Series

Diabetes 2
Cardiovascular outcome trials of glucose-lowering drugs
or strategies in type 2 diabetes
Rury R Holman, Harald Sourij, Robert M Cali
Lancet 2014; 383: 200817
This online publication has
been corrected. The corrected
version rst appeared at
thelancet.com on June 27, 2014
See Editorial page 1945
This is the second in a Series of
two papers about diabetes
Diabetes Trials Unit, University
of Oxford, Oxford, UK
(Prof R R Holman FRCP,
H Sourij MD); Division of
Endocrinology and
Metabolism, Medical
University of Graz, Graz,
Austria (H Sourij); and Duke
Clinical Research Institute,
Duke University School of
Medicine, Durham, NC, USA
(Prof R M Cali MD)
Correspondence to:
Prof Rury R Holman, Diabetes
Trials Unit, OCDEM, Churchill
Hospital, Old Road, Headington,
Oxford OX3 7LJ, UK
rury.holman@dtu.ox.ac.uk

Few trials of glucose-lowering drugs or strategies in people with type 2 diabetes have investigated cardiovascular
outcomes, even though most patients die from cardiovascular causes despite the benecial eects of lipid-reducing
and blood pressure-lowering treatments. The evidence-based reduction in risk of microvascular disease with glucose
lowering has resulted in guidelines worldwide recommending optimisation of glycosylated haemoglobin, but no trial
results have shown unequivocal cardiovascular risk reduction with glucose lowering. However, results of the post-trial
follow-up of the UK Prospective Diabetes Study, and of a meta-analysis of the four glucose-lowering outcome trials
completed to date, suggest about a 15% cardiovascular relative risk reduction per 1% decrement in HbA1c. The 2008
US Food and Drug Administration industry guidance for licensing of antidiabetic drugs greatly increased the number
of cardiovascular outcome trials in diabetes, but most trials opted for non-inferiority designs aiming primarily to
show absence of cardiovascular toxicity in the shortest possible time. This unintended consequence of the new
regulations has meant that the potential long-term benets, and the possible risks of new therapies, are not being
assessed eectively. Also, essential head-to-head trials of therapies for this complex progressive disease, to answer
issues such as how best to achieve and maintain optimum glycaemia without promoting weight gain or hypoglycaemia,
are not being undertaken. In this Series paper, we summarise randomised controlled cardiovascular outcome trials in
type 2 diabetes, provide an overview of ongoing trials and their limitations, and speculate on how future trials could
be made more ecient and eective.

Introduction
The rapidly emerging diabetes pandemic is one of the
most challenging noncommunicable disease threats to
public health in the 21st century. More than 380 million
people worldwide have type 2 diabetes,1 most of whom
will die from cardiovascular disease, the second most
common cause of death being their increased risk of
cancer.2 People with type 2 diabetes have a roughly

Search strategy and selection criteria

We searched for cardiovascular outcome trials of
glucose-lowering drugs or strategies in patients with type 2
diabetes with the following denition: randomised, controlled
clinical trial, primary study endpoint cardiovascular outcome,
sample size >1 000, patients followed-up for 12 months. We
searched the Cochrane Central Register of Controlled Trials,
Medline, and Embase for reports published in English between
Jan 1, 1996, and Jan 4, 2014, with the search terms diabetes,
in combination with the terms cardiovascular,
complications, and mortality. Table 1 lists the results tting
the search criteria. To identify ongoing trials we used the
advanced search tab at www.ClinicalTrials.gov on
Jan 13, 2014, to search interventional trials, phase 3 or higher,
with the keyword diabetes. We included ongoing trials
registered between March 1, 2008, and Jan 10, 2014,
investigating glucose-lowering pharmacological interventions,
with a planned follow-up of at least 18 months and a recent
update on the study status on www.ClinicalTrials.gov. Table 2
lists the results tting these criteria.

2008

doubled risk of cardiovascular disease, compared with

people without diabetes, even after adjustment for
established cardiovascular risk factors.2 After diagnosis,
patients have a lifetime of attempted strict lifestyle,
escalating drug therapy to oset glycaemic progression,
and management of several risk factors to minimise the risk of diabetes-related complications. Life
expectancy of a person diagnosed with type 2 diabetes
at age 40 years is estimated to be shortened by about
67 years, compared with people without type 2
diabetes.2 A major need exists for evidence-based type 2
diabetes treatment strategies that better manage this
complex condition by reducing disease progression,
improving quality of life, and extending longevity,
without causing unwanted or harmful side-eects.
Crucially, the precise ability of glucose-lowering per se
to reduce the risk of cardiovascular disease in type 2
diabetes has not been elucidated.
Despite the many people aected by type 2 diabetes,
the huge disease burden, and the major health economic
costs to society, few denitive outcome trials to properly
inform clinical eectiveness of licensed therapies have
been done. Before publication of the results of the UK
Prospective Diabetes Study (UKPDS)3 and the Diabetes
Control and Complications Trial,4 the prevailing opinion
was that diabetic complications were primarily genetic in
origin and unrelated to the carbohydrate derangements
of diabetes mellitus.5 Although the UKPDS ndings in
type 2 diabetes showed that improved glycaemic control3
and blood pressure control6 could reduce the risk of
diabetic complications, most drug trials undertaken for
www.thelancet.com Vol 383 June 7, 2014

Series

Year of
Population
publication

Follow up

Intervention

Outcome

UKPDS3,12

1998

Newly diagnosed T2DM

4209

Median
100 years

Intensive versus conventional

glucose lowering strategy

Fatal and nonfatal myocardial Infarction: Intensive (SFU/

Insulin) versus conventional: RRR 16% (p=0052); intensive
(metformin) versus conventional: RRR 39% (p=0010)
All-cause mortality: intensive (SFU/Insulin) versus
conventional: RRR 6% (p=044); intensive (metformin)
versus conventional: RRR 36% (p=0011)

DIGAMI 213

2005

T2DM and acute MI

1253

Median
21 years

Intensive versus usual

metabolic control versus after
MI

All-cause mortality
Group 1 versus 2: HR 103 (95% CI 079134)
Group 2 versus 3: HR 123 (95% CI 089169)

PROactiv14

2005

T2DM with
macrovascular disease

5238

Mean
345 months

Addition of pioglitazone or
placebo to usual diabetes care

All-cause mortality, non-fatal MI, stroke:

HR 084 (95% CI 072098)

ADVANCE15

2008

T2DM, at least 55 years

old, history of
macrovascular or
microvascular disease or
at least one other CV
risk factor

11 140

Median
5 years

Intensive (gliclazide plus other MACE: HR 094 (95% CI 084106)

drugs) versus standard glucose
control strategy

ACCORD16

2008

T2DM with established

CVD or additional CV
risk factors

10 251

Mean
35 years

Intensive versus standard

glucose control strategy

MACE: HR 090 (95% CI 078104)

All-cause mortality: HR 122 (95% CI 101146)

