| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Department of Internal Medicine IV (Endocrinology, Metabolism, Angiology, Pathobiochemistry and Clinical Chemistry), University of Tübingen, D-72076 Tübingen, Germany
 |
Abstract |
|---|
 Top Abstract I. IntroductionII. Pathophysiology and...III. Pharmacological TreatmentIV. PerspectivesV. Summary and ConclusionReferences |
|---|
-glucosidase inhibitors, biguanides,
thiazolidinediones, sulfonylureas, and insulin. Finally, a rational
treatment approach is proposed based on the dynamic pathophysiological
abnormalities of this highly heterogeneous and progressive disease.
I. Introduction
II. Pathophysiology and Pathogenesis of Insulin Resistance
A. Introduction
B. Multiple sites of insulin resistance: muscle, liver, and adipose tissue
C. Pathogenesis of insulin resistance
D. Inactivity-related insulin resistance
E. Molecular events in obesity-related insulin resistance
III. Pharmacological Treatment
A. -Glucosidase Inhibitors
B. Biguanides
C. Thiazolidinediones
D. Insulinotropic agents
E. Insulin
IV. Perspectives
A. Agents to enhance insulin action
B. Agents to increase insulin secretion
C. Agents to inhibit fatty acid oxidation
V. Summary and Conclusion
|
I. Introduction |
|---|
|
TopAbstractI. IntroductionII. Pathophysiology and...III. Pharmacological TreatmentIV. PerspectivesV. Summary and ConclusionReferences |
|---|
) of long existing metabolic
disturbances with deleterious effects on the vascular system, tissues,
and organs. Consequently, urgent efforts are required to avoid the
growing number of patients with this form of a "silently killing"
metabolic disease.
|
After an introduction into the pathophysiology and molecular
pathogenesis of insulin resistance, this review will focus on the
mechanism of insulin action and the capability of the available
antihyperglycemic pharmacological agents to treat insulin resistance.
In the conclusion a rational treatment approach, based on the dynamic
pathophysiological abnormalities of the disease, is proposed. The
importance of optimal treatment of other abnormalities often associated
with type 2 diabetes, i.e., obesity, hypertension,
dyslipidemia, disturbances in the fibrinolytic system, becomes evident
when the pathophysiology of the type 2 diabetic syndrome is examined
closely (see Fig. 1
). These essential aspects in the medical care of
type 2 diabetic patients have been recently reviewed (14, 15, 16, 17, 18, 19, 20) and thus
will not be covered in this review.
|
II. Pathophysiology and Pathogenesis of Insulin Resistance |
|---|
|
TopAbstractI. IntroductionII. Pathophysiology and...III. Pharmacological TreatmentIV. PerspectivesV. Summary and ConclusionReferences |
|---|
B. Multiple sites of insulin resistance: muscle, liver, and adipose
tissue
The term "insulin resistance" in humans is frequently used
synonymously with impaired insulin-stimulated glucose disposal (3, 22, 23) as measured with the hyperinsulinemic-euglycemic clamp technique
(24). Consequently, basic research in the area of insulin resistance as
a fundamental component of the pathogenesis of type 2 diabetes has
focused on tissues responsible for insulin-mediated glucose uptake,
namely muscle and, to a minor degree, adipose tissue (5). However, it
is well known that not only muscle glucose uptake but also adipose
tissue lipolysis and suppression of glucose production are regulated by
insulin.
1. The euglycemic-hyperinsulinemic clamp for the assessment of insulin resistance in vivo. It is unquestioned that in conditions commonly associated with the term "insulin resistance," such as obesity or type 2 diabetes, peripheral glucose disposal, as measured by a hyperinsulinemic-euglycemic clamp, is lower for the level of hyperinsulinemia achieved compared with healthy subjects. However, it is important to point out that, in people with type 2 diabetes, glucose uptake by skeletal muscle, both in the fasting state and postprandially, although inefficient for prevailing insulin levels, is not reduced in an absolute sense (25, 26, 27). In an attempt to identify "insulin resistance genes" underlying the disease (9, 28, 29), the hyperinsulinemic-euglycemic clamp has also been used to determine insulin resistance in healthy subjects with a first-degree family history of type 2 diabetes, who are of normal weight and whose glucose tolerance is normal. The classical hyperinsulinemic-euglycemic clamp, however, generates insulin levels above those these subjects usually experience and may therefore fail to reveal potential abnormalities of processes regulated by lower insulin concentrations. The manner in which insulin sensitivity is determined during the hyperinsulinemic-euglycemic clamp (using MCR, i.e., glucose infusion rate divided by plasma glucose at steady state) is based upon the assumption [unless appropriate tracer techniques are used (30)] that endogenous glucose production [largely attributable to liver, less so to kidney (31)] is completely shut off by the insulin infusion. This implies, however, that suppression of glucose production is regulated by much lower insulin concentrations than stimulation of glucose uptake. This should make the liver (and kidney) a target for insulin resistance whose effects on glucose homeostasis would be at least as important as those of muscle insulin resistance. In fact, excessive basal glucose production in the presence of fasting hyperinsulinemia is a key feature of type 2 diabetes (32, 33, 34, 35). Moreover, defective suppression of endogenous glucose production by normal or elevated insulin levels has been observed in type 2 diabetes (36, 37). Both observations demonstrate that insulin resistance of glucose production is involved in the pathogenesis of type 2 diabetes.
2. Potential role of adipose tissue for insulin resistance. In addition to muscle and liver, adipose tissue is the third metabolically relevant site of insulin action. While insulin-stimulated glucose disposal in adipose tissue is of little quantitative importance compared with that in muscle, regulation of lipolysis with subsequent release of glycerol and FFA into the circulation by insulin has major implications for glucose homeostasis. It is widely accepted that increased availability and utilization of FFA contribute to the development of skeletal muscle insulin resistance (38, 39, 40). Moreover, FFA have been shown to increase endogenous glucose production both by stimulating key enzymes and by providing energy for gluconeogenesis (41). Finally, the glycerol released during triglyceride hydrolysis serves as a gluconeogenic substrate (42). Consequently, resistance to the antilipolytic action of insulin in adipose tissue resulting in excessive release of FFA and glycerol would have deleterious effects on glucose homeostasis.
