Efficacy and Safety of IDegLira Versus Basal-Bolus Insulin Therapy in Patients With Type 2 Diabetes Uncontrolled on Metformin and Basal Insulin: The DUAL VII Randomized Clinical Trial

Progression of type 2 diabetes means adequate blood glucose (BG) control with basal insulin may deteriorate over time. By the time basal insulin is initiated, patients on average have had a type 2 diabetes diagnosis for 9.2 years and a glycated hemoglobin (HbA_1c) level of 9.5% (80 mmol/mol) (1). In one study, only 29% of patients could maintain HbA_1c <7.0% (53 mmol/mol) 3 years after basal insulin initiation (2). Current guidelines recommend therapy intensification if HbA_1c targets are not reached after 3–6 months of basal insulin (3). However, many patients remain without therapy intensification (4) due to concerns over potential increased hypoglycemia risk, weight gain, and treatment complexity (5,6).

Glucagon-like peptide 1 receptor agonists (GLP-1RAs) are incretin mimetics that lower fasting plasma glucose (FPG) and postprandial glucose via stimulation of endogenous insulin and inhibition of glucagon secretion, with weight-reducing effects through appetite suppression (7). Using basal insulin with GLP-1RAs is a recognized therapeutic option after failure to achieve glycemic control with basal insulin (3). Comparison of basal insulin and GLP-1RA given in separate injections versus basal insulin alone showed a significantly greater HbA_1c reduction and weight loss with the combination therapy, with significantly more hypoglycemic events (8). Comparison with basal-bolus showed noninferior HbA_1c reduction with basal insulin and GLP-1RA loose combination as well as more desirable hypoglycemia rates and weight management (9). This evidence supports combination therapies such as IDegLira, a fixed-ratio combination of insulin degludec (degludec) and liraglutide.

IDegLira efficacy and safety were established in the Dual Action of Liraglutide and Insulin Degludec in Type 2 Diabetes (DUAL) clinical trial program in patients with uncontrolled type 2 diabetes on oral antidiabetic drugs (DUAL I, IV, and VI) (10–12), GLP-1RAs (DUAL III) (13), and basal insulin (DUAL II and V) (14,15). DUAL VII assessed IDegLira efficacy and safety versus basal-bolus.

DUAL VII was a phase 3b, multinational (Supplementary Table 1), open-label, two-arm parallel, randomized trial in patients with type 2 diabetes conducted at 89 sites in 12 countries from July 2015 to October 2016. The trial lasted 32 weeks, with 2-week screening, 26-week treatment, and 4-week follow-up periods. It was conducted in accordance with the International Conference on Harmonisation Guideline for Good Clinical Practice (16) and the Declaration of Helsinki (17).

Patients included were ≥18 years old with type 2 diabetes, HbA_1c 7.0–10.0% (53–86 mmol/mol), BMI ≤40 kg/m², and on stable daily doses of insulin glargine 100 units/mL (IGlar U100) 20–50 units and metformin ≥1,500 mg or maximum tolerated dose for >90 days prior to screening. Exclusion criteria included treatment with any medication for diabetes or obesity other than stated in the inclusion criteria during 90 calendar days before screening, anticipated initiation or change in concomitant medications known to affect weight or glucose metabolism, and renal impairment (estimated glomerular filtration rate <60 mL/min/1.73 m²) (Supplementary Table 2).

A treat-to-target approach ensured optimal insulin titration and improved glycemic control in all patients.

Patients were randomized 1:1 using a centralized allocation via an interactive voice web response system.

Patients were randomized to either IDegLira (insulin degludec 100 units/mL and liraglutide 3.6 mg/mL, in a 3-mL prefilled FlexTouch pen for subcutaneous injection) or IGlar U100 (insulin glargine 100 units/mL solution, in a 3-mL prefilled Solostar pen for subcutaneous injection) and insulin aspart (IAsp) (100 units/mL solution in a 3-mL prefilled FlexPen for subcutaneous injection). Metformin was continued at pretrial doses.