HEART2D17

2009

T2DM after acute MI

1115

Mean
26 years

Prandial versus basal insulin

strategy

Composite of CV death, MI, stroke, coronary revascularisation,

or hospitalisation for ACS: HR 104 (95% CI 078137)

VADT18

2009

T2DM

1791

Median
56 years

Intensive versus standard

glucose control strategy

Composite of MI, stroke, CV death, new or worsening

congestive heart failure, surgical intervention for cardiac,
cerebrovascular or PAD, inoperable CAD, amputation for
ischaemic gangrene): HR 088 (95% CI 074105)

RECORD19

2009

T2DM on maximum
tolerated dose of
metformin or SFU

4447

Mean
55 years

Addition of rosiglitazone or
combination of metformin
and sulfonylurea

Composite of CV death and CV hospitalisations:

HR 093 (074115)

BARI 2D20

2009

T2DM and heart disease

2368

Mean
53 years

Insulin-sensitisation versus
insulin-provision treatment
strategy

No dierence in survival: 882% versus 879%, p=089

ADDITION21

2011

Screen-detected T2DM

3055

Mean
53 years

Routine versus intensied

multifactorial risk factor
intervention

Composite of CV death, CV morbidity, revascularisation,

and non-traumatic amputation: HR 083 (95% CI 065105)

ORIGIN22

2012

CV risk factors plus T2DM

or IFG or IGT

12 537

Median
62 years

Insulin glargine or standard

glucose control

MACE: HR 102 (95% CI 094111)

SAVOR-TIMI5323

2013

T2DM with history of

CV event or at risk for

16 492

Median
21 years

Addition of Saxagliptin versus

placebo to usual diabetes care

MACE: HR 100 (95% CI 089112)

EXAMINE24

2013

T2DM with ACS within

15 to 90 before
randomisation

5380

Median
15 years

Addition of Alogliptin versus

placebo to usual diabetes care

MACE: HR 096 (upper bound of 95% CI 116)

LOOK-AHEAD25

2013

Overweight or obese
T2DM with or without
CVD history

5145

Median
96 years

Intensive versus standard

lifestyle intervention strategy

Composite of CV death, MI, stroke and hospitalisation for

angina: HR 095 (95% CI 083109)

T2DM=type 2 diabetes mellitus. SFU=sulfonylurea. RRR=relative risk reduction. MI=myocardial infarction. CV=cardiovascular. MACE=cardiovascular mortality, myocardial infarction, stroke. CVD=cardiovascular
disease. PAD=peripheral artery disease. CAD=coronary artery disease. IFG=impaired fasting glucose. IGT=impaired glucose tolerance. ACS=acute coronary syndrome. HR=hazard ratio.

Table 1: Randomised, controlled, cardiovascular outcome trials (>1000 subjects, >1 year of follow up) of glucose-lowering drugs or strategies in people with type 2 diabetes

regulatory purposes continued to be short term, usually

in selected populations, and often focused primarily on
biomarkers rather than on hard outcomes. This situation
changed after the 2008 US Food and Drug Administration
(FDA) mandate and subsequent European Medicines
Agency requirements for cardiovascular outcome trials
in licensing of new glucose-lowering drugs.7,8 These
regulatory requirements, driven primarily by response to
cardiovascular safety concerns associated with the
antidiabetic drug rosiglitazone,9,10 have changed the
diabetes trials landscape. The FDA require that clinical
www.thelancet.com Vol 383 June 7, 2014

trials done before drug approval show a two-sided 95%

condence interval upper boundary of 18 risk ratio for
major adverse cardiovascular events, versus the control
group, with subsequent outcomes trials having an upper
boundary of 13.7 Bethel and Sourij11 reviewed the eect
of this two-stage route to type 2 diabetes drug approval.
The regulatory need to obtain robust cardiovascular
safety data to approve new diabetes drugs has led to a
substantial increase in the number of type 2 diabetes
cardiovascular outcome trials. However, most of these
trials have simple, placebo-controlled, non-inferiority
2009

Series

designs aiming to show an absence of cardiovascular

toxicity to satisfy regulatory requirements in the shortest
possible time. This approach has meant that many
essential questions are unanswered, such as comparative
eectiveness, estimation of the balance of benets to
risks over time, identication of heterogeneity of
treatment eects to establish best patient selection, and
assessment of possible drug-related adverse events in the
long term.
In this Series paper, we review published cardiovascular
outcome trials of glucose-lowering drugs or strategies in
type 2 diabetes (table 1), provide an overview of ongoing
trials and their limitations, and speculate on potentially
more eective and ecient trial designs.

Published cardiovascular outcome trials in

type 2 diabetes
The University Group Diabetes Programme, published
in 1970,26 was the rst randomised multicentre head-tohead eectiveness trial of available type 2 diabetes
glucose-lowering treatments that assessed cardiovascular
outcomes. The trial randomly assigned about 200 patients
to each of variable-dose insulin, standard-dose insulin,
sulfonylurea (tolbutamide), biguanide (phenformin), and
placebo. The trial was halted because of concerns that all
of the therapies seemed to increase cardiovascular risk.26
Although the University Group Diabetes Programme
was grossly underpowered, controversy still exists
regarding its ndings, and concerns about unexpected
adverse outcomes might have inhibited industry funding
of outcome trials for many years. The introduction of
metformin to North America was delayed until 1994, and
even in 2014, sulfonylureas carry a so-called black box in
the USA, signifying FDA concern that, as a class, they
might increase cardiovascular risk.
The 1998 UK Prospective Diabetes Study3 provided the
rst robust outcome data to inform clinical practice
about the ecacy and safety of the then-licensed type 2
diabetes treatments. The UKPDS randomly assigned
4209 of 5102 patients with newly-diagnosed type 2
diabetes to one of two dierent glucose-lowering
strategies. The 1138 patients randomly assigned to the
conventional treatment group were allocated to diet
alone, with secondary randomisation to active glucoselowering treatments only if their fasting plasma glucose
subsequently became greater than 15 mmol per L. The
3867 patients randomly assigned to the intensive
treatment group were allocated to monotherapy with
insulin (n=1156), sulfonylurea (n=1573), or metformin
(only in those with an ideal bodyweight >120%, n=342).
Further rescue treatments were added in both trial
groups in the event that fasting plasma glucose
concentrations again exceeded 15 mmol per L. A highly
signicant 25% relative risk reduction (p=00099) was
noted in microvascular disease (retinopathy necessitating
photocoagulation, vitreous haemorrhage, or fatal or nonfatal renal failure, or a combination) after a median
2010

