy, proportion of patients achieving A1C <7%, and change in 24-h urinary glucose-to-creatinine ratio.

Study visits occurred at screening; days −14 and 1; weeks 1, 2, 4, 6, 8, 10, and 12; and follow-up weeks 14 and 16. Fasting blood and urine samples were collected after a minimum 10-h fast. During oral glucose tolerance testing, blood was drawn at 0, 30, 60, 120, and 180 min after an oral glucose challenge (75 g). Samples were centrally assessed (Quintiles Laboratories, Smyrna, GA). Glucose area under the curve (AUC) was calculated by trapezoidal methodology. Vital signs, brief physical examination, and adverse event assessment were performed at each visit. Complete physical examination and electrocardiograms were performed at lead-in and week 12.

Adverse events were summarized by preferred term (Medical Dictionary for Regulatory Activities [MedDRA], version 10). Safety topics of special interest were summarized by interest categories.

Fifty patients per treatment group provided 82% power to detect a mean 0.7% difference in A1C between dapagliflozin groups and placebo, assuming 1% SD. Comparisons between dapagliflozin and placebo were performed at the 0.012 level using Dunnett's adjustment so that overall type 1 error rate was controlled at 0.05 significance. Statistical analyses were performed on all randomly assigned and treated patients. Missing values were imputed by last observation carried forward (LOCF). Week 12 primary and secondary efficacy analyses for A1C, FPG, and 24-h urinary glucose-to-creatinine ratio were performed by ANCOVA with treatment group as the effect and baseline value as covariate. Linear trend tests were performed to assess dose-response relationships among dapagliflozin groups for A1C change from baseline after 12 weeks. Fisher's exact test was used to compare the proportion of subjects achieving A1C <7.0% between dapagliflozin groups and placebo.

A total of 389 patients were randomly assigned to receive dapagliflozin, metformin, or placebo (Fig. 1); 348 completed week 12, and 41 discontinued. The most common reason for discontinuation was withdrawal of consent (12 patients). Baseline demographics and disease characteristics were similar among all groups (Table 1).

At week 12, all dapagliflozin groups achieved significant reductions in mean A1C change from baseline versus placebo (Fig. 2,A and Table 2). Adjusted mean reductions ranged from −0.55 to −0.90% (dapagliflozin), −0.18% (placebo), and −0.73% (metformin). No log-linear dose-response relationship was demonstrated (P_trend = 0.41).

FPG reductions were apparent by week 1 in all dapagliflozin groups. By week 12, adjusted mean FPG reductions were −16 to −31 mg/dl (dapagliflozin), −6 mg/dl (placebo), and −18 mg/dl (metformin), demonstrating dose-related FPG decreases and statistically significant reductions in the 5- to 50-mg dapagliflozin groups versus placebo (Fig. 2,B). Adjusted mean postprandial plasma glucose (PPG) AUC reductions from baseline were −7,053 to −10,149 mg · min⁻¹ · dl⁻¹ (dapagliflozin), −3,182 mg · min⁻¹ · dl⁻¹ (placebo), and −5,891 mg · min⁻¹ · dl⁻¹ (metformin) (Fig. 2,C and Table 2).

Proportions of patients achieving A1C <7% at week 12 ranged from 40 to 59% (dapagliflozin), 32% (placebo), and 54% (metformin). The comparison versus placebo was statistically significant only for the 50-mg group (P < 0.01).

Urinary glucose excretion increased in all dapagliflozin groups. Adjusted mean changes in 24-h urinary glucose-to-creatinine ratios at week 12 were 32–65 g/g versus −0.2 g/g for placebo (P < 0.001 for each dapagliflozin group) (Fig. 2,D and Table 2). Total mean urinary glucose excreted per 24 h at week 12 ranged from 52 to 85 g with dapagliflozin (Table 2).

Total body weight reductions occurred in all groups (Fig. 3,A). Mean percent reductions at week 12 were −2.5 to −3.4% (dapagliflozin), −1.2% (placebo), and −1.7% (metformin) (Fig. 3,B and Table 2). More patients achieved >5% reductions with dapagliflozin (range 15.3–29.1%) than with placebo (7.7%); the proportion with metformin was 16.1%. Mean percent changes in waist circumference were −1.6 to −3.5% (dapagliflozin), −1.2% (placebo), and −2.2% (metformin).

