Efficacy and Safety ofCanagliflozin, a Sodium–GlucoseCotransporter 2 Inhibitor, asAdd-on to Insulin in PatientsWith Type 1 DiabetesDiabetes Care 2015;38:2258–2265 | DOI: 10.2337/dc15-1730

OBJECTIVE

This study assessed the efficacy and safety of canagliflozin, a sodium–glucosecotransporter 2 inhibitor, as add-on to insulin in adults with type 1 diabetes.

RESEARCH DESIGN AND METHODS

This 18-week, double-blind, phase 2 study randomized 351 patients (HbA1c 7.0–9.0% [53–75 mmol/mol]) on multiple daily insulin injections or continuous sub-cutaneous insulin infusion to canagliflozin 100 or 300 mg or placebo. The primaryend point was the proportion of patients achieving at week 18 both HbA1c re-duction from baseline of ‡0.4% (‡4.4 mmol/mol) and no increase in body weight.Other end points included changes in HbA1c, bodyweight, and insulin dose, as wellas hypoglycemia incidence. Safety was assessed by adverse event (AE) reports.

RESULTS

More patients had both HbA1c reduction ‡0.4% and no increase in body weightwith canagliflozin 100 and 300 mg versus placebo at week 18 (36.9%, 41.4%,14.5%, respectively; P < 0.001). Both canagliflozin doses provided reductions inHbA1c, body weight, and insulin dose versus placebo over 18 weeks. The incidenceof hypoglycemia was similar across groups; severe hypoglycemia rates were low(1.7–6.8%). Overall incidence of AEs was 55.6%, 67.5%, and 54.7% with canagli-flozin 100 and 300 mg and placebo; discontinuation rates were low (0.9–1.3%).Increased incidence of ketone-related AEs (5.1%, 9.4%, 0%), including the specificAE of diabetic ketoacidosis (DKA) (4.3%, 6.0%, 0%), was seenwith canagliflozin 100and 300 mg versus placebo.

CONCLUSIONS

Canagliflozin provided reductions in HbA1c, body weight, and insulin dose with noincrease in hypoglycemia, but increased rates of ketone-related AEs, includingDKA, in adults with type 1 diabetes inadequately controlled with insulin.

Type 1 diabetes is an autoimmune disease characterized by progressive destructionof pancreatic b-cells, resulting in loss of endogenous insulin production and hyper-glycemia (1). Consequently, treatment of type 1 diabetes requires lifelong insulintherapy to maintain normal blood glucose levels. Type 1 diabetes is associated withincreased morbidity and mortality related to microvascular and macrovascular

1Center for Metabolic Research, VA San DiegoHealthcare System, San Diego, CA2University of California, San Diego, La Jolla, CA3Janssen Research & Development, LLC, Raritan,NJ4Janssen Research & Development, LLC, SanDiego, CA

Corresponding author: Robert R. Henry, rrhenry@outlook.com.

Received 6 August 2015 and accepted 16 Sep-tember 2015.

Clinical trial reg. no. NCT02139943, clinicaltrials.gov.

This article contains Supplementary Data onlineat http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc15-1730/-/DC1.

A slide set summarizing this article is availableonline.

© 2015 by the American Diabetes Association.Readersmayuse this article as longas thework isproperly cited, the use is educational and not forprofit, and the work is not altered.

See accompanying articles, pp. 2200,2204, 2211, 2217, 2226, 2234, 2237,2241, 2250, 2266, 2274, and 2282.

Robert R. Henry,1,2 Payal Thakkar,3

Cindy Tong,3 David Polidori,4 and

Maria Alba3

2258 Diabetes Care Volume 38, December 2015

SPEC

IALARTICLE

COLLECTION:INSU

LIN

complications such as renal insufficiency,peripheral neuropathy, retinopathy, cor-onary artery disease, and stroke (1,2).These complications reduce the esti-mated life expectancy for patients withtype 1 diabetes by ;11 years in menand 13 years in women (3). Patientswith type 1 diabetes are also at increasedrisk of diabetic ketoacidosis (DKA), a se-rious condition caused by absolute orrelative deficiency of circulating insulinlevels, commonly due to precipitatingfactors such as intercurrent illness or in-terruption of insulin therapy (4).Intensive insulin treatment has been

shown to reduce the onset and/or pro-gression of microvascular and macrovas-cular complications in patientswith type1diabetes (5,6); however, even with thedevelopment of rapid- and long-acting in-sulin analogs and improvements in insulin-delivery devices, ;75% of adults withtype 1 diabetes fail to achieve the targetof HbA1c,7.0% (,53mmol/mol) recom-mended by the American Diabetes Asso-ciation (7). Limitations of insulin therapyinclude increased risk of hypoglycemia,excessive glucose fluctuations, andweight gain, which may deter patientsfrom adequately titrating their daily insu-lin dosage to meet target HbA1c goals(1,2). The only approved adjunctive treat-ment for type 1 diabetes is pramlintide,an injectable amylin analog that inhibitsglucagon secretion and delays gastricemptying; however, pramlintide is associ-ated with an increased risk of severe hy-poglycemia (8,9). Thus, new therapies areneeded that help patients with type 1 di-abetes improve glycemic control withoutincreasing body weight or the risk ofhypoglycemia.Canagliflozin is a sodium–glucose co-

transporter 2 (SGLT2) inhibitor approvedfor the treatment of adults with type 2diabetes (10) but not type 1 diabetes.Canagliflozin reduces blood glucose levelsthrough an insulin-independent mecha-nism by lowering the renal threshold forglucose and increasing urinary glucose ex-cretion, which results in a mild osmoticdiuresis and net caloric loss (11–13). Inphase 3 studies, canagliflozin improvedglycemic control, reduced body weightand blood pressure, and was generallywell tolerated across a broad range of pa-tients with type 2 diabetes (10,14). Theincidence of hypoglycemia was low withcanagliflozin in patients not on backgroundantihyperglycemicagent therapyassociated

with an increased risk of hypoglycemia; inpatients on background therapies associ-ated with an increased risk of hypogly-cemia (i.e., insulin or sulfonylurea), theincidence of hypoglycemiawith canagliflo-zin was higher (15).

