Exploring the Potential of the SGLT2 Inhibitor Dapagliflozin in Type 1 Diabetes: A Randomized, Double-Blind, Placebo-Controlled Pilot Study

Insulin therapy for type 1 diabetes presents many challenges to patients and physicians. On a day-to-day basis, marked daily glucose fluctuations and frequent hypoglycemia limit insulin dosing (1). In the long-term, increasing the insulin dose to achieve and/or maintain glycemic control is associated with weight gain (2), which may contribute to potential cardiovascular complications (3).

Dapagliflozin, a highly selective, orally active inhibitor of sodium–glucose cotransporter 2 (SGLT2), reduces hyperglycemia by inhibiting renal glucose reabsorption independently of insulin. In patients with insulin-treated type 2 diabetes, dapagliflozin therapy has been shown to improve glycemic control with attenuation of insulin-associated weight gain and without increasing rates of major hypoglycemia, and also with consistent reductions of systolic blood pressure (4,5). This efficacy profile suggests that dapagliflozin may offer a potential therapeutic advance as an adjunct to insulin therapy in patients with type 1 diabetes.

Since the introduction of insulin in 1922, only one new injectable therapeutic class (amylin analog) has been approved for the treatment of type 1 diabetes, but with a black box warning for hypoglycemia. In addition, there are no approved oral therapies for type 1 diabetes, therefore, an unmet need exists for novel therapies in type 1 diabetes. The kidney might become a therapeutic target with the advent of insulin-independent selective orally active glucosuric agents, which could potentially influence the maladaptive reabsorption of glucose, which may perpetuate hyperglycemia, that has been observed in these patients (6).

This pilot study sought to answer the following research questions. First, is the short-term safety and tolerability of dapagliflozin acceptable in patients with type 1 diabetes? Second, are the pharmacokinetic profile and pharmacodynamic activity of dapagliflozin in patients with type 1 diabetes comparable to those previously described in other populations? Third, do the effects of dapagliflozin on efficacy parameters after 7 days of treatment in patients with type 1 diabetes suggest that it may have the potential to be efficacious in these patients?

This was a 2-week randomized, double-blind, parallel-group, placebo-controlled, exploratory phase 2a pilot study, evaluating the safety, tolerability, pharmacokinetics, and pharmacodynamics of dapagliflozin in patients with type 1 diabetes inadequately controlled with insulin; it was conducted from 3 February 2012 to 17 October 2012. The study complied with the Declaration of Helsinki and the International Conference on Harmonisation/Good Clinical Practice Guidelines, and was approved by institutional review boards and independent ethics committees for the participating centers (clinical trial reg. no. NCT01498185). All participants provided informed consent.

Inclusion criteria included the following: age range 18–65 years; type 1 diabetes treated with insulin monotherapy, either by multiple daily injections consisting of long-acting (basal) plus short-acting prandial (bolus) insulin or continuous subcutaneous insulin infusion pump, for ≥12 months; and HbA_1c ≥7.0% and ≤10.0% at screening. Exclusion criteria included the following: a history of type 2 diabetes; treatment with oral antihyperglycemic agents within 12 months of study commencement or diabetic ketoacidosis; hospitalization for poor glycemic control; and frequent episodes of hypoglycemia within the previous 24 weeks (see Supplementary Data for detailed lists of the inclusion and exclusion criteria).

Three days prior to randomization, participants were admitted to an inpatient facility for collection of baseline data and confirmation of eligibility. They were subsequently randomly assigned by a computerized interactive voice response system to receive daily doses of dapagliflozin (1, 2.5, 5, or 10 mg) or matching doses of placebo in a 1:1:1:1:1 ratio in addition to their insulin regimen. Randomization was stratified by BMI category (≤23 or >23 kg/m²) and insulin administration method (i.e., multiple daily injection or continuous subcutaneous insulin infusion). The investigator, sponsor, and patients remained blinded to treatment allocation throughout the double-blind treatment and the follow-up periods.

Inpatient double-blind treatment continued for a period of 7 days, during which data were collected and analyzed. During inpatient treatment, standardized diets and guidance on insulin dose adjustment were given. Insulin dose was not proactively reduced at study drug initiation. However, patients and investigators were advised to adjust insulin dosing as needed to avoid hypoglycemia and ensure patient safety. Outpatient double-blind treatment continued for a further 7 days, and, on cessation of double-blind treatment, follow-up data were collected over another 7 days.

