Empagliflozin as Add-On to Metformin in Patients With Type 2 Diabetes: A 24-Week, Randomized, Double-Blind, Placebo-Controlled Trial

Metformin is recommended as first-line pharmacotherapy for patients with type 2 diabetes who have not achieved or are unlikely to achieve glycemic control through lifestyle modifications (1). Metformin mainly acts by reducing hepatic glucose production through inhibition of gluconeogenesis and may increase glucose uptake in peripheral tissue (2,3). Although initially effective, metformin alone frequently fails to maintain glycemic control as type 2 diabetes progresses (1,4). After 2 years of metformin monotherapy, 40–50% of patients fail to achieve treatment goals (5,6), increasing to ∼70% after 3 years (6,7) and ∼90% after 9 years (7). When glycemic control can no longer be maintained with monotherapy, additional therapies are required, but several of the agents used as second-line treatment are associated with tolerability issues, such as gastrointestinal side effects, hypoglycemia, and weight gain (8). There is, therefore, a need for new agents that are effective and well-tolerated, and can be added to metformin therapy when metformin alone fails to maintain glycemic control.

The kidney has emerged as a therapeutic target in the treatment of type 2 diabetes. Approximately 90% of glucose filtered by the kidney is reabsorbed by the sodium glucose cotransporter 2 (SGLT2), located in the proximal tubule of the nephron (9). SGLT2 inhibitors are novel oral antidiabetes agents that, by reducing glucose reabsorption, increase urinary glucose excretion and reduce hyperglycemia independent of β-cell function and insulin resistance (10). This mechanism carries a low risk of hypoglycemia, with additional benefits of weight loss and reductions in blood pressure (BP) (10,11).

Empagliflozin is an orally active, potent, selective inhibitor of SGLT2 (12). In phase II trials, treatment with empagliflozin as an add-on to metformin therapy was shown to be well-tolerated, with low rates of hypoglycemia, and to lead to significant reductions in HbA_1c and body weight that were maintained over 90 weeks (13,14). Phase III trials have demonstrated the efficacy and tolerability of empagliflozin over 24 weeks when given as monotherapy, as an add-on to therapy with metformin plus sulfonylurea, and as an add-on to therapy with pioglitazone (15–17).

The aim of this study (EMPA-REG MET) was to evaluate the efficacy, safety, and tolerability of empagliflozin (10 and 25 mg once daily) versus placebo over 24 weeks as an add-on to metformin therapy in patients with type 2 diabetes who have inadequate glycemic control. In addition, the efficacy and safety of empagliflozin 25 mg were investigated in patients with poor glycemic control with an HbA_1c level of >10% (>86 mmol/mol) in an open-label treatment arm.

This was a randomized, placebo-controlled, double-blind phase III study conducted from July 2010 to February 2012 in 148 centers in 12 countries (Canada, China, France, Germany, India, Korea, Mexico, Slovakia, Slovenia, Taiwan, Turkey, and the U.S.). The clinical trial protocol was approved by the institutional review boards, independent ethics committees, and competent authorities of the participating centers, and complied with the Declaration of Helsinki in accordance with the International Conference on Harmonization Harmonized Tripartite Guideline for Good Clinical Practice. All patients provided written informed consent. The trial was registered with ClinicalTrials.gov (NCT01159600).

Patients (aged ≥18 years; BMI ≤45 kg/m²) with inadequately controlled type 2 diabetes (HbA_1c ≥7% to ≤10% [≥53 to ≤86 mmol/mol]) despite undergoing a diet and exercise program and a stable (unchanged for ≥12 weeks prior to randomization) immediate-release metformin regimen were enrolled into the study. Patients with an HbA_1c level >10% (>86 mmol/mol) were eligible to participate in an open-label treatment arm.

Exclusion criteria included uncontrolled hyperglycemia (glucose level >13.3 mmol/L) after an overnight fast confirmed by a second measurement; acute coronary syndrome, stroke, or transient ischemic attack within 3 months prior to informed consent; indication of liver disease (alanine aminotransferase, alkaline aminotransferase, or alkaline phosphatase levels more than three times the upper limit of normal); impaired kidney function (estimated glomerular filtration rate [eGFR] <30 mL/min/1.73 m²) during screening or run-in; contraindications to metformin according to the local label; bariatric surgery or other gastrointestinal surgeries that induce chronic malabsorption; medical history of cancer (except for basal cell carcinoma) or treatment for cancer within the last 5 years; blood dyscrasias or any disorders causing hemolysis or unstable erythrocytes; treatment with antiobesity drugs 3 months prior to consent; use of any treatment at screening leading to unstable body weight; treatment with systemic steroids at the time of consent; change in the dosage of thyroid hormones within 6 weeks prior to consent; alcohol or drug abuse within 3 months of consent; and investigational drug intake in another trial within 30 days prior to the current trial.