In a randomized, double-blind, phase 3trial in patients with type 2 diabetes and ahistory or high risk of cardiovascular dis-ease, canagliflozin 100 and 300 mg pro-vided reductions in HbA1c, body weight,and blood pressure versus placebo over52 weeks when used in conjunction withinsulin therapy ($20 IU/day) (16). Rates ofdocumented hypoglycemia (i.e., episodeswith fingerstick or plasma glucose#3.9 mmol/L [#70 mg/dL], irrespectiveof symptoms, and severe episodes [i.e.,episodes requiring assistance from an-other person or those resulting in seizureor loss of consciousness]) were 59%,57%, and 48% with canagliflozin 100and 300 mg and placebo, respectively;the incidence of severe hypoglycemiawas low and similar across treatmentgroups (4–6%).

Because their mechanism of action isindependent of insulin secretion, SGLT2inhibitorsmay represent a complementarytreatment option to insulin for patientswith type 1 diabetes. Several small pilotstudies with SGLT2 inhibitors, rangingfrom 2 to 8 weeks, have been performedin patients with type 1 diabetes (17–19).These studies have generally shown thattreatment with SGLT2 inhibitors providesimproved glucose control and reductionsin the insulin dose without increasinghypoglycemic events. This phase 2 studyevaluated the efficacy and safety of cana-gliflozin compared with placebo over 18weeks in patients with type 1 diabetesinadequately controlled with insulintherapy.

RESEARCH DESIGN AND METHODS

Study Design and PatientsThis double-blind, parallel-group, multi-center, phase 2 study consisted of a2-week period before randomization,followed by an 18-week double-blindtreatment phase and 2 weeks of safetyfollow-up for all patients. Eligible patientswere aged 25–65 years with a BMI of21–35 kg/m2, a fasting C-peptide levelof ,0.2 pmol/L (,0.6 ng/mL) at screen-ing, and type 1 diabetes for 1 year ormore that was inadequately controlledwith multiple daily insulin (MDI) injec-tions or continuous subcutaneous insulin

infusion (CSII) (HbA1c of 7.0–9.0%[53–75 mmol/mol] at screening). Thelower age limit was chosen because pa-tients ,25 years may be less compliantwith following treatment instructionsand adhering to protocol procedures(20). The lower BMI limit was chosento avoid weight loss in patients with analready low body weight.

Patients were required to be on a sta-ble insulin regimen and a stable methodof insulin administration (MDI or CSII)for$8 weeks before screening. Patientswere excluded if they had a history oftype 2 diabetes; had a severe hypoglyce-mic event (defined as an event that re-quired assistance from another person orresulted in seizure or loss of conscious-ness) or DKA (self-reported by patients)within 6 months before randomization;had a myocardial infarction, unstable an-gina, revascularization procedure, orcerebrovascular accident #12 weeksbefore screening; history of New YorkHeart Association Functional Clas-sification III–IV cardiac disease; haduncontrolled hypertension; had an esti-mated glomerular filtration rate (eGFR),70 mL/min/1.73 m2; or were takingany antihyperglycemic agent other thaninsulin within 12 weeks before screen-ing. Patients were discontinued if theireGFRwas,60mL/min/1.73m2 during thestudy.

The study was conducted in accordancewith the ethical principles that have theirorigin in theDeclarationofHelsinki and areconsistent with Good Clinical Practice andapplicable regulatory requirements. Ap-proval was obtained from institutional re-view boards and independent ethicscommittees for each participating center.Patients provided informed written con-sent before participation.

Insulin TherapyTo mitigate a potential increased riskof hypoglycemia due to the addition ofcanagliflozin, patients with HbA1c #8.0%(#64 mmol/mol) at screening were rec-ommended, at the discretionof the inves-tigator, to reduce their basal insulin doseby 20%before randomization; similarly, pa-tients with HbA1c.8.0% (.64 mmol/mol)at screening were recommended to re-duce their basal insulin dose by 10%.During the 18-week treatment period,patients were instructed to titrate theirbasal insulin dose toachieveaprebreakfastblood glucose level of 4.4 to,6.7 mmol/L

care.diabetesjournals.org Henry and Associates 2259

(80 to ,120 mg/dL) and to titrate theirbolus insulin dose to achieve a prelunch,predinner, and bedtime blood glucoselevel of 4.4 to ,6.7 mmol/L (80 to,120 mg/dL). Algorithms for titratingbasal and bolus insulin doses to achievetarget glycemic goals were provided as ageneral guideline (Supplementary Table1). Investigators were allowed to usetheir judgement when recommendinginsulin adjustments based on theirknowledge of the patient’s individualhistory and risk of hypoglycemia. Pa-tients were required to record dailyglucose measurements and concomi-tant doses of basal and bolus insulinat specified time points to assess com-pliance with insulin dosing and to re-ceive titration instructions.

Randomization and BlindingA computer-generated randomizationschedule prepared by the sponsor be-fore the study was used to randomizepatients (1:1:1) to receive canagliflozin100 or 300 mg or placebo once dailybefore the first meal of the day over18 weeks. Randomization was stratifiedaccording to patients’ use of MDI versusCSII. Fasting plasma glucose (FPG) andHbA1c values were masked to the studysites and to the sponsor.

