Clinical Care/Education/Nutrition/Psychosocial Research ORIGINAL ARTICLE

Canagliflozin Lowers Postprandial and by Delaying Intestinal Glucose Absorption in Addition to Increasing Urinary Glucose Excretion Results of a randomized, placebo-controlled study

1 2 DAVID POLIDORI, PHD NICOLE VACCARO, BS low-capacity, high-affinity transporter ex- 2 2 SUE SHA, MD, PHD KRISTIN FARRELL, BS 3 2 pressed in the distal segment of the prox- SUNDER MUDALIAR, MD PAUL ROTHENBERG, MD, PHD 3 3 imal tubule (1), in the intestinal mucosa THEODORE P. CIARALDI, PHD ROBERT R. HENRY, MD 2 of the small intestine (2), and in other tis- ATALANTA GHOSH, PHD sues to a lesser extent (3). Although SGLT1 plays a smaller role in renal glu- cose absorption than SGLT2, SGLT1 is OBJECTIVEdCanagliflozin, a sodium glucose cotransporter 2 (SGLT2) inhibitor, is also a low-potency SGLT1 inhibitor. This study tested the hypothesis that intestinal canagliflozin levels the primary pathway involved in intesti- postdose are sufficiently high to transiently inhibit intestinal SGLT1, thereby delaying intestinal nal glucose and galactose absorption glucose absorption. (2,4,5). Pharmacologic inhibition of SGLT2 RESEARCH DESIGN AND METHODSdThis two-period, crossover study evaluated is a novel approach to lowering plasma fl effects of canagli ozin on intestinal glucose absorption in 20 healthy subjects using a dual-tracer glucose in hyperglycemic individuals by method. Placebo or canagliflozin 300 mg was given 20 min before a 600-kcal mixed-meal tolerance test. Plasma glucose, 3H-glucose, 14C-glucose, and insulin were measured frequently blocking renal glucose reabsorption, low- ering the renal threshold for glucose for 6 h to calculate rates of appearance of oral glucose (RaO) in plasma, endogenous glucose production, and glucose disposal. (RTG), and thereby markedly increasing urinary glucose excretion (UGE). Cana- RESULTSdCompared with placebo, canagliflozin treatment reduced postprandial plasma gliflozin, an SGLT2 inhibitor in develop- glucose and insulin excursions (incremental 0- to 2-h area under the curve [AUC0–2h] reductions ment for the treatment of patients with of 35% and 43%, respectively; P , 0.001 for both), increased 0- to 6-h urinary glucose excretion – 6 , P , fl type 2 (6 10), is also a low-potency (UGE0–6h,18.2 5.6 vs 0.2 g; 0.001), and delayed RaO. Canagli ozin reduced AUC RaO fl by 31% over 0 to 1 h (geometric means, 264 vs. 381 mg/kg; P , 0.001) and by 20% over 0 to 2 h SGLT1 inhibitor. In vitro, canagli ozin in- P fl hibited sodium-dependent 14C-a-methyl- (576 vs. 723 mg/kg; = 0.002). Over 2 to 6 h, canagli ozin increased RaO such that total AUC , P RaO over 0 to 6 h was 6% lower versus placebo (960 vs. 1,018 mg/kg; = 0.003). A modest glucoside uptake in cells expressing (;10%) reduction in acetaminophen absorption was observed over the first 2 h, but this differ- human SGLT2 or SGLT1 with half-maximal fi ence was not suf cient to explain the reduction in RaO. Total glucose disposal over 0 to 6 h was inhibitory concentrations (IC50) of 4.4 6 similar across groups. 1.2 and 684 6 159 nmol/L, respectively (8). Because the maximum plasma con- CONCLUSIONSdCanagliflozin reduces postprandial plasma glucose and insulin by increas- centrations of unbound canagliflozin in ing UGE (via renal SGLT2 inhibition) and delaying RaO, likely due to intestinal SGLT1 inhibition. subjects treated with canagliflozin 300 mg once-daily are ;100 nmol/L (maximum plasma concentrations are ;10 mmol/L ; he sodium glucose cotransporter 2 (1). SGLT2 is a high-capacity, low-affinity (11) and protein binding is 99% [un- (SGLT2) is the major transporter transporter expressed primarily at the lu- published data]), only minimal systemic T inhibition of SGLT1 is expected in sub- responsible for reabsorption of glu- minal membrane of the early segments of fl cose filtered through the renal glomerulus the proximal renal tubules (1). SGLT1 is a jects treated with canagli ozin 300 mg. In clinical studies in healthy subjects and subjects with , treat- ccccccccccccccccccccccccccccccccccccccccccccccccc ment with canagliflozin provided dose- 1 2 From Janssen Research & Development, LLC, San Diego, California; Janssen Research & Development, LLC, dependent increases in UGE compared Raritan, New Jersey; and the 3University of California, San Diego, San Diego, California Corresponding author: David Polidori, [email protected]. with placebo (7,9). In healthy subjects Received 15 November 2012 and accepted 07 January 2013. treated with escalating doses of canagliflo- DOI: 10.2337/dc12-2391. Clinical trial reg. no. NCT01173549, clinicaltrials.gov. zin given 10 min before a mixed meal, This article contains Supplementary Data online at http://care.diabetesjournals.org/lookup/suppl/doi:10 doses of canagliflozin higher than 200 .2337/dc12-2391/-/DC1. © 2013 