The n e w e ng l a n d j o u r na l of m e dic i n e

Original Article

Lixisenatide in Patients with Type 2 Diabetes

and Acute Coronary Syndrome
MarcA. Pfeffer, M.D., Ph.D., Brian Claggett, Ph.D., Rafael Diaz, M.D.,
Kenneth Dickstein, M.D., Ph.D., HertzelC. Gerstein, M.D., LarsV. Kber, M.D.,
FrancescaC. Lawson, M.D., Lin Ping, M.D., Xiaodan Wei, Ph.D.,
EldrinF. Lewis, M.D., M.P.H., AldoP. Maggioni, M.D.,
JohnJ.V. McMurray, M.D., Ph.D., JeffreyL. Probstfield, M.D.,
MatthewC. Riddle, M.D., ScottD. Solomon, M.D., and JeanClaude Tardif, M.D.,
for the ELIXA Investigators*

A BS T R AC T

BACKGROUND
Cardiovascular morbidity and mortality are higher among patients with type 2 From the Cardiovascular Division, Brigham
diabetes, particularly those with concomitant cardiovascular diseases, than in and Womens Hospital, and Harvard Medi
cal School both in Boston (M.A.P., B.C.,
most other populations. We assessed the effects of lixisenatide, a glucagon-like E.F.L., S.D.S.); Estudios Clnicos Latino
peptide 1receptor agonist, on cardiovascular outcomes in patients with type 2 amrica, Rosario, Argentina (R.D.); Univer
diabetes who had had a recent acute coronary event. sity of Bergen, Stavanger University Hos
pital, Stavanger, Norway (K.D.); McMaster
University, Hamilton, ON, Canada (H.C.G.);
METHODS Rigshospitalet Copenhagen University Hos
We randomly assigned patients with type 2 diabetes who had had a myocardial pital, Copenhagen (L.V.K.); Sanofi U.S.,
infarction or who had been hospitalized for unstable angina within the previous Bridgewater, NJ (F.C.L., L.P., X.W.); Re
search Center of the Italian Association of
180 days to receive lixisenatide or placebo in addition to locally determined stan- Hospital Cardiologists, Florence (A.P.M.);
dards of care. The trial was designed with adequate statistical power to assess British Heart Foundation Cardiovascular
whether lixisenatide was noninferior as well as superior to placebo, as defined by Research Centre, University of Glasgow,
Glasgow, United Kingdom (J.J.V.M.); Uni
an upper boundary of the 95% confidence interval for the hazard ratio of less than versity of Washington Medical Center,
1.3 and 1.0, respectively, for the primary composite end point of cardiovascular Seattle (J.L.P.); Oregon Health and Sci
death, myocardial infarction, stroke, or hospitalization for unstable angina. ence University, Portland (M.C.R.); and
Montreal Heart Institute, Universit de
RESULTS Montral, Montreal (J.C.T.). Address re
print requests to Dr. Pfeffer at the Cardio
The 6068 patients who underwent randomization were followed for a median of vascular Division, Brigham and Womens
25 months. A primary end-point event occurred in 406 patients (13.4%) in the Hospital, 75 Francis St., Boston, MA 02115,
lixisenatide group and in 399 (13.2%) in the placebo group (hazard ratio, 1.02; or at [email protected].

95% confidence interval [CI], 0.89 to 1.17), which showed the noninferiority of * A list of the investigators and committee
lixisenatide to placebo (P<0.001) but did not show superiority (P=0.81). There were members in the Evaluation of Lixisenatide
in Acute Coronary Syndrome (ELIXA)
no significant between-group differences in the rate of hospitalization for heart trial is provided in the Supplementary
failure (hazard ratio in the lixisenatide group, 0.96; 95% CI, 0.75 to 1.23) or the rate Appendix, available at NEJM.org.
of death (hazard ratio, 0.94; 95% CI, 0.78 to 1.13). Lixisenatide was not associated N Engl J Med 2015;373:2247-57.
with a higher rate of serious adverse events or severe hypoglycemia, pancreatitis, DOI: 10.1056/NEJMoa1509225
pancreatic neoplasms, or allergic reactions than was placebo. Copyright 2015 Massachusetts Medical Society.

CONCLUSIONS
In patients with type 2 diabetes and a recent acute coronary syndrome, the addi-
tion of lixisenatide to usual care did not significantly alter the rate of major car-
diovascular events or other serious adverse events. (Funded by Sanofi; ELIXA
ClinicalTrials.gov number, NCT01147250.)

n engl j med 373;23nejm.org December 3, 2015 2247

The New England Journal of Medicine
Downloaded from nejm.org on April 9, 2017. For personal use only. No other uses without permission.
Copyright 2015 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