End Points and AssessmentsPatients with type 1 diabetes often failto achieve adequate glycemic controldue to fear of weight gain associatedwith an intense insulin treatment regi-men. Hence, the proportion of patientsat week 18 with HbA1c reduction$0.4%($4.4 mmol/mol) and no increase inbody weight relative to baseline in thecontext of intensified insulin treatmentwas selected to be a clinically meaning-ful prespecified primary end point inthis study. Because some natural day-to-day variability occurs in body weight,the proportion of patients at week18 with an HbA1c reduction $0.4%($4.4 mmol/mol) and a change inbody weight,1 kg was also calculatedto assess how sensitive the conclusionswere to the specific value of weightchange used in the end point. Key sec-ondary end points included changefrom baseline in HbA1c and FPG, pro-portion of patients with HbA1c ,7.0%(,53 mmol/mol), percentage changefrom baseline in body weight, changefrom baseline in basal and bolus insulindosage requirements (after the initial

10–20% downtitration), and the inci-dence and event rate of documentedhypoglycemia and documented symp-tomatic hypoglycemia at week 18. Pa-tients were instructed to recordinformation on all events of possiblehypoglycemia, as well as associated fin-gerstick glucose measurements, if avail-able. Documented hypoglycemia episodesincluded biochemically confirmed epi-sodes (concurrent fingerstick or plasmaglucose #3.9 mmol/L [#70 mg/dL]) orsevere episodes (i.e., episodes requir-ing assistance from another person orthose resulting in seizure or loss of con-sciousness). Documented symptomatichypoglycemia included episodes withsymptoms consistent with hypoglycemiawith concurrent documented plasmaglucose #3.9 mmol/L (#70 mg/dL) orsevere episodes.

Overall safety and tolerability were as-sessed based on adverse event (AE) re-ports, safety laboratory tests, vital signsmeasurements, and physical examina-tions. Analysis of ketone-related AEs wasperformed using a prespecified list ofpreferred terms (i.e., acidosis, blood ke-tone body increased, blood ketone bodypresent,DKA,diabetic ketoacidotic hyper-glycemic coma, ketoacidosis, ketonemia,ketonuria, ketosis, metabolic acidosis,and urine ketone body present).

Statistical AnalysesSample size was determined based onthe assumption that 40% of patientswould meet the primary end point witheach canagliflozin dose compared with20% of patients in the placebo group. Atotal of 100 patients per group were es-timated to be required to achieve;80%power for the comparison of each cana-gliflozin dose with placebo, with a two-sided family-wise type I error rate of0.05. Approximately 110 patients wereplanned for inclusion in each treatmentgroup to account for potential earlydiscontinuations.

Efficacy and safety analyses were con-ducted using the modified intent-to-treat analysis set (i.e., all patients whowere randomized and received$1 dosedouble-blind study drug). The primaryefficacy end point was analyzed longitu-dinally using a generalized linear mixedmodel that included treatment, stratifi-cation factor (i.e., use of MDI vs. CSII),visit, and treatment-by-visit interactionas fixed categorical effects and baseline

HbA1c, baseline body weight, and base-line-by-visit interaction as continuousfixed covariates. Change from baselinein HbA1c, FPG, and body weight wereanalyzed with a mixed-model for re-peated measures using a restrictedmaximum likelihood approach. Thisanalysis was based on observed datathat included treatment, stratificationfactor, visit, and treatment-by-visit inter-action as fixed categorical effects andbaseline value and baseline-by-visit in-teraction as continuous fixed covariates.Least squares (LS) mean differences and95% CIs were estimated at week 18 foreach canagliflozin dose versus placebo.The proportion of patients withHbA1c ,7.0% (,53 mmol/mol) wasanalyzed using a logistic regressionmodel including the fixed categoricaleffects of treatment and stratificationfactor (use of MDI vs. CSII) and the fixedcontinuous covariate of baseline HbA1c.Change from baseline in insulin dose wasassessed for total daily insulin dose,basal insulin dose, and bolus insulindose. Similar to the analysis of HbA1cand body weight, the change from base-line in insulin dose was analyzed using themixed-model for repeated-measuresapproach. The incidence of docu-mented hypoglycemia and documentedsymptomatic hypoglycemia was as-sessed at week 18. The hypoglycemicevent rate was calculated as the numberof hypoglycemia episodes per patient-year exposure. Statistical testing of com-parisons of canagliflozin versus placebowas not performed (not prespecified)for end points other than the primaryend point; therefore, P values are onlyreported for the primary end point, and95% CIs are provided for descriptive pur-poses for end points that were not pre-specified for hypothesis testing. Safetyanalyses included all reported AEswith onset during the treatmentphase (i.e., treatment-emergent AEs).

RESULTS

PatientsA total of 351 patients were randomizedand received one or more doses of thestudy drug, comprising the modifiedintent-to-treat analysis set; of these,328 (93.4%) completed 18 weeks oftreatment (Supplementary Fig. 1).Rates of discontinuation were 5.1%,6.0%, and 8.5% with canagliflozin 100and 300 mg and placebo, respectively.

2260 Canagliflozin in Patients With Type 1 Diabetes Diabetes Care Volume 38, December 2015

Lost to follow-up (2.0%) and withdrawalof consent (2.0%)were themost commonreasons for discontinuation. Baselinedemographic and disease characteristicswere generally similar across treatmentgroups (Table 1). Patients had a meanbaseline HbA1c of 7.9% (63 mmol/mol)and BMI of 28.1 kg/m2. A greater propor-tion of patients were using CSII versusMDI (62.4% vs. 37.6%). The mean dura-tion of type 1 diabetes was 22 years;14.8% and 12.0% of patients had a his-tory of severe hypoglycemia or DKA,respectively.

