Articles

Efficacy and safety of a novel dual GIP and GLP-1 receptor

agonist tirzepatide in patients with type 2 diabetes
(SURPASS-1): a double-blind, randomised, phase 3 trial
Julio Rosenstock, Carol Wysham, Juan P Frías, Shizuka Kaneko, Clare J Lee, Laura Fernández Landó, Huzhang Mao, Xuewei Cui,
Chrisanthi A Karanikas, Vivian T Thieu

Summary
Background Despite advancements in care, many people with type 2 diabetes do not meet treatment goals; thus, Lancet 2021; 398: 143–55
development of new therapies is needed. We aimed to assess efficacy, safety, and tolerability of novel dual glucose- Published Online
dependent insulinotropic polypeptide and GLP-1 receptor agonist tirzepatide monotherapy versus placebo in people June 26, 2021
https://doi.org/10.1016/
with type 2 diabetes inadequately controlled by diet and exercise alone.
S0140-6736(21)01324-6
See Comment page 95
Methods We did a 40-week, double-blind, randomised, placebo-controlled, phase 3 trial (SURPASS-1), at
Dallas Diabetes Research Center
52 medical research centres and hospitals in India, Japan, Mexico, and the USA. Adult participants (≥18 years) were at Medical City, Dallas, TX, USA
included if they had type 2 diabetes inadequately controlled by diet and exercise alone and if they were naive to injectable (J Rosenstock MD); MultiCare
diabetes therapy. Participants were randomly assigned (1:1:1:1) via computer-generated random sequence to once a week Rockwood Clinic, Spokane, WA,
tirzepatide (5, 10, or 15 mg), or placebo. All participants, investigators, and the sponsor were masked to treatment USA (C Wysham MD); National
Research Institute, Los Angeles,
assignment. The primary endpoint was the mean change in glycated haemoglobin (HbA1c) from baseline at 40 weeks. CA, USA (J P Frías MD); Takatsuki
This study is registered with ClinicalTrials.gov, NCT03954834. Red Cross Hospital, Osaka,
Japan (S Kaneko MD); Eli Lilly
and Company, Indianapolis, IN,
Findings From June 3, 2019, to Oct 28, 2020, of 705 individuals assessed for eligibility, 478 (mean baseline HbA1c 7∙9%
USA (C J Lee MD,
[63 mmol/mol], age 54∙1 years [SD 11∙9], 231 [48%] women, diabetes duration 4∙7 years, and body-mass L Fernández Landó MD,
index 31∙9 kg/m²) were randomly assigned to tirzepatide 5 mg (n=121 [25%]), tirzepatide 10 mg (n=121 [25%]), H Mao PhD, X Cui PhD,
tirzepatide 15 mg (n=121 [25%]), or placebo (n=115 [24%]). 66 (14%) participants discontinued the study drug and C A Karanikas MS, V T Thieu PhD)
50 (10%) discontinued the study prematurely. At 40 weeks, all tirzepatide doses were superior to placebo for changes Correspondence to:
from baseline in HbA1c, fasting serum glucose, bodyweight, and HbA1c targets of less than 7∙0% (<53 mmol/mol) and Dr Julio Rosenstock, Dallas
Diabetes Research Center at
less than 5∙7% (<39 mmol/mol). Mean HbA1c decreased from baseline by 1∙87% (20 mmol/mol) with tirzepatide Medical City, Dallas, TX 75230,
5 mg, 1∙89% (21 mmol/mol) with tirzepatide 10 mg, and 2∙07% (23 mmol/mol) with tirzepatide 15 mg versus +0∙04% USA
with placebo (+0∙4 mmol/mol), resulting in estimated treatment differences versus placebo of –1∙91% (–21 mmol/mol) juliorosenstock@
dallasdiabetes.com
with tirzepatide 5 mg, –1∙93% (–21 mmol/mol) with tirzepatide 10 mg, and –2∙11% (–23 mmol/mol) with tirzepatide
15 mg (all p<0∙0001). More participants on tirzepatide than on placebo met HbA1c targets of less than 7∙0%
(<53 mmol/mol; 87–92% vs 20%) and 6∙5% or less (≤48 mmol/mol; 81–86% vs 10%) and 31–52% of patients on
tirzepatide versus 1% on placebo reached an HbA1c of less than 5∙7% (<39 mmol/mol). Tirzepatide induced a dose-
dependent bodyweight loss ranging from 7∙0 to 9∙5 kg. The most frequent adverse events with tirzepatide were mild
to moderate and transient gastrointestinal events, including nausea (12–18% vs 6%), diarrhoea (12–14% vs 8%), and
vomiting (2–6% vs 2%). No clinically significant (<54 mg/dL [<3 mmol/L]) or severe hypoglycaemia were reported
with tirzepatide. One death occurred in the placebo group.

Interpretation Tirzepatide showed robust improvements in glycaemic control and bodyweight, without increased risk
of hypoglycaemia. The safety profile was consistent with GLP-1 receptor agonists, indicating a potential monotherapy
use of tirzepatide for type 2 diabetes treatment.

Funding Eli Lilly and Company.

Copyright © 2021 Elsevier Ltd. All rights reserved.

Introduction manner. It is the dominant incretin hormone in

GLP-1 receptor agonists deliver substantial improve­ healthy individuals, contributing the majority of the
ments in glucose control, and some, especially once a incretin effect,3 but the insulin response after GIP
week GLP-1 receptor agonists dulaglutide and secretion in type 2 diabetes is diminished.4 GIP differs
semaglutide, can induce clinically significant but variable from GLP-1 in its role of stimulating glucagon secre­
weight loss.1,2 Similar to GLP-1, glucose-dependent tion during hypoglycaemia and exerting a direct effect on
insulinotropic polypeptide (GIP) also enhances meal- lipid homoeostasis.2,5,6 GIP might also promote weight
stimulated insulin secretion in a glucose-dependent loss by signalling satiety through its receptors present in

www.thelancet.com Vol 398 July 10, 2021 143

 Articles

Research in context
Evidence before this study whom 20% reached less than 7∙0%, 10% reached 6∙5% or less,
We searched PubMed on Sept 10, 2020, using the terms and 1% reached normoglycaemia. Notably, HbA1c targets were
“tirzepatide”, “liraglutide”, “exenatide”, “lixisenatide”, reached with tirzepatide without severe or clinically significant
“dulaglutide”, “albiglutide”, “semaglutide”, “glucose-dependent cases of hypoglycaemia. Significant bodyweight loss of
insulinotropic polypeptide”, “glucagon-like peptide 1 receptor –7·0 kg to –9·5 kg was observed with tirzepatide treatment
agonist”, and “type 2 diabetes” with no date or study duration (vs –0∙7 kg with placebo), with 67–78%, 31–47%, and 13–27% of
restrictions. Non-English references were excluded. participants given tirzepatide compared with 14%, 1%,
We designed this study based on the evidence obtained from and 0% of participants given placebo reaching bodyweight
the clinical development of tirzepatide, including in-vitro and reductions of 5% or greater, 10% or greater, and 15% or greater,
preclinical studies, phase 1 single-dose and multi-dose respectively. The effect of tirzepatide on bodyweight loss was
pharmacokinetic and pharmacodynamic studies, a phase 2 dose-dependent and progressive, and did not plateau at the end
efficacy and safety trial, and a phase 2 dose-escalation trial. of the 40-week treatment period, similar to the safety and
Although there are several long-acting GLP-1 receptor agonists tolerability profile of selective GLP-1 receptor agonists. The most
approved for the treatment of type 2 diabetes, tirzepatide is a frequent adverse events with tirzepatide versus placebo were
novel once a week dual GIP and GLP-1 receptor agonist that mild to moderate gastrointestinal events, including nausea
integrates the actions of both incretins into a single molecule, (12–18% vs 6%), diarrhoea (12–14% vs 8%), and vomiting
representing a promising medication for type 2 diabetes (2–6% vs 2%), which decreased over time.
treatment.
Implications of all the available evidence
Added value of this study To our knowledge, SURPASS-1 is the first randomised
Findings from this study showed that 40 weeks of tirzepatide controlled trial of tirzepatide, a once a week dual GIP and
treatment resulted in robust and clinically significant reductions GLP-1 receptor agonist, which showed potent glucose-lowering
in glycated haemoglobin (HbA1c) at all doses compared with effects towards near-normal ranges with robust weight loss of a
placebo. Furthermore, all three doses of tirzepatide led to magnitude not previously reported in people with type 2
significantly greater improvements in fasting serum glucose as diabetes and without increased risk of severe or clinically
early as 4 weeks and postprandial self-monitored blood glucose significant hypoglycaemia (<54 mg/dL). The advantageous
profiles as early as 12 weeks. Most patients given tirzepatide therapeutic effects for glycaemic control and bodyweight
reached HbA1c targets, with 87–92% reaching an HbA1c reduction will be further assessed in the SURPASS clinical trial
concentration of less than 7∙0% and 81–86% reaching programme when compared with the selective GLP-1 receptor
6∙5% or less, and 31–52% reaching normoglycaemia agonist semaglutide in people with type 2 diabetes.
(HbA1c <5∙7%), compared with participants given placebo in

