A Randomized, Double-Blind, Phase 3 Trial of Triple Therapy With

Diabetes Care 1

Chantal Mathieu,1 Aurelian Emil Ranetti,2 A Randomized, Double-Blind, Danshi Li,3 Ella Ekholm,4 William Cook,5 CARE/EDUCATION/NUTRITION/PSYCHOSOCIAL CLIN Boaz Hirshberg,6 Hungta Chen,5 Phase 3 Trial of Triple Therapy Lars Hansen,3 and Nayyar Iqbal5 With Dapagliflozin Add-on to Saxagliptin Plus Metformin in Type 2 Diabetes DOI: 10.2337/dc15-0779

OBJECTIVE To compare the efﬁcacy and safety of treatment with dapagliﬂozin versus that with placebo add-on to saxagliptin plus metformin in patients whose type 2 diabetes is inadequately controlled with saxagliptin plus metformin treatment.

RESEARCH DESIGN AND METHODS

Patients receiving treatment with stable metformin (stratum A; screening HbA1c level 8.0–11.5% [64–102 mmol/mol]) or stable metformin and a dipeptidyl peptidase-4 (DPP-4) inhibitor (stratum B; HbA1c 7.5–10.5% [58–91 mmol/mol]) for ‡8 weeks received open-label saxagliptin 5 mg/day and metformin for 16 weeks (stratum A) or 8 weeks (stratum B, saxagliptin replaced any DPP-4 inhibitor).

Patients with inadequate glycemic control (HbA1c 7–10.5% [53–91 mmol/mol]) were randomized to receive placebo or dapagliflozin 10 mg/day plus saxagliptin and metformin. The primary end point was the change in HbA1c from baseline to week 24. Secondary end points included fasting plasma glucose (FPG) level, 2-h 1Katholieke Universiteit Leuven, Leuven, Belgium postprandial glucose (PPG) level, body weight, and proportion of patients achieving 2Carol Davila University of Medicine and Pharmacy, Bucharest, Romania an HbA1c level of <7% (53 mmol/mol). 3Bristol-Myers Squibb, Princeton, NJ 4 RESULTS AstraZeneca, Molndal, Sweden 5 fl AstraZeneca, Gaithersburg, MD Treatment with dapagli ozin add-on to saxagliptin plus metformin resulted in a 6MedImmune, Gaithersburg, MD 2 2 2 2 greater mean HbA1c reduction than placebo ( 0.82% vs. 0.10% [ 9vs. 1.1 Corresponding author: Chantal Mathieu, mmol/mol], P < 0.0001). Significantly greater reductions in FPG level, 2-h PPG [email protected]. level, and body weight were observed, and more patients achieved an HbA1c Received 13 April 2015 and accepted 13 July level of <7% (53 mmol/mol) with treatment with dapagliflozin versus placebo. 2015. Adverse events were similar across treatment groups, with a low overall risk of Clinical trial reg. no. NCT01646320, clinicaltrials hypoglycemia (∼1%). Genital infections developed in more patients with dapagliflozin .gov. treatment (5%) than with placebo (0.6%). This article contains Supplementary Data online at http://care.diabetesjournals.org/lookup/ CONCLUSION suppl/doi:10.2337/dc15-0779/-/DC1. Triple therapy with dapagliflozin add-on to saxagliptin plus metformin improves © 2015 by the American Diabetes Association. Readers may use this article as long as the work is glycemic control and is well tolerated in patients whose type 2 diabetes was properly cited, the use is educational and not for inadequately controlled with saxagliptin plus metformin therapy. profit, and the work is not altered. Diabetes Care Publish Ahead of Print, published online August 5, 2015 2 Dapagliflozin Add-on to Saxagliptin-Metformin Diabetes Care