Efficacy

HbA1c, FPG, and Body Weight

At week 18, a greater proportion ofpatients had HbA1c reduction $0.4%($4.4 mmol/mol) and no increase inbody weight with canagliflozin 100and 300 mg compared with placebo(36.9%, 41.4%, and 14.5%, respectively;P , 0.001 for both comparisons). Withrespect to the individual componentsof the composite end point, 45.0%,43.2%, and 22.7%of patients had a reduc-tion in HbA1c $0.4% ($4.4 mmol/mol)at week 18 with canagliflozin 100 and300 mg and placebo, respectively;83.8%, 96.4%, and 49.1% of patients, re-spectively, had no increase in body weight

(Fig. 1A). Results of the sensitivity analysisof the proportion of patients with HbA1creduction $0.4% ($4.4 mmol/mol) andchange in body weight,1 kg were con-sistent with the primary end point, withmore patients achieving this compositeend point with canagliflozin versusplacebo (Supplementary Table 2).

Consistent with the data for the com-posite end point, canagliflozin 100 and300 mg provided greater reductions inHbA1c comparedwith placebo at week 18.Placebo-subtracted LS mean reductionsin HbA1c from baseline were 20.29%(23.2 mmol/mol) with canagliflozin 100mg and 20.25% (22.7 mmol/mol) withcanagliflozin 300 mg (Fig. 1B). Theproportion of patients achieving HbA1c,7.0% (,53 mmol/mol) at week18 was 10.1%, 11.2%, and 5.7% withcanagliflozin 100 and 300 mg and pla-cebo, respectively. Dose-dependentreductions in body weight were seenwith canagliflozin 100 and 300 mgcompared with placebo at week 18(placebo-subtracted LS mean percent-age changes of 23.4% and 25.3%, re-spectively; Fig.1C). Canagliflozin 100and 300 mg also provided reductionsin FPG compared with placebo at week18 (placebo-subtracted LS meanchanges of 20.5 mmol/L [28.5 mg/dL]

and20.6mmol/L [211.5mg/dL], respec-tively; Fig. 1D).

Insulin Dose

Canagliflozin 100 and 300 mg were as-sociated with reductions from baselinein the daily insulin dose at week 18(Fig. 1E). Placebo-subtracted absoluteLS mean reductions (mean percentagechange) in the total insulin dose were–4.1 IU/day (–8.9%) and –7.6 IU/day(–12.9%) with canagliflozin 100 and300 mg, respectively. In the placebogroup, small increases from baseline inbasal insulin dose were observed atweek 18 that restored the basal insulindoses to nearly the same level they werebefore pretreatment downtitration. Incontrast, basal insulin doses werefurther decreased at week 18 withcanagliflozin 100 and 300 mg; placebo-subtracted absolute LS mean reductions(mean percentage change) in basal in-sulin dose were 24.3 IU/day (219.0%)and 25.3 IU/day (222.4%), respec-tively. Placebo-subtracted absolute LSmean reductions (mean percentagechange) in the bolus insulin dose atweek 18 were –0.3 IU/day (+6.1%) and–3.2 IU/day (212.1%) with canagliflozin100 and 300 mg, respectively.

Hypoglycemia

Almost all patients experienced at leastone documented hypoglycemic episodeduring the study (98.3%, 99.1%, and96.6% with canagliflozin 100 and 300 mgand placebo, respectively; Table 2).The event rate per patient-year of expo-sure was 70.7, 79.2, and 80.6 with can-agliflozin 100 and 300 mg and placebo,respectively. The incidence of severe hy-poglycemic episodes was 2.6%, 6.8%, and1.7% with canagliflozin 100 and 300 mgand placebo, respectively. Because therisk of hypoglycemia is known to be asso-ciated with insulin titration and insulintitrationwas expected tobemore intenseearly in the study, a separate analysis wasperformed excluding hypoglycemic epi-sodes between weeks 1 and 4. In thisanalysis, the incidence of documentedhy-poglycemia was 94.9%, 93.2%, and 88.9%with canagliflozin 100 and 300 mg andplacebo, respectively, and the event rateper patient-year of exposure was 62.5,67.5, and 75.8, respectively.

Safety and TolerabilityThe overall incidence of AEs was 55.6%,67.5%, and 54.7% with canagliflozin

Table 1—Baseline demographic and disease characteristics

Canagliflozin

Placebo(n = 117)

100 mg(n = 117)

300 mg(n = 117)

Total(N = 351)

SexMale 63 (53.8) 69 (59.0) 65 (55.6) 197 (56.1)Female 54 (46.2) 48 (41.0) 52 (44.4) 154 (43.9)

Age, years 42.0 6 11.9 42.0 6 11.6 42.8 6 11.0 42.3 6 11.5

Race*White 106 (90.6) 111 (94.9) 102 (87.2) 319 (90.9)Black/African American 7 (6.0) 5 (4.3) 5 (4.3) 17 (4.8)Asian 2 (1.7) 0 5 (4.3) 7 (2.0)Other† 2 (1.7) 1 (0.9) 5 (4.3) 8 (2.3)

HbA1c, % 7.9 6 0.6 7.9 6 0.5 8.0 6 0.5 7.9 6 0.5

HbA1c, mmol/mol 63 6 6.6 63 6 5.5 64 6 5.5 63 6 5.5

Body weight, kg 83.0 6 15.4 84.1 6 14.2 82.9 6 15.0 83.4 6 14.8

BMI, kg/m2 28.0 6 3.6 28.0 6 3.9 28.1 6 3.9 28.1 6 3.8

eGFR, mL/min/1.73 m2 96.0 6 14.8 97.4 6 14.9 95.8 6 16.5 96.4 6 15.4

Duration of type 1 diabetes, years 23.3 6 11.0 22.0 6 11.5 21.9 6 10.6 22.4 6 11.0

