Evaluation of Efficacy and Safety of the Glucagon Receptor Antagonist LY2409021 in Patients with Type 2 Diabetes

Diabetes Care Volume 39, July 2016 1241

Christof M. Kazda,1 Ying Ding,2 Evaluation of Efficacy and Safety Ronan P. Kelly,3 Parag Garhyan,2 Chunxue Shi,4 Chay Ngee Lim,2 Haoda Fu,2 of the Glucagon Receptor David E. Watson,2 Andrew J. Lewin,5* William H. Landschulz,2 Mark A. Deeg,2 Antagonist LY2409021 in Patients David E. Moller,2 and Thomas A. Hardy2 With Type 2 Diabetes: 12- and 24-Week Phase 2 Studies Diabetes Care 2016;39:1241–1249 | DOI: 10.2337/dc15-1643 MRIGTCNLGE N THERAPEUTICS AND TECHNOLOGIES EMERGING

OBJECTIVE Type 2 diabetes pathophysiology is characterized by dysregulated glucagon secretion. LY2409021, a potent, selective small-molecule glucagon receptor antagonist that lowers glucose was evaluated for efﬁcacy and safety in patients with type 2 diabetes.

RESEARCH DESIGN AND METHODS

The efficacy (HbA1c and glucose) and safety (serum aminotransferase) of once- daily oral administration of LY2409021 was assessed in two double-blind studies. 1Eli Lilly and Company, Suresnes, France 2 Phase 2a study patients were randomized to 10, 30, or 60 mg of LY2409021 or Eli Lilly and Company, Indianapolis, IN 3Lilly-NUS Centre for Clinical Pharmacology, placebo for 12 weeks. Phase 2b study patients were randomized to 2.5, 10, or Singapore 20 mg LY2409021 or placebo for 24 weeks. 4inVentiv Health Clinical, Clarksville, MD 5National Research Institute, Los Angeles, CA RESULTS Corresponding author: Christof M. Kazda, [email protected]. LY2409021 produced reductions in HbA1c that were significantly different from placebo over both 12 and 24 weeks. After 12 weeks, least squares (LS) mean Received 27 July 2015 and accepted 30 March 2016. change from baseline in HbA1c was –0.83% (10 mg), –0.65% (30 mg), and –0.66% (60 mg) (all P < 0.05) vs. placebo, 0.11%. After 24 weeks, LS mean change from Clinical trial reg. nos. NCT01241448 (phase 2a) and NCT00871572 (phase 2b), clinicaltrials.gov. baseline in HbA1c was –0.45% (2.5 mg), –0.78% (10 mg, P < 0.05), –0.92% (20 mg, P < – This article contains Supplementary Data online 0.05), and 0.15% with placebo. Increases in serum aminotransferase, fasting at http://care.diabetesjournals.org/lookup/ glucagon, and total fasting glucagon-like peptide-1 (GLP-1) were observed; levels suppl/doi:10.2337/dc15-1643/-/DC1. returned to baseline after drug washout. Fasting glucose was also lowered with Y.D. is currently affiliated with the Department of LY2409021 at doses associated with only modest increases in aminotransferases Biostatistics, University of Pittsburgh, Pitts- (mean increase in alanine aminotransferase [ALT] £10 units/L). The incidence burgh, PA, and C.N.L. is currently affiliated with the Department of Experimental and Clinical of hypoglycemia in the LY2409021 groups was not statistically different from Pharmacology, University of Minnesota Twin placebo. Cities, Minneapolis, MN. *Retired. CONCLUSIONS © 2016 by the American Diabetes Association. In patients with type 2 diabetes, glucagon receptor antagonist treatment signif- Readers may use this article as long as the work is icantly lowered HbA1c and glucose levels with good overall tolerability and a low properly cited, the use is educational and not for risk for hypoglycemia. Modest, reversible increases in serum aminotransferases profit, and the work is not altered. were observed. See accompanying article, p. 1075. 1242 Efficacy and Safety of LY2409021 Diabetes Care Volume 39, July 2016