IDegLira was administered once daily at any time, independent of meals, repeated approximately the same time each day. Patients were initiated on 16 units (16 units degludec/0.58 mg liraglutide) and titrated twice weekly, aiming for a mean prebreakfast self-monitored plasma glucose (SMPG) target range of 4.0–5.0 mmol/L (72–90 mg/dL) (Supplementary Table 3A). The maximum dose of IDegLira was 50 units (50 units degludec/1.8 mg liraglutide). IGlar U100 was administered once daily according to local labeling and was initiated at a dose equivalent to the pretrial dose, titrated using the same algorithm as IDegLira. IAsp was initiated from the day of randomization at a starting dose of 4 units per main meal (≤4 times per day) and titrated twice weekly, aiming for a mean preprandial and bedtime SMPG target range of 4.0–6.0 mmol/L (72–108 mg/dL) (Supplementary Table 3B). The principal investigator at each study site could adjust the titration per clinical judgment. There were no protocols given to patients regarding meal size or constituents.

Patients who prematurely discontinued treatment had telephone contacts every 4 weeks to collect information on serious adverse events (AEs) (death, life-threatening experience, inpatient hospitalization or prolongation of existing hospitalization, persistent/significant disability/incapacity, congenital anomaly, event that jeopardizes the patient requiring medical/surgical intervention, and suspicion of transmission of infectious agents via trial product), major cardiovascular events, and antidiabetic medication. These patients were also invited to the week-26 visit for data collection on these parameters, as well as HbA_1c. A small number of patients completed the week-26 visit as a telephone contact, with no HbA_1c collected.

The primary end point was change in HbA_1c from baseline to week 26 of treatment. Confirmatory secondary end points included number of treatment-emergent severe or BG-confirmed symptomatic hypoglycemic episodes during 26 weeks of treatment and change in body weight from baseline after 26 weeks of treatment. Hypoglycemic events were defined as either severe according to the American Diabetes Association (ADA) classification (requiring assistance of another person to take corrective actions) (18) or symptomatic BG-confirmed (<3.1 mmol/L [56 mg/dL] accompanied by glycopenic symptoms). Supportive secondary efficacy end points included total, basal, and bolus daily insulin doses; responders for HbA_1c <7.0% (53 mmol/mol) and ≤6.5% (48 mmol/mol) after 26 weeks of treatment; and proportion of patients achieving these HbA_1c targets without severe or BG-confirmed symptomatic hypoglycemia in the last 12 weeks and/or weight gain. Hypoglycemia rates during the last 12 weeks were included as this was considered a maintenance period with more stable insulin doses and comparable hypoglycemia rates. Other glycemia-related secondary end points included changes in FPG and mean difference in the nine-point SMPG profile.

Supportive secondary safety end points included number of treatment-emergent AEs and treatment-emergent nocturnal severe or BG-confirmed symptomatic hypoglycemic episodes (severe or BG-confirmed symptomatic hypoglycemia occurring between 0001 h and 0559 h inclusive) (Supplementary Fig. 1) and changes from baseline in blood pressure, heart rate, laboratory assessments of biochemistry, hematology, and fasting lipid profile.

The trial was powered for the primary objective of confirming noninferiority with respect to HbA_1c change from baseline, using a one-sided Student t test with the following assumptions: a mean treatment difference of 0.0%, an SD from HbA_1c change from baseline of 1.0%, a noninferiority margin of 0.30%, and up to 15% of randomized patients excluded from the per-protocol analysis set. Sample size was set to 250 per treatment arm, ensuring a power of at least 85% for confirming the primary objective in the per-protocol analysis set.

All efficacy end points were summarized using the full analysis set (FAS), and safety end points were summarized using the safety analysis set (SAS). All statistical analyses of efficacy and safety end points were based on the FAS (analysis set definitions are detailed in Supplementary Table 4).