3. Glucose uptake vs. glucose production: comparison
of EC50s. Only one study has examined the entire
insulin dose response characteristics for stimulation of glucose uptake
and suppression of glucose production in normal and type 2 diabetic
subjects (43). This study showed a significant right shift of the
dose-response curve for glucose uptake with an
EC50 for glucose uptake (58 µU/ml) more than
double that for glucose production (26 µU/ml). In another study also
using the stepwise hyperinsulinemic-euglycemic clamp, a plateau for
glucose uptake was not reached at the highest insulin concentration.
Thus, dose response characteristics could only be approximated but
appeared to range in the same order (44). These findings clearly
demonstrate that with low physiological increments in plasma insulin,
the liver is the primary determinant of whole body glucose homeostasis.
In patients with type 2 diabetes the dose-response curves for both
glucose uptake and glucose production were markedly shifted to the
right (43, 44). The EC50 values for glucose
uptake (EC50, 118 µU/ml) and glucose production
(EC50, 66 µU/ml) in the patients with type 2
diabetes, however, were increased similarly (
2-fold) compared with
normal subjects. Similarly, in obesity a parallel right shift of the
dose-response curve for both glucose disposal and production was found
(45). Thus, the relative impairment in the sensitivity of glucose
uptake and suppression of glucose production was not different, which
suggests that both processes are equally resistant to insulin in type 2
diabetes. At plasma insulin concentrations below 50 µU/ml, however,
impaired suppression of glucose production appears to contribute
quantitatively more than defective glucose uptake to the abnormal
glucose homeostasis of type 2 diabetes.
4. Lipolysis is most sensitive to insulin. More data on whole body lipolysis as determined by isotopic measurements of glycerol appearance in plasma are available from stepwise hyperinsulinemic-euglycemic clamps. The insulin EC50 values for suppression of lipolysis in normal subjects ranged between 7 and 16 µU/ml (44, 46, 47, 48, 49), placing the dose-response curve of adipose tissue distinctly left of those for glucose production and glucose uptake. From these studies it becomes evident, that lipolysis is the process most sensitive to the action of insulin with a greater than 90% effect well within the physiological insulin range. In obese subjects and patients with type 2 diabetes the EC50 values are increased 2- to 3-fold (44, 47, 49), indicating that adipose tissue lipolysis is at least as resistant to the action of insulin as muscle and liver. Since failure to adequately turn off lipolysis directly affects liver (and kidney) and muscle metabolism while the reverse does not hold true, it is tempting to speculate that adipose tissue might even be a primary site for the defect leading to insulin resistance elsewhere and, ultimately, to type 2 diabetes.
To summarize, lipolysis is the most insulin-sensitive process followed by glucose production and, far behind, glucose uptake with EC50 values in the physiological range only for insulin-induced inhibition of lipolysis and glucose production, but not for insulin-stimulated glucose uptake. In full-blown diabetes mellitus, insulin sensitivity in muscle, liver (and kidney), and adipose tissue is compromised to a similar degree. While several studies unanimously showed defective insulin action on glucose uptake and lipolysis in prediabetic states [obesity, impaired glucose tolerance (IGT)], data for suppression of glucose production are somewhat divergent. During an oral glucose tolerance test, suppression of endogenous glucose production was significantly diminished in IGT (50), whereas during a hyperinsulinemic-euglycemic clamp, suppression was comparable between IGT and normal glucose-tolerant (NGT) (51).
C. Pathogenesis of insulin resistance
Insulin resistance, as determined by the
euglycemichyperinsulinemic clamp technique, reflects defective
insulin action predominantly in skeletal muscle and liver. The major
causes of skeletal muscle insulin resistance in the prediabetic state
may be grouped into genetic background-related and as obesity- and
physical inactivity-related (Fig. 2).
Despite intensive research efforts, there is, so far, no clear
understanding of the factors that define the genetic accessibility of
insulin resistance. One approach to analysis of the genetic background
is to define candidate genes based on the present knowledge of the
insulin-signaling chain. We have recently reviewed the present
knowledge of the insulin-signaling chain (52). Abnormalities in insulin
signaling that may induce insulin resistance in type 2 diabetes will be
discussed in the following section.
|
shows a simplified draft of the
signaling steps from insulin receptor binding to glucose transport
activation. Insulin signaling at the cellular level is mediated by
binding of insulin to a specific receptor. Insulin binding to the
receptor stimulates autophosphorylation of the intracellular region of
the receptor ß-subunit (53). A reduced autoactivation status of the
insulin receptor from skeletal muscle and adipocytes of type 2 diabetic
patients has been described by several but not all investigators
(54, 55, 56, 57, 58, 59, 60, 61, 62). Some of these studies have shown that obesity was a major
contributory factor for the development of a reduced insulin receptor
activity (56, 63). This could suggest that the defective insulin
receptor kinase activity is secondarily acquired due to obesity and
metabolic changes such as hyperinsulinemia and hyperglycemia.
|
b. PI 3-kinase and protein kinase B (PKB).
At the molecular
level, insulin causes activation of the insulin receptor and
phosphorylation of IRS proteins on tyrosine residues. Phosphorylation
of IRS proteins creates binding sites for PI 3kinase and enables
activation of PI 3-kinase. The activated PI 3-kinase converts PI 4- or
PI 4,5-phosphate into PI 3,4- and PI 3,4,5-phosphate
(PIP3). PIP3 can bind
PKB/AKT (for cellular homolog of the transforming oncogene v-akt) and
phosphatidylinositol-3,4,5-phosphate-dependent kinase-1 (PDK-1)
by their PH (pleckstrin homologous) domains (72). Colocalization
of PKB/AKT and PDK-1 at the plasma membrane region enables
phosphorylation of PKB/AKT on threonine308 by
PDK-1. PKB/AKT regulates several protein kinase cascades involved in
insulin signal transduction to glucose uptake, to glycolysis, to
glycogen synthesis as well as to protein synthesis (73). In addition to
phosphorylation of PKB/AKT, there is evidence that PDK-1 is also able
to phosphorylate protein kinase C (PKC) isoforms (74, 75).
Insulin-dependent activation of the atypical PKC
isoform has been
demonstrated recently (76). Recent evidence suggests that PDK-1
mediates insulin-dependent activation of atypical PKC through
phosphorylation on threonine410 in the activation
loop (74, 75). In addition, insulin-dependent stimulation of atypical
PKC has been shown to mediate insulin effects on protein synthesis
(76). Moreover, there is evidence that the atypical PKC isoforms
and 
are involved in coupling of the insulin signal to the glucose
transport system (77, 78). This demonstrates that insulin-stimulated
glucose transport can be mediated via different signaling cascades.
