Clinical Care/Education/Nutrition/Psychosocial Research ORIGINAL ARTICLE

Canagliflozin Compared With for Patients With Type 2 Who Do Not Have Adequate Glycemic Control With Plus A 52-week randomized trial

1 5 GUNTRAM SCHERNTHANER, MD JACQUELINE YEE, MS (2). With further decline in glycemic con- 2 5 JORGE L. GROSS, MD MASATO KAWAGUCHI, MD 3 5 trol (3,4), the addition of a third oral agent JULIO ROSENSTOCK, MD WILLIAM CANOVATCHEL, MD 4 5 is increasingly common. Currently avail- MICHAEL GUARISCO, MD GARY MEININGER, MD 5 able classes of AHAs, such as dipeptidyl MIN FU, MS peptidase-4 inhibitors, peroxisome proliferator–activated receptor (PPAR)g agonists, and , have distinct OBJECTIVEdTo evaluate the efficacy and safety of canagliflozin, a sodium glucose cotrans- fi fi porter 2 inhibitor, compared with sitagliptin in subjects with inadequately risk/bene tpro les (2,5). A recent posi- controlled with metformin plus sulfonylurea. tion statement by the American Diabetes Association and the European Associa- RESEARCH DESIGN AND METHODSdIn this 52-week, randomized, double-blind, tion for the Study of Diabetes recom- active-controlled, phase 3 study, subjects using stable metformin plus sulfonylurea (N =755) mends individualization of treatment received canagliflozin 300 mg or sitagliptin 100 mg daily. Primary end point was change from for patients and suggests the use of phar- baseline in A1C at 52 weeks. Secondary end points included change in fasting plasma glucose macologic agents with complemen- (FPG) and systolic blood pressure (BP), and percent change in body weight, triglycerides, and tary mechanisms of action in triple HDL cholesterol. Safety was assessed based on adverse event (AE) reports. therapy combinations if A1C targets RESULTSdAt 52 weeks, canagliflozin 300 mg demonstrated noninferiority and, in a subsequent are not attained with dual combination assessment, showed superiority to sitagliptin 100 mg in reducing A1C (21.03% [211.3 mmol/mol] therapy (2). and 20.66% [27.2 mmol/mol], respectively; least squares mean difference between groups, Canagliflozin is an inhibitor of the 20.37% [95% CI, 20.50 to 20.25] or 24.0 mmol/mol [25.5 to 22.7]). Greater reductions in sodium glucose cotransporter 2 (SGLT2) FPG, body weight, and systolic BP were observed with canagliflozin versus sitagliptin (P , in development for the treatment of pa- fl 0.001). Overall AE rates were similar with canagli ozin (76.7%) and sitagliptin (77.5%); incidence tients with type 2 diabetes (6–10). SGLT2 of serious AEs and AE-related discontinuations was low for both groups. Higher incidences of is responsible for the majority of glucose genital mycotic infections and osmotic –related AEs were observed with canagliflozin, which led to one discontinuation. rates were similar in both groups. reabsorption in the (11). Almost all glucose is reabsorbed from the tubules CONCLUSIONSdFindings suggest that canagliflozin may be a new therapeutic tool providing until renal tubular resorptive capacity is better improvement in glycemic control and body weight reduction than sitagliptin, but with in- exceeded and urinary glucose creased genital infections in subjects with type 2 diabetes using metformin plus sulfonylurea. (UGE) ensues; the glucose concentration at which this occurs is referred to as the – Diabetes Care 36:2508 2515, 2013 renal threshold for glucose. Canagliflozin lowers the renal threshold for glucose, atients with type 2 diabetes often is the recommended first-line pharmaco- markedly increasing UGE and thereby re- Prequire combinations of antihyper- logic therapy for type 2 diabetes (1,2). For ducing blood glucose concentrations in glycemic agents (AHAs) to maintain patients who do not achieve or sustain patients with hyperglycemia. The in- glycemic control because of the progres- sufficient glycemic control with met- crease in UGE results in a mild osmotic sive nature of the disease (1,2). Metformin formin, a second AHA is often added diuresis and also provides a net caloric loss (with most patients with type 2 diabetes ccccccccccccccccccccccccccccccccccccccccccccccccc losing an average of 80–120 g/day) (12). From the 1Rudolfstiftung Hospital-Vienna, Vienna, Austria; the 2Department of Internal Medicine, Federal This mechanism of action, distinct from University of Rio Grande do Sul, Porto Alegre, Brasil; the 3Dallas Diabetes and Endocrine Center at Medical the mechanisms of glucose-lowering of City, Dallas, Texas; the 4Stanocola Medical Clinic, Baton Rouge, Louisiana; and 5Janssen Research and current AHA classes and independent of Development, LLC, Raritan, New Jersey. , should provide additive glycemic Corresponding author: Guntram Schernthaner, [email protected]. Received 30 November 2012 and accepted 21 February 2013. control across stages of type 2 diabetes and DOI: 10.2337/dc12-2491. reg. no. NCT01137812, clinicaltrials.gov. range of classes, including add-on to the This article contains Supplementary Data online at http://care.diabetesjournals.org/lookup/suppl/doi:10 combination of metformin and a sulfonyl- .2337/dc12-2491/-/DC1. urea agent. This 52-week Canagliflozin A slide set summarizing this article is available online. – © 2013 by the American Diabetes Association. Readers may use this article as long as the work is properly Treatment and Trial Analysis dipeptidyl cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons.org/ peptidase-4 inhibitor (CANTATA-D2; sec- licenses/by-nc-nd/3.0/ for details. ond comparator trial) study evaluated the

