Rev Endocr Metab Disord (2010) 11:21–30
DOI 10.1007/s11154-010-9130-8
Impact of cardiovascular outcomes on the development
and approval of medications for the treatment
of diabetes mellitus
Hylton V. Joffe & Mary H. Parks & Robert Temple
Published online: 2 March 2010
# US Government 2010
Abstract All medications currently approved by the Food
and Drug Administration (FDA) for the treatment of type 2
diabetes mellitus are indicated to improve glycemic control.
Since 1995, FDA has used HbA1c as the primary basis for
approval of these therapies because a reduction in blood
glucose lessens the symptoms of hyperglycemia and
lowering of HbA1c has been shown to reduce the risk for
some of the chronic complications of diabetes. Despite
evidence of clinical benefit with therapies that reduce
HbA1c, concerns have been raised that some diabetes
medications may increase cardiovascular risk in a patient
population that is already vulnerable to cardiovascular
disease. Therefore, FDA convened a public advisory
committee meeting to discuss the role of cardiovascular
assessment in the pre-approval and post-approval settings
for medications developed for the treatment of type 2
diabetes. After considering the advisory panel’s recommen-
dations and other data, FDA published a guidance
document requesting evidence showing that new treatments
for type 2 diabetes do not result in an unacceptable increase
H. V. Joffe : M. H. Parks
Division of Metabolism and Endocrinology Products,
Center for Drug Evaluation and Research,
U.S. Food and Drug Administration,
Silver Spring, MD, USA
R. Temple
Center for Drug Evaluation and Research,
U.S. Food and Drug Administration,
Silver Spring, MD, USA
H. V. Joffe (*)
Division of Metabolism and Endocrinology Products,
U.S. Food and Drug Administration,
10903 New Hampshire Avenue, Building 22, Room 3340,
Silver Spring, MD 20993-0002, USA
e-mail: hylton.joffe@fda.hhs.gov
in cardiovascular risk. This review article begins by
summarizing the events leading up to publication of this
guidance. Subsequent sections discuss the guidance itself as
well as general considerations for implementing the new
cardiovascular recommendations. The new approach to
developing medications for the treatment of type 2 diabetes
will lead to evaluation in patients more representative of
those who will use these therapies, if approved, and will
help healthcare providers make informed decisions when
choosing a medication within the growing treatment
armamentarium for type 2 diabetes.
Keywords Diabetes mellitus . Cardiovascular disease .
Drug development . Food and Drug Administration
1 Introduction
Diabetes mellitus affects more than 170 million people
worldwide and more than 24 million people in the United
States [1, 2]. Type 2 diabetes, which accounts for
approximately 90% of cases, is increasing at alarming
rates, in part because of the obesity epidemic and
widespread physical inactivity [2]. People with diabetes
have an increased risk of cardiovascular disease. For
example, people with type 2 diabetes and no prior history
of myocardial infarction may have as high a risk of
myocardial infarction as non-diabetic people with a history
of previous myocardial infarction [3]. The risk of death
from cardiovascular disease is 2–4 times higher among
people with type 2 diabetes than among people without
diabetes and cardiovascular disease and stroke account for
approximately two-thirds of deaths in people with diabetes
[1, 4–9]. Long-term microvascular complications of diabe-
tes are also important causes of morbidity and mortality
22
[10]. In the United States, diabetes is the leading cause of
blindness, end-stage renal disease, and non-traumatic
lower-extremity amputations [1]. Neuropathy can cause
chronic pain and involvement of the autonomic nervous
system can result in other disabling conditions, such as
diabetic gastroparesis and orthostatic hypotension. Diabetes
can also lead to diabetic ketoacidosis or non-ketotic
hyperosmolar coma and treatment of diabetes can lead to
hypoglycemia, which is sometimes severe.
All medications currently approved by the U.S. Food
and Drug Administration (FDA) for the treatment of type 2
diabetes are indicated to improve glycemic control. Since
1995, these medications have been approved on the basis of
lowering hemoglobin A1c (HbA1c). HbA1c is formed by
the non-enzymatic, irreversible attachment of glucose to
hemoglobin and is directly proportional to the ambient
glucose concentration [11]. HbA1c reflects mean blood
glucose over the lifespan of the red blood cell, but
correlates better with mean blood glucose over the
preceding 12 weeks. FDA has used HbA1c as the primary
basis for approval of these therapies because a reduction in
blood glucose lessens the symptoms of hyperglycemia (e.g.,
polyuria, polydipsia) and lowering of HbA1c has been
shown to reduce the risk for some of the chronic
complications of diabetes. FDA’s decision to begin accept-
ing HbA1c as a surrogate was primarily based on the
findings from the landmark Diabetes Control and Compli-
cations Trial (DCCT), which demonstrated that intensive
glycemic control (as assessed by HbA1c) with insulin in
patients with type 1 diabetes reduces the risk of onset and
progression of retinopathy, nephropathy, and neuropathy
[12]. In 1998, the results from the United Kingdom
Prospective Diabetes Study (UKPDS) demonstrated a
similar finding in patients with type 2 diabetes [13]. In this
study, patients who were randomized to intensive glycemic
control with a sulfonylurea, insulin, or a combination of
these agents had a lower risk of microvascular complica-
tions than patients randomized to standard glycemic
control, which was initially comprised of diet and exercise.
Despite evidence of clinical benefit with therapies that
reduce HbA1c, concerns have been raised that some
diabetes medications may increase cardiovascular risk in a
patient population that is already vulnerable to cardiovas-
cular disease. For example, all sulfonylurea package inserts
contain a bolded warning regarding increased cardiovascu-
lar mortality based on a finding seen with the first-
generation sulfonylurea, tolbutamide, in the University
Group Diabetes Program [14]. The repaglinide package
insert describes a numerical imbalance in serious myocar-
dial ischemic events when repaglinide is concomitantly
administered with NPH insulin and also describes an
increased rate of serious cardiovascular events with
repaglinide compared to sulfonylureas [15]. More recently,
Rev Endocr Metab Disord (2010) 11:21–30
a meta-analysis of mostly postmarketing, short-term clinical
trials with rosiglitazone noted an increased risk for
myocardial ischemic events, although an increase with
rosiglitazone in cardiovascular death, non-fatal myocardial
infarction, and stroke has not been confirmed in long-term
clinical trials [16]. For these reasons, interest has emerged
in showing that chronic medications developed to treat type
2 diabetes do not further increase the vulnerability of
patients with diabetes to cardiovascular disease [17, 18].