R
andomized trials involving pa- Me thods
tients with new or established type 2 dia-
betes have shown that improved glucose Study Design
control reduces the risk of microvascular com- The Evaluation of Lixisenatide in Acute Coronary
plications,1-3 with modest cardiovascular bene- Syndrome (ELIXA) trial was a multicenter, ran-
fits suggested by meta-analyses and extended domized, double-blind, placebo-controlled trial
follow-up of clinical trials.4-7 However, various involving patients with type 2 diabetes who had
studies indicate that, despite being effective in had a recent acute coronary syndrome. The trial
lowering the glucose and glycated hemoglobin was designed to assess the effects of lixisenatide
levels, some hypoglycemic medications may in- on cardiovascular morbidity and mortality. De-
crease, rather than reduce, the risk of cardio- tails of the trial design and the demographic
vascular events.8-10 These unexpected findings and clinical characteristics of the patients at
prompted the reexamination of the regulatory baseline have been reported previously.26 The
approval processes for new antidiabetic thera- study protocol, available with the full text of this
pies, which had been based primarily on the article at NEJM.org, was designed by the execu-
surrogate measure of glucose lowering with tive committee, which also oversaw the conduct
limited clinical-outcomes data. Since 2008, reg- of the trial (see the Supplementary Appendix,
ulatory agencies have required robust cardiovas- available at NEJM.org).
cular-outcome data from randomized, controlled The chair of the executive committee wrote
trials in order to grant and sustain approvals of the first draft of the manuscript, and all the
drugs or biologic agents for the treatment or coauthors contributed to the subsequent revi-
prevention of diabetes mellitus.11-13 sions. Data were analyzed and confirmed at two
Glucagon-like peptide 1 (GLP-1)receptor ago- statistical centers (one at the sponsors location
nists are a class of parenteral glucose-lowering and one at Brigham and Womens Hospital). Each
drugs that activate the receptor for the endoge- author could query the database by requesting
nous incretin GLP-1. These drugs lower glucose specific analyses or clarifications of definitions.
levels by inhibiting the secretion of glucagon, All the authors edited and approved the manu-
promoting the release of insulin in response to script and assume full responsibility for the
hyperglycemia, slowing gastric emptying, and accuracy and completeness of the data and for
augmenting satiety.14 On the basis of some evi- the fidelity of this report to the study protocol,
dence of cardioprotection in preliminary studies which was approved by the appropriate national
in animal models15,16 and in pilot studies of myo- and institutional regulatory and ethics boards. All
cardial ischemia17,18 and heart failure in hu- the patients provided written informed consent.
mans,19 it was postulated20,21 that in addition to
improving glycemic control and promoting weight Study Patients
loss, GLP-1receptor agonists may improve car- Eligible patients had type 2 diabetes and had
diovascular outcomes. had an acute coronary event within 180 days
Lixisenatide, a once-daily GLP-1receptor ago- before screening. Major exclusion criteria were
nist, is effective in reducing the glycated hemo- an age of less than 30 years, percutaneous coro-
globin level in patients with type 2 diabetes by nary intervention within the previous 15 days,
lowering both the fasting and the postprandial coronary-artery bypass graft surgery for the
blood glucose levels.22-25 However, the studies qualifying event, planned coronary revasculariza-
showing such glycemic control and other favorable tion procedure within 90 days after screening, an
metabolic effects of lixisenatide were not designed estimated glomerular filtration rate (eGFR) of
to accrue sufficient clinical events to adequately less than 30 ml per minute per 1.73 m2 of body-
show its cardiovascular safety. Although lixisena- surface area, a glycated hemoglobin level of less
tide and several other GLP-1receptor agonists than 5.5% or more than 11.0%, or an inability
are approved in many countries for use as glu- to provide written informed consent.26 Before
cose-lowering agents in patients with type 2 dia- randomization, there was a 1-week run-in period
betes, large cardiovascular-outcome trials with during which eligible consenting patients were
any agent in this class have been lacking. trained in the self-administration of daily subcu-

2248 n engl j med 373;23nejm.org December 3, 2015

The New England Journal of Medicine

Downloaded from nejm.org on April 9, 2017. For personal use only. No other uses without permission.
Copyright 2015 Massachusetts Medical Society. All rights reserved.
 Lixisenatide in Type 2 Diabetes and Acute Coronary Syndrome

taneous injections of unblinded placebo, with Study Oversight

the volume matched to that of the injections The data and safety monitoring committee,
they would receive after randomization. whose members had access to unblinded data,
monitored the safety of the participants. When
Study Interventions 144 adjudicated primary end-point events had
Patients were randomly assigned, in a double- occurred, the data and safety monitoring com-
blind manner, to once-daily subcutaneous injec- mittee reviewed a prespecified, independently
tions of lixisenatide or volume-matched placebo. performed analysis of the unblinded data and
Randomization was performed with the use of a informed the executive committee and the spon-
centralized assignment system. A starting dose sor that the upper boundary of the resulting 96%
of 10 g of lixisenatide per day or volume- confidence interval of the hazard ratio for the
matched placebo was administered during the comparison of lixisenatide with placebo was less
first 2 weeks and then increased, at the investi- than 1.8 (see the Supplementary Appendix).9,11,27
gators discretion, to a maximum dose of 20 g At that time, a group from the sponsor, which
of lixisenatide per day or volume-matched pla- was not in communication with the ELIXA team,
cebo. Glycemic control was managed by the in- prepared and subsequently filed a New Drug Ap-
vestigators in accordance with local clinical prac- plication with the Food and Drug Administra-
tice guidelines by the adjustment of concomitant tion, which was later withdrawn to avoid public
glucose-lowering agents or the addition of new disclosure of preliminary data.
antidiabetic medications with the exception of
other incretin therapies. This approach was ex- Statistical Analysis
pected to yield similar glycemic control in the The primary analysis was conducted in the inten-
two study groups. tion-to-treat population with the use of the Cox
proportional-hazards model, with study group
Study End Points and geographic region as the covariates, to esti-
The primary end point in the time-to-event mate the hazard ratio for the comparison of
analysis was a composite of the first occurrence lixisenatide with placebo. The noninferiority of
of any of the following: death from cardiovascu- lixisenatide to placebo would be shown if the
lar causes, nonfatal myocardial infarction, non- upper boundary of the 95% confidence interval
fatal stroke, or hospitalization for unstable an- of the hazard ratio was less than 1.3 and the
gina. Secondary end points included a composite superiority would be shown if the upper bound-
of the primary end point or hospitalization for ary was less than 1.0.11 The P values for the tests
heart failure and a composite of the primary end of noninferiority and superiority are derived from
point, hospitalization for heart failure, or coro- the Cox proportional-hazards model.
nary revascularization procedures. All-cause mor- The trial was event-driven. We estimated that
tality and the rates of the components of each of 6000 patients would need to be enrolled for us
the composite end points were examined. to observe 844 primary end-point events, which
Changes in the glycated hemoglobin level, would provide the study with 96% power to
urinary albumin-to-creatinine ratio, body weight, show noninferiority, assuming a true hazard
heart rate, and blood pressure were evaluated. ratio of 1.0, and with 90% power to show supe-
Safety end points, including the incidence of riority, assuming a true hazard ratio of 0.80.
hypoglycemia and severe hypoglycemia (requiring More complete details regarding sample-size
the assistance of another person), pancreatitis, assumptions have been published previously.26
cancer, and systemic allergic reactions, were as- Sensitivity analyses were conducted in which
sessed in patients who underwent randomiza- events that occurred more than 30 days after
tion and received at least one dose of lixisenatide the discontinuation of lixisenatide or placebo
or placebo. Separate independent committees were excluded; in addition, a post hoc Cox
whose members were unaware of the study- proportional-hazards analysis was conducted
group assignments adjudicated potential cardio- with a model that was adjusted for nominally
vascular, pancreatic, and allergic events (see the significant baseline imbalances. Postbaseline
Supplementary Appendix). measures were analyzed with the use of non-