Type 2 diabetes is a complex, chronic weight reduction. This phase 3 study saxagliptin 5 mg/day. They received 8 disease characterized by hyperglycemia was designed to compare the safety weeks of open-label treatment with (1) and an increased risk of microvascular and efficacy of dapagliflozin therapy this regimen. and macrovascular complications (2). versus placebo add-on to saxagliptin Major exclusion criteria for screening Because of the progressive nature of plus metformin therapy in patients with and the open-label treatment period type 2 diabetes (3), achieving and type 2 diabetes who had inadequate included pregnancy, cardiovascular maintaining glycemic treatment goals glycemic control with saxagliptin plus events within 3 months of screening, may be challenging (4,5). Metformin, metformin therapy. an estimated glomerular filtration rate if tolerated and not contraindicated, of ,60 mL/min/1.73 m2 or a serum cre- is generally considered the drug of RESEARCH DESIGN AND METHODS atinine level of $1.5 mg/dL in men or choice for initiating oral therapy in pa- Study Design $1.4 mg/dL in women, microscopic he- tients with type 2 diabetes (6–8). As This was a multicenter, randomized, double- maturiawithnoknowncauseinmen, glycemic control worsens over time, blind, placebo-controlled, parallel-group, and significant hepatic disease. Patients other classes of antidiabetes medica- phase 3 study (Clinical trial reg. no. were also excluded if they received any tions with different mechanisms of ac- NCT01646320). It was designed and antidiabetes medication, other than tion are typically added to metformin monitored in accordance with the ethical metformin and DPP-4 inhibitors, for in a stepwise manner as dual or triple principles of Good Clinical Practice as .14 days during the 12 weeks before therapy as recommended by current defined by the International Confer- screening. Patients with uncontrolled treatment guidelines (6–8). ence on Harmonization and the Decla- hypertension (systolic blood pressure Clinical practice guidelines recom- ration of Helsinki. Institutional review $160 mm Hg and/or diastolic blood mend a patient-centered approach boards or ethics committees at each pressure $100 mm Hg) were allowed when selecting add-on therapy for pa- study site approved the protocol, to enter the open-label period provided tients whose blood glucose levels are and all patients gave written informed that their antihypertensive therapy was not controlled with metformin therapy consent. adjusted appropriately; if their blood (6–8). Among these options are dipep- The study design consisted of a pressure fell below these limits, they tidyl peptidase-4 (DPP-4) inhibitors and screening and open-label treatment could be considered for randomization sodium–glucose cotransporter 2 (SGLT2) period followed by a randomized, to double-blind treatment. inhibitors (6,7). Saxagliptin, a DPP-4 in- 24-week, short-term, double-blind For the open-label period, the enroll- hibitor, enhances glucose-mediated treatment period (Supplementary Fig. ment of patients with an HbA1c level at insulin secretion and suppresses gluca- 1); this was followed with a long-term the lower end of the inclusion range (8– gon secretion by inhibiting the rapid extension of an additional 28 weeks 9% [64–75 mmol/mol] for patients in degradation of glucagon-like peptide-1 for a total of 52 weeks of triple therapy. stratum A, 7.5–8.5% [58–69 mmol/mol] (9). Dapagliflozin, an SGLT2 inhibitor, The results of the long-term extension for patients in stratum B) was limited acts independently of insulin and inhibits will be presented in a subsequent re- to ;50%. The other ;50% of patients renal glucose reabsorption, thus increas- port. Two groups of patients ($18 years had an HbA1c level of .9–11.5% (75– ing urinary glucose excretion and reduc- of age) with type 2 diabetes and inade- 102 mmol/mol) (stratum A) or .8.5– ingplasmaglucoseconcentrations quate glycemic control were included in 10.5% (69–91 mmol/mol) (stratum B). (10). By acting on distinct and separate the open-label treatment period based At week 210 and week 22, if the fasting pathways that affect different defects on DPP-4 inhibitor use. Patients in stra- plasma glucose (FPG) level was .270 involved in the pathophysiology of type tum A had a glycated hemoglobin mg/dL, the patient could not be ran- 2 diabetes, saxagliptin and dapagliflozin (HbA1c)levelof8.0–11.5% (64–102 domized and was discontinued from provide a treatment option that has a mmol/mol) at screening and were re- the study. complementary mechanism of action ceiving stable metformin therapy (im- For inclusion in the 24-week double- whenaddedtometformin(11). mediate release [IR] or extended blind treatment period, patients in both Saxagliptin and dapagliflozin, when release [XR] $1,500 mg/day) for at least stratum A and B had to have an HbA1c used as monotherapy or as an add-on 8 weeks before screening. These pa- level of 7.0–10.5% (53–91 mmol/mol) to metformin therapy in patients with tients were switched to the nearest when assessed at week 22. Following type 2 diabetes, improve glycemic con- lowerorhighermultipleofmetforminIR the open-label period, eligible patients trol, and have favorable safety and tol- 500-mg tablets and saxagliptin 5 mg/day were randomly assigned by an interac- erability profiles (12–16). Moreover, for 16 weeks of open-label treatment. tive voice response system in a centrally both agents have a low propensity for Patients in stratum B had an HbA1c level blocked 1:1 ratio within each stratum to hypoglycemia and are weight neutral of 7.5–10.5% (58–91 mmol/mol) and double-blind treatment with placebo or (saxagliptin) (13,15,16) or cause weight were receiving stable metformin (IR or dapagliflozin 10 mg/day for 24 weeks. reduction of 2–3 kg (dapagliflozin) XR $1,500 mg/day) and a DPP-4 inhibi- Patients completing the 24-week, (12,17). Thus, the addition of dapagliflozin tor at the maximum approved dose for short-term, double-blind period could to saxagliptin and metformin results in a at least 8 weeks before the screening enter a 28-week long-term extension. triple combination therapy that in- visit.Thesepatientswereswitchedto cludes components with complemen- thenearestlowerorhighermultiple Efficacy End Points tary mechanisms of action, a low risk of metformin IR 500-mg tablets, and The primary end point was the mean of hypoglycemia, and the potential for any DPP-4 inhibitor was replaced by change from baseline in HbA1c level care.diabetesjournals.org Mathieu and Associates 3