This interest prompted a 2-day, public advisory committee
meeting in July 2008 [19] and the subsequent publication of
an FDA guidance in December 2008 recommending
cardiovascular assessment in the pre-approval and post-
approval settings for drugs and biologics developed for the
treatment of type 2 diabetes [20]. To place these new
guidelines into perspective, we first present an overview of
prior diabetes phase 2/3 development programs.
2 Prior diabetes development programs
In February 2008, FDA published a draft guidance for
industry, entitled Diabetes Mellitus: Developing Drugs and
Therapeutic Biologics for Treatment and Prevention [21].
Guidance documents represent FDA’s current thinking on
various topics and offer non-binding recommendations. The
February 2008 draft guidance document covers a wide
range of issues related to both type 1 and type 2 diabetes
and both oral anti-diabetic medications and insulin prod-
ucts. Pertinent aspects pertaining to phase 2 and phase 3
development of non-insulin antidiabetic medications for
type 2 diabetes are summarized below. Note that this draft
guidance was issued prior to the development of the
cardiovascular guidelines that will be discussed later in this
article.
2.1 Phase 2 and phase 3 clinical trials
Phase 2 programs for oral anti-diabetic medications have
typically consisted of 1 or 2 dose-finding trials, usually
conducted in patients who are not taking other anti-diabetic
therapy or who are taking a stable dose of metformin. These
studies have usually randomized a few hundred patients to
12 weeks of treatment with placebo or to several doses of
the investigational agent. The doses that show the most
promising preliminary efficacy and safety profiles are
selected for further evaluation in phase 3.
The Phase 3 program has typically consisted of several
6-month placebo controlled trials followed by 6–18 month
extensions. The investigational agent has usually been
studied in different treatment settings, such as monotherapy
and in combination with maximal or near-maximal doses of
commonly used anti-diabetic medications. The extension
Rev Endocr Metab Disord (2010) 11:21–30
trials provide long-term data for these chronically used
therapies. In the past, the extension phases were typically
uncontrolled, with all participating patients taking the
investigational agent after the primary efficacy timepoint.
Although this approach maximizes the number of patients
exposed to the investigational agent, which may be useful
for identifying rare, serious adverse events (e.g., agranulo-
cytosis or hepatotoxicity), it is now strongly discouraged
because the lack of a control group substantially limits
conclusions on efficacy and safety. In addition, when
programs did have controlled extensions, the extension
periods were often unblinded and did not include all patients,
raising the potential for bias. Patients who developed
unacceptable hyperglycemia in the controlled treatment
periods were often withdrawn from the clinical trials, reducing
the number of patients contributing to long-term controlled
data. The February draft guidance strongly encourages
placebo- or active-controlled extension trials and also recom-
mends that patients requiring glycemic rescue therapy not be
withdrawn [21]. Instead, those patients should receive open-
label rescue therapy and continue in the trials, thereby
bolstering the long-term controlled safety database.
2.2 Sample sizes
The International Conference on Harmonisation (ICH)
guideline for industry E1 The Extent of Population
Exposure to Assess Clinical Safety For Drugs Intended
for Long-Term Treatment of Non-Life-Threatening Condi-
tions recommends a total exposure to investigational agent
of at least 1,500 patients with 300–600 for 6 months, and
100 for 1 year for the safety assessment of chronically
administered drugs developed for the treatment of non-life-
threatening conditions [22]. The February 2008 guidance
recommends that phase 2/3 trial data include at least 2,500
patients exposed to the investigational product with at least
1,300 to 1,500 patients exposed to the investigational
product for 1 year or more and at least 300 to 500 subjects
exposed to the investigational product for 18 months or
more [21]. The February guidance recommended greater
patient exposure than ICH E1 because of the large and
growing population with type 2 diabetes and the increasing
complexity of treatment regimens as additional anti-diabetic
therapies have become available. Note that even larger
exposures may be needed, based on premarketing safety
signals, safety concerns with related compounds, or the
product’s mechanism of action.
2.3 Limitations
These prior diabetes programs typically excluded patients at
high cardiovascular risk and usually enrolled patients with
early diabetes (i.e., patients who had failed diet and exercise
23
or who had failed a single anti-diabetic medication). Few
patients with longstanding diabetes, few patients >75 years
old, and few patients with moderate or severe renal
impairment were enrolled. Patients with a history of
significant cardiovascular disease, such as myocardial
infarction or stroke, were required to have a remote event
(e.g., >6 months prior to study entry), were excluded
entirely, or were not actively recruited. These development
programs were, therefore, not fully representative of the
population of patients with diabetes who would use the
investigational medication, if approved.
Until recently, diabetes development programs did not
carefully quantify the cardiovascular risk of the investiga-
tional medications. Cardiovascular events were reported by
investigators without any attempt to prospectively define
cardiovascular endpoints of interest; blinded adjudication of
events, typical of outcome effectiveness trials, was not
utilized. The pharmaceutical companies and FDA did, of
course, review investigator-reported cardiovascular adverse
events and took note of any imbalances. The number of
cardiovascular events was typically low, however, for the
reasons explained above (e.g., exclusion of patients at high
cardiovascular risk, little controlled data beyond 6 months).
Nonetheless, no further evaluation of cardiovascular risk
was deemed necessary, so long as the few cardiovascular
events were well-balanced between treatment groups and
there were no unfavorable findings with regard to well-
validated cardiovascular risk surrogates (e.g., increased
blood pressure and adverse effects on lipid parameters).