Following a 2-week, open-label, placebo run-in period, eligible patients were randomized (1:1:1) to receive empagliflozin 10 mg, empagliflozin 25 mg, or placebo. Randomization was performed using a third-party interactive voice and web response system, and was stratified by HbA_1c level (<8.5% [<69 mmol/mol]; ≥8.5% [≥69 mmol/mol]), eGFR (≥90, 60–89, and 30–59 mL/min/1.73 m² [using the Modification of Diet in Renal Disease (MDRD) equation]), and region (Europe, Asia, North America, and Latin America).

During the 24-week treatment period, patients received once-daily therapy (in the morning with water) with empagliflozin 10 mg, empagliflozin 25 mg, or placebo as an add-on to metformin therapy (≥1,500 mg/day, the maximum tolerated dose, or the maximum dose according to the local label). Patients allocated to the open-label arm received empagliflozin 25 mg for 24 weeks without run-in. Study visits were scheduled at screening; at the start of the placebo run-in period (randomized patients only); and at weeks 0, 6, 12, 18, and 24; and a follow-up visit occurred 7 days after the last administration of the study drug.

Rescue medication treatment was initiated during the treatment period if, between weeks 1 and 12, a patient had a glucose level >13.3 mmol/L after an overnight fast; between weeks 12 and 24 a patient had a glucose level >11.1 mmol/L after an overnight fast; or an HbA_1c level >8.5% (>69 mmol/mol). The initiation, choice, and dosage of rescue medication used were at the discretion of the investigator, according to local prescribing information. In cases of hypoglycemia, rescue medication was to be reduced or discontinued. Where hyperglycemia or hypoglycemia could not be controlled, the patient was discontinued from the trial.

The primary end point was the change from baseline in HbA_1c level at week 24. Key secondary end points were the change from baseline to week 24 in body weight and weighted mean daily glucose (MDG) level using an 8-point blood glucose profile. Exploratory end points included the following: percentage of patients with baseline HbA_1c level ≥7.0% (≥53 mmol/mol) who had an HbA_1c level <7% (<53 mmol/mol) at week 24; changes from baseline to week 24 in fasting plasma glucose (FPG), waist circumference, systolic BP (SBP), and diastolic BP (DBP); percentage of patients with >5% reduction in body weight at week 24; percentage of patients with uncontrolled BP at baseline who had controlled BP (SBP <130 and DBP <80 mmHg) at week 24; and use of rescue medication. Change from baseline in 2-h postprandial glucose (PPG) was assessed in a subset of patients based on a meal tolerance test (MTT) at baseline and week 24.

Safety end points included vital signs, clinical laboratory parameters, 12-lead electrocardiogram findings, and adverse events (AEs; preferred terms were coded according to the Medical Dictionary for Drug Regulatory Activities version 14.1). AEs included all events with an onset after the first dose of trial medication up to a period of 7 days after the last dose. AEs of special interest included confirmed hypoglycemic AEs (plasma glucose ≤3.9 mmol/L and/or requiring assistance), and events consistent with urinary tract infection (UTI) and genital infection (identified from AEs reported spontaneously by the investigator using prospectively defined search categories based on 67 and 87 preferred terms, respectively).

Efficacy analysis was performed on the full analysis set (FAS), which included all randomized patients treated with one or more doses of study drug who had a baseline HbA_1c value. The 2-h PPG level was analyzed in the MTT set (patients in the FAS with valid baseline and one or more MTT measurements performed during treatment). Safety and lipid parameters were analyzed in the treated set (patients treated with one or more doses of study drug).