n engl j med 373;23nejm.org December 3, 2015 2249

The New England Journal of Medicine
Downloaded from nejm.org on April 9, 2017. For personal use only. No other uses without permission.
Copyright 2015 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

Table 1. Characteristics of the Patients at Baseline.*

Characteristic Placebo (N=3034) Lixisenatide (N=3034)

Age yr 60.69.6 59.99.7
Age 65 yr no. (%) 1040 (34.3) 1003 (33.1)
Female sex no. (%) 938 (30.9) 923 (30.4)
Duration of diabetes yr 9.48.3 9.28.2
Glycated hemoglobin % 7.61.3 7.71.3
Retinopathy no. (%) 331 (10.9) 320 (10.5)
Neuropathy no. (%) 498 (16.4) 512 (16.9)
Body weight kg 85.119.6 84.619.2
Body-mass index 30.25.8 30.15.6
Race no. (%)
Asian 367 (12.1) 404 (13.3)
Black 103 (3.4) 118 (3.9)
Other 246 (8.1) 254 (8.4)
White 2318 (76.4) 2258 (74.4)
Hispanic ethnic group no. (%) 903 (29.8) 865 (28.5)
Geographic region no. (%)
Africa or Near East 142 (4.7) 154 (5.1)
Asia Pacific 329 (10.8) 374 (12.3)
Eastern Europe 811 (26.7) 776 (25.6)
North America 403 (13.3) 404 (13.3)
South or Central America 972 (32.0) 972 (32.0)
Western Europe 377 (12.4) 354 (11.7)
Current smoking no. (%) 354 (11.7) 355 (11.7)
Myocardial infarction before index ACS no. (%) 672 (22.1) 672 (22.1)
Medical history at randomization no. (%)
Hypertension 2340 (77.1) 2295 (75.6)
Percutaneous coronary intervention 2027 (66.8) 2052 (67.6)
Coronary-artery bypass grafting 249 (8.2) 258 (8.5)
Heart failure 676 (22.3) 682 (22.5)
Stroke 188 (6.2) 143 (4.7)
Peripheral arterial disease 229 (7.5) 237 (7.8)
Atrial fibrillation 190 (6.3) 176 (5.8)
Systolic blood pressure mm Hg 13017 12917
Heart rate beats/min 70.29.9 70.210.1
HDL cholesterol mg/dl 42.910.9 43.010.8
LDL cholesterol mg/dl 78.235.2 78.835.4
eGFR ml/min/1.73 m2 75.221.4 76.721.3
Qualifying ACS event no. (%)
NSTEMI 1183 (39.0) 1165 (38.4)
STEMI 1317 (43.4) 1349 (44.5)
Unstable angina 528 (17.4) 514 (16.9)
Unclassified 6 (0.2) 6 (0.2)

2250 n engl j med 373;23nejm.org December 3, 2015

The New England Journal of Medicine

Downloaded from nejm.org on April 9, 2017. For personal use only. No other uses without permission.
Copyright 2015 Massachusetts Medical Society. All rights reserved.
 Lixisenatide in Type 2 Diabetes and Acute Coronary Syndrome

Table 1. (Continued.)

Characteristic Placebo (N=3034) Lixisenatide (N=3034)

Days from ACS to randomization 72.243.9 71.843.4
Urinary albumin:creatinine ratio
Median 10.5 10.2
Interquartile range 6.033.6 6.029.6

* Plusminus values are means SD. There were no significant between-group differences at baseline, except with re
spect to age (P=0.005), estimated glomerular filtration rate (eGFR; P=0.006), glycated hemoglobin level (P=0.02), and
prior stroke (P=0.01). To convert the values for cholesterol to millimoles per liter, multiply by 0.02586. ACS denotes
acute coronary syndrome, HDL high-density lipoprotein, LDL low-density lipoprotein, NSTEMI nonST-segment eleva
tion myocardial infarction, and STEMI ST-segment elevation myocardial infarction.
The body-mass index is the weight in kilograms divided by the square of the height in meters.
Race and ethnic group were self-reported.
Peripheral arterial disease included amputation due to a cause other than trauma.
Albumin was measured in milligrams and creatinine in grams.