CSII use 72 (61.5) 74 (63.2) 73 (62.4) 219 (62.4)

MDI use 45 (38.5) 43 (36.8) 44 (37.6) 132 (37.6)

Prior severe hypoglycemia 18 (15.4) 15 (12.8) 19 (16.2) 52 (14.8)

Prior DKA 14 (12.0) 13 (11.1) 15 (12.8) 42 (12.0)

Data are mean 6 SD or n (%). *Percentages may not total 100.0 due to rounding. †Includesmultiple, other, and not reported.

care.diabetesjournals.org Henry and Associates 2261

Figure 1—Proportion of patients with HbA1c reduction$0.4% ($4.4mmol/mol) from baseline and noweight gain (A), change from baseline in HbA1c(B), percentage change from baseline in body weight (C), change from baseline in FPG (D), and change from baseline in insulin dose at week 18 (E).Data are LS mean change6 SE from baseline unless otherwise indicated. Statistical testing of comparisons of canagliflozin versus placebo was notperformed (not prespecified) for end points other than the primary end point. *P, 0.001 vs. placebo. †Total baseline insulin dose is the sum of thebaseline basal and the baseline bolus insulin dose. ‡Baseline basal insulin dose is the total daily dose of basal insulin after downtitration before day 1.§Baseline bolus insulin dose is the mean of the total bolus insulin doses per day before day 1 during week 21.

2262 Canagliflozin in Patients With Type 1 Diabetes Diabetes Care Volume 38, December 2015

100 and 300 mg and placebo, respec-tively, over 18 weeks (Table 3). Rates ofAEs leading to discontinuation werelow across treatment groups. The inci-dence of serious AEs was 7.7% and 6.8%with canagliflozin 100 and 300 mg, re-spectively; no patients treated withplacebo had a serious AE. Canagliflozinwas associated with a higher incidenceof AEs related to the mechanism ofSGLT2 inhibition that have been ob-served in studies of patients with type 2diabetes, including urinary tract infec-tions, and AEs related to osmotic diure-sis and volume depletion. The incidenceof female genital mycotic infectionswas higher with canagliflozin 300 mg(21.2%) than with canagliflozin 100 mg

(4.2%) and placebo (5.6%); no malepatients experienced a genital mycoticinfection.

At week 18, the incidence of ketone-related AEs was 5.1% (6 of 117) withcanagliflozin 100 mg and 9.4% (11 of117)with canagliflozin 300mg; nopatientsin the placebo group had a ketone-related AE (Table 3). Most of the ketone-related AEs (12 of 18) occurred after 1month of treatment. Of the 17 patientswith 18 ketone-related AEs, 12 werewomen, and 4 (24%) had a history ofDKA compared with 24 of 217 (11%) ofcanagliflozin-treated patients without aketone-related AE. A serious AE of DKAthat required hospitalization occurred in12 patients (5 [4.3%] with canagliflozin

100 mg and 7 [6.0%] with canagliflozin300 mg). Among the 12 patients with aserious AE of DKA, blood glucose levels atthe timeof hospitalization ranged from9.4to .44.4 mmol/L (170 to .800 mg/dL),and 5 patients had blood glucose levels,13.9 mmol/L (,250 mg/dL), the classiccut point in defining DKA (4). All seriousAEs of DKA were associated with precipi-tating factors at the time of the event (e.g.,flu, pneumonia, insulin pump failure ormalfunction, inappropriate insulin use);none were fatal and none led to discontin-uation of study drug. One patient with aserious AE of DKA discontinued due to asubsequent nonserious AE of increasedurine ketones.

CONCLUSIONS

Canagliflozin 100 and 300 mg providedreductions in HbA1c, FPG, and bodyweight versus placebo in conjunctionwith insulin therapy over 18 weeks inpatients with type 1 diabetes. The pro-portion of patients achieving the pri-mary composite end point of HbA1c

reduction $0.4% ($4.4 mmol/mol)and no increase in body weight was sim-ilar with canagliflozin 100 and 300mg andgreater compared with patients takingplacebo. In addition to glycemic improve-ments and body weight reductions,decreases in the total daily insulindose were seen with canagliflozin over18 weeks that were mainly driven by de-creases in the basal insulin dose. Greaterreductions in total, basal, and bolus insulindoses were seenwith canagliflozin 300mgcompared with canagliflozin 100 mg andplacebo. Canagliflozin treatment was notassociated with an increase in hypogly-cemic episodes over 18 weeks, despitethe greater reductions in HbA1c seen withcanagliflozin 100 and 300 mg versus pla-cebo. The reductions in HbA1c, FPG, body

Table 2—Treatment-emergent hypoglycemia episodes over 18 weeks

PlaceboCanagliflozin

(n = 117) 100 mg (n = 117) 300 mg (n = 117)

Patients with anyDocumented hypoglycemia, n (%)* 113 (96.6) 115 (98.3) 116 (99.1)Event rate per patient-year of exposure (95% CI) 80.6 (77.8, 83.4) 70.7 (68.1, 73.3) 79.2 (76.4, 82.0)

Documented symptomatic hypoglycemia, n (%)† 109 (93.2) 112 (95.7) 111 (94.9)Event rate per patient-year of exposure (95% CI) 56.1 (53.8, 58.4) 50.6 (48.4, 52.8) 47.3 (45.2, 49.5)

Severe hypoglycemia, n (%)‡ 2 (1.7) 3 (2.6) 8 (6.8)

*Concurrent fingerstick or plasma glucose#3.9 mmol/L (#70 mg/dL) or a severe hypoglycemic episode. †Symptoms consistent with hypoglycemiawith concurrent biochemically documented plasma glucose#3.9 mmol/L (#70 mg/dL) or a severe hypoglycemic episode. ‡Requiring assistance fromanother person or those resulting in seizure or loss of consciousness.