Type 2 diabetes is characterized by active treatment period, and a 4-week meaningful efficacy with acceptable he- chronic hyperglycemia as a result of blinded posttreatment washout/follow- patic safety over the 24-week treatment impaired insulin secretion and action. up period (Supplementary Fig. 1A). Pa- period. Dysregulated glucagon secretion (1–3) tients were randomly assigned to one Study Participants and resultant hepatic glucose overpro- of the four treatment groups (Supple- Patients were adults (aged 18–70 years, duction are also characteristic patho- mentary Fig. 1A) in a 1.5:2.1:1:1 ratio inclusive) with diagnoses of type 2 dia- physiological features and contribute (60 mg:30 mg:10 mg:placebo). Unequal betes according to the World Health to hyperglycemia in type 2 diabetes allocation was used to maximize the ex- Organization diagnostic criteria (19) who (4–10). Previous studies with LY2409021 posure to 30 mg LY2409021. Stratifica- had been treated with diet and exercise have demonstrated that glucagon re- tion was based on the absence or alone or in combination with a stable ceptor antagonism inhibits glucose presence of preexisting metformin ther- dose of metformin ($1,000 mg/day) production in response to exogenous apy and on study site. The phase 2b study for at least 3 months before screening. hyperglucagonemia in healthy volun- was a multicenter, randomized, double- Patients had HbA values of 6.5–10% teers and improves hyperglycemia in blind, placebo-controlled, parallel-group 1c (48–86 mmol/mol) and BMI between patients with type 2 diabetes through study consisting of a 1-week single-blind 25 and 40 kg/m2, inclusive (phase 2a reductions in both fasting and postpran- placebo lead-in period, a 24-week ac- study), and 7.0–10.5% (53–91 mmol/mol) dial glucose (11,12). A 28-day phase 1 tive treatment period, and a 4-week blin- and a BMI between 25 and 45 kg/m2,in- study of LY2409021 in patients with ded posttreatment washout/follow-up clusive (phase 2b study). Patients agreed type 2 diabetes demonstrated improve- period (Supplementary Fig. 1B). to self-monitor blood glucose (SMBG) lev- ment in glycemic parameters at doses Patients were randomly assigned to els, complete study diaries, and maintain of 5–90 mg administered once daily one of the four treatment groups consistent dietary, physical activity, and (12). However, reversible, dose-related (Supplementary Fig. 1B) in a 1:1:1:1 ra- sleeping patterns throughout the study. increases in the levels of serum amino- tio (20 mg:10 mg:2.5 mg:placebo). Stable doses of antihypertensives or transferases were observed in that study Stratification was based on preexisting lipid-lowering medications were allowed. (12). Similar reversible increases in se- metformin therapy, baseline HbA 1c Clinical signs or symptoms of liver disease, rum aminotransferases have been noted level (, or $8.2% [66 mmol/mol]), previous diagnosis or serologic evidence with other glucagon receptor antago- and baseline alanine aminotransferase of hepatitis B or C infection, and use of nists, including a human glucagon recep- (ALT) levels (# or . the upper limit of any antihyperglycemic medication other tor monoclonal antibody (13), other normal [ULN]). Patients were discontin- than metformin within the 3-month pe- small molecules (14–17), and an anti- ued from study participation if they had riod before screening were exclusionary sense oligonucleotide targeting gluca- fasting glucose levels .270 mg/dL for both studies. Additional exclusion cri- gon receptor expression (18). onthreeormoreseparatedaysover teria at screening included aminotrans- In the present phase 2a and 2b studies, any 2-week period between randomiza- ferase levels greater than two times the we further examined the glucose-lowering tion and week 6 of treatment, fasting glu- ULN and fasting triglycerides .400 mg/dL efficacy and safety of LY2409021 by com- cose levels .240 mg/dL on three or more in the phase 2a study and aminotransfer- paring the relationships between doses separate days over any 2-week period be- ase levels .2.5 times the ULN, nonfasting of LY2409021 that lower HbA and glu- tween weeks 6 and 12, or fasting glucose 1c triglyceride levels .600 mg/dL, and use cose levels (efficacy) and those that in- levels .200 mg/dL or HbA levels .8.0% 1c of insulin or glucagon-like peptide-1 crease serum aminotransferase levels (64 mmol/mol) from week 12 to week 24 (GLP-1) agonist for .5dayswithinthe (hepatic safety). The phase 2b study re- of treatment. The criteria for the two stud- 3-month period before screening in the ported here represents the largest and ies were the same through 12 weeks of phase 2b study. longest clinical trial of a glucagon recep- treatment; the phase 2b study had more tor antagonist reported to date. stringent criteria after 12 weeks. Outcome Measures The doses of LY2409021 used in the The primary efficacy end point was RESEARCH DESIGN AND METHODS phase 2b study were selected on the mean change in HbA1c level from base- Study Design basis of the exposure response analysis line to end of the 12-week LY2409021 Both phase 2a (NCT00871572) and of a phase 1, 28-day, multiple-dose study treatment period (phase 2a study) and phase 2b (NCT01241448) studies were that indicated that doses of LY2409021 to the end of the 24-week treatment conducted in accordance with regula- ranging from 30 to 90 mg/day produced period (phase 2b study). Secondary tory standards of good clinical practice, substantial lowering of fasting glucose end points (in both studies) included the Declaration of Helsinki, and all appli- levels in patients with type 2 diabetes changes in fasting serum glucose, gluca- cable local regulations. Study protocols (12) and the current phase 2a study gon, and GLP-1 (total and active [7–36]) were approved by each site’s ethical re- that demonstrated reductions in fasting levels and fasting serum insulin levels. view board. All patients provided writ- glucose and HbA1c levels in patients with Seven-point SMBG profiles consisted of ten informed consent before initiation type 2 diabetes at all dose levels (10, 30, blood glucose values obtained before of study procedures. The phase 2a study and 60 mg). The doses of LY2409021 se- and 2 h after each meal and at bedtime was a multicenter, randomized, double- lected for the phase 2b study (2.5, 10, (phase 2a study) or at 0300 h (phase 2b blind, placebo-controlled, parallel-group and 20 mg once daily) were used to iden- study) on two separate days during the study with a 1-week single-blind placebo tify the minimal efficacious dose and max- 5-day period before the respective lead-in period, a 12-week double-blind imize the ability to demonstrate clinically study visits. For the phase 2a study, care.diabetesjournals.org Kazda and Associates 1243