In order to establish the short-term safety and tolerability of dapagliflozin in combination with insulin, data on safety and tolerability were analyzed after a total of 14 days of exposure. Key parameters included general adverse events (AEs), treatment discontinuations, hypoglycemic events, and genital and urinary tract infections, as well other parameters of interest, such as total daily fluid intake, total daily urine output, body weight, blood pressure, and daily urine ketones.

In order to establish whether the pharmacokinetic profile and pharmacodynamic effect on urinary glucose excretion was similar to that already well described in other populations, the following analyses were conducted. Pharmacokinetic parameters for dapagliflozin and its major inactive metabolite dapagliflozin 3-O-glucuronide (D3OG) analyzed were as follows: maximum observed plasma concentration (C_max); time of maximum observed plasma concentration (T_max); area under the concentration-time curve in one dosing interval (AUCτ); and ratio of metabolite to parent AUC (corrected for molecular weight). The pharmacodynamic parameter analyzed was the mean change from baseline in 24-h urine glucose at day 7.

To explore the potential of dapagliflozin as an adjunctive treatment for type 1 diabetes and to guide future clinical study design considerations, the efficacy parameters analyzed were the mean changes from baseline at day 7 in the following: mean glucose based upon 7-point central laboratory glucose measurements obtained before and after each meal and at bedtime; fasting plasma glucose (FPG) ; daily average glucose (DAG); SD of DAG; and mean amplitude of glucose excursion (MAGE). The latter three parameters were derived from 24-h continuous glucose monitoring (CGM) using a Dexcom SEVEN PLUS device. The CGM system, which remained blinded to the patient and site staff during the recording, recorded data approximately every 5 min and downloaded the data wirelessly into a data file. Site staff and patients were fully trained to operate and calibrate the CGM device according to the manufacturer’s instructions. Also analyzed were the mean percent changes from baseline at day 7 in daily basal, bolus, and total insulin doses.

In this exploratory pilot study, mean changes from baseline and two-sided 95% CIs were calculated. No statistical hypothesis testing was performed, and no P values were calculated. For the analysis of pharmacokinetic parameters, geometric means and coefficients of variation were summarized for C_max, AUCτ, and the ratio of metabolite to parent AUC. Medians and ranges were summarized for T_max. For the pharmacodynamic analysis of 24-h urine glucose, no dapagliflozin versus placebo mean difference calculations were planned. For the efficacy parameter, 7-point mean glucose, dapagliflozin versus placebo mean differences, and 95% CIs were prespecified. For the remaining efficacy parameters, no dapagliflozin versus placebo mean difference calculations were planned. For the analysis of the percent changes in insulin dose, values were log transformed prior to analysis; resulting geometric mean estimates and 95% CIs on the log scale were then back-transformed using the formula 100 × (e^Estimate − 1) to obtain the mean percent changes from baseline and their associated 95% CIs.

Overall, the baseline mean ± SD HbA_1c was 8.46 ± 0.81% and the mean age was 35.3 ± 12.9 years. Baseline urinary glucose, FPG, and HbA_1c trended higher and were more variable in the placebo group compared with the dapagliflozin groups (Table 1). Of the 70 patients randomized, 62 (88.6%) completed the study (Supplementary Fig. 1).

AEs are summarized in Table 2. One serious AE (gastroparesis) not related to treatment was reported in the dapagliflozin 5 mg group in a patient with a history of gastroparesis, which led to discontinuation of treatment. A high percentage of patients in all treatment groups experienced at least one episode of hypoglycemia with no apparent relationship to dapagliflozin dose (Table 2); one event on day 6 in the dapagliflozin 10 mg group was major (defined as requiring third-party assistance and a plasma glucose value <3.0 mmol/L), which was related to the noncompliance of the patient to reduce insulin dosing as instructed by the investigator. This event led to discontinuation of treatment on day 8 because of the investigator’s concerns regarding the patient’s compliance with insulin dosing instructions and nutrition during the outpatient period. One genitourinary infection occurred in each of the placebo and dapagliflozin 1, 2.5, and 5 mg groups (Table 2).

There were no apparent effects of dapagliflozin on fluid intake, body weight, or blood pressure in this relatively normal weight and normotensive population (Supplementary Table 1). Total daily urine output tended to show numeric increases in all groups, but this increase was larger in the placebo group compared with the dapagliflozin groups (Supplementary Table 1). Positive results of urine ketone tests were observed in all groups at baseline and appeared to decrease in the placebo and dapagliflozin 1 and 2.5 mg groups during study drug administration. No instance of diabetic ketoacidosis occurred.