This signaling diversity potentially opens compensatory mechanisms if
gene mutations were to occur, for example, in PKB/AKT or atypical PKCs.
Expression level and possible gene mutations of PI 3kinase and PKB in insulin-resistant and diabetic patients have been investigated by a small number of studies. Decreased activation of PKB in skeletal muscle of type 2 diabetic patients in spite of normal protein levels has been described (79). However, these results are controversial since another recently published study described normal PKB/AKT activation in skeletal muscle of type 2 diabetic patients (80). In skeletal muscle of lean and obese type 2 diabetic patients, decreased IRS-1 phosphorylation and PI 3-kinase activity as well as a 50–60% reduction in the protein expression level of IRS-1 and PI 3-kinase have been shown (81, 82). Thus, decreased expression and phosphorylation level of early insulin signaling elements (i.e., IRS, PI 3-kinase, and PKB) have been demonstrated in insulin target tissues of type 2 diabetic patients. This may contribute to insulin resistance in type 2 diabetic patients. However, it does not necessarily mean that this represents a genetic defect since it is not clear to what extent metabolic disturbances are able to induce the above mentioned signaling defects.
2. Possible genes for insulin resistance and obesity. Substantial evidence that type 2 diabetes is an inherited disease was demonstrated by twin studies, familial clustering of type 2 diabetes, and the high prevalence of this disease in some ethnic groups. Efforts to identify potential type 2 diabetes and insulin resistance genes have been undertaken by screening of different candidate genes and genome scans.
a. Candidate gene studies for insulin resistance and type 2
diabetes.
1. Insulin receptor. In the candidate gene approach, several genes have
been screened for their potential role in the development of insulin
resistance. A wide spectrum of insulin effects is mediated by the
insulin receptor and its substrates IRS-1 and IRS-2 as well as the PI
3-kinase (reviewed in Ref. 5). Therefore, these genes have been tested
for their potential role in the pathogenesis of insulin resistance. It
is now well established that mutations of both insulin receptor alleles
occur in very rare cases and cause severe syndromes of insulin
resistance (e.g., leprechaunism, and Rabson-Mendenhall
syndrome), which, in most patients, result in death during the first
year (reviewed in Refs. 83, 84). Other insulin receptor mutations
affecting only one allele are compatible with life and cause severe
insulin resistance syndromes (called "type A insulin resistance")
often without developing hyperglycemia during young adulthood. Since
these early reports have clearly demonstrated that insulin receptor
mutations could induce insulin resistance, patients with common type 2
diabetes were also screened for the presence of insulin receptor gene
mutations. Thus far, several insulin receptor mutations (Lys1068Glu,
Arg1152Gln, and Val985 Met) have been identified in about 1–5% of
patients with common type 2 diabetes (85, 86, 87). Only one
population-based study in the Netherlands could demonstrate a Val985
Met mutation of the insulin receptor at a relatively high rate of 5.6%
(87), which was not found in other population groups (85). Functional
characterization of the described insulin receptor mutations in type 2
diabetes revealed only mild degrees of insulin-signaling defects.
However, overt diabetes may develop in combination with other genetic
defects. In summary, insulin receptor mutations were not commonly found
in random type 2 diabetes, and only a small number of individuals may
have mutations that could contribute to insulin resistance, probably in
concert with other genetic defects which are not yet identified.
2. IRS-1 and -2. Mutations of IRS-1 and IRS-2 have also been described
in humans. However, these mutations were found with the same frequency
in nondiabetic compared with diabetic individuals (12% for
Gly972Arg mutation in the IRS-1 gene and 33% for Gly1057Asp in the
IRS-2 gene) (88, 89). Cell culture studies indicated that the mutation
in codon 972 of IRS-1 impairs insulin-stimulated signaling (90).
Whether this mutation is correlated with insulin resistance in
vivo seems contradictory at present (88, 91, 92). It appears,
however, that Gly972Arg is associated with a slightly lower insulin
secretion rate (88, 91, 93), which has recently been confirmed by
in vitro studies (94) and which might also contribute to
the development of type 2 diabetes. In addition to IRS-1, an amino acid
polymorphism of the IRS-2 gene causing replacement of glycine to
aspartate at position 1,057 was found at a high frequency of 33% in an
unselected Scandinavian population. This amino acid exchange, however,
was not associated with type 2 diabetes (89). Furthermore, genome
screening of the IRS-2 locus has been performed in families with
early-onset autosomal dominant type 2 diabetes (95). The results of
this study did not suggest that the IRS-2 gene represents a major
pathogenetic factor in this highly selected group. In summary,
mutations of the IRS-1 and IRS-2 genes seem to occur at a relatively
high rate of 12–33% in nonobese healthy, as well as type 2 diabetic,
human subjects (88, 89). Although some data suggest impaired insulin
action by these mutations, the high prevalence in healthy subjects does
not support a major role in the development of type 2 diabetes in
humans.
3. PI 3-kinase. Gene mutations in the PI 3-kinase gene have also been
studied. Screening for mutations in the PI 3-kinase gene could be
complicated by the existence of several isoforms of the PI 3-kinase
regulatory and catalytic subunit (reviewed in Ref. 96). In human
skeletal muscle more than four different regulatory subunit variants
are expressed and differently regulated by insulin (97). It has been
shown that a splice variant of approximately 50 kDa of the p85
regulatory subunit of PI 3-kinase is highly sensitive upon insulin
stimulation in human skeletal muscle (97). Although PI 3kinase
activation seems crucial for insulin-dependent glucose uptake, mice
lacking the p85 subunit of PI 3-kinase are surprisingly more insulin
sensitive and mildly hypoglycemic (98). This has been explained by a
switch from p85 to the p50 subunit expression and activation
which led to increased generation of phosphatidylinositol
3,4,5-phosphate (98). These results demonstrate that interpretation of
potential mutations in the regulatory subunit of PI 3-kinase is
difficult without the knowledge of total PI 3-kinase activity and the
functional status and expression level of other regulatory isoform
subunits. Screening for PI 3-kinase mutations in human subjects has
revealed a mutation at codon 326 replacing methionine by isoleucine in
the regulatory subunit. This mutation was found in a Scandinavian
insulin-resistant population at a frequency of approximately 30% in
its heterozygous form and 2% in its homozygous form. The homozygous
mutation was found to be associated with a significant reduction of
insulin sensitivity (99). However, this could not be found in Japanese
type 2 diabetic patients (100). Moreover, in Pima Indians, this
mutation was not associated with insulin resistance but rather with an
increased acute insulin response after a glucose challenge test (101).