fasting lipid profiles were collected at diastolic blood pressure (SBP and DBP) RESULTS initiation of treatment (week 0) and at were measured with an automated cuff Patient Disposition, Demographics, end point (week 12); fasting body while the patient was in a sitting position. and Baseline Characteristics weight and sitting blood pressure mea- In the phase 2a study, 87 patients were surements were collected before ran- Data Analyses randomized, 68 (78.2%) completed the domization and weekly throughout the For both phase 2a and 2b studies, effi- study, and 21.8% of patients prematurely treatment period. For the phase 2b cacy analyses were conducted on the discontinued (Supplementary Fig. 2A). study, fasting lipid profiles were col- modified intent-to-treat population, de- The most commonly reported reason for lected at baseline and weeks 8, 12, 16, fined as all randomized patients with at study discontinuation was subject deci- and 24; fasting body weight and sitting least one postbaseline measurement ac- sion (8.0%). Discontinuations due to ad- blood pressure measurements were cording to the treatment the patients verse events (five patients, 5.9%) were collected before randomization and bi- were assigned. Mixed-effects model re- driven by abnormalities in laboratory val- weekly throughout the treatment period. peated measures (MMRM) analysis was ues (two events of liver aminotransferase Hepatic safety assessment included used for the primary analysis of HbA1c level increases and three events of serum weekly (phase2a study) or biweekly(phase levels for both studies. The least squares amylase and/or lipase level increases) 2b study) monitoring of ALT; aspartate (LS) mean differences, 90% CIs, and P without any clinical symptoms. All discon- aminotransferase (AST); g-glutamyl trans- values reported were based on a model tinuations due to adverse events oc- ferase (GGT); alkaline phosphatase (ALK); that included baseline HbA1c as a covari- curred in patients receiving LY2409021. and total, direct, and indirect bilirubin ate and metformin use, visit, treatment, In the phase 2b study, 254 patients levels. and visit-by-treatment interaction as were randomized and 151 (59%) com- fixed effects. Change in HbA level from 1c pleted the study: 29 in the placebo group, baseline to end point was also analyzed Analytical Methods 36 in the 2.5-mg group, 43 in the 10-mg Blood samples for glucagon and GLP-1 using ANCOVA with baseline as a cova- group, and 43 in the 20-mg group analysis were collected in prechilled riate and treatment and metformin use (Supplementary Fig. 2B). Approximately tubes containing EDTA. After collection, as fixed effects. There were no adjust- 40% of patients (103 of 254) discontinued aprotinin/Trasylol (Bayer) (glucagon ments of multiplicity for any analyses early. The most common reasons for early samples) and dipeptidyl peptidase-4 in- unless otherwise stated. Missing end discontinuation were loss of glycemic hibitor (GLP-1 samples) additives were points were imputed using last obser- control based on protocol discontinua- added as a preservative. For the phase vation carried forward. For the phase tion criteria (49 [19%]), protocol violation 2a study, plasma glucagon concentra- 2b study, categorical analyses on HbA 1c (24 [9%]), and subject decision (18 [7%]). tions were measured by radioimmuno- were also performed to compare the assay by ALPCO Diagnostics (Salem, NH). proportions of patients who achieved Most of the per-protocol discontinuations due to inadequate glycemic control For the phase 2b study, plasma glucagon HbA1c values #6.5% (48 mmol/mol) concentrations were measured using an and ,7.0% (53 mmol/mol) at week occurred at week 14 when hyperglycemia electrochemiluminescence sandwich 24 by treatment group. A Cochran- discontinuation criteria became more fi immunoassay (20). Both assays have Mantel-Haenszel test was used and stringent. Speci cally, eleven placebo patients (17%), eight 2.5-mg patients ,1% cross-reactivity to oxyntomodulin stratified by baseline HbA1c values or gut glucagon-like immunoreactivity. (,8.2%, $8.2% [66 mmol/mol]). For (13%), seven 10-mg patients (11%), LY2409021 does not cross-react with im- bothphase2aand2bstudies,analyses and four 20-mg patients (6%) discon- munoassays where glucagon and GLP-1 of secondary efficacy measures with tinued the study at week 14 because are measured. LY2409021 binds only to continuous data were performed using of protocol discontinuation criteria re- the receptor, not to ligands of the re- similar statistical models. Fisher exact lated to inadequate glycemic control. ceptor or related receptors. For both test also compared each treatment The mean and median days of treat- phase 2a and 2b studies, total and active group with the placebo group for cat- ment were 130 and 167, respectively. (7–36) plasma GLP-1 levels were mea- egorical outcomes. An ANCOVA was Baseline characteristics for phase 2a sured using an ELISA (Millipore Corp., performed on the change from base- study participants (Supplementary Table 1) St. Charles, MO) and electrochemi- line aminotransferase values at each showed that HbA1c values (mean, % luminescence sandwich immunoassay visit up to the end of the posttreat- [mmol/mol]) at week 0 for the LY2409021 (Meso Scale Discovery, Rockville, MD), ment washout period and analyzed treatment groups were highest in the respectively; fasting serum glucose lev- using an MMRM model. The model 10-mg group (8.0% [64 mmol/mol]) and els were measured using a commercially consisted of the baseline value as a were 7.5% (58 mmol/mol) for the 30-mg validated method; serum insulin levels covariate and visit, treatment, and group, 7.6% (60 mmol/mol) for the 60-mg were measured using direct chemilumi- visit-by-treatment interaction as fixed group, and 7.9% (62 mmol/mol) for the nescent immunoassays (ADVIA Centaur; effects. Treatment comparisons were placebo group. The most common con- Siemens, New York, NY); patient 7-point reported as the treatment LS mean, comitant medication was metformin, SMBG levels were measured using a cali- 95% CI, and P value. Safety analyses used by 58.6% of patients. brated glucose meter; total cholesterol, were performed on the safety popula- Baseline characteristics for phase 2b HDL cholesterol, and triglyceride levels tion, which was defined as all random- study participants (Supplementary Table 2) were determined directly and LDL choles- ized patients who received at least one showed overall similarity in treatment terol was calculated and systolic and dose of the assigned study drug. group demographics. The mean HbA1c 1244 Efficacy and Safety of LY2409021 Diabetes Care Volume 39, July 2016