by the American Diabetes Association. Readers may use this article as long as the work is properly mg reduced postprandial plasma glucose cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons.org/ and insulin concentrations to a greater ex- licenses/by-nc-nd/3.0/ for details. tent than lower doses of canagliflozin, care.diabetesjournals.org DIABETES CARE 1 Diabetes Care Publish Ahead of Print, published online February 14, 2013 Canagliflozin delays gut glucose absorption even when compared with doses that pro- Safety analyses 180, 210, 240, 270, 300, and 360 min for vided similar UGE during the postpran- Vital sign measurements, 12-lead electro- measurements of plasma glucose (labeled dial period (7). These pronounced cardiograms, physical examinations, and and unlabeled) and insulin. Additional reductions in postprandial glucose and clinical laboratory tests were performed at blood samples were drawn at predefined insulin excursions observed with canagli- predefined time points throughout the time points for determination of canagli- flozin doses higher than 200 mg were study. Adverse events (AEs) were moni- flozin and acetaminophen concentrations only observed for the first meal after dos- tored from the signing of informed consent and for analysis of other pharmacody- ing; similar reductions beyond that ex- until completion of the last study-related namic markers, including concentrations pected on the basis of increased UGE procedure. of glucose-dependent insulinotropic pep- were not observed after later meals (lunch tide (GIP), glucagon-like peptide-1 (GLP- and dinner) given on the same day (7). On Study design 1), and peptide YY (PYY). the basis of these observations, it was hy- This was a randomized, double-blind, pothesized that after dosing and during placebo-controlled, two-period crossover Bioanalytical analyses drug absorption, canagliflozin concentra- study consisting of a screening phase, a Plasma and urine glucose concentrations tions within the lumen of the intestinal 25-day double-blind treatment phase (in- were measured using a hexokinase en- tract could be sufficiently high to provide cluding two 1-day treatment periods zymatic assay, and plasma insulin con- transient inhibition of intestinal SGLT1- and a washout period of 7–21 days be- centration was determined using an mediated glucose absorption, thereby tween periods 1 and 2), and a follow-up electrochemiluminescent sandwich im- lowering postprandial plasma glucose phase of up to 10 days after period 2. munoassay (Roche Diagnostics, Indian- and insulin concentrations. Subjects were randomized to one of two apolis, IN). 3H-glucose and 14C-glucose The current study investigated the ef- treatment sequences: canagliflozin 300 specific activities were determined using fects of a single 300-mg oral dose of canagli- mg in period 1, followed by matching pla- the assays described by Mudaliar et al. flozin on intestinal glucose absorption cebo in period 2, or vice versa. On days –3 (14) and Kreisberg et al. (15), respec- and metabolism in healthy subjects (Clin- and –2 of each period, subjects were tively. Recycling of 14C-glucose over the icalTrials.gov Identifier: NCT01173549). counseled to adhere to a specified diet course of the procedure was negligible, This study used a dual-tracer method to (;55% carbohydrate, 30% fat, 15% pro- with recycled 14C generally below the test the hypothesis that canagliflozin 300 mg tein, and total caloric intake of ;30 kcal/kg limit of detection, similar to the observa- slows the rate of systemic appearance of body weight). On the morning of day –1 tions of others (16,17). Active and total orally administered glucose (R O) during a of each study period, subjects were admit- a plasma GLP-1 levels were measured using mixed-meal tolerance test (MMTT) com- ted to the clinical research unit in a fasting an electrochemiluminescent sandwich pared with placebo. state for safety analyses, followed by stan- immunoassay (Meso Scale Discovery, dardized meals. Gaithersburg, MD). Total plasma GIP RESEARCH DESIGN AND On the morning of day 1 of each study was measured using an enzyme-linked METHODS period, after an overnight fast of at least 8 h, immunosorbent assay, and total plasma subjects received a primed (25 mCi), PYY was measured using a radioimmuno- Study population continuous intravenous infusion of assay (Millipore, Billerica, MA). Plasma This study was conducted from 22 No- 3H-glucose (0.25 mCi/min) for approxi- acetaminophen concentration was deter- vember 2010 to 29 September 2011 at a mately 9 h. Three hours after starting the mined using a validated high-performance single center in San Diego, California. The intravenous infusion and 20 min after ad- liquid chromatography (HPLC) method study protocol and all amendments were ministration of study drug, subjects with ultraviolet