2508 DIABETES CARE, VOLUME 36, SEPTEMBER 2013 care.diabetesjournals.org Schernthaner and Associates efficacy and safety of canagliflozin 300 mg At week 22, all subjects received single- efficacy end points included change from compared with sitagliptin 100 mg as add- blind placebo capsules matching the baseline in FPG and systolic blood pres- on therapy in subjects with type 2 diabetes double-blind study drug for once-daily sure (BP), and percent change from base- inadequately controlled with metformin administration. After the placebo run-in line in body weight, triglycerides, and plus a sulfonylurea agent. period, subjects were randomly assigned HDL cholesterol (HDL-C). Additional to receive oral doses of canagliflozin 300 efficacy measures included proportion RESEARCH DESIGN AND mg or sitagliptin 100 mg once daily (1:1) of subjects reaching A1C ,7.0% (53 METHODS using an Interactive Voice Response System/ mmol/mol) and ,6.5% (48 mmol/mol), Interactive Web Response System. The change in diastolic BP, and percent Subjects and study design computer-generated randomization change in other fasting plasma lipids. This randomized, double-blind, active- schedule was prepared by the sponsor Indices of b-cell function (bCF) were controlled, phase 3 study was conducted before the study, and randomization evaluated. Homeostasis model assess- at 140 centers in 17 countries. The study was balanced using permuted blocks ment of b-cell function (HOMA2-%B), consisted of a 2-week single-blind pla- with the following two stratification cri- proinsulin/insulin ratio, and proinsulin/ cebo run-in period, a 52-week double- teria: whether the prerandomization A1C C-peptide ratio (post hoc analysis) were blind treatment phase, and a 4-week was $9.0% (75 mmol/mol) and whether assessed in the overall study population. follow-up period. a subject underwent the frequently The ratio of C-peptide area under the Eligible subjects were men and sampled mixed-meal tolerance test curve (AUCC) to glucose AUC (AUCG) women 18 years of age or older with (FS-MMTT). was assessed in a subset of subjects who type 2 diabetes using stable metformin After randomization, A1C and FPG underwent a FS-MMTT (;700 kcal and and sulfonylurea therapy. Subjects at values and all glucose levels from the FS- 100 g carbohydrates) on day 1 and at screening already using the combination MMTT were masked to the study centers, week 52. Postprandial glucose measure- of metformin and sulfonylurea with both unless FPG/A1C values met specificstudy ments were also performed in these sub- agents at maximally or near-maximally criteria for discontinuation; similarly, jects. For FS-MMTT end points, blood effective doses (metformin $2,000 urine glucose results were not reported samples were collected 15 min before mg/day [or $1,500 mg/day if unable to for urine dipsticks. Subjects, investiga- and immediately before the meal (time tolerate a higher dose]; sulfonylurea at tors, and local sponsor personnel re- 0), and 30, 60, 90, 120, and 180 min after half-maximal labeled dose or more), who mained blinded to treatment assignment the meal. had A1C $7.0% (53 mmol/mol) and and urine samples for glucosuria until all Safety evaluations included adverse #10.5% (91 mmol/mol), and who subjects completed the study (week 52 events (AEs), clinical laboratory tests, met all other enrollment criteria directly visit) and the final database was locked. vital sign measurements, physical exami- entered the 2-week single-blind placebo Subjects meeting prespecified glyce- nations, self-monitored blood glucose, run-in period before randomization. miccriteria(FPG.15.0 mmol/L [270 12-lead electrocardiograms, and docu- All other subjects underwent an AHA ad- mg/dL] after day 1 to week 6, .13.3 mentation of hypoglycemic episodes. justment period of up to 12 weeks (in- mmol/L [240 mg/dL] after week 6 to AEs prespecified for additional data col- cluding an 8-week dose-stable period) to week 12, .11.1 mmol/L [200 mg/dL] af- lection and analysis included genital my- use maximally or near-maximally ef- ter week 12 to week 26, and A1C .8.0% cotic infections and urinary tract fective doses of metformin and a sulfonyl- [64 mmol/mol] after week 26) were dis- infections (UTIs). The incidence of hypo- urea agent. These subjects then entered continued (no rescue therapy was imple- glycemia was assessed by the number of the 2-week single-blind placebo run-in mented). Subjects with serum creatinine subjects with documented events (i.e., period if they had A1C of $7.0% (53 $133 mmol/L (men) or $124 mmol/L concurrent finger stick glucose #3.9 mmol/mol) and #10.5% (91 mmol/mol) (women), eGFR ,50 mL/min/1.73 m2, mmol/L [70 mg/dL]) and severe events and met all enrollment criteria before be- or with eGFR values that constituted con- (i.e., requiring the assistance of another ing randomized. traindications to metformin use in the or resulting in seizure or loss of con- Exclusion criteria included the fol- country of the investigational site (,60 sciousness). lowing: repeated fasting plasma glucose mL/min/1.73 m2 in some countries) also (FPG) or fasting self-monitored blood were discontinued from the study. Statistical analyses glucose measurements $16.7 mmol/L The study was conducted in accor- The primary hypothesis for this study was (300 mg/dL), or both, during the pretreat- dance with ethical principles that comply that canagliflozin 300 mg was noninferior ment phase; history of , with the Declaration of Helsinki and is to sitagliptin 100 mg in reducing A1C cardiovascular disease, or uncontrolled consistent with Good Clinical Practices from baseline to week 52. The primary hypertension; treatment with either a and