Values observed after a patient started receiving rescue medication were set to missing. The last observation carried forward (LOCF) approach was used to impute missing continuous efficacy data. If no subsequent measures were available, the last observed value was used to replace subsequent missing values. If a patient had no observations obtained during treatment, then the value from baseline was carried forward. Missing values within a course of measurements during treatment were interpolated based on the last observed value before the missing visit and the first observed value after the missing visit. MDG was also analyzed based on observed cases (OCs). Categorical efficacy variables were analyzed using noncompleters considered failure imputation. LOCF-IR imputation (i.e., the LOCF without setting values after rescue therapy to missing) was used for the analysis of lipid parameters. Analyses of efficacy end points in the open-label set were based on OCs.

The primary end point was assessed using an ANCOVA model, with treatment, region and eGFR at baseline as fixed effects, and baseline HbA_1c level as a linear covariate. Key secondary and continuous exploratory end points were analyzed using the statistical model described for the primary end point, with the baseline value for the end point in question as an additional linear covariate. Changes over time in HbA_1c, FPG, and BP were analyzed using restricted maximum likelihood–based mixed-model repeated measures. Categorical change in HbA_1c level and the proportion of patients with a >5% weight loss were analyzed using logistic regression.

Treatment differences versus placebo in primary and key secondary end points were tested using a hierarchical testing approach for each dose at a significance level of 2.5% (two-sided) to maintain the overall type I error at 5%. All other exploratory tests were two-sided at a 5% level (no multiplicity adjustment). Safety analyses and analyses of efficacy end points in the open-label group were descriptive.

Further details on statistical analysis including sample size calculation are given in the Supplementary Data.

A total of 638 patients with a background of metformin therapy were randomized to receive double-blind study medication (empagliflozin 10 mg, n = 217; empagliflozin 25 mg, n = 214; placebo, n = 207) (Fig. 1). One patient assigned to receive empagliflozin 25 mg was not treated; thus, the FAS comprised 637 patients, of whom 591 (92.8%) completed the study. An additional 69 patients with HbA_1c level >10% (>86 mmol/mol) received open-label empagliflozin 25 mg, and 58 of these patients (84.1%) completed the 24-week treatment period (Fig. 1). The 2-h PPG level was evaluated in a subset of 167 patients who had undergone an MTT (placebo, n = 57; empagliflozin 10 mg, n = 52; empagliflozin 25 mg, n = 58).

Demographic and baseline characteristics were balanced across treatment groups (Table 1). The mean (SD) age of randomized patients was 55.7 years (9.9 years), BMI was 29.2 kg/m² (5.5 kg/m²), and HbA_1c level was 7.90% (0.85%) (63 mmol/mol [9.3 mmol/mol]) at baseline.

Reductions from baseline in HbA_1c level at week 24 were significantly greater in the empagliflozin groups than in the placebo group (Fig. 2A). Differences in adjusted mean values versus placebo were −0.57% (95% CI −0.70 to −0.43) (−6.2 mmol/mol [95% CI −7.7 to −4.7]) for empagliflozin 10 mg and −0.64% (95% CI −0.77 to −0.50) (−7.0 mmol/mol [95% CI −8.4 to −5.5]) for empagliflozin 25 mg; P < 0.001 for both doses (Supplementary Table 1). The adjusted mean HbA_1c level in the randomized patients over the 24-week treatment period are shown in Fig. 2B. In patients with baseline HbA_1c levels ≥7.0% (≥53 mmol/mol), more patients in the empagliflozin 10 and 25 mg groups had an HbA_1c level <7.0% (<53 mmol/mol) at week 24 compared with those in the placebo group (37.7%, 38.7%, and 12.5%, respectively; both P < 0.001 vs. placebo) (Supplementary Table 1 and Supplementary Fig. 1A).

Treatment with empagliflozin resulted in significant reductions in MDG level versus placebo. The adjusted mean (SE) changes from baseline to week 24 in MDG level using LOCF were −0.11 mmol/L (0.11 mmol/L) with placebo, −0.54 mmol/L (0.10 mmol/L) with empagliflozin 10 mg, and −0.80 mmol/L (0.10 mmol/L) with empagliflozin 25 mg (differences in adjusted means vs. placebo were −0.42 mmol/L [95% CI −0.72 to −0.13] for empagliflozin 10 mg [P = 0.006] and −0.69 mmol/L [95% CI −0.99 to −0.39] for empagliflozin 25 mg [P < 0.001]; Supplementary Table 1). Because many of the patients analyzed (38.3%) had missing or invalid MDG measurements at week 24 and so had their baseline values carried forward according to the LOCF imputation method, results based on the prespecified OC analysis may reflect the effect of empagliflozin on MDG level more accurately. The results of this analysis are shown in Fig. 2D.

