The publication of the U.K. Prospective Diabetes Study (UKPDS) in 1998 helped to shape
the management of type 2 diabetes in recent years (1). The study demonstrated several
points. First, sulfonylureas are as safe as insulin in controlling blood glucose.
Second, metformin reduced cardiovascular disease in an overweight subgroup. Third,
the same benefit of glycemic control in reducing microvascular disease (previously
noted in type 1 diabetes) is applied equally to patients with type 2 diabetes. A separation
in A1C of ∼1% in the UKPDS reduced the risk of microvascular disease (largely diabetic
retinopathy) by ∼25%. This reflected the data from the Diabetes Control and Complications
Trial, where a separation in A1C of 2% in intensive and standard groups led to a reduction
in microvascular disease of ∼50% (2).
A fourth demonstration was that there was no significant reduction in macrovascular
disease but a trend toward fewer myocardial infarctions with more intensive glucose
control. Fifth, using the current treatment of the time (first-generation sulfonylureas,
human ultratard insulin, or metformin), it proved impossible to maintain glucose control,
which tended to deteriorate throughout the study. It is now generally believed that
the progressive fall in endogenous insulin production as β-cell numbers decline makes
it difficult, if not impossible, to maintain tight control using standard treatment.
Sixth, the UKPDS also showed that in those patients with hypertension, lowering blood
pressure (BP) to moderate levels with either captopril or atenolol could reduce microvascular
disease (3).
In a subsequent study, the UKPDS investigators presented the rates of both micro-
and macrovascular disease according to the achieved levels of A1C during the study
(4). They showed a linear relationship between A1C and both groups of complications.
The implication of the article was that if glycemic control could be tightened below
the levels achieved in the UKPDS, then it might be possible to reduce rates, not only
of microvascular complications, but also cardiovascular disease as well.
The aim of the glucose arm of the Action in Diabetes and Vascular Disease: Preterax
and Diamicron MR Controlled Evaluation (ADVANCE) trial (5) was to build on the information
gained by the UKPDS and to answer the question as to whether intensifying glucose
control to achieve an A1C of <6.5% would provide additional benefit in reducing the
risk of both micro- and macrovascular disease.
ADVANCE also asked questions about BP lowering in patients with type 2 diabetes. The
aims of the BP arm were to establish whether routine provision of BP-lowering therapy
produced additional benefits in terms of macro- and microvascular disease, irrespective
of baseline BP, and added to the benefits produced by other cardiovascular preventive
therapies, including ACE inhibitors.
TRIAL DESIGN
The trial involved 215 collaborating centers in 20 countries from Asia, Australia,
Europe, and North America. It was designed to randomize over 10,000 patients with
established type 2 diabetes in a factorial design (Table 1), resulting in treatment
in four categories: 1) intensive glucose lowering (including gliclazide MR) and additional
“routine” BP lowering (perindopril/indapamide combination), 2) standard glucose therapy
and routine BP lowering, 3) intensive glucose lowering and placebo; and 4) standard
glucose therapy and placebo (6).
Table 1
Factorial design: description of the possible assigned therapies
Intensive glucose control
Standard glucose control
Routine BP-lowering therapy
Routine BP-lowering therapy
Intensive glucose control
Standard glucose control
Placebo
Placebo
Patients were included if they had type 2 diabetes and were aged ≥55 years with an
additional risk factor for a vascular event, any level of BP, and any level of glucose
control with no immediate indication for insulin treatment. After a 6-week prerandomization
run-in, during which they received a fixed combination of perindopril (2 mg) and indapamide
(0.625 mg) and standard guidelines-based blood glucose control, patients were randomized
in a factorial design to either a perindopril/indapamide combination or placebo and
either an intensive glucose-lowering strategy aiming for an A1C target of ≤6.5% or
standard glucose-lowering strategy.
Subjects in the intensive glucose control arm received gliclazide MR and any other
additional therapy to achieve these glucose targets. Subjects in the standard arm
received therapy according to local guidelines; although if they required a sulfonylurea,
then any medication other than gliclazide could be used.
Patients in the intensive glucose arm made clinic visits more frequently, attending
three monthly after the first 6 months compared with 6 monthly visits in patients
allocated to standard therapy.
Other measures that were used to tighten control in the intensive arm included encouraging
investigators to promote lifestyle management such as weight loss and exercise; maximizing
the dose of gliclazide MR, adding other oral agents; and adding long-acting insulin,
used as basal bedtime insulin, with rapid-acting insulin added as required, starting
with the main meal of the day.
Those allocated to standard therapy received treatment according to local practice.
Outcomes were recorded at 6 monthly intervals with the exception of retinopathy, albumin-to-creatinine
ratio, mini mental scores, and quality of life, which were recorded at 2 years, 4
years, and at the end of the study.
The primary outcome was a composite of macrovascular (myocardial infarction, stroke,
or cardiovascular death) and microvascular (retinopathy or nephropathy) events. Secondary
outcomes included heart failure, hospitalization, all-cause mortality, and dementia.