prespecified repeated-measures, mixed-effects observed that 805 patients had a confirmed pri-
linear regression models with patient-specific mary end-point event, which provided the study
random-intercept terms and an independent with more than 95% power for the test of non-
within-patient residual covariance structure. inferiority and more than 88% power for the test
Study group and visit were included in the of superiority.
model as factors, and baseline value was in- The characteristics of the study groups were
cluded as a covariate. Other comparisons were generally balanced at baseline (Table1, and Ta-
performed with the use of Students t-tests, ble S1 in the Supplementary Appendix). Nomi-
Wilcoxon rank-sum tests, and chi-square tests nally significant between-group differences were
and were not prespecified or adjusted for mul- observed in 4 of the 35 baseline comparisons
tiplicity. Summary statistics include all avail- (those regarding age, eGFR, glycated hemoglo-
able postbaseline data, whereas the figures bin level, and prior stroke) (Table1, and Table S1
exclude data beyond 40 months of follow-up in the Supplementary Appendix).
(<10% of patients had longer follow-up). The median follow-up was 25 months in each
study group. Among patients who did not die,
96.3% of the patients in the lixisenatide group
R e sult s
and 96.1% of those in the placebo group com-
Patients and Interventions pleted the study. Vital status was ascertained at
Between July 9, 2010, and August 2, 2013, we the end of the study in all but 29 patients (1.0%)
enrolled 6068 patients from 49 countries and in the lixisenatide group and in all but 42 (1.4%)
randomly assigned 3034 patients to lixisenatide in the placebo group.
and 3034 to placebo. All but 5 patients (3 patients The mean duration of exposure to study medi-
in the lixisenatide group and 2 in the placebo cation was 690 days in the lixisenatide group and
group) received at least one dose of lixisenatide 712 days in the placebo group. Excluding the 51
or placebo. patients in the lixisenatide group and 41 in the
At the recommendation of the data and safety placebo group who were taking the study medi-
monitoring committee, the study went to com- cation on the day of death, the study medication
pletion. On the basis of projections made by the was permanently discontinued in 833 patients
sponsor and shared with the executive commit- (27.5%) in the lixisenatide group and in 727
tee without knowledge of the study-group assign- (24.0%) in the placebo group (P=0.002). Among
ments, it was estimated that the initiation of patients who received at least one dose of lixisena-
end-of-study visits on November 11, 2014, would tide or placebo, 2591 of 3031 patients (85.5%) in
allow for the observation of the prespecified the lixisenatide group were taking the maximum
number of adjudicated events. The last patient dose of 20 g at the time of their last dose of
visit occurred on February 11, 2015. Overall, we lixisenatide, and 2926 of 3032 patients (96.5%)

n engl j med 373;23nejm.org December 3, 2015 2251

The New England Journal of Medicine
Downloaded from nejm.org on April 9, 2017. For personal use only. No other uses without permission.
Copyright 2015 Massachusetts Medical Society. All rights reserved.
 2252
Table 2. Incidence Rates and Hazard Ratios, with Adjustment for Geographic Region, for the Primary Composite End Point, Its Components, and Other Efficacy Outcomes.

Adjusted Hazard Ratio

End Point Placebo (N=3034) Lixisenatide (N=3034) (95% CI) P Value

Patients No. of Events/ Patients No. of Events/

with Event 100 Patient-Yr with Event 100 Patient-Yr
Primary end point: death from cardiovascular causes, 399 (13.2) 6.3 406 (13.4) 6.4 1.02 (0.891.17) 0.81
nonfatal stroke, nonfatal myocardial infarction,
The

or unstable angina no. (%)

Components of primary end point no./total no. (%)
Death from cardiovascular causes 93/399 (23.3) 88/406 (21.7)
Nonfatal myocardial infarction 247/399 (61.9) 255/406 (62.8)
Nonfatal stroke 49/399 (12.3) 54/406 (13.3)
Unstable angina 10/399 (2.5) 9/406 (2.2)
Patients with each primary end-point event no. (%)*
Death from cardiovascular causes 158 (5.2) 2.4 156 (5.1) 2.3 0.98 (0.781.22) 0.85
Myocardial infarction 261 (8.6) 4.1 270 (8.9) 4.2 1.03 (0.871.22) 0.71
n e w e ng l a n d j o u r na l

Stroke 60 (2.0) 0.9 67 (2.2) 1.0 1.12 (0.791.58) 0.54

The New England Journal of Medicine

of

Unstable angina 10 (0.3) 0.1 11 (0.4) 0.2 1.11 (0.472.62) 0.81

Secondary end points no. (%)
Primary end-point event or hospitalization for heart 469 (15.5) 7.6 456 (15.0) 7.3 0.97 (0.851.10) 0.63

n engl j med 373;23nejm.org December 3, 2015

failure

Copyright 2015 Massachusetts Medical Society. All rights reserved.

m e dic i n e

Primary end-point event, hospitalization for heart 659 (21.7) 11.2 661 (21.8) 11.1 1.00 (0.901.11) 0.96
failure, or revascularization
Additional end points no. (%)
Hospitalization for heart failure 127 (4.2) 1.9 122 (4.0) 1.8 0.96 (0.751.23) 0.75

Downloaded from nejm.org on April 9, 2017. For personal use only. No other uses without permission.
Death from any cause 223 (7.4) 3.3 211 (7.0) 3.1 0.94 (0.781.13) 0.50

* Some patients had more than one component of the primary end point. In the analyses for the separate components, they were included once for each end point they had, regardless of
whether it was their first event.
 Lixisenatide in Type 2 Diabetes and Acute Coronary Syndrome

in the placebo group were taking the volume-

matched equivalent for this maximum dose. 100 20 Hazard ratio, 1.02 (95% CI, 0.891.17)
90
Lixisenatide
Cardiovascular End Points 80 15

The primary end point occurred in 406 patients 70 Placebo

(13.4%) in the lixisenatide group and in 399 10

Patients (%)
60
(13.2%) in the placebo group. For this first- 50
occurrence composite end point in the time-to- 40
5

event analysis, the contributions from each com- 30

ponent event are shown in Table2. The hazard 0
20 0 12 24 36
ratio for the primary end point in the lixisena-
10
tide group as compared with the placebo group
0
was 1.02 (95% confidence interval [CI], 0.89 to 0 12 24 36
1.17); the upper boundary of the 95% confidence Months
interval excluded 1.3 but not 1.0, which showed No. at Risk
the noninferiority of lixisenatide to placebo Placebo 3034 2759 1566 476
(P<0.001) but not superiority (P=0.81) (Fig.1). Lixisenatide 3034 2785 1558 484