level was 8.0% (64 mmol/mol) at study compared with placebo (Fig. 1C). The 30- LY2409021 treatment also led to entry. Concomitant metformin use was and 60-mg dose groups showed reduc- dose-related higher levels of fasting to- similar among the treatment groups tions (P , 0.05) in fasting glucose levels tal GLP-1 than placebo did; fasting total (ranging from 85.7 to 87.5%). that were sustained throughout the ac- GLP-1 levels followed a pattern similar tive treatment period. The LS mean to that of fasting glucagon levels in both Glucose-Related Efficacy change from baseline in fasting glucose studies (Fig. 3C and D). Total fasting Measurements levels for the phase 2b study is shown GLP-1 levels returned to baseline (Fig. In the phase 2a study, after 12 weeks of in Fig. 1D. LY2409021 produced dose- 3C) during the posttreatment washout treatment, LS mean (90% CI) change dependent, statistically significantly period in the phase 2a study (GLP-1 lev- from baseline for HbA1c level in the greater changes from baseline in fasting els were not measured during washout three LY2409021 dose groups was sig- glucose levels than placebo; maximal in the phase 2b study). There were no nificantly greater than that in the pla- decreases occurred within 2 weeks of statistically significant differences in cebo group: –0.83% (–1.28, –0.38) dosing. A small increase in fasting glu- fasting active GLP-1 levels between (–9.1 mmol/mol) for the 10-mg group, cose level was observed from week 12 LY2409021 treatment at any dose and –0.65% (–0.93, –0.37) (–7.1 mmol/mol) totheendoftheactivetreatmentpe- placebo for either study. Fasting insulin for the 30-mg group, and –0.66% (–1.00, riod (week 24) in the 10- and 20-mg values were not significantly different –0.31) (–7.2 mmol/mol) for the 60-mg dose groups (Fig. 1D). between LY2409021 treatment and pla- group (P =0.03,P = 0.04, and P =0.05, In the phase 2a study, the means of cebo (data not shown). respectively). The change from baseline 7-point SMBG levels (Fig. 2A) decreased for the placebo group was 0.11% (–0.44, from baseline in all three LY2409021 Safety and Tolerability 0.65) (1.2 mmol/mol). The LS mean treatment groups at all pre- and post- Dose-related, reversible increases in se- changes from baseline in HbA1c values meal times and at all visits but generally rum ALT levels were observed during for all LY2409021 dose groups were sig- increased from baseline in the placebo active treatment with LY2409021 (Fig. 4). nificantly greater compared with those group. In the phase 2b study, at baseline In the phase 2a study, 4 of 85 patients had of the placebo group from week 6 (week 0), all treatment groups had sim- ALT levels three or more times the ULN: 1 through week 12 (Fig. 1A). In the phase ilar fasting, preprandial, postprandial, patient in the 30-mg group; 2 patients in 2b study, after 24 weeks of treatment, and 0300 h mean blood glucose values the 60-mg group; and 1 patient in the 10 and 20 mg LY2409021 produced sta- (Fig. 2B). By study end point (week 24), 10-mg group, whose ALT level reached tistically significantly lower (P , 0.001) blood glucose levels were lower for the five or more times the ULN. There was HbA1c values than placebo (Fig. 1B). The 10- and 20-mg LY2409021 doses than no dose-dependent change from baseline LS mean (90% CI) change from baseline for the placebo group across the time in GGT (data not shown). There were no for HbA1c value was –0.92% (–1.12, points (Fig. 2B); the 20-mg group clinically significant changes in body –0.73) (–10.1 mmol/mol) at 20 mg, showed significantly lower glucose weight, lipid levels, blood pressure, heart –0.78% (–0.97, –0.58) (–8.5 mmol/mol) levels than the placebo group at most rate, or electrocardiogram measurements at 10 mg, –0.45% (–0.65, –0.25) preprandial and postprandial time observed in the phase 2a study. (–4.9 mmol/mol) at 2.5 mg, and points. In the phase 2b study, no placebo sub- –0.15% (–0.37, 0.06) (–1.6 mmol/mol) for jects and 8 of 191 patients receiving the placebo group. A small increase in Non–Glucose-Related LY2409021 had ALT levels three or more HbA1c level was seen from week 12 to Pharmacodynamic Measurements times the ULN: 2 patients in the 2.5-mg the end of the active treatment period In both studies, dose-dependent in- group, 4 patients in the 10-mg group, and (week 24) in the 10- and 20-mg groups creases in fasting glucagon levels were 2 patients in the 20-mg group. One pa- (Fig. 1B). MMRM was used for the primary observed with LY2409021 treatment tient in the 20-mg group had an ALT level analysis of HbA1c levels for both studies. and appeared to plateau after ;4 weeks five or more times the ULN. No subject The ANCOVA (last observation carried for- of treatment (Fig. 3). Mean changes with elevated ALT exhibited concomitant ward) analysis demonstrated a greater from baseline in fasting glucagon levels elevation of bilirubin in either study, and magnitude of reduction in HbA1c level (Fig. 3A, phase 2a study) were largest in no clinical signs or symptoms of liver in- across treatment groups (not shown) the 60-mg treatment group (4.5-fold injury were reported. In all patients, the ALT than the MMRM analysis. Nonetheless, crease at week 4). An apparent increase levels returned to baseline either with the results for the MMRM and ANCOVA in fasting glucagon levels in the placebo continued treatment or after cessation analyses were generally similar. group at weeks 6–10 was in part the re- of treatment (Fig. 4). Changes in AST lev- In the phase 2b study, one-third of the sult of a postanalytical error in a single els showed a similar dose-related trend as 20-mg treatment group achieved HbA1c patient’s samples. In the phase 2b study changes in ALT levels but were less pro- levels #6.5% (48 mmol/mol; P , 0.001), (Fig. 3B), the LS mean changes from nounced (data not shown). In the phase and approximately half achieved HbA1c baseline in fasting glucagon levels were 2b study, GGT values increased beginning levels ,7% (53 mmol/mol; P = 0.004) largest in the 20-mg group, with a 4.5- at week 2 and continued to rise through compared with the placebo (Supplemental fold increase at week 4 similar to that in week 24 in the 20-mg group. During the Fig. 3B). the phase 2a study. Fasting glucagon 4-week posttreatment washout period, In the phase 2a study, LY2409021 pro- levels returned to baseline in all treat- the GGT values of the 20-mg group began duced statistically and clinically signifi- ment groups by the end of the post- to reverse but were still above baseline cant reductions in fasting glucose levels treatment washout period (Fig. 3). values and statistically significantly care.diabetesjournals.org Kazda and Associates 1245