the hypothalamus, as shown in preclinical animal study in type 2 diabetes, tirzepatide significantly reduced
models.7,8 glycated haemoglobin (HbA1c) and bodyweight compared
Notably, in preclinical studies, the combined effect of with placebo and the selective GLP-1 receptor agonist
GIP and GLP-1 has shown robust weight loss effects dulaglutide 1·5 mg, with or without metformin.10 More
independent of insulin sensitivity and lipid metabolism.9 gastrointestinal events and decreased appetite of mild to
In diet-induced obese mice, combined GIP and GLP-1 moderate severity were reported with tirzepatide than
agonism showed greater anorectic action compared with with placebo.10 More patients reported gastrointestinal
semaglutide by enhancing satiety and satiation, reducing events and decreased appetite with tirzepatide 15 mg
preference for high-fat diets, and lowering sweet taste than with dulaglutide 1·5 mg.10 To further improve
preference.8 In addition, combined agonism of GIP and gastrointestinal tolerability, we adopted an optimised
GLP-1 receptors has shown greater improvements in dose-escalation scheme in the phase 3 SURPASS clinical
glycaemia and bodyweight than selective GLP-1 receptor programme, with a lower starting dose of tirzepatide and
agonists in clinical studies, and thus it serves as the slower monthly dose escalation.12
rationale for targeting both incretin hormones to treat In this trial, we aimed to assess the glycaemic efficacy
type 2 diabetes.10 and safety of three doses of once a week tirzepatide (5, 10,
Tirzepatide is a novel once a week dual GIP and and 15 mg) as monotherapy versus placebo in people
GLP-1 receptor agonist. It is a 39-amino acid synthetic with type 2 diabetes.
peptide with agonist activity at both GIP and GLP-1
receptors.11 Its structure is based on the GIP sequence and Methods
includes a C20 fatty di-acid moiety, helping with half-life Study design and participants
extension (half-life of around 5 days) allowing once a week This 40-week, multicentre, randomised, double-blind,
subcutaneous administration.11 In a 26-week phase 2B placebo-controlled, parallel group trial (SURPASS-1) was

144 www.thelancet.com Vol 398 July 10, 2021

 Articles

done in 52 medical research centres and hospitals in Participants given tirzepatide followed a slow dose
India, Japan, Mexico, and the USA. Key inclusion criteria escalation regimen fixed at 2∙5 mg-dose increments of
included adult participants (≥18 years) with type 2 diabetes tirzepatide every 4 weeks until the maintenance dose was
inadequately controlled with diet and exercise alone reached. The maintenance doses of 5, 10, and 15 mg were
and who were naive to injectable diabetes therapy. achieved at 4, 12, and 20 weeks in respective tirzepatide
Eligible participants had an HbA1c of 7∙0% or more groups. Vital sign measurements were assessed at
(≥53 mmol/mol) to 9∙5% or less (≤80 mmol/mol) at the fasting state, except for visit one, and taken before
screening, body-mass index (BMI) of 23 kg/m² or greater, obtaining an electrocardiogram tracing and before
and with stable weight (no change outside of 5%) during collection of blood samples for laboratory testing, at visits
the previous 3 months with agreement to not initiate a when required. The participant sat quietly for 5 min
diet or exercise programme during the study with the before blood pressure and heart rate measurements were
intent of reducing bodyweight other than the lifestyle and taken. For each parameter, each measure was assessed
dietary measures for diabetes treatment. Additionally, twice using the same arm and the recordings were taken
per the protocol, sites needed to educate participants at least 1 min apart. Blood pressure was taken with an
on diabetes management including management of automated blood pressure machine. Initiation of new
hypoglycaemia, hyperglycaemia, and general counselling antihyperglycaemic medications was allowed according
on diet and exercise. Key exclusion criteria included type 1 to specific rescue criteria: (1) for protocol-defined
diabetes, history of pancreatitis, history of proliferative persistent hyperglycaemia; (2) in those participants who
diabetic retinopathy, diabetic maculopathy, or non- require permanent discontinuation of study drug, but
proliferative diabetic retinopathy that requires acute remain in the study; or (3) during the safety follow-up
treatment, estimated glomerular filtration rate of less period, if the participants needed additional glycaemic
than 30 mL/min per 1∙73 m², and use of any oral control. Additional details on the initiation of new
antihyperglycaemia medication for 3 months before antihyperglycaemic medications are described in the
screening. A full list of eligibility criteria is provided in appendix (pp 3–4).
the appendix (pp 1–3). See Online for appendix
The protocol was approved by local institutional review Outcomes
boards and the trial was done in accordance with the The primary endpoint was mean change from baseline in
Declaration of Helsinki guidelines on good clinical practice. HbA1c at 40 weeks. Key secondary endpoints, included in
All participants provided written informed consent. the type 1 error control at 40 weeks (appendix pp 6, 17),
were mean change from baseline in fasting serum
Randomisation and masking glucose, the proportion of participants with HbA1c target
Participants were randomly assigned (1:1:1:1) to receive values of less than 7∙0% (<53 mmol/mol) and less
once a week tirzepatide (5, 10, or 15 mg) or volume- than 5∙7% (<39 mmol/mol), and mean change from
matched placebo in a single dose pen. All pens were baseline in bodyweight. Other efficacy endpoints were
similar in appearance. Assignment to treatment group the proportion of participants with an HbA1c target of
was determined by a computer-generated random 6∙5% or less (≤48 mmol/mol), proportions reaching
sequence using the Eli Lilly and Company interactive 5% or greater, 10% or greater, and 15% or greater weight
web-response system. This system is externally validated loss, and mean change from baseline in daily mean seven-
and compliant with the Code of Federal Regulations 21 point self-monitored blood glucose (SMBG) profiles at
part 11. 40 weeks.
Participants were stratified at randomisation based Safety endpoints were treatment-emergent adverse
on country, baseline HbA1c (≤8∙5% [≤69 mmol/mol] or events; study drug discontinuation due to adverse events;
>8∙5% [>69 mmol/mol]) and previous use of any oral adjudicated pancreatic adverse events (appendix p 4);
antihyperglycaemia medication (yes or no). All partici­ serum calcitonin, allergic and hypersensitivity reactions,
pants, investigators, and the sponsor were masked to and treatment-emergent anti­ drug antibodies for
treatment assignment. The randomisation data was only tirzepatide; mean changes from baseline in pulse rate,
assessed by a very limited group of personnel (interactive systolic, and diastolic blood pressure; hypoglycaemia
web-response system support team, Data Movement events of blood glucose less than 70 mg/dL (<4 mmol/L)
Group associates, demand forecasters and clinical and clinically significant concentrations less than
supply coordinators, and unmasked analysts for expedite 54 mg/dL (<3 mmol/L); and severe hypoglycaemia.
pharmacokinetic and pharmacodynamic analysis).
Statistical analysis
Procedures The sample size calculation assumed at least –0·65% dif­
After a 3-week screening or lead-in period, participants ference of mean change from baseline in HbA1c between
received subcutaneous injections of tirzepatide or tirzepatide groups and placebo, a common SD of 1∙1%
matched placebo once a week for 40 weeks, followed by a up to 25%, and 35% dropout rate in respective tirzepatide
4-week safety follow-up period (appendix p 16). and placebo groups. It was estimated that 472 participants

