Diabetes Care 1

Bo Ahrén,1 Susan L. Johnson,2

HARMONY 3: 104-Week Murray Stewart,3 Deborah T. Cirkel,4

CLIN CARE/EDUCATION/NUTRITION/PSYCHOSOCIAL
Fred Yang,5 Caroline Perry,5 and
Randomized, Double-Blind, Mark N. Feinglos,6 for the HARMONY 3

Placebo- and Active-Controlled Study Group

Trial Assessing the Efficacy and

Safety of Albiglutide Compared
With Placebo, Sitagliptin, and
Glimepiride in Patients With Type 2
Diabetes Taking Metformin
DOI: 10.2337/dc14-0024

OBJECTIVE
To compare the efficacy and safety of weekly albiglutide with daily sitagliptin,
daily glimepiride, and placebo.

RESEARCH DESIGN AND METHODS

Patients with type 2 diabetes receiving metformin were randomized to albiglutide
(30 mg), sitagliptin (100 mg), glimepiride (2 mg), or placebo. Blinded dose titration
for albiglutide (to 50 mg) and glimepiride (to 4 mg) was based on predefined
hyperglycemia criteria. The primary end point was change in HbA1c from baseline
at week 104. Secondary end points included fasting plasma glucose (FPG), weight,
and time to hyperglycemic rescue.

RESULTS
Baseline characteristics were similar among the albiglutide (n = 302), glimepiride 1
Department of Clinical Sciences, Lund Univer-
(n = 307), sitagliptin (n = 302), and placebo (n = 101) groups. Baseline HbA1c was sity, Lund, Sweden
8.1% (65.0 mmol/mol); mean age was 54.5 years. The mean doses for albiglutide 2
Metabolic Pathways and Cardiovascular Unit,
and glimepiride at week 104 were 40.5 and 3.1 mg, respectively. At week 104, GlaxoSmithKline, Research Triangle Park, NC
3
Metabolic Pathways and Cardiovascular Unit,
albiglutide significantly reduced HbA1c compared with placebo (20.9% [29.8
GlaxoSmithKline, Upper Merion, PA
mmol/mol]; P < 0.0001), sitagliptin (20.4% [24.4 mmol/mol]; P = 0.0001), and 4
RD Projects Clinical Platforms & Sciences,
glimepiride (20.3% [23.3 mmol/mol]; P = 0.0033). Outcomes for FPG and HbA1c GlaxoSmithKline, Stevenage, Herts, U.K.
5
were similar. Weight change from baseline for each were as follows: albiglutide Alternate Development Program, GlaxoSmithKline,
Upper Merion, PA
21.21 kg (95% CI 21.68 to 20.74), placebo 21.00 kg (95% CI 21.81 to 20.20), 6
Department of Medicine, Division of Endocrinol-
sitagliptin 20.86 kg (95% CI 21.32 to 20.39), glimepiride 1.17 kg (95% CI 0.70–1.63). ogy, Metabolism, and Nutrition, Duke University
The difference between albiglutide and glimepiride was statistically significant (P < Medical Center, Durham, NC
0.0001). Hyperglycemic rescue rate at week 104 was 25.8% for albiglutide compared Corresponding author: Mark N. Feinglos, mark.
with 59.2% (P < 0.0001), 36.4% (P = 0.0118), and 32.7% (P = 0.1504) for placebo, feinglos@duke.edu.
sitagliptin, and glimepiride, respectively. Rates of serious adverse events in the Received 4 January 2014 and accepted 11 April
albiglutide group were similar to comparison groups. Diarrhea (albiglutide 12.9%, 2014.
other groups 8.6–10.9%) and nausea (albiglutide 10.3%, other groups 6.2–10.9%) Clinical trial reg. no. NCT00838903, clinicaltrials
.gov.
were generally the most frequently reported gastrointestinal events.
This article contains Supplementary Data online
CONCLUSION at http://care.diabetesjournals.org/lookup/
suppl/doi:10.2337/dc14-0024/-/DC1.
Added to metformin, albiglutide was well-tolerated; produced superior reduc-
© 2014 by the American Diabetes Association.
tions in HbA1c and FPG at week 104 compared with placebo, sitagliptin, and See http://creativecommons.org/licenses/by-
glimepiride; and resulted in weight loss compared with glimepiride. nc-nd/3.0/ for details.
Diabetes Care Publish Ahead of Print, published online June 4, 2014
2 HARMONY 3: 2-Year Albiglutide Efficacy/Safety Diabetes Care