It has been suggested by these investigators that the Met326Iso
mutation might even protect homozygous carriers in the female Pima
population against the development of type 2 diabetes. This may also
agree with the data from p85 knockout mice, which are characterized
by increased insulin sensitivity and hypoglycemia instead of
developing diabetes (98).
4. Other candidate genes. In addition to these early insulin-signaling elements, mutations of the liver glucokinase promoter, of GLUT4, glycogen synthase, and the protein phosphatase-1, among others, have also been identified, but these mutations were not associated with insulin resistance or type 2 diabetes apart from a very few cases (reviewed in Ref. 102).
Although a large number of genes remain to be screened for their potential role in insulin resistance, it can be concluded from present studies that heterozygous mutations in insulin signaling molecules are often found with a high frequency in human subjects. In most cases these mutations are not sufficient to cause insulin resistance or type 2 diabetes. However, if these mutations are in rare cases homozygous or occur together with mutations of other insulin-signaling proteins or obesity, this combination of different disturbances might ultimately lead to insulin resistance and type 2 diabetes. This would also be in agreement with the postulated polygenic pathogenesis of type 2 diabetes.
b. Genome scans for susceptibility genes for insulin resistance and
type 2 diabetes.
While the candidate gene approach serves to
identify mutations of known genes, the method of genome scanning in
family cohorts or sib pairs could reveal previously undetected type 2
diabetes genes (103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119). This method has identified new diabetes loci
on different chromosomes, which are listed in Table 1 103–115). The chromosomal loci are
partially located in the vicinity of known genes such as the hepatic
nuclear factor 1 (HNF 1), the sulfonylurea receptor, the
apolipoprotein A-2, and others. To date, most of the genome scan
studies suggested that there are different chromosomal linkage regions
for type 2 diabetes confirming the role of diabetes susceptibility
genes. However, these gene loci are often restricted to a special trait
and ethnic group, which means that they are probably not a major gene
locus for the large group of common type 2 diabetic patients. A more
general impact for common type 2 diabetes has recently been discussed
for a gene locus on chromosome 20 near the hepatic nuclear factor 1
gene, but more studies are needed to identify this gene and to evaluate
its potential role for the development of diabetes.
|
E. Molecular events in obesity-related insulin resistance
The negative impact of increased body fat mass on insulin
sensitivity can be clearly shown for the vast majority of individuals.
Furthermore, the insulin-sensitizing effect of weight reduction and
physical training is well documented (reviewed in Refs. 10, 11).
1. The role of FFA in obesity-related insulin resistance. Among the signaling molecules that are derived from adipocytes, FFA have been implicated in the pathogenesis of insulin resistance (40, 126). FFA are generated via lipolysis mainly in fat cells. In insulin resistant and obese subjects increased FFA release into plasma can occur. Obesity-related insulin resistance leads to reduced antilipolytic effect of insulin (123). Another mechanism by which obesity could contribute to increased FFA production is overactivity of the sympathetic nervous system, which has been demonstrated in obese human subjects and type 2 diabetic patients (127, 128, 129). FFA are taken up by liver and skeletal muscle cells. They counteract the effects of insulin by increasing hepatic gluconeogenesis and by inhibiting glucose uptake and oxidation in skeletal muscle (130, 131, 132). This fatty acid-induced insulin resistance in liver and skeletal muscle has been suggested to be a result of increased acetyl-CoA production and inhibition of glucose oxidation by FFA (130, 133). The concept of a glucose-fatty acid cycle, which was originally described by Randle et al. (130), has now been called into question by Wolfe (134). While Randle et al. suggested that increased availability of FFA and fatty acid oxidation regulates glucose oxidation, Wolfe has developed a vice versa concept in which the rate of glycolysis rather than the availability of fatty acids regulates fatty acid oxidation. Evidence that glucose oxidation could directly regulate fatty acid oxidation by inhibition of fatty acid transport to the mitochondria was provided by this group recently (134). Nevertheless, increased fatty acids can regulate this process by reducing intracellular glucose availability through inhibition of glucose uptake. Therefore, the initiating effect of fatty acids to induce insulin resistance could be inhibition of glucose uptake, which would be followed by a decrease of intracellular glucose availability and glucose oxidation. This would consequently lead to increased fatty acid oxidation according to the concept of Wolfe.
2. The role of leptin in obesity-related insulin resistance. Leptin has gained much attention recently in the study of the underlying mechanisms of insulin resistance in obesity. In 1994, it was identified as an adipocyte-derived hormone in by Friedman and colleagues (135). Leptin reduces body weight via specific receptors in hypothalamic areas regulating energy expenditure and satiety (136, 137, 138). Secretion of leptin from fat cells is strongly dependent on body fat mass (reviewed in Ref. 138). Leptin deficiency and receptor defects in rodents cause marked obesity as well as hyperinsulinemia and hyperglycemia (139). Thus, many studies have focused on the effects of leptin on insulin resistance and insulin secretion. Both inhibition and stimulation of insulin action have been shown by leptin in different cell systems (140, 141, 142, 143). In addition, several groups have shown inhibitory effects of leptin on insulin secretion in isolated cell lines and perfused pancreatic islets (128), while others have found that leptin stimulates insulin secretion (144, 145, 146, 147, 148, 149, 150). Therefore, the conclusion that leptin causes a defect in the insulin-signaling chain or that it is capable of improving ß-cell dysfunction in human subjects cannot yet be made on the basis of these studies.
In human subjects congenital leptin deficiency or mutations of the leptin receptor occur in extremely rare cases. These mutations have been associated with severe obesity but not with diabetes (151, 152, 153). However, it must be considered that these few cases reported recently were of young age, and it remains to be seen whether IGT or diabetes may still develop with advancing age.
3. The role of tumor necrosis factor- (TNF) in
obesity-related insulin resistance. A great number of studies have
been performed in the last years to elucidate the role of TNF for
obesity-related insulin resistance. Spiegelman and co-workers recently
proposed that TNF may contribute to insulin resistance in obese
subjects (reviewed in Ref. 154). Several studies have shown that TNF
is able to impair insulin signaling through serine kinase and tyrosine
phosphatasedependent modulation of the insulin-signaling chain
(155, 156, 157). However, these studies have been performed in isolated cell
systems, and to date there is no evidence that these mechanisms are
relevant in type 2 diabetic patients. While the data from isolated cell
systems and animal models provide a plausible molecular basis for
TNF-induced insulin resistance, clinical results from different
insulin-resistant populations so far do not support a major role of
TNF on insulin resistance in humans (158, 159). However, one study
has shown increased adipose tissue expression of TNF in obese
premenopausal women when compared with control subjects (506).
