It is now common knowledge that we are in the midst of a worldwide epidemic of type
2 diabetes (T2D). This disease occurs disproportionately in African Americans, Hispanic
Americans, and other minorities. T2D is a major cause of cardiovascular disease morbidity
and mortality, and these unfavorable outcomes are greatly increased in the presence
of T2D (1–4). Impaired glucose tolerance (IGT) is itself a risk factor for cardiovascular
disease (5). Obesity and IGT are each major risk factors for T2D, with yearly incidence
of T2D in individuals with IGT ranging from 3.9 to 8.7% in various American populations,
with an average of 5.7% (6), and averaging 6.5% in a meta-analysis of global studies
(7). Both insulin resistance and diminished β-cell function (insulin secretion) are
crucial pathogenetic factors in the progression from IGT to T2D, with the proportional
impact of these defects varying in different reports and populations (8). The rate
of progression can be reduced by lifestyle modification leading to weight loss (9)
and the use of metformin (9), the thiazolidinedione drugs pioglitazone (PIO) (10)
and rosiglitazone (11), α-glucosidase inhibitors (12), sulfonylureas (13), and insulin
(14). Selection of therapeutic agents could be enhanced by matching their dominant
actions on insulin resistance and/or β-cell dysfunction to the relative dominance
of these abnormalities in specific individuals or populations with IGT.
For the above reasons, the study by DeFronzo et al. (15) in this issue is of particular
interest. In the Actos Now for the prevention of diabetes (ACT NOW) study (10), which
included 602 high-risk IGT patients, PIO reduced the progression to T2D over 2.4 years
from an annual incidence of 7.6% in the placebo (PLAC) group by 72% (95% CI 51–84%);
48% of the PIO group reverted to normal glucose tolerance (NGT) compared with 28%
in the PLAC group (P < 0.001). In the current report, 441 (73%) of these subjects
had oral glucose tolerance tests at baseline and at study completion, with sampling
of plasma glucose, insulin, and C-peptide. The protocol permitted calculation of insulin
sensitivity by the Matsuda Index (16) as well as insulin secretion rate using plasma
insulin and deconvolution of plasma C-peptide (17). Insulin sensitivity increased
92% in the PIO group compared with 17% in the PLAC group (P = 0.002), whereas insulin
secretion rate increased 33% in the PLAC group compared with 11% in the PIO group
(P < 0.001). The Disposition Index (insulin secretion rate/insulin resistance) increased
66% with PIO compared with 12% with PLAC (P < 0.001). The focus of DeFronzo et al.,
however, was on the relative changes from baseline of these derived parameters of
glucose metabolism. Subjects were grouped into three categories according to their
follow-up oral glucose tolerance tests: progression to T2D (N = 15 PIO and 45 PLAC),
no change in IGT status, and reversion to NGT. In subjects who reverted to NGT, insulin
sensitivity was increased 96% with PIO compared with PLAC, whereas insulin sensitivity
changed little in those who progressed to T2D in either group. Insulin secretion in
response to glucose declined with PIO and PLAC in those who progressed to T2D. Of
importance, the Disposition Index fell slightly and similarly in both these groups.
In an analytical tour de force, a plot of incidence of T2D in the combined PIO and
PLAC groups was negatively associated with octiles of the decrease in the natural
logarithm of the Disposition Index with an extraordinary correlation coefficient of
0.99. The use of the somewhat unusual octiles reflected a curvilinear association
between the variables. While this exceedingly strong association was driven by an
outlier value at the lowest octile of the Disposition Index that corresponded to the
highest incidence of diabetes, the remaining 7 points yielded a visually convincing
but much more shallow curvilinear association. IGT subjects who experienced a >80%
improvement in the Disposition Index had an approximate 2% incidence of diabetes whereas
the incidence was 14% in those whose Disposition Index declined 60–80%.
Better final glucose tolerance status (NGT > IGT > T2D) was categorically associated
with increase in insulin sensitivity (odds ratio [OR] = 0.61), insulin secretion (OR
= 0.61), and Disposition Index (OR = 0.26). In multivariate analyses of factors associated
with end of study outcome, NGT and IGT data were combined and compared with T2D. In
the PIO group, the combined favorable outcome was associated with a 21% increase in
insulin secretion rate, and a 70% increase in insulin secretion in response to glucose.
There was a 109% increase in insulin sensitivity comparing only the group who reverted
from IGT to NGT to the group that did not.
In assessing all of these impressive results, several points should be considered.
The authors strongly assert that the overall conclusions indicate that improvement
in β-cell function, i.e., the Disposition Index, is the dominant factor associated
with final glucose tolerance status, and that the beneficial effect of PIO on reduced
progression to T2D or increased reversion to NGT is largely attributable to the thiazolidinedione
effect on this factor (OR = 0.26). This conclusion is consistent with most previous
reports showing that insulin resistance is a factor in the earlier stages of IGT (8,18)
whereas deteriorating insulin secretion in response to glucose is a later determinant
of worsening glucose tolerance and severity of T2D (19). Hence, the decrease in conversion
rate of IGT to T2D reflects less deterioration of β-cell function in the PIO group.
However, the Disposition Index incorporates both of these functional parameters of
glucose metabolism, and increases in both insulin sensitivity and insulin secretion
were individually associated with the final impact of PIO therapy.
This should not be surprising. Although the temporal sequence of insulin resistance
and diminished β-cell responsiveness cited above is widely accepted and appears logical,
this does not mean that they are independent events. Several animal models support
a homeostatic relationship between insulin sensitivity and insulin secretion (Fig.
1). Decreased sensitivity to insulin elicits production, especially by the liver,
of betatrophin, a molecule that stimulates proliferation of β-cells, expansion of
β-cell mass, and improvement in glucose tolerance in experimental animals (20). Moreover,
insulin resistance—specifically at the level of the β-cell—impairs first phase insulin
secretion in response to glucose (21).
FIG. 1.
Established and speculative beneficial effects of PIO.
Thus, there are two ways to interpret PIO’s effects on the Disposition Index and its
individual components (Fig. 1). The most straightforward is that PIO is a thiazolidinedione
and has positive actions on both insulin sensitivity in its target tissues as well
as on β-cell function (22). Plasma glucose homeostasis is thereby maintained at an
acceptable level because PIO only rarely causes hypoglycemia. A second speculative
construct is to posit a brain regulating center with overall control of total body
insulin action (Fig. 1). When there is need for more total body insulin action on
all its target tissues, in addition to its effects on glucose metabolism, this center
would simultaneously increase insulin sensitivity and insulin secretion. Of course,
there would be need for a negative feedback plasma/intracellular glucose brake on
the output of such neurons to prevent hypoglycemia. PIO has been shown to affect glucose
metabolism in brain cells (23). The possibility that the ultimate seat of T2D could
lie in the brain has also been suggested before (24), given the importance of insulin
in regulating energy expenditure and storage through central as well as peripheral
effects. PIO could presumably act by stimulating the output of such putative neurons.
These two alternative mechanisms of PIO action, peripheral and central, need not be
mutually exclusive.
These speculative ruminations aside, the main importance of the observations by DeFronzo
et al. may be to encourage further development of T2D treatments that have ameliorative
effects on glucose dysmetabolism, but less severe and less frequent adverse effects
on the risks of congestive heart failure (25), fractures (26), and possibly bladder
cancer (27). Their observations on the natural history of progression or reversion
of IGT could also be applied to modification or development of other therapeutic agents.
From bedside to bench and back to bedside should be a productive therapeutic path
to follow.