Sensitivity analyses that excluded events oc-

Figure 1. KaplanMeier Plot of the First Confirmed Primary End-Point Event.
curring more than 30 days after discontinuation
The composite primary end point in the time-to-event analysis was death
of lixisenatide or placebo and that were adjusted
from cardiovascular causes, nonfatal myocardial infarction, nonfatal stroke,
for baseline imbalances produced similar results or hospitalization for unstable angina. The inset shows the same data on
(Table S2 in the Supplementary Appendix). No an enlarged y axis.
significant study-group interactions were observed
for the primary end point in the prespecified
subgroups or in the post hoc subgroups, includ- panded composite end point resulted in 661 pa-
ing the subgroup defined according to history or tients (21.8%) in the lixisenatide group and 659
no history of heart failure (Fig. S1 in the Supple- (21.7%) in the placebo group having at least one
mentary Appendix). of these adjudicated cardiovascular end points
Regarding the separate cardiovascular com- (hazard ratio, 1.00; 95% CI, 0.90 to 1.11) (Table2).
ponents of the primary end point, the frequency As a separate end point, hospitalization for
of each was similar in the two study groups heart failure occurred in 122 patients (4.0%) in
(Table2). A total of 156 patients in the lixisena- the lixisenatide group and in 127 (4.2%) in the
tide group and 158 in the placebo group died placebo group (hazard ratio, 0.96; 95% CI, 0.75
from cardiovascular causes (hazard ratio, 0.98; to 1.23) (Table2). Death from any cause was
95% CI, 0.78 to 1.22), a total of 270 patients in reported in 211 patients (7.0%) in the lixisena-
the lixisenatide group and 261 in the placebo tide group and in 223 (7.4%) in the placebo
group had a fatal or nonfatal myocardial infarc- group (hazard ratio, 0.94; 95% CI, 0.78 to 1.13)
tion (hazard ratio, 1.03; 95% CI, 0.87 to 1.22), a (Table2). Deaths that were adjudicated as having
total of 67 patients in the lixisenatide group and resulted from noncardiovascular causes oc-
60 in the placebo group had a fatal or nonfatal curred in 46 patients (1.5%) in the lixisenatide
stroke (hazard ratio, 1.12; 95% CI, 0.79 to 1.58), group and in 58 (1.9%) in the placebo group.
and a total of 11 patients in the lixisenatide
group and 10 in the placebo group were hospi- Clinical and Metabolic Effects
talized for unstable angina (hazard ratio, 1.11; Although the management of glucose levels in
95% CI, 0.47 to 2.62). individual patients was at the discretion of the
When hospitalization for heart failure was treating physicians, the reduction in glycated
added to the primary composite end point, 456 hemoglobin values from baseline to week 12 was
patients (15.0%) in the lixisenatide group and significantly greater among patients randomly
469 (15.5%) in the placebo group had an event assigned to lixisenatide than among those as-
in this expanded end point (hazard ratio, 0.97; signed to placebo (0.6 percentage points vs.
95% CI, 0.85 to 1.10). The further addition of 0.2 percentage points, P<0.001). The between-
coronary revascularization procedure to the ex- group difference (lixisenatide minus placebo)

n engl j med 373;23nejm.org December 3, 2015 2253

The New England Journal of Medicine
Downloaded from nejm.org on April 9, 2017. For personal use only. No other uses without permission.
Copyright 2015 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

isenatide group than in the placebo group across

8.0 all visits (P=0.01). This result appeared to be
due predominantly to an increase of 1 beat per
Placebo minute at weeks 2 and 6 that was not detected
Mean Glycated Hemoglobin (%)