follow-up of 10 years in those allocated to intensive

treatment with insulin or sulfonylurea. For myocardial
infarction, relative risk reduction was almost signicant
at 16% (p=0052).3 However, after 10 years of post-trial
observation, this dierence became statistically
signicant with a 15% relative risk reduction for
myocardial infarction (p=001), and an emerging 13%
relative risk reduction for all-cause mortality (p=0007).27
The trialists interpreted these ndings as showing a socalled legacy eect of the earlier allocation to intensive
glucose control within the trial period. In patients with
an ideal bodyweight greater than 120% randomly
assigned at diagnosis to metformin, a 39% relative risk
reduction occurred in fatal and nonfatal myocardial
infarction (p=0010), and a 36% relative risk reduction in
all-cause mortality (p=0011).12 These welcome but
unexpected ndings, albeit in a small cohort (342 allocated
metformin, 411 allocated diet), have helped support the
adoption by most diabetes management guidelines of
metformin as a foundation treatment for type 2 diabetes.28
Regrettably, no further cardiovascular outcome trials of
metformin have been commissioned until now, with the
Glucose Lowering in non-diabetic hyperglycaemia Trial
study about to commence (ISRCTN 34875079).
Results of the 1997 Diabetes Insulin-Glucose in Acute
Myocardial Infarction (DIGAMI) study of patients with
type 2 diabetes and acute myocardial infarction29 showed a
30% decrease in 1 year mortality with intensive insulin
treatment, compared with usual care. The DIGAMI-2
trial13 randomly assigned a similar population to acute
insulin followed by long-term insulin treatment,
acuteinsulin followed by standard long-term treatment, or
routine management, but did not show a dierence in the
primary all-cause mortality outcome. Although the
interpretation of these two trials is controversial, notably in
DIGAMI-2 the predened degree of separation in glucose
control between groups could not be achieved, and only
1253 of the necessary 3000 patients were recruited.
The ACCORD,16 ADVANCE,15 and VADT18 trials,
published in 2008, and 2009, each compared intensive
versus standard glucose-lowering strategies. Despite their
large sample sizes and up to 5 years of follow-up, all three
trials showed only non-signicant trends towards a
reduced risk in their primary cardiovascular composite
endpoints with allocation to intensive therapy. However, a
meta-analysis of these three trials, done together with the
rst 5 years of follow-up data from the UKPDS, showed a
signicant 15% risk reduction in fatal and non-fatal
myocardial infarction (95% CI 1416%).30 Results of the
ACCORD trial showed a signicant decrease in the rate of
non-fatal myocardial infarction, but was stopped
prematurely because of an unexpected 22% increased
relative risk of all-cause mortality from cardiovascular
deaths in the intensive treatment group (95% CI 146%).
Despite many analyses, investigators have not shown any
conclusive explanatory relationships between the mortality
dierence and weight gain, diabetes duration, or
www.thelancet.com Vol 383 June 7, 2014

Series

polypharmacy, although it is well known that identication

of an empirical causal pathway within a clinical trial can
be dicult. Patients in the intensive treatment group in
which more deaths occurred had increased concentrations
of glycaemia.31 Post-hoc analyses of ACCORD, ADVANCE,
and ORIGIN have shown that episodes of severe
hypoglycaemia are associated with adverse future
cardiovascular outcome for the overall populations,3234 but
between-treatment group analyses in ACCORD conclude
that previous episodes of severe hypoglycaemia do not
account for the mortality dierence recorded in this trial.
The Hyperglycemia and Its Eect After Acute Myocardial
Infarction on Cardiovascular Outcomes in Patients With
Type 2 Diabetes Mellitus (HEART 2D) trial compared
basal and prandial insulin treatment strategies,17 and the
Bypass Angioplasty Revascularization Investigation 2
Diabetes (BARI 2D) trial compared insulin-sensitising and
insulin-providing treatment strategies in patients with
type 2 diabetes and cardiovascular disease.20 Neither trial
found a dierence in primary cardiovascular composite
endpoints between treatment groups.
The PROspective pioglitAzone Clinical Trial In
macroVascular Events (PROactive),14 published in 2005,
compared the eect of the addition of pioglitazone versus
placebo to usual diabetes care. A non-signicant trend
(p=0095) for a 10% relative risk reduction was recorded
in the primary composite cardiovascular endpoint (allcause mortality, non-fatal myocardial infarction, stroke,
acute coronary syndrome, endovascular or surgical
intervention on the coronary or leg arteries, or
amputations above the ankle). However, raised rates of
heart failure and peripheral fractures in women,
combined with a small but worrying excess of bladder
cancer, are concerns for this thiazolidinedione
compound.35,36
Results of several inter-related meta-analyses of
rosiglitazone trials have suggested that rosiglitazone
might increase the risk of myocardial infarction and
heart failure.23,24 These ndings have been confounded by
allegations of improper study conduct and many
conicting observational studies, some of which
suggested harmful eects. The only substantive
randomised controlled cardiovascular outcome trial of
rosiglitazone in type 2 diabetes is the Rosiglitazone
Evaluated for Cardiac Outcomes and Regulation of
Glycaemia in Diabetes (RECORD) trial.19 In this postmarketing study, no excess risk of cardiovascular events
was noted, other than heart failure, and meta-analyses
including RECORD have not shown convincing harm
beyond the previously known excess risk of heart failure.37
RECORD was criticised for its open-label design and its
low statistical power, especially because the cardiovascular
event rate was substantially lower than was anticipated.19
The Thiazolidinedione Intervention with Vitamin D
Evaluation TIDE Trial,38 commissioned by the FDA, was
designed to directly compare rosiglitazone, pioglitazone,
and placebo when added double-blind to usual care. The
www.thelancet.com Vol 383 June 7, 2014

study was terminated before any meaningful results

could be obtained because of controversy about the safety
of thiazolidinediones and the tight restrictions made to
the rosiglitazone label. The FDA has since relaxed their
rosiglitazone restrictions on the basis of a thorough reexamination of all the data, including a re-adjudication of
the RECORD endpoints. Unfortunately, no denitive
cardiovascular outcome data exist and no new trials are
planned.39
The Outcome Reduction with an Initial Glargine
Intervention (ORIGIN) trial randomly assigned
12 537 patients with type 2 diabetes, impaired fasting
glucose, or impaired glucose tolerance with additional
cardiovascular disease risk factors to insulin glargine or to
standard glucose control.22 No dierence was recorded
between these two groups in the primary cardiovascular
disease outcome. Reassuringly, no increased risk of cancer
morbidity or mortality with insulin glargine was noted, as
epidemiological data had previously suggested.40,41
The Examination of Cardiovascular Outcomes with
Alogliptin versus Standard of Care (EXAMINE),24 and the
Saxagliptin Assessment of Vascular Outcomes Recorded
in Patients with Diabetes Mellitus (SAVOR)
Thrombolysis in Myocardial Infarction (TIMI) 53 study,23
both published in 2013, were the rst two major
cardiovascular outcome trials spawned by the 2008 FDA
cardiovascular safety regulations to be reported. These
dipeptidyl-peptidase-4 inhibitor trials both showed noninferiority with regard to cardiovascular safety, compared
with placebo, when added to usual diabetes care. These
ndings overturned previous systematic reviews of
small-scale dipeptidyl-peptidase-4 inhibitor trials, the
results of which had suggested cardiovascular benet.42,43
The analysis of pre-specied SAVOR Trial secondary
endpoints showed an unexpected 27% signicantly
increased relative risk of hospitalisation for heart failure
in patients treated with saxagliptin (p=0007).23