4. The peroxisome proliferator-activated receptor-
(PPAR):
potential role for insulin resistance and ß-cell function.
Thiazolidinediones are pharmacological compounds that reduce insulin
resistance both in prediabetic as well as diabetic individuals (see
also Section III.C.). Thiazolidinediones are ligands of
the PPAR2 (160). PPAR2 is predominantly expressed in adipocytes,
intestine, and macrophages (161). There is some evidence that a low
level expression might also occur in muscle cells. The PPAR receptor
is a transcription factor that controls the expression of numerous
genes. It is assumed that the effect of thiazolidinediones on insulin
sensitivity is mediated through altered expression of
PPAR2-dependent genes (reviewed in Refs. 162, 163). Recently, the
Pro12Ala and two other polymorphisms were described in the PPAR2
receptor (164). It appears that the Pro12Ala polymorphism in its
heterozygous form occurs in approximately 30% of humans. Auwerx and
colleagues (165) have shown that this polymorphism appears to be
functionally relevant, leading to a reduced transcriptional activity
and improved insulin sensitivity (165). In our studies, an obese
subgroup with a body mass index (BMI) >35 kg/m2 carrying
this polymorphism in the heterozygous form appears to be less insulin
resistant compared with individuals without this PPAR2 mutation
(166). Although the total number of subjects studied is still very low,
it might be speculated that this polymorphism protects against the
negative influences of obesity on insulin sensitivity. Interestingly,
we also found differences in insulin secretion measured during an oral
glucose tolerance test. Individuals carrying the polymorphism showed
lower insulin secretion at 60 and 120 min. The lower secretion might be
interpreted as a consequence of the increased insulin sensitivity of
these individuals. Alternatively, it might be speculated that the
PPAR2 polymorphism directly interferes with ß-cell function. In
agreement with this, direct effects of PPAR agonists on ß-cells
have been demonstrated. Studies on isolated pancreatic islets and on a
hamster ß-cell line have shown that thiazolidinediones could enhance
glucose- and glibenclamide-induced insulin release (167). Although the
underlying mechanism for this direct effect on ß-cells is not
completely clear, it has been suggested that thiazolidinediones
stimulate insulin release by increase of glucose uptake in the ß-cell
(167). In animal studies, treatment with thiazolidinediones resulted in
improvement of pancreatic islet cell integrity and hyperplasia (168, 169). Moreover, it has been demonstrated that PPAR activation
reduces triglyceride content in islets of Zucker diabetic fatty rats,
leading to a significant increase in insulin secretion (170). However,
in humans only preliminary data are available concerning a direct
effect of PPAR agonists on pancreatic islets. One study demonstrated
that troglitazone treatment in humans could increase glucose-stimulated
insulin secretion (171). Thus, in addition to the insulin-sensitizing
effects, PPAR agonists may directly improve ß-cell dysfunction in
humans. However, clearly more studies are needed to investigate direct
effects of PPAR agonists on pancreatic ß-cells in humans.
|
III. Pharmacological Treatment |
|---|
|
TopAbstractI. IntroductionII. Pathophysiology and...III. Pharmacological TreatmentIV. PerspectivesV. Summary and ConclusionReferences |
|---|
A. -Glucosidase inhibitors
1. Mechanism of action. These agents delay digestion of
complex carbohydrates and disaccharides (starch, dextrin, sucrose) to
absorbable monosaccharides by reversibly inhibiting -glucosidases
within the intestinal brush border (glucoamylase, sucrase, maltase, and
isomaltase). This leads to a reduction of glucose absorption and,
subsequently, the rise of postprandial hyperglycemia is attenuated. The
currently available -glucosidase inhibitors are acarbose, miglitol,
and voglibose. Extensive and excellent reviews about their pharmacology
have been published (175, 176, 177, 178, 179, 180, 181, 182, 183).
2. Effect of -glucosidase inhibitors on hyperglycemia in
patients with type 2 diabetes mellitus. The effect of monotherapy
with -glucosidase inhibitors (usually 100 mg three times daily) on
postprandial hyperglycemia is well documented in numerous randomized
placebo-controlled studies, and the decrease of postprandial glycemia
averages about 3 mmol/liter (184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203). The effect of -glucosidase
inhibitors on fasting plasma glucose levels is less pronounced and
averages -1.3 mmol/liter. The overall effect of -glucosidase
inhibitors on glycemia of diet-pretreated subjects with type 2
diabetes, as determined by HbA1c-measurements, averages 0.9% (range,
0.6–1.4), as recently reviewed by Lebovitz (204).
Addition of acarbose to type 2 diabetic subjects pretreated with insulin, metformin, or sulfonylureas causes a reduction of HbA1c levels between 0.5 and 0.8%. This beneficial effect seems to last for at least 3 yr as has been recently shown in the UKPDS study. During the last 3 yr of this long-term trial, 379 patients were additionally treated with acarbose in a placebo-controlled design. This resulted in a mean reduction of the HbA1c by 0.5% in the group of patients who still took acarbose after 3 yr. This significant effect was sustained over the 3-yr time period (205). However, at 3 yr a significant lower proportion of patients were taking acarbose compared with placebo (39 vs. 58%), the main reasons for noncompliance being flatulence and diarrhea. Intention to treat analysis showed that all patients allocated to acarbose, compared with placebo, had 0.2% significantly lower HbA1c at 3 yr (205).
3. Effect of -glucosidase-inhibitors on insulin
sensitivity. Eight randomized placebo-controlled studies have been
published examining the effect of -glucosidase inhibitors on insulin
sensitivity in patients with IGT or type 2 diabetes mellitus (Table 2). In subjects with IGT, Chiasson
et al. (206) demonstrated that acarbose (100 mg three times
daily) for 4 months caused a 21% decrease in steady-state plasma
glucose (SSPG) during an insulin suppression test using somatostatin,
glucose, and insulin infusions. Similar results were obtained by Laube
et al. (207), who reported that 12 weeks of acarbose
treatment (100 mg three times daily) increased steady-state glucose
infusion rate (SSGIR) by 45%. In addition, Shinozaki et al.