Figure 1—Time course for LS mean (90% CI) change from baseline in HbA1c level by week and treatment with LY2409021 or placebo over the 12-week phase 2a study (A) and 24-week phase 2b study (B) treatment periods. Time course for LS mean (95% CI) change from baseline in fasting glucose level by week and treatment with LY2409021 or placebo over the 12-week phase 2a study (C) and 24-week phase 2b study (D) treatment periods. *P , 0.001 and †P , 0.05, compared with placebo. **P = 0.05, compared with placebo. greater than those for the placebo group 4.9 mmHg; P =0.038).TheLSmean cellulitis, which was considered not re- (data not shown). There were no dose- change from baseline in DBP for all three lated to study drug, and one patient in dependent changes from baseline in ALK LY2409021 cohorts was ,1mmHgat the 10-mg group had elevated ALT, AST, or in total, direct, or indirect bilirubin in week 24. Changes in body weight after and GGT levels. The investigator consid- either study (data not shown). 24 weeks of treatment were as follows: ered the elevations related to study In the phase 2b study, there were no placebo, –1.07 kg; 2.5 mg, –0.33 kg; drug; however, the subject had unde- statistically significant differences in 10 mg, 0.55 kg; and 20 mg, 0.07 kg. tectable levels of LY2409021 at the plasma lipid levels between any of the Only the change for the 10-mg dose time of the event. Given the very long LY2409021 treatment groups and the was statistically significantly different half-life of the study drug (;60 h), it ap- placebo group. In 8 of 12 biweekly visits from the change for placebo (P = 0.033). pears that the subject had not been re- in the phase 2b study, there were no cently compliant with treatment. Both statistically significant differences in LS Adverse Events SAEs were of moderate intensity and mean change from baseline in SBP or There were no severe treatment-emergent were resolved. DBP between any of the LY2409021 adverse events (TEAEs) or deaths report- In the phase 2b study, the percent- treatment groups compared with the ed in either study. The percentages of ages of patients who experienced any placebo group. On four visits, the LS patients who experienced any TEAE in TEAE (Supplementary Table 4) were mean change from baseline for SBP the phase 2a study were similar across slightly higher in the LY2409021 groups and DBP was significantly greater for treatment groups (Supplementary Table than in the placebo group. The most one or more LY2409021 dose groups 3). Nausea and hypoglycemia (4.7% of common TEAEs of all causes were ALT compared with the placebo group. After patients for each) were the most fre- and AST level increases (6.7 and 5.1%, 24 weeks of treatment, only the 2.5-mg quent TEAEs, but there was no dose de- respectively) and headache (5.5%). A to- group showed a statistically significant pendency and no severe hypoglycemic tal of seven patients reported SAEs: two difference in LS mean change from base- events observed in any group. Two seri- patients in the 10-mg group (AST level line to end point in SBP relative to the ous adverse events (SAEs) were report- increased and atrial fibrillation), two placebo group (LS mean difference, ed: one subject in the 60-mg group had patients in the 20-mg group (ALT level 1246 Efficacy and Safety of LY2409021 Diabetes Care Volume 39, July 2016

Figure 2—LS mean (95% CI) change from baseline in SMBG level by time point and treatment with LY2409021 or placebo at end point week 12 (phase 2a study) (A) and week 24 (phase 2b study) (B). The SMBG level at each visit consisted of blood glucose values collected before and 2 h after each of the three main meals and at bedtime (phase 2a study) or at 0300 h (phase 2b study) as described in RESEARCH DESIGN AND METHODS.*P , 0.05, compared with placebo at indicated time points (A).*P , 0.001 and †P , 0.05, compared with placebo at indicated time points (B).

increased and AST level increased), and glucagon receptor antagonist published the current phase 2b study would pro- three placebo-treated patients (ankle to date. We have shown that once-daily duce ;67% blockade of the gluca- fracture [one patient], cholecystectomy treatment with LY2409021 at doses of gon receptor at steady state, and [one patient], and gastritis and iron 10–60 mg for up to 24 weeks produced maximal reductions in fasting glucose deficiency anemia [one patient]). The statistically significant and clinically rel- levels were consistently produced only three SAEs related to increased ALT evant reductions in HbA1c levels and at doses of 60 mg and higher in a pre- and/or AST levels were all asymptom- secondary efficacy measures in patients vious study (12). Finally, the efficacy ob- atic. The incidence of hypoglycemia with type 2 diabetes without signifi- served at the end of 24 weeks of (21) was low and not statistically differ- cantly increasing the risk of hypogly- treatment in the phase 2b study was at- ent across drug treatment and placebo cemia. The improvements in glycemic tenuated by the slight rise in HbA1c and groups; 17 patients (7%) reported a total parameters in the present studies are fasting glucose levels after week 12 in of 20 events (placebo, 2 patients; generally consistent with the findings the 10- and 20-mg groups. Whether 2.5 mg, 4 patients; 10 mg, 6 patients; of our previous studies with LY2409021 this reflects a waning of efficacy, an arti- 20 mg, 5 patients). None of the cases (12), including reduction in glycemia at fact of the higher discontinuation rate was severe, and all cases were self- all time points during SMBG testing. during this phase of the study, or some- treated with an oral glucose source. This suggests that glucagon signaling thing else is unclear. contributes to hyperglycemia in type 2 LY2409021 produced dose-related in- CONCLUSIONS diabetes in both the fasting and post- creases in total GLP-1 levels that re- The role of insulin-glucagon imbalance prandial states, as has been reported turned to baseline over the course of continues to be of interest in under- by others (9,23). The reductions in fast- the 4-week posttreatment washout standing the pathophysiology of type 2 ing glucose and HbA1c in the current period. Consistent with our previous diabetes. Clinical data to support the studies are less pronounced than those findings (12), levels of fasting active potential utility of glucagon receptor an- that have been reported with some GLP-1 did not change significantly with tagonism in the treatment of diabetes, other glucagon receptor antagonists LY2409021 treatment. Both glucagon however, have lagged (10,22). Over 10 (14,18). This may be due, in part, to and GLP-1 are derived from the proglu- years has elapsed between publication thefactthatbaselineHbA1c values cagon precursor, and the parallel in- of the first clinical study of a glucagon were higher in those studies than in creases in total GLP-1 and glucagon receptor antagonist (Bay 27-9955) (22) the current report. Glucose-lowering ef- suggest the possibility that both gluca- and the more recent studies of glucagon ficacy in the current studies is also likely gon and GLP-1 are secreted from pancre- receptor antagonists LY2409021 (11,12), to have been reduced, in part, by our atic a-cells as a result of an endocrine LY2786890 (13), MK-0893 (14,15), PF- efforts to find doses that are not associ- feedback loop upon glucagon receptor 06691874 (16), LGD-6972 (17), and ISIS- ated with aminotransferase elevations. blockade (24). Support for such a sys- GCGRRX (18). Pharmacokinetic-pharmacodynamic temic feedback loop comes from results The current report details findings modeling using data from a study of showing that liver-selective “knockout” from two studies that together compose LY2409021-mediated antagonism of of the glucagon receptor in mice can pro- the largest and longest experience eval- exogenous hyperglucagonemia (11) mote hyperglucagonemia (25). How- uating the safety and efficacy of a suggests that the 20-mg dose used in ever, the acute nature of the increase care.diabetesjournals.org Kazda and Associates 1247