The management of type 2 diabetes has RESEARCH DESIGN AND METHODS glimepiride 2 mg, or placebo. Matching
become an increasingly complex prac- Study Design placebos for albiglutide, sitagliptin, and
tice given the number of treatments This was a phase 3, randomized, double- glimepiride were used to maintain
available (1). Despite the availability of blind, placebo- and active-controlled blinding to treatment.
new therapeutic agents, many patients parallel-group study that occurred be- After randomization, patients with
with type 2 diabetes continue to have tween 17 February 2009 and 21 March persistent hyperglycemia qualified to
uncontrolled glycemia (2–4), and treat- 2013; the study comprised 4 study pe- undergo dose titration and/or hypergly-
ment adherence varies but remains rel- riods: screening, run-in/stabilization (4 cemia rescue. Blinded up-titration of al-
atively poor (5–9). weeks), treatment (156 weeks; 104- biglutide from 30 to 50 mg once weekly
Among patients taking metformin but week data are reported here), and or glimepiride from 2 to 4 mg once daily
who do not have adequately controlled posttreatment follow-up (8 weeks) occurred if patients exceeded prede-
diabetes, data are limited regarding (Fig. 1A); the trial is registered as fined FPG or HbA1c thresholds. The final
next-step concomitant treatment (10). NCT00838903 at clinicaltrials.gov. Eligi- titration threshold, from week 12 until
As a result, extended head-to-head ble patients were stratified by HbA 1c week 143, was HbA1c 7.5% (58.5 mmol/mol);
comparisons of type 2 diabetes medica- level (,8.0% [,63.9 mmol/mol] vs. there was no titration from week 143
tions mirroring treatment algorithms $8.0% [$63.9 mmol/mol]), history of to week 156. In general, 4 weeks after
(i.e., combinations with metformin) myocardial infarction (MI), and age up-titration, patients meeting criteria in-
are needed to aid clinicians in making (,65 vs. $65 years) and were ran- cluding predefined FPG or HbA1c thresh-
treatment decisions (11,12). Recent domly assigned (3:3:3:1) to receive, in olds could receive hyperglycemic rescue
treatment guidelines have positioned addition to their background metfor- therapy, in addition to study medication,
incretin-based therapies, such as GLP-1 min, 1 of 4 treatments at baseline: albi- and remain in the trial. Rescue thresh-
receptor agonists (GLP-1RAs), as alter- glutide 30 mg, sitagliptin 100 mg, olds early in the trial were based
native first-line therapies in certain
clinical settings and as second-line ther-
apies following metformin because of
their substantial effectiveness in im-
proving glycemic control as well as
other positive effects, such as weight
loss and low hypoglycemia rates
(10,13,14).
Albiglutide is a novel, once-weekly,
long-acting GLP-1RA composed of a di-
peptidyl peptidase-4 (DPP-4)2resistant
GLP-1 dimer fused to recombinant hu-
man albumin. This structure affords an
extended half-life of ;5 days and, as a
consequence, once-weekly dosing
(15,16). In a multinational phase 2b
study of type 2 diabetes, albiglutide 30 mg
once-weekly reduced HbA1c by 20.87%
(29.5 mmol/mol) and also reduced fast-
ing plasma glucose (FPG) and weight,
with a low incidence of gastrointestinal
(GI) adverse events (AEs) (16).
The HARMONY program for albiglu-
tide includes eight pivotal phase 3 stud-
ies designed to evaluate the efficacy and
safety of albiglutide compared with pla-
cebo, oral antidiabetic medications, in-
sulin glargine, and another GLP-1RA
in a typical type 2 diabetes popula-
tion (17,18). Here, we present 2-year
primary end point data for HARMONY
3, which is a 3-year phase 3 study
comparing the efficacy and safety of
weekly albiglutide with daily sitagliptin,
daily glimepiride, and placebo in pa-
tients with diabetes receiving metfor-
min but not adequately controlled by Figure 1—A: HARMONY 3 study design. QD, once daily; QW, once weekly. B: CONSORT flow diagram
the medication. of the HARMONY 3 trial through week 104. DC, discontinue; FU, follow-up; ITT, intent to treat.
care.diabetesjournals.org
on FPG ($280 mg/dL [15.6 mmol/L]
from week 2 to week 4; $250 mg/dL
[13.9 mmol/L] from week 4 to week
12) and, later, on HbA1c ($8.5% [69.4
mmol/mol] and a #0.5% reduction
from baseline from week 12 to week
24; $8.5% [69.4 mmol/mol] from
week 24 to week 104).
Patients
Inclusion Criteria
Patients were $18 years of age, had
type 2 diabetes, and were experiencing
inadequate glycemic control while tak-
ing background metformin ($1500 mg
or maximum tolerated dose) $3 months
before screening. They also had a baseline
HbA1c of 7.0% (53.0 mmol/mol) to 10.0%
(85.8 mmol/mol), inclusive; BMI 20 to 45
kg/m2; creatinine clearance .60 mL/min
(determined using the Cockcroft-Gault
formula); and normal thyroid-stimulating
hormone concentration or were clinically
euthyroid.
Exclusion Criteria
Major exclusion criteria were current
ongoing symptomatic biliary disease or
history of pancreatitis, recent clinically
significant cardiovascular and/or cere-
brovascular disease (#2 months before
screening), treated gastroparesis, his-
tory of GI surgery thought to signifi-
cantly affect upper GI function, history
of most cancers not in remission for at
least 3 years, personal or family history