For the primary end point, noninferiority was considered confirmed if the upper bound of the two-sided 95% CI for estimated mean between-treatment difference in change from baseline in HbA_1c was <0.30%. All postbaseline HbA_1c on-treatment measurements obtained at scheduled visits were analyzed using a linear mixed normal model using an unstructured residual covariance matrix for HbA_1c measurements within the same patient. The model included treatment, visit, and region as fixed factors and baseline HbA_1c as covariate, with interactions between visit and all factors and covariates included. In order to control the overall type I error on a 5% level with regard to confirmatory secondary end points, a hierarchical testing procedure was used (Statistical Analyses Detail in Supplementary Data).

A negative binomial model with a log-link function and the log of the time period in which a hypoglycemic episode is considered treatment emergent as offset was used to analyze the number of treatment-emergent severe or BG-confirmed symptomatic hypoglycemic episodes, including treatment and region as fixed factors. Risk of hypoglycemia versus no hypoglycemia was analyzed using a generalized linear regression model with a log link, including treatment and region as fixed factors.

A mixed model for repeated measurements (MMRM) with an unstructured covariance matrix was used for the continuous confirmatory and supportive secondary end points, including treatment, visit, and region as fixed factors and corresponding baseline values as covariates. Interactions between visit and all factors and the covariate were also included in the model. Insulin dose was analyzed using a compound symmetry covariance matrix and included IGlar U100 dose at screening as an additional covariate.

A number of sensitivity analyses, including reference-based multiple imputation, were performed to assess the robustness of primary and confirmatory secondary analyses (Supplementary Table 5).

The nine-point SMPG profile was analyzed using a linear mixed-effect model fitted to data at week 26. The model included treatment, region, time (within nine-point profile), and interactions between treatment and time and between time and region as fixed factors and patient as a random effect, where measurements within patients were assumed correlated using a compound symmetry covariance matrix.

Responder end points were analyzed using a logistic regression model with treatment and region as fixed factors and baseline HbA_1c with or without baseline body weight as covariates. Missing response data were imputed from the MMRM analysis of the primary end point change in HbA_1c from baseline with or without the confirmatory secondary end point change in body weight. AEs were summarized descriptively.

Overall, 672 patients were screened and 506 were randomized, with 99.2% of those randomized to IDegLira and 98.0% of those randomized to basal-bolus completing the trial. Furthermore, 94.4% and 91.7% completed 26 weeks of treatment with IDegLira and basal-bolus, respectively (Fig. 1). Baseline characteristics were similar between the two arms (Table 1).

After 26 weeks of treatment, mean HbA_1c decreased from 8.2% (66 mmol/mol) at baseline to 6.7% (50 mmol/mol) at end of trial (EOT) with IDegLira and from 8.2% (67 mmol/mol) to 6.7% (50 mmol/mol) with basal-bolus. The estimated treatment difference (ETD) was −0.02% (95% CI −0.16, 0.12) (−0.2 mmol/mol [95% CI −1.7, 1.3]) (Fig. 2A), confirming noninferiority of IDegLira versus basal-bolus (P < 0.0001). In total, 489 patients (244 and 245 in the IDegLira and basal-bolus arms, respectively) contributed to the primary analysis. All preplanned sensitivity analyses were in agreement with the primary analysis (Supplementary Tables 5 and 6 and Supplementary Fig. 2).