applicable regulatory requirements. analysis was based on the modified intent- PPARg agonist, ongoing insulin therapy, Approval was obtained from Institutional to-treat population (all randomized another SGLT2 inhibitor, or any other Review Boards and independent Ethics subjects who received one or more doses AHA (other than metformin and a sulfo- Committees for participating centers. All of study drug) with a last observation nylurea) within 12 weeks before screen- participants provided written informed carried forward approach to impute ing; or estimated glomerular filtration rate consent. missing data at the end point. Assuming (eGFR) ,55 mL/min/1.73 m2 (or ,60 no difference between canagliflozin and mL/min/1.73 m2 ifbasedonrestriction Study end points and assessments sitagliptin in A1C-lowering efficacy of metformin use in the metformin local The prespecified primary efficacy end and a common SD of 1.0% with respect label); or serum creatinine $124 mmol/L point was change in A1C from baseline to change in A1C, it was estimated that (men) and $115 mmol/L (women). through week 52. Prespecified secondary 234 subjects per treatment group would care.diabetesjournals.org DIABETES CARE, VOLUME 36, SEPTEMBER 2013 2509 Canagliflozin versus sitagliptin provide ;90% power to demonstrate the RESULTS were similar with canagliflozin and sitagliptin noninferiority of canagliflozin compared (Table 1). Baseline mean A1C was 8.1% with sitagliptin. In addition, per-protocol Subject disposition and baseline (65 mmol/mol [A1C values in mmol/mol analysis (subjects completing the characteristics were directly converted from values in 52-week study and without protocol de- This study was conducted from 30 June percent]) and mean known duration of viations that could impact efficacy assess- 2010 through 9 March 2012. Of 1,672 type 2 diabetes was 9.6 years. Details of ment) was conducted to further support subjects screened, 916 (54.8%) were ex- background sulfonylurea therapies are the noninferiority assessment. To provide cluded from the study (Fig. 1). Of 756 shown in Supplementary Fig. 1. 90% power for the per-protocol analysis, randomized subjects, 755 received one assuming a discontinuation rate of 35% or more doses of study drug and were in- Efficacy over 52 weeks, the sample size was in- cluded in the modified intent-to-treat At 52 weeks, canagliflozin 300 mg dem- creased to 360 subjects per treatment analysis set; 464 (61%) completed the onstrated noninferiority to sitagliptin 100 group. 52-week treatment period. A higher rate mg in A1C lowering (LS mean change Safety analyses and the primary effi- of discontinuation was observed with of 21.03% [211.3 mmol/mol] and cacy analysis were conducted using the sitagliptin 100 mg (44.4%) than with 20.66% [27.2 mmol/mol], respec- modified intent-to-treat population. The canagliflozin 300 mg (32.6%); the largest tively), with a difference in LS means of last observation carried forward approach contribution was the discontinuation of 20.37% (95% CI, 20.50 to 20.25) or was used for the primary analysis of subjects who met glycemic withdrawal 24.0 mmol/mol (25.5 to 22.7); upper efficacy data. All statistical tests were criteria (22.5% [sitagliptin] and 10.6% limit of the 95% CI was less than the pre- interpreted at a two-sided significance [canagliflozin]), because glycemic rescue specified margin of 0.3% and thus met the level of 5%, and all CIs were interpreted therapy was not provided in this study, primary hypothesis of the study. Further, at a two-sided confidence level of 95%. with the majority (88%) of these subjects canagliflozin demonstrated superiority in Primary and continuous secondary end discontinued after week 26. After meet- reducing A1C compared with sitagliptin points were assessed using an ANCOVA ing glycemic withdrawal criteria, the (upper limit of the 95% CI ,0.0%) (Fig. model, including treatment and stratifi- most common reasons for withdrawal 2A, B). A smaller between-group differ- cation factors as fixed effects and the were meeting creatinine or eGFR with- ence was observed for the per-protocol corresponding baseline value as a cova- drawal criteria or AEs (Fig. 1). Subject analysis (20.21% [95% CI, 20.34 to riate. The least squares (LS) mean differences demographic and baseline characteristics 20.08] or 22.3 mmol/mol [23.7 to and two-sided 95% CIs were estimated for the comparisons of canagliflozin versus sitagliptin. Noninferiority of canagliflozin to sitagliptin was assessed based on a prespecified margin of 0.3% for the upper limit of the two-sided 95% CI for the comparison in the primary last observation carried forward analysis. If noninferiority was demonstrated, then superiority was assessed, as determined by an upper bound of the 95% CI around the between-group difference (canagliflozin minus sitagliptin) of ,0.0%. A prespeci- fied hierarchical testing sequence was im- plemented to strongly control overall type I error attributable to multiplicity; P values are reported for prespecified comparisons only. For subgroup analysis, descriptive statistics and 95% CIs for the change from baseline in A1C were provided for subgroups of subjects with baseline A1C of ,8.0% (64 mmol/mol), $8.0% (64 mmol/mol) to ,9.0% (75 mmol/mol), and $9.0% (75 mmol/mol). For indices of bCF, descriptive statistics and 95% CIs for the changes from baseline were pro- vided; comparisons of canagliflozin with sitagliptin for changes from baseline at week 52 were assessed using an ANCOVA model with treatment and stratification factors as fixed effects and the corresponding baseline value as a covariate. Figure 1dStudy flow diagram. SITA, sitagliptin; CANA, canagliflozin.