of medullary thyroid carcinoma or mul-
tiple endocrine neoplasia type 2, resting
systolic blood pressure (SBP) .160
mmHg and/or diastolic blood pressure
(DBP) .100 mmHg, lipase above the up-
per limit of normal (ULN), hemoglobin-
opathy that could affect HbA 1c , and
alanine aminotransferase or aspartate
aminotransferase more than two and a
half times the ULN.
Withdrawal Criteria
Patients withdrew or were removed from
the study drug because of loss to follow-
up, protocol violation, noncompliance,
withdrawal of consent, or an AE or labo-
ratory result requiring withdrawal, includ-
ing QTc prolongation, elevation of liver
function test results, severe potential al-
lergic reactions, confirmed pancreatitis,
severe or repeated hypoglycemia, and
calcitonin .100 pg/mL.
Primary and Secondary End Points
The primary end point was the change in
model-adjusted HbA1c from baseline to
week 104 between albiglutide and
the comparators. Secondary end points
included changes in HbA 1c , FPG, and
weight from baseline over time; the pro-
portion of patients who achieved HbA1c
treatment goals (i.e., ,6.5% [,47.5
mmol/mol], ,7.0% [,53.0 mmol/mol],
and ,7.5% [,58.5 mmol/mol]); and
time to hyperglycemic rescue.
Safety and tolerability were assessed,
including AEs and serious AEs (SAEs);
safety events of special interest (i.e., GI
or hypoglycemic events, injection-site
reactions, pancreatitis, thyroid tumors,
potential systemic allergic reactions
[SARs], and CV events); clinical labora-
tory evaluations; physical examinations;
12-lead electrocardiograms; vital sign
measurements; and immunogenicity.
All major CV AEs occurring post-
randomization were adjudicated by a
clinical end point committee and are
part of an ongoing meta-analysis (results
will be reported separately). An indepen-
dent, blinded pancreatitis adjudication
committee (PAC) comprising three GI
specialists adjudicated AEs suggesting
pancreatitis and all laboratory elevations
of lipase and/or amylase more than or
equal to three times the ULN. The PAC
adjudicated both the probability of
events being pancreatitis (definite, prob-
able, possible, not likely) and the likeli-
hood of a relationship to the study drug
(definite, probable, possible, unlikely al-
ternate etiology).
To assess any association of study med-
ication with potential SARs, investigators
flagged AEs that were considered SARs.
Additional evaluation of the AE data for
SARs was conducted using standardized
Medical Dictionary for Regulatory Activi-
ties queries for angioedema, anaphylaxis,
and severe cutaneous reaction. Hypogly-
cemia was defined and assessed using
American Diabetes Association criteria
(19) and evaluated before hyperglycemic
rescue.
Statistical Methods
The planned sample size provided
.90% power to demonstrate superior-
ity versus placebo and noninferiority
versus sitagliptin and glimepiride (non-
inferiority margin = 0.3%). Superiority of
albiglutide versus sitagliptin and glime-
piride was tested if noninferiority was
established. The primary efficacy end
point (intent-to-treat population/last
observation carried forward algorithm)
Ahrén and Associates 3
was analyzed using an ANCOVA model
to compare treatment effects adjusting
for region, history of previous MI, age
category, and baseline HbA1c. Patients
rescued from hyperglycemia before
week 104 had their latest HbA1c value
before rescue carried forward for pri-
mary analyses. Those patients who
achieved treatment goals were evalu-
ated by nonparametric ANCOVA. Time
to rescue was evaluated by Kaplan-
Meier curves and log-rank tests. For
primary end point analysis, log-rank
tests for the duration up to the primary
end point were added post hoc. A mul-
tiple comparisons adjustment strategy
was implemented for various inferen-
tial tests among the primary and key
secondary objectives; AEs were ana-
lyzed by incidence proportion and inci-
dence density rate overall and before
rescue (with additional type 2 diabetes
medication); in this article, overall in-
cidence/rate is used for all events ex-
cept hypoglycemia.
Study Conduct
This study was conducted in accordance
with good clinical practice standards, all
applicable privacy requirements, and
the guiding principles of the Declaration
of Helsinki. Written informed consent
was obtained from each patient before
participation.
RESULTS
This study was conducted at 289 centers
in 10 countries. Demographics and base-
line characteristics were similar among
the albiglutide (n = 302), glimepiride (n =
307), sitagliptin (n = 302), and placebo
(n = 101) treatment groups (Supplemen-
tary Table 1).The mean age for the
groups ranged from 54.3 to 56.1 years
(84.3% were ,65 years old); approxi-
mately half of patients were men and
the majority were white (63.4–74.5%).
Approximately 67.0% of the total popu-
lation were moderately to severely
obese (BMI $30 kg/m2); mean weight
in all treatment groups ranged from
89.6 to 91.8 kg, and mean duration of
diabetes ranged from 5.8 to 6.7 years. In
all treatment groups at baseline, HbA1c
levels were similar (8.1–8.2% [65.0–66.1
mmol/mol]), as were FPG concentra-
tions (9.0–9.3 mmol/L [162.8–167.4
mg/dL]).
Overall, 67.4% of patients completed
active treatment through week 104;
4 HARMONY 3: 2-Year Albiglutide Efficacy/Safety Diabetes Care