Trials not exclusively investigating glucoselowering drugs or strategies

ADDITION, a cluster-randomised trial, compared
multifactorial intensive treatment with routine care in
screen-detected type 2 diabetes.21 The results showed a
non-signicant 17% relative risk reduction in the
composite cardiovascular endpoint (95% CI 355%) in
favour of multifactorial intensive risk-factor management.
This result might be because only marginal dierences in
cardiovascular risk-factor levels were achieved between
the two groups.
The LOOK-AHEAD trial compared the eect of
randomly assigning people with type 2 diabetes to an
intensive lifestyle intervention aiming for at least 70%
weight loss, or to a diabetes support and education
programme.25 Despite a greater weight reduction in the
intensive lifestyle intervention group than in the
support and education programme group throughout
the study, the trial was stopped because of futility after a
2011

Series

median follow-up of 96 years with no dierence in the

composite cardiovascular primary outcome (HR for
intensive lifestyle intervention 095; 95% CI 083109).
This dierence might be because of greater use of
proven risk-factor-reducing drugs in the diabetes
education and support group.

Ongoing cardiovascular outcome trials in type 2

diabetes
Increased research
The 2008 FDA and European Medicines Agency guidance
requiring proof of cardiovascular safety in addition to
HbA1c lowering, as a prerequisite for approval of new
antidiabetic drugs, changed the type 2 diabetes clinical

TECOS

trial landscape. Initial concerns that the costs of such

large-scale trials might deter pharmaceutical companies
from further investing in the diabetes specialty were
proved wrong, with at least 16 cardiovascular outcome
trials (that will report by 2020) ongoing in more than
150 000 patients with type 2 diabetes (table 2). These
studies include four dipeptidyl-peptidase-4 inhibitor
trials, six glucagon-like peptide(GLP)-1 receptor agonist
trials, four sodium-glucose linked transporter 2-inhibitor
trials, one trial comparing a thiazolidinedione with a
sulfonylurea, and one trial comparing two long-acting
insulin preparations. Driven by the new regulatory
requirements, more cardiovascular outcome trials in
type 2 diabetes have been commenced in the past 10 years

Intervention

Population

Primary outcome

Study status

Start and estimated

end date

ClinicalTrials.
gov identier

14 000

Sitagliptin versus placebo

T2DM; HbA1c 6580%; 50 years;

CVD history

CV death, MI, UA, or stroke

Ongoing,
not recruiting

12/2008Q3/2014

NCT00790205

TOSCA IT

3371

Pioglitazone versus
sulfonylurea

T2DM; HbA1c 70% and 90%;

metformin monotherapy

Death, MI, stroke or

coronary revascularisation

Recruiting

09/200812/2018

NCT00700856

CANVAS

4330

Canagliozin 100 mg
versus canagliozin
300 mg versus placebo

T2DM; 30 years; HbA1c 70105%;

History of/high risk of CVD

CV death, MI, UA, or stroke

Ongoing,
not recruiting

12/200903/2017

NCT01032629

ELIXA

6000

Lixisenatide versus
placebo

T2DM; HbA1c 55110%; ACS

CV death, MI, UA, or stroke

Ongoing,
not recruiting

06/201001/2015

NCT01147250

14 000

Exenatide once weekly

versus placebo

T2DM; HbA1c 65100%; CVD in

about 60%

CV death, MI, or stroke

Recruiting

06/201012/2017

NCT01144338

BI 10773 trial

7000

Empagliozin 10 mg
versus empagliozin
25 mg versus placebo

T2DM; 18 years; HbA1c 70100%;

(7090% drug nave); high CV risk

CV death, MI, or stroke

Ongoing,
not recruiting

07/201003/2018

NCT01131676

LEADER

9340

Liraglutide versus placebo

T2DM; HbA1c 70%; 50 years+CVD; CV death, MI, or stroke

60 years+CV risk factors

Ongoing,
not recruiting

08/201010/2015

NCT01179048

CAROLINA

6000

Linagliptin versus
glimepiride

T2DM; HbA1c 6585%; 4085 years; CV death, MI, UA, or stroke

CVD/CV risk factors/diabetes end
organ damage

Ongoing,
not recruiting

10/201009/2018

NCT01243424

REWIND

9622

Dulaglutide versus
placebo

T2DM; HbA1c 95%; 50 years+CVD;

55 years+subclinical CVD;
60 years+CV risk factors

Ongoing,
not recruiting

07/201104/2019

NCT01394952

MK-3102 trial

4000

MK-3102 versus placebo

T2DM; CVD history

CV death, MI, UA, or stroke

Recruiting

10/201210/2017

NCT01703208

SUSTAIN6

3260

Semaglutide 05 mg
versus semaglutide
10 mg versus placebo

T2DM; HbA1c 70%;

age 50 years+CVD;
age 60 years+subclinical CVD

CV death, MI, or stroke

Ongoing,
not recruiting

02/201301/2016

NCT01720446

ITCA650 trial

2000

ITCA 650 (exenatide in

DUROS) versus placebo

T2DM; HbA1c >65%; CVD history

CV death, MI, UA, or stroke

Recruiting

03/201307/2018

NCT01455896

Dapagliozin 10 mg
versus placebo

T2DM; 40 years; high CV risk

CV death, MI, or stroke

Recruiting

04/201304/2019

NCT01730534

Recruiting

07/201301/2018

NCT01897532

EXSCEL

DECLARE-TIMI 58

17 150

CV death, MI, or stroke

CARMELINA

8300

Linagliptin versus placebo

CV death, MI, UA, or stroke

T2DM; HbA1c 6510%; 18 years;
microalbuminuri or
macroalbuminuria and previous
macrovascular disease; impaired
renal function with predened UACR

DEVOTE

7500

Insulin degludec versus

insulin glargine

T2DM; HbA1c 70%; or HbA1c

70% and insulin treatment;
50 years and CV or renal disease
or 60 years and CV risk factors

CV death, MI or stroke

Recruiting

10/201311/2018

NCT01959529

Ertugliozin trial

3900

Ertugliozin 5 mg versus
ertugliozin 15 mg versus
placebo

T2DM; HbA1c 70 105%;

CVD history

CV death, MI, or stroke

Recruiting

11/201306/2020

NCT01986881

T2DM=type 2 diabetes. CVD=cardiovascular disease. CV=cardiovascular. MI=myocardial infarction. UA=unstable angina. ACS=acute coronary syndrome.