7.5
at later time points (Fig. S4 in the Supplemen-
tary Appendix).
Lixisenatide
Although the prespecified analysis of the
percentage change in the urinary albumin-to-
7.0 creatinine ratio (with albumin measured in milli-
grams and creatinine measured in grams) from
baseline to 108 weeks showed a modest differ-
ence in favor of lixisenatide over placebo (24%
6.5 vs. 34%, P=0.004), the median values at baseline
0 (ratio, 10 in each study group) and follow-up
0 12 24 36
(ratio, 12 in the lixisenatide group and 13 in the
Months
placebo group) were clinically similar (Table S3
in the Supplementary Appendix). In a post hoc
Figure 2. Mean Glycated Hemoglobin Level, According to Study Visit.
model that included further adjustment for the
I bars indicate 95% confidence intervals.
glycated hemoglobin level as measured at base-
line and at 3 months after randomization, this
difference was attenuated (P=0.07) (Table S3 in
was sustained, with an average difference across the Supplementary Appendix).
all visits of 0.27 percentage points (95% CI,
0.31 to 0.22; P<0.001) (Fig.2). Hypoglycemic Adverse Events
episodes during the study were reported in 504 Adverse events that led to the permanent discon-
patients (16.6%) in the lixisenatide group and in tinuation of lixisenatide or placebo occurred in
462 (15.2%) in the placebo group (P=0.14). Seri- 347 patients (11.4%) in the lixisenatide group
ous hypoglycemic episodes (requiring assistance and in 217 (7.2%) in the placebo group (P<0.001).
from another person) were numerically less fre- The most common adverse event leading to dis-
quent with lixisenatide (14 patients reporting 16 continuation was a gastrointestinal event, which
events) than with placebo (24 patients reporting was reported in 149 patients (4.9%) in the lix-
37 events). isenatide group and in 37 (1.2%) in the placebo
A modest but significant between-group dif- group (P<0.001) (Table S4 in the Supplementary
ference in the change in body weight from base- Appendix). Within this category, nausea or vom-
line was apparent at 12 weeks (0.6 kg in the iting accounted for most of the between-group
lixisenatide group vs. 0.0 kg in the placebo group, difference in the discontinuation of lixisenatide
P<0.001). This relative weight difference was or placebo, with 91 patients (3.0%) in the lix-
sustained throughout the follow-up period, with isenatide group and 11 (0.4%) in the placebo
an average between-group difference (lixisena- group discontinuing because of nausea and 33
tide minus placebo) across all visits of 0.7 kg (1.1%) in the lixisenatide group and 5 (0.2%) in
(95% CI, 0.9 to 0.5; P<0.001) (Fig. S2 in the the placebo group discontinuing because of
Supplementary Appendix). vomiting (P<0.001 for both comparisons).
A modest relative difference (lixisenatide minus Serious adverse events were reported in 625
placebo) in systolic blood pressure in the lixisena- patients (20.6%) in the lixisenatide group and in
tide group as compared with the placebo group 669 (22.1%) in the placebo group. Serious ad-
was sustained throughout follow-up, with an av- verse events that were attributed to the gastro-
erage difference across all visits of 0.8 mm Hg intestinal system were reported in 66 patients
(95% CI, 1.3 to 0.3) in favor of lixisenatide (2.2%) in the lixisenatide group and in 81 (2.7%)
(P=0.001) (Fig. S3 in the Supplementary Appen- in the placebo group (Table3). While the patients
dix). A small but significant difference in heart were receiving lixisenatide or placebo, pancreati-
rate was observed, with, on average, 0.4 more tis occurred in 5 patients in the lixisenatide
beats per minute (95% CI, 0.1 to 0.6) in the lix- group and in 8 in the placebo group; 3 patients

2254 n engl j med 373;23nejm.org December 3, 2015

The New England Journal of Medicine

Downloaded from nejm.org on April 9, 2017. For personal use only. No other uses without permission.
Copyright 2015 Massachusetts Medical Society. All rights reserved.
 Lixisenatide in Type 2 Diabetes and Acute Coronary Syndrome

in the lixisenatide group and 9 in the placebo Table 3. Serious Adverse Events.*
group received a diagnosis of pancreatic cancer.
Systemic allergic reactions occurred in 27 pa- Placebo Lixisenatide
Event (N=3032) (N=3031)
tients in the lixisenatide group and in 25 in the
placebo group. no. of patients with event (%)
Any event 669 (22.1) 625 (20.6)
Discussion Blood or lymphatic event 14 (0.5) 14 (0.5)

In patients with type 2 diabetes and a recent Cardiac event 107 (3.5) 83 (2.7)
acute coronary syndrome, treatment with the Ear or labyrinth event 4 (0.1) 5 (0.2)
GLP-1receptor agonist lixisenatide, added to Endocrine event 3 (0.1) 2 (0.1)
conventional therapy, was not associated with a Eye event 13 (0.4) 9 (0.3)
significant difference in rates of cardiovascular Gastrointestinal event 81 (2.7) 66 (2.2)
events as compared with conventional therapy
General event 58 (1.9) 64 (2.1)
plus placebo. This conclusion is supported by
additional sensitivity analyses that excluded events Hepatobiliary event 28 (0.9) 36 (1.2)
occurring more than 30 days after the discon- Immune system event 2 (0.1) 4 (0.1)
tinuation of lixisenatide or placebo, as well as by Infection 186 (6.1) 173 (5.7)
a post hoc analysis in which the model was ad- Injury or poisoning 50 (1.6) 44 (1.5)
justed for minor between-group imbalances ob- Investigations 19 (0.6) 10 (0.3)
served at baseline (Table S2 in the Supplemen-
Metabolism or nutrition event 57 (1.9) 33 (1.1)
tary Appendix).
Musculoskeletal event 35 (1.2) 32 (1.1)
This neutral influence of lixisenatide on car-
diovascular end points was reflected in the indi- Neoplasm 61 (2.0) 72 (2.4)
vidual components of the primary composite end Nervous system event 53 (1.7) 47 (1.6)
point and was consistent across multiple pre- Psychiatric event 5 (0.2) 9 (0.3)
specified subgroups. The targeting of a popula- Renal or urinary event 48 (1.6) 48 (1.6)
tion that had type 2 diabetes and a high cardio- Reproductive system event 5 (0.2) 13 (0.4)
vascular risk due to a recent acute coronary
Respiratory or thoracic event 58 (1.9) 58 (1.9)
syndrome with a protocol that permitted the
Skin or subcutaneous tissue event 18 (0.6) 14 (0.5)
adjustment of glucose-lowering therapies in each
group, with the expectation of similar glycemic Social circumstances 0 1 (<0.1)
control, resulted in modest differences in the gly- Surgical or medical procedure 6 (0.2) 6 (0.2)
cated hemoglobin level and allowed us to focus on Vascular event 71 (2.3) 59 (1.9)
cardiovascular actions of the GLP-1receptor
agonist that were unrelated to glycemic factors.26 * Events were assessed with the use of the classifications in the Medical Dictionary
for Regulatory Activities (MedDRA), version 15.0.
Heart failure is a major cardiovascular burden Adjudicated cardiovascular events were not required to be submitted as serious
with severe adverse consequences, including a adverse events unless they were considered by the investigator to be possibly
substantial increase in the risk of death among drug-related.
Investigations included abnormal results on laboratory tests or physical exami
patients with type 2 diabetes.10,28 Particularly nation. The most common MedDRA preferred terms used were alanine amino
worrisome is a meta-analysis of 14 trials of transferase increased (in seven patients in the placebo group and in two in
glucose-lowering therapies that had enrolled the lixisenatide group) and blood glucose increased (in four in the placebo
group and in one in the lixisenatide group).
more than 95,000 patients with type 2 diabetes
and showed that both the peroxisome prolifera-
tor-activated receptor and dipeptidyl peptidase-4 heart failure.30 Our finding of a neutral effect on
inhibitor (DPP-4) classes of glucose-lowering the incidence of hospitalization for heart failure
therapies were associated with increased rates of among patients randomly assigned to lixisena-
heart failure.29 tide, which was consistent in the subgroups of
However, subsequent to that analysis, a recent patients with a history of heart failure and those
major trial showed that the use of sitagliptin without such a history (Table S5 in the Supple-
(another DPP-4 inhibitor) was not associated mentary Appendix), offers new safety informa-
with changes in the rates of hospitalization for tion regarding this GLP-1receptor agonist.