Table 3—Summary of overall safety and selected AEs over 18 weeks

PlaceboCanagliflozin

(n = 117) 100 mg (n = 117) 300 mg (n = 117)

Any AE 64 (54.7) 65 (55.6) 79 (67.5)

AEs leading to discontinuation 0 1 (0.9) 2 (1.7)

AEs related to the study drug* 15 (12.8) 19 (16.2) 34 (29.1)

Serious AEs 0 9 (7.7) 8 (6.8)

Deaths 0 0 0

Urinary tract infections 2 (1.7) 5 (4.3) 6 (5.1)

Genital mycotic infectionsMale 0 0 0Female†‡ 3 (5.6) 2 (4.2) 11 (21.2)

Osmotic diuresis–related AEs§ 3 (2.6) 9 (7.7) 11 (9.4)

Volume depletion–related AEs| 0 4 (3.4) 1 (0.9)

Ketone-related AEs¶ 0 6 (5.1) 11 (9.4)#Serious DKA AEs** 0 5 (4.3) 7 (6.0)Nonserious AEs†† 0 1 (0.9) 5 (4.3)

Data are n (%). *Possibly, probably, or very likely related to study drug, as assessed byinvestigators. †Placebo, n = 54; canagliflozin 100 mg, n = 48; canagliflozin 300 mg, n = 52.‡Including vaginal infection, vulvovaginal candidiasis, and vulvovaginal mycotic infection.§Including dry mouth, nocturia, pollakiuria, polydipsia, polyuria, thirst, and urine outputincreased. |Including dehydration, dizziness postural, hypotension, and syncope. ¶IncludingDKA, ketoacidosis, and urine ketone body present. #One patient had an initial serious DKA eventand a subsequent nonserious DKA event of increased urine ketones. **Requiring hospitalization.††Including increased urinary ketones and mild and moderate DKA or acidosis.

care.diabetesjournals.org Henry and Associates 2263

weight, insulin dose, and the incidenceof hypoglycemia with canagliflozin treat-ment are consistent with the effectsreported with other SGLT2 inhibitorsadded on to insulin therapy in patientswith type 1 diabetes (17,18).Dose-related reductions in HbA1c

were seen with canagliflozin acrossphase 3 studies in patients with type 2diabetes (10), but no dose response wasobserved in the current study. In thisstudy, HbA1c lowering with canagliflozin300 mg may have been blunted by re-ductions in the insulin dose, whichtended to be greater than those seenwith canagliflozin 100 mg and may ex-plain why a dose response in HbA1c

lowering was not seen in patients withtype1diabetes.Althoughdose-dependentreductions in insulin may minimize theglycemic effects of canagliflozin, theywould be expected to further increasebody weight reduction because insulinis associated with weight gain. Consis-tent with this, dose-related reductionsin body weight were seen with canagli-flozin in this study, similar to findingsfrom the phase 3 program in type 2 di-abetes (10). The high baseline BMI inthis study (28.1 kg/m2) reflects an over-weight population, which may partly bedue toweight gain with intensive insulintherapy and suggests that patients withtype 1 diabetes may benefit from weightloss, similar to patients with type 2diabetes.The safety and tolerability profile of

canagliflozin was generally similar tothat seen in patients with type 2 diabe-tes (10,14), with the exception ofketone-related AEs, including DKA. Theincidence of overall AEs was higher withcanagliflozin 300 mg relative to canagli-flozin 100 mg and placebo; rates of AEsleading to discontinuation were low(,2.0%) with both canagliflozin doses.A higher rate of serious AEs was seenwith canagliflozin 100 and 300 mg ver-sus placebo that was driven by an in-crease in serious AEs of DKA.The incidence of nonserious and seri-

ous ketone-related AEs was higher withboth canagliflozin doses versus placebo.All patients with a serious AE of DKA hadprecipitating factors that likely contrib-uted to the development of DKA (e.g.,infection, pump failure, cessation of insu-lin therapy) (21). A few patients alsoreported changes in diet, including strictcarbohydrate restriction, which previously

was shown to causeDKA in a patient treat-ed with the SGLT2 inhibitor ipragliflozin(22). DKA events have also been reportedin pilot studies with other SGLT2 inhibitorsin patients with type 1 diabetes. In a 4-week study with sotagliflozin, 2 of 16 pa-tients in the sotagliflozin armhad a seriousAE of DKA compared with no patients inthe placebo arm (19). In an 8-week open-label study with empagliflozin, 2 of 40 pa-tients experienced a DKA AE (18). Thesefour reported DKA AEs occurred in thepresence of other precipitating factors(e.g., insulin pump failure, gastroenteritis).

The U.S. Food and Drug Administra-tion and European Medicines Agencyhave issued statements cautioning thattreatment with SGLT2 inhibitors may beassociated with an increased risk of DKA(23,24). Rare but serious cases of DKAhave been reported in patients withtype 1 or type 2 diabetes treated withSGLT2 inhibitors, including canagliflozin(25). These cases of DKA were not typi-cal because patients had nearly normalblood glucose levels and showedfew symptoms. Furthermore, a recentmeta-analysis of clinical trial data dem-onstrated an increased risk of ketoacido-sis with SGLT2 inhibitor treatment inpatients with type 2 diabetes (26). A sep-arate analysis based on data from 17,596patients reported a low frequency(,0.1%) of serious DKA events acrosstreatment groups in the canagliflozintype 2 diabetes clinical trial program(27). Among the 12 patients with seriousDKA events in that report, all had a lowb-cell reserve and 7 patients treatedwith canagliflozin were on insulin ther-apy. Most of the patients with DKA alsohad DKA-precipitating factors, including6 with evidence of autoimmune diabe-tes (i.e., latent autoimmune diabetes ofadulthood, type 1 diabetes, or positivefor GAD65 antibody). We speculatethat patients with type 1 or type 2 dia-betes who have no or low b-cell reserve,coupled with a potential SGLT2 inhibitor–associated increase in glucagon (28–30)and other metabolic factors, such as ele-vated free fatty acids, may not take ormay be unable to produce sufficient in-sulin to suppress hepatic ketogenesis(28–30), which can progress to DKA inthe setting of an acute illness, inadequatecarbohydrate intake, and an associatedincrease in insulin resistance.