Steady-state pharmacokinetic results showed that both C_max and AUCτ appeared to be dose proportional across all dose groups for both dapagliflozin and its metabolite D3OG. Dapagliflozin was rapidly absorbed after oral administration, with median T_max ranging from 0.50 to 1.00 h. Similarly, the median T_max values for D3OG ranged from 1.00 to 2.00 h. The ratio of metabolite to parent AUC was consistent across all dose groups, ranging from 0.66 to 0.76 (Supplementary Table 2).

Consistent with its mechanism of action, a dose-dependent increase in 24-h urine glucose was observed with dapagliflozin (Fig. 1A), but this was not statistically tested. In contrast, the 24-h urine glucose decreased in the placebo group.

Glycemic efficacy parameters at day 7 are summarized in Fig. 1B–F. For mean glucose measurements derived from 7-point glucose monitoring (Fig. 1B and Supplementary Table 3), no differences between the placebo and dapagliflozin groups were observed. For all other glycemic efficacy parameters (Fig. 1C–F), the 95% CIs of the mean for all dapagliflozin doses overlapped those of placebo. Among the dapagliflozin groups, the following dose-related trends were suggested: FPG, DAG, SD of DAG, and MAGE showed greater numeric reductions at day 7 with the dapagliflozin 5 and 10 mg doses versus the lower dapagliflozin doses (Fig. 1C–F). In addition, the dapagliflozin 5 and 10 mg doses were associated with reductions in the mean percent change from baseline at day 7 in total insulin dose of −19.3% (95% CI −30.1 to −6.8) and −16.2% (95% CI −29.4 to −0.5), respectively; however, it should again be noted that the 95% CIs for all dapagliflozin doses overlapped those for placebo (Fig. 2).

No unexpected short-term safety concerns were identified with dapagliflozin therapy in patients with type 1 diabetes. A high percentage of patients experienced at least one hypoglycemic event across all treatment groups, with the number of events varying substantially between groups. Despite this variation, hypoglycemia rates appeared to show no relationship to dapagliflozin dose, and it was reassuring that there was no evidence of a dramatic increase in major hypoglycemic events in patients who received dapagliflozin in combination with insulin; of 57 dapagliflozin-treated patients, only one patient experienced a major hypoglycemia event. It should be noted that insulin doses were not proactively reduced in this study; however, on initiation of dapagliflozin therapy, caution and/or proactive insulin dose reduction may be required, given that the insulin dose was reduced in the dapagliflozin 5 and 10 mg groups. In addition, further larger studies of longer duration are required to fully assess hypoglycemia rates with dapagliflozin added to insulin in patients with type 1 diabetes. Numeric increases in urine output were observed in all treatment groups, but there was no evidence of volume depletion with dapagliflozin therapy, and no patient experienced ketoacidosis.

In patients with type 1 diabetes, dapagliflozin exhibited a pharmacokinetic profile similar to that previously observed in patients with type 2 diabetes (7,8) and dose-dependently increased urinary glucose excretion, as would be expected from its mechanism of action.

No differences between dapagliflozin and placebo were noted for 7-point mean glucose measurements, and differences from placebo were not estimated for other exploratory efficacy parameters. While the study was not powered to test for between-group differences in changes in efficacy parameters, and dapagliflozin 95% CIs overlapped those of placebo, dose-related trends in the magnitude of effect for FPG, DAG, SD of DAG, and MAGE were suggested with dapagliflozin therapy (Fig. 1C–F). In addition to reductions in 24-h average glucose levels and reductions in glycemic variability, lower total insulin doses, particularly with the 5 and 10 mg doses of dapagliflozin, were suggested (Fig. 2C).

This study has acknowledged limitations that affect the interpretability of many of the changes in efficacy parameters. As the initial safety experience in type 1 diabetes, study duration was very short and sample sizes in each group were small, which may have led to randomization imbalances contributing to the higher baseline values in the placebo group versus the dapagliflozin groups for 24-h urine glucose (30.4 vs. 6.6–12.2 g/24 h) and HbA_1c (8.7% vs. 8.2–8.5%). Furthermore, given the uncertain safety risks of using an SGLT2 inhibitor in the treatment of type 1 diabetes, additional measures to ensure patient safety were necessary, such as very close monitoring in an inpatient setting with guidance on insulin dose adjustments. These features of the study design may have contributed to improvements in glucose control in all patients, including patients receiving placebo for whom reductions in glucosuria occurred.