Table 2: Ongoing cardiovascular outcome trials of glucose-lowering drugs or strategies (in order of starting date)

2012

www.thelancet.com Vol 383 June 7, 2014

Series

Possible pleotropic eects

With so many new treatment methods being assessed, a
clear need exists to be alert for possible non-glycaemic
benets, such as improvements in blood pressure, lipid
proles, and bodyweight with GLP-1 receptor agonists,
and the potential for o-target eects leading to
unexpected benecial or harmful outcomes. Equally, if
glucose-lowering per se does reduce cardiovascular
disease risk, the need to ensure glycaemic equipoise in
studies of novel drugs will be of paramount importance
in understanding of the results obtained.

Need for long-term follow-up

The need for ongoing trials to satisfy new regulatory
requirements has led to a design culture of risk-avoidance
studies, coupled with a need for trials to be completed
within the shortest possible time to maximise nancial
return on pharmaceutical investment. Type 2 diabetes is a
chronic, progressive disease, with patients at increasing
risk over time of both microvascular and macrovascular
complications. The vascular eects of glucose-lowering
treatments took about 5 years to emerge in both the
UKPDS3 and the DCCT,4 and other eects such as cancer
and fractures will probably take a similar period to be
evident compared with placebo or comparators. These
ndings suggest that, although short-term trials can
gather information on rapidly occurring adverse events,
such as hypoglycaemia, infections, or peripheral oedema,
medium-to-long-term trials could be essential in type 2
diabetes to obtain a fuller appreciation of the probable
benets and risks of glucose-lowering interventions. That
most ongoing cardiovascular outcome trials are being
carried out in populations at high cardiovascular risk who
have had type 2 diabetes for many years is also a concern,
perhaps limiting the trials ability to assess treatments
that might be more eective for primary cardiovascular
disease prevention, and identifying potentially greater
benets of intervention earlier in the progression of
diabetes.27 The median follow-up of 15 years in the
EXAMINE trial and 21 years in the SAVOR-TIMI 53 trial
are testament to this concern. To motivate companies to
invest in long-term trials, nancial incentives, such as
oering a longer patent life, could be considered.
The number of trial participants needed to achieve the
number of primary events ordained by the sample size
calculation needs to be balanced by a median duration of
follow-up to fully capture the probable benets and risks.
Age and disease severity-adjusted cardiovascular event
rates are decreasing worldwide as evidence-based
www.thelancet.com Vol 383 June 7, 2014

management of cardiovascular risk factors increases. In

view of the lower event rates, the need for enough events
to exclude a clinically meaningful excess of cardiovascular
events, and the cardiovascular outcome studies driven by
nancial time pressure are required to enrol larger
cohorts with more extensive pre-existing disease. Shortterm follow-up might be satisfactory when seeking to
rule out direct cardiotoxicity, but short trials have little
ability to fully assess potential long-term cardiovascular
risks or benets, or other type 2 diabetes complications.
Equally, extended follow-up is necessary to examine the
possible eect on incident cancer or emergence of
unanticipated side-eects, as in the detection in the
ADOPT trial of the unanticipated increased risk of bone
fractures in women.45

Scarcity of active comparators

A particular strength of the UKPDS was its head-to-head
comparison of all licensed type 2 diabetes drugs during the
trial period. Unfortunately, most new trials are testing
novel drugs primarily for non-inferiority against placebo
when administered in addition to usual diabetes care,
thereby limiting their clinical interpretation, and providing
no direct comparison with existing treatments. In this way
the CAROLINA trial stands out, being an active comparator
study of linagliptin versus glimepiride. Unfortunately,
CAROLINA has no placebo group to calibrate any ndings
to the population being studied. If linagliptin were shown
to be superior to glimepiride it would not be known
whether linagliptin would be better than usual care
(assuming glimepiride is the benchmark), or the same as
usual care in the event that glimepiride has an adverse
cardiovascular prole as the UGDP sulfonylurea data
suggested.26 The FDA noted this omission and required
the company to run an additional placebo-controlled trial
250
Cumulative number of cardiovascular trial participants
(thousands)

than in the entire history of diabetes research (gure).

Trial sample sizes have also substantially increased from
median (IQR) 1116 (3004447) to 6000 (40829313), and
an increase in the average number (range) of countries
per trial from 17 (120) to 27 (635),25,44 providing data for
a wider range of racial and ethnic groups and variations
in practice patterns than previously.

200

150

100

50

0
1996

2000

2004

2008
Year

2012

2016

2020

Figure: Cumulative number of participants in cardiovascular outcome trials over time

Numbers of trial participants are added at the time of publication for historical trials (solid line) and at the
estimated time of reporting for ongoing trials (dotted line). The red circle indicates when the new US Food and
Drug Administration guidance for industry was issued.

2013

Series

(CARMELINA) to comply with their safety requirements.46

The DEVOTE trial is comparing insulin degludec directly
with insulin glargine, after FDA cardiovascular safety
concerns raised by post-hoc analyses of earlier regulatory
studies,47 but a placebo group is not feasible in view of the
need to maintain acceptable levels of glycaemia in patients
who need insulin.
The ongoing GRADE trial (ClinicalTrials.gov:
NCT01794143) is a comparative eectiveness trial, examining the randomised addition of a DPP-4 inhibitor, a
GLP-1 receptor agonist, a sulfonylurea, or long-acting
insulin to metformin. Unfortunately the trial is not
powered for cardiovascular outcomes, with the primary
outcome limited to time to second-line treatment failure.

Challenges for cardiovascular outcome trials

going forward
A downside of long-term trials is that retention and
adherence are increasingly dicult over time to
maintain. Retention and adherence rates vary
substantially among the trials we describe, with no clear
trend or pattern, but with permanent drug discontinuation rates reaching about 10% per year in
EXAMINE38 and SAVOR-TIMI 53.39 Although these drug
discontinuation rates are consistent with clinical practice
patterns, the traditional clinical trial method seeks to
optimise adherence in trials more so than in clinical
practice. Several challenges exist to participant retention
and adherence: unwillingness to stick to a protocol for
several years; the requirement to travel regularly to study
sites; concerns about absence of eective treatment if
participants perceive that they might be taking a placebo;
open-label availability of the study drug as a routine
treatment option, especially if reimbursed; and life
events that might aect patients ability or willingness to
continue in a trial. It is crucial that trial managers choose
participating sites carefully, using previous retention
and adherence metrics as a key selection parameter.
Sites need to be supported during the trial by adequate
nancial and personal resources to retain participants
over an extended period. Expert opinion from the
Institute of Medicine48 and publicly expressed FDA
opinion point out the analytical dilemma caused by
missing data, which introduces an unresolvable potential
bias to analyses. Although discontinuation of therapy
often represents the appropriate therapeutic option or
patient choice, maintainenance of follow-up and
measurement of outcomes to the fullest extent possible
in all study participants is essential. A misconception
often exists among investigators and study coordinators,
possibly fuelled by experience in early-phase trials, that
participants in clinical outcome trials who discontinue
study medication need no longer be followed.49 Constant
emphasis is crucial to ensure that all participants
continue to be followed as closely as possible until study
end. Vital status should always be available, because in
previous trials where legal opinion was sought the view
2014

was that no right of condentiality about death exists

because the death is part of the public record.50 True
withdrawal of consent to follow-up should be exceptional,
needs to be discussed in detail with participants, and
with written records obtained should they genuinely
wish to withdraw completely from the study.