detection at PRA Interna- reviewed and approved by the University received a standard 600-kcal MMTT tional, Assen, The Netherlands (calibra- of California, San Diego Institutional Re- (55% carbohydrate, 30% fat, 15% protein). tion range = 0.500–50 mg/mL). view Board. This study was conducted in The liquid component of the MMTT con- accordance with the ethical principles sisted of an oral solution of 75 g glucose, that have their origin in the Declaration which was mixed with 75 mCi 14C-glucose, Glucose flux analysis of Helsinki and that are consistent with and an acetaminophen solution (960 mg in RaO, the rate of endogenous glucose pro- good clinical practices and applicable 30 mL); acetaminophen absorption was duction (EGP), and the rate of total glu- regulatory requirements. All subjects pro- used as an indirect measure of gastric emp- cose disposal (Rd) were determined from vided written informed consent to partic- tying (12,13). Subjects returned for a final the measured plasma glucose, 3H-glucose, ipate in the study after having been follow-up visit for safety analyses 7 to 10 and 14C-glucose profiles using a circula- informed about the nature and purpose days after discharge on day 1 of period 2. tory model of glucose kinetics (18–20). of the study, participation/termination The infused 3H-glucose profile was used conditions, and the possible risks and Clinical evaluations to determine glucose clearance over time, benefits of treatment. On day 1, subjects emptied their bladder and the 14C-glucose and plasma glucose This study enrolled healthy men aged before and after the 3-h 3H-glucose iso- profiles were used to determine the rela- 18 to 45 years with a BMI of $20 and tope equilibration period (from t = –3hto tive amounts of ingested and endogenous #27 kg/m2, stable body weight of $50 t = 0 [the start of the meal]), and urine was glucose in the circulation. The resulting kg (,5% change during the 3 months be- collected over the intervals of 0 to 2 h and rates of appearance of endogenous and fore screening), and fasting plasma glucose 2 to 6 h for determination of urinary glu- oral glucose were determined by fitting (FPG) of ,6.1 mmol/L. Further eligibility cose and creatinine concentrations. Blood the model of glucose kinetics to the oral criteria are described at ClinicalTrials.gov samples were collected at t = –20, –10, 0, and endogenous glucose profiles. All cal- (NCT01173549). 15, 30, 40, 50, 60, 75, 90, 105, 120, 150, culations were performed using the

2 DIABETES CARE care.diabetesjournals.org Polidori and Associates

GLUTRAN toolbox (licensed from Mari statistics for each treatment. Mean (95% Subject baseline and demographic char- and colleagues at the Institute of Biomed- CIs) differences between canagliflozin acteristics are summarized in Table 1. ical Engineering, National Research 300 mg and placebo were determined Council, Padova, Italy) in Matlab 7.10 for all pharmacodynamic variables. To- Postprandial plasma glucose and software (18). tal Rd was analyzed using a mixed-effects insulin responses To separate total Rd into tissue glu- linear model based on log-transformed After ingestion of the standardized meal, cose disposal (tissue Rd) and UGE, the data. The model included the logarithm thepostprandialplasmaglucoseandin- rate of UGE at each time point was esti- of total Rd as the dependent variable; se- sulin excursions were reduced with can- mated from the UGE collections over the quence group, period, and treatment as agliflozin 300 mg compared with placebo 0- to 2-h and 2- to 6-h intervals. This was fixed effects; and subject as a random (Fig. 1A and B). Mean postprandial done by calculating RTG over these inter- effect. plasma glucose ΔAUC values were vals, as previously described (7,9), and Acetaminophen pharmacokinetic pa- ;44%, 35%, and 26% lower during the then estimating UGE at each interval us- rameters (Cmax and AUCs) were com- 0- to 1-h, 0- to 2-h, and 0- to 6-h postmeal ing the following equation: pared between canagliflozin and placebo intervals, respectively, after administra- groups by constructing a 90% CI for the tion of canagliflozin 300 mg (arithmetic rate of UGE ðmmol=minÞ  ratio of geometric means using mixed- mean 6 SD ΔAUC of 1.88 6 0.77, 3.67 6 GFR ðL=minÞ3ðPG ðmMÞ2 RTGðmMÞÞ if PG.RTG ¼ effects modeling of the data with the loga- 1.41, and 4.80 6 1.90 [mmol/L] z h, re- PG#RT 0ifG rithm of the pharmacokinetic parameters spectively) compared with placebo as the dependent variable; sequence, pe- (3.34 6 1.43, 5.61 6 2.62, and 6.44 6 fi GFR where the glomerular ltration rate ( ) riod, and treatment as fixed effects; and 2.57 [mmol/L] z h, respectively). Simi- fi was estimated using the Modi cation of subject as a random effect. The relation- larly, postprandial plasma insulin ΔAUC Diet in Renal Disease equation and PG is ship between