after 24 weeks of double-blind treat- probability of drug-induced liver was summarized by treatment group ment with dapagliflozin versus placebo injury as the cause for liver-related and analyzed using previously published add-on to saxagliptin plus metformin. abnormalities. methods (18,19). Secondary end points were the mean Statistical Analysis change from baseline at 24 weeks in RESULTS FPG level, 2-h postprandial glucose With 133 patients per treatment group, Patients (PPG) level following a liquid meal toler- there was 90% power to detect a differ- For the 24-week short-term phase of this ance test (MTT), and body weight, ence in the mean changes from baseline study, the first patient visit occurred on and the mean proportion of patients in HbA level at 24 weeks of 0.4% (4.4 1c 21 September 2012, and the last patient achieving a therapeutic glycemic re- mmol/mol) between the dapagliflozin visit occurred on 7 August 2014. Of the sponse, definedasanHbA level of plus saxagliptin plus metformin treat- 1c 818 enrolled patients, 483 entered the ,7.0% (53 mmol/mol), after 24 weeks. ment group and the placebo plus saxa- open-label period, and 83% (402) com- The composition of the liquid MTT was gliptin plus metformin treatment group, pleted this period (Supplementary Fig. dependent on the investigational site assuming an SD of 1.0%. Assuming that 2). Three hundred twenty patients (219 and consisted of 360–375 kcal, 14.0– 5% of patients would not have a post- from stratum A and 101 from stratum 28.2 g of protein, 10.5–14.0 g of baseline assessment, a total of ;280 B) were eligible for randomization and fat, 42–45 g of carbohydrates; 16.8– patients (140 patients per treatment entered the 24-week, double-blind treat- 22.0 g of sugars as a component of car- arm) needed to be randomized. Because ment period, and 94.1% (301) completed bohydrates. Other end points included patients in stratum B were assumed to the short-term period of the study. Pa- the proportion of patients rescued or have better glycemic control based on tient demographics and baseline charac- discontinued from the study for lack of their prior exposure to dual therapy, ;33% teristics were balanced across treatment efficacy, the change from baseline in the of randomized patients were expected groups (Table 1). Most patients (93%) PPG area under the concentration-time to be from stratum B. were white, and 54% were women. The curve from 0 to 180 min (AUC – ) The primary efficacy data set included 0 180 min mean age was 55 years. The mean dura- during a liquid MTT, and the change from all randomized patients who received tion of diabetes was 7.6 years, the mean baseline in serum lipid levels. one or more doses of study medication HbA level was 8.2% (66 mmol/mol), and during the double-blind treatment pe- 1c the mean FPG level was 178 mg/dL at Safety riod. The primary efficacy analysis was baseline. The proportion of patients Safety and tolerability were evaluated performed using a longitudinal re- from each stratum was also balanced in based on adverse events (AEs), hypogly- peated-measures analysis with terms the two treatment groups. The baseline cemia, laboratory abnormalities, and vi- for baseline value, treatment group, HbA level at randomization was similar tal signs. Hypoglycemia episodes were time, stratum, the interaction of treat- 1c between the two treatment groups classified as minor (symptomatic or ment group and time, and the interac- (8.17% [66 mmol/mol] and 8.24% [67 asymptomatic with plasma glucose con- tion of baseline value and time, mmol/mol]; Table 1). centration of ,63 mg/dL, regardless of including only observations made be- the need for external assistance), major fore rescue. Point estimates and 95% Efficacy (symptomatic requiring third-party as- CIs were calculated for the adjusted The addition of dapagliflozin to saxagliptin sistance because of severe impairment mean changes within each treatment plus metformin therapy resulted in a sig- in consciousness or behavior with or group and for the differences in ad- nificantly greater adjusted mean 6 SE without a plasma glucose concentration justed mean changes between treat- reduction from baseline in HbA1c of ,54 mg/dL and prompt recovery af- ment groups. level at 24 weeks (20.82% 6 0.07% ter glucose or glucagon administration), The interpretation of the statistical [29.0 6 0.8 mmol/mol]) compared with and other (suggestive episode not meet- significance of treatment comparisons placebo (20.10% 6 0.07% [21.1 6 0.8 ing the criteria for major or minor). AEs for each secondary efficacy end point mmol/mol]) (Fig. 1A, Table 2). The differ- of special interest included severe cuta- was performed using a stepwise proce- ence (95% CI) in change from baseline neous events, decreased lymphocyte dure to protect the overall type I error in HbA1c level between dapagliflozin count, decreased thrombocyte count, rate. Analysis of the mean change from and placebo add-on to saxagliptin plus opportunistic infection, pancreatitis, baseline at week 24 for FPG level and metformin therapy was 20.72% ([95% pancreatic cancer, fracture, severe hy- body weight was performed using the CI 20.91%, 20.53%] 27.9 mmol/mol persensitivity, worsening renal function, same longitudinal repeated-measures [95% CI 29.9, 25.8], P , 0.0001). genital infections, urinary tract infec- model as for the primary efficacy end Thus, the primary end point of statistical tions, bladder neoplasm, breast neo- point. The analysis of the mean change superiority of dapagliflozin add-on to plasm, volume depletion, and heart from baseline at week 24 for 2-h PPG saxagliptin plus metformin therapy ver- failure. Suspected cardiovascular AEs level during a liquid MTT was based on sus placebo add-on to saxagliptin plus were blindly adjudicated by a clinical an ANCOVA model using last observa- metformin therapy was met. A clear event committee managed by the Mon- tion carried forward methodology with separation in HbA1c level was observed treal Heart Institute. Reported liver in- terms for treatment group, stratum, and between the treatment groups starting jury AEs were blindly adjudicated baseline value in the model. The propor- at week 6, the earliest time point as- by an independent hepatic adjudi- tion of patients achieving an HbA1c level sessed, and continuing to the end of cation committee who determined the of ,7.0% (,53 mmol/mol) at 24 weeks the study (Fig. 1B). 4 Dapagliflozin Add-on to Saxagliptin-Metformin Diabetes Care

Table 1—Demographics and baseline characteristics at randomization Placebo add-on to saxagliptin Dapagliﬂozin add-on to saxagliptin plus metformin (n =160) plus metformin (n = 160) Total (n =320) Age, years 55.0 6 9.6 55.2 6 8.6 55.1 6 9.1 Women 84 (52.5) 90 (56.3) 174 (54.4) Race White 147 (91.9) 150 (93.8) 297 (92.8) African American 10 (6.3) 8 (5.0) 18 (5.6) Asian 1 (0.6) 1 (0.6) 2 (0.6) Other 2 (1.3) 1 (0.6) 3 (0.9) BMI, kg/m2 32.2 6 5.3 31.2 6 4.7 31.7 6 5.1 Duration of diabetes, years 8.0 6 6.6 7.2 6 5.7 7.6 6 6.1

HbA1c %8.176 0.98 8.24 6 0.96 8.20 6 0.97 mmol/mol 66 6 10.7 67 6 10.5 66 6 10.6

HbA1c category ,8% (64 mmol/mol) 75 (46.9) 69 (43.1) 144 (45.0) $8% and ,9% ($64 and ,75 mmol/mol) 53 (33.1) 51 (31.9) 104 (32.5) $9% (75 mmol/mol) 32 (20.0) 40 (25.0) 72 (22.5) FPG, mg/dL 177 6 46.8 179 6 48.9 178 6 47.8 2-h PPG, mg/dL 243 6 57.5 (n = 154) 242 6 60.9 (n = 156) 242 6 59.1 (n =310) Fasting C-peptide, ng/mL 2.6 6 1.1 2.5 6 1.1 2.6 6 1.1 eGFR, mL/min/1.73 m2 91.6 6 23.2 93.5 6 20.8 92.5 6 22.0 Randomization stratiﬁcation Stratum A 109 (68.1) 110 (68.8) 219 (68.4) Stratum B 51 (31.9) 50 (31.3) 101 (31.6) Data are mean 6 SD or n (%), unless otherwise indicated. eGFR, estimated glomerular ﬁltration rate calculated by MDRD formula.