n engl j med 373;23nejm.org December 3, 2015 2255

The New England Journal of Medicine
Downloaded from nejm.org on April 9, 2017. For personal use only. No other uses without permission.
Copyright 2015 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

Mortality among patients with type 2 diabetes, ficient numbers of cardiovascular events were
although higher than in the general population, observed so that we could reasonably exclude a
varies widely depending on cardiovascular and major nonglycemic cardiovascular benefit21 as
renal coexisting conditions.31,32 The overall an- well as an unanticipated harm of lixisenatide.
nualized mortality in our trial was on the higher In conclusion, among patients with type 2
end of the risk spectrum that has been observed diabetes and a recent acute coronary syndrome,
in other clinical trials involving patients with treatment with lixisenatide resulted in rates of
diabetes. Strides have been made to address the major cardiovascular events, including heart fail-
nonglycemic risk factors for these patients.33-36 ure and death from any cause, that were similar
These advances in the standard of care have to those observed with placebo. The neutral car-
increased the difficulty of showing incremental diovascular profile associated with lixisenatide
improvements in cardiovascular prognosis with will inform physicians and patients decisions
intensification of glucose-lowering or lifestyle regarding the use of this agent as an adjunctive
measures.2,37 The few favorable demonstrations therapy to control the glycated hemoglobin level
of reduced rates of cardiovascular events with safely, with no observed augmentation of the
predominantly glucose-lowering therapies have risks of hypoglycemia or pancreatitis.
required more protracted follow-up than routinely
Supported by Sanofi.
occurs in clinical trials.4,7 Although our study Disclosure forms provided by the authors are available with
had only approximately 2 years of follow-up, suf- the full text of this article at NEJM.org.

References
1. UK Prospective Diabetes Study (UKPDS) RR, Pfeffer MA. Heart failure: a cardio- farction undergoing primary percutane-
Group. Intensive blood-glucose control with vascular outcome in diabetes that can no ous coronary intervention: results of
sulphonylureas or insulin compared with longer be ignored. Lancet Diabetes Endo- exenatide myocardial protection in revas-
conventional treatment and risk of com- crinol 2014;2:843-51. cularization study. Arterioscler Thromb
plications in patients with type 2 diabetes 11. Diabetes mellitus evaluating car- Vasc Biol 2013;33:2252-60.
(UKPDS 33). Lancet 1998;352:837-53. diovascular risk in new antidiabetic ther- 19. Nathanson D, Ullman B, Lfstrm U,
2. Gerstein HC, Miller ME, Byington RP, apies to treat type 2 diabetes. Silver et al. Effects of intravenous exenatide in
et al. Effects of intensive glucose lowering Spring, MD:Department of Health and type 2 diabetic patients with congestive
in type 2 diabetes. N Engl J Med 2008;358: Human Services, Center for Drug Evalua- heart failure: a double-blind, randomised
2545-59. tion and Research, 2008. controlled clinical trial of efficacy and
3. Chew EY, Ambrosius WT, Davis MD, 12. European Medicines Agency guide- safety. Diabetologia 2012;55:926-35.
et al. Effects of medical therapies on reti- line on clinical investigation of medicinal 20. Saraiva FK, Sposito AC. Cardiovascular
nopathy progression in type 2 diabetes. products in the treatment or prevention of effects of glucagon-like peptide 1 (GLP-1)
N Engl J Med 2010;363:233-44. diabetes mellitus. London:Committee for receptor agonists. Cardiovasc Diabetol
4. Holman RR, Paul SK, Bethel MA, Medicinal Products for Human Use, 2012. 2014;13:142.
Matthews DR, Neil HA. 10-Year follow-up 13. Hiatt WR, Kaul S, Smith RJ. The car- 21. Smilowitz NR, Donnino R, Schwartz-
of intensive glucose control in type 2 dia- diovascular safety of diabetes drugs in- bard A. Glucagon-like peptide-1 receptor
betes. N Engl J Med 2008;359:1577-89. sights from the rosiglitazone experience. agonists for diabetes mellitus: a role in
5. Gerstein HC, Miller ME, Ismail-Beigi F, N Engl J Med 2013;369:1285-7. cardiovascular disease. Circulation 2014;
et al. Effects of intensive glycaemic con- 14. Drucker DJ. The biology of incretin 129:2305-12.
trol on ischaemic heart disease: analysis hormones. Cell Metab 2006;3:153-65. 22. Fonseca VA, Alvarado-Ruiz R, Raccah
of data from the randomised, controlled 15. Noyan-Ashraf MH, Momen MA, Ban K, D, Boka G, Miossec P, Gerich JE. Efficacy
ACCORD trial. Lancet 2014;384:1936-41. et al. GLP-1R agonist liraglutide activates and safety of the once-daily GLP-1 recep-
6. Turnbull FM, Abraira C, Anderson RJ, cytoprotective pathways and improves tor agonist lixisenatide in monotherapy:
et al. Intensive glucose control and macro- outcomes after experimental myocardial a randomized, double-blind, placebo-
vascular outcomes in type 2 diabetes. Dia- infarction in mice. Diabetes 2009;58:975- controlled trial in patients with type 2
betologia 2009;52:2288-98. 83. diabetes (GetGoal-Mono). Diabetes Care
7. Hayward RA, Reaven PD, Wiitala WL, 16. Timmers L, Henriques JP, de Kleijn 2012;35:1225-31.
et al. Follow-up of glycemic control and DP, et al. Exenatide reduces infarct size 23. Bolli GB, Munteanu M, Dotsenko S,
cardiovascular outcomes in type 2 diabe- and improves cardiac function in a por- et al. Efficacy and safety of lixisenatide
tes. N Engl J Med 2015;372:2197-206. cine model of ischemia and reperfusion once daily vs. placebo in people with type 2
8. Lago RM, Singh PP, Nesto RW. Con- injury. J Am Coll Cardiol 2009;53:501-10. diabetes insufficiently controlled on met-
gestive heart failure and cardiovascular 17. Nikolaidis LA, Mankad S, Sokos GG, formin (GetGoal-F1). Diabet Med 2014;
death in patients with prediabetes and et al. Effects of glucagon-like peptide-1 in 31:176-84.
type 2 diabetes given thiazolidinediones: patients with acute myocardial infarction 24. Riddle MC, Aronson R, Home P, et al.
a meta-analysis of randomised clinical and left ventricular dysfunction after suc- Adding once-daily lixisenatide for type 2
trials. Lancet 2007;370:1129-36. cessful reperfusion. Circulation 2004;109: diabetes inadequately controlled by es-
9. Goldfine AB. Assessing the cardiovas- 962-5. tablished basal insulin: a 24-week, ran-
cular safety of diabetes therapies. N Engl 18. Woo JS, Kim W, Ha SJ, et al. Cardio- domized, placebo-controlled comparison
J Med 2008;359:1092-5. protective effects of exenatide in patients (GetGoal-L). Diabetes Care 2013;36:2489-
10. McMurray JJ, Gerstein HC, Holman with ST-segment-elevation myocardial in- 96.