(208) treated subjects with IGT with a different glucosidase
inhibitor, voglibose (0.2 mg three times daily), for 12 weeks, and
showed that SSPG levels decreased significantly after voglibose
treatment. Thus, these data suggest that -glucosidase inhibitors
improve insulin sensitivity in subjects with IGT and hyperinsulinemia
possibly secondary to an amelioration of glucose-induced insulin
resistance by reducing hyperglycemia in the postprandial period. In
contrast to studies in subjects with IGT, studies examining the effect
of -glucosidase inhibitors on insulin sensitivity in patients with
type 2 diabetes showed no amelioration of insulin resistance despite
decreased postprandial glycemia (209, 210, 211, 212, 213). Thus, these data are in
support of the notion that -glucosidase inhibitors improve insulin
sensitivity in subjects with IGT but have no effect on insulin
sensitivity in subjects with overt type 2 diabetes.
|
-Glucosidase inhibitors in type 2 diabetes prevention
studies. Currently, three type 2 diabetes prevention trials
examining the effect of acarbose on the conversion rate from IGT to
type 2 diabetes are under way. The Early Diabetes Intervention Trial
(EDIT), the Dutch Acarbose Intervention Trial (DAISI), and the Study to
Prevent NIDDM (STOP-NIDDM). The STOP-NIDDM is the largest trial
including more than 1,400 IGT-subjects recruited until February 1998.
The study has a randomized double-blind placebo-controlled design, and
the recently published preliminary screening data (214) provide
interesting information on the population under study. In a preliminary
subset of 3,919 screened subjects, preselected by known risk factors to
develop type 2 diabetes (BMI > 27 kg/m2,
history of diabetes, hypertension, dyslipidemia, and gestational
diabetes in women) 13.3% had previously undetected diabetes and 17.3%
had IGT. A total of 1.418 IGT subjects identified during the screening
procedure were included in the study for a predictive median follow-up
period of 3.9 yr. The results will be available by 2002, and it will be
interesting to see whether treatment with acarbose is able to decrease
the conversion rate of IGT to manifest type 2 diabetes mellitus in a
higher proportion than nonpharmacological intervention protocols
including dietary advice and exercise in the Da Qing study (215). In addition, two other multicenter studies are investigating the effect of diet, increased physical activity or metformin [Diabetes Prevention Program (DPP)] and diet, increased physical activity and sulfonylurea [Fasting Hyperglycemia Study (216)] to prevent type 2 diabetes mellitus. The results of these long-term studies will be available between 2002 and 2004.
5. Adverse effects of -glucosidase inhibitors.
-Glucosidase inhibitors have not been associated with
life-threatening adverse effects, possibly due to the low systemic
absorption.
a. Gastrointestinal adverse effects.
The major adverse effects
associated with acarbose therapy are gastrointestinal complaints,
including flatulence and abdominal discomfort, resulting from
malabsorption and consequently increased fermentation of carbohydrates.
Depending on the acarbose dosage used (300–900 mg/day), the frequency
of gastrointestinal effects was as high as 56–76% (placebo, 32–37%)
in earlier studies (217). When the new recommendations for use of
-glucosidase inhibitors were considered in the study protocols (low
acarbose starting dose of 50 mg/day, slow increase of dosage over
weeks, maximum dose 100 mg three times daily), the incidence of
gastrointestinal adverse effects were reported to be as low as 7.5%
(203). Furthermore, it has been shown that the incidence of
gastrointestinal side effects decreased during long-term treatment
(218).
b. Systemic effects.
The systemic availability of
nonmetabolized acarbose is reported to be 0.5–1.7% (217, 219, 220).
Due to the low systemic absorption of acarbose, systemic
effects are rare. However, liver transaminase elevations [defined as
treatment-induced increases of alanine aminotransferase (ALT) and/or
aspartate aminotransferase (AST) > 1.8-fold the upper limit of
the normal range] were documented in 3.8% of acarbose recipients
(placebo 0.9%) in the early studies carried out in the United States,
using high acarbose daily dosage (900 mg/day) (221). Animal studies on
ethanolinduced hepatotoxicity revealed that high-dose acarbose
treatment augmented ethanol-induced hepatotoxicity (222). However, in
all major acarbose trials using 100 mg three times daily as the maximum
dose, hepatic transaminase elevations were extremely rare (203, 223)
and in the five cases published, transaminase levels were reversible
upon withdrawal of the drug (224, 225, 226, 227). Furthermore, a recent study
from Japan demonstrated that acarbose treatment in patients with
chronic liver disease and diabetes mellitus was effective and caused no
significant alterations in hepatic transaminase levels after 8 weeks of
treatment (228). Recently, it has been reported that acarbose induced a
generalized erythema multiforme in a middle-aged Japanese type 2
diabetic patient (229).
6. Guidelines for the clinical use of -glucosidase inhibitors.
a. Selection of the most appropriate patients.
Postprandial
hyperglycemia represents a major metabolic disturbance of carbohydrate
metabolism in IGT and early phase type 2 diabetic subjects. Since
-glucosidase inhibitors decrease postprandial glycemia these
patients are suitable candidates for treatment with -glucosidase
inhibitors, provided that the individual therapeutic goal was not
achieved by dietary advice and increased physical activity. In type 2
diabetic patients suffering predominantly from fasting hyperglycemia,
glucosidase inhibitors are less effective but may be used in
combination with other antihyperglycemic agents, such as metformin,
sulfonylureas, or insulin. The results of the UKPDS have shown that
combination therapy using these drugs is effective and safe over at
least 3 yr. In patients remaining on their allocated therapy, the
HbA1c-difference at 3 yr was 0.5% lower in the acarbose study group
compared with placebo (205).
B. Biguanides
1. Introduction. There is now a large body of data
documenting the clinical efficacy of metformin in the treatment of type
2 diabetes (230), and most of its clinical, pharmacological, and basic
cellular aspects have been addressed in several excellent reviews
published during the past 20 yr (231, 232, 233, 234, 235, 236, 237, 238). Recently, the UKPDS showed
that metformin is particularly effective in overweight type 2 diabetic
subjects, a condition usually characterized by insulin resistance
(239). Moreover, in essentially all clinical studies the improvement of
hyperglycemia with metformin occurred in the presence of unaltered or
reduced plasma insulin concentrations (e.g., Refs. 240, 241). Taken collectively, these findings indicate the potential of
metformin as an insulin-sensitizing or insulin-mimetic drug, which is
the focus of the following.