www.thelancet.com Vol 398 July 10, 2021 145

 Articles

provided at least 90% power to establish superiority for a proportion of patients with HbA1c target values of less
tirzepatide dose compared with placebo at a two-sided than 5∙7% (39 mmol/mol) at 40 weeks. This procedure
significance level of 0∙0167. Detailed type 1 error control was repeated to test superiority of primary and key
strategy is shown in the appendix (pp 6, 17). secondary endpoints for the tirzepatide 10 mg group
Efficacy analyses were done in the modified intention- versus placebo comparison and tirzepatide 5 mg group
to-treat population, comprised of all randomly assigned versus placebo comparison. Safety analyses were done
participants who were exposed to at least one dose of on the modified intention-to-treat population with all
study drug. Participants who discontinued study drug data from start of treatment to end of safety follow-up.
due to inadvertent enrolment were excluded from Statistical analyses were done using SAS version 9.4,
efficacy analysis. Analyses of efficacy measures to align unless otherwise specified. Additional statistical analyses
with the efficacy estimand, representing on-treatment methods are provided in the appendix (pp 4–6). This
efficacy without the influence of rescue therapy, was study is registered with ClinicalTrials.gov, NCT03954834.
done using the efficacy analysis set. In addition, primary
and key secondary objectives were assessed to align with Role of the funding source
the treatment-regimen estimand, representing efficacy The funder of the study had a role in study design, data
regardless of premature study drug discontinuation and collection, data analysis, data interpretation, and writing
use of rescue medication, using the full analysis set data of the report.
and with missing endpoint values imputed by multiple
imputation based on the placebo group. Type 1 error rate Results
was controlled at a level of 0·05 within each estimand for Between June 3, 2019, and Oct 28, 2020, 705 participants
evaluation of primary and key secondary objectives via were assessed for eligibility; 478 participants (mean
a graphical testing approach (appendix pp 6, 17). The baseline HbA1c 7∙94% [63 mmol/mol], age 54∙1 years
graphical testing scheme started with testing the [SD 11∙9], 231 [48%] women, diabetes duration 4∙7 years,
superiority of mean change from baseline to 40 weeks in and BMI 31∙9 kg/m²) were randomly assigned and
HbA1c for the tirzepatide 15 mg group versus placebo received at least one dose of tirzepatide 5 mg
at a significance level of 0·0167, followed by testing (n=121 [25%]), tirzepatide 10 mg (n=121 [25%]),
superiority of key secondary endpoints for the tirzepatide tirzepatide 15 mg (n=121 [25%]), or placebo (n=115 [24%];
15 mg group versus placebo in the order of mean figure 1). 66 (14%) participants discontinued the study
change from baseline to 40 weeks in bodyweight, drug and 50 (10%) discontinued the study prematurely.
proportion of participants with HbA1c target values of less Time to study drug discontinuation and time to study
than 7∙0% (53 mmol/mol) at 40 weeks, mean change discon­tinuation are shown in Kaplan-Meier plots in the
from baseline to 40 weeks in fasting serum glucose, and appendix (p 18). Reasons for premature study drug

705 assessed for eligibility

227 excluded
204 screen failure
2 lost to follow-up
1 physician decision
4 other
16 withdrawal by participant

478 randomised

121 assigned to tirzepatide 5 mg 121 assigned to tirzepatide 10 mg 121 assigned to tirzepatide 15 mg 115 assigned to placebo

11 discontinued study 12 discontinued study 26 discontinued study 17 discontinued study

treatment treatment treatment treatment
7 discontinued study 9 discontinued study 18 discontinued study 16 discontinued study

110 completed treatment 109 completed treatment 95 completed treatment 98 completed treatment
114 completed study 112 completed study 103 completed study 99 completed study

Figure 1: Trial profile

Reasons for study discontinuation up until safety follow-up were death, loss to follow-up, withdrawal by participant, or other.

146 www.thelancet.com Vol 398 July 10, 2021

 Articles

Tirzepatide 5 mg Tirzepatide 10 mg Tirzepatide 15 mg Placebo Total

(n=121) (n=121) (n=121) (n=115) (n=478)
Age, years 54∙1 (11∙9) 55∙8 (10∙4) 52∙9 (12∙3) 53∙6 (12∙8) 54∙1 (11∙9)
Sex
Women 65 (54%) 49 (40%) 58 (48%) 59 (51%) 231 (48%)
Men 56 (46%) 72 (60%) 63 ( 52%) 56 (49%) 247 (52%)
Race
American Indian or Alaska Native 31 (26%) 31 (26%) 30 (25%) 26 (23%) 118 (25%)
Asian 45 (37%) 43 (36%) 42 (35%) 38 (33%) 168 (35%)
Black or African American 7 (6%) 4 (3%) 6 (5%) 5 (4%) 22 (5%)
White 38 (31%) 43 (36%) 43 (36%) 46 (40%) 170 (36%)
Ethnicity
Hispanic or Latino 50 (41%) 54 (45%) 55 (45%) 48 (42%) 207 (43%)
Not Hispanic or Latino 48 (40%) 49 (40%) 42 (35%) 45 (39%) 184 (38%)
Not reported 23 (19%) 18 (15%) 24 (20%) 22 (19%) 87 (18%)
Country
India 18 (15%) 20 (17%) 18 (15%) 17 (15%) 73 (15%)
Japan 23 (19%) 22 (18%) 23 (19%) 21 (18%) 89 (19%)
Mexico 42 (35%) 40 (33%) 42 (35%) 40 (35%) 164 (34%)
USA 38 (31%) 39 (32%) 38 (31%) 37 (32%) 152 (32%)
HbA1c concentration
In percentage 7∙97 (0∙84) 7∙90 (0∙78) 7∙85 (1∙02) 8∙05 (0∙80) 7∙94 (0∙87)
In mmol/mol 63∙59 (9∙19) 62∙86 (8∙50) 62∙27 (11∙11) 64∙52 (8∙75) 63∙29 (9∙46)
≤8∙5% 95 (79%) 98 (81%) 98 (81%) 87 (76%) 378 (79%)
>8∙5% 26 (21%) 23 (19%) 23 (19%) 28 (24%) 100 (21%)
Fasting serum glucose concentration
In mg/dL 153∙7 (37∙3) 152∙6 (41∙7) 153∙3 (40∙4) 154∙8 (40∙3) 153∙6 (39∙8)
In mmol/L 8∙5 (2∙1) 8∙5 (2∙3) 8∙5 (2∙2) 8∙6 (2∙2) 8∙5 (2∙2)
Body-mass index, kg/m² 32∙2 (7∙0) 32∙2 (7∙6) 31∙5 (5∙5) 31∙7 (6∙1) 31∙9 (6∙6)
Weight, kg 87∙0 (21∙2) 86∙2 (19∙5) 85∙4 (18∙5) 84∙8 (20∙0) 85∙9 (19∙8)
eGFR, CKD-EPI calculation, mL/min 94∙7 (20∙6) 92∙1 (18∙2) 96∙2 (19∙7) 93∙4 (20∙2) 94∙1 (19∙7)
per 1∙73 m²
Diabetes duration, years 4∙6 (5∙1) 4∙9 (5∙6) 4∙8 (5∙0) 4∙5 (5∙9) 4∙7 (5∙4)
Previous oral antihyperglycaemic 55 (45%) 53 (44%) 56 (46%) 55 (48%) 219 (46%)
medication use
Systolic blood pressure, mm Hg 128∙2 (15∙7) 127∙8 (12∙6) 126∙8 (13∙8) 127∙8 (14∙1) 127∙6 (14∙1)
Diastolic blood pressure, mm Hg 79∙9 (9∙0) 78∙7 (8∙2) 79∙2 (8∙8) 79∙7 (9∙3) 79∙4 (8∙8)
Pulse rate, beats per min 72∙7 (9∙3) 73∙0 (9∙8) 74∙3 (8∙3) 74∙9 (10∙0) 73∙7 (9∙4)
Data are mean (SD) or n (%). Data are for the all-randomised population. n=all randomly assigned participants who took at least one dose of study drug (modified intention-
to-treat population). CKD-EPI=chronic kidney disease-epidemiology. eGFR=estimated glomerular filtration rate. HbA1c=glycated haemoglobin.

Table 1: Baseline characteristics

discontinuation were adverse event, death, did not meet Demographics and clinical characteristics were similar
randomisation criteria, loss to follow-up, physician between groups (table 1). The overall mean duration
decision, protocol deviation, withdrawal by participant, of diabetes was 4∙7 years with a mean HbA1c of 7∙94%
or other. A greater number of study drug discontinua­ (63 mmol/mol) and BMI of 31∙9 kg/m². 54% of partici­
tions were observed with tirzepatide 15 mg, and most pants had no previous use of oral antihyperglycaemia
discontinuations in the tirzepatide 15 mg group were medication, which was similarly distributed across
for reasons other than adverse events. Three participants groups.
(tirzepatide 15 mg [n=1], placebo [n=2]) discontinued Data presented are based on the efficacy analysis set.
study drug due to inadvertent enrolment for violation At 40 weeks, HbA1c significantly decreased from baseline
of exclusion criteria related to gastric abnormalities by a least squares mean (LSM) of –1∙87% (SE 0∙09;
or retinopathy. All other participants who were –20 [1 mmol/mol]) with tirzepatide 5 mg, –1∙89%
inadvertently enrolled continued in the study on study (0∙10; –21 [1 mmol/mol]) with tirzepatide 10 mg, and
drug. –2∙07% (0∙10; –23 [1 mmol/mol]) with tirzepatide 15 mg,