rates of completion were similar among

the active treatment groups (sitagliptin
67.7%, glimepiride 68.8%, and albiglu-
tide 68.3%) and lower for the placebo
group (59.6%). The main reasons for dis-
continuing active treatment were with-
drawal of consent (placebo 14.4%,
sitagliptin 13.1%, glimepiride 13.2%, al-
biglutide 12.1%) and AEs (placebo 4.8%,
sitagliptin 3.2%, glimepiride 4.1%, albi-
glutide 6.3%). Patient flow is presented
in Fig. 1B.
Although all treatment groups went
through a blinded up-titration proce-
dure, only the albiglutide and glimepir-
ide groups actually up-titrated to 50 and
4 mg, respectively. By week 104, a
higher proportion of patients went
through the blinded up-titration process
in the placebo (;69%) and sitagliptin
(;59%) groups compared with the gli-
mepiride (;54%) and albiglutide (;53%)
groups. Mean doses for albiglutide and
glimepiride at week 104 were 40.5 and
3.1 mg, respectively.

Primary Outcome: HbA1c

HbA1c reductions from baseline to week
104 were 20.63% (26.9 mmol/mol) for
albiglutide, 20.28% (23.1 mmol/mol)
for sitagliptin, 20.36% (23.9 mmol/mol)
for glimepiride, and +0.27% (3.0 mmol/mol)
for placebo. The model-adjusted mean
difference (albiglutide vs. comparator
treatment) demonstrated that albiglutide,
when added to metformin, was clinically
and statistically superior to placebo
(20.9% [29.8 mmol/mol]; 95% CI 21.2
to 20.7; P , 0.0001), sitagliptin (20.4%
[24.4 mmol/mol]; 95% CI 20.5 to 20.2;
P = 0.0001), and glimepiride (20.3%
[23.3 mmol/mol]; 95% CI 20.5 to
20.1; P = 0.0033) (Fig. 2A). Albiglutide
had the most durable effect, with a mean
HbA 1c of 7.5% (58.5 mmol/mol) at
week 104 compared with sitagliptin
(7.8% [61.7 mmol/mol]), glimepiride
(7.8% [61.7 mmol/mol]), and placebo
(8.4% [68.3 mmol/mol]) (Fig. 2B). Figure 2—Change in HbA1c from baseline (A), mean HbA1c (B), FPG (C), and weight (D) over
More albiglutide-treated patients time through week 104. Intent-to-treat population data were determined using the last
reached HbA1c treatment thresholds at observation carried forward method. Data in Fig. 2A denote difference (95% CI). *P ,
each level (,6.5% [,47.5 mmol/mol], 0.0001 vs. placebo; †P , 0.0001, noninferiority of albiglutide vs. sitagliptin or glimepiride;
‡P = 0.0033, superiority of albiglutide vs. glimepiride; §P = 0.0001, superiority of albiglutide vs.
,7.0% [,53.0 mmol/mol], ,7.5% sitagliptin. BL, baseline.
[,58.5 mmol/mol]) compared with pla-
cebo, sitagliptin, and glimepiride at
week 104 (Supplementary Fig. 1). The sitagliptin and glimepiride at the respectively. Subgroup analyses for
difference between albiglutide and the ,7.5% (,58.5 mmol/mol; P # 0.04) age, race, ethnicity, sex, baseline BMI,
comparators was statistically significant level. For the albiglutide group, these and baseline HbA1c were all consistent
for placebo at all threshold levels (P # treatment thresholds were met by with the primary end point (Supplemen-
0.02) and was comparable with only 17.1%, 38.6%, and 58.7% of patients, tary Fig. 2).
care.diabetesjournals.org
Secondary Outcomes
FPG
Changes in FPG from baseline at week
104 were consistent with the primary
end point (Fig. 2C). The treatment group
difference (albiglutide – comparator
treatment) for a model-adjusted least
squares mean decrease from baseline
to week 104 demonstrated that, when
added to metformin, albiglutide was
statistically superior to placebo (21.5
mmol/L [228 mg/dL]; P , 0.0001), sita-
gliptin (20.9 mmol/L [216 mg/dL]; P =
0.0002), and glimepiride (20.6 mmol/L
[210 mg/dL]; P = 0.0133).
Weight
At week 104, weight loss occurred
among patients taking albiglutide
(21.21 kg), placebo (21.0 kg), and sita-
gliptin (20.86 kg), whereas weight gain
occurred in glimepiride patients (+1.17
kg). The treatment difference between
albiglutide and glimepiride was statisti-
cally significant (P , 0.0001) (Fig. 2D).
Time to Hyperglycemic Rescue
The albiglutide group had the most du-
rable glycemic control. Per Kaplan-
Meier estimates, fewer patients received
hyperglycemic rescue in the albiglutide
group (25.8%) compared with the pla-
cebo (59.2%), sitagliptin (36.4%), and gli-
mepiride groups (32.7%). The difference
between baseline and time to rescue
with albiglutide was statistically signifi-
cantly longer compared with sitagliptin
(P = 0.0118) and placebo (P , 0.00001)
(Fig. 3).
General Safety and Tolerability
For overall safety, the proportion of pa-
tients experiencing AEs, SAEs, and AEs
leading to withdrawal while receiving
therapy was relatively balanced among
treatment groups (Table 1). Most AEs
were mild or moderate in intensity in
all treatment groups. The most frequent
AEs (occurring in .7% of patients taking
albiglutide) were upper respiratory tract
infection, diarrhea, nausea, injection-site
reaction, hypertension, nasopharyngi-
tis, and cough. The type of AEs overall
and those occurring before rescue
(i.e., before the addition of hyperglycemic
rescue medication) were similar, and the
majority of AEs occurred before rescue
(Supplementary Table 2). The rate of
AEs was higher with albiglutide and
generally due to the higher rate
of injection-site reactions (45 of 100
patient-years for albiglutide vs. 1 of 100
Figure 3—Kaplan-Meier plot of time to hyperglycemic rescue (intent-to-treat population). Time
to rescue was evaluated by log-rank tests and Kaplan-Meier curves.
patient-years for placebo). Fewer pa-
tients experienced SAEs in the albiglutide
group (12.9%) than in the placebo group
(15.8%) but were comparable with those
who experienced SAEs in the sitagliptin
and glimepiride groups (9.9% and
11.7%, respectively). Overall, eight fatal
SAEs occurred through week 104 (three
each for albiglutide and glimepiride and
one each for sitagliptin and placebo);
none were considered by the investiga-
tors to be related to study medication.
Events of Special Interest (Through
Week 104)
GI Events
GI AEs were experienced by a similar
proportion of patients in the albiglutide
(n = 110; 36.4%) and placebo groups (n =
38; 37.6%) and fewer in the sitagliptin
(n = 75; 24.8%) and glimepiride groups
(n = 85; 27.7%). Few GI events were se-
vere in intensity (1.0%, 1.3%, 1.3%, and
4.0% for placebo, sitagliptin, glimepir-
ide, and albiglutide, respectively) or led
to withdrawal of the study drug (1.0%,
1.0%, 0.7%, and 2.0%, respectively). The
most common GI event was diarrhea in
the sitagliptin, glimepiride, and albiglu-
tide groups and constipation in the pla-
cebo group.
At week 104, nausea event rates
were comparable between treatment
groups (7.9, 5.3, 3.7, and 9.3 events
per 100 patient-years for placebo,
Ahrén and Associates 5
sitagliptin, glimepiride, and albiglutide,
respectively), even though the inci-
dence was highest in the placebo group
(10.9%) compared with sitagliptin
(6.6%), glimepiride (6.2%), and albiglu-
tide (10.3%). The incidence and event
rate of vomiting over the 104 weeks
was (5.6%, 4.4 events per 100 patient-
years) in the albiglutide group com-
pared with placebo (1.0%, 0.6 events
per 100 patient-years), sitagliptin
(4.3%, 3.2 events per 100 patient-
years), and glimepiride (3.6%, 2.1 events
per 100 patient-years). The per-visit prev-
alence of nausea/vomiting among pa-
tients receiving albiglutide was ,5% at
each time point (Supplementary Fig. 3).
Hypoglycemia
Documented symptomatic hypoglyce-
mia (#3.9 mmol/L [#70 mg/dL]) before
rescue with albiglutide was low (3.0%)
and was similar to placebo (4.0%) and
sitagliptin (1.7%) compared with glime-
piride (17.9%) (Table 1). No severe
hypoglycemic events were reported be-
fore rescue.
Injection-Site Reactions
Events classified by investigators as
injection-site reactions occurred more
frequently with albiglutide (17.2%; n =
52) compared with glimepiride (7.8%;
n = 24), sitagliptin (6.3%; n = 19), and
placebo (5%; n = 5). Injection-site reac-
tions for albiglutide were mostly mild to
6 HARMONY 3: 2-Year Albiglutide Efficacy/Safety Diabetes Care