Figure 3—A: Time course for mean (6SE) change from baseline in fasting glucagon level (pmol/L) by week and treatment with LY2409021 or placebo over the 12-week phase 2a study treatment period. B: Time course for LS mean (95% CI) change from baseline in fasting glucagon level (pmol/L) by week and treatment with LY2409021 or placebo over the 24-week phase 2b study treatment period. C: Time course for mean (6SE) change from baseline in fasting total GLP-1 level (pmol/L) by week and treatment with LY2409021 or placebo over the 12-week phase 2a study treatment period. D: Time course for LS mean (95% CI) change from baseline in fasting total GLP-1 level (pmol/L) by week and treatment with LY2409021 or placebo over the 24-week phase 2b study treatment period. *P , 0.05, compared with placebo at indicated time points (A). *P , 0.001 and †P , 0.05, compared with placebo at indicated time points (B and D). No statistical analysis was conducted for fasting total GLP-1 level in C. in circulating glucagon also suggests heterozygous knockout of the glucagon dependent, but some statistically signif- that partial blockade of the glucagon receptor gene have very modest in- icant differences from baseline were ob- receptor is most likely driven by a direct creases in glucagon and do not develop served in the larger phase 2b study. systemic feedback loop but might also a-cell hyperplasia (29). In the present These parameters will continue to be be impacted by decreased overall glu- phase 2a and 2b studies, fasting gluca- of key interest in future studies. A study cagon clearance. Studies in mice have gon was increased in a dose-dependent to investigate the effect of LY2409021 provided conflicting results regarding manner; maximum mean increases were on 24-h blood pressure profiles as mea- the role of GLP-1 in the improved glu- ;4.5-fold. The changes in glucagon sured by ambulatory blood pressure cose metabolism observed in glucagon levels seen with LY2409021 probably measurement is under way (31). receptor knockouts (26,27). Our results reflect a direct pharmacologic effect of It is our view that hepatic safety re- suggest that GLP-1 is unlikely to con- glucagon receptor antagonism rather mains the most important safety ques- tribute to the efficacy of LY2409021 than an effect on a-cell number: the ef- tion facing the development of glucagon because active hormone levels are fect occurred early, was nonprogressive receptor antagonists for the treatment unchanged. during treatment, and reversed com- of type 2 diabetes. A key objective for a-Cell hyperplasia has been raised pletely during posttreatment washout. both studies reported here was to eval- as a potential concern with pharmaco- Body weight, blood pressure, and uate the therapeutic margin between logic antagonism of glucagon action lipid increases have been reported for doses of LY2409021 that lower glucose (28). Mice with homozygous disruption other glucagon receptor antagonists levels and those that increase serum of the glucagon receptor gene develop (14,15). In the current studies, there aminotransferase levels. We were un- dramatic elevations in plasma gluca- were no significant changes in plasma able to identify an efficacious dose of gon levels (.1,000-fold), hyperplasia lipid levels with LY2409021 treatment. LY2409021 that was not associated of a-cells (29), and even neuroendo- Changes in body weight and blood pres- with an increase in mean serum amino- crine tumors (30). However, mice with sure were generally small and not dose transferase levels. However, changes in 1248 Efficacy and Safety of LY2409021 Diabetes Care Volume 39, July 2016

Figure 4—A: Time course for mean (6SE) change from baseline in ALT level (units/L) by week and treatment with LY2409021 or placebo over the 12- week phase 2a study treatment period. The ULNs for ALT level were 43 units/L (dashed line, male) and 34 units/L (dotted line, female). B:Timecourse for LS mean change (95% CI) from baseline in ALT level (units/L) by week and treatment with LY2409021 or placebo over the 24-week phase 2b study treatment period. Baseline ALT level = 32 units/L. *P , 0.05 and **P , 0.10, compared with placebo at indicated time points (A). *P , 0.05, compared with placebo at indicated time points (B).