www.thelancet.com Vol 398 July 10, 2021 147

 Articles

versus +0∙04% (SE 0∙11; +0∙4 [1 mmol/mol]) with 5 mg, –7∙8 kg (0∙53) with tirzepatide 10 mg, and
placebo (figure 2A, B, table 2). Estimated mean treat­ –9∙5 kg (0∙54) with tirzepatide 15 mg, versus
ment differences versus placebo were –1∙91% (95% CI –0∙7 kg (0∙57) with placebo (figure 3A, C, table 2).
–2∙18 to –1∙63; –21 mmol/mol [–24 to –18]) with Estimated mean treatment differences versus placebo
tirzepatide 5 mg, –1∙93% (–2∙21 to –1∙65; –21 mmol/mol were –6∙3 kg (95% CI –7∙8 to –4∙7) with tirzepatide
[–24 to –18]) with tirzepatide 10 mg, and –2∙11% 5 mg, –7∙1 (–8∙6 to –5∙5) with tirzepatide 10 mg, and
(–2∙39 to –1∙83; –23 mmol/mol [–26 to –20]) with –8∙8 kg (–10∙3 to –7∙2) with tirzepatide 15 mg (all
tirzepatide 15 mg (all p<0∙0001). Significant decreases p<0∙0001). The reduction in weight was observed by
in HbA1c with tirzepatide versus placebo were evident week 4 and did not plateau by week 40 (figure 3A, C).
by the first assessment at week 4 (figure 2A). Mean The effect of tirzepatide on bodyweight was progressive
HbA1c concentrations reached near-normal values of and dose dependent. A greater proportion of par­ticipants
6∙08% (43 mmol/mol) with tirzepatide 5 mg, 6∙06% had bodyweight reductions of 5% or greater (67–78%),
(43 mmol/mol) with tirzepatide 10 mg, and 5∙88% 10% or greater (31–47%), and 15% or greater (13–27%)
(41 mmol/mol) with tirzepatide 15 mg at week 40 with tirzepatide versus 14%, 1%, and 0% with placebo
(figure 2A). At 40 weeks, previous use of oral anti- (p≤0∙011, all tirzepatide groups; figure 3D, table 2).
hyperglycaemic medication did not affect HbA1c change Similarly, mean BMI and waist circumference reduced
from baseline with tirzepatide (treatment by oral anti- significantly from baseline with tirzepatide versus
hyperglycaemic medication status interaction p=0∙33; placebo (p<0∙0001 all tirzepatide groups; appendix p 22).
appendix p 20). At 40 weeks, total cholesterol, triglycerides, and
A significantly higher proportion of participants given serum VLDL cholesterol decreased from baseline and
tirzepatide reached HbA1c target concentrations of less HDL cholesterol increased from baseline with all
than 7∙0% (<53 mmol/mol), 6∙5% or less (≤48 mmol/mol), tirzepatide doses versus placebo (figure 3E, appendix
and less than 5∙7% (<39 mmol/mol) at 40 weeks versus pp 7–8). LDL cholesterol decreased with tirzepatide
placebo (p<0∙0001, all tirzepatide groups; figure 2F). An 15 mg versus placebo. Homoeostatic model assessment
HbA1c concentration of less than 7∙0% (<53 mmol/mol) of insulin resistance decreased with all tirzepatide
was reached in 87–92% of participants with tirzepatide doses versus placebo (appendix p 9). No significant
versus 19% with placebo, an HbA1c concentration of 6∙5% differences were seen in study drug discontinuation due
or less (≤48 mmol/mol) was reached in 81–86% of to adverse events across treatment groups (table 3). The
participants with tirzepatide versus 10% with placebo, and proportion of participants reporting any adverse events
an HbA1c concentration of less than 5∙7% (<39 mmol/mol) and the number of reported serious adverse events
was reached in 31–50% of participants with tirzepatide were similar between groups (table 3). Study drug
versus 1% with placebo. discontinuations due to gastrointestinal adverse events
At 40 weeks, fasting serum glucose significantly were 2–7% with tirzepatide versus 1% with placebo.
decreased from baseline by an LSM of –43∙6 mg/dL The most frequent adverse events with tirzepatide were
(SE 3∙4; –2 [0∙2 mmol/L]) with tirzepatide 5 mg, gastrointestinal. Nausea was reported in 14 (12%)
–45∙9 mg/dL (3∙5; –3 [0∙2 mmol/L]) with tirzepatide participants who received tirzepatide 5 mg, 16 (13%) who
10 mg, and –49∙3 mg/dL (3∙6; –3 [0∙2 mmol/L]) with received tirzepatide 10 mg, and 22 (18%) who received
tirzepatide 15 mg, versus +12∙9 mg/dL (4∙0; +1 tirzepatide 15 mg, versus seven (6%) participants who
[0∙2 mmol/L]) with placebo (figure 2D, E, table 2). received placebo (table 3). Diarrhoea was reported in
Estimated mean treatment differences versus placebo 14 (12%) participants in each group who received
were –56∙5 mg/dL (95% CI –66∙8 to –46∙1; –3 mmol/L tirzepatide 5 mg and 15 mg, and in 17 (14%) partici­
[–4 to –3]) with tirzepatide 5 mg, –58∙8 mg/dL pants who received tirzepatide 10 mg, versus
(–69∙2 to –48∙4; –3 mmol/L [–4 to –3]) with tirzepatide nine (8%) participants who received placebo. Vomiting
10 mg, and –62∙1 mg/dL (–72∙7 to –51∙5; –3 mmol/L was reported in four (3%) participants who received
[–4 to –3]) with tirzepatide 15 mg (all p<0∙0001). The tirzepatide 5 mg, three (2%) who received tirzepatide
seven-point SMBG profiles showed significant lowering 10 mg, and seven (6%) who received tirzepatide 15 mg,
of the mean daily, pre-meal, and 2-h post-meal glucose versus two (2%) participants who received placebo. Most
values with tirzepatide versus placebo at 12 weeks and reports of nausea, vomiting, and diarrhoea were mild to
40 weeks (p<0∙0001, all tirzepatide groups at both time­ moderate in severity and decreased over time in all groups
points; figure 2F, appendix p 21). Decreases in mean 2-h (appendix pp 10–12, 23–24).
post-meal glucose values ranged from –61 to –65 mg/dL One death occurred in a participant randomly assigned
(–3 to –4 mmol/L) with tirzepatide versus –11 (–1 mmol/L) to placebo due to myocardial infarction confirmed by
with placebo. Mean post-meal glucose values in par­ adjudication (table 2).
ticipants given tirzepatide were less than 140 mg/dL Clinically significant (<54 mg/dL [<3 mmol/L]) or
(<8 mmol/L) based on SMBG profiles (figure 2F). severe hypoglycaemia was not reported in patients
At 40 weeks, bodyweight significantly decreased from given tirzepatide (table 3). Hypoglycaemia (<70 mg/dL
baseline by an LSM of –7∙0 kg (SE 0∙52) with tirzepatide [<4 mmol/L]) was reported in seven (6%) participants

148 www.thelancet.com Vol 398 July 10, 2021

 Articles

A B
ETD –1·91 (95% CI –2·18 to –1·63); p<0·0001
Tirzepatide 5mg
Tirzepatide 10mg ETD –1·93 (95% CI –2·21 to –1·65); p<0·0001
Tirzepatide 15mg
Placebo ETD –2·11 (95% CI –2·39 to –1·83); p<0·0001
Overall mean baseline HbA1c=7·95% 0·5 5·5

HbA1c change from baseline (mmol/mol)

0·04
10 85·8

HbA1c change from baseline (%)

0 0

HbA1c (mmol/mol)
9 74·9 –0·5 –5·5
HbA1c (%)

8 7·99% 63·9 –1·0 –10·9

7·0%
7 53·0 –1·5 –16·4
6·5%
6·08%
6 6·06% 42·1 –2·0 –21·9
5·88% –1·87 –1·89
–2·07
5 31·1 –2·5 –27·3
0 4 8 12 16 20 24 32 40

Time (weeks)
C D
100 * 200 Overall mean baseline FSG=154 mg/dL 11·1
* 92 * *
Participants reaching HbA1c target (%)

87 88 * * 86
82 81
80 166·9 mg/dL
150 8·3

FSG (mmol/L)
FSG (mg/dL)

60 *
52
* * 110·5 mg/dL
40 34 31 108·1 mg/dL
100 104·8 mg/dL 5·6
20
20
10
1
0 50 2·8
HbA1c <7·0% HbA1c ≤6·5% HbA1c <5·7% 0 4 8 12 16 20 24 32 40

Time (weeks)

E F
ETD –56·5 (95% CI –66·8 to –46·1); p<0·0001 250 Baseline 13·9
Seven-point SMBG (mmol/L)
Seven-point SMBG (mg/dL)

At 12 weeks
ETD –58·8 (95% CI –69·2 to –48·4); p<0·0001
200 11·1
ETD –62·1 (95% CI –72·7 to –51·5); p<0·0001