3 The July 2008 public advisory committee meeting
FDA convened a 2-day advisory committee meeting in July
2008 to discuss the role of cardiovascular assessment in the
pre-approval and post-approval settings for drugs and bio-
logics developed for the treatment of type 2 diabetes mellitus
[19]. The advisory panel was populated by members of the
Endocrinologic and Metabolic Drugs Advisory Committee,
diabetologists, cardiologists, statisticians, and members of the
Drug Safety and Risk Management Committee. After hearing
presentations by experts in endocrinology and cardiology, the
panel members extensively discussed the issues at hand and
were then asked to vote on the following question: “It should
be assumed that an anti-diabetic therapy with a concerning
CV [cardiovascular] safety signal during Phase 2/3 develop-
ment will be required to conduct a long-term cardiovascular
trial. For those drugs or biologics without such a signal,
should there be a requirement to conduct a long-term
cardiovascular trial, or to provide other equivalent evidence
to rule out an unacceptable cardiovascular risk?” Fourteen
panel members voted “yes” to this question and two voted
“no”. Complete transcripts are publicly available [19].
24
4 The December 2008 guidance for industry
After considering the recommendations of the advisory
panel and other data, FDA published a guidance for
industry in December 2008 entitled Diabetes Mellitus—
Evaluating Cardiovascular Risk in New Antidiabetic
Therapies to Treat Type 2 Diabetes [20]. The December
2008 guidance reaffirms HbA1c as the primary efficacy
endpoint for glucose reduction but emphasizes the vulner-
ability of patients with diabetes to cardiovascular disease.
The guidance, therefore, asks that pharmaceutical compa-
nies developing new drugs or biologics for the treatment of
type 2 diabetes show that these therapies do not result in an
unacceptable increase in cardiovascular risk. The guidance
applies to all unapproved therapies, regardless of the stage
of development, even those with completed or ongoing
programs at the time the guidance was issued. This
guidance does not discuss cardiovascular assessment for
already approved treatments for diabetes, which will be
addressed separately in the near future.
Of note, the guidance calls for evaluation of whether
the treatment increases cardiovascular risk; it does not
ask pharmaceutical companies to show a cardiovascular
benefit with their product. Such a requirement would
have, in all likelihood, prevented development of new
treatments for type 2 diabetes, as it has not been
possible, even in large outcome trials, to conclusively
show such benefit in type 2 diabetes for any therapies,
including insulin. Nonetheless, the value of controlling
hyperglycemia is clear and there is good evidence, which
was discussed and accepted by the July 2008 advisory
committee, that control of hyperglycemia reduces the rate
of microvascular complications.
The need for evaluation of cardiovascular risk applies
even if there is no known cardiovascular signal with the
investigational agent in animals or humans, or a history of
concern with the pharmacologic class. The goal of this new
approach is to identify off-target cardiovascular toxicity.
Prior to publication of the guidance, there was reliance on
non-clinical studies in healthy animals (which might not be
representative of patients with diabetes and its comorbid-
ities) and cardiovascular assessment based on adverse
events reported by investigators. However, as explained
above, patients at high cardiovascular risk were typically
excluded from the clinical trials and long-term extensions
were often uncontrolled, yielding few cardiovascular events
of interest during the controlled portions of the trials, and,
therefore, little assurance that risk was not increased.
To generate reliable data on cardiovascular risk, the
guidance recommends that pharmaceutical companies es-
tablish an independent cardiovascular endpoint committee
to prospectively and blindly adjudicate major cardiovascu-
lar events during phase 2 and 3 clinical trials. The guidance
Rev Endocr Metab Disord (2010) 11:21–30
also recommends that the phase 2 and 3 clinical trials be
designed so that a pre-specified meta-analysis of major
cardiovascular events can be reliably performed. The
guidance notes that the phase 3 trials should be of sufficient
duration to ensure that an adequate number of cardiovas-
cular events are accrued and to obtain long-term safety data
for these therapies that will be used chronically, if
approved. In addition, the guidance recommends that the
phase 3 program enroll patients at increased cardiovascular
risk. For example, enrolling elderly patients, those with
prior cardiovascular disease and risk factors, those with
longstanding diabetes, and those with renal impairment
increases the likelihood that a sufficient number of
cardiovascular events will occur and improves generaliz-
ability of results.
4.1 Defining unacceptable cardiovascular risk
The guidance recommends that pharmaceutical companies
compare the incidence of major cardiovascular events with
the investigational drug to the incidence of these events
with comparators. Based on this comparison, the company
should calculate the point estimate for the risk ratio and its
corresponding two-sided 95% confidence interval. The
upper bound of the 95% confidence interval represents a
reasonable estimate of the worst case for increased
cardiovascular risk. For example, an upper bound of 1.8
means that the estimated increased risk of cardiovascular
events with the investigational drug is very unlikely to be
worse than 1.8 times (or 80% greater) the risk with
comparator. For the cardiovascular data to support approv-
ability of a new diabetes drug or biologic, this upper bound
should be less than 1.8 with a reassuring point estimate
(Table 1). If this upper bound is between 1.3 and 1.8 and
the overall risk-benefit analysis supports approval, a
postmarketing cardiovascular trial generally will be re-
quired to definitively show that this upper bound is less
than 1.3. If the premarketing data were to show that this
upper bound is less than 1.3, and the overall risk-benefit
analysis supports approval, a postmarketing cardiovascular
trial generally may not be necessary.
The postmarketing trial to show that the upper bound of
the risk ratio or hazard ratio is less than 1.3 would be a
“postmarketing requirement” under the authorities of the
FDA Amendments Act of 2007 [23]. With these new
authorities, FDA ensures that the trials are adequately
designed to answer the safety question of interest within a
publicly posted timeline. This timeline has three compo-
nents: the date by which the final protocol must be
submitted to FDA, the completion date for the trial, and
the date by which the final study report must be submitted
to FDA for review. Delays in any of these timelines are
publicly posted and can lead to civil penalties.