Changes from baseline in FPG level over time are shown in Supplementary Figure 1B. There were significant reductions in FPG level with both doses of empagliflozin compared with placebo at week 24; the adjusted mean (SE) changes from baseline were 0.35 mmol/L (0.10 mmol/L) with placebo, −1.11 mmol/L (0.10 mmol/L) with empagliflozin 10 mg, and −1.24 mmol/L (0.10 mmol/L) with empagliflozin 25 mg (both P < 0.001 vs. placebo) (Supplementary Table 1 and Supplementary Fig. 1C). There were significant reductions in the 2-h PPG level with both doses of empagliflozin compared with placebo; the adjusted mean (SE) changes from baseline were 0.33 mmol/L (0.34 mmol/L) with placebo, −2.55 mmol/L (0.35 mmol/L) with empagliflozin 10 mg, and −2.47 mmol/L (0.33 mmol/L) with empagliflozin 25 mg (both P < 0.001) (Supplementary Table 1 and Supplementary Fig. 1D).

Both doses of empagliflozin resulted in significant reductions in body weight compared with placebo (Fig. 2E; differences of adjusted means versus placebo were −1.63 kg [95% CI −2.11 to −1.15] for empagliflozin 10 mg and −2.01 kg [95% CI −2.49 to −1.53] for empagliflozin 25 mg; P < 0.001 for both doses; Supplementary Table 1). Compared with placebo, more patients achieved a >5% reduction in body weight with empagliflozin 10 mg and 25 mg (4.8%, 21.2%, and 23.0%, respectively; both P < 0.001) (Supplementary Table 2 and Supplementary Fig. 2A). The decreases in body weight with empagliflozin 10 and 25 mg were accompanied by significant reductions in waist circumference (adjusted mean [SE] changes of −1.55 cm [0.25 cm] and −1.57 cm [0.25 cm], respectively) compared with placebo (−0.54 cm [0.25 cm]; P = 0.005 and P = 0.004, respectively, for empagliflozin 10 mg and 25 mg vs. placebo) (Supplementary Table 2 and Supplementary Fig. 2B).

Empagliflozin therapy was associated with significantly greater reductions from baseline in SBP and DBP at week 24 than placebo (Fig. 2F and G, and Supplementary Table 3). Changes from baseline in SBP and DBP over time are shown in Supplementary Figure 3A and B. The percentage of patients with uncontrolled BP at baseline who had controlled BP (SBP <130 mmHg and DBP <80 mmHg) at week 24 was higher with empagliflozin 10 mg (35.9%) and 25 mg (30.4%) than with placebo (13.2%; P < 0.001 for both doses) (Supplementary Table 3). Despite reductions in BP with empagliflozin therapy, there were no changes in pulse rate (data not shown).

Rescue therapy was administered to a total of 48 patients (7.5%) in the randomized groups; more patients received rescue therapy in the placebo group (29 patients [14.0%]) than in the empagliflozin 10 mg group (12 patients [5.5%]) and the empagliflozin 25 mg group (7 patients [3.3%]).

Figure 2C shows the change in HbA_1c level in the open-label empagliflozin 25 mg group (patients with baseline HbA_1c level >10% [>86 mmol/mol]). The mean (SE) change from baseline to week 24 in HbA_1c level was −3.23% (0.22%) (−35.3 mmol/mol [2.4 mmol/mol]). The mean (SE) changes from baseline to week 24 in MDG and FPG levels were −4.23 mmol/L (0.53 mmol/L) and −3.02 mmol/L (0.57 mmol/L), respectively (Supplementary Table 1). The mean (SE) changes from baseline in body weight and waist circumference were −1.91 kg (0.59 kg) and −2.52 cm (1.26 cm), respectively; 15.9% of patients had a >5% reduction in body weight at week 24 (Supplementary Table 2). For SBP and DBP, the mean (SE) changes from baseline to week 24 were −2.4 mmHg (1.6 mmHg) and −3.6 mmHg (1.3 mmHg), respectively; and 36.2% of patients with uncontrolled BP at baseline had controlled BP (SBP <130 mmHg, DBP <80 mmHg) at week 24 (Supplementary Table 3). In the open-label group, 14.5% of patients received rescue therapy.