Despite almost 40 yr of research, the precise cellular mechanism of metformin action is still not entirely understood. Several cellular mechanisms have been described but a single unifying site of action, such as a receptor, an enzyme, or a transcription factor, has yet to be identified. Nevertheless, it is generally undisputed that metformin has no effect on the pancreatic ß-cell in stimulating insulin secretion (234). Mild increases in glucose-stimulated insulin secretion after metformin treatment (242) are thought to be the result of reduced glucose toxicity on the ß-cell secondary to improved glycemic control (243).
2. Mechanisms of action in humans.
a. Glucose production.
Accelerated endogenous glucose
production is thought to be a key factor in the development of fasting
hyperglycemia in type 2 diabetes (244, 245). In patients with type 2
diabetes, metformin has been shown to inhibit endogenous glucose
production in most studies (246, 247, 248, 249, 250, 251, 252). This could be accounted for
largely by inhibition of gluconeogenesis (247), although an additional
inhibitory effect of metformin on glycogen breakdown is likely (247, 248). The observation in many studies that, in the basal postabsorptive
state, overall glucose disposal (metabolic plasma clearance rate of
glucose) did not change while endogenous glucose production decreased
(246, 247, 248, 251, 252, 253) suggests that the improvement in glycemic control
is largely attributable to the effect of metformin on glucose
production.
b. Peripheral glucose metabolism.
Many (246, 249, 251, 252, 254, 255, 256), but not all, studies (248, 250, 253, 257) using the
hyperinsulinemic-euglycemic clamp technique have shown a
metformin-induced increase in insulin-stimulated glucose disposal in
patients with type 2 diabetes. Since muscle represents a major site of
insulin-mediated glucose uptake (244, 258), metformin must, either
directly or via indirect mechanisms, have an insulin-like or
insulin-sensitizing effect on this tissue. In humans, the increase in
insulin-stimulated glucose disposal is mostly accounted for by
nonoxidative pathways (252, 255, 259). Nonoxidative glucose metabolism
includes storage as glycogen, conversion to lactate, and incorporation
into triglycerides. While no effect on lactate production is observed
(247, 248), implications on net triglyceride synthesis cannot be drawn.
Nevertheless, it appears reasonable to propose that in human muscle
glucose transport and, possibly as a consequence, glycogen synthesis
are the major targets of metformin action in the insulin-stimulated
state. However, in the basal state, metformin had no effect on glucose
clearance or whole-body glucose oxidation, although the proportion of
glucose turnover undergoing oxidation was increased (247). Moreover,
forearm glucose uptake in the postabsorptive state was not
significantly altered (247).
c. Metabolic effects independent of improved glycemia.
The
interpretation of the above experiments is limited by the fact that
treatment with metformin was always accompanied by improvement in
glycemic control and sometimes also by reduction of body weight. It
cannot be excluded, therefore, that the effects on endogenous glucose
production and glucose disposal, at least in part, were secondary to
reduced glucose toxicity (243) and/or weight loss (260) rather than
metformin per se. Only four studies have examined the
metabolic actions of metformin in the absence of any changes in
glycemic control or body weight.
In one study, 1 g of metformin was administered acutely to patients with type 2 diabetes; after 12 h no effect on insulin-stimulated glucose disposal was seen while the excessive endogenous glucose production in the basal state was significantly reduced (253). This suggests that in patients with type 2 diabetes, improvement in insulin-stimulated glucose disposal is predominantly due to alleviation of glucose toxicity while endogenous glucose production is immediately affected by metformin. In another study, lean, normal glucose-tolerant, insulin-resistant first-degree relatives of patients with type 2 diabetes acutely received 1 g of metformin and the opposite effect was observed (259). In subjects with IGT, 6-week metformin treatment improved basal (HOMA) but not insulin-stimulated glucose disposal or glucose oxidation (261). In this study both fasting glucose and insulin decreased significantly. In android obese subjects with IGT, increased insulin sensitivity (using an iv glucose tolerance) was observed after only 2 days of metformin treatment (1,700 mg/day) (262). In obese women with the polycystic ovary syndrome (PCOS) 6 months treatment with metformin also significantly improved insulin-stimulated glucose disposal (263, 264). In another study in obese women with PCOS, the decrease in serum insulin levels was associated with an increased ovulatory response to clomiphene (265). Glucose production was not assessed in the latter study. These apparent discrepancies could be explained by differences in the type of insulin resistance. In the highly selected group of lean, first-degree relatives and women with PCOS, mechanisms may contribute to insulin resistance that are different than those in garden-variety type 2 diabetes in which insulin resistance is predominantly the result of obesity and longstanding hyperglycemia. Moreover, the reduction in endogenous glucose production after metformin treatment may only be seen in subjects in whom it was increased to begin with, such as patients with type 2 diabetes. The latter is supported by observations showing that metformin alone does not cause hypoglycemia or lower blood glucose in nondiabetic subjects (266, 267). The effect of metformin on endogenous glucose production in nondiabetic humans has not yet been studied.
Additional evidence for improved insulin action comes from studies
combining insulin therapy and metformin. It was shown that requirements
of exogenous insulin are reduced (by 30%) by addition of metformin
in obese patients with type 2 diabetes (268, 269, 270) and in some patients
with type 1 diabetes in whom glycemic control was unaltered (271, 272, 273).
d. Other mechanisms of action.
It has been suggested that part
of the antihyperglycemic effect of metformin is due to decreased
release of FFA from adipose tissue and/or decreased lipid oxidation
(253, 274). However, reduced FFA levels after metformin treatment have
been shown in some (251, 257, 274) but not all studies (247, 248, 259).
Moreover, in vitro studies have shown that metformin does
not enhance the antilipolytic action of insulin on adipose tissue
(275). Only two studies have examined FFA turnover using isotope
techniques and found either no difference (247) or a 17% reduction
(255) after metformin treatment. In the latter study, the effect was
seen in the basal state but not in the insulin-stimulated state in
which FFA flux was largely suppressed. Thus, the metformin effect on
peripheral glucose uptake may, at least in part, be mediated by
suppression of FFA and lipid oxidation. In contrast, a causal
relationship with endogenous glucose production is unlikely, since
distinctly greater reductions in circulating FFA levels with acipimox
failed to lower glucose production (276, 277).
Evidence for other proposed mechanisms of metformin action is less convincing. Increased intestinal utilization of glucose has been suggested by animal studies (278, 279, 280). More recently, in vivo treatment with metformin increased gene expression of the energy-dependent sodium-glucose cotransporter (SGLT1) in rat intestine (281). However, such a mechanism has not been confirmed in humans (250).
e. Weight loss.