Generally, adverse events were reported at similar frequencies across all groups (Table 2). No deaths or drug-related serious adverse events occurred. Hypoglycemic events were reported in 6–10% of dapagliflozin-treated patients with no dose relationship, in 4% of placebo-treated patients, and in 9% of metformin-treated patients (Table 2). There were no symptomatic hypoglycemic events with a fingerstick glucose ≤50 mg/dl.

Relevant adverse events were grouped into special interest categories. Events relating to each category were pooled (e.g., preferred terms “urinary tract infection” and “cystitis” were pooled as “infections of the urinary tract”) (Table 2). Infections of the urinary tract were seen in 5–12% of dapagliflozin-treated patients with no clear dose relationship versus 6% of placebo-treated patients and 9% of metformin-treated patients. Genital infections were seen in 2–7% of dapagliflozin-treated patients, 0% of placebo-treated patients, and 2% of metformin-treated patients.

Hypotensive events were seen in 0–2% of dapagliflozin-treated patients versus 2% of placebo-treated patients and 4% of metformin-treated patients. Decreased blood pressure was observed in all dapagliflozin groups (Table 2). Mean changes from baseline in supine systolic blood pressure (sBP) at week 12 ranged from −2.6 to −6.4 mmHg with no clear dose relationship. Similar changes occurred for standing sBP. Changes in diastolic blood pressure (dBP) and heart rate were small and inconsistent across dapagliflozin groups.

The diuretic effect of dapagliflozin was assessed by 24-h urine volume, hematocrit, and serum blood urea nitrogen (BUN) and creatinine (Table 2). Small dose-related increases in 24-h urine volumes (range 107–470 ml above baseline of 1.8–2.2 l) were demonstrated at week 12. Increases in hematocrit were also dose-related (range 1.5–2.9%). There were small changes from baseline in serum BUN and no change in serum creatinine at week 12 across dapagliflozin doses. Mean percent increases at week 12 in the BUN-to-creatinine ratio ranged from 10.4 to 18.3%, with no apparent dose relationship. Changes in urine volume, hematocrit, and BUN-to-creatinine ratio returned toward baseline during follow-up. There was no clinically meaningful change in estimated glomerular filtration rate (Modification of Diet in Renal Disease formula) (19) in any group. All groups experienced a small decrease in 24-h creatinine clearance.

A small increase of ∼0.1 mEq/l above the baseline mean (1.7 mEq/l) in serum magnesium and a larger relative decrease of ∼1.0 mg/dl below the baseline mean (5.5 mg/dl) in serum uric acid were observed, returning toward baseline after discontinuation of dapagliflozin. Serum phosphate increased in a dose-related manner for doses ≥5 mg (range −0.01 to +0.24 mg/dl from baseline of 3.6–3.8 mg/dl), although these changes were not statistically different from placebo (0.08 mg/dl) (Table 2). There were no clinically relevant mean changes from baseline in serum sodium, potassium, and calcium (Table 2).

With respect to bone metabolism, serum 1,25-dihydroxyvitamin D and 25-hydroxyvitamin D values were unchanged from baseline. Mean changes in the 24-h urinary calcium-to-creatinine ratio were similar to those with placebo. Small increases in mean parathyroid hormone concentrations (range 0.6–7.0 pg/ml above baseline of 31.1–35.0 pg/ml) were noted, which were generally greater than the 0.8 pg/ml increase for placebo. There was no clear treatment effect of dapagliflozin on fasting lipid parameters in this 12-week study.

Glucose reabsorption by the kidney is necessary from an evolutionary standpoint to retain calories but becomes detrimental in type 2 diabetes by contributing to perpetuation of hyperglycemia and caloric excess. Paradoxically, the glucose resorptive capacity of the kidney may increase in type 2 diabetes (20). Therefore, limiting renal glucose reabsorption through the inhibition of SGLT2 represents a new approach to treating hyperglycemia in type 2 diabetic patients.