RaO AUC and AUC acet- values were ;43%, 43%, and 33% lower plasma glucose. Tissue Rd was then cal- aminophen was assessed by linear during the 0- to 1-h, 0- to 2-h, and 0- to 2 culated as total Rd UGE. regression analysis. Comparison of the re- 6-h postmeal intervals, respectively, for gression lines for canagliflozin versus pla- canagliflozin (arithmetic mean 6 SD Pharmacodynamic parameters cebo was performed using an ANCOVA ΔAUC of 271 6 232, 514 6 273, and The total amount of glucose appearance model in Prism 5.01 software (GraphPad 738 6 335 [pmol/L] z h, respectively) com- and disappearance over the intervals from Software, Inc., La Jolla, CA). pared with placebo (474 6 263, 906 6 521, 0to1h,0to2h,2to6h,and0to6hwere For all subjects, based on individual and 1,104 6 624 [pmol/L] z h, respectively). determined by calculating the area under plasma concentration-time profiles, the The oral 14C-glucose tracer concen- the curve (AUC) of the associated rates of total and incremental AUCs for glucose, tration was reduced with canagliflozin glucose appearance- and disappearance- insulin, PYY, GIP, and GLP-1 from 0 to 1 compared with placebo over the first 2 h fi time pro les over each time interval. h, 0 to 2 h, and 0 to 6 h (glucose and after the MMTT (Fig. 1C). The infused D Incremental AUCs (denoted as AUC) insulin only) were calculated using the 3H-glucose tracer concentration was sim- for plasma glucose, insulin, and gut pep- trapezoid rule using WinNonlin 5.2.1 ilar between treatments for the first 3 h fi tides were de ned as the positive area software (Pharsight Corporation, Moun- after the MMTT; between 3 and 6 h after above the premeal value. tain View, CA). Unless otherwise speci- the MMTT, the plasma concentration of fied, all results shown are mean 6 SD. 3H-glucose was lower with canagliflozin Statistical analysis treatment than with placebo (Fig. 1D). For the primary pharmacodynamic pa- RESULTSdThis study enrolled 20 sub- rameters of RaOAUC0–1h and RaOAUC0–2h, jects and 19 completed both treatment RaO in plasma data were log-transformed for analysis. periods (1 subject was withdrawn for Canagliflozin treatment blunted and de- Mixed-effects models were fitted with the noncompliance with study prohibitions). layed the postprandial increase in RaO logarithm of the parameter of interest as the dependent variable; sequence, period, d and treatment as fixed effects; and subject Table 1 Subject baseline and demographic characteristics as a random effect. The null hypothesis that the mean RaO AUCs (on the log-scale) Sequence 1 Sequence 2 are equal for canagliflozin 300 mg and CANA 300 mg → placebo Placebo → CANA 300 mg placebo was tested using a one-sided Characteristic (n =10) (n = 10) (left-sided) a level of 5%. Using the esti- Male, n (%) 10 (100) 10 (100) mated least squares (LS) means and intra- Age (years), mean 6 SD 25.9 6 4.4 26.9 6 7.0 subject SD from the mixed-effects model, Race, n (%) 90% CIs were constructed for the differ- White 8 (80) 7 (70) ence in means on the log scale between Black or African American 0 1 (10) canagliflozin 300 mg and placebo. The Native Hawaiian/other Pacific CI limits for the difference in mean Islander 1 (10) 1 (10) AUCs were exponentiated to yield the Multiple 1 (10) 1 (10) 90% CI for the ratio of geometric mean Body weight (kg), mean 6 SD 82.1 6 9.5 74.2 6 8.9 R O AUCs of canagliflozin to placebo. a BMI (kg/m2), mean 6 SD 24.4 6 2.0 23.4 6 2.2 All secondary pharmacodynamic var- iables were summarized with descriptive CANA, canagliflozin. care.diabetesjournals.org DIABETES CARE 3 Canagliflozin delays gut glucose absorption

Figure 1dMean 6 SEM concentration-time profiles from predose to 6 h after the standard meal for plasma glucose (A), insulin (B), oral 14C- glucose tracer (C), and infused 3H-glucose tracer (D); rate of oral glucose appearance (E); and glucose absorption as a function of time after the standard meal (F). CANA, canagliflozin.

4 DIABETES CARE care.diabetesjournals.org Polidori and Associates compared with placebo (Fig. 1E). Treat- ment with canagliflozin reduced the amount of oral glucose absorption (AUC RaO) compared with placebo by 31% over the 0- to 1-h interval (ratio of LS geomet- ric means [90% CI] of 0.69 [0.60–0.80]; P , 0.001) and by 20% over the 0- to 2-h interval (0.80 [0.71–0.89]; P , 0.01; Fig. 1F). However, this decrease in AUC RaO over the first 2 h with canagliflozin was nearly matched by a 34% increase in AUC RaO in the 2- to 6-h interval for canagli- flozin compared with placebo (1.34 [1.19–1.51]), such that the AUC RaO over 0 to 6 h was only ;6% lower for canagliflozin compared with placebo (0.94 [0.91–0.98]; P = 0.003; Fig. 1F). Administration of canagliflozin also re- duced the maximum RaObyapproxi- mately 24% compared with placebo (LS geometric means of 7.16 and 9.38 mg/kg/min, respectively).