Therewerestatisticallysignificantly vs. placebo add-on 213.6% [95% CI therapy was numerically greater larger reductions from baseline to 219.4%, 27.7%]). The adjusted mean (20.96% [95% CI 21.19%, 20.72%] in week 24 in the dapagliflozin add-on (SE) change from baseline at week 24 patients who received a DPP-4 inhib- to saxagliptin plusmetformingroup in PPG AUC0–180 min was greater with itor and metformin before screening compared with the placebo add-on to dapagliflozin add-on to saxagliptin plus (stratum B) than in those patients who saxagliptin plus metformin group in metformin 211,353 mg/dL/min (676.3 received only metformin before screen- FPG level (difference for dapagliflozin mg/dL/min) than with placebo add-on ing (20.75% [95% CI 20.94%, 20.59%], vs. placebo add-on 228 mg/dL [95% CI to saxagliptin plus metformin 25,162 stratum A). A numerically greater change 235.4, 219.6 mg/dL], P , 0.0001), 2-h mg/dL/min (SE 681.7 mg/dL/min) (dif- from baseline in HbA1c level was also PPG level (difference for dapagliflozin ference for dapagliflozin vs. placebo, seen in the placebo group in stratum B vs. placebo add-on 236 mg/dL [95% CI 26,191 mg/dL/min [95% CI 28,027.1, (20.24% [95% CI 20.48%, 0.01%]) 246.3, 224.7 mg/dL], P , 0.0001), and 24,354.6 mg/dL/min]). The mean per- compared with stratum A (20.03% body weight (difference for dapagliflozin centage changes from baseline at week [95% CI 20.20%, 0.13%]). vs. placebo add-on 21.5 kg [95% CI 24 in fasting serum lipids were small and 22.12, 20.89 kg], P , 0.0001) (Table similar in the two treatment groups Safety 2). In addition, a statistically significantly (Supplementary Table 1). The proportion of patients with AEs was higher proportion of patients in the In a post hoc subgroup analysis of the similar across treatment groups (Table dapagliflozin add-on to saxagliptin plus change from baseline in HbA1c level in the 3). Few patients were discontinued metformin group (38.0%) achieved an two strata, the difference (95% CI) in fromthestudy,andtherewereno HbA1c level of ,7% at week 24 than in the adjusted mean change from baseline deaths. The most common AEs in both the placebo add-on to saxagliptin plus in HbA1c level between dapagliflozin treatment groups ($5% of patients) metformin group (12.4%) (Table 2). add-on to saxagliptin plus metformin ther- were headache, urinary tract infections, Fewer patients were rescued or dis- apyandplaceboadd-ontosaxagliptin and influenza. The proportion of pa- continued from the study for lack of plus metformin therapy was identical tients with serious AEs (SAEs) was small glycemic control in the dapagliflozin (stratum A 20.72% [95% CI 20.94%, and similar in both treatment groups. add-on to saxagliptin plus metformin 20.49%], stratum B 20.72% [95% CI Only one SAE (thrombocytopenia in group (3 of 160 patients) compared with 21.06%, 20.38%]). However, the ad- the dapagliflozin add-on to saxagliptin the placebo add-on to saxagliptin plus justed mean change (95% CI) from base- plus metformin group) was considered metformin group (24 of 160 patients; ad- line in HbA1c level with dapagliflozin by the investigators to be related to justed percent difference for dapagliflozin add-on to saxagliptin plus metformin study medication. Discontinuations because care.diabetesjournals.org Mathieu and Associates 5

add-on to saxagliptin plus metformin therapy had elevated levels of alanine aminotransferase and/or aspartate aminotransferase. There were no AEs of pancreatitis, fracture, severe cuta- neous events, opportunistic infection, hypersensitivity, bladder neoplasm, pancreatic cancer, decreased lymphocyte count, or hypotension, dehydration, and hypovolemia. There were small decreases from baseline to week 24 in systolic blood pressure (adjusted mean change 21.9 vs. 2.0 mmHg; difference 23.9 [95% CI 26.6, 21.2], P = 0.005) and diastolic blood pressure (21.7 vs. 0.3 mmHg; 22.0 [95% CI 23.7, 20.3], P = 0.024) with dapagliﬂozin add-on to saxagliptin plus metformin therapy compared with placebo add-on to saxagliptin plus metformin therapy.

CONCLUSIONS Acknowledging the progressive nature of type 2 diabetes, with an underlying failure of b-cell function, current clinical guidelines (6,7) recommend combination therapies with glucose-lowering agents. In the absence of contraindications, metformin remains the drug of ﬁrst choice in patients who can tolerate it. Treatment guidelines recommend a patient-centered approach when de- ciding what should be added when metformin monotherapy fails to achieve or sustain glycemic control (6,7). DPP-4

Figure 1—Change in HbA1c level. Adjusted mean change in HbA1c level from baseline to 24 weeks inhibitors are becoming a preferred sec- (A) and time course of adjusted mean change from baseline in HbA1c level (B). *Randomized ond agent for many prescribers because patients with baseline and one or more postbaseline measurements. †Randomized patients with of their low risk of hypoglycemia, lack of ‡ baseline and week 24 measurement. Randomized patients who received one or more doses of weight gain, and once-daily administra- double-blind medication. DAPA, dapagliflozin; MET, metformin; PBO, placebo; SAXA, saxagliptin. tion (20). However, mechanistically, in addition to their effect to inhibit gluca- of AEs were higher with dapagliflozin add-on to saxagliptin plus metformin gon secretion, DPP-4 inhibitors rely on add-on to saxagliptin plus metformin therapy (0.6%); the occurrence of uri- residual b-cell function to lower hyper- therapy (5.0%) than with placebo add- nary tract infections was similar for glycemia (9). Although some data sug- on to saxagliptin plus metformin ther- dapagliflozin add-on to saxagliptin plus gest that these agents may have the apy (1.3%). In the dapagliflozin add-on metformingroup(5%)andplacebo ability to slow the deterioration of to saxagliptin plus metformin group, add-on to saxagliptin plus metformin b-cell function (21), the progressive na- reasons for discontinuations were group (6.3%). Hypoglycemia episodes ture of diabetes will eventually lead to thrombocytopenia (n =1),breastcancer were infrequent, and there were no b-cell failure (3) and loss of glycemic (n = 1), heart failure (n = 1), recurrent cases of major hypoglycemia. Two cardio- control, regardless of which antidiabetes vulvovaginal and urinary tract infec- vascular events confirmed by adjudica- agent is used. Thus, further combinations tions (n =1),adecreaseinglomerular tion (ventricular tachycardia and heart with additional glucose-lowering thera- filtration rate (n =1),renalimpairment failure) occurred in the dapagliflozin pies are needed to maintain glycemic (n = 1), and pollakiuria (n =2).Ahigher add-on to saxagliptin plus metformin control. Although combination inject- proportion of patients, all of whom group. One patient in the dapagliflozin able therapy with insulin with or without were women, receiving dapagliflozin add-on to saxagliptin plus metformin glucagon-like peptide 1 receptor agonist add-on to saxagliptin plus metformin group had elevated alanine aminotrans- is recommended in patients with severe therapy had genital infections (5%) ferase and/or aspartate aminotrans- hyperglycemia (HbA1c level $10%) (7), compared with those receiving placebo ferases. Three patients receiving placebo for patients with less severe disease, the 6 Dapagliflozin Add-on to Saxagliptin-Metformin Diabetes Care