During 26 weeks of treatment, 19.8% of patients on IDegLira experienced one or more severe or BG-confirmed symptomatic hypoglycemic episodes vs. 52.6% with basal-bolus treatment, corresponding to a 61% lower risk with IDegLira compared with basal-bolus (estimated risk ratio 0.39 [95% CI 0.29, 0.51], P < 0.0001). Patients on IDegLira experienced 129 episodes in total (rate of 1.07 episodes per patient year of exposure [PYE)] versus patients on basal-bolus who experienced 975 episodes in total (rate of 8.17 episodes per PYE). This resulted in an 89% lower rate of severe or BG-confirmed symptomatic hypoglycemic episodes with IDegLira versus basal-bolus (estimated rate ratio 0.11 [95% CI 0.08, 0.17], P < 0.0001), confirming the superiority of IDegLira versus basal-bolus. Figure 2B demonstrates an increasing divergence in the cumulative hypoglycemia rate from randomization until EOT. Severe hypoglycemic episodes occurred in 1.2% of patients on IDegLira throughout the treatment period (three episodes) at a rate of 0.03 episodes per PYE vs. 1.6% of patients on basal-bolus (10 episodes) at a rate of 0.08 episodes per PYE. There was a 34% lower risk (estimated risk ratio 0.76 [95% CI 0.17, 3.35], P = 0.7173) and 72% lower rate (estimated rate ratio 0.28 [0.04, 1.98], P = 0.2034) of severe hypoglycemia in IDegLira compared with basal-bolus. Over 26 weeks of treatment, observed mean body weight decreased by 0.9 kg with IDegLira from 87.2 kg and increased with basal-bolus by 2.6 kg from 88.2 kg (ETD −3.6 kg [95% CI −4.2, −2.9], P < 0.0001) (Fig. 2C), confirming the superiority of IDegLira over basal-bolus.

All preplanned sensitivity analyses, including reference-based multiple imputation, were in agreement with conclusions from confirmatory analyses (Supplementary Tables 5 and 6 and Supplementary Fig. 2).

Total daily insulin dose increased to a mean EOT dose of 40 and 84 units with IDegLira and basal-bolus, respectively (ETD −44.5 units [95% CI −48.3, −40.7], P < 0.0001) (Fig. 2D). Mean basal insulin dose with basal-bolus increased from 34 units in week 1 to 52 units at EOT vs. a mean EOT dose of 40 units with IDegLira (corresponding to 40 units degludec/1.44 mg liraglutide) (ETD −12.6 units [95% CI −14.9, −10.3], P < 0.0001). Total daily bolus insulin dose in the basal-bolus arm at EOT was 32 units of bolus insulin. After 26 weeks of treatment, 86 patients on IDegLira were on the 50-unit maximum dose, with a mean HbA_1c at week 26 of 7.0% (53 mmol/mol); 57% of these patients achieved HbA_1c <7.0% (53 mmol/mol), with a mean HbA_1c at week 26 of 6.4% (46 mmol/mol). Those who did not achieve an HbA_1c <7.0% on 50 units IDegLira had a mean HbA_1c at week 26 of 7.8% (61 mmol/mol).

By week 26, ∼90% of patients on basal-bolus reported taking at least three insulin injections per day, including the once-daily basal injection (Supplementary Fig. 3) versus the once-daily single injection with IDegLira.

Patients on IDegLira had a lower rate of nocturnal hypoglycemia compared with basal-bolus. During the 26 treatment weeks, 4.8% of patients on IDegLira experienced one or more nocturnal severe or BG-confirmed symptomatic hypoglycemic episodes (16 episodes) at a rate of 0.13 episodes per PYE vs. 19.4% of patients on basal-bolus (198 episodes) at a rate of 1.66 per PYE. There was a 75% lower risk and 92% rate reduction in nocturnal hypoglycemia with IDegLira versus basal-bolus (estimated risk ratio 0.25 [95% CI 0.13, 0.45], estimated rate ratio 0.08 [95% CI 0.04, 0.17]; both P < 0.0001). No patients on IDegLira experienced any nocturnal severe hypoglycemic episodes versus one episode in the basal-bolus arm.