2510 DIABETES CARE, VOLUME 36, SEPTEMBER 2013 care.diabetesjournals.org Schernthaner and Associates

Table 1dBaseline demographics and disease characteristics

SITA 100 mg (n = 378) CANA 300 mg (n = 377) Total (N = 755) Age, years 56.7 6 9.3 56.6 6 9.6 56.7 6 9.5 Sex, n (%) Male 215 (56.9) 207 (54.9) 422 (55.9) Female 163 (43.1) 170 (45.1) 333 (44.1) Race, n (%)* White 240 (63.5) 245 (65.0) 485 (64.2) Black or African American 45 (11.9) 43 (11.4) 88 (11.7) Asian 65 (17.2) 67 (17.8) 132 (17.5) Other† 28 (7.4) 22 (5.8) 50 (6.6) Ethnicity, n (%)* Hispanic or Latino 80 (21.2) 79 (21.0) 159 (21.1) Not Hispanic or Latino 296 (78.3) 298 (79.0) 594 (78.7) Other‡ 2 (0.5) 0 2 (0.3) Body weight, kg 89.1 6 23.2 87.4 6 23.2 88.3 6 23.2 BMI, kg/m2 31.7 6 6.9 31.5 6 6.9 31.6 6 6.9 A1C, % (mmol/mol) 8.1 6 0.9 (65 6 9.8) 8.1 6 0.9 (65 6 9.8) 8.1 6 0.9 (65 6 9.8) FPG, mmol/L (mg/dL) 9.2 6 2.5 (165.8 6 44.9) 9.4 6 2.6 (169.4 6 47.4) 9.3 6 2.6 (167.6 6 46.1) Duration of type 2 diabetes, years 9.7 6 6.3 9.4 6 6.1 9.6 6 6.2

Data are mean 6 SD unless otherwise indicated. SITA, sitagliptin; CANA, canagliflozin. *Percentages may not total 100% because of rounding. †Includes American Indian or Alaska Native, Native Hawaiian or other Pacific islander, multiple, other, unknown, or not reported. ‡Includes unknown or not reported.