Table 2—Adjusted mean change from baseline at 24 weeks for primary and secondary efﬁcacy end points Placebo add-on to saxagliptin plus Dapagliﬂozin add-on to saxagliptin metformin (n =160) plus metformin (n =160)

HbA1c Baseline mean 6 SD, % [mmol/mol] 8.16 6 0.99 [66 6 10.8] 8.24 6 0.97 [67 6 10.6] Change from baseline, % [mmol/mol] 20.10 (20.24, 0.04) 20.82 (20.96, 20.69) [21.1 (22.6, 0. 4)] [29(210.5, 27.5)] Difference dapagliflozin vs. placebo, % [mmol/mol] 20.72 (20.91, 20.53) [27.9 (29.9, 25.8)] P , 0.0001 FPG, mg/dL Baseline, mean 6 SD 177 6 46.8 179 6 48.7 Change from baseline 25(211.1, 0.60) 233 (238.3, 227.2) Difference dapagliflozin vs. placebo 228 (235.4, 219.6) P , 0.0001 2-h PPG, mg/dL Baseline, mean 6 SD 241 6 57.1 240 6 60.9 Change from baseline 238 (246.1, 229.9) 274 ( 281.5, 265.5) Difference dapagliflozin vs. placebo 236 (246.3, 224.7) P , 0.0001 Body weight, kg Baseline, mean 6 SD 88.2 6 18.1 85.8 6 18.4 Change from baseline 20.4 (20.86, 0.04) 21.9 (22.34, 21.48) Difference dapagliflozin vs. placebo 21.5 (22.12, 20.89) P , 0.0001

Patients with HbA1c ,7% (53 mmol/mol) x/n 21/158 58/158 Adjusted % 12.4 (7.0, 17.9) 38.0 (30.9, 45.1) Difference vs. saxagliptin plus metformin, % 25.5 (16.7, 34.4) P , 0.0001 Data are adjusted mean change from baseline (95% CI), unless otherwise indicated. PPG, postprandial plasma glucose; x/n, number of responders/ number of patients with nonmissing baseline and week 24 values (LOCF, last observation carried forward).

recently introduced class of SGLT2 inhib- to saxagliptin plus metformin compared in the latter study had lower baseline itors offers the advantage of reducing with placebo add-on to saxagliptin plus HbA1c levels (7.8–7.9% [62–63 mmol/mol]) hyperglycemia independent of b-cell metformin in patients with type 2 diabe- than patients in the current study (8.2% [66 function and circulating insulin levels tes who were not at a glycemic goal mmol/mol]), which may account for the (22) and can thus, theoretically, be com- HbA1c level of ,7% (53 mmol/mol), greater HbA1c response observed in this bined with all other glucose-lowering had poor glycemic control, and remained study. agents. In addition, adding an SGLT2 in- in the HbA1c range of 7–10.5% (53–91 Triple therapy with dapagliflozin add- hibitor to a combination therapy with mmol/mol) after open-label treatment on to saxagliptin plus metformin was metformin and a DPP-4 inhibitor results with saxagliptin and metformin. well tolerated, and the proportion of in a combination that works complemen- We demonstrate that treatment with patients with AEs was similar in the tarily to address several of the pathologic dapagliflozin add-on to saxagliptin plus two treatment groups during the 24- defects of type 2 diabetes (11), leading to metformin improved all glycemic week double-blind period. The overall clinically relevant reductions in hyper- measures in this study (fasting and post- safety observed in the current study glycemia in patients whose glycemic con- prandial glycemia), resulting in a statis- was similar to that in the dapagliflozin trol has not been maintained with two tically significantly larger reduction in add-on to sitagliptin with or without oral agents and without the burden of HbA1c from baseline. This difference be- metformin study (23) and the dual hypoglycemia or weight gain that one came apparent from the earliest time add-on of dapagliflozin and saxagliptin would expect with sulfonylureas or insu- point at which HbA1c was measured to metformin study (26). This study con- lins (23,24). Moreover, a DPP-4 inhibitor (6 weeks), and HbA1c levels remained firms the increased risk for genital infec- may reduce the increase in glucagon se- stable throughout the study. The de- tions with dapagliflozin therapy (5.0% cretion induced by SGLT2 inhibitors (25). gree of HbA1c lowering observed in compared with 0.6% in placebo-treated Although major clinical practice guide- this study (placebo-corrected, 20.72% patients) as observed in previous stud- lines recommend the use of such triple [27.9 mmol/mol]) was greater than ies (14,17,27); however, no increased therapy based on HbA1c levels, evidence that seen in a similarly designed study risk in urinary tract infections was ob- from randomized clinical trials remains (23) in a cohort of patients receiving served (5.3% vs. 6.0% in placebo-treated scarce (7,8). The purpose of this study treatment with dapagliflozin add-on to patients). was to assess the efficacy and safety of sitagliptin plus metformin (20.4% These findings with triple therapy of triple therapy with dapagliflozin add-on [24.4 mmol/mol]). However, the patients dapagliflozin add-on to saxagliptin plus care.diabetesjournals.org Mathieu and Associates 7