2256 n engl j med 373;23nejm.org December 3, 2015

The New England Journal of Medicine

Downloaded from nejm.org on April 9, 2017. For personal use only. No other uses without permission.
Copyright 2015 Massachusetts Medical Society. All rights reserved.
 Lixisenatide in Type 2 Diabetes and Acute Coronary Syndrome

25. Meier JJ. GLP-1 receptor agonists for Chatterjee S, Farkouh ME, Scirica BM. 34. Griffin SJ, Borch-Johnsen K, Davies MJ,
individualized treatment of type 2 diabe- Glucose-lowering drugs or strategies and et al. Effect of early intensive multifactorial
tes mellitus. Nat Rev Endocrinol 2012;8: cardiovascular outcomes in patients with therapy on 5-year cardiovascular outcomes
728-42. or at risk for type 2 diabetes: a meta-analy- in individuals with type 2 diabetes de-
26. Bentley-Lewis R, Aguilar D, Riddle MC, sis of randomised controlled trials. Lan- tected by screening (ADDITION-Europe):
et al. Rationale, design, and baseline char- cet Diabetes Endocrinol 2015;3:356-66. a cluster-randomised trial. Lancet 2011;
acteristics in Evaluation of LIXisenatide 30. Green JB, Bethel MA, Armstrong PW, 378:156-67.
in Acute Coronary Syndrome, a long-term et al. Effect of sitagliptin on cardiovascu- 35. American Diabetes Association. Stan-
cardiovascular end point trial of lixisena- lar outcomes in type 2 diabetes. N Engl J dards of medical care in diabetes 2012.
tide versus placebo. Am Heart J 2015; Med 2015;373:232-42. Diabetes Care 2012;35:Suppl 1:S11-63.
169(5):631-638.e7. 31. Seshasai SR, Kaptoge S, Thompson A, 36. Fox CS, Golden SH, Anderson C, et al.
27. Yang F, Stewart M, Ye J, DeMets D. et al. Diabetes mellitus, fasting glucose, Update on prevention of cardiovascular
Type 2 diabetes mellitus development and risk of cause-specific death. N Engl J disease in adults with type 2 diabetes
programs in the new regulatory environ- Med 2011;364:829-41. mellitus in light of recent evidence: a sci-
ment with cardiovascular safety require- 32. Barkoudah E, Skali H, Uno H, Solo- entific statement from the American Heart
ments. Diabetes Metab Syndr Obes 2015; mon SD, Pfeffer MA. Mortality rates in Association and the American Diabetes
8:315-25. trials of subjects with type 2 diabetes. Association. Circulation 2015;132:691-718.
28. Jhund PS, McMurray JJV, Chaturvedi N, J Am Heart Assoc 2012;1:8-15. 37. Wing RR, Bolin P, Brancati FL, et al.
et al. Mortality following a cardiovascular 33. Ali MK, Bullard KM, Saaddine JB, Cardiovascular effects of intensive lifestyle
or renal event in patients with type 2 dia- Cowie CC, Imperatore G, Gregg EW. intervention in type 2 diabetes. N Engl J
betes in the ALTITUDE trial. Eur Heart J Achievement of goals in U.S. diabetes Med 2013;369:145-54.
2015;36:2463-9. care, 19992010. N Engl J Med 2013;368: Copyright 2015 Massachusetts Medical Society.
29. Udell JA, Cavender MA, Bhatt DL, 1613-24.

n engl j med 373;23nejm.org December 3, 2015 2257

The New England Journal of Medicine
Downloaded from nejm.org on April 9, 2017. For personal use only. No other uses without permission.
Copyright 2015 Massachusetts Medical Society. All rights reserved.