Table 1—AEs while receiving therapy, to week 104

Placebo + metformin Sitagliptin + metformin Glimepiride + metformin Albiglutide + metformin
(n = 101) (n = 302) (n = 307) (n = 302)
Overall safety to week 104
Any AEs 79.2/283.0 79.1/250.2 83.1/261.1 83.8/350.7
SAEs 12.9/10.7 8.9/5.6 9.4/6.5 11.9/10.1
Related AEs 20.8/24.4 17.9/30.6 17.6/19.4 31.1/101.4
AEs leading to withdrawal 5.0/3.1 3.6/2.1 4.6/2.6 6.6/3.8
Most common AEs (.7% with albiglutide)
Upper respiratory tract infection 9.9/7.3 9.3/7.0 8.5/6.5 16.2/12.4
Diarrhea 10.9/7.3 8.6/6.1 9.1/6.5 12.6/9.1
Nausea 10.9/7.9 6.6/5.3 6.2/3.7 10.3/9.3
Injection-site reaction 2.0/1.2 1.7/1.0 2.6/1.7 9.6/48.1
Hypertension 5.0/3.1 8.6/5.3 6.5/4.3 7.9/4.6
Nasopharyngitis 7.9/4.9 9.3/7.6 9.1/6.5 7.6/7.0
Cough 5.9/4.3 6.3/4.6 7.5/4.9 7.3/4.6
Hypoglycemia before rescue*
Severe 0 0 0 0
Documented symptomatic 4 (4.0)/3.6 5 (1.7)/2.8 55 (17.9)/60.7 9 (3.0)/2.6
Asymptomatic 1 (1.0)/1.8 4 (1.3)/1.4 3 (1.0)/0.7 4 (1.3)/1.3
Data are incidence (percentage of patients with event) and event rate (events per 100 person-years). The order of AEs is based on albiglutide
proportion in the overall data. *Patients with more than one hypoglycemic event were counted in all severity categories reported. Percentages are
based on the number of patients in each treatment group for the study being summarized. Severity was derived using the American Diabetes
Association guidelines for categorization of hypoglycemic events, as follows: severe = required assistance of another person; documented
symptomatic = typical symptoms accompanied by a plasma glucose concentration of #3.9 mmol/L; and asymptomatic = no symptoms but plasma
glucose concentration #3.9 mmol/L.