Table 3—AEs Placebo add-on to saxagliptin plus Dapagliflozin add-on to saxagliptin metformin (n =160) plus metformin (n =160) At least 1 AE 94 (58.8) 90 (56.3) At least 1 SAE 3 (1.9) 5 (3.1) AE leading to discontinuation 2 (1.3) 8 (5.0) SAE leading to discontinuation 2 (1.3) 3 (1.9) Most common AEs ($5% of patients) Headache 9 (5.6) 9 (5.6) Influenza 8 (5.0) 8 (5.0) Diarrhea 8 (5.0) 3 (1.9) AEs of special interest Urinary tract infections 10 (6.3) 8 (5.0) Genital infections 1 (0.6) 8 (5.0) GFR decrease 0 2 (1.3) Fractures 2 (1.3) 0 Decreased thrombocyte count 1 (0.6) 2 (1.3) Breast neoplasms 0 1 (0.6) Heart failure 0 2 (1.3) Hypoglycemia* 0 2 (1.3) Major 0 0 Minor 0 0 Other 0 2 (1.3) Adjudicated cardiovascular events 0 2 (1.3) Adjudicated liver AEs 3 (1.9) 1 (0.6) Data are n (%). GFR, glomerular filtration rate.*Hypoglycemia includes minor (symptomatic or asymptomatic with plasma glucose concentration ,63 mg/dL, regardless of need for external assistance), major (symptomatic requiring third-party assistance due to severe impairment in consciousness or behavior with plasma glucose concentration ,54 mg/dL, and prompt recovery after glucose or glucagon administration), and other (suggestive episode not meeting the criteria for major or minor) episodes. Data after the receipt of rescue medication were excluded; one patient receiving placebo reported a minor hypoglycemia episode while receiving glimepiride rescue medication. metformin are generally consistent (14–16) have demonstrated a low risk in patients whose conditions were inade- with the mechanisms of action of of hypoglycemia with saxagliptin or quately controlled with therapy using dapagliflozin and saxagliptin and with dapagliflozin therapy when studied indi- metformin plus a sulfonylurea, the add- results from previous clinical studies in- vidually. Likewise, in the current study, on of sitagliptin or canagliflozin produced vestigating the efficacy and safety of hypoglycemia events were infrequent in reductions in HbA1c levels at 52 weeks of these agents as monotherapy (14–16), the dapagliflozin, saxagliptin, and met- 20.66% (27.2 mmol/mol) and 21.03% as add-on therapy to oral antidiabetes formin triple therapy group, and oc- (211.3 mmol/mol), respectively. Finally, medications (12,13,27–30), or when curred with a frequency similar to that in patients whose conditions were inade- used as triple therapy with dual add- in the placebo group. When comparing quately controlled with therapy using on of saxagliptin and dapagliflozin to the current study to a previous one metformin and pioglitazone, the addition metformin (26). where dapagliflozin was added on to of a DPP-4 inhibitor or SGLT2 inhibitor Current treatment recommendations therapy with metformin plus a sulfo- reduced HbA1c levels by 20.48% to position triple therapy using metformin, nylurea, the mean reductions with 20.77% (25.2 to 28.4 mmol/mol) rela- a DPP-4 inhibitor, and an SGLT2 inhibitor dapagliflozin compared with placebo in tive to placebo at 24–26 weeks (41–45). as an alternative to metformin, plus a HbA1c levels, FPG levels, and body In summary, this study demonstrated sulfonylurea, and a DPP-4 inhibitor or weight were similar, but the rates of that triple therapy using a well- an SGLT2 inhibitor. The major issue hypoglycemia in the study with the sul- tolerated combination of dapagliflozin when using a sulfonylurea in combina- fonylurea combination were much higher add-on to saxagliptin plus metformin in tion therapy is the risk of hypoglycemia than those in the current study (24). patients with type 2 diabetes who were (31). In addition, the use of DPP-4 inhib- The results of the current study are in not at their goal for glycemic control itors with sulfonylureas mechanistically line with the results of several recent resulted in clinically meaningful reduc- relies on the residual b-cell function, studies that have examined the safety tions in HbA1c level, and in a greater pro- whereas the SGLT2 inhibitors work com- and efficacy of DPP-4 inhibitors or portion of patients achieving an HbA1c plementarily with the action of DPP-4 SGLT2 inhibitors as a component of triple level of ,7% (53 mmol/mol) compared inhibitors, reducing hyperglycemia oral therapy, mainly as add-on to metformin with those patients receiving placebo through renal excretion of glucose inde- plus a sulfonylurea. In these studies, add-on to saxagliptin plus metformin pendent of insulin secretion, resulting placebo-corrected changes from baseline therapy. The significantly improved in a lessening of glucotoxicity (32,33) in HbA1c level at 24–26 weeks were in the glycemic control with dapagliflozin add- that relieves some of the burden on range of 20.59% to 20.93% (26.4 to on to saxagliptin plus metformin therapy the failing b-cells. Previous studies 210.2 mmol/mol). In another study (40) was achieved without an increased risk 8 Dapagliflozin Add-on to Saxagliptin-Metformin Diabetes Care