UGE and RTG Canagliflozin treatment increased UGE (Fig. 2A). Mean UGE after treatment with canagliflozin was 5.9 and 12.2 g dur- ing the 0- to 2-h and 2- to 6-h intervals, respectively, compared with less than 0.15 g in each interval with placebo. In fl canagli ozin-treated subjects, mean RTG was 4.1 6 1.8 mmol/L over the 0- to 2-h interval and 2.4 6 0.8mmol/Loverthe2- to 6-h interval. RTG values could not be determined in placebo-treated subjects because most had only minimal (,200 mg) UGE, which is consistent with ex- pectations based on the plasma glucose profiles shown in Fig. 1 and with the com- monly reported RTG values of ;10.0 to 11.1 mmol/L in untreated, healthy sub- jects (21,22).

Rates of EGP and Rd Before the MMTT, the rate of EGP was ;2 mg/kg/mininbothtreatmentgroups, Figure 2dEffects of canagliflozin (CANA) treatment on UGE rate (A), EGP rate (B), total R consistent with values reported in other d (C), and tissue Rd (D) over 6 h after the standard meal and on total glucose turnover from 0 to 2 h studies in healthy subjects (23,24). EGP (E)and0to6h(F) postmeal. The calculated mean total amount of oral glucose absorption over was rapidly suppressed after the MMTT, 0 to 6 h was 79 g with placebo and 75 g with canagliflozin. The mean value in the placebo group is with similar suppression of EGP observed slightly higher than the 75-g oral glucose load that was ingested, giving a calculated bio- with canagliflozin or placebo treatments availability of glucose that was slightly .100%. This slight overestimation may be due partly to 14 (Fig. 2B). From 3 to 6 h postmeal (when recycling of the C glucose tracer through the Cori cycle that could not be accounted for due to limitations of assay sensitivity not allowing accurate detection of the recycled tracer amounts (see plasma glucose concentrations were gen- 6 erally back to premeal values), the rate of RESEARCH DESIGN AND METHODS). Values shown are mean SEMinA-DandmeaninE-F. EGP was modestly higher with canagliflo- zin than with placebo (Fig. 2B), and the calculated total amount of EGP over 2 to 6 placebo (Fig. 2A), total and tissue Rd were compared with placebo (P =0.026and hwas;20% higher with canagliflozin both lower with canagliflozin treatment P = 0.009, respectively). Thus, despite than placebo (geometric mean ratio = than with placebo for ;2hafterthe the increase in UGE with canagliflozin P 1.20, = 0.005). MMTT (Fig. 2C and D), with AUC0–1h treatment, the reduction in postprandial fl Although canagli ozin treatment dra- and AUC0–2h for total Rd reduced by plasma glucose and insulin over the first 2 matically increased UGE compared with ;18% and 19% with canagliflozin h is not explained by increased glucose care.diabetesjournals.org DIABETES CARE 5 Canagliflozin delays gut glucose absorption disposal because Rd over that time inter- val is lower with canagliflozin than with placebo. Total Rd increased by ;25% over the 2- to 6-h interval with canagliflo- zin compared with placebo (P , 0.001) so that the total amount of glucose dis- posaloverthe6-hperiod(totalRd AUC0–6h) was essentially identical with canagliflozin and placebo (P = 0.78).

Total glucose turnover The total amounts of glucose appearance and disappearance over the 0- to 2-h and 2- to 6-h intervals are shown in Fig. 2. Over the 0- to 2-h interval, treatment with canagliflozin reduced mean RaO AUC (i.e., the amount of oral glucose ab- sorbed) by 11 g, which was almost twice as large as the increase in UGE during this interval (6 g; Fig. 2E). This suggests that most of the observed reductions in plasma glucose and insulin excursions with canagliflozin treatment are due to decreased intestinal glucose absorption over the first 2 h after the meal. However, the total amounts of glucose appearance and disappearance over the 0- to 6-h in- terval were essentially identical between F treatments (Fig. 2 ). Figure 3dMean 6 SEM plasma concentration-time profiles of GIP (A), PYY (B), total GLP-1 (C), and active GLP-1 (D). CANA, canagliflozin. Gastric emptying and postprandial plasma GIP, PYY, and GLP-1 responses Mean plasma acetaminophen concentra- 30.1 6 11.1 and 63.5 6 20.2 pmol z h/L, intestinal lumen during the period of tions were ;10% lower over the first 2 h respectively). The incremental postpran- drug absorption. after the meal with canagliflozin com- dial PYY concentration was ;60% higher Although canagliflozin reduced intes- pared with placebo (LS geometric mean from 0 to 2 h with canagliflozin than with tinal glucose absorption by ;31% over m z AUC0–2h of 9.21 and 10.29 g h/mL, placebo (21.9 6 13.5 and 13.6 6 10.8, the first hour and by ;20% over the first respectively, giving a LS geometric mean respectively), and total GLP-1 was ;35% 2 h after a meal, this initial reduction was ratio of 0.90 (90% CI 0.84–0.95; P = higher with canagliflozin than with placebo almost entirely compensated for by an in- 0.004; Supplementary Fig. 1A). However, (13.7 