The number of patients reporting one or more AEs was similar across groups (Table 2). Most patients with one or more AEs (95%) reported only events of mild or moderate intensity. There were no deaths. There were very few patients with confirmed hypoglycemic AEs (one patient who received placebo [0.5%]; four patients receiving empagliflozin 10 mg [1.8%]; and three patients receiving empagliflozin 25 mg [1.4%]); none of the events required assistance (Table 2). In the open-label group, confirmed hypoglycemic AEs were reported in two patients (2.9%).

The proportion of patients with events consistent with UTI was comparable across the treatment groups (placebo group, n = 10 [4.9%]; empagliflozin 10 mg group, n = 11 [5.1%]; empagliflozin 25 mg group, n = 12 [5.6%]) (Table 2). Most patients who reported events consistent with UTI (79%) reported only events of mild intensity; none of the events were severe or led to study discontinuation. No case of urosepsis or pyelonephritis was reported. Most patients who reported any events consistent with UTI reported just one event; two patients in the empagliflozin 10 mg group reported two events, and one patient in the empagliflozin 10 mg group reported three to four events. In the empagliflozin groups, events consistent with UTI were reported in more female patients (11.8–12.0%) than male patients (0.0–0.8%). In the open-label group, events consistent with UTI were reported in five patients (7.2%).

Events consistent with genital infection were reported by 8 patients (3.7%) in the empagliflozin 10 mg group, 10 patients (4.7%) in the empagliflozin 25 mg group, and no patients in the placebo group. In the empagliflozin groups, events consistent with genital infection were reported in more female patients (7.6–9.7%) than male patients (0.8%) (Table 2). All of the patients who reported any events consistent with genital infection reported just one event, except for two patients in the empagliflozin 10 mg group, who reported two events. Most patients with events consistent with genital infection (56%) reported only mild events, and none of the events were severe. Two patients (one patient in each empagliflozin group) prematurely discontinued study medication because of events consistent with genital infection. In the open-label group, one patient (1.4%) reported an event consistent with genital infection.

Changes in laboratory parameters are shown in Supplementary Table 4. In the randomized empagliflozin groups, hematocrit increased by 2.4–2.7% from baseline, and serum uric acid concentration decreased by 45–56 μmol/L from baseline. There were small changes in eGFR and no changes in electrolyte levels across groups. There was a small increase from baseline in HDL cholesterol level with empagliflozin therapy (adjusted mean [SE] 0.08 mmol/L [0.01 mmol/L] for empagliflozin 10 mg and 0.06 mmol/L [0.01 mmol/L] for empagliflozin 25 mg) compared with placebo (0.00 mmol/L [0.01 mmol/L]; P ≤ 0.001 for both), a small increase in total cholesterol level with empagliflozin 10 mg therapy (adjusted mean [SE] 0.23 mmol/L [0.05 mmol/L]) compared with placebo (0.09 mmol/L [0.05 mmol/L]; P = 0.043), and a small increase in LDL cholesterol level with empagliflozin therapy (adjusted mean [SE] 0.15 mmol/L [0.04 mmol/L] for both doses) compared with placebo (0.03 mmol/L [0.04 mmol/L]; P = 0.043 for empagliflozin 10 mg, P = 0.032 for empagliflozin 25 mg). No major differences in adjusted mean changes from baseline in triglyceride levels were noted between placebo and empagliflozin therapy.

The aim of this study was to evaluate the efficacy, safety, and tolerability of empagliflozin as an add-on to metformin therapy in patients with type 2 diabetes. Current treatment guidelines for type 2 diabetes advocate the addition of a second agent in patients who do not maintain an HbA_1c target level with metformin monotherapy (1).

The results of this trial demonstrate that empagliflozin used as an add-on to metformin therapy improves glycemic control with a low risk of hypoglycemia, and leads to weight loss and BP reduction. Treatment with empagliflozin resulted in significant reductions in HbA_1c levels compared with placebo, and the proportion of patients reaching an HbA_1c level of <7% (<53 mmol/mol) was higher with empagliflozin therapy (38–39%) than with placebo (13%). In addition, empagliflozin treatment led to significant improvements in FPG, MDG, and 2-h PPG levels compared with placebo. In patients with very poor glycemic control (mean baseline HbA_1c 11.07% [98 mmol/mol]), mean HbA_1c level was reduced by 3.23% (35.3 mmol/mol) after 24 weeks of treatment with empagliflozin to an HbA_1c level of 7.86% (62 mmol/mol).