of hypoglycemia and was associated with (UKPDS 35): prospective observational study. placebo-controlled trial. Lancet 2010;375: the additional benefit of body weight re- BMJ 2000;321:405–412 2223–2233 fi duction. Taken together, dapagliflozin 3. Fonseca VA. De ning and characterizing the 18. Tsiatis AA, Davidian M, Zhang M, Lu X. Co- progression of type 2 diabetes. Diabetes Care variate adjustment for two-sample treatment add-on to saxagliptin plus metformin 2009;32(Suppl. 2):S151–S156 comparisons in randomized clinical trials: a prin- therapy provides a new treatment op- 4. Kahn SE, Haffner SM, Heise MA, et al.; ADOPT cipled yet flexible approach. Stat Med 2008;27: tion of triple oral therapy with weight Study Group. Glycemic durability of rosiglitazone, 4658–4677 reduction and low risk of hypoglycemia. metformin, or glyburide monotherapy. N Engl J 19. Zhang M, Tsiatis AA, Davidian M. Improving – fi The 28-week long-term extension of the Med 2006;355:2427 2443 ef ciency of inferences in randomized clinical 5. UK Prospective Diabetes Study (UKPDS) Group. trials using auxiliary covariates. Biometrics study is currently ongoing and will pro- Intensive blood-glucose control with sulphonylureas 2008;64:707–715 vide further characterization regarding or insulin compared with conventional treatment 20. Karagiannis T, Paschos P, Paletas K, the long-term safety of triple therapy and risk of complications in patients with type 2 Matthews DR, Tsapas A. Dipeptidyl peptidase-4 with dapagliflozin add-on to saxagliptin diabetes (UKPDS 33). Lancet 1998;352:837–853 inhibitors for treatment of type 2 diabetes plus metformin. 6. Garber AJ, Abrahamson MJ, Barzilay JI, et al.; mellitus in the clinical setting: systematic review American Association of Clinical Endocrinologists. and meta-analysis. BMJ 2012;344:e1369 AACE comprehensive diabetes management al- 21. Leibowitz G, Cahn A, Bhatt DL, et al. Impact gorithm 2013. Endocr Pract 2013;19:327–336 of treatment with saxagliptin on glycaemic sta- Funding. This study was funded by Bristol- 7. American Diabetes Association. Approaches bility and b-cell function in the SAVOR-TIMI 53 Myers Squibb and AstraZeneca. Medical writing to glycemic treatment. Diabetes Care 2015;38 study. Diabetes Obes Metab 2015;17:487–494 support for the preparation of this manuscript (Suppl. 1):S41–S48 22. Ferrannini E, Solini A. SGLT2 inhibition in was provided by Richard Edwards, PhD, and 8. Cheng AY; Canadian Diabetes Association diabetes mellitus: rationale and clinical pros- – Janet Matsuura, PhD, from Complete Healthcare Clinical Practice Guidelines Expert Committee. pects. Nat Rev Endocrinol 2012;8:495 502 Communications, Inc. (Chadds Ford, PA), with Canadian Diabetes Association 2013 clinical 23. Jabbour SA, Hardy E, Sugg J, Parikh S; Study fl funding from AstraZeneca. practice guidelines for the prevention and man- 10 Group. Dapagli ozin is effective as add-on Duality of Interest. C.M. serves or has served agement of diabetes in Canada. Introduction. therapy to sitagliptin with or without metformin: fi on advisory panels for Novo Nordisk, Sano , Can J Diabetes 2013;37(Suppl. 1):S1–S3 a 24-week, multicenter, randomized, double- Merck Sharp and Dohme, Eli Lilly and Company, 9. Drucker DJ. The biology of incretin hormones. blind, placebo-controlled study. Diabetes Care – Novartis, Bristol-Myers Squibb, AstraZeneca, Cell Metab 2006;3:153–165 2014;37:740 750 fi P zer, Johnson and Johnson, Boehringer 10. Jabbour SA. SGLT2 inhibitors to control 24. Matthaei S, Bowering K, Rohwedder K, Ingelheim, Hanmi and Mannkind, and Katholieke fl glycemia in type 2 diabetes mellitus: a new ap- Grohl A, Parikh S; Study 05 Group. Dapagli ozin Universiteit Leuven, whichhas receivedresearch proach to an old problem. Postgrad Med 2014; improves glycemic control and reduces body support from Novo Nordisk, Sanofi, Merck Sharp weight as add-on therapy to metformin plus 126:111–117 and Dohme, Eli Lilly and Company, Roche, sulfonylurea: a 24-week randomized, double- 11. DeFronzo RA, Hompesch M, Kasichayanula Abbott, and Novartis; and also serves or has blind clinical trial. Diabetes Care 2015;38: S, et al. Characterization of renal glucose reab- served on speakers bureaus for Novo Nordisk, 365–372 sorption in response to dapagliflozin in healthy Sanofi, Merck Sharp and Dohme, Eli Lilly and 25. Hansen L, Iqbal N, Ekholm E, Cook W, subjects and subjects with type 2 diabetes. Di- Company, Astra Zeneca, and Novartis. A.E.R. Hirshberg B. Postprandial dynamics of plasma abetes Care 2013;36:3169–3176 serves on the National Ethical Commission for glucose, insulin, and glucagon in patients with 12. Bailey CJ, Morales Villegas EC, Woo V, Tang the clinical study of dapagliflozin (Endocrinology type 2 diabetes treated with saxagliptin plus W, Ptaszynska A, List JF. Efficacy and safety of Specialist); has received research support form dapagliflozin add-on to metformin therapy. En- dapagliflozin monotherapy in people with type Eli Lilly and Company, Servier, and Pfizer; and docr Pract 2014;20:1187–1197 2 diabetes: a randomized double-blind placebo- serves or has served on speakers bureaus for Alfa 26. Rosenstock J, Hansen L, Zee P, et al. Dual Wasserman,EliLillyandCompany,andAbbott.D.L., controlled 102-week trial. Diabet Med 2015;32: add-on therapy in type 2 diabetes poorly con- – L.H., and N.I. are employees of Bristol-Myers 531 541 trolled with metformin monotherapy: a ran- Squibb. E.E., W.C., B.H., and H.C. are employees of 13. DeFronzo RA, Hissa MN, Garber AJ, et al.; domized double-blind trial of saxagliptin plus fi AstraZeneca. No other potential conflicts of in- Saxagliptin 014 Study Group. The ef cacy and dapagliflozin addition versus single addition of terest relevant to this article were reported. safety of saxagliptin when added to metformin saxagliptin or dapagliflozin to metformin. Dia- Author Contributions. C.M., A.E.R., D.L., E.E., therapy in patients with inadequately con- betes Care 2015;38:376–383 W.C., B.H., L.H., and N.I. interpreted the data, trolled type 2 diabetes with metformin alone. 27. Rosenstock J, Vico M, Wei L, Salsali A, List – reviewed and edited the article, and contributed Diabetes Care 2009;32:1649 1655 JF. Effects of dapagliflozin, an SGLT2 inhibitor, to the discussion. H.C. reviewed the statistics 14. Ferrannini E, Ramos SJ, Salsali A, Tang W, on HbA(1c), body weight, and hypoglycemia risk fl and data, reviewed and edited the article, and List JF. Dapagli ozin monotherapy in type 2 in patients with type 2 diabetes inadequately contributed to the discussion. All authors ap- diabetic patients with inadequate glycemic controlled on pioglitazone monotherapy. proved the final version of the manuscript. C.M. control by diet and exercise: a randomized, Diabetes Care 2012;35:1473–1478 is the guarantor of this work and, as such, had full double-blind, placebo-controlled, phase 3 trial. 28. Chacra AR, Tan GH, Apanovitch A, access to all the data in the study and takes Diabetes Care 2010;33:2217–2224 Ravichandran S, List J, Chen R; CV181-040 Inves- responsibilityfortheintegrityofthedataand the 15. Rosenstock J, Aguilar-Salinas C, Klein E, tigators. Saxagliptin added to a submaximal accuracy of the data analysis. Nepal S, List J, Chen R; CV181-011 Study Inves- dose of sulphonylurea improves glycaemic con- Prior Presentation. Parts of this study were tigators. Effect of saxagliptin monotherapy in trol compared with uptitration of sulphonylurea fi presented in abstract form at the 75th Scienti c treatment-na¨ıve patients with type 2 diabetes. in patients with type 2 diabetes: a randomised Sessions of the American Diabetes Association, Curr Med Res Opin 2009;25:2401–2411 controlled trial. Int J Clin Pract 2009;63:1395– – Boston, MA, 5 9 June 2015. 16. Frederich R, McNeill R, Berglind N, Fleming 1406 D, Chen R. The efficacy and safety of the dipep- 29. Hollander P, Li J, Allen E, Chen R; CV181-013 References tidyl peptidase-4 inhibitor saxagliptin in treat- Investigators. Saxagliptin added to a thiazolidi- 1. Defronzo RA. Banting lecture. From the tri- ment-na¨ıve patients with type 2 diabetes nedione improves glycemic control in patients umvirate to the ominous octet: a new paradigm mellitus: a randomized controlled trial. Diabetol with type 2 diabetes and inadequate control on for the treatment of type 2 diabetes mellitus. Metab Syndr 2012;4:36 thiazolidinedione alone. J Clin Endocrinol Metab Diabetes 2009;58:773–795 17. Bailey CJ, Gross JL, Pieters A, Bastien A, List 2009;94:4810–4819 2. Stratton IM, Adler AI, Neil HA, et al. Associa- JF. Effect of dapagliflozin in patients with type 30. Strojek K, Yoon KH, Hruba V, Elze M, tion of glycaemia with macrovascular and mi- 2 diabetes who have inadequate glycaemic con- Langkilde AM, Parikh S. Effect of dapagliflozin crovascular complications of type 2 diabetes trol with metformin: a randomised, double-blind, in patients with type 2 diabetes who have care.diabetesjournals.org Mathieu and Associates 9