ALT and AST levels were nonprogressive development of LY2409021 and possibly of the manuscript. C.S., C.N.L., and H.F. contrib- over time, reversible with treatment dis- other glucagon receptor antagonists. uted to drafting the manuscript and critical continuation, modest in magnitude, and revisions. For the phase 2b study, C.M.K., R.P.K., P.G., D.E.W., A.J.L., D.E.M., and T.A.H. contributed not associated with other signs or symp- to the study concept and design, analysis and toms of hepatic dysfunction. Changes in interpretation of data, drafting the manuscript, Publisher’sNote.An early version of this serum aminotransferases observed with article had been inadvertently accepted for and critical revisions. Y.D. contributed to the study LY2409021 are probably related directly publication and published online before the analysis, interpretation of data, drafting the man- revision process was completed. The American uscript, and critical revisions. The authors certify to antagonism of glucagon action; simi- thatthismanuscriptrepresentsvalidworkandthat lar, reversible increases in aminotrans- Diabetes Association, the publisher of Diabetes Care, apologizes for any confusion this error this manuscript has not been published and is not ferase levels have been seen in clinical may have caused. being considered for publication elsewhere. All studies with other small-molecule glu- authors contributed to the writing and review fi cagon receptor antagonists (14–17), process and approved the nalmanuscript. C.M.K. is the guarantor of this work and, as such, had full Acknowledgments. The authors thank the with the human glucagon receptor mono- access to all the data in the study and takes principal investigators and their clinical staff as clonal antibody LY2786890 (13), and with responsibility for the integrity of the data and well as the many study participants who gener- the accuracy of the data analysis. an antisense oligonucleotide targeting ously agreed to participate in these clinical trials. Prior Presentation. The phase 2a study was glucagon receptor gene expression (18). The authors also thank the clinical operations presented in part in poster form at the 72nd The cellular mechanisms by which these staff for their excellent trial implementation and Scientific Sessions of the American Diabetes support. The authors thank Lakechie Turnipseed changes occur remain unknown. Com- Association, Philadelphia, PA, 8–12 June 2012, (Eli Lilly and Company) for contributions to study plete abrogation of glucagon signaling and at the 48th Annual Meeting of the European design, implementation, and management as Association for the Study of Diabetes, Berlin, has been reported to increase susceptibil- the clinical trial manager for the phase 2a study. Germany, 1–5 October 2012. The phase 2b ity of mice to experimental liver injury The authors also thank Dr. Robert Panek (INC study was presented in part in poster form at (32,33). Hepatic steatosis has also been Research, Raleigh, NC) who provided medical the 73rd Scientific Sessions of the American Di- writing assistance. suggested as a possible consequence abetes Association, Chicago, IL, 21–25 June Duality of Interest. This study was sponsored of blocking glucagon action (33). How- 2013. by Eli Lilly and Company and/or one of its ever, results with pharmacologic or subsidiaries. C.M.K., R.P.K., P.G., H.F., D.E.W., genetic disruption of glucagon signal- W.H.L., M.A.D., D.E.M., and T.A.H. are em- ing in rodents have been contradictory ployees and stockholders of Eli Lilly and Com- References on this point (32,33,35–37). In light of pany and/or one of its subsidiaries. Y.D. and 1. Woerle HJ, Szoke E, Meyer C, et al. Mecha- these unresolved questions, we have C.N.L. were employees of Eli Lilly and Company nisms for abnormal postprandial glucose me- and/or one of its subsidiaries. A.J.L. was an tabolism in type 2 diabetes. Am J Physiol initiated a 12-month placebo- and active employee of the National Research Institute, Endocrinol Metab 2006;290:E67–E77 comparator–controlled hepatic safety Los Angeles, CA, at the time of this work and is 2. Sloop KW, Michael MD, Moyers JS. Glucagon study of LY2409021 that includes mag- currently retired. No other potential conflicts of as a target for the treatment of type 2 diabetes. netic resonance imaging to assess poten- interest relevant to this article were reported. Expert Opin Ther Targets 2005;9:593–600 Author Contributions. For the phase 2a study, 3. Unger RH, Cherrington AD. Glucagonocentric tial changes in hepatic fat fraction (34). C.M.K., R.P.K., P.G., D.E.W., W.H.L., and M.A.D. restructuring of diabetes: a pathophysiologic The results of this study could have sig- contributed to the study concept and design, and therapeutic makeover. J Clin Invest 2012; nificant implications for the future analysis and interpretation of data, and drafting 122:4–12 care.diabetesjournals.org Kazda and Associates 1249