Rev Endocr Metab Disord (2010) 11:21–30
Table 1 Acceptable and unacceptable cardiovascular risk
Upper bound of 95% Conclusion
confidence interval for
risk ratio or hazard ratio
>1.8 Inadequate to support approval
>1.3 but <1.8a Postmarketing cardiovascular trial(s)
needed to show definitively <1.3
<1.3a Postmarketing cardiovascular trial(s)
generally not necessary
a
with a reassuring point estimate
The 1.3 goalpost has been used by FDA in other settings
to exclude unacceptable cardiovascular risk (e.g., COX-2
selective non-steroidal anti-inflammatory drugs) [24]. The
1.3 goalpost is feasible but requires a sizeable study and
meeting this criterion pre-approval would significantly
delay the availability of new anti-diabetic medications. As
shown in Table 2, to have 80% power to exclude the upper
bound of 1.3 for the hazard ratio requires approximately
450 major cardiovascular events (based on the logrank test
statistic with 1:1 randomization, and an assumed true
hazard ratio of 1.0). In contrast, 90 events of interest are
needed to have 80% power to exclude 1.8. FDA is prepared
to tolerate additional uncertainty of cardiovascular risk at
the time of approval (with uncertainty capped with an upper
bound <1.8) because glycemic control (which is the basis
for approval of these medications) has inherent benefits,
including short-term symptomatic relief from hyperglyce-
mia and reduction in long-term microvascular complica-
tions of diabetes [12, 13]. There is, moreover, an unmet
need for new anti-diabetic medications, justifying the two-
step approach to ruling out cardiovascular risk. For
example, a recent publication showed that approximately
45% of adults with diagnosed diabetes are not at the
American Diabetes Association HbA1c target of <7%
despite the widespread availability of multiple classes of
antidiabetic medications [25]. Although there may be many
reasons for the low proportion of patients meeting HbA1c
goals, the limitations of current therapies may play a role
[26]. For example, insulin typically requires injection and is
associated with weight gain and hypoglycemia. Metformin
Table 2 Number of cardiovascular events of interest needed based on
the non-inferiority margin and desired statistical powera
Non-inferiority margin Power
80% 85% 90%
1.8 90 105 125
1.3 455 525 615
a
Using the logrank statistic, assuming 1:1 randomization and a true
hazard ratio of 1.0
25
is recommended as the first-line treatment for type 2
diabetes but is contraindicated in patients with renal
impairment and thiazolidinediones can cause or exacerbate
heart failure in susceptible individuals.
Table 3 shows the sample sizes needed for excluding
cardiovascular risk ratios of 1.2 and 1.3 with 90% power
for various annual event rates for a 5-year clinical trial that
has a 2-year recruitment period. This table includes data for
a risk ratio of 1.2 only to illustrate the effect of seeking to
rule out an even slightly smaller risk ratio (1.2 is not
discussed in the December 2008 guidance). As shown,
sample sizes for excluding 1.2 are considerably larger than
those for excluding 1.3 and the clinical trials needed are not
feasible if the true annual event rate with investigational
drug is even slightly larger than the true annual event rate
with comparator.
4.2 Insulins and fixed-dose combination products
The 1.8 and 1.3 goalposts apply to insulins that are
intended to be used like non-insulin therapies (e.g., oral
insulin), but do not apply to other insulin products.
However, even when the 1.8 and 1.3 goalposts do not
apply, pharmaceutical companies should still collect reli-
able cardiovascular data (e.g., with pre-specified endpoints,
blinded adjudication) and include pre-specified cardiovas-
cular analyses in the marketing application. Insulin is
generally not held to the 1.8 and 1.3 goalposts because
insulin is the only life-saving treatment available for
patients with type 1 diabetes and is the last-line treatment
for patients with type 2 diabetes who have failed all other
available therapies. In contrast, the wide range of other
therapies for type 2 diabetes provides an opportunity to
evaluate the effect of these therapies on cardiovascular risk,
allowing a more informed decision when choosing between
medications for the management of type 2 diabetes.
Table 3 Sample sizes for excluding cardiovascular riska
Annual event rate Annual event rate Total sample size to
(Drug) (Comparator) rule out increased risk
of 1.2 or 1.3
1.2b 1.3
2% 2% 16,500 8,000
2% 1.75% >100,000 34,000
3% 3% 11,200 5,400
3% 2.8% 29,300 10,100
a
Assumes a 5-year trial with a 2-year recruitment period, 90% power,
and alpha of 0.05
b
1.2 is included only to illustrate concepts related to the choice of the
margin and is not discussed in the December 2008 diabetes
cardiovascular guidance
26
Fixed-dose combination products contain two or more
approved medications within a single tablet. These products
do not create a need for additional cardiovascular outcome
data provided there is adequate evidence of cardiovascular
safety for the individual components and there is no
pharmacological interaction between components that could
adversely impact cardiovascular safety. As mentioned
above for insulins, clinical trials that form the basis for a
marketing application for fixed-dose combination should
still collect reliable cardiovascular data and include pre-
specified cardiovascular analyses even if the product will
not be held to the 1.8 and 1.3 goalposts.
4.3 Sample sizes
As noted earlier, the February 2008 diabetes guidance
recommends that, to support approvability, phase 2/3 trial
data should be available for at least 2,500 patients exposed
to the investigational product with at least 1,300 to 1,500 of
these patients exposed to the investigational product for
1 year or more and at least 300–500 patients exposed to the
investigational product for 18 months or more [21].
Therefore, these are the minimum exposures that should
be used to determine whether the 1.8 criterion has been
met. Although it may be possible to follow patients at very
high cardiovascular risk for shorter durations and still
obtain the necessary number of cardiovascular events, this
approach is not recommended for several reasons. First,
there is interest in longer-term efficacy and safety informa-
tion in addition to cardiovascular risk, because the product
is intended for chronic use. Also, there is concern that
people at very high risk, such as those who have had a
myocardial infarction or episode of unstable angina within
a few weeks prior to randomization might have another
cardiovascular event soon that is independent of an effect of
treatment. These early events could add “noise” in a trial
intended to rule out a specified increase (similar to a non-
inferiority effectiveness trial) and potentially bias results
toward showing no difference. Less acute situations,
however, such as a myocardial infarction occurring several
months prior to randomization, help obtain a population
with an adequate event rate.