Concepts for future outcome trials in patients

with type 2 diabetes
The concept of randomised controlled trials is the gold
standard in clinical research to investigate outcomes
attributable to a certain intervention, because of the
many advantages of randomisation over mere registry
analyses. Intervention allocation is too important to leave
to chance. Randomisation helps to ensure that the groups
being compared are balanced in terms of the known and
unknown confounders, which information is not
accessible to registry analyses.
However, incremental changes in trial methods will
not be sucient to address the deciencies of the present
generation of clinical trials assessing clinical outcomes
in patients with type 2 diabetestransformational
change is needed. The present approach produces trials
that are too short, with measurement of limited outcomes
at a high cost per patient. The genesis of these costs is a
clinical trials system that initiates studies and collects
data outside the routine system of clinical care, and relies
on audit functions rather than quality by design to
attempt to ensure reliable results. These costs can only
be borne by industry as a function of its need to obtain
approval for marketing. The huge market for type 2
diabetes drugs has enabled industry to bear these costs,
but the situation prevents trials of adequate length from
being truly informative for a lifelong disease, inhibits
head-to-head comparisons and factorial designs, and
constrains the exploration of more comprehensive
clinical outcome assessments.
Many countries are developing comprehensive
electronic health record systems which enable various
functions that will greatly reduce the cost of trials and
enable comprehensive follow-up within the context of
the health-care delivery system. Using so-called
computable phenotypes derived from diagnostic codes
in electronic health records, populations with diabetes
can be identied, baseline characteristics produced,
and follow-up events ascertained. Before these trials
can begin, various measures must be taken to engage
health system leaders, clinicians, and patients, in more
of a systematic approach to learning about best
therapies in practice. Regulatory systems must also be
adapted to develop a more eective approach to quality
by design. By reducing the cost of trials by an order of
magnitude or more, these approaches could open up
the system to the kinds of trials that would provide the
much-needed information. The realistic probability of a
transition in trial conduct is emphasised by the
formation of the Patient Centered Outcomes Research
www.thelancet.com Vol 383 June 7, 2014

Series

Network in the USA, with up to 100 million Americans

in a network designed to do these types of trials.51 Proofof-concept has been achieved in several cardiovascular
outcomes trials.5254
Future trials should plan on long-term follow up,
factorial designs, and embedded head-to-head comparisons. An especially interesting design beginning to
be used in many trials in other disease areas begins
with many groups that are within standard of care.
Furthermore, continuous follow-up with electronic
health records will enable multidimensional collection
of various outcomes, including cancer, fractures, and
psychiatric and neurological outcomes. Sample sizes
can also be much larger because of the lower cost per
patient when followed in usual care, thus enabling
powerful assessments of heterogeneity of treatment
eect. If dierent classes of drugs are found to be better
for patients with dierent characteristics, such trials
will be crucial in realisation of the aspiration for
personalised medication.
Another important element of future trials will be the
direct involvement of patients and their advocates in the
prioritisation, design, conduct, analysis, and dissemination
of trials.55 This, combined with the wide-scale availability
of personal monitoring devices, should enable a range of
microvascular outcomes, symptoms, and quality-of-life
factors to be measured.56 Because advocacy organisations
are increasingly realising the importance of clinical
research to guide decision making about best treatment,
direct involvement of many aected individuals should
foster the broad-scale enrolment of many participants
prepared to share data to advance knowledge of diabetes
and its treatment.57 The resulting mega-networks should
also enable replication of ndings across networks to
ensure the validity of observations.
This new world of large-scale, long-term trials will need
the development of methods for assessment of several
comparisons, appropriate handling of missing data, and
examination of time-dependent variations in the benetto-risk balance. As increasingly relevant populations are
enrolled in real-world settings across dierent countries
and cultures, the heterogeneity of treatment eect will
become a multidimensional construct.58 This new
approach to clinical trials will need to be phased in as
systems develop and methods are tested and validated. In
the interim, hybrid trials, in which electronic health
records data are used to identify cohorts and measure
some outcomes, will become more common.

Conclusions
A critical view of the large cardiovascular trials done over
the last four decades shows that, until recently, few had
adequate statistical power to inform practice, and that the
designs of the trials constrained the range of questions
that could be addressed. The UKPDS taught us that
improved glycaemic control in newly-diagnosed type 2
diabetes patients reduces the risk of microvascular
www.thelancet.com Vol 383 June 7, 2014

complications, and, in the long term, of cardiovascular

events and all-cause mortality.3,7 Findings from
ACCORD,11 ADVANCE,10 and VADT13 showed that
medium-term trials of intensive glucose lowering, in
participants with long-established diabetes and with a
history of cardiovascular disease, did not show similar
benecial cardiovascular eects. A clear need exists for
better-designed trials with sucient length of follow-up
to more eectively address the crucial questions of the
comparative balance of risk and benet during
substantial periods of follow-up. Smarter trials are
necessarysuch as factorial designs and trials nested
into the rapidly evolving environment of electronic
health records and data warehousesthat permit novel
treatments to be compared more appropriately and more
cost-eectively with existing treatments, and that seek to
answer a wider range of clinically-relevant questions
than do present trials.
Contributors
HS did the search, wrote the rst draft, and revised the Series paper.
RRH and RMC planned and revised the Series paper.
Declaration of interests
RRH received grants from Bristol-Myers Squibb, grants and personal fees
from Bayer and Merck, and personal fees from Novartis and Jansen, all
outside the submitted work. RMC received grants from Amylin, Novartis,
Schering-Plough Research Institute, Scios, Eli Lilly, Johnsons & Johnson/
Scios, Aterovax, Bayer, the NIH, and the Patient-Centered Outcomes
Research Institute; grants and personal fees from Bristol-Myers Squibb,
Janssen Research & Development, Merck, and Roche; personal fees from
Genentech, GlaxoSmithKline, Heart.org/Daiichi Sankyo, Kowa, Servier,
Medscape, Regeneron, TMC, Pzer, Gambro, Gilead, DSI-Lilly, CV Sight,
Heart.org/Bayer, Bayer Pharma AG, Bayer Healthcare, Parkview, Pozen,
Orexigen, Nile, and WebMD; and other nancial support from N30
Pharma, Portola, and Nitrox LLC, all outside the submitted work.
HS declares no conicts of interest.
References
1
International Diabetes Federation. Diabetes Atlas. http://www.idf.
org/diabetesatlas (accessed Feb 23, 2014).
2
Seshasai SR, Kaptoge S, Thompson A, et al, and the Emerging Risk
Factors Collaboration. Diabetes mellitus, fasting glucose, and risk of
cause-specic death. N Engl J Med 2011; 364: 82941.
3
UK Prospective Diabetes Study (UKPDS) Group. Intensive
blood-glucose control with sulphonylureas or insulin compared
with conventional treatment and risk of complications in patients
with type 2 diabetes (UKPDS 33). Lancet 1998; 352: 83753.
4
Nathan DM, Cleary PA, Backlund JY, et al, and the Diabetes
Control and Complications Trial/Epidemiology of Diabetes
Interventions and Complications (DCCT/EDIC) Study Research
Group. Intensive diabetes treatment and cardiovascular disease in
patients with type 1 diabetes. N Engl J Med 2005; 353: 264353.
5
Siperstein MD, Unger RH, Madison LL. Studies of muscle capillary
basement membranes in normal subjects, diabetic, and prediabetic
patients. J Clin Invest 1968; 47: 197399.
6
UK Prospective Diabetes Study Group. Tight blood pressure control
and risk of macrovascular and microvascular complications in
type 2 diabetes: UKPDS 38. BMJ 1998; 317: 70313.
7
US Department of Health and Human Services, Food and Drug
Administration, Center for Drug Evaluation and Research (CDER).
Guidance for industry: diabetes mellitusevaluating cardiovascular
risk in new antidiabetic therapies to treat type 2 diabetes. http://www.
fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ucm071627.pdf (accessed Feb 23, 2014).
8
European Medicines Agency. Guideline on clinical investigation of
medicinal products in the treatment or prevention of diabetes
mellitus. May 2012. http://www.ema.europa.eu/docs/en_GB/
document_library/Scientic_guideline/2012/06/WC500129256.pdf
(accessed Feb 23, 2014).