Canagliflozin is not indicated for usein patients with type 1 diabetes. However,

mitigation strategies to reduce the riskof DKA in patients with type 1 diabetestreated with canagliflozin will be neededfor use in future clinical trials. Proposedstrategies include more frequent moni-toring of ketones, interruption of treat-ment during periods of stress or illness,and use of lower canagliflozin doses. Fur-thermore, treating physicians must bemade aware that DKA can occur withmild hyperglycemia, which may increasethe risk of delayed diagnosis and treat-ment in patients treated with SGLT2inhibitors.

In summary, canagliflozin 100 and 300mg improved glycemic control and re-duced body weight compared with pla-cebo over 18 weeks in patients withtype 1 diabetes in conjunction with insu-lin therapy. In addition, canagliflozintreatment was not associated with anincreased risk of hypoglycemia. The insulin-independent mechanism of action ofcanagliflozin offers a unique benefitprofile in patients with type 1 diabetes,which is different from that of stan-dard insulin therapies that are associ-ated with hypoglycemia and weightgain. An increase in DKA events wasseen with canagliflozin that may be re-lated to the presence of a condition char-acterized by reduced insulin dose orinsulin resistance (e.g., illness). Imple-mentation of additional mitigation strat-egies in future studies may allow forearlier identification of elevated ketonelevels and earlier intervention that maysubstantially reduce DKA risk in patientswith type 1 diabetes.

Acknowledgments. The authors thank all in-vestigators, study teams, and patients for par-ticipating in this study.Funding and Duality of Interest. Editorialsupport was provided by Kimberly Fuller, PhD,of MedErgy, and was funded by Janssen GlobalServices, LLC. Canagliflozin has been developedbyJanssen Research & Development, LLC, in collab-oration with Mitsubishi Tanabe Pharma Corpo-ration. R.R.H. has received support from aDepartment of Veterans Affairs Merit Reviewgrant and the VA San Diego Healthcare System.

This study was supported by Janssen Re-search&Development, LLC. R.R.H. has receivedgrant support from Hitachi, Janssen Re-search & Development, LLC, Eli Lilly and Com-pany, Sanofi, and ViaCyte, Inc., and is aconsultant/advisory board member for Alere,Amgen Inc., AstraZeneca, Boehringer IngelheimGmbH, ClinMet, Eisai Co., Ltd., ElcelyxTherapeutics,Inc., Gilead, Intarcia Therapeutics, Inc., Isis Pharma-ceuticals, Inc., Janssen Research & Development,

2264 Canagliflozin in Patients With Type 1 Diabetes Diabetes Care Volume 38, December 2015

LLC, Merck & Co., Inc., Novo Nordisk, Sanofi,and VIVUS Inc. P.T., C.T., D.P., and M.A. are full-time employees of Janssen Research & Develop-ment, LLC. No other potential conflicts of interestrelevant to this article were reported.AuthorContributions. R.R.H. and P.T. contrib-uted to the conduct of the study; contributed tothe acquisition, analysis, and interpretation ofdata; and drafted, reviewed, and approved themanuscript. C.T. and D.P. contributed to theanalysis and interpretation of data and drafted,reviewed, and approved the manuscript. M.A.contributed to the design and conduct of thestudy; contributed to the acquisition, analysis,and interpretation of data; and drafted, re-viewed, and approved the manuscript. R.R.H.andM.A. are the guarantors of this work and, assuch, had full access to all of the data in the studyand take responsibility for the integrity of thedata and the accuracy of the data analysis.

References1. Chiang JL, Kirkman MS, Laffel LM, Peters AL;Type 1 Diabetes Sourcebook Authors. Type 1diabetes through the life span: a position state-ment of the American Diabetes Association.Diabetes Care 2014;37:2034–20542. Aschner P, Horton E, Leiter LA, Munro N,Skyler JS; Global Partnership for Effective Dia-betes Management. Practical steps to improv-ing the management of type 1 diabetes:recommendations from the Global Partnershipfor Effective Diabetes Management. Int J ClinPract 2010;64:305–3153. Livingstone SJ, Levin D, Looker HC, et al.;Scottish Diabetes Research Network epidemiol-ogy group; Scottish Renal Registry. Estimatedlife expectancy in a Scottish cohort with type 1diabetes, 2008-2010. JAMA 2015;313:37–444. Trachtenbarg DE. Diabetic ketoacidosis. AmFam Physician 2005;71:1705–17145. The Diabetes Control and Complications TrialResearch Group. The effect of intensive treatmentof diabetes on the development and progressionof long-term complications in insulin-dependentdiabetesmellitus. N Engl JMed 1993;329:977–9866. Nathan DM, Cleary PA, Backlund JY, et al.;Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions andComplications (DCCT/EDIC) Study ResearchGroup.Intensive diabetes treatment and cardiovasculardisease in patients with type 1 diabetes. N Engl JMed 2005;353:2643–26537. Miller KM, Foster NC, Beck RW, et al.; T1DExchange Clinic Network. Current state of type 1diabetes treatment in the U.S.: updated data