moderate in intensity (98.1%), were not
progressive, and most patients (n = 32/
52; 61.5%) experiencing injection-site
reactions had 1 or 2 events that lasted
#1 week (median duration). These re-
actions led to the withdrawal of 4 of
302 patients (1.3%) at 2 years. Most
patients (41 of 52; 79.0%) treated
with albiglutide who experienced in-
jection-site reactions were negative
for antialbiglutide antibodies.
Pancreatitis
Six patients (albiglutide, n = 4; glimepir-
ide, n = 2) were evaluated by a blinded,
independent PAC because of a reported
AE or a lipase/amylase concentration
more than or equal to three times the
ULN. Adjudication determined that nei-
ther patient from the glimepiride group
had pancreatitis. Among the four pa-
tients receiving albiglutide, adjudication
determined that one event was unlikely
to reflect pancreatitis, one event was
considered to reflect possible pancrea-
titis (laboratory elevations alone) un-
likely related to the study drug, and
two events were probable pancreatitis
(laboratory elevations and suggestive
symptoms), and both were considered
possibly related to the study drug.
Thyroid Cancer
One patient treated with albiglutide de-
veloped follicular cancer on day 243.
Two patients receiving sitagliptin devel-
oped thyroid cancer: one experienced
follicular papillary carcinoma on day 48
and the other papillary thyroid cancer
on day 549. In these three patients, thy-
roid cancer diagnoses were considered
by the investigator to be not related to
the study drug, and calcitonin concen-
trations were not abnormal.
Potential SARs
A search of the standardized Medical
Dictionary for Regulatory Activities
queries identified two patients (pla-
cebo, n = 1; glimepiride, n = 1) who ex-
perienced angioedema as an SAE. The
investigator considered neither to be re-
lated to the study drug but instead re-
lated to ACE inhibitor therapy. Both
patients remained in the study after
discontinuing ACE inhibitor therapy.
Neither patient had positive antialbi-
glutide antibodies, and no other cases
of angioedema, anaphylaxis, pharyngeal
edema, or laryngeal edema occurred.
The investigators did not flag any addi-
tional events of interest.
Immunogenicity
Antialbiglutide antibody incidence
among albiglutide-treated subjects was
7.0% (21 of 302 subjects), and included
1 subject (0.4%) with preexisting anti-
bodies (positive at baseline). Antibodies
were nonneutralizing and showed
reactivity with GLP-1 in 16 of the 21
antibody-positive subjects and with al-
bumin in 5 antibody-positive subjects.
No samples tested positive for antialbi-
glutide IgE antibodies, and no SARs were
reported for antibody-positive subjects.
Cardiovascular Parameters
At baseline, lipids and blood pressure
were well controlled and mean values
were similar across the four treatment
groups (Supplementary Table 3). There
were no clinically meaningful mean
changes to lipids throughout the study.
Minor changes in SBP and DBP, respec-
tively, from baseline occurred in all treat-
ment groups at week 104: placebo (2.2
and 0 mmHg), sitagliptin (0.2 and 0.2
mmHg), glimepiride (1.5 and 1.0
mmHg), albiglutide (21.0 and 20.7
mmHg). Treatment differences in SBP
and DBP at week 104 between albiglutide
and the three comparators were not sta-
tistically significant. Similarly, there were
no meaningful changes from baseline at
week 104 in mean heart rate: placebo, 2.3
bpm; sitagliptin, 0.8 bpm; glimepiride,
20.5 bpm; albiglutide, 1.3 bpm.
CONCLUSIONS
When added to metformin, albiglutide
produced clinically and statistically su-
perior and more sustained reduction in
HbA 1c at week 104 compared with
care.diabetesjournals.org
placebo, sitagliptin, and glimepiride.
The sustained treatment effect of albi-
glutide also is supported by the time to
hyperglycemic rescue, which showed
that fewer subjects taking albiglutide re-
quired rescue by week 104 compared
with patients treated with placebo, sita-
gliptin, and glimepiride; the difference
was statistically significant for albiglu-
tide versus placebo and sitagliptin. In-
terestingly, the reduction in HbA1c was
similar during the early treatment pe-
riod (;4 weeks). This might not have
been anticipated given the longer time
to a steady state for the once-weekly
albiglutide and the quick mechanism of
action of sulfonylureas. Similar clinically
and statistically superior results with al-
biglutide compared with all other treat-
ment arms were observed for change in
FPG from baseline and the proportion of
patients meeting clinically relevant
HbA1c treatment goals. In addition, pa-
tients receiving albiglutide, sitagliptin,
and placebo lost weight through week
104, whereas patients treated with gli-
mepiride gained weight; the difference
between albiglutide and glimepiride
was statistically significant (P , 0.0001).
The sustained effect of diabetes ther-
apy is a critical clinical issue. Previous
type 2 diabetes comparator trials have
demonstrated the superiority or nonin-
feriority of GLP-1RAs with or without
metformin but have been shorter 26-
week trials (20–22). The Diabetes Ther-
apy Utilization: Researching Change in
A1C, Weight, and Other Factors Through
Intervention with Exenatide Once-
Weekly (DURATION)-2 and Liraglutide
Effect Action in Diabetes (LEAD)-2 trials,
which included metformin in all treat-
ment arms, demonstrated the superior-
ity of exenatide over sitagliptin and
pioglitazone and of liraglutide over pla-
cebo, respectively, with regard to HbA1c
reductions (20–22). In another 26-week
study (HARMONY 8), once-weekly albi-
glutide provided superior glycemic im-
provement, similar tolerability, and
better patient compliance compared
with sitagliptin therapy in patients with
type 2 diabetes and renal impairment
(21). Understanding the long-term ef-
fects on type 2 diabetes of different
combination therapies has led to the
5-year Glycemia Reduction Approaches
in Diabetes: A Comparative Effective-