More than 80% of patients with type 2 diabetes are overweight or obese (18). Overweight/obesity is a significant cardiovascular risk factor (19) and can complicate the management of type 2 diabetes because it increases insulin resistance and glucose intolerance (20,21). Weight control is challenging for patients with type 2 diabetes to achieve and may be complicated by medication side effects (22,23). Treatment with empagliflozin resulted in significant reductions in body weight of 2.1–2.5 kg and led to body weight reductions of >5%, which have been shown to improve insulin action and glucose control (24), in 21–23% of patients, compared with 5% of patients who received placebo. The weight loss observed with SGLT2 inhibitor therapy may be due to a loss of calories as a result of urinary glucose excretion (10), and has been reported to reflect the loss of both subcutaneous and visceral fat (25).

Hypertension affects more than half of patients with diabetes (26). Empagliflozin as an add-on to metformin therapy significantly reduced SBP by 4.1–4.8 mmHg, and DBP by 1.6–1.9 mmHg compared with placebo without increases in pulse rate. These reductions may be due to mild osmotic diuresis and weight loss as a result of urinary glucose excretion (11). Consistent with glucose-induced osmotic diuresis, treatment with empagliflozin resulted in a small increase in hematocrit. Treatment with empagliflozin resulted in a numerical reduction in uric acid levels compared with a small increase in patients receiving placebo.

Empagliflozin doses of 10 and 25 mg as add-ons to metformin therapy were well-tolerated. The number of patients with confirmed hypoglycemic AEs was low, and no hypoglycemic events required assistance. Hypoglycemia represents a safety risk for patients with type 2 diabetes, with implications for morbidity, mortality, quality of life, and treatment adherence (27–29). Therefore, the low risk of hypoglycemia is an important safety attribute of empagliflozin therapy. Patients with type 2 diabetes are at increased risk of UTIs and genital infections (30–32). While the proportion of patients with events consistent with UTI was comparable in patients treated with empagliflozin or placebo, an increased risk of genital infection was observed in patients treated with empagliflozin. However, all such events were mild or moderate, and only one patient in each empagliflozin group discontinued study medication due to an event consistent with genital infection.

Empagliflozin and metformin act via complementary mechanisms of action, resulting in significant and clinically meaningful improvements in glucose control, accompanied by a modest weight reduction. Because metformin therapy alone frequently fails to maintain glycemic control as type 2 diabetes progresses (1,4), a combination of metformin and empagliflozin may emerge as an attractive new treatment option. This trial demonstrates the potential of empagliflozin to be used as add-on therapy in patients with type 2 diabetes who fail to achieve glycemic control while receiving metformin monotherapy. A phase III study (NCT01719003, clinicaltrials.gov) will illuminate the safety and efficacy of an initial combination of empagliflozin with metformin in patients with type 2 diabetes.

Acknowledgments. The authors thank Elizabeth Ng and Wendy Morris of Fleishman Hillard Group, Ltd. for medical writing assistance during the preparation of this article.

Funding. This research was supported by Boehringer Ingelheim and Eli Lilly. Medical writing assistance was supported by Boehringer Ingelheim.

Duality of Interest. H.-U.H. is a member of the advisory boards for Daiichi Sankyo, Sanofi, Boehringer Ingelheim, and Roche. L.M. is an investigator, consultant, and/or speaker without any direct financial benefit to him under contracts between his employer and the following companies: AstraZeneca, Boehringer Ingelheim, Bristol-Myers Squibb, Eli Lilly, Merck Sharp & Dohme, Novo Nordisk, and Roche Pharma. Neither he nor his family members hold stocks in pharmaceutical or device companies. E.S.-B., M.W., T.M., U.C.B., and H.J.W. are employees of Boehringer Ingelheim.

Author Contributions. H.-U.H. and L.M. contributed to the acquisition and interpretation of the data, drafted the manuscript, approved the final version of the manuscript before submission for publication, had full access to the study data, and were responsible for the final decision to submit the manuscript for publication. E.S.-B., M.W., T.M., U.C.B., and H.J.W. contributed to the study design and interpretation of the data, reviewed and edited the manuscript, contributed to the writing of the manuscript, approved the final version of the manuscript before submission for publication, had full access to the study data, and were responsible for the final decision to submit the manuscript for publication. M.W. 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, 21–25 June 2013.