6 6.88 and 10.1 6 5.38, respec- crease in glucose absorption from 2 to 6 h this reduction in the gastric emptying rate tively). after the meal. As a result, there was only a is insufficient to explain the decrease in small difference (,6%) in the total ap- fl RaO observed with canagli ozin com- Safety pearance of orally ingested glucose in pared with placebo, because the relation- Canagliflozin was well tolerated, with no plasma over the full 6-h period after the ship between acetaminophen absorption discontinuations due to AEs. No subjects meal between canagliflozin and placebo and glucose absorption was altered by reported symptoms suggestive of glucose treatments. These findings demonstrate canagliflozin treatment (Supplementary malabsorption or gastrointestinal adverse that canagliflozin treatment is not associ- Fig.1B). Over both the 0- to 1-h and 0- effects. ated with any meaningful glucose malab- to 2-h intervals, the relationship between sorption, and consistent with this, no RaO and acetaminophen absorption was CONCLUSIONSdResults of this symptoms of malabsorption were reported shifted downward with canagliflozin study show that in healthy subjects, a in this study. No increase in glucose mal- treatment, demonstrating that the reduc- single 300-mg dose of canagliflozin ad- absorption (assessed using a hydrogen tion in RaO could not be explained solely ministered before a meal reduced post- breath test) was observed in subjects by reductions in gastric emptying. prandial plasma glucose excursions by with type 2 diabetes treated with canagli- Changes in gut peptide concentration two mechanisms: increased UGE due to flozin 300 mg twice daily for 4 weeks (9). during the 0- to 2-h postprandial period renal SGLT2 inhibition and delayed ab- This is in contrast to serious malabsorp- were consistent with delayed intestinal sorption of ingested glucose. The delay in tion symptoms observed in individuals glucose absorption (Fig. 3). Incremental oral glucose absorption is likely due to with inactivating genetic mutations in postprandial GIP was reduced by ;50% local and transient intestinal SGLT1 in- SGLT1 (3). This lack of glucose malab- with canagliflozin compared with placebo hibition resulting from high intestinal sorption in subjects treated with canagli- (arithmetic mean 6 SD ΔAUC0–2h of canagliflozin concentrations within the flozin is likely due to the rapid absorption

6 DIABETES CARE care.diabetesjournals.org Polidori and Associates of canagliflozin, such that intestinal cana- a small contribution of delayed gastric LP, Bristol-Myers Squibb Company, Johnson gliflozin concentrations are only tran- emptying, possibly related to the in- & Johnson, Lilly Pharmaceuticals, Novartis siently sufficiently high to inhibit SGLT1. creased GLP-1 levels. Pharmaceuticals Corporation, and Sanofi; The changes in gut peptide secretion Endogenous glucose production was served as a consultant for Amgen, Inc., Boeh- observed with canagliflozin treatment are suppressed to a similar extent with both ringer Ingelheim Pharmaceuticals, Inc., Isis Pharmaceuticals, Inc., Novo Nordisk Phar- generally consistent with expectations treatments after the meal, but EGP was fl maceuticals, Inc., Roche Pharmaceuticals, Sa- based on intestinal SGLT1 inhibition higher with canagli ozin than with pla- nofi, and Vivus; and served on advisory boards leading to delayed glucose absorption. cebo over the 3- to 6-h period after the for Amgen, Inc., Pharmaceuticals, GIP-secreting K cells are primarily found meal (Fig. 2). The elevated EGP is likely a Inc., Boehringer Ingelheim Pharmaceuticals, in the proximal small intestine, and the compensatory response that enables nor- Inc., Eli Lilly and Company, Gilead, Intarcia observed reduction in plasma GIP con- mal plasma glucose concentrations to be Therapeutics, Inc., Merck & Co., Novo Nor- centrations with canagliflozin treatment is maintained in the presence of sustained disk Pharmaceuticals, Inc., Roche Pharma- consistent with recent data showing that UGE, but the mechanism(s) leading to ceuticals, Sanofi, and Versartis, Inc. D.P., S.S., SGLT1-mediated glucose uptake is piv- the increased EGP is not known. A.G., N.V., K.F., and P.R. are full-time em- otal for GIP secretion (5). Similarly, the Although this study demonstrated ployees of Janssen Research & Development, LLC. S.M. has received research support from increase in plasma PYY and total GLP-1 that a single 300-mg dose of canagliflozin t Bristol-Myers Squibb Company, AstraZeneca observed after = 30 min suggests in- delays intestinal glucose absorption in Pharmaceuticals LP, Daiichi-Sankyo, Inc., and creased glucose absorption in the more healthy subjects, there are some impor- Novo Nordisk Pharmaceuticals, Inc.; served distal intestine where the GLP-1–secreting tant limitations. Because only one dose as a consultant for Bristol-Myers Squibb L cells are primarily found. However, it strength