5 Potential impact of the December 2008 guidance
The recommendations in the December 2008 guidance are
expected to lead to clinical development programs that are
at least somewhat larger and longer than completed
programs that predated this guidance. It should be appre-
ciated, however, that the phased assurance (ruling out a 1.8
risk ratio pre-marketing and 1.3 postmarketing), which
followed the July 2008 advisory committee recommenda-
Rev Endocr Metab Disord (2010) 11:21–30
tion, is not expected to increase pre-marketing exposure
markedly if the recommended higher risk populations are
included. FDA believes the new approach outlined in the
December 2008 guidance will provide important safety
information and help better inform treatment decisions for
type 2 diabetes. Nonetheless, it could be asked whether the
added cost, time and, perhaps, uncertainty of the new drug
development process will discourage development of new
anti-diabetic medications. This may prove to be the case for
a drug when there are already several drugs in the same
class on the market; the incremental benefit to a company
of further developing such a drug may be unclear. This new
guidance, however, would not be expected to discourage
development of novel therapies for type 2 diabetes because
there is a significant need for new diabetes medications.
There is a large and growing patient population that would
use new therapies [1, 2]. Diabetes therapies, moreover, are
used chronically and there is often need for new add-on
therapies to maintain adequate glycemic control given the
progressive nature of the disease. Although there are
currently 11 classes of approved anti-diabetic medications
(Table 4), as noted, many patients do not achieve adequate
glycemic control [25] and there are important limitations of
available therapies, such as the contraindication to use
metformin in patients with renal impairment and the heart
failure limitations related to thiazolidinediones [26]. For
these reasons, it is likely that pharmaceutical companies
will continue to have important incentives for developing
promising anti-diabetic medications.
6 Trial design challenges and considerations
Table 5 summarizes key design issues and challenges for
diabetes cardiovascular trials. These aspects are discussed
in detail in the sections below.
6.1 Study design
The current relatively poor information on cardiovascular
risks of available therapies for type 2 diabetes makes
planning the clinical trials a formidable challenge, espe-
cially the dedicated cardiovascular trials. The most obvious
trial would be one comparing the new drug to placebo in
patients with less than optimal glycemic control, with the
two treatments added to whatever anti-diabetic medication
the patient is already receiving. If, as described above, the
patient population is one with long-standing diabetes,
background anti-diabetic therapy is virtually inevitable.
This design will generally lead to better glycemic control in
the investigational drug group, which would clearly be a
concern if better glycemic control reduced cardiovascular
risk, but given results to date, this does not appear to be the
Rev Endocr Metab Disord (2010) 11:21–30 27

Table 4 Classes of medications approved for the treatment of problem only if it has an adverse cardiovascular effect
diabetes mellitusa
(perhaps decreasing the apparent adverse effect of the new
Drug class with Presumed primary mechanism of action drug) or a favorable cardiovascular effect (creating an
examples apparent adverse effect of the new drug). Given the lack of
evidence of clearly favorable or adverse cardiovascular
Alpha-glucosidase Delay glucose absorption from the
inhibitors gastrointestinal tract by inhibiting enzymes
effects of current therapy, this should be manageable.
- Acarbose that convert carbohydrates to Another possible design is a comparison of the new drug
- Miglitol monosaccharides with an active control in patients not responding adequately
Amylin analogues Slow gastric emptying and suppress the to prior anti-diabetic therapy. The risk in this design is the
-Pramlintide acetate postprandial rise in plasma glucagon possible cardiovascular effect of the control agent, which, if
Biguanides Decrease hepatic glucose production present, could make the investigational drug look either
- Metformin better or worse. Again, to date, most potential control agents
Bile acid Unknown seem relatively neutral with regard to cardiovascular risk.
sequestrants
- Colesevelam
hydrochloride 6.2 Glycemic rescue therapy
Dipeptidyl-peptidase Slow inactivation of incretin hormones (e.g.,
4 inhibitors GLP-1), which stimulate glucose-dependent Diabetes trials will probably need to be several years in
- Saxagliptin insulin secretion duration to meet the 1.3 relative risk criterion recommended
- Sitagliptin in the December 2008 guidance. Because diabetes is a
phosphate progressive disease, it is unlikely that many patients will be
Dopamine receptor Unknown able to maintain adequate long-term glycemic control with
agonists
- Bromocriptine the therapies initiated at the start of a multi-year trial. For
mesylate example, ADOPT (A Diabetes Outcome Progression Trial)
Glinides Stimulate insulin release by inhibiting randomized patients with type 2 diabetes diagnosed within
- Nateglinide ATP-dependent potassium channels the preceding 3 years to monotherapy with metformin, a
- Repaglinide on pancreatic beta-cells (glinides are sulfonylurea, or rosiglitazone and followed these patients
structurally distinct to sulfonylureas
and exert effects via a different receptor)
for up to 6 years. Over this duration of follow-up,
GLP-1b receptor Stimulate glucose-dependent insulin
agonists secretion Table 5 Key issues in cardiovascular trial design
- Exenatide
Justify the choice of comparator, ideally its cardiovascular profile
- Liraglutide
should be known
Insulins Stimulate peripheral glucose uptake by
Design and execute safety comparison trials well, otherwise there is
- Insulin Aspart skeletal muscle and fat and inhibit hepatic
potential to bias towards showing no difference between treatment
- Insulin Glulisine glucose production
groups
- Insulin Lispro The need for glycemic rescue therapy is expected in multi-year trials.
Sulfonylureas Stimulate insulin release by inhibiting the Patients requiring glycemic rescue therapy should not be discontinued
- Glipizide ATP-dependent potassium channel on from the trials.