20.9]) (Supplementary Fig. 1), but respectively; difference in LS means, decrease in the proinsulin/C-peptide ratio canagliflozin met both noninferiority 25.9 mmHg [95% CI, 27.6 to 24.2]; with canagliflozin, whereas an increase was to sitagliptin (upper limit of the 95% P , 0.001) (Supplementary Table 2). seen with sitagliptin. Both groups showed CI ,0.3%) and superiority to sitagliptin Diastolic BP also was reduced with similar increases from baseline in the (upper limit of the 95% CI ,0.0%) in this canagliflozin compared with sitagliptin AUCC/AUCG ratio. analysis. (23.0 and 20.3 mmHg, respectively; dif- A greater proportion of subjects treated ference in LS means, 22.7 mmHg [95% Safety with canagliflozin compared with sitagliptin CI, 23.8 to 21.7]). No meaningful change The overall incidences of AEs, serious achieved A1C ,7.0% (53 mmol/mol; in heart rate was observed with canagliflozin AEs, and study discontinuations attribut- 47.6 vs. 35.3%, respectively) and A1C or sitagliptin (mean change of 20.1 and able to AEs were similar for canagliflozin ,6.5% (48 mmol/mol; 22.5 vs. 18.9%, 0.7 bpm, respectively). An increase in 300 mg and sitagliptin 100 mg (Table 2). respectively) at week 52. HDL-C was seen with canagliflozin and Canagliflozin was associated with higher Subgroup analyses showed stepwise a minimal change was seen with sitagliptin rates of genital mycotic infections in fe- greater A1C reductions from baseline for (LS mean percent change of 7.6 and 0.6%, males (e.g., vulvovaginitis) and males canagliflozin relative to sitagliptin with respectively; difference in LS means, 7.0% (e.g., balanitis) compared with sitagliptin, increasing categories of baseline A1C as [95% CI, 4.6 to 9.3]); increases in LDL-C which led to one study discontinuation follows: ,8.0% (64 mmol/mol); $8.0% were seen in both groups, with a larger (male subject) and no serious AEs, and re- (64 mmol/mol) to ,9.0% (75 mmol/mol); increase seen with canagliflozin relative to sponded to usual treatment with antifungal and $9.0% (75 mmol/mol) (Supplemen- sitagliptin (11.7 and 5.2%, respectively; agents. Among female subjects with genital tary Fig. 2). difference in LS means, 6.4% [95% CI, mycotic infections, 8 of 26 and 2 of 7 sub- Significantly greater reductions from 1.7 to 11.2]). Smaller increases in non– jects in the canagliflozin and sitagliptin baseline in FPG were observed at week 52 HDL-C were observed in both groups, groups, respectively, had recurrences; 2 of with canagliflozin 300 mg compared with with no notable difference in the change 19 male subjects with genital mycotic infec- sitagliptin 100 mg (P , 0.001; Fig. 2C). from baseline in the LDL-C/HDL-C ratio tions had recurrences in the canagliflozin Among subjects who underwent the between groups. Modest increases in trigly- group and no recurrences were reported in FS-MMTT, canagliflozin provided greater cerides, similar in both groups, also were the sitagliptin group. Incidences of UTIs reductions from baseline in 2-h postpran- observed (Supplementary Table 2). were similar for canagliflozin and sitagliptin, dial glucose compared with sitagliptin At week 52, a greater increase from with no serious AEs of UTI reported with (difference in LS means: –1.0 mmol/L baseline in HOMA2-%B was observed canagliflozin. Low incidences (,2%) of [95% CI, 21.9 to 20.1]) (Fig. 2D). with canagliflozin 300 mg compared AEs related to osmotic diuresis (i.e., thirst, Canagliflozin 300 mg provided sig- with sitagliptin 100 mg (Supplementary pollakiuria) were seen with canagliflozin nificant reductions in body weight at Table 3). Modest increases in the proin- treatment. The proportion of subjects having week 52 relative to sitagliptin 100 mg sulin/insulin ratio were observed in both at least one documented hypoglycemic (P , 0.001; Fig. 2E). Canagliflozin signif- groups, with a numerically greater increase episode was similar with canagliflozin icantly decreased systolic BP compared seen with canagliflozin compared with (43.2%) and sitagliptin (40.7%). Incidence with sitagliptin (25.1 and 0.9 mmHg, sitagliptin. Post hoc analysis showed a of severe hypoglycemic episodes was similar care.diabetesjournals.org DIABETES CARE, VOLUME 36, SEPTEMBER 2013 2511 Canagliflozin versus sitagliptin

Figure 2dChanges in efficacy parameters (last observation carried forward [LOCF]). Change in A1C (A), mean A1C over time (B), change in FPG (C), change in postprandial glucose (PPG) (D), and percent change in body weight (E). SITA, sitagliptin; CANA, canagliflozin. *Statistical comparison for CANA 300 mg versus SITA 100 mg not performed (not prespecified). with canagliflozin (4.0%) and sitagliptin Similar small reductions in eGFR were is a novel insulin-independent approach (3.4%). observed in both groups. to lowering plasma glucose in patients Overall, there were only minor differ- with type 2 diabetes that is associated ences observed in laboratory parameters CONCLUSIONSdFor patients with with caloric loss and low risk of hypoglyce- with canagliflozin 300 mg compared with type 2 diabetes to achieve and maintain mia (14). SGLT2 inhibitors therefore may sitagliptin 100 mg (Supplementary Table glycemic control, combination therapies present a new therapeutic option that 4). Small to moderate decreases in alanine with AHAs are often needed (1,2). How- provides a mechanism of action that is com- aminotransferase, g-glutamyl transferase, ever, currently available AHAs have de- plementary to those of other AHAs. and serum urate and small to moderate creased efficacy over time and some In this phase 3 study of subjects with increases in hemoglobin, bilirubin, and have side effects, including weight gain type 2 diabetes inadequately controlled blood urea nitrogen were observed with or increased risk of hypoglycemia (2,3,13). with metformin plus sulfonylurea, the ad- canagliflozin compared with sitagliptin. Inhibition of SGLT2 and increasing UGE dition of canagliflozin 300 mg provided