from the T1D Exchange clinic registry. DiabetesCare 2015;38:971–9788. SYMLIN (pramlintide acetate) injection for sub-cutaneous use [package insert]. Wilmington, DE,AstraZeneca Pharmaceuticals LP; 20159. Kruger DF, Gloster MA. Pramlintide for thetreatment of insulin-requiring diabetes melli-tus: rationale and review of clinical data. Drugs2004;64:1419–143210. Rosenthal N, Meininger G, Ways K, et al.Canagliflozin: a sodium glucose co-transporter2 inhibitor for the treatment of type 2 diabetesmellitus. Ann N Y Acad Sci 25 August 2015 [Epubahead of print]. DOI: 10.1111/nyas.1285211. Devineni D, Morrow L, Hompesch M, et al.Canagliflozin improves glycaemic control over28 days in subjects with type 2 diabetes notoptimally controlled on insulin. Diabetes ObesMetab 2012;14:539–54512. Sha S,DevineniD,GhoshA, et al. Canagliflozin,a novel inhibitor of sodium glucose co-transporter2, dose dependently reduces calculated renalthreshold for glucose excretion and increases uri-nary glucose excretion in healthy subjects. Diabe-tes Obes Metab 2011;13:669–67213. Polidori D, Sha S, Ghosh A, Plum-MorschelL, Heise T, Rothenberg P. Validation of a novelmethod for determining the renal threshold forglucose excretion in untreated and canagliflozin-treated subjects with type 2 diabetes mellitus.J Clin Endocrinol Metab 2013;98:E867–E87114. Usiskin K, Kline I, Fung A,Mayer C,MeiningerG. Safety and tolerability of canagliflozin in pa-tients with type 2 diabetes mellitus: pooled anal-ysis of phase 3 study results. PostgradMed 2014;126:16–3415. INVOKANA (canagliflozin) tablets, for oraluse [package insert]. Titusville, NJ, Janssen Phar-maceuticals; 201416. Neal B, Perkovic V, de ZeeuwD, et al.; CANVASTrial Collaborative Group. Efficacy and safety ofcanagliflozin, an inhibitor of sodium-glucose co-transporter 2, when used in conjunction with in-sulin therapy in patients with type 2 diabetes.Diabetes Care 2015;38:403–41117. Henry RR, Rosenstock J, Edelman S, et al.Exploring the potential of the SGLT2 inhibitordapagliflozin in type 1 diabetes: a randomized,double-blind, placebo-controlled pilot study.Diabetes Care 2015;38:412–41918. Perkins BA, Cherney DZ, Partridge H, et al.Sodium-glucose cotransporter 2 inhibition andglycemic control in type 1 diabetes: results of an8-week open-label proof-of-concept trial. Dia-betes Care 2014;37:1480–148319. Sands AT, Zambrowicz BP, Rosenstock J,et al. Sotagliflozin, a dual SGLT1 and SGLT2

inhibitor, as adjunct therapy to insulin intype 1 diabetes. Diabetes Care 2015;38:1181–118820. Borus JS, Laffel L. Adherence challenges inthe management of type 1 diabetes in adoles-cents: prevention and intervention. Curr OpinPediatr 2010;22:405–41121. Perilli G, Saraceni C, Daniels MN, Ahmad A.Diabetic ketoacidosis: a review and update. CurrEmerg Hosp Med Rep 2013;1:10–1722. Hayami T, Kato Y, Kamiya H, et al. Case of ke-toacidosis by a sodium-glucose cotransporter 2 in-hibitor in adiabetic patientwith a low-carbohydratediet. J Diabetes Invest 2015;6:587–59023. Food and Drug Administration. FDA drugsafety communication: FDA warns that SGLT2inhibitors for diabetes may result in a seriouscondition of too much acid in the blood [articleonline], 2015. Available from http://www.fda.gov/Drugs/DrugSafety/ucm446845.htm.Accessed 19 May 201524. EuropeanMedicinesAgency (EMA). Reviewofdiabetesmedicines called SGLT2 inhibitors started.Risk of diabetic ketoacidosis to be examined [arti-cle online], 2015. Available from http://www.ema.europa.eu/ema/index.jsp?curl=pages/medicines/human/referrals/SGLT2_inhibitors/human_referral_prac_000052.jsp&mid=WC0b01ac05805c516f. Ac-cessed 25 June 201525. Peters AL, Buschur EO, Buse JB, Cohan P,Diner JC, Hirsch IB. Euglycemic diabetic ketoaci-dosis: a potential complication of treatmentwith sodium-glucose cotransporter 2 inhibition.Diabetes Care 2015;38:1687–169326. Yabe D, Nishikino R, Kaneko M, Iwasaki M,Seino Y. Short-term impacts of sodium/glucoseco-transporter 2 inhibitors in Japanese clinicalpractice: considerations for their appropriateuse to avoid serious adverse events. ExpertOpin Drug Saf 2015;14:795–80027. Erondu N, Desai M, Ways K, Meininger G.Diabetic ketoacidosis and related events in thecanagliflozin type 2 diabetes clinical program.Diabetes Care 2015;38:1680–168628. Ferrannini E, Muscelli E, Frascerra S, et al.Metabolic response tosodium-glucosecotransporter2 inhibition in type 2 diabetic patients. J ClinInvest 2014;124:499–50829. Merovci A, Solis-Herrera C, Daniele G, et al.Dapagliflozin improves muscle insulin sensitiv-ity but enhances endogenous glucose produc-tion. J Clin Invest 2014;124:509–51430. Bonner C, Kerr-Conte J, Gmyr V, et al.Inhibition of the glucose transporter SGLT2with dapagliflozin in pancreatic alpha cellstriggers glucagon secretion. Nat Med 2015;21:512–517

care.diabetesjournals.org Henry and Associates 2265