- Glyburide pancreatic beta-cells Pre-specify the definitions for all cardiovascular events of interest and
Thiazolidinediones Insulin sensitizers have events blindly adjudicated by an independent party
- Pioglitazone Modulate the transcription of genes involved Manage all cardiovascular risk factors to current standards-of-care
- Rosiglitazone in glucose metabolism guidelines
Consider trial efficiencies, such as waiving expedited reporting of
a
listed in alphabetical order adverse events that are part of the cardiovascular primary endpoint
b and limiting centralized laboratories
GLP-1=glucagon-like peptide 1
The primary cardiovascular endpoint should consist of cardiovascular
death, non-fatal myocardial infarction, and non-fatal stroke. Note the
caveats mentioned in the text if hospitalization for unstable angina is
case in trials of several years duration. It must be included.
acknowledged, however, that as study duration lengthens, Structured case report forms are recommended to ensure critical
there is increasing concern with inadequate glycemic cardiovascular data are captured.
control because of adverse effects of hyperglycemia on Power for meeting the 1.8 and 1.3 goalposts is predominantly driven
microvascular disease [12, 13]. Patients who develop by the event rate. Estimate event rates conservatively and design the
inadequate glycemic control would, therefore, clearly need trial as event-driven so that additional events can be accrued if the
event rate is unexpectedly low. Seek patients at high cardiovascular
glycemic rescue medication and this would be more likely
risk.
to occur in the control group. The added therapy is a
28
approximately 10–25% of patients across the treatment
groups met the primary endpoint of inadequate glycemic
control (e.g., consecutive fasting plasma glucose >180 mg/
dL after at least 6 weeks of treatment at the maximum
tolerated dose of study medication) [27]. Therefore, many
patients with type 2 diabetes participating in a multi-year
cardiovascular trial are likely to require initiation of
glycemic rescue therapy at some point during the treatment
period. If the investigational agent is compared to an active
comparator with similar glycemic efficacy, it would be
expected that the extent of glycemic rescue would be
similar across treatment groups.
6.3 Management of other cardiovascular risk factors
In these trials, cardiovascular risk factors (e.g., blood
pressure, lipids, anti-platelet medications) should be treated
to current, regional standards-of-care. The impact of
regional differences should be minimized by randomiza-
tion. For trials that also have an objective of showing
glycemic efficacy with the investigational agent based on
change from baseline in HbA1c, it is acceptable to not treat
HbA1c to standard-of-care guidelines (e.g., <7% if follow-
ing the American Diabetes Association guidelines) for the
first 6 months of the trial so as not to bias the efficacy
assessment. However, glycemic rescue therapy should still
be initiated during these 6 months for patients with
significant hyperglycemia.
6.4 Control of type 1 error
Control of type 1 error (here meaning the false negative rate
or the rate of not detecting a true upper bound of risk
greater than 1.8 or 1.3) is an important consideration, and
should be addressed if there will be interim analyses to
evaluate whether the 1.8 and 1.3 goalposts have been met.
There may be particular interest in conducting interim
analyses for the 1.8 value so that pharmaceutical companies
can learn as soon as possible when this threshold has been
met to limit delays in bringing the new product to market. It
is important to control type 1 error for the 1.8 analyses
because this information will be used to decide whether the
product can be approved from a cardiovascular standpoint.
If type 1 error is inflated from multiple looks without
correction, there may be an erroneous conclusion that there
is adequate cardiovascular safety when in fact there is not.
6.5 Sample trial designs
To date, there have been several different proposals by
pharmaceutical companies for meeting the 1.8 and 1.3
goalposts. Design proposals have included, but are not
limited to, a dedicated cardiovascular safety trial in patients
Rev Endocr Metab Disord (2010) 11:21–30
with recent acute coronary syndrome, combining of data
from a standard phase 2/3 program plus a dedicated
cardiovascular safety trial, and a single dedicated phase 3
cardiovascular trial containing substudies for assessing
glycemic efficacy. Some pharmaceutical companies may
choose to use the same dedicated cardiovascular trial to
meet 1.8 to support approvability then continue this same
trial in the postmarketing setting to meet 1.3. Others may
choose to use different trials to meet 1.8 and 1.3. The
guidance allows for flexibility when designing a program to
assess cardiovascular risk; therefore, all of these options are
acceptable provided that the plan and analyses are pre-
specified and statistical rigor is maintained. Such plans
should be reviewed by FDA prior to implementation.
6.6 Efficiencies in trial design
Several approaches can be used to help efficiently imple-
ment the December 2008 guidance without undue burden.
For example, there should be no expedited reporting to
FDA of adverse events that are part of the cardiovascular
primary endpoint because such events are expected, will be
monitored by the data monitoring committee, and cannot be
considered possibly-related to the treatment in isolation.
Another approach includes performing only critical labora-
tory tests at centralized facilities. With agreement of the
review division, adverse events collected in the dedicated
cardiovascular trial can be limited to serious adverse events,
withdrawals due to adverse events, and adverse events of
particular interest.
6.7 Cardiovascular endpoints
Trials designed to show cardiovascular benefit often, but not
always, use a primary composite endpoint of Major Adverse
Cardiovascular Events (MACE) that includes cardiovascular
death, non-fatal myocardial infarction and non-fatal stroke.
Trials sometimes include other endpoints in the primary
composite, such as hospitalization for unstable angina, urgent
coronary revascularization, and hospitalization for heart
failure [28, 29]. Some of these other endpoints are
challenging to define and have a more subjective component
than cardiovascular death, myocardial infarction, and stroke.
For example, coronary revascularization is often performed
to alleviate symptoms of myocardial ischemia and the
decision to perform this procedure as well as its timing
may vary between investigators. Although the trial is
randomized, the variability of a somewhat subjective
endpoint introduces “noise” that can bias results towards
showing no difference between treatments.