This study provides evidence that inducing controlled glucosuria through selective SGLT2 inhibition improves hyperglycemia consistently over 12 weeks of treatment in type 2 diabetic patients. Dapagliflozin produced decreases in A1C, FPG, and PPG after 12 weeks, with reductions in FPG apparent by week 1. Changes in FPG were dose-related; however, there was little evidence of a dose response for either PPG or A1C. These observations apparently reflect an intrinsic property of dapagliflozin as an SGLT2 inhibitor. The impact of SGLT2 inhibition was relatively greater on PPG than on FPG, with renal glucose excretion acting as a relief valve to blunt postprandial hyperglycemia. Even the lowest dapagliflozin dose (2.5 mg) produced a near-maximal effect on PPG, consistent with reductions observed in a clinical ward study (16). In contrast, the effect on FPG, measured at the trough drug concentration, was dose-ordered and corresponded to expected residual trough pharmacodynamic activity (16).

Dapagliflozin exhibited a diuretic effect, with small dose-dependent increases in urine volume equivalent to ∼0.3–1.5 voids/day, small increases in BUN, and small dose-dependent increases in hematocrit. No clinical safety signals for dehydration were observed. The observed decrease in sBP was consistent with a diuretic action. The relevance of this diuresis in type 2 diabetic patients, who often require diuretics for controlling hypertension (21), warrants further investigation. Although no effect on renal function was observed, longer-term studies and exploratory renal biomarker assessments are being undertaken.

Dapagliflozin-treated patients experienced total body weight reductions. Veterinary literature suggests that chronic administration of phlorizin in lactating cows induces lipolysis (22), and dapagliflozin in obese rats induces reduced adiposity (23). During treatment, all doses induced progressive weight reductions, consistent with steady caloric loss through glucosuria. Weight loss was more pronounced during week 1 with dapagliflozin, particularly at higher doses. This observation, coupled with a rapid partial rebound in weight after discontinuation of higher doses, suggests that diuresis may contribute to some weight loss. Overall, it appears likely that acute weight reduction during week 1 represents fluid loss, which may also result in lower sBP, whereas continued gradual weight loss represents decreased fat mass. Longer-term clinical and body composition studies will help to establish the relative contribution of diuresis versus adiposity reduction to total weight loss.

Daily dapagliflozin was well tolerated with no major difference in adverse events across treatment groups. The hypoglycemia experience supports the potential for dapagliflozin to achieve meaningful glycemic efficacy with relatively low hypoglycemic risk. The number of reported urinary tract infections was similar among dapagliflozin, metformin, and placebo groups and is consistent with rates reported in type 2 diabetic patients (24). The incidence of genital infections was higher with dapagliflozin versus placebo, especially at higher doses, but without statistical significance for comparison. Of note is the lower rate of genital infections reported for placebo group patients than previously reported for type 2 diabetic patients (25).

Dapagliflozin increased serum phosphate at higher doses, and all arms including placebo and metformin demonstrated increased serum parathyroid hormone. Additional data are needed to understand the long-term effects of chronic glucosuria and dapagliflozin treatment on skeletal metabolism.

This study demonstrated the clinical efficacy of inhibiting renal glucose reabsorption with dapagliflozin in type 2 diabetic patients and relative safety across numerous doses. Our results suggest that dapagliflozin, as the first in a new class of SGLT inhibitors, can improve glycemic and weight status of type 2 diabetic patients. Although we evaluated monotherapy, the insulin-independent mechanism of dapagliflozin may complement other type 2 diabetes agents that act through insulin signaling pathways and thus enhance combination therapy. Although human genetic case reports are reassuring, the chronic effects of pharmacologically induced glucosuria are unknown and require long-term assessment. On the basis of evidence to date, further clinical study of dapagliflozin is warranted to develop a more definitive benefit/risk profile for this novel therapeutic agent.

This study was supported by Bristol-Myers Squibb and AstraZeneca. Editorial support was funded by Bristol-Myers Squibb.

V.W. has received lecture or consulting fees from Bristol-Myers Squibb. E.M. received a grant from Bristol-Myers Squibb to conduct clinical research related to this article. No other potential conflicts of interest relevant to this article were reported.

Parts of this study were presented in abstract form at the 68th Scientific Sessions of the American Diabetes Association, San Francisco, California, 6–10 June 2008 and the 44th annual meeting of the European Association for the Study of Diabetes, Rome, Italy, 7–11 July 2008.

We thank Sean Gregory, Health Science Communications, for writing and editorial assistance.