of canagliflozin was tested, the Company, AstraZeneca Pharmaceuticals LP, remains uncertain why similarly rapid in- dose-response relationship for the effect and Daiichi-Sankyo, Inc.; and served as a creases in GLP-1 and PYY were observed of canagliflozin on intestinal glucose ab- speaker for Boehringer Ingelheim Pharma- in the first 30 min in both treatment sorption cannot be established from this ceuticals, Inc. No other potential conflicts of groups. The early increments in GLP-1 study. Results from an earlier study sug- interest relevant to this article were reported. and PYY (occurring before ingested nu- gested that doses .200 mg of canagliflo- D.P., S.M., T.P.C., and R.R.H. contributed to the design and conduct of the study and to trients have reached the more distal intes- zin were required to see pronounced the acquisition, analysis, and interpretation of tine) have been commonly reported in reductions in postprandial glucose excur- data, and reviewed and approved the manu- other studies (25), and the source of this sions (7), but it is not known whether script. A.G. contributed to the design of the early GLP-1 and PYY secretion remains greater delays in intestinal glucose ab- study and to the analysis and interpretation of uncertain, with possibilities including sorption could be achieved with higher data and reviewed and approved the manu- 1) a neural signal from the proximal small doses of canagliflozin or whether malab- script. S.S., N.V., K.F., and P.R. contributed to intestine that reaches L cells in the more sorption would be observed with higher the design and conduct of the study and to the distal intestine (26), 2)releasefromthe doses. In addition, because this mecha- analysis and interpretation of data, and re- small number of L cells in the proximal nistic study only characterized the re- viewed and approved the manuscript. D.P. is small intestine (27), and/or 3) release sponsetoasingledoseinhealthy the guarantor of this work and, as such, had full access to all the data in the study and takes by a subset of enteroendocrine cells that subjects, further studies will be required responsibility for the integrity of the data and cosecrete GIP and GLP-1 (28). Because to characterize the effects of sustained fl the accuracy of the data analysis. SGLT1 is also reported to be essential canagli ozin treatment on intestinal glu- Parts of this study were presented in abstract for GLP-1 secretion (5) and no reduction cose absorption in subjects with type 2 form at the 72nd Scientific Sessions of the in early GLP-1 secretion was observed diabetes. American Diabetes Association, Philadelphia, with canagliflozin, it is not clear that the In conclusion, canagliflozin 300 mg PA, 8–12 June 2012, and at the 48th Annual early increment in GLP-1 secretion can be reduces postprandial plasma glucose and Meeting of the European Association for the explained by GLP-1 secretion from cells insulin concentration in healthy subjects Study of Diabetes, Berlin, Germany, 1–5Oc- in the proximal small intestine. The ob- by two distinct mechanisms: 1) increasing tober 2012. served reduction in serum insulin is likely UGE due to renal SGLT2 inhibition and due to decreased plasma glucose, because 2)delayingRaO, which is likely due to the relationship between plasma glucose transient intestinal SGLT1 inhibition. References 1. Lee YJ, Lee YJ, Han HJ. Regulatory and the insulin secretion rate was un- fl mechanisms of Na(+)/glucose cotrans- changed by canagli ozin treatment (data porters in renal proximal tubule cells. not shown). AcknowledgmentsdThis study was sup- Kidney Int Suppl 2007;Aug:S27–S35 Although a slight delay in gastric ported by Janssen Research & Development, 2. Drozdowski LA, Thomson AB. Intestinal emptying (estimated by plasma acetamin- LLC. Editorial support for the writing of this sugar transport. World J Gastroenterol ophen concentrations) was observed with manuscript was provided by Cherie Koch, 2006;12:1657–1670 canagliflozin treatment compared with PhD, of MedErgy, and was funded by Janssen 3. Wright EM, Loo DD, Hirayama BA. Bi- placebo, the reductions in R Oobserved Global Services, LLC. Canagliflozin is being ology of human sodium glucose trans- a – with canagliflozin are greater than can be developed by Janssen Research & De- porters. Physiol Rev 2011;91:733 794 accounted for by the modest delay in gas- velopment, LLC, in collaboration with Mitsu- 4. Abdul-Ghani MA, Norton L, Defronzo bishi Tanabe Pharma Corporation. RA. Role of sodium-glucose cotransporter tric emptying. The observed effects of fl R.R.H. has received grants or research sup- 2 (SGLT 2) inhibitors in the treatment of canagli ozin on RaO are consistent with port from the National Institutes of Health, type 2 diabetes. Endocr Rev 2011;32: the hypothesis that canagliflozin 300 mg National Institute of Diabetes and Digestive 515–531 transiently inhibits intestinal SGLT1- and Kidney Diseases, DVA, Amylin Pharma- 5. Gorboulev V, Schurmann€ A, Vallon V, mediated glucose absorption, with perhaps ceuticals, Inc., AstraZeneca Pharmaceuticals et al. Na(+)-D-glucose cotransporter SGLT1 care.diabetesjournals.org DIABETES CARE 7 Canagliflozin delays gut glucose absorption