2512 DIABETES CARE, VOLUME 36, SEPTEMBER 2013 care.diabetesjournals.org Schernthaner and Associates

Table 2dSummary of overall safety and selected AEs with canagliflozin. Postmeal improvement in bCF based on the AUCC/AUCG ratio Subjects, n (%) SITA 100 mg (n = 378) CANA 300 mg (n =377) (assessed during the FS-MMTT) also was seen with both agents. Modest increases in Any AE 293 (77.5) 289 (76.7) the proinsulin/insulin ratio were observed AEs leading to discontinuation 11 (2.9) 20 (5.3) with canagliflozin and sitagliptin. The AEs related to study drug* 105 (27.8) 128 (34.0) proinsulin/C-peptide ratio has been re- Serious AEs 21 (5.6) 24 (6.4) ported to be a more accurate reflection of Deaths 0 2 (0.5)† the degree of disproportional hyperproin- AEs of special interest sulinemia (18); this was assessed in the Genital mycotic infection current study and showed a reduction Male‡x 1 (0.5) 19 (9.2) with canagliflozin, suggesting improved Female|{ 7 (4.3) 26 (15.3) insulin processing efficiency with no UTI 21 (5.6) 15 (4.0) discernible change seen with sitagliptin. Osmotic diuresis–related AEs Because b-cells are not known to express Pollakiuria** 5 (1.3) 6 (1.6) the SGLT2 transporter, an indirect mech- Polyuria†† 0 3 (0.8) anism of these improvements in bCF Volume-related AEs seems likely. Now classic studies of the Postural dizziness 2 (0.5) 0 nonselective SGLT inhibitor, , Orthostatic hypotension 1 (0.3) 0 by Rossetti and DeFronzo et al. (19) in a SITA, sitagliptin; CANA, canagliflozin. *Possibly, probably, or very likely related to study drug, as assessed by diabetic rodent model demonstrated investigators. †One death was attributable to respiratory arrest and cardiac arrest, and the other death was marked bCF improvements considered to attributable to cardiac arrest. Both deaths were considered by the investigator to be doubtfully related to be related to reversal of glucotoxicity. The the study drug. ‡SITA 100 mg, n = 215; CANA 300 mg, n =207.xIncluding balanitis, balanitis candida, results of these studies with phlorizin are | n n balanoposthitis, genital candidiasis, and genital infection fungal. SITA 100 mg, = 163; CANA 300 mg, = 170. supported by experimental studies in mice {Including vaginal infection, vulvovaginal candidiasis, vulvovaginal mycotic infection, and vulvovaginitis. **Increased urine frequency. ††Increased urine volume. harboring a gene knockout of SGLT2 (20), preclinical studies of SGLT inhibitors in diabetic rodent models (21–23), and pre- asignificantly greater reduction in A1C attributable to the marked increase in vious clinical studies of canagliflozin dem- compared with sitagliptin 100 mg at UGE, because each gram of glucose lost onstrating improvements in bCF (10). In 52 weeks. Significantly greater reductions equates to the loss of 4 kcal; previous addition to reversal of glucotoxicity, by re- in FPG also were observed with canagliflozin studies of canagliflozin have demonstrated ducing plasma glucose concentrations compared with sitagliptin. Both medica- that this agent leads to 80–120 g/day of through a noninsulin-dependent mecha- tions lowered postmeal glucose levels, UGE in subjects with type 2 diabetes, nism, the b cells may be “unloaded,” also with numerically greater reductions seen or a deficit of ;400 kcal/day; note that contributing to improved function (24). with canagliflozin, as shown by similar de- studies of canagliflozin have shown that Furthermore, improvements in b-cell func- creases in incremental glucose after the meal the augmentation of UGE with this agent tion could be related to the weight loss seen with both agents; however, canagliflozin is sustained (9,10). This caloric deficit sus- with canagliflozin, with potential improve- provided substantially greater premeal reduc- tained over 52 weeks might predict an ments in insulin sensitivity reducing b-cell tions in glucose. The effects of canagliflozin in even greater weight loss than observed, demand, thus improving insulin secretion lowering A1C and FPG were sustained over suggesting that compensatory mecha- (25,26). 52 weeks of treatment, whereas increases nisms (e.g., increased food intake) likely The overall rate of study discontinua- from the achieved nadirs in A1C and FPG limit weight loss. An important observation tion (38.5%) was high but consistent with were observed with sitagliptin. The re- was that although the weight loss plateaued expectations given the absence of glycemic sults with sitagliptin in the current study, with canagliflozin, no meaningful weight rescue therapy in this study; nearly 80% of both with regard to the pattern of change regain occurred. Body composition anal- subjects in both groups could have com- over time and the extent of A1C-lowering yses conducted in other phase 3 studies pleted the study had rescue therapy rather efficacy, are consistent with the 52-week of subjects with type 2 diabetes have than withdrawal been specified for poorer results seen in a previous sitagliptin study shown that the weight loss observed with glycemic control. The higher discontinua- (in particular, comparing the primary per- canagliflozin is predominantly (approxi- tion rate observed with sitagliptin was protocol analysis reported in the previous mately two-thirds) from the loss of fat mass primarily attributable to the discontinua- study with the per-protocol analysis results (17). Canagliflozin also lowered systolic BP tion of subjects who met glycemic with- from the current study) (Supplementary at week 52, with no meaningful change drawal criteria. Rates of discontinuation Fig. 1) considering baseline differences in seen with sitagliptin. The BP-lowering ef- observed with sitagliptin are consistent glycemic control (15,16). fect with canagliflozin is likely related to the with those previously reported in the In addition to improvements in gly- mild osmotic response and some 52-week study comparing sitagliptin with cemic parameters, canagliflozin provided natriuretic effect, but weight loss also may (15), discussed here, that also did statistically significant and clinically contribute. not provide rescue therapy. In that study, relevant reductions in body weight Canagliflozin and sitagliptin were asso- the rate of discontinuations attributable to over 52 weeks, whereas a slight increase ciated with improvements in HOMA2-%B, reasons other than meeting glucose dis- was observed with sitagliptin. The effect an index of fasting insulin secretion, continuation criteria was similar to that of canagliflozin on body weight is likely with a numerically greater increase seen observed in the current study. care.diabetesjournals.org DIABETES CARE, VOLUME 36, SEPTEMBER 2013 2513 Canagliflozin versus sitagliptin