150 8·3
20 12·9 1·1
FSG change from baseline (mmol/L)
FSG change from baseline (mg/dL)

100 5·6
0 0
250 Baseline 13·9
Seven-point SMBG (mmol/L)
Seven-point SMBG (mg/dL)

At 40 weeks
–20 –1·1
200 11·1

–40 150 8·3

–2·2
–43·6 –45·9
–49·3
–60 –3·3 100 5·6
t

ea t

lu t

er

di t

e
t

er
as

br os

os

os
as

im
nc

nc

nn

nn
kf

kf
hp

hp

hp
lu

dt
di
ea

e-

Be
2-

2-

2-
e-
br

Pr

Pr
e-
Pr

Timepoint

Figure 2: Proportion of participants reaching HbA1c targets, FSG, and seven-point SMBG profiles
Data are LSM (SE), unless otherwise noted. Estimated treatment differences are LSM (95% CI) at 40 weeks, modified intention-to-treat population (efficacy analysis set).
Arrows indicate when the maintenance dose of tirzepatide 5, 10, and 15 mg was reached. (A) HbA1c values over time from MMRM analysis. (B) Change from baseline in
HbA1c at 40 weeks from MMRM analysis. (C) Proportion of participants reaching HbA1c targets (<7·0%, ≤6·5%, and <5·7%) from logistic regression analysis. (D) FSG values
over time from MMRM analysis. (E) Change from baseline in FSG at 40 weeks from MMRM analysis. (F) Seven-point SMBG profiles at baseline, 12 weeks, and 40 weeks.
Dotted lines represent baseline values. ANOVA analysis (baseline) and MMRM analysis (12 weeks and 40 weeks). Data are estimates. ETD=estimated treatment
difference. FSG=fasting serum glucose. HbA1c=glycated haemoglobin. LSM=least squares mean. MMRM=mixed model repeated measures. SMBG=self-monitored blood
glucose. *p<0·0001 versus placebo at 40 weeks.

www.thelancet.com Vol 398 July 10, 2021 149

 Articles

Tirzepatide 5 mg (n=121) Tirzepatide 10 mg (n=121) Tirzepatide 15 mg (n=120) Placebo (n=113)

Mean* p value Mean* p value Mean* p value Mean* p value
HbA1c, %
Baseline 7∙97 (0∙08) ∙∙ 7∙88 (0∙08) ∙∙ 7∙88 (0∙08) ∙∙ 8∙08 (0∙08) ∙∙
Change from baseline –1∙87 (0∙09) <0∙0001 –1∙89 (0∙10) <0∙0001 –2∙07 (0∙10) <0∙0001 0∙04 (0∙11) 0∙72
Versus placebo –1∙91 <0∙0001 –1∙93 <0∙0001 –2∙11 <0∙0001 ∙∙ ∙∙
(–2∙18 to –1∙63) (–2∙21 to –1∙65) (–2∙39 to –1∙83)
HbA1c, mmol/mol
Baseline 63∙60 (0∙86) ∙∙ 62∙60 (0∙87) ∙∙ 62∙60 (0∙88) ∙∙ 64∙80 (0∙89) ∙∙
Change from baseline –20∙40 (1∙03) <0∙0001 –20∙70 (1∙05) <0∙0001 –22∙70 (1∙07) <0∙0001 0∙40 (1∙15) 0∙72
Versus placebo –20∙80 <0∙0001 –21∙10 <0∙0001 –23∙10 <0∙0001 ∙∙ ∙∙
(–23∙90 to –17∙80) (–24∙10 to –18∙00) (–26∙20 to –20∙00)
Weight, kg
Baseline 87∙0 (1∙8) ∙∙ 85∙7 (1∙8) ∙∙ 85∙9 (1∙8) ∙∙ 84∙4 (1∙9) ∙∙
Change from baseline –7∙0 (0∙5) <0∙0001 –7∙8 (0∙5) <0∙0001 –9∙5 (0∙5) <0∙0001 –0∙7 (0∙6) 0∙22
Versus placebo –6∙3 <0∙0001 –7∙1 <0∙0001 –8∙8 <0∙0001 ∙∙ ∙∙
(–7∙8 to –4∙7) (–8∙6 to –5∙5) (–10∙3 to –7∙2)
Participants reaching 105 (87%) ∙∙ 108 (92%) ∙∙ 102 (88%) ∙∙ 22 (19%) ∙∙
HbA1c <7∙0%†
Versus placebo 49∙0 <0∙0001 80∙4 <0∙0001 52∙9 <0∙0001 ∙∙ ∙∙
(21∙1 to 113∙7) (31∙8 to 203∙2) (22∙3 to 125∙7)
Fasting serum glucose, mg/dL
Baseline 153∙7 (3∙7) ∙∙ 152∙6 (3∙7) ∙∙ 154∙6 (3∙7) ∙∙ 155∙2 (3∙8) ∙∙
Change from baseline –43∙6 (3∙4) <0∙0001 –45∙9 (3∙5) <0∙0001 –49∙3 (3∙6) <0∙0001 12∙9 (4∙0) 0∙0014
Versus placebo –56∙5 <0∙0001 –58∙8 <0∙0001 –62∙1 <0∙0001 ∙∙ ∙∙
(–66∙8 to –46∙1) (–69∙2 to –48∙4) (–72∙7 to –51∙5)
Fasting serum glucose, mmol/L
Baseline 8∙5 (0∙2) ∙∙ 8∙5 (0∙2) ∙∙ 8∙6 (0∙2) ∙∙ 8∙6 (0∙2) ∙∙
Change from baseline –2∙4 (0∙2) <0∙0001 –2∙6 (0∙2) <0∙0001 –2∙7 (0∙2) <0∙0001 0∙7 (0∙2) 0∙0014
Versus placebo –3∙1 <0∙0001 –3∙3 <0∙0001 –3∙4 <0∙0001 ∙∙ ∙∙
(–3∙7 to –2∙6) (–3∙8 to –2∙7) (–4∙0 to –2∙9)
Participants with HbA1c 41 (34%) ∙∙ 36 (31%) ∙∙ 60 (52%) ∙∙ 1 (1%) ∙∙
<5∙7%†
Versus placebo 40∙3 <0∙0001 34∙1 <0∙0001 85∙1 <0∙0001 ∙∙ ∙∙
(7∙7 to 209∙7) (6∙5 to 178∙2) (16∙4 to 443∙1)
Participants with HbA1c 99 (82%) ∙∙ 96 (81%) ∙∙ 100 (86%) ∙∙ 11 (10%) ∙∙
≤6∙5%†
Versus placebo 74∙8 <0∙0001 69∙1 <0∙0001 105∙8 <0∙0001 ∙∙ ∙∙
(30∙6 to 183∙0) (28∙5 to 167∙6) (41∙3 to 271∙4)
Participants with ≥5% weight 81 (67%) ∙∙ 92 (78%) ∙∙ 89 (77%) ∙∙ 16 (14%) ∙∙
loss†
Versus placebo 12∙4 <0∙0001 21∙1 <0∙0001 20∙1 <0∙0001 ∙∙ ∙∙
(6∙4 to 23∙9) (10∙6 to 42∙2) (10∙1 to 40∙0)
Participants with ≥10% weight 37 (31%) ∙∙ 47 (40%) ∙∙ 55 (47%) ∙∙ 1 (1%) ∙∙
loss†
Versus placebo 34∙9 <0∙0001 50∙6 <0∙0001 71∙5 <0∙0001 ∙∙ ∙∙
(6∙8 to 180∙5) (9∙8 to 260∙4) (13∙9 to 368∙4)
Participants with 16 (13%) ∙∙ 20 (17%) ∙∙ 31 (27%) ∙∙ 0 ∙∙
≥15% weight loss†
Versus placebo 35∙6 0∙011 46∙5 0∙0063 83∙6 0∙0016 ∙∙ ∙∙
(2∙2 to 565∙3) (3∙0 to 733∙2) (5∙4 to >999∙0)

MMRM (mean) or logistic regression (odds ratio) analysis with missing value imputed by MMRM using the efficacy analysis set at 40 weeks, mITT population (efficacy
analysis set). HbA1c=glycated haemoglobin. LSM=least squares mean. n=mITT population on treatment without rescue therapy and excluding patients who discontinued
study drug due to inadvertent enrolment (efficacy analysis set). mITT=modified intention-to-treat. MMRM=mixed model repeated measures. *Baseline and change from
baseline at 40 weeks data are LSM (SE) and treatment differences are LSM (95% CI), except for HbA1c targets and weight loss of 5% or more, 10% or more, or 15% or more
versus placebo, which are n (%) or odds ratio (95% CI). †Proportion of people reaching HbA1c or weight loss goal were obtained by dividing the number reaching respective
goals at endpoint by the number with baseline value and at least one non-missing post-baseline value in mITT efficacy analysis set (121, 118, 116, 112 patients were the
denominator for tirzepatide 5, 10, and 15 mg, and placebo, respectively). No participants had weight loss of 15% or more with placebo; therefore, 95% CI was not calculated.