The December 2008 guidance discusses the possibility of
including events other than cardiovascular death, non-fatal
myocardial infarction and non-fatal stroke in the primary
Rev Endocr Metab Disord (2010) 11:21–30
cardiovascular endpoint. However, a more traditional MACE
composite is probably preferable because of the methodolog-
ical issues discussed above. To date, only hospitalization for
unstable angina has been accepted as an additional endpoint
for inclusion with the traditional MACE endpoints. Pharma-
ceutical companies who want to include other endpoints
should be aware of the above caveats. Regardless of which
components are included in the primary composite endpoint,
FDA will assess the contribution of the individual components
to the overall findings. If there are favorable trends with
hospitalization for unstable angina but the more traditional
MACE endpoints trend adversely, there may not be adequate
evidence of cardiovascular safety even if the composite
endpoint meets the 1.8 or 1.3 goalposts.
The reason pharmaceutical companies might prefer to
use endpoints broader than the MACE composite is to
increase event rates and reduce the sample size needed
to meet the recommendations of the December 2008
guidance. In recent diabetes trials, such as Action to
Control Cardiovascular Risk in Diabetes (ACCORD) [30]
and Action in Diabetes and Vascular Disease (ADVANCE)
[31], the event rates for traditional MACE have averaged
approximately 2% per year in patients intended to be at
increased risk for cardiovascular disease. In contrast, the
annual event rate was approximately 5% in the Veterans
Affairs Diabetes Trial (VADT) for an expanded endpoint
that also included heart failure, surgery for vascular disease,
inoperable coronary artery disease, and amputation for
ischemic gangrene [32]. Even in high-risk populations,
cardiovascular event rates have fallen over time as effective
therapies (e.g., statins, blood pressure medications) have
become available.
To help facilitate efficient implementation of the December
2008 guidance, FDA together with stakeholders, such as
academia and industry, has been working to standardize
definitions for cardiovascular events of interest. When
finalized, these standardized definitions will be shared with
pharmaceutical companies that are developing treatments for
type 2 diabetes.
Pharmaceutical companies should pre-specify how cardio-
vascular events of interest will be identified and captured and
which events will be sent to the adjudication committee. It is
recommended that pharmaceutical companies develop a
checkbox form that facilitates structured collection of critical
information needed to assess whether a cardiovascular event
of interest has occurred. There should not be reliance on the
Medical Dictionary for Regulatory Affairs (MedDRA) alone,
because events may not be coded correctly and inadequate
information may be collected when an unstructured approach
is used. However, it is still appropriate to search through
reported MedDRA preferred terms to identify any potential
cardiovascular events of interest that were missed by the
checkbox approach.
29
7 Summary
The new diabetes cardiovascular guidance will ensure that
new pharmaceuticals developed for the treatment of type 2
diabetes do not unacceptably increase cardiovascular risk
in a patient population that is already vulnerable to
cardiovascular disease. There is a two-step approach to
rule out cardiovascular risk so that new anti-diabetic
therapies are not unduly delayed. Pharmaceutical compa-
nies should pre-specify the approach that will be used to
assess cardiovascular risk to ensure that cardiovascular
events of interest are appropriately defined, captured, and
analyzed with sufficient rigor. The end-of-phase 2 meeting
is the ideal time for pharmaceutical companies to discuss
their plans for assessing cardiovascular risk with FDA.
The new approach to developing medications for the
treatment of type 2 diabetes will encourage evaluation in
patients representative of those who will use these
therapies, if approved, and will help healthcare providers
make informed decisions when choosing a medication
within the treatment armamentarium for type 2 diabetes.
Acknowledgements The authors thank Ilan Irony, M.D. for his
critical review of the manuscript.
References
1. National estimates on diabetes. In: National diabetes statistics,
2007. National Institutes of Health. 2007. http://diabetes.niddk.
nih.gov/DM/PUBS/statistics. Accessed 21 Feb 2010.
2. Stumvoll M, Goldstein BJ, van Haeften TW. Type 2 diabetes:
principles of pathogenesis and therapy. Lancet. 2005;365:1333–
46.
3. Haffner SM, Lehto S, Ronnemaa T, Pyorala K, Laakso M.
Mortality from coronary heart disease in subjects with type 2
diabetes and in nondiabetic subjects with and without prior
myocardial infarction. N Engl J Med. 1998;339:229–34.
4. Manson JE, Colditz GA, Stampfer MJ, Willett WC, Krolewski
AS, Rosner B, et al. A prospective study of maturity-onset
diabetes mellitus and risk of coronary heart disease and stroke in
women. Arch Intern Med. 1991;151:1141–7.
5. Kannel WB, McGee DL. Diabetes and cardiovascular disease.
The Framingham study. JAMA. 1979;241:2035–8.
6. Stamler J, Vaccaro O, Neaton JD, Wentworth D. Diabetes, other
risk factors, and 12-yr cardiovascular mortality for men screened
in the Multiple Risk Factor Intervention Trial. Diabetes Care.
1993;16:434–44.
7. Goldbourt U, Yaari S, Medalie JH. Factors predictive of long-term
coronary heart disease mortality among 10, 059 male Israeli civil
servants and municipal employees. A 23-year mortality follow-up
in the Israeli Ischemic Heart Disease Study. Cardiology.
1993;82:100–21.
8. Fuller JH, Shipley MJ, Rose G, Jarrett RJ, Keen H. Mortality from
coronary heart disease and stroke in relation to degree of
glycaemia: the Whitehall study. Br Med J (Clin Res Ed).
1983;287:867–70.
9. Barrett-Connor EL, Cohn BA, Wingard DL, Edelstein SL. Why is
diabetes mellitus a stronger risk factor for fatal ischemic heart
30
disease in women than in men? The Rancho Bernardo Study.
JAMA. 1991;265:627–31.
10. Standards of medical care in diabetes-2010. Diabetes Care
2010;33 Suppl 1:S11–-61.
11. Goldstein DE, Little RR, Lorenz RA, Malone JI, Nathan D,
Peterson CM, et al. Tests of glycemia in diabetes. Diabetes Care.
2004;27:1761–73.
12. The effect of intensive treatment of diabetes on the development
and progression of long-term complications in insulin-dependent
diabetes mellitus. The Diabetes Control and Complications Trial
Research Group. N Engl J Med 1993;329:977–86.