4. Muller¨ WA, Faloona GR, Aguilar-Parada E, 16. Kazierad DJ, Bergman A, Tan B, Somayaji V, 27. Gu W, Winters KA, Motani AS, et al. Gluca- Unger RH. Abnormal alpha-cell function in di- Lee DS, Rolph T. Pharmacokinetics (PK) and gon receptor antagonist-mediated improve- abetes. Response to carbohydrate and protein pharmacodynamics (PD) of PF-06291874 (PF), ments in glycemic control are dependent on ingestion. N Engl J Med 1970;283:109–115 a glucagon receptor antagonist, in subjects functional pancreatic GLP-1 receptor. Am J 5. Hollander PM, Asplin CM, Palmer JP. Glucose with T2DM (Abstract 1202-P). Diabetes 2015; Physiol Endocrinol Metab 2010;299:E624–E632 modulation of insulin and glucagon secretion in 64(Suppl. 1):A310–A311 28. Lefebvre` PJ, Paquot N, Scheen AJ. Inhibiting nondiabetic and diabetic man. Diabetes 1982; 17. Vajda EG, Logan D, Lasseter K, et al. Phar- or antagonizing glucagon: making progress in 31:489–495 macokinetics and pharmacodynamics of the diabetes care. Diabetes Obes Metab 2015;17: 6. Ward WK, Best JD, Halter JB, Porte D Jr. Pro- glucagon receptor antagonist LGD-6972 in a 720–725 longed infusion of somatostatin with glucagon multi-dose clinical trial (Abstract 1193-P). Dia- 29. Gelling RW, Du XQ, Dichmann DS, et al. replacement increases plasma glucose and glu- betes 2015;64(Suppl. 1):A308 Lower blood glucose, hyperglucagonemia, and cose turnover in man. J Clin Endocrinol Metab 18. Morgan E, Smith A, Watts L, et al. ISIS-GCGRRX, pancreatic alpha cell hyperplasia in glucagon re- 1984;58:449–453 an antisense glucagon receptor antagonist, caused ceptor knockout mice. Proc Natl Acad Sci U S A 7. Dunning BE, Gerich JE. The role of alpha-cell rapid, robust, and sustained improvements in gly- 2003;100:1438–1443 dysregulation in fasting and postprandial hyper- cemic control without changes in BW, BP, lipids, or 30. Yu R, Dhall D, Nissen NN, Zhou C, Ren SG. glycemia in type 2 diabetes and therapeutic im- hypoglycemia in T2DM patients on stable metfor- Pancreatic neuroendocrine tumors in glucagon plications. Endocr Rev 2007;28:253–283 min therapy (Abstract 109-LB). Diabetes 2014; receptor-deficient mice. PLoS One 2011;6: 8. Gastaldelli A, Baldi S, Pettiti M, et al. Influence 63(Suppl. 1A):LB28 e23397 of obesity and type 2 diabetes on gluconeogenesis 19. Alberti KG, Zimmet PZ. Definition, diagnosis 31. Eli Lilly and Company. A study of LY2409021 and glucose output in humans: a quantitative and classification of diabetes mellitus and its on blood pressure and pulse rate in participants study. Diabetes 2000;49:1367–1373 complications. Part 1: diagnosis and classification with type 2 diabetes mellitus. In: ClinicalTrials.gov 9. Shah P, Vella A, Basu A, Basu R, Schwenk WF, of diabetes mellitus provisional report of a WHO [Internet]. Bethesda, MD, National Library of Med- Rizza RA. Lack of suppression of glucagon con- consultation. Diabet Med 1998;15:539–553 icine, 2014. Available from http://clinicaltrials.gov/ tributes to postprandial hyperglycemia in sub- 20. Sloan JH, Siegel RW, Ivanova-Cox YT, show/NCT02091362 NLM Identifier NCT02091362. jects with type 2 diabetes mellitus. J Clin Watson DE, Deeg MA, Konrad RJ. A novel high- Accessed 17 March 2016 Endocrinol Metab 2000;85:4053–4059 sensitivity electrochemiluminescence (ECL) 32. Sinclair EM, Yusta B, Streutker C, et al. Glu- 10. Bagger JI, Knop FK, Holst JJ, Vilsbøll T. Glu- sandwich immunoassay for the specific quanti- cagon receptor signaling is essential for control cagon antagonism as a potential therapeutic tative measurement of plasma glucagon. Clin of murine hepatocyte survival. Gastroenterol- target in type 2 diabetes. Diabetes Obes Metab Biochem 2012;45:1640–1644 ogy 2008;135:2096–2106 2011;13:965–971 21. Expert Committee on the Diagnosis and 33. Conarello SL, Jiang G, Mu J, et al. Glucagon 11. Tham LS, Abu-Raddad EJ, Lim CN. The glu- Classification of Diabetes Mellitus. Report of receptor knockout mice are resistant to diet- cagon receptor antagonist LY2409021 attenu- the expert committee on the diagnosis and clas- induced obesity and streptozotocin-mediated ates increases in hepatic glucose output (HGO) sification of diabetes mellitus. Diabetes Care beta cell loss and hyperglycaemia. Diabetologia and blood glucose during hyperglucagonemia in 2003;26(Suppl. 1):S5–S20 2007;50:142–150 healthy male patients (Abstract 416-PP). Diabe- 22. Petersen KF, Sullivan JT. Effects of a novel 34. Eli Lilly and Company. A phase 2, double- tes 2011;60(Suppl. 1):A115 glucagon receptor antagonist (Bay 27-9955) on blind, placebo-controlled trial to evaluate the 12. Kelly RP, Garhyan P, Raddad E, et al. Short- glucagon-stimulated glucose production in hu- safety and efficacy of LY2409021 compared to term administration of the glucagon receptor mans. Diabetologia 2001;44:2018–2024 sitagliptin in subjects with type 2 diabetes mel- antagonist LY2409021 lowers blood glucose in 23. Baron AD, Schaeffer L, Shragg P, Kolterman litus. In: ClinicalTrials.gov [Internet]. Bethesda, healthy people and in those with type 2 diabe- OG. Role of hyperglucagonemia in mainte- MD, National Library of Medicine, 2014. Available tes. Diabetes Obes Metab 2015;17:414–422 nance of increased rates of hepatic glucose out- from http://clinicaltrials.gov/show/NCT02111096 13. Kelly RP, Garhyan P, Reynolds VL, et al. put in type II diabetics. Diabetes 1987;36: NLM Identifier NCT00004451. Accessed 12 January Glucagon receptor antibody LY2786890 re- 274–283 2016 duced glucose levels in type 2 diabetes mellitus 24. Whalley NM, Pritchard LE, Smith DM, White 35. Liang Y, Osborne MC, Monia BP, et al. Re- patients (Abstract 106-LB). Diabetes 2015;64(Suppl. A. Processing of proglucagon to GLP-1 in pan- duction in glucagon receptor expression by an 1A):LB27 creatic a-cells: is this a paracrine mechanism antisense oligonucleotide ameliorates diabetic 14. Engel SS, Xu L, Andryuk PJ, et al. Efficacy and enabling GLP-1 to act on b-cells? J Endocrinol syndrome in db/db mice. Diabetes 2004;53: tolerability of MK-0893, a glucagon receptor an- 2011;211:99–106 410–417 tagonist (GRA), in patients with type 2 diabetes 25. Longuet C, Robledo AM, Dean ED, et al. 36. Hancock AS, Du A, Liu J, Miller M, May CL. Glu- (T2DM) (Abstract 309-OR). Diabetes 2011;60 Liver-specific disruption of the murine glucagon cagon deficiency reduces hepatic glucose produc- (Suppl. 1):A85 receptor produces a-cell hyperplasia: evidence tion and improves glucose tolerance in adult 15. Ruddy M, Pramanik B, Lunceford J, et al. for a circulating a-cell growth factor. Diabetes mice. Mol Endocrinol 2010;24:1605–1614 Inhibition of glucagon-induced hyperglycemia 2013;62:1196–1205 37. Sloop KW, Cao JX, Siesky AM, et al. Hepatic predicts glucose lowering efficacy of a glucagon 26. Ali S, Lamont BJ, Charron MJ, Drucker DJ. and glucagon-like peptide-1-mediated reversal receptor antagonist, MK-0893, in type 2 diabe- Dual elimination of the glucagon and GLP-1 re- of diabetes by glucagon receptor antisense oligo- tes (T2DM) (Abstract 311-OR). Diabetes 2011; ceptors in mice reveals plasticity in the incretin nucleotide inhibitors. J Clin Invest 2004;113: 60(Suppl. 1):A85–A86 axis. J Clin Invest 2011;121:1917–1929 1571–1581