inadequate glycaemic control with glimepiride: saxagliptin as add-on therapy in patients with ongoing metformin and pioglitazone com- a randomized, 24-week, double-blind, placebo- type 2 diabetes and inadequate glycaemic con- bination therapy in a randomized, placebo- controlled trial. Diabetes Obes Metab 2011;13: trol on metformin plus a sulphonylurea. Diabe- controlled, 26-week trial in patients with 928–938 tes Obes Metab 2014;16:443–450 type 2 diabetes. J Diabetes Complications 31. Bennett WL, Maruthur NM, Singh S, et al. 37. Owens DR, Swallow R, Dugi KA, Woerle HJ. 2013;27:177–183 Comparative effectiveness and safety of medi- Efficacy and safety of linagliptin in persons with 42. Bajaj M, Gilman R, Patel S, Kempthorne- cations for type 2 diabetes: an update including type 2 diabetes inadequately controlled by a Rawson J, Lewis-D’Agostino D, Woerle HJ. new drugs and 2-drug combinations. Ann Intern combination of metformin and sulphonylurea: Linagliptin improved glycaemic control without Med 2011;154:602–613 a 24-week randomized study. Diabet Med 2011; weight gain or hypoglycaemia in patients with – 32. Ferrannini E, Muscelli E, Frascerra S, et al. 28:1352 1361 type 2 diabetes inadequately controlled by a Metabolic response to sodium-glucose cotrans- 38. Haring¨ HU, Merker L, Seewaldt-Becker E, combination of metformin and pioglitazone: porter 2 inhibition in type 2 diabetic patients. J et al.; EMPA-REG METSU Trial Investigators. a 24-week randomized, double-blind study. Di- Clin Invest 2014;124:499–508 Empagliflozin as add-on to metformin plus abet Med 2014;31:1505–1514 33. Merovci A, Solis-Herrera C, Daniele G, et al. sulfonylurea in patients with type 2 diabetes: 43. DeFronzo RA, Burant CF, Fleck P, Wilson C, Dapagliflozin improves muscle insulin sensitiv- a 24-week, randomized, double-blind, placebo- Mekki Q, Pratley RE. Efficacy and tolerability of ity but enhances endogenous glucose produc- controlled trial. Diabetes Care 2013;36:3396– the DPP-4 inhibitor alogliptin combined with tion. J Clin Invest 2014;124:509–514 3404 pioglitazone, in metformin-treated patients 34. Hermansen K, Kipnes M, Luo E, Fanurik D, 39. Wilding JP, Charpentier G, Hollander P, with type 2 diabetes. J Clin Endocrinol Metab Khatami H, Stein P; Sitagliptin Study 035 Group. et al. Efficacy and safety of canagliflozin in – Efficacy and safety of the dipeptidyl peptidase-4 patients with type 2 diabetes mellitus in- 2012;97:1615 1622 inhibitor, sitagliptin, in patients with type 2 di- adequately controlled with metformin and 44. Forst T, Guthrie R, Goldenberg R, et al. Ef- fi fl abetes mellitus inadequately controlled on gli- sulphonylurea: a randomised trial. Int J Clin cacy and safety of canagli ozin over 52 weeks mepiride alone or on glimepiride and metformin. Pract 2013;67:1267–1282 in patients with type 2 diabetes on background Diabetes Obes Metab 2007;9:733–745 40. Schernthaner G, Gross JL, Rosenstock J, metformin and pioglitazone. Diabetes Obes 35. Lukashevich V, Del Prato S, Araga M, Kothny et al. Canagliflozin compared with sitagliptin Metab 2014;16:467–477 W. Efficacy and safety of vildagliptin in patients for patients with type 2 diabetes who do 45. Kovacs CS, Seshiah V, Swallow R, et al.; with type 2 diabetes mellitus inadequately con- not have adequate glycemic control with EMPA-REG PIO Trial Investigators. Empagliflozin trolled with dual combination of metformin and metformin plus sulfonylurea: a 52-week improves glycaemic and weight control as add- sulphonylurea. Diabetes Obes Metab 2014;16: randomized trial. Diabetes Care 2013;36: on therapy to pioglitazone or pioglitazone plus 403–409 2508–2515 metformin in patients with type 2 diabetes: 36. Moses RG, Kalra S, Brook D, et al. A random- 41. Fonseca V, Staels B, Morgan JD 2nd, et al. a 24-week, randomized, placebo-controlled tri- ized controlled trial of the efficacy and safety of Efficacy and safety of sitagliptin added to al. Diabetes Obes Metab 2014;16:147–158