ness (GRADE) study. This trial is compar-
ing the efficacy and durability of four
classes of medications (DPP-4, sulfonyl-
ureas, GLP-1RAs, and long-acting insulin)
in patients with type 2 diabetes who are
receiving metformin (11). The results re-
ported here may give a window to the
eventual results of that trial.
Albiglutide was generally well toler-
ated; the albiglutide group showed rates
of SAEs and AEs leading to withdrawal
similar to those of the comparator
groups. The percentage of patients re-
porting AEs while receiving therapy was
similar for albiglutide, sitagliptin, and
glimepiride and slightly lower for pla-
cebo. The types of AEs noted for albiglu-
tide are generally consistent with the
type 2 diabetic population and the
known profile of GLP-1RAs (23,24).
The 104-week incidence of nausea/
vomiting with albiglutide (10.6%/5.6%)
was lower than values observed in pre-
vious studies using other GLP-1RAs, par-
ticularly in the early weeks of therapy. In
clinical practice, GI AEs occurring early in
the course of treatment with GLP-1RAs
may contribute to the discontinuation
rates seen with this class, and agents
with reduced GI effects may improve
patient adherence. With respect to hy-
poglycemia, albiglutide had rates similar
to sitagliptin and placebo and lower rates
compared with glimepiride. Injection-site
reactions were more common in the
albiglutide-treated subjects, although a
few (1.3%) withdrew because of reac-
tions during the 104-week period. The
proportion of patients developing anti-
albiglutide antibodies was low, and
none were nonneutralizing.
HARMONY 3 is a long-term, 3-year tri-
al with unique design features that must
be considered when interpreting re-
sults. First, up-titration was based on
glucose thresholds established by pro-
tocol: the time point of titration was
not prescribed or standardized, reflect-
ing up-titration methods in clinical prac-
tice; thus, the efficacy and safety data
reflect a mixture of 30- and 50-mg albi-
glutide doses and 2- and 4-mg glimepir-
ide doses. In addition, there was no
increased dose for placebo and sitaglip-
tin despite patients who underwent the
blinded up-titration procedure. Because
of this design feature, many patients
randomized to glimepiride and albiglu-
tide were not receiving the maximum
dose, as evidenced by the mean doses
of each drug. Therefore, the maximum
effect in these arms may not have been
Ahrén and Associates 7
achieved, depending on the dose curve
for each agent, and there is a possibility
that the relationship between the arms
may have shifted at higher doses.
Second, rescue medications were
used in the program to permit patients
to remain in the studies while receiving
a blinded investigational product or con-
trol through 3 years. This represents a
more “real-life” situation, where addi-
tional medications are added to regi-
mens that are not meeting treatment
goals. While the primary efficacy assess-
ment was based on end points before
rescue, the main safety assessment in-
cluded data after rescue. These design
features, however, may have compli-
cated the interpretation of some results.
Finally, interpretation of the proportion
of patients reaching the treatment goal
was complex; up-titration occurred at
7.5% and not 7.0%, and the full benefit
of 50 mg may not be seen since the end
point contains a mix of patients taking
30- and 50-mg albiglutide.
In conclusion, when added to metfor-
min, albiglutide produced clinically and
statistically significant reductions in
HbA1c and FPG at week 104 compared
with placebo, sitagliptin, and glimepir-
ide. Patients receiving albiglutide, sita-
gliptin, and placebo lost body weight
through week 104, whereas patients
taking glimepiride gained weight; the
difference between albiglutide and gli-
mepiride was statistically significant. Al-
biglutide was generally well tolerated,
and rates of SAEs were similar across
treatment groups. The most frequent
AEs for albiglutide were largely consistent
with the known profile for GLP-1RAs.
Acknowledgments. The authors thank Douglas
L. Wicks, MPH, CMPP, of GlaxoSmithKline, for
managing manuscript development. Editorial
support was provided by Leonard Lionnet, PhD,
of PharmaWrite, LLC (assistance with the pro-
duction of a draft outline and the first draft of
the manuscript, assembling tables and figures,
and collating author comments); Diana Talag,
ELS, of PharmaWrite, LLC (assistance with
copyediting and fact-checking); and Shana
Cambareri of PharmaWrite, LLC (art direction).
Duality of Interest. B.A. received honoraria for
lecturing and holds advisory board membership
for Bristol-Myers Squibb, GlaxoSmithKline,
Lilly, Novartis, Novo Nordisk, Merck, and Sanofi,
which all are companies producing GLP-1RAs
or DPP-4 inhibitors. S.L.J., M.S., D.T.C., F.Y., and
C.P. are employed by and shareholders of
GlaxoSmithKline. M.N.F. has received research
support and has served either as an investigator
8 HARMONY 3: 2-Year Albiglutide Efficacy/Safety
or principal investigator for Amylin, AstraZeneca,
Eli Lilly and Company, GlaxoSmithKline, Merck &
Co., Inc., Medtronic Inc., Novo Nordisk, Proctor &
Gamble Co., Prodigy Diabetes Care, Sanofi, and
Tethys, some of which are companies producing
GLP-1RAs or DPP-4 inhibitors. M.N.F. also has
served as a consultant for GlaxoSmithKline, Pfizer,
and Proctor & Gamble Co. No other potential
conflicts of interest relevant to this article were
reported.
The sponsor of the study participated in the
study design, data collection, data review, data
analysis, and writing of the report. All authors
had full access to all the data in the study. The
corresponding author reviewed the trial report
(signatory investigator), had full access to all
data in the study, and had final responsibility for
the decision to submit for publication.
Author Contributions. B.A. and M.N.F. pro-
vided study patients. S.L.J., M.S., D.T.C., F.Y.,