Table 2: Efficacy measures at 40 weeks

150 www.thelancet.com Vol 398 July 10, 2021

 Articles

A B ETD –6·3 (95% CI –7·8 to –4·7); p<0·0001

Tirzepatide 5mg
Tirzepatide 10mg ETD –7·1 (95% CI –8·6 to –5·5); p<0·0001
Tirzepatide 15mg
ETD –8·8 (95% CI –10·3 to –7·2);
Placebo
p<0·0001
2 Overall mean baseline weight=85·8 kg 0

Weight change from baseline (kg)

0
Weight change from baseline (%)

–0·9% –3
–2 –0·7

–4 –6 –7·0

–6 –9
–8 –7·9% –7·8
–9·3% –12
–10 –9·5
–11·0%
–12 –15
0 4 8 12 16 20 24 32 40

Time (weeks)

C D
90 100
Participants reaching bodyweight loss target (%)

† †
78 77
80 †
85·5 kg 67
85
60
Weight (kg)

†
† 47
40
40 †
31 *
80 27
79·2 kg *
78·4 kg 20 14 * 17
13
76·7 kg 1 0
75 0
0 4 8 12 16 20 24 32 40 ≥5% weight loss ≥10% weight loss ≥15% weight loss

Time (weeks) Weight loss (%)

E
15
†
4·7 † 7·5
10 *
4·8
Lipids change from baseline (%)

3·2
5

–5 –0·8
–3·8 –1·6
–5·5
–10 –6·3 –6·7
* * –8·4 –7·6
†
–15
–12·4
*
–20
–18·5
–25 † –18·2
† –21·0
†
Triglycerides Total HDL LDL
cholesterol cholesterol cholesterol

Figure 3: Bodyweight and proportion of participants reaching weight loss aims and lipid profile
Data are LSM (SE), unless otherwise noted. Estimated treatment differences are LSM (95% CI) at 40 weeks, modified intention-to-treat population (efficacy analysis
set). Arrows indicate when the maintenance dose of tirzepatide 5, 10, and 15 mg was reached. (A) Percentage change from baseline in bodyweight over time from
MMRM analysis. (B) Change from baseline in bodyweight at 40 weeks from MMRM analysis. (C) Bodyweight values over time from MMRM analysis. (D) Proportion of
participants reaching weight loss aims (≤5%, ≤10%, and ≤15%) was obtained by dividing the number of patients reaching respective goals at week 40 by the number
of patients with baseline value and at least one non-missing post-baseline value. Missing value at week 40 was predicted from MMRM analysis. Data are estimates.
(E) Percentage change from baseline in lipids at 40 weeks. Data are estimated percentage means (SE) using log transformation. ETD=estimated treatment difference.
MMRM=mixed model repeated measures. *p<0·05 versus placebo at 40 weeks. †p<0·0001 versus placebo at 40 weeks.

www.thelancet.com Vol 398 July 10, 2021 151

 Articles

Tirzepatide 5 mg Tirzepatide 10 mg Tirzepatide 15 mg Placebo Total

(n=121) (n=121) (n=121) (n=115) (n=478)
Participants with ≥1 treatment-emergent adverse event 83 (69%) 81 (67%) 77 (64%) 76 (66%) 317 (66%)
Serious adverse events 5 (4%) 2 (2%) 1 (1%) 3 (3%) 11 (2%)
Deaths* 0 0 0 1 (1%) 1 (<1%)
Adverse event leading to study drug discontinuation 4 (3%) 6 (5%) 8 (7%) 3 (3%) 21 (4%)
Gastrointestinal disorder (system order class) 3 (2%) 6 (5%) 8 (7%) 1 (1%) 18 (4%)
Gastrointestinal disorder (preferred term) 1 (1%) 2 (2%) 2 (2%) 0 5 (1%)
Diarrhoea 0 2 (2%) 2 (2%) 0 4 (1%)
Nausea 0 2 (2%) 1 (1%) 1 (1%) 4 (1%)
Abdominal discomfort 0 0 2 (2%) 0 2 (<1%)
Dyspepsia 1 (1%) 0 1 (1%) 0 2 (<1%)
Colitis ischaemic 1 (1%) 0 0 0 1 (<1%)
Treatment-emergent adverse events occurring in ≥5% of participants in any treatment group (preferred term)
Nausea 14 (12%; 31) 16 (13%; 82) 22 (18%; 50) 7 (6%; 8) 59 (12%; 171)
Diarrhoea 14 (12%; 21) 17 (14%; 19) 14 (12%; 20) 9 (8%; 15) 54 (11%; 75)
Hyperglycaemia 4 (3%) 5 (4%) 3 (2%) 31 (27%) 43 (9%)
Nasopharyngitis 7 (6%) 8 (7%) 8 (7%) 10 (9%) 33 (7%)
Dyspepsia 11 (9%) 8 (7%) 7 (6%) 4 (3%) 30 (6%)
Decreased appetite 5 (4%) 8 (7%) 10 (8%) 1 (1%) 24 (5%)
Headache 5 (4%) 4 (3%) 5 (4%) 9 (8%) 23 (5%)
Constipation 7 (6%) 6 (5%) 8 (7%) 1 (1%) 22 (5%)
Vomiting 4 (3%; 6) 3 (2%; 3) 7 (6%; 9) 2 (2%; 3) 16 (3%; 21)
Influenza 7 (6%) 3 (2%) 0 2 (2%) 12 (3%)
Gastritis 6 (5%) 0 3 (2%) 0 9 (2%)
All gastrointestinal adverse events 46 (38%) 50 (41%) 50 (41%) 22 (19%) 168 (35%)
Other adverse events
Hypoglycaemia (blood glucose <70 mg/dL) 7 (6%; 16) 8 (7%; 19) 8 (7%; 19) 1 (1%; 6) 24 (5%; 60)
Hypoglycaemia (blood glucose <54 mg/dL) 0 0 0 1 (1%; 3) 1 (<1%; 3)
Severe hypoglycaemia 0 0 0 0 0
Injection site reactions 4 (3%; 6) 4 (3%; 31) 3 (2%; 15) 0 11 (2%; 52)
Adjudicated pancreatitis† 0 0 0 0 0
Pancreatic cancer† 1 (1%) 0 0 0 1 (<1%)
Cholelithiasis† 1 (1%) 0 0 0 1 (<1%)
Hypersensitivity‡ 3 (2%; 3) 2 (2%; 2) 1 (1%; 1) 1 (1%; 1) 7 (1%; 7)
Data are n (%), or n (%; number of episodes). Patients could be counted in more than one category. Number of episodes were reported if available. n=all randomly assigned
participants who took at least one dose of study drug (modified intention-to-treat population). *Deaths are also included as serious adverse events and discontinuations due to
adverse events. †Medical Dictionary for Regulatory Activities preferred term. ‡Includes immediate (≤24 h after study drug administration) and non-immediate (>24 h after study
drug administration) hypersensitivity events. One immediate event was reported in the tirzepatide 15 mg group.

Table 3: Adverse events

who received tirzepatide 5 mg and eight (7%) participants normal range (p≤0∙0019, all tirzepatide groups; appendix
in each group who received tirzepatide 10 mg and 15 mg, p 13). One case of pancreatic cancer was diagnosed
versus one (1%) with placebo. Rescue therapy for during evaluation of haematuria in the tirzepatide 5 mg
persistent hyperglycaemia was required for two (2%) group during the safety follow-up period and one case of
participants who received tirzepatide 5 mg, four (3%) who cholelithiasis was reported in the same individual. No
received tirzepatide 10 mg, and two (2%) who received clinically relevant changes in mean calcitonin con­
tirzepatide 10 mg and 15 mg, versus 29 (25%) with centrations were observed. No cases of medullary thyroid
placebo. Metformin was the most com­ mon anti­ cancer or treatment-emergent diabetic retinopathy were
hyperglycaemic medication used as rescue therapy for reported.
persistent hyperglycaemia during the study. At 40 weeks, increases in mean pulse rate of 1–2 beats
No adjudicated-confirmed cases of pancreatitis were per min were observed with tirzepatide (appendix
seen. At 40 weeks, increases from baseline in lipase and pp 13, 25). At 40 weeks, decreases in mean systolic blood
pancreatic amylase concentrations were significant with pressure ranged from –4·7 to –5·2 mm Hg with
tirzepatide versus placebo, while remaining within the tirzepatide versus with –2·0 mm Hg placebo, and differed