This pilot study provides an initial evaluation of the safety, pharmacokinetic profile, and pharmacodynamic effect on urinary glucose excretion of dapagliflozin in patients with type 1 diabetes, as well as an exploratory evaluation of efficacy parameters to guide future study design considerations for the further examination of the potential role of dapagliflozin therapy in this population. There were several suggestive changes in efficacy parameters of interest, which provide hope that, in larger randomized controlled trials, SGLT2 inhibition may prove to be efficacious in reducing hyperglycemia in patients with type 1 diabetes.

Acknowledgments. The authors thank the study participants and investigators. In addition to R.R.H. (Center for Metabolic Research, VA San Diego Healthcare System and University of California, San Diego, La Jolla, CA) and J.R. (Dallas Diabetes and Endocrine Center at Medical City, Dallas, TX), the authors thank Martin Kankam (Vince And Associates Clinical Research, Overland Park, KS), Eva-Maria Heurich (Compass Research, LLC, Orlando, FL), Elaine Watkins (Profil Institute for Clinical Research, Inc., Chula Vista, CA), James Vanderlugt (Jasper Clinic, Inc., Kalamazoo, MI), Eli Ipp (Los Angeles Biomedical Research Institute at Harbor-UCLA Medical Center, Torrance, CA), Alex White (Progressive Medical Research, Port Orange, FL), Gautam Desai (Kansas City University of Medicine and Biosciences, Kansas City, MO), James Borders (Central Kentucky Research Associates, Inc., Lexington, KY), John Palmer (Regional Medical Clinic-Endocrinology, Rapid City, SD), and Ramon Vargas (Louisiana Research Associates, Inc., New Orleans, LA), who served as principal investigators. The authors also thank Julian Martins of inScience Communications, Springer Healthcare, for medical writing and editorial assistance.

Duality of Interest. This study was supported by AstraZenecahttp://dx.doi.org/10.13039/100004325 and Bristol-Myers Squibbhttp://dx.doi.org/10.13039/100002491. Medical writing and editorial assistance was funded by AstraZeneca and Bristol-Myers Squibb.

R.R.H. has served on scientific advisory boards of and received honoraria or consulting fees from the following insulin or SGLT2 manufacturers: Sanofi, Novo Nordisk, Boehringer Ingelheim, Bristol-Myers Squibb/AstraZeneca, Janssen, and Merck. His institution (Veterans Medical Research Foundation) has managed grants/research support from the following insulin or SGLT2 manufacturers: Sanofi, Novo Nordisk, Eli Lilly, Bristol-Myers Squibb/AstraZeneca, Boehringer Ingelheim, Pfizer, and Janssen. J.R. has served on scientific advisory boards of and received honoraria or consulting fees from the following insulin or SGLT2 manufacturers: Sanofi, Novo Nordisk, Eli Lilly, Merck, Janssen, Boehringer Ingelheim, and Lexicon. He has also received grants/research support from the following insulin or SGLT2 manufacturers: Sanofi, Novo Nordisk, Eli Lilly, Bristol-Myers Squibb, AstraZeneca, Merck, Pfizer, Johnson & Johnson, Janssen, Boehringer Ingelheim, and Lexicon. S.E. has served on scientific advisory boards of and received honoraria or consulting fees from Abbott, Animas, AstraZeneca, Bayer, Bristol-Myers Squibb, Daiichi Sankyo, Dexcom, Insulet, LifeScan, Lilly, Medtronic, Merck, Novo Nordisk, Sanofi Aventis, and Tandem. S.M. has served on scientific advisory boards of and received honoraria or consulting fees from Bristol-Myers Squibb and AstraZeneca. A.-G.C., S.K., A.B., N.I., J.L., and S.C.G. are employees of Bristol-Myers Squibb. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. R.R.H., S.E., and S.M. helped formulate the study concept and design; participated in the acquisition, analysis, and interpretation of the data; and contributed to writing and revising of the manuscript. J.R. participated in the acquisition, analysis, and interpretation of the data and contributed to the writing and revising of the manuscript. A.-G.C., S.K., A.B., and S.C.G. helped to formulate the study concept and design; participated in study supervision and analysis and interpretation of the data; and contributed to the writing and revising of the manuscript. N.I. and J.L. helped to formulate the study concept and design; participated in the analysis and interpretation of the data; and contributed to the writing and revising of the manuscript. R.R.H. is the guarantor of this work and, as such, had full access to all the data in the 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 73rd Scientific Sessions of the American Diabetes Association, Chicago, IL, June 21–25, 2013; and at the 49th Annual Meeting of the European Association for the Study of Diabetes, Barcelona, Spain, September 23–27, 2013.