Overall incidences of AEs were similar diabetes inadequately controlled with met- Boehringer Ingelheim, Takeda, and Johnson for canagliflozin and sitagliptin. Canagliflozin formin plus sulfonylurea suggests that this & Johnson, and is a speaker for Eli Lilly, was associated with an increased incidence agent may be a beneficial option when a Boehringer Ingelheim, Takeda, AstraZeneca, fi of genital mycotic infections; however, third agent is needed. Although the current Bristol-Myers Squibb, Sano , Merck, and Novo these were generally assessed by the inves- study specifically evaluated the potential Nordisk. J.L.G. has received grants or research fl support from Eli Lilly, Bristol-Myers Squibb, tigators as mild to moderate in intensity, use of canagli ozin as an add-on therapy to Boehringer Ingelheim, GlaxoSmithKline, Novo responded to usual antifungal therapy, and metformin plus a sulfonylurea, other stud- fi Nordisk, and Janssen, and has served as a board led to few discontinuations. There was a ies have assessed the ef cacy and safety of member for Boehringer Ingelheim, Eli Lilly, and higher rate of AEs related to osmotic canagliflozin in subjects inadequately con- Novo Nordisk. J.R. has received grants or re- diuresis (i.e., polyuria) compared with trolled with diet or exercise (monotherapy) search support from Merck, Pfizer, Sanofi,Novo sitagliptin, with an overall low incidence or metformin (10,32,33). Canagliflozin Nordisk, Roche, Bristol-Myers Squibb, Eli and infrequently leading to discontinua- also may represent an alternative to the ad- Lilly, GlaxoSmithKline, Takeda, Novartis, tion. Incidences of UTIs were similar be- dition of injectable AHAs that have shown AstraZeneca, , Johnson & Johnson, tween groups. Despite greater reductions in efficacy as add-on therapies to metformin Daiichi Sankyo, MannKind, Lexicon, and A1C and FPG with canagliflozin compared plus sulfonylurea, such as -like Boehringer Ingelheim, and has served on advi- sory boards for and received honoraria or con- with sitagliptin, the incidence of documen- peptide 1 receptor agonists (e.g., , sulting fees from Novo Nordisk, Eli Lilly, Sanofi, ted and severe hypoglycemia was similar in ) without gastrointestinal symp- MannKind, GlaxoSmithKline, Takeda, Daiichi both groups and consistent with other toms or insulin without weight gain or fre- Sankyo, Johnson & Johnson, Boehringer studies conducted on the background of quent hypoglycemia (beyond that expected Ingelheim, and Lexicon. M.G. has no proprietary AHAs associated with hypoglycemia (e.g., from improved glycemic control) (29,34). interest in the tested product, does not have a sulfonylureas, insulin) (27–31). In studies A limitation of this study was its 1-year significant equity interest in the sponsor of the of subjects not using background AHAs as- duration, because longer-term studies will covered study, and has not received significant sociated with hypoglycemia, the low risk of be needed to evaluate the durability of the payments of other sorts from the sponsor. M.F., hypoglycemia observed with canagliflozin effects of canagliflozin. This study enrolled J.Y., M.K., W.C., and G.M. are full-time em- is consistent with its mechanism of action, subjects using background therapy with ployees of Janssen Research & Development, LLC. No other potential conflicts of interest because the renal threshold for glucose is metformin plus a sulfonylurea and in- relevant to this article were reported. typically reduced to 4.4–5.0 mmol/L cluded subjects with a reasonably wide – G.S., J.L.G., J.R., M.G., J.Y., and M.K. con- (;80 90 mg/dL) in patients with type 2 range of baseline A1C values; therefore, the tributed to the conduct of the study, con- diabetes, above the hypoglycemia threshold findings may not be generalizable to pa- tributed to the acquisition, analysis, and (9,10). A different profile of lipid changes tients using other AHA regimens or those interpretation of data, and reviewed and ap- wasobservedwithcanagliflozin relative to with milder or more severe hyperglycemia proved the manuscript. M.F. and W.C. con- sitagliptin; there were increases in HDL-C at baseline. Additionally, studies including tributed to the analysis and interpretation of and LDL-C seen with canagliflozin, and no other AHAs as active comparators will be data, and reviewed and approved the manu- change in HDL-C and a small increase in useful for evaluating the relative efficacy/ script. G.M. contributed to the design and LDL-C seen with sitagliptin, resulting in no safety profile of canagliflozin compared conduct of the study, contributed to the acqui- sition, analysis, and interpretation of data, and difference in the change from baseline in with other AHAs commonly used as third reviewed and approved the manuscript. G.S. the LDL-C/HDL-C ratio between treatment agents in combination therapy with met- and G.M. are the guarantors of this work and, as groups. formin plus sulfonylurea. such, had full access to all the data in the study The recent American Diabetes Asso- In conclusion, canagliflozin significan- and take responsibility for the integrity of the ciation and European Association for the tly improved glycemic control and reduced data and the accuracy of the data analysis. Study of Diabetes position statement rec- body weight and systolic BP compared Parts of this study were previously pre- ommends that selection of AHA therapy with sitagliptin over 52 weeks, with an sented in abstract form at the 72nd Scientific should be individualized to meet the increase in genital infections and osmotic Sessions of the American Diabetes Association, – needs of each patient based on the risk/ diuresis–related AEs, but it was generally Philadelphia, Pennsylvania, 8 12 June 2012, benefitprofiles of the various AHA classes well-tolerated in subjects with type 2 diabe- and at the 48th Annual Meeting of the Euro- pean Association for the Study of Diabetes, (2). Findings from this study demon- tes inadequately controlled with metformin Berlin, Germany, 1–5 October 2012. strated that in patients using metformin plus sulfonylurea. These results support the fl The authors acknowledge Sharmila Patel, and a sulfonylurea, relative to sitagliptin, potential value of canagli ozin in triple PhD, Janssen Global Services, LLC, for her as- canagliflozin provided a greater and more combination therapy with metformin sistance and contribution to data review, clinical sustained reduction in A1C, weight loss, a plus a sulfonylurea for the treatment of pa- management, and the study report. The authors reductioninsystolicBP,andasimilarinci- tients with type 2 diabetes. thank all investigators, study teams, and patients dence of hypoglycemia. However, these ben- for participating in this study. efits of canagliflozin relative to sitagliptin were associated with a higher incidence of AcknowledgmentsdThis study was sup- female and male genital infections and AEs ported by Janssen Research & Development, References related to osmotic diuresis. Thus, in selecting LLC. Editorial support was provided by Kimberly 1. American Diabetes Association. Standards Dittmar, PhD, of MedErgy, and was funded by of medical care in diabetes–2012. Diabetes treatments, physicians will have to weigh the fl – fi Janssen Global Services, LLC. Canagli ozin Care 2012;35:S11 S63 bene ts with the side effects, taking into is being developed by Janssen Research & 2. Inzucchi SE, Bergenstal RM, Buse JB, et al.; account the needs and tolerances of the pa- Development, LLC, in collaboration with American Diabetes Association (ADA); tient based on age and comorbidities. Mitsubishi Tanabe Pharma Corporation. European Association for the Study of Di- The overall profile of canagliflozin in G.S. has served on advisory boards for abetes (EASD). Management of hyperglyce- the current study of patients with type 2 Eli Lilly, AstraZeneca, Bristol-Myers Squibb, mia in type 2 diabetes: a patient-centered