The sample size was calculated based on epidemiological studies reporting the association
of A1C with vascular events and assumed a 1% reduction in A1C between groups and 6
mmHg reduction in systolic BP, a 16% reduction in microvascular or macrovascular events
based on an annual event rate of 3% for each. The analysis was based on the intention-to-treat
principle. After 3 years of the trial, overall event rates were occurring at around
2% and separation of A1C was <1%. Without access to unblinded data, it was therefore
decided to combine the analysis of the primary end point (i.e., macrovascular and
microvascular disease) and extend the duration of the glucose-control arm by 18 months.
RESULTS
Patient characteristics
The progress of subjects through the trial is shown in Fig. 1. A total of 11,140 patients
were randomized into the two glucose arms with a median follow-up of 5 years. Baseline
characteristics are shown in Table 1. The mean age was 66 years, known duration of
diabetes 8 years, and BMI 28 kg/m2. Mean A1C at entry was 7.5%. About one-third of
the patients had a history of previous macrovascular disease, 71% were taking sulfonylureas
at entry, and 60% were on metformin.
Figure 1
Enrollment, randomization, and follow-up of study participants.
The BP arm of the study ran for an average of 4.3 years: the results, published in
The Lancet in 2007 (7), showed that a combination of indapamide and perindopril reduced
mortality, coronary events, and diabetic nephropathy regardless of the initial BP.
A1C and glucose-lowering therapy
By the end of follow-up, A1C in the intensive group had fallen to a mean of 6.5% compared
with 7.3% in the standard group (Fig. 2). A1C fell progressively in the intensive
group, reaching 6.5% after 2–3 years' duration of the trial. By the end of the trial,
over 90% of patients were still taking a sulfonylurea (gliclazide MR) in the intensive
group compared with 69% in those allocated to standard therapy. A total of 74% were
taking metformin (vs. 67% in the standard group), 40% were on insulin (vs. 24%), and
17% were taking a glitazone compared with 11% in the standard group.
Figure 2
A1C at baseline and during follow-up, according to glucose-control strategy. Data
are shown for mean glycated hemoglobin. The average difference between the intensive-control
group and the standard-control group for the follow-up period was 0.67 percentage
points (95% CI 0.64–0.70).
Effect of glycemic control on other risk factors
Systolic BP was significantly lower in individuals allocated to intensive glucose
control by the end of follow-up (135.5 vs. 137.9 mmHg, average difference during follow-up
1.6 mmHg; P < 0.001). The difference was present at the 3-month visit and all subsequent
visits. There were no differences in diastolic BP or lipids. The weight of those assigned
to intensive control remained generally stable, whereas those in the standard group
had a mean reduction in weight of 0.69 kg compared with the intensive group (P < 0.001).
Primary outcomes
A total of 2,125 participants had a major macrovascular or microvascular event during
follow-up: 18.1% in the intensive control group and 20.0% in the standard control
group (hazard ratio 0.90 [95% CI 0.82–0.98]; P = 0.013) (Fig. 3). There were no significant
differences in the number of macrovascular events between the two groups during the
trial (hazard ratio 0.94 [0.84–1.06]; P = 0.32) (Fig. 3). Thus, the differences were
due to fewer microvascular events (14% relative risk reduction, P = 0.01), essentially
a reduction in diabetic nephropathy (Fig. 4).
Figure 3
Cumulative incidences of events, according to glucose-control strategy. The hazard
ratios for intensive glucose control compared with standard glucose control were as
follows: for combined major macrovascular or microvascular events, 0.90 (95% CI 0.82–0.98)
(A); for major macrovascular events, 0.94 (0.84–1.06) (B); for major microvascular
events, 0.86 (0.77–0.97) (C); and for death from any cause, 0.93 (0.83–1.06) (D).
The vertical dashed lines indicate the 24- and 48-month study visits, at which additional
data on microvascular events were collected, specifically the ratio of urinary albumin
to creatinine and results of a retinal examination. For events relating to these data,
the event time was recorded as the date of the visit. The curves were truncated at
month 66, by which time 99% of the events had occurred. The effects of treatment (hazard
ratios and P values) were estimated from unadjusted Cox proportional-hazard models
that used all the available data.
Figure 4
Relative effects of glucose-control strategy on microvascular disease. The diamonds
incorporate the point estimates, represented by the vertical dashed lines, and the
95% CIs of the overall effects within categories; for subcategories, black squares
represent point estimates (with the area of the square proportional to the number
of events), and horizontal lines represent 95% CIs. The hazard ratios and relative
risk reductions are given for intensive glucose control compared with standard glucose
control.
Secondary outcomes (including sudden death)
There were 1,031 deaths during the trial, but there were no significant differences
between the two groups (8.9% in the intensive control group and 9.6% in the standard
control group (hazard ratio 0.93 [0.83–1.06]; P = 0.28) (Fig. 3). Hospitalization
was more frequent in the intensive control group, but there were no other significant
differences in secondary end points.