2015

Series

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

2016

Nissen SE, Wolski K. Eect of rosiglitazone on the risk of

myocardial infarction and death from cardiovascular causes.
N Engl J Med 2007; 356: 245771.
Nissen SE, Wolski K. Rosiglitazone revisited: an updated
meta-analysis of risk for myocardial infarction and cardiovascular
mortality. Arch Intern Med 2010; 170: 1191201.
Bethel MA, Sourij H. Impact of FDA guidance for developing
diabetes drugs on trial design: from policy to practice.
Curr Cardiol Rep 2012; 14: 5969.
UK Prospective Diabetes Study (UKPDS) Group. Eect of intensive
blood-glucose control with metformin on complications in overweight
patients with type 2 diabetes (UKPDS 34). Lancet 1998; 352: 85465.
Malmberg K, Rydn L, Wedel H, et al, and the DIGAMI 2
Investigators. Intense metabolic control by means of insulin in
patients with diabetes mellitus and acute myocardial infarction
(DIGAMI 2): eects on mortality and morbidity. Eur Heart J 2005;
26: 65061.
Dormandy JA, Charbonnel B, Eckland DJ, et al, and the PROactive
investigators. Secondary prevention of macrovascular events in
patients with type 2 diabetes in the PROactive Study (PROspective
pioglitAzone Clinical Trial In macroVascular Events): a randomised
controlled trial. Lancet 2005; 366: 127989.
Patel A, MacMahon S, Chalmers J, et al, and the ADVANCE
Collaborative Group. Intensive blood glucose control and vascular
outcomes in patients with type 2 diabetes. N Engl J Med 2008;
358: 256072.
Gerstein HC, Miller ME, Byington RP, et al, and the Action to
Control Cardiovascular Risk in Diabetes Study Group. Eects of
intensive glucose lowering in type 2 diabetes. N Engl J Med 2008;
358: 254559.
Raz I, Wilson PW, Strojek K, et al. Eects of prandial versus fasting
glycemia on cardiovascular outcomes in type 2 diabetes: the
HEART2D trial. Diabetes Care 2009; 32: 38186.
Duckworth W, Abraira C, Moritz T, et al, and the VADT
Investigators. Glucose control and vascular complications in
veterans with type 2 diabetes. N Engl J Med 2009; 360: 12939.
Home PD, Pocock SJ, Beck-Nielsen H, et al, and the RECORD
Study Team. Rosiglitazone evaluated for cardiovascular outcomes
in oral agent combination therapy for type 2 diabetes (RECORD):
a multicentre, randomised, open-label trial. Lancet 2009;
373: 212535.
Frye RL, August P, Brooks MM, et al, and the BARI 2D Study
Group. A randomized trial of therapies for type 2 diabetes and
coronary artery disease. N Engl J Med 2009; 360: 250315.
Grin SJ, Borch-Johnsen K, Davies MJ, et al. Eect of early
intensive multifactorial therapy on 5-year cardiovascular outcomes
in individuals with type 2 diabetes detected by screening
(ADDITION-Europe): a cluster-randomised trial. Lancet 2011;
378: 15667.
Gerstein HC, Bosch J, Dagenais GR, et al, and the ORIGIN Trial
Investigators. Basal insulin and cardiovascular and other outcomes
in dysglycemia. N Engl J Med 2012; 367: 31928.
Scirica BM, Bhatt DL, Braunwald E, et al, and the SAVOR-TIMI 53
Steering Committee and Investigators. Saxagliptin and
cardiovascular outcomes in patients with type 2 diabetes mellitus.
N Engl J Med 2013; 369: 131726.
White WB, Cannon CP, Heller SR, et al, and the EXAMINE
Investigators. Alogliptin after acute coronary syndrome in patients
with type 2 diabetes. N Engl J Med 2013; 369: 132735.
Wing RR, Bolin P, Brancati FL, et al, and the Look AHEAD
Research Group. Cardiovascular eects of intensive lifestyle
intervention in type 2 diabetes. N Engl J Med 2013; 369: 14554.
Meinert CL, Knatterud GL, Prout TE, Klimt CR. A study of the
eects of hypoglycemic agents on vascular complications in
patients with adult-onset diabetes. II. Mortality results. Diabetes
1970; 19 (suppl): 789830.
Holman RR, Paul SK, Bethel MA, Matthews DR, Neil HA. 10-year
follow-up of intensive glucose control in type 2 diabetes.
N Engl J Med 2008; 359: 157789.
Inzucchi SE, Bergenstal RM, Buse JB, et al. Management of
hyperglycaemia in type 2 diabetes: a patient-centered approach.
Position statement of the American Diabetes Association (ADA)
and the European Association for the Study of Diabetes (EASD).
Diabetologia 2012; 55: 157796.