and C.P. analyzed and interpreted data. B.A.,
S.L.J., D.T.C., F.Y., and M.N.F. wrote and revised
the manuscript. F.Y. performed the statistical
analysis. All authors revised the manuscript,
reviewed the final manuscript, and approved
the manuscript for submission. M.N.F. is the
guarantor of this work and, as such, had full
access to all the data in the study and takes
responsibility for the integrity of the data and
the accuracy of the data analysis.
Prior Presentation. Data from this article were
presented at the 73rd Scientific Sessions of the
American Diabetes Association, Chicago, IL, 21–
25 June 2013, and published in Diabetes
2013;62(Suppl 1A):LB14 (Abstract 52-LB). Data
also were presented via oral presentation at the
49th Annual Meeting of the European Associa-
tion for the Study of Diabetes, Barcelona, Spain,
23–27 September 2013, and published in Dia-
betologia 2013;56(Suppl 1):S8.
References
1. Setji T, Feinglos M. Albiglutide: clinical over-
view of a long-acting GLP-1 receptor agonist in
the treatment of type 2 diabetes. Expert Rev
Endocrinol Metab 2013;8:229–238
2. Ong KL, Cheung BM, Wong LY, Wat NM, Tan
KC, Lam KS. Prevalence, treatment, and control
of diagnosed diabetes in the U.S. National
Health and Nutrition Examination Survey
1999-2004. Ann Epidemiol 2008;18:222–229
3. Saaddine JB, Cadwell B, Gregg EW, et al. Im-
provements in diabetes processes of care and
intermediate outcomes: United States, 1988–
2002. Ann Intern Med 2006;144:465–474
4. American Diabetes Association. Standards of
medical care in diabetesd2009. Diabetes Care
2009;32(Suppl 1):S13–S61
5. Aikens JE, Piette JD. Longitudinal association
between medication adherence and glycaemic
control in Type 2 diabetes. Diabet Med 2013;30:
338–344
6. Cramer JA. A systematic review of adherence
with medications for diabetes. Diabetes Care
2004;27:1218–1224
7. Bailey CJ, Kodack M. Patient adherence to
medication requirements for therapy of type 2
diabetes. Int J Clin Pract 2011;65:314–322
8. Davies MJ, Gagliardino JJ, Gray LJ, Khunti K,
Mohan V, Hughes R. Real-world factors affect-
ing adherence to insulin therapy in patients with
Type 1 or Type 2 diabetes mellitus: a systematic
review. Diabet Med 2013;30:512–524
9. DiMatteo MR. Variations in patients’ adher-
ence to medical recommendations: a quantita-
tive review of 50 years of research. Med Care
2004;42:200–209
10. Inzucchi SE, Bergenstal RM, Buse JB, et al;
American Diabetes Association (ADA); European
Association for the Study of Diabetes (EASD).
Management of hyperglycemia in type 2 diabetes:
a patient-centered approach: position statement
of the American Diabetes Association (ADA) and
the European Association for the Study of Diabe-
tes (EASD). Diabetes Care 2012;35:1364–1379
11. NIH begins recruitment for long-term study
of diabetes drug efficacy [press release online],
2013. Bethesda, MD, National Institutes of
Health. Available from http://www.nih.gov/
news/health/jun2013/niddk-03.htm. Accessed
20 August 2013.
12. Scheen AJ. DPP-4 inhibitors in the manage-
ment of type 2 diabetes: a critical review of
head-to-head trials. Diabetes Metab 2012;38:
89–101
13. St Onge EL, Miller SA. Albiglutide: a new
GLP-1 analog for the treatment of type 2 diabe-
tes. Expert Opin Biol Ther 2010;10:801–806
14. American Diabetes Association. Standards
of medical care in diabetesd2014. Diabetes
Care 2014;37(Suppl 1):S14–S80
15. Matthews JE, Stewart MW, De Boever EH,
et al. Pharmacodynamics, pharmacokinetics,
safety and tolerability of albiglutide, a long-
acting glucagon-like peptide-1 mimetic, in pa-
tients with type 2 diabetes. J Clin Endocrinol
Metab 2008;93:4810–4817
Diabetes Care
16. Rosenstock J, Reusch J, Bush M, Yang F,
Stewart M; Albiglutide Study Group. Potential
of albiglutide, a long-acting GLP-1 receptor ag-
onist, in type 2 diabetes: a randomized con-
trolled trial exploring weekly, biweekly, and
monthly dosing. Diabetes Care 2009;32:1880–
1886
17. Rosenstock J, Fonseca V, Gross JL, et al. Ad-
vancing basal insulin replacement in type 2 di-
abetes inadequately controlled with insulin
glargine plus oral agents: a comparison of add-
ing albiglutide, a weekly glp-1 receptor agonist
versus thrice daily prandial insulin lispro. Lan-
cet. In press
18. Pratley RE, Nauck MA, Barnett AH, et al;
HARMONY 7 Study Group. Once-weekly albiglu-
tide versus once-daily liraglutide in patients with
type 2 diabetes inadequately controlled on oral
agents (HARMONY 7): a randomised, open-label,
multicentre, non-inferiority phase 3 study. Lan-
cet Diabetes Endocrinol 2014;2:289–297
19. Workgroup on Hypoglycemia, American Di-
abetes Association. Defining and reporting hy-
poglycemia in diabetes: a report from the
American Diabetes Association Workgroup on
Hypoglycemia. Diabetes Care 2005;28:1245–
1249
20. Bergenstal RM, Wysham C, Macconell L,
et al; DURATION-2 Study Group. Efficacy and
safety of exenatide once weekly versus sitaglip-
tin or pioglitazone as an adjunct to metformin
for treatment of type 2 diabetes (DURATION-
2): a randomised trial. Lancet 2010;376:431–
439
21. Nauck M, Frid A, Hermansen K, et al; LEAD-2
Study Group. Efficacy and safety comparison of
liraglutide, glimepiride, and placebo, all in com-
bination with metformin, in type 2 diabetes: the
LEAD (liraglutide effect and action in diabetes)-2
study. Diabetes Care 2009;32:84–90
22. Charbonnel B, Steinberg H, Eymard E, et al.
Efficacy and safety over 26 weeks of an oral
treatment strategy including sitagliptin com-
pared with an injectable treatment strategy
with liraglutide in patients with type 2 diabetes
mellitus inadequately controlled on metformin:
a randomised clinical trial. Diabetologia 2013;
56:1503–1511
23. Victoza (liraglutide [rDNA origin] injection)
[prescribing information]. Princeton, NJ, Novo
Nordisk Inc., April 2013.
24. Byetta (exenatide) injection [prescribing in-
formation]. San Diego, CA, Amylin Pharmaceuti-
cals, Inc., February 2013.