152 www.thelancet.com Vol 398 July 10, 2021

 Articles

significantly with tirzepatide 10 mg (appendix pp 13, 25). 14% of participants in the placebo group, probably due to
Decreases in diastolic blood pressure did not differ from the placebo effect. Weight loss occurred irrespective of
placebo. Decreases in alanine aminotransferase (ALT) participants reporting gastro­intestinal adverse events and
and aspartate aminotransferase (AST) concentrations did not plateau over time.
differed significantly with tirzepatide, except with In this study, HbA1c reduction plateaued at around
tirzepatide 10 mg for AST, compared with placebo 20 weeks, whereas weight reductions were progressive
(appendix p 13). during the trial, suggesting that HbA1c improved via both
Injection site reactions occurred in 11 (2–3%) par­ weight-dependent and weight-independent mechanisms.
ticipants with tirzepatide versus no reports with placebo A mechanism of action study is ongoing, which will help
(table 3). Hypersensitivity reactions occurred in 1–2% of to further elucidate the relationship between HbA1c and
participants given tirzepatide and 1% with placebo; no bodyweight reduction (NCT03951753).
systemic hypersensitivity events were reported. No Similarly, GLP-1 receptor agonist monotherapy trials
hypersensitivity cases and injection site reactions were evaluating dulaglutide injection (0∙75 mg and 1∙5 mg
severe or serious and none of the cases led to once a week), semaglutide injection (0∙5 mg and 1∙0 mg
discontinuation of study drug. Overall, there was no once a week), and oral semaglutide (7 mg and 14 mg once
evidence of a diminished effect of tirzepatide in a day) also showed small differences in HbA1c reduction
tirzepatide pharmacokinetics or HbA1c. None of the with the higher doses of each drug (an increase of
participants who showed treatment-emergent antidrug around 0∙1–0∙2%).15–17 Compared with other incretin-
antibody-positivity had severe or serious hypersensitivity based monotherapy trials, our results showed a higher
or injection site reactions (data not shown). proportion of participants reaching HbA1c targets of less
Overall, results from the treatment-regimen estimand than 7∙0%1 or 6∙5% or less18,19 with tirzepatide treatment.
for HbA1c, fasting serum glucose, bodyweight, and The proportion of participants with HbA1c concentrations
proportion of participants reaching HbA1c targets and of less than 7∙0% was 87–92% for tirzepatide (5, 10, and
weight loss aims were similar to those reported from the 15 mg), 63% for exenatide once a week (2∙0 mg),20 61–63%
efficacy estimand (appendix pp 14–15). for dulaglutide (0∙75 mg and 1∙5 mg),15 and 72–74% for
semaglutide once a week (0∙5 mg and 1∙0 mg).16
Discussion Near-normalisation or complete normalisation of
SURPASS-1 showed significant improvement in glucose concentrations (HbA1c <5·7%) without increasing
glycaemic control and robust bodyweight reductions hypoglycaemic risk is expected to further reduce the
with all three tirzepatide doses compared with placebo risk of complications. An analysis of UK Prospective
in people with type 2 diabetes treated with diet and Diabetes Study data reported a continuous association of
exercise alone. Importantly, 87–92% of participants microvascular and macrovascular diabetes complications
given tirzepatide reached the American Diabetes with the lowest risk being in those with HbA1c values in
Association (ADA)-recommended HbA1c target of less the normal range (<6·0%).21 Newer, glucose-dependent,
than 7∙0% (<53 mmol/mol). Furthermore, 31–52% of glucose-lowering agents (such as tirzepatide) that safely
participants reached normoglycaemia (HbA1c <5∙7% lower HbA1c without increasing the risk of hypo­
[<39 mmol/mol]) without an increased risk of cli­ glycaemia could help to reduce long-term microvascular
nically significant (<54 mg/dL [<3 mmol/L]) or severe and macrovascular complications. The ongoing trial
hypoglycaemia. This result was further supported by the SURPASS-CVOT (NCT04255433) of tirzepatide versus
significant improvements in fasting serum glucose and dulaglutide 1·5 mg will help to shed light on the potential
SMBG profiles towards the normal range, which were clinical relevance of reaching and maintaining near-
observed as early as 12 weeks after study initiation and normoglycaemia in type 2 diabetes management and the
were maintained at 40 weeks. The marked improve­ment effect of tirzepatide on cardiovascular outcomes.
in postprandial hyperglycaemia observed in SMBG In SURPASS-1, tirzepatide treatment resulted in
data are probably achieved by the role of tirzepatide robust HbA1c reductions in a population with early type 2
in glucose-dependent insulin secretion and are likely diabetes and 54% of participants were naive to any
to be unrelated to slowed gastric emptying given the diabetes treatment, thus the legacy effect of tirzepatide
persistence of improved postprandial hyperglycaemia at remains to be investigated. Furthermore, we observed
40 weeks, well after the transient effect of tirzepatide in favourable changes in fasting lipids, insulin sensitivity,
delaying gastric emptying has waned.13,14 and systolic blood pressure with tirzepatide. The cardio­
In addition to robust glycaemic effects, significant, dose- vascular effects of GIP receptor agonism are under
dependent weight reductions were observed at 40 weeks, debate and are being assessed.9
with 13–27% of the participants given tirzepatide having a Robust HbA1c reductions with tirzepatide 5 mg observed
15% or greater weight loss versus none with placebo. in this study are likely to be related to the relatively early
Importantly, 67–78% of participants given tirzepatide course of type 2 diabetes in the participants studied and
reached the ADA-recommend weight loss of 5% or greater the greater preservation of pancreatic β cell function
in individuals with type 2 diabetes and obesity,1,2 versus expected in this population. All three tirzepatide doses