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

Malmberg K. Prospective randomised study of intensive insulin

treatment on long term survival after acute myocardial infarction in
patients with diabetes mellitus. DIGAMI (Diabetes Mellitus,
Insulin Glucose Infusion in Acute Myocardial Infarction) Study
Group. BMJ 1997; 314: 151215.
Turnbull FM, Abraira C, Anderson RJ, et al, and the Control Group.
Intensive glucose control and macrovascular outcomes in type 2
diabetes. Diabetologia 2009; 52: 228898.
Riddle MC, Ambrosius WT, Brillon DJ, et al, and the Action to
Control Cardiovascular Risk in Diabetes Investigators.
Epidemiologic relationships between A1C and all-cause mortality
during a median 3.4-year follow-up of glycemic treatment in the
ACCORD trial. Diabetes Care 2010; 33: 98390.
Bonds DE, Miller ME, Bergenstal RM, et al. The association
between symptomatic, severe hypoglycaemia and mortality in type 2
diabetes: retrospective epidemiological analysis of the ACCORD
study. BMJ 2010; 340: b4909.
Zoungas S, Patel A, Chalmers J, et al, and the ADVANCE
Collaborative Group. Severe hypoglycemia and risks of vascular
events and death. N Engl J Med 2010; 363: 141018.
Mellbin LG, Rydn L, Riddle MC, et al, and the ORIGIN Trial
Investigators. Does hypoglycaemia increase the risk of cardiovascular
events? A report from the ORIGIN trial. Eur Heart J 2013; 34: 313744.
Ferwana M, Firwana B, Hasan R, et al. Pioglitazone and risk of
bladder cancer: a meta-analysis of controlled studies. Diabet Med
2013; 30: 102632.
Dormandy J, Bhattacharya M, van Troostenburg de Bruyn AR,
investigators PROactive. Safety and tolerability of pioglitazone in
high-risk patients with type 2 diabetes: an overview of data from
PROactive. Drug Saf 2009; 32: 187202.
US Department of Health and Human Services, Food and Drug
Administration. Brieng Document: Readjudication of the
rosiglitazone evaluated for cardiovascular outcomes and regulation of
glycemia in diabetes trial (RECORD). Joint Meeting of the
Endocrinologic and metabolic drugs advisory committee and the drug
safety and risk management advisory committee, June 56, 2013.
http://www.fda.gov/downloads/advisorycommittees/
committeesmeetingmaterials/drugs/endocrinologicandmetabolicdrugsadvisorycommittee/ucm354859.pdf (accessed Feb 23, 2014).
Punthakee Z, Bosch J, Dagenais G, et al, and the TIDE Trial
Investigators. Design, history and results of the Thiazolidinedione
Intervention with vitamin D Evaluation (TIDE) randomised
controlled trial. Diabetologia 2012; 55: 3645.
US Department of Health and Human Services, Food and Drug
Administration. FDA press release: FDA requires removal of
certain restrictions on the diabetes drug Avandia. http://www.fda.
gov/NewsEvents/Newsroom/PressAnnouncements/ucm376516.
htm (accessed Feb 23, 2014).
Hemkens LG, Grouven U, Bender R, et al. Risk of malignancies in
patients with diabetes treated with human insulin or insulin
analogues: a cohort study. Diabetologia 2009; 52: 173244.
Jonasson JM, Ljung R, Talbck M, Haglund B, Gudbjrnsdttir S,
Steineck G. Insulin glargine use and short-term incidence of
malignancies-a population-based follow-up study in Sweden.
Diabetologia 2009; 52: 174554.
Frederich R, Alexander JH, Fiedorek FT, et al. A systematic
assessment of cardiovascular outcomes in the saxagliptin drug
development program for type 2 diabetes. Postgrad Med 2010;
122: 1627.
Monami M, Iacomelli I, Marchionni N, Mannucci E. Dipeptydil
peptidase-4 inhibitors in type 2 diabetes: a meta-analysis of
randomized clinical trials. Nutr Metab Cardiovasc Dis 2010;
20: 22435.
Bethel MA, Sourij H. Positive impact of revised FDA guidance on
clinical trial design in diabetes (abstr). Diabetes 2012;
61 (supp 1): A264.
Kahn SE, Zinman B, Lachin JM, et al, and the Diabetes Outcome
Progression Trial (ADOPT) Study Group. Rosiglitazone-associated
fractures in type 2 diabetes: an Analysis from A Diabetes Outcome
Progression Trial (ADOPT). Diabetes Care 2008; 31: 84551.
US Department of Health and Human Services, Food and Drug
Administration. Summary review Linagliptin. http://www.
accessdata.fda.gov/drugsatfda_docs/nda/2011/201280Orig
1s000SumR.pdf (accessed Feb 23, 2014).

www.thelancet.com Vol 383 June 7, 2014

Series

47

48

49

50

51
52

Endocrinologic and Metabolic Drug Advisory Committee,

Novo Nordisk. Brieng document: insulin degludec and insulin
degludec/insulin aspart treatment to improve glycemic control in
patients with diabetes mellitus. http://www.fda.gov/downloads/
AdvisoryCommittees/CommitteesMeetingMaterials/Drugs/
EndocrinologicandMetabolicDrugsAdvisoryCommittee/
UCM327017.pdf (accessed Feb 23, 2014).
Little RJ, DAgostino R, Cohen ML, et al. The prevention and
treatment of missing data in clinical trials. N Engl J Med 2012;
367: 135560.
Cobo E, Stephen S, Elmore M. Withdrawing from the treatment
does not mean from the study. Bull World Health Organ 2008;
86: C.
Cleland JG, Torp-Pedersen C, Coletta AP, Lammiman MJ.
A method to reduce loss to follow-up in clinical trials: informed,
withdrawal of consent. Eur J Heart Fail 2004; 6: 12.
Selby JV, Lipstein SH. PCORI at 3 yearsprogress, lessons, and
plans. N Engl J Med 2014; 370: 59295.
Frbert O, Lagerqvist B, Olivecrona GK, et al, and the TASTE Trial.
Thrombus aspiration during ST-segment elevation myocardial
infarction. N Engl J Med 2013; 369: 158797.

www.thelancet.com Vol 383 June 7, 2014

53

54

55

56
57

58

Hess CN, Rao SV, Kong DF, et al. Embedding a randomized clinical
trial into an ongoing registry infrastructure: unique opportunities
for eciency in design of the Study of Access site For Enhancement
of Percutaneous Coronary Intervention for Women (SAFE-PCI for
Women). Am Heart J 2013; 166: 42128.
Lauer MS, DAgostino RB Sr. The randomized registry trialthe
next disruptive technology in clinical research? N Engl J Med 2013;
369: 157981.
Barker AD, Sigman CC, Kello GJ, Hylton NM, Berry DA,
Esserman LJ. I-SPY 2: an adaptive breast cancer trial design in the
setting of neoadjuvant chemotherapy. Clin Pharmacol Ther 2009;
86: 97100.
Topol EJ. Individualized medicine from prewomb to tomb. Cell
2014; 157: 24153.
Gallin EK, Bond E, Cali RM, et al. Forging stronger partnerships
between academic health centers and patient-driven organizations.
Acad Med 2013; 88: 122024.
Kent DM, Rothwell PM, Ioannidis JP, Altman DG, Hayward RA.
Assessing and reporting heterogeneity in treatment eects in
clinical trials: a proposal. Trials 2010; 11: 85.

2017