Similar proportions of patients in each treatment arm achieved an HbA_1c <7.0% (53 mmol/mol): 66% in IDegLira and 67% in basal-bolus (Fig. 3). Likewise, 49.6% and 44.6% of patients on IDegLira and basal-bolus, respectively, achieved HbA_1c ≤6.5% (48 mmol/mol) (ETDs not statistically significant for both targets) (Supplementary Fig. 4). More patients reached the triple composite HbA_1c targets (<7.0% [<53 mmol/mol]) without hypoglycemic episodes in the last 12 weeks of treatment and without weight gain on IDegLira versus basal-bolus (odds ratio [OR] 10.39 [95% CI 5.76, 18.75], P < 0.0001) (Fig. 3). Odds for HbA_1c targets <7.0% (53 mmol/mol) without hypoglycemic episodes in 26 weeks of treatment and without weight gain were higher with IDegLira versus basal-bolus (OR 12.56 [95% CI 6.46, 24.45], P < 0.0001) (Fig. 3).

The observed mean FPG decreased with IDegLira by 2.4 mmol/L and by 1.9 mmol/L with basal-bolus (ETD −0.31 mmol/L [95% CI −0.67, 0.05], P = 0.0936) (Fig. 2E).

The nine-point SMPG profile decreased with both treatments at all time points (Fig. 2F) but to a larger extent with basal-bolus than with IDegLira. The mean of the nine-point SMPG decreased from 9.7 to 7.2 mmol/L with IDegLira and from 9.7 to 6.6 mmol/L with basal-bolus (ETD 0.57 mmol/L [95% CI 0.31, 0.83], P < 0.0001; in favor of basal-bolus). Basal-bolus insulin treatment resulted in statistically significantly lower postlunch, dinner, postdinner, and bedtime SMPG readings (Fig. 2F). The mean prandial increments decreased in both arms at all three meals but were only statistically significantly different in favor of basal-bolus at the evening meal (ETD 0.49 mmol/L [95% CI 0.06, 0.92], P = 0.0261).

A greater decrease in mean systolic blood pressure was observed with IDegLira (−4.5 mmHg) versus basal-bolus (−1.2 mmHg) (ETD −3.70 [95% CI −5.68, −1.72], P = 0.0003), but no statistically significant difference was observed between the two arms in terms of changes in diastolic blood pressure. There was a greater difference in increase in heart rate with IDegLira (3.5 bpm) versus basal-bolus (1.3 bpm) (ETD 2.73 [95% CI 1.30, 4.16], P = 0.0002). Small but statistically significant differences were seen between the two arms for various lipid profile parameters (Supplementary Table 8).

AEs occurred in similar proportions of patients in both treatment arms: 59.1% of IDegLira patients at a rate of 4.1 events per PYE vs. 56.9% of patients on basal-bolus at a rate of 3.5 events per PYE (Supplementary Table 7). The majority of AEs were nonserious, and there was no clustering of events: 13 and 11 serious AEs with IDegLira and basal-bolus, respectively (Supplementary Table 7). There were no fatal AEs. The most common AE with IDegLira was nausea, with 11.1% vs. 1.6% of patients on basal-bolus reporting one or more events. At all time points, the reported nausea rate was <3% in either treatment arm (Supplementary Fig. 5). The most common AE on basal-bolus was nasopharyngitis, with 11.9% vs. 4.8% of patients on IDegLira reporting one or more events. There was one confirmed adjudicated cardiovascular event in the IDegLira arm, confirmed as unstable angina pectoris. There was one confirmed benign neoplasm (a colorectal polyp) with IDegLira and no confirmed neoplasms on basal-bolus. There were no confirmed events of pancreatitis or thyroid disease or major cardiovascular events in the trial.

This is the only trial thus far that compares a fixed-ratio combination of basal insulin and GLP-1RA with basal-bolus. With once-daily versus multiple daily injections, IDegLira was noninferior to basal-bolus with respect to HbA_1c reduction and statistically superior with respect to lower hypoglycemia rate and change in body weight in patients on IGlar U100 and metformin with uncontrolled BG.