is pivotal for intestinal glucose absorption (T2DM) (Abstract). Clin Pharmacol Ther glucose loading (Staub-Traugott effect): and glucose-dependent incretin secretion. 2012;91:S98 size and mechanisms. Am J Physiol En- Diabetes 2012;61:187–196 12. Nimmo WS, Heading RC, Wilson J, docrinol Metab 2009;297:E532–E537 6. Nomura S, Sakamaki S, Hongu M, et al. Tothill P, Prescott LF. Inhibition of gastric 21. Nair S, Wilding JP. Sodium glucose co- Discovery of canagliflozin, a novel emptying and drug absorption by narcotic transporter 2 inhibitors as a new treat- C-glucoside with thiophene ring, as analgesics. Br J Clin Pharmacol 1975;2: ment for diabetes mellitus. J Clin sodium-dependent glucose cotransporter 509–513 Endocrinol Metab 2010;95:34–42 2 inhibitor for the treatment of type 2 13. Medhus AW, Lofthus CM, Bredesen J, 22. Rave K, Nosek L, Posner J, Heise T, diabetes mellitus. J Med Chem 2010;53: Husebye E. Gastric emptying: the validity Roggen K, van Hoogdalem EJ. Renal glu- 6355–6360 of the paracetamol absorption test adjusted cose excretion as a function of blood 7. Sha S, Devineni D, Ghosh A, et al. for individual pharmacokinetics. Neuro- glucose concentration in subjects with Canagliflozin, a novel inhibitor of sodium gastroenterol Motil 2001;13:179–185 type 2 diabetesdresults of a hyper- glucose co-transporter 2, dose depen- 14. Mudaliar S, Mohideen P, Deutsch R, et al. glycaemic glucose clamp study. Nephrol dently reduces calculated renal threshold Intravenous glargine and regular insulin Dial Transplant 2006;21:2166–2171 for glucose excretion and increases uri- have similar effects on endogenous glu- 23. DeFronzo RA, Ferrannini E, Simonson nary glucose excretion in healthy sub- cose output and peripheral activation/ DC. Fasting hyperglycemia in non-insulin- jects. Diabetes Obes Metab 2011;13: deactivation kinetic profiles. Diabetes dependent diabetes mellitus: contributions 669–672 Care 2002;25:1597–1602 of excessive hepatic glucose production 8. Liang Y, Arakawa K, Ueta K, et al. Effect of 15. Kreisberg RA, Siegal AM, Owen WC. Al- and impaired tissue glucose uptake. Me- canagliflozin on renal threshold for glu- anine and gluconeogenesis in man: effect tabolism 1989;38:387–395 cose, glycemia, and body weight in nor- of ethanol. J Clin Endocrinol Metab 1972; 24. DeFronzo RA. Lilly lecture 1987. The mal and diabetic animal models. PLoS 34:876–883 triumvirate: beta-cell, muscle, liver. A ONE 2012;7:e30555 16. Ferrannini E, Bjorkman O, Reichard GA collusion responsible for NIDDM. Di- 9. Devineni D, Morrow L, Hompesch M, Jr, et al. The disposal of an oral glucose abetes 1988;37:667–687 et al. Canagliflozin improves glycaemic load in healthy subjects. A quantitative 25. De Silva A, Bloom SR. Gut hormones and control over 28 days in subjects with type study. Diabetes 1985;34:580–588 appetite control: a focus on PYY and GLP-1 2 diabetes not optimally controlled on 17. Kelley D, Mitrakou A, Marsh H, et al. as therapeutic targets in obesity. Gut Liver insulin. Diabetes Obes Metab 2012;14: Skeletal muscle glycolysis, oxidation, and 2012;6:10–20 539–545 storage of an oral glucose load. J Clin In- 26. Lim GE, Brubaker PL. Glucagon-like 10.RosenstockJ,AggarwalN,PolidoriD, vest 1988;81:1563–1571 peptide 1 secretion by the L-cell the view et al.; Canagliflozin DIA 2001 Study Group. 18. Mari A, Stojanovska L, Proietto J, from within. Diabetes 2006;55:S70–S77 Dose-ranging effects of canagliflozin, a Thorburn AW. A circulatory model for 27. Nauck MA, Siemsgluss€ J, Orskov C, Holst sodium-glucose cotransporter 2 inhibitor, calculating non-steady-state glucose fluxes. JJ. Release of glucagon-like peptide 1 as add-on to in subjects with Validationandcomparisonwithcompart- (GLP-1 [7-36 amide]), gastric inhibitory type 2 diabetes. Diabetes Care 2012;35: mental models. Comput Methods Programs polypeptide (GIP) and insulin in response 1232–1238 Biomed 2003;71:269–281 to oral glucose after upper and lower in- 11. Devineni D, Curtin CR, Polidori D, 19. Bandsma RH, Spoelstra MN, Mari A, et al. testinal resections. Z Gastroenterol 1996; Murphy J, Rusch S, Rothenberg PL. Can- Impaired glucose absorption in children 34:159–166 agliflozin (sodium glucose co-transporter with severe malnutrition. J Pediatr 2011; 28. Mortensen K, Christensen LL, Holst JJ, 2 [SGLT2] inhibitor): pharmacokinetics 158:282–287, e1 Orskov C. GLP-1 and GIP are colocalized (PK) and pharmacodynamics (PD) in 20. Bonuccelli S, Muscelli E, Gastaldelli A, in a subset of endocrine cells in the small subjects with type 2 diabetes mellitus et al. Improved tolerance to sequential intestine. Regul Pept 2003;114:189–196

8 DIABETES CARE care.diabetesjournals.org