www.thelancet.com Vol 398 July 10, 2021 153

 Articles

leading to near-normoglycaemia (mean HbA1c of received honoraria or consulting fees from Applied Therapeutics, Eli Lilly
5∙9–6∙1%) at 40 weeks and presumably reaching a floor and Company, Sanofi, Novo Nordisk, Hanmi, Oramed, Boehringer
Ingelheim, and Intarcia; and also received grants and research support
effect (ie, reaching HbA1c values within the normal range) from Applied Therapeutics, Merck, Pfizer, Sanofi, Novo Nordisk, Eli Lilly
with no room for further reductions might explain the and Company, GlaxoSmithKline, Genentech, Hanmi, Oramed, Janssen,
absence of apparent dose-response on HbA1c reductions. Lexicon, Boehringer Ingelheim, and Intarcia. CW has received consulting
Similarly, a less pronounced dose-response has also been and advisory board fees from AstraZeneca, Janssen, and Sanofi; research
support from AstraZeneca and Novo Nordisk; and speaker fees from
observed in GLP-1 receptor agonist monotherapy trials AstraZeneca, Boehringer Ingelheim, Eli Lilly and Company, Janssen,
with lesser glucose-lowering effects.15–17,20 Novo Nordisk, and Sanofi. JPF declares grants from Eli Lilly and
The safety profile in the SURPASS-1 trial was similar to Company, AbbVie, Akcea, Allergan, AstraZeneca, Boehringer Ingelheim,
that of selective GLP-1 receptor agonists with gastro­ Bristol Myers Squibb, Cirius, CymaBay, Enanta, Genentech, Intercept,
Janssen, Johnson and Johnson, Lexicon, Ligand, Madrigal, Merck, Mylan,
intestinal side-effects as the most frequently reported NGM Biopharmaceuticals, Novartis, Novo Nordisk, Pfizer, Sanofi, and
adverse events. The optimised slow dose escalation Theracos; has served on advisory boards and received consulting fees
scheme used in the phase 3 programme resulted in an from Boehringer Ingelheim, Gilead, Johnson and Johnson, Eli Lilly and
improved tolerability profile compared with the phase 2 Company, Merck, Novo Nordisk, and Sanofi; and served on a speaker
bureau for Merck and Sanofi. SK received research support from
studies.10,12 Overall, gastrointestinal adverse events were Novo Nordisk, Poxel SA, and Eli Lilly Japan; received honoraria for
reported more frequently in the tirzepatide groups than lectures from Sumitomo Dainippon Pharma, Novo Nordisk, Eli Lilly
for placebo. Increases in pulse rate reported in this study Japan, AstraZeneca, Boehringer Ingelheim, and Mitsubishi Tanabe
were consistent with previous data for GLP-1 receptor Pharma; and received consulting and advisory board fees from
Novo Nordisk.
agonists.15,16,20,22 Reductions in ALT and AST were also
observed with tirzepatide, supporting the potential role Data sharing
Eli Lilly provides access to all individual participant data collected during
of tirzepatide in fatty liver disease.23 No clinically rele­ the trial, after anonymisation, with the exception of pharmacokinetic or
vant changes in mean calcitonin con­ centrations were genetic data. Data are available to request 6 months after the indication
observed and no cases of medullary thyroid cancer were studied has been approved in the USA and EU and after primary
publication acceptance, whichever is later. No expiration date of data
reported. Of note, there were no signals of hyperplasia or
requests is currently set once data are made available. Access is provided
neoplasia in the thyroid of rats or monkeys administered after a proposal has been approved by an independent review committee
tirzepatide for up to 6 months in duration or in a 6-month identified for this purpose and after receipt of a signed data sharing
carcinogenicity study in rasH2 mice (unpublished). A agreement. Data and documents, including the study protocol, statistical
analysis plan, clinical study report, and blank or annotated case report
2-year carcinogenicity study in rats is ongoing.
forms, will be provided in a secure data sharing environment. For details
For details see https://vivli.org/ One limitation of the SURPASS-1 trial was the relatively on submitting a request, see the instructions provided online.
short duration of 40 weeks. A longer study duration could
Acknowledgments
provide further insights on the full weight loss effects of We thank Weiguo Zhu and Xiaolei Zhang (Eli Lilly and Company) for
tirzepatide. Additionally, gastrointestinal adverse events their contributions on data preparation and analysis. Partial data from
were self-reported. Study strengths include the ran­ this study were presented at the American Diabetes Association
81st Virtual Scientific Sessions, held June 25–29, 2021.
domised, placebo-controlled design with use of a placebo
single dose pen that was similar in appearance to the References
1 American Diabetes Association. Standards of medical care in
tirzepatide single dose pen. diabetes-2021. Diabetes Care 2021; 44 (suppl 1): S1–2.
In conclusion, tirzepatide once a week, a novel dual 2 Buse JB, Wexler DJ, Tsapas A, et al. 2019 Update to: management of
GIP and GLP-1 receptor agonist, at doses of 5, 10, and hyperglycemia in type 2 diabetes, 2018. A consensus report by the
American Diabetes Association (ADA) and the European Association
15 mg as monotherapy for type 2 diabetes, showed robust for the Study of Diabetes (EASD). Diabetes Care 2020; 43: 487–93.
reductions compared with placebo in glycaemic control 3 Gasbjerg LS, Helsted MM, Hartmann B, et al. Separate and
with 31–52% of participants reaching normoglycaemia combined glucometabolic effects of endogenous glucose-dependent
insulinotropic polypeptide and glucagon-like peptide 1 in healthy
(HbA1c <5∙7% [<39 mmol/mol]), and meaningful individuals. Diabetes 2019; 68: 906–17.
reductions in bodyweight, without increased risk of 4 Nauck MA, Heimesaat MM, Orskov C, Holst JJ, Ebert R,
clinically significant (<54 mg/dL [<3 mmol/L]) or severe Creutzfeldt W. Preserved incretin activity of glucagon-like peptide 1
[7–36 amide] but not of synthetic human gastric inhibitory
hypoglycaemia, and a safety profile consistent with polypeptide in patients with type-2 diabetes mellitus. J Clin Invest
GLP-1 receptor agonists, indicating a potential use of 1993; 91: 301–07.
tirzepatide as an option for type 2 diabetes treatment. 5 Christensen M, Vedtofte L, Holst JJ, Vilsbøll T, Knop FK.
Glucose-dependent insulinotropic polypeptide: a bifunctional
Contributors glucose-dependent regulator of glucagon and insulin secretion in
VTT, LFL, and XC contributed to the study design. JR, VTT, LFL, and humans. Diabetes 2011; 60: 3103–09.
CJL conducted and provided medical oversight during the trial. XC and 6 Campbell JE, Drucker DJ. Pharmacology, physiology, and
HM were responsible for the statistical analyses. CJL, LFL, HM, XC, mechanisms of incretin hormone action. Cell Metab 2013;
CAK, and VTT are the guarantors of this work and, as such, take 17: 819–37.
responsibility for the integrity of the data and the accuracy of the data 7 Adriaenssens AE, Biggs EK, Darwish T, et al. Glucose-dependent
analysis. All authors participated in data interpretation and critical review insulinotropic polypeptide receptor-expressing cells in the
of the manuscript, had full access to all the data in the study, and hypothalamus regulate food intake. Cell Metab 2019; 30: 987–96.
approved of this manuscript to be submitted for publication. 8 Cosgrove RJ, Antonellis MP, Droz BA, et al. Tirzepatide, a dual GIP
and GLP-1 receptor agonist, mediates its anorectic effect in
Declarations of interest preclinical models of obesity due to a reduction in homeostatic and
CJL, LFL, HM, XC, CAK, and VTT are employees and shareholders of hedonic related feeding. Diabetologia 2020; 63 (suppl 1): 1–485.
Eli Lilly and Company. JR serves on scientific advisory boards and

154 www.thelancet.com Vol 398 July 10, 2021

 Articles

9 Baggio LL, Drucker DJ. Glucagon-like peptide-1 receptor co-agonists 17 Aroda VR, Rosenstock J, Terauchi Y, et al. PIONEER 1: randomized
for treating metabolic disease. Mol Metab 2021; 46: 101090. clinical trial of the efficacy and safety of oral semaglutide
10 Frías JP, Nauck MA, Van J, et al. Efficacy and safety of LY3298176, monotherapy in comparison with placebo in patients with type 2
a novel dual GIP and GLP-1 receptor agonist, in patients with type 2 diabetes. Diabetes Care 2019; 42: 1724–32.
diabetes: a randomised, placebo-controlled and active comparator- 18 Rodbard HW, Blonde L, Braithwaite SS, et al. American Association
controlled phase 2 trial. Lancet 2018; 392: 2180–93. of Clinical Endocrinologists medical guidelines for clinical practice
11 Coskun T, Sloop KW, Loghin C, et al. LY3298176, a novel dual GIP for the management of diabetes mellitus. Endocr Pract 2007;
and GLP-1 receptor agonist for the treatment of type 2 diabetes 13 (suppl 1): 1–68.
mellitus: From discovery to clinical proof of concept. Mol Metab 19 National Institute for Clinical Excellence. Type 2 diabetes in adults:
2018; 18: 3–14. management (NG28). Updated Dec 16, 2020. https://www.nice.org.
12 Frías JP, Nauck MA, Van J, et al. Efficacy and tolerability of uk/guidance/ng28 (accessed May 28, 2021).
tirzepatide, a dual glucose-dependent insulinotropic peptide and 20 Russell-Jones D, Cuddihy RM, Hanefeld M, et al. Efficacy and safety
glucagon-like peptide-1 receptor agonist in patients with type 2 of exenatide once weekly versus metformin, pioglitazone, and
diabetes: a 12-week, randomized, double-blind, placebo-controlled sitagliptin used as monotherapy in drug-naive patients with type 2
study to evaluate different dose-escalation regimens. diabetes (DURATION-4): a 26-week double-blind study.
Diabetes Obes Metab 2020; 22: 938–46. Diabetes Care 2012; 35: 252–58.
13 Nauck MA, Baller B, Meier JJ. Gastric inhibitory polypeptide and 21 Stratton IM, Adler AI, Neil HAW, et al. Association of glycaemia
glucagon-like peptide-1 in the pathogenesis of type 2 diabetes. with macrovascular and microvascular complications of type 2
Diabetes 2004; 53 (suppl 3): S190–96. diabetes (UKPDS 35): prospective observational study. BMJ 2000;
14 Urva S, Coskun T, Loghin C, et al. The novel dual glucose- 321: 405–12.
dependent insulinotropic polypeptide and glucagon-like peptide-1 22 Lorenz M, Lawson F, Owens D, et al. Differential effects of
(GLP-1) receptor agonist tirzepatide transiently delays gastric glucagon-like peptide-1 receptor agonists on heart rate.
emptying similarly to selective long-acting GLP-1 receptor agonists. Cardiovasc Diabetol 2017; 16: 6.
Diabetes Obes Metab 2020; 22: 1886–91. 23 Samms RJ, Coghlan MP, Sloop KW. How may GIP enhance the
15 Umpierrez G, Tofé Povedano S, Pérez Manghi F, Shurzinske L, therapeutic efficacy of GLP-1? Trends Endocrinol Metab 2020;
Pechtner V. Efficacy and safety of dulaglutide monotherapy versus 31: 410–21.
metformin in type 2 diabetes in a randomized controlled trial
(AWARD-3). Diabetes Care 2014; 37: 2168–76.
16 Sorli C, Harashima SI, Tsoukas GM, et al. Efficacy and safety of
once-weekly semaglutide monotherapy versus placebo in patients
with type 2 diabetes (SUSTAIN 1): a double-blind, randomised,
placebo-controlled, parallel-group, multinational, multicentre
phase 3a trial. Lancet Diabetes Endocrinol 2017; 5: 251–60.

www.thelancet.com Vol 398 July 10, 2021 155