2514 DIABETES CARE, VOLUME 36, SEPTEMBER 2013 care.diabetesjournals.org Schernthaner and Associates

approach: position statement of the metformin or sulphonylureas often fails to 25. Kahleova H, Mari A, Nofrate V, et al. Im- American Diabetes Association (ADA) and achieve or maintain current glycaemic goals provement in b-cell function after diet- the European Association for the Study of in patients with Type 2 diabetes in UK pri- induced weight loss is associated with Diabetes (EASD). Diabetes Care 2012;35: mary care. Diabet Med 2007;24:350–358 decrease in pancreatic polypeptide in 1364–1379 14. Neumiller JJ, White JR Jr, Campbell RK. subjects with type 2 diabetes. J Diabetes 3. Cook MN, Girman CJ, Stein PP, Alexander Sodium-glucose co-transport inhibitors: Complications 2012;26:442–449 CM, Holman RR. Glycemic control con- progress and therapeutic potential in type 2 26. Bradley D, Conte C, Mittendorfer B, et al. tinues to deteriorate after sulfonylureas are diabetes mellitus. Drugs 2010;70:377–385 Gastric bypass and banding equally im- added to metformin among patients with 15. Nauck MA, Meininger G, Sheng D, prove insulin sensitivity and b cell func- type 2 diabetes. Diabetes Care 2005;28: Terranella L, Stein PP; Sitagliptin Study tion. J Clin Invest 2012;122:4667–4674 995–1000 024 Group. Efficacy and safety of the di- 27. Buse JB, Henry RR, Han J, Kim DD, 4. Gross JL, Kramer CK, Leitão CB, et al.; Di- peptidyl peptidase-4 inhibitor, sitagliptin, Fineman MS, Baron AD; Exenatide-113 abetes and Endocrinology Meta-analysis compared with the sulfonylurea, glipizide, Clinical Study Group. Effects of exenatide Group (DEMA). Effect of antihyperglycemic in patients with type 2 diabetes inade- (exendin-4) on glycemic control over agents added to metformin and a sulfonyl- quately controlled on metformin alone: 30 weeks in sulfonylurea-treated patients urea on glycemic control and weight gain in a randomized, double-blind, non-inferiority with type 2 diabetes. Diabetes Care 2004; type 2 diabetes: a network meta-analysis. trial. Diabetes Obes Metab 2007;9:194–205 27:2628–2635 Ann Intern Med 2011;154:672–679 16. Bloomgarden ZT, Dodis R, Viscoli CM, 28. Hermansen K, Kipnes M, Luo E, Fanurik 5. National Institute for Health and Care Holmboe ES, Inzucchi SE. Lower baseline D, Khatami H, Stein P; Sitagliptin Study Excellence. Type 2 diabetes: newer agents glycemia reduces apparent oral agent 035 Group. Efficacy and safety of the di- for blood glucose control in type 2 diabetes. glucose-lowering efficacy: a meta-regression peptidyl peptidase-4 inhibitor, sitagliptin, NICE short clinical guideline 87.2009 analysis. Diabetes Care 2006;29:2137– in patients with type 2 diabetes mellitus 6. Nomura S, Sakamaki S, Hongu M, et al. 2139 inadequately controlled on Discovery of canagliflozin, a novel 17. Toubro S, Cefalu WT, Xie J, et al. alone or on glimepiride and metformin. C-glucoside with thiophene ring, as sodium- Canagliflozin, a sodium glucose co- Diabetes Obes Metab 2007;9:733–745 dependent glucose cotransporter 2 inhibitor transporter 2 inhibitor, reduces body weight 29. Kendall DM, Riddle MC, Rosenstock J, for the treatment of type 2 diabetes melli- mainly through loss of fat mass in subjects et al. Effects of exenatide (exendin-4) on tus. J Med Chem 2010;53:6355–6360 with type 2 diabetes. Diabetologia 2012;55 glycemic control over 30 weeks in patients 7. Sha S, Devineni D, Ghosh A, et al. (Suppl. 1):S313–S314 [Abstract] with type 2 diabetes treated with metfor- Canagliflozin, a novel inhibitor of sodium 18. Loopstra-Masters RC, Haffner SM, Lorenzo min and a sulfonylurea. Diabetes Care glucose co-transporter 2, dose dependently C, Wagenknecht LE, Hanley AJ. Proinsulin- 2005;28:1083–1091 reduces calculated renal threshold for glu- to-C-peptide ratio versus proinsulin-to- 30. Strojek K, Yoon KH, Hruba V, Elze M, cose excretion and increases urinary glu- insulin ratio in the prediction of incident Langkilde AM, Parikh S. Effect of cose excretion in healthy subjects. Diabetes diabetes: the Insulin Resistance Athero- dapagliflozin in patients with type 2 di- Obes Metab 2011;13:669–672 sclerosis Study (IRAS). Diabetologia 2011; abetes who have inadequate glycaemic 8. Liang Y, Arakawa K, Ueta K, et al. Effect of 54:3047–3054 control with glimepiride: a randomized, canagliflozin on renal threshold for glucose, 19. Rossetti L, Smith D, Shulman GI, 24-week, double-blind, placebo-controlled glycemia, and body weight in normal and Papachristou D, DeFronzo RA. Correction trial. Diabetes Obes Metab 2011;13:928– diabetic animal models. PLoS ONE 2012;7: of hyperglycemia with phlorizin normalizes 938 e30555 tissue sensitivity to insulin in diabetic rats. 31. Wilding JP, Woo V, Soler NG, et al.; 9. Devineni D, Morrow L, Hompesch M, J Clin Invest 1987;79:1510–1515 Dapagliflozin 006 Study Group. Long-term et al. Canagliflozin improves glycaemic 20. Jurczak MJ, Lee HY, Birkenfeld AL, et al. efficacy of dapagliflozin in patients with control over 28 days in subjects with type 2 SGLT2 deletion improves glucose ho- type 2 diabetes mellitus receiving high diabetes not optimally controlled on insulin. meostasis and preserves pancreatic beta- doses of insulin: a randomized trial. Ann Diabetes Obes Metab 2012;14:539–545 cell function. Diabetes 2011;60:890–898 Intern Med 2012;156:405–415 10. Rosenstock J, Aggarwal N, Polidori D, et al.; 21. Yasuda K, Okamoto Y, Nunoi K, et al. 32. Niskanen L, Cefalu WT, Leiter LA, et al. Canagliflozin DIA 2001 Study Group. Dose- Normalization of cytoplasmic calcium re- Efficacy and safety of canagliflozin, a so- ranging effects of canagliflozin, a sodium- sponse in pancreatic beta-cells of sponta- dium glucose co-transporter 2 inhibitor, glucose cotransporter 2 inhibitor, as add-on neously diabetic GK rat by the treatment compared with glimepiride in patients to metformin in subjects with type 2 di- with T-1095, a specific inhibitor of renal with type 2 diabetes on background met- abetes. Diabetes Care 2012;35:1232–1238 Na+-glucose co-transporters. Horm Metab formin. Diabetologia 2012;55(Suppl. 1): 11. Bakris GL, Fonseca VA, Sharma K, Wright Res 2002;34:217–221 S314 [Abstract] EM. Renal sodium-glucose transport: role in 22. Katsuno K, Fujimori Y, Ishikawa-Takemura 33. Stenlöf K, Cefalu WT, Kim K-A, et al. Ef- diabetes mellitus and potential clinical im- Y, Isaji M. Long-term treatment with ficacy and safety of canagliflozin mono- plications. Kidney Int 2009;75:1272–1277 sergliflozin etabonate improves disturbed therapy in subjects with type 2 diabetes 12. Rothenberg P, Devineni D, Ghosh A, et al. glucose metabolism in KK-A(y) mice. Eur mellitus inadequately controlled with diet Canagliflozin, a Novel Inhibitor of Sodium J Pharmacol 2009;618:98–104 and exercise. Diabetes Obes Metab 2013; Glucose Co-Transporter 2, Improved Glu- 23. Macdonald FR, Peel JE, Jones HB, et al. 15:372–382 cose Control in Subjects With Type 2 Di- The novel sodium 2 34. Russell-Jones D, Vaag A, Schmitz O, et al.; abetes: Results of a Phase 1b Study. Poster inhibitor dapagliflozin sustains pancreatic Liraglutide Effect and Action in Diabetes 5 presented at: the 46th Annual Meeting of function and preserves islet morphology (LEAD-5) met+SU Study Group. Liraglutide the European Association for the Study of in obese, diabetic rats. Diabetes Obes Metab vs and placebo in combi- Diabetes (EASD); September 20-24, 2010; 2010;12:1004–1012 nation with metformin and sulfonylurea Stockholm, Sweden 2012 24. Wajchenberg BL. Clinical approaches to therapy in type 2 diabetes mellitus (LEAD-5; 13. Cook MN, Girman CJ, Stein PP, Alexander preserve beta-cell function in diabetes. met+SU): a randomised controlled trial. CM. Initial monotherapy with either Adv Exp Med Biol 2010;654:515–535 Diabetologia 2009;52:2046–2055

care.diabetesjournals.org DIABETES CARE, VOLUME 36, SEPTEMBER 2013 2515