Other trials have investigated combinations of insulin and GLP-1RA given separately. In the BEGIN: VICTOZA ADD-ON trial, addition of liraglutide to degludec in separate injections resulted in a significantly greater reduction in HbA_1c versus addition of IAsp at one meal to degludec, with weight loss and a reduced risk of hypoglycemia (19). Addition of once-weekly albiglutide to IGlar U100 resulted in similar HbA_1c reductions versus IGlar U100 and prandial insulin lispro, with weight loss and a reduced risk of hypoglycemia (9). These results are mirrored in this trial. Furthermore, the fixed combination achieved these outcomes with only one injection per day (vs. ≤5 in the basal-bolus arm) and lower rates of nausea, likely attributable to the ability to slowly titrate the IDegLira dose in contrast to the separate injections (9,19).

An insulin-sparing effect was demonstrated with IDegLira versus basal-bolus. Patients achieved similar HbA_1c reductions with a mean daily dose of ∼13 units of basal insulin less with IDegLira and nearly 45 units of total daily insulin less.

Despite being an efficacious glucose-lowering therapy, basal-bolus treatment is associated with a higher rate of hypoglycemia versus other diabetes therapies (20). Although resulting in similar HbA_1c reductions, IDegLira was confirmed to be statistically superior to basal-bolus, with an 89% lower rate of severe or BG-confirmed symptomatic hypoglycemic episodes and 92% lower rate of nocturnal hypoglycemia. The striking difference in the cumulative incidence of hypoglycemia was noted early in the treatment period, with an immediate split between the two curves at randomization and continuing divergence toward EOT. This difference is likely to be explained by the lower total daily insulin dose with IDegLira; the glucose-stimulated insulin secretion promoted by the GLP-1RA component of IDegLira (7), which reduces the risk of low BG (21); and the lower rate of hypoglycemia with degludec versus IGlar U100 (22).

Weight gain is another concern when initiating basal-bolus treatment, and increasing doses of insulin can promote weight gain (3,6). Here, IDegLira was confirmed to be statistically superior to basal-bolus in terms of change in body weight. This is likely to be due to the weight-reducing and insulin-sparing effects of liraglutide mitigating the risk of weight gain with insulin (3).

There were no unexpected safety or tolerability issues identified with IDegLira in this trial, and the safety profile was as expected based on results from previous DUAL and monocomponent trials, including the LEADER (Liraglutide Effect and Action in Diabetes: Evaluation of Cardiovascular Outcome Results—A Long Term Evaluation) and DEVOTE (Trial Comparing Cardiovascular Safety of Insulin Degludec Versus Insulin Glargine in Subjects With Type 2 Diabetes at High Risk of Cardiovascular Events) cardiovascular outcomes trials (23,24).

The main limitation in this trial was its open-label design, which may have affected reporting of AEs and other results. However, it was necessary to avoid additional placebo injections with IDegLira, which were deemed an unacceptable burden to patients.

Trial design meant patients treated with basal-bolus continued IGlar U100, whereas those treated with IDegLira transferred from IGlar U100 to degludec as a basal insulin. The comparison of different basal insulins may be seen as a limitation as they have different pharmacokinetic properties (25). Notably, degludec has an efficacy similar to IGlar U100 in terms of HbA_1c lowering (24). However, IGlar U100 was the chosen comparator because it was the most prescribed basal insulin at the time of study design.

Additionally, this trial used more stringent titration targets than some current guidelines. The titration target for preprandial SMPG (4.0–6.0 mmol/L [72–108 mg/dL]) with IAsp was lower than the ADA-recommended target of 4.4–7.2 mmol/L (80–130 mg/dL) (20). Thus, higher doses of preprandial insulin may have been used in this trial, contributing to the difference in total daily insulin dose, weight, and hypoglycemia at EOT between arms. Nonetheless, a similar significant lowering of nocturnal hypoglycemia was observed with IDegLira versus basal-bolus when fasting treatment targets were lower than ADA-recommended targets, but there was no difference in FPG levels at EOT between treatment arms.

Medication compliance may be higher in this study than might be expected in the real world, where compliance decreases as the number of insulin injections increases (26). This trial contains real-world features, including the choice of basal comparison, no specific meal plan, and open-label design. Real-world studies are warranted to determine whether the results are replicated in clinical practice.