Hypoglycemia
Severe hypoglycemia was more frequent in the intensive-control group than in the standard-control
group: 150 patients (2.7%) had at least one episode of severe hypoglycemia compared
with 81 (1.5%) in the standard group (hazard ratio 1.86 [1.42–2.40]; P < 0.001).
Minor hypoglycemia was also more frequent in the intensive control group (120 events
per 100 patients per year, versus 90 with standard control).
CONCLUSIONS
The ADVANCE study has answered a number of important questions regarding glycemic
control in patients with type 2 diabetes. It has clearly demonstrated that it is possible
to achieve tight levels of glycemic control safely using conventional agents. The
glucose-lowering approach reflected conventional “real-life” strategies using sulfonylureas
(gliclazide MR), metformin, and insulin. Yet, A1C levels fell progressively to a mean
of 6.5% and remained there for the duration of the study. This is in contrast to the
glycemic profile of the UKPDS, where glucose control deteriorated gradually for the
length of the trial (2). It is possible that the differences between the populations
of patients studied in the two trials might account for some of these findings. Patients
in the UKPDS were newly diagnosed, and it is conceivable that in recruiting for ADVANCE,
local investigators excluded those whose glycemic control was likely to deteriorate.
The ADVANCE study also included a large group of individuals from Asia whose diabetes
might be expected to behave differently from a U.K. population. However, preliminary
analyses do not suggest that the European patients in ADVANCE behaved significantly
differently in terms of glycemic control. It may be that the incremental approach
to glycemic therapy in type 2 diabetes that has developed over the last 15 years and
uses metformin early together with basal insulin in combination with oral agents is
reasonably effective at maintaining tight levels of glycemic control.
Importantly, the intensive glucose-lowering strategy used in ADVANCE was apparently
safe. The incremental approach with increasing number of oral agents and the use of
basal insulin lowered A1C levels in the intensive group gradually and the target A1C
of 6.5% was achieved over 2 years after the start of the trial. Perhaps as a result
of this relatively slow fall in A1C, the side effects of glucose-lowering therapy
were relatively limited. Severe hypoglycemia was not common and less than that reported
in the UKPDS. Indeed, rates of severe hypoglycemia in those patients assigned to intensive
control in ADVANCE were below those reported in the standard arm of the UKPDS. It
is not clear which aspects of the glucose control strategy were responsible for the
low risk of severe hypoglycemia. It is possible that the use of basal insulin rather
than quick-acting insulin, which is associated with lower rates of hypoglycemia compared
with other insulin regimens (8), partly explains this finding.
The lack of weight gain in the trial was noteworthy and unexpected. It is possible
that the use of basal insulin as a strategy in combination with metformin and sulfonylureas
was partially responsible (9). The high proportion of Asian patients and thus the
slightly lower BMI in the trial compared with other large studies of type 2 diabetes
might also have contributed. However, perhaps the relatively low A1C in the recruited
population on entry into the trial is the most likely explanation. Such individuals
would be losing relatively small amounts of glucose in their urine so that tightening
their control would not have caused weight gain due to a loss of glycosuria, an important
cause of increased weight among patients with type 2 diabetes when diabetic control
is tightened (10).
This trial and others (11) have demonstrated clearly that at least in the short term
(3–5 years), tightening glycemic control from moderate to fairly tight levels of A1C
(6–6.5%) has no significant effect in reducing macrovascular disease. The results
of ADVANCE do not exclude a minor effect on cardiovascular disease; indeed cardiovascular
mortality was reduced by 12%, albeit nonsignificantly. However, clinicians can now
concentrate on targeting other aspects of type 2 diabetes, particularly BP and cholesterol
reduction in the clear knowledge that this will have a larger effect in preventing
macrovascular complications.
Nevertheless, it may be that the fall in the incidence of nephropathy that was demonstrated
in ADVANCE might have longer-term downstream benefits on cardiovascular disease. Patients
with diabetic nephropathy have a much higher risk of macrovascular disease, and so
a 20% reduction could be expected to affect cardiovascular disease over a longer time
frame (12). It is noteworthy that the STENO-2 trial, which tested an aggressive strategy
targeting glucose, BP, and lipids in high-risk individuals with type 2 diabetes, only
demonstrated clinically relevant reductions in cardiovascular disease after 8 years
and mortality after 13 years (13).
Clinicians will need to decide whether the reduction in diabetic nephropathy among
patients with type 2 diabetes when A1C levels are lowered to ∼6.5% is worthwhile.
However, since glycemic control was achieved in ADVANCE across multiple centers worldwide
with relatively few side effects, this level of glycemic control appears to be both
practical and provides useful clinical benefit. Current clinical guidelines that recommend
A1C levels of between 6.5 and 7% are supported by the data from ADVANCE.
In conclusion, the ADVANCE trial has demonstrated in over 10,000 patients with type
2 diabetes, that an intensive strategy with conventional agents can achieve mean A1C
levels of 6.5% safely with no increase in mortality and has no significant effect
in reducing macrovascular disease, but reduces diabetic nephropathy by ∼20%.