13. Intensive blood-glucose control with sulphonylureas or insulin
compared with conventional treatment and risk of complications
in patients with type 2 diabetes (UKPDS 33). UK Prospective
Diabetes Study (UKPDS) Group. Lancet 1998;352:837–53.
14. Meinert CL, Knatterud GL, Prout TE, Klimt CR. A study of the
effects of hypoglycemic agents on vascular complications in
patients with adult-onset diabetes. II. Mortality results. Diabetes.
1970;19:s747–830.
15. Approved repaglinide package insert. In: Drugs@FDA. United
States Food and Drug Administration. 2008. http://www.
accessdata.fda.gov/drugsatfda_docs/label/2009/020741s035lbl.
pdf. Accessed 21 Feb 2010.
16. Cardiovascular ischemic events with rosiglitazone. In: FDA
briefing document for July 30, 2007, joint meeting of the
endocrinologic and metabolic drugs advisory committee and the
drug safety and risk management advisory committee meeting.
United States Food and Drug Administration. 2007. http://www.
fda.gov/ohrms/dockets/ac/07/briefing/2007-4308b1-02-fda-back
grounder.pdf. Accessed 21 Feb 2010.
17. Rosen CJ. The rosiglitazone story—lessons from an FDA
Advisory Committee meeting. N Engl J Med. 2007;357:844–6.
18. Joffe HV, Parks MH, Meyer RJ, Jenkins JK, Temple R.
Rosiglitazone and the FDA. N Engl J Med. 2007;357:1775–6.
19. Role of cardiovascular assessment in the preapproval and postap-
proval settings for drugs and biologics developed for the treatment of
type 2 diabetes mellitus. In: Dockets for July 1–2, 2008, endocrino-
logic and metabolic drugs advisory commitee meeting. United States
Food and Drug Administration. 2008. http://www.fda.gov/ohrms/
dockets/ac/cder08.html#EndocrinologicMetabolic. Accessed 21 Feb
2010.
20. Guidance for industry: Diabetes mellitus—evaluating cardiovascular
risk in new antidiabetic therapies to treat type 2 diabetes. In:
Guidances (drugs). United States Food and Drug Administration.
2008. http://www.fda.gov/downloads/Drugs/GuidanceCompliance
RegulatoryInformation/Guidances/ucm071627.pdf. Accessed 21
Feb 2010.
21. Guidance for industry: Diabetes mellitus—developing drugs
and therapeutic biologics for treatment and prevention (draft).
Rev Endocr Metab Disord (2010) 11:21–30
In: Guidances (Drugs). United States Food and Drug Admin-
istration. 2008. http://www.fda.gov/downloads/Drugs/Guidance
ComplianceRegulatoryInformation/Guidances/ucm071624.pdf.
Accessed 21 Feb 2010.
22. E1: The extent of population exposure to assess clinical safety for
drugs intended for long-term treatment of non-life-threatening
conditions. In: ICH guidelines. International Conference on Harmo-
nisation. 1994. http://www.ich.org/LOB/media/MEDIA435.pdf.
Accessed 21 Feb 2010.
23. Full text of FDAAA law. In: public law 110-85. 110th congress. 2007.
http://frwebgate.access.gpo.gov/cgi-bin/getdoc.cgi?dbname=110_
cong_public_laws&docid=f:publ085.110. Accessed 21 Feb 2010.
24. Becker MC, Wang TH, Wisniewski L, Wolski K, Libby P,
Luscher TF, et al. Rationale, design, and governance of
Prospective Randomized Evaluation of Celecoxib Integrated
Safety versus Ibuprofen Or Naproxen (PRECISION), a cardio-
vascular end point trial of nonsteroidal antiinflammatory agents in
patients with arthritis. Am Heart J. 2009;157:606–12.
25. Ford ES, Li C, Little RR, Mokdad AH. Trends in A1C
concentrations among U.S. adults with diagnosed diabetes from
1999 to 2004. Diabetes Care. 2008;31:102–4.
26. Nathan DM, Buse JB, Davidson MB, Heine RJ, Holman RR,
Sherwin R, et al. Management of hyperglycemia in type 2
diabetes: a consensus algorithm for the initiation and adjustment
of therapy: a consensus statement from the American Diabetes
Association and the European Association for the Study of
Diabetes. Diabetes Care. 2006;29:1963–72.
27. Kahn SE, Haffner SM, Heise MA, Herman WH, Holman RR,
Jones NP, et al. Glycemic durability of rosiglitazone, metformin,
or glyburide monotherapy. N Engl J Med. 2006;355:2427–43.
28. Colhoun HM, Betteridge DJ, Durrington PN, Hitman GA, Neil
HA, Livingstone SJ, et al. Primary prevention of cardiovascular
disease with atorvastatin in type 2 diabetes in the Collaborative
Atorvastatin Diabetes Study (CARDS): multicentre randomised
placebo-controlled trial. Lancet. 2004;364:685–96.
29. Downs JR, Clearfield M, Weis S, Whitney E, Shapiro DR, Beere
PA, et al. Primary prevention of acute coronary events with
lovastatin in men and women with average cholesterol levels:
results of AFCAPS/TexCAPS. Air Force/Texas Coronary Athero-
sclerosis Prevention Study. JAMA. 1998;279:1615–22.
30. Gerstein HC, Miller ME, Byington RP, Goff Jr DC, Bigger JT, et
al. Effects of intensive glucose lowering in type 2 diabetes. N
Engl J Med. 2008;358:2545–59.
31. Patel A, MacMahon S, Chalmers J, Neal B, Billot L, Woodward
M, et al. Intensive blood glucose control and vascular outcomes in
patients with type 2 diabetes. N Engl J Med. 2008;358:2560–72.
32. Duckworth W, Abraira C, Moritz T, Reda D, Emanuele N, Reaven
PD, et al. Glucose control and vascular complications in veterans
with type 2 diabetes. N Engl J Med. 2009;360:129–39.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.