IDegLira provides an efficacious intensification option with noninferior glycemic control versus basal-bolus in patients with type 2 diabetes on 20–50 units of basal insulin with no renal impairment and HbA_1c levels of 7.0–10.0%. Compared with basal-bolus, IDegLira offers an alternative well-tolerated treatment with fewer injections, doses taken independently of meals, lower total daily insulin dose, reduced monitoring, weight loss, and reduced rate of hypoglycemic episodes. This alternative might help overcome the inertia that currently leaves many patients with poor glycemic control.

Acknowledgments. The authors thank the investigators, research coordinators, and patients in the trial as well as Bue F. Ross Agner and Jakob Langer (both of Novo Nordisk A/S) for their review and input to the manuscript. The authors also thank Victoria Stone and Germanicus Hansa-Wilkinson (Watermeadow Medical, an Ashfield Company, Oxford, U.K.) for providing medical writing and editorial support.

Funding and Duality of Interest. This trial was funded by Novo Nordisk A/S in accordance with Good Publication Practice (GPP3) guidelines (www.ismpp.org/gpp3). L.K.B. has participated in advisory panels for Novo Nordisk and is on the speakers’ bureau for Novo Nordisk. A.D. has participated in research trials as a principal investigator for Novo Nordisk, AstraZeneca, Eli Lilly and Company, Pfizer, and Sanofi. D.G. has received research support from Novo Nordisk, Eli Lilly and Company, Boehringer Ingelheim, Medtronic, Johnson & Johnson, and Sanofi. A.O. has received research support from Novo Nordisk and is on the speakers’ bureau for Novo Nordisk. H.W.R. has participated in advisory panels for AstraZeneca, Boehringer Ingelheim, Eli Lilly and Company, Janssen, Merck, Novo Nordisk, Sanofi, and Regeneron; is a consultant for AstraZeneca, Boehringer Ingelheim, Eli Lilly and Company, Janssen, Merck, Novo Nordisk, Sanofi, and Regeneron; has received research support from AstraZeneca, Bristol-Myers Squibb, Boehringer Ingelheim, Eli Lilly and Company, Janssen, Lexicon, Merck, Novo Nordisk, Sanofi, and Regeneron; and is on the speakers’ bureau for AstraZeneca, Eli Lilly and Company, Merck, Novo Nordisk, and Sanofi. N.T. has participated in advisory panels for Merck, AstraZeneca, Sanofi, Novo Nordisk, ELPEN, Eli Lilly and Company, Boehringer Ingelheim, and Novartis and has received research support from Merck, Eli Lilly and Company, Novo Nordisk, Sanofi, Pfizer, AstraZeneca, Janssen, Cilag, GlaxoSmithKline, and Novartis. R.G. is an employee of Novo Nordisk. N.H. is an employee of Novo Nordisk. E.J. has participated in advisory panels for Janssen, Eli Lilly and Company, Merck, and Novo Nordisk; has received research support from Janssen, Eli Lilly and Company, Merck, Novo Nordisk, Pfizer, and Sanofi; and is on the speakers’ bureau for Janssen, Eli Lilly and Company, Merck, and Novo Nordisk.

Author Contributions. L.K.B., A.D., D.G., A.O., and H.W.R. participated in drafting the manuscript or revising it critically for important intellectual content. N.T., R.G., N.H., and E.J. made substantial contributions to conception and design, and/or acquisition of data, and/or analysis and interpretation of data and participated in drafting the manuscript or revising it critically for important intellectual content. All authors gave final approval of the version of the manuscript to be submitted and any revised version. Novo Nordisk was involved in the trial design and protocol development, provided logistical support, and obtained the data, which were evaluated jointly by the authors and the sponsor. L.K.B. 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 orally at the 77th Scientific Sessions of the ADA, San Diego, CA, 9–13 June 2017, and as a poster at the 53rd Annual Meeting of the European Association for the Study of Diabetes, Lisbon, Portugal, 11–15 September 2017.