Unlike other pharmacological therapies for
type 2 diabetes (sulfonylureas, insulin), metformin treatment is not
associated with weight gain. Clinical studies have consistently shown
either a small but significant decrease in body weight (240, 251) or a
significantly smaller increase in body weight compared with other forms
of treatment (268). One study has shown that weight loss during
metformin treatment was largely accounted for by loss of adipose tissue
(247). This was explained by differential effects of metformin on
adipose tissue and muscle. While metformin improves insulin sensitivity
in muscle, it does not affect the antilipolytic action of insulin on
adipose tissue (282). The overall effect of metformin on body weight is
attributed to a reduction in caloric intake (268, 283) rather than an
increase in energy expenditure (247, 253, 284). Since reduction in body
weight per se reduces insulin resistance, this may also
represent a mechanism by which metformin improves insulin resistance.
To summarize, the partly divergent observations from the numerous
metabolic studies regarding metformin’s effect on muscle and liver
(Table 3, A and B) may
reflect different mechanisms of metformin action in the basal
vs. the insulin-stimulated state. In the basal,
postabsorptive state, the improvement of fasting hyperglycemia is
mostly due to a decrease of the accelerated endogenous glucose
production. This results from inhibition of both gluconeogenesis and
glycogen breakdown. Direct or indirect effects on regulatory enzymes
are likely to be involved. No data are available for suppression of
glucose production during experimental hyperinsulinemia. However, the
fact that reduction in basal glucose production occurs in the presence
of lower or unaltered insulin levels suggests that glucose production
in liver and kidney (285, 286) is more sensitive to the restrictive
action of insulin after treatment with metformin.
|
|
3. Clinical efficacy of metformin in patients with type 2 diabetes
mellitus.
a. Glycemic control.
The glucose-lowering effect of metformin,
monotherapy or in combination, has been extensively reviewed
(231, 232, 233). In a recent meta-analysis (230), all randomized, controlled
clinical trials comparing metformin with placebo (239, 240, 252, 289, 290, 291, 292, 293, 294) and sulfonylurea (239, 240, 295, 296, 297, 298, 299, 300, 301) were evaluated. The
weighted mean difference between metformin and placebo after treatment
(median treatment duration, 4.5 months) for fasting blood glucose was
-2.0 mM and for HbA1c -0.9%. Body weight was not
significantly changed after treatment. Sulfonylureas and metformin
lowered blood glucose (-2.0 and -1.8 mM, respectively)
and HbA1c (-1.1 and -1.3%, respectively) equally (median treatment
duration, 6 months). However, whereas after sulfonylurea treatment body
weight increased by 2.9 kg, there was a decrease of 1.2 kg after
metformin. In a retrospective study of 9,875 patients with type 2
diabetes mellitus who attended a large health maintenance organization,
metformin treatment improved the mean HbA1c by 1.41% over a 20-month
period (302). Among obese patients treated by intensive blood
glucose control within the UKPDS, metformin showed a significantly
greater effect than chlorpropamide, glibenclamide, or insulin for any
diabetes-related endpoint, all-cause mortality, and stroke (239). In
summary, metformin is as effective as sulfonylureas in improving
glycemic control but, especially in overweight/obese patients,
advantageous with respect to body weight, diabetes-related endpoints,
and frequency of hypoglycemia.
b. Lipid profile and cardiovascular system.
In addition to
improving glycemic control, metformin has been shown to reduce serum
lipid levels. Metformin treatment results in a moderate (10–20%)
reduction in circulating triglyceride levels, particularly in patients
with marked hypertriglyceridemia and hyperglycemia (247, 257, 303), but
also in nondiabetic subjects (304, 305). This has been attributed to a
reduction in hepatic very low density lipoprotein (VLDL) synthesis
(257, 292, 306). Small (5–10%) decreases in total circulating
cholesterol have also been reported (286, 289, 290, 291) that were
essentially attributed to reductions in low density lipoprotein (LDL)
levels (307, 308, 309) since high-density lipoprotein (HDL) cholesterol
levels were either increased (304) or unchanged (309).
In addition to the improvement of the lipid profile, metformin appears to have potentially beneficial hemostaseological effects. Fibrinolysis is increased (305, 307, 308) and the fibrinolysis inhibitor plasminogen-activator inhibitor 1 (PAI1) is decreased (292, 305, 310). Moreover, a decrease in platelet aggregability and density has been demonstrated (296, 311). These additional effects of metformin, which have been extensively reviewed elsewhere (231, 232), may explain the advantage of metformin over sulfonylurea or insulin treatment with respect to macrovascular endpoints shown in the UKPDS (239).
c. Combination therapies: metformin plus sulfonylureas and
metformin plus insulin.
Metformin is also used in combination with
other antihyperglycemic agents. Because of its unique mechanisms of
action, a synergistic effect on glycemic control has been observed in
combination with sulfonylureas (e.g., Refs. 240, 312, 313), troglitazone (Ref. 314 and see next chapter), and insulin
where a dose-sparing effect was consistently demonstrated (268, 269, 270, 314, 315, 316). Interestingly, in patients in whom sulfonylurea therapy has
failed to satisfactory glycemic control, the combination of bedtime
NPH-insulin with metformin was advantageous compared with other
combinations (316). In contrast to insulin alone, insulin plus
sulfonylurea, and sulfonylurea alone, when bedtime NPH-insulin was
combined with metformin, a decrease in HBA1c was achieved
without significant weight gain (315, 316).
4. Adverse effects. While mild gastrointestinal disturbances are the most common side effects, lactic acidosis, although rare, is the most serious side effect of metformin treatment (317). In 9,875 patients one case of probable lactic acidosis was observed in 20 treatment months (302). The incidence of lactic acidosis is 10 to 20 times lower than with phenformin. This is explained by the necessity to hydroxylate phenformin before renal excretion, a step that is genetically defective in 10% of whites (318, 319). Metformin, in contrast, is excreted unmetabolized. In addition, in contrast to phenformin (320), metformin neither increases peripheral lactate production nor decreases lactate oxidation (247, 248), making lactate accumulation unlikely. One study investigating individual cases of metformin-associated lactic acidosis showed that in these patients metformin should never have been started or should have been discontinued with the onset of acute illness (321). Thus, strict adherence to the exclusion criteria of metformin treatment (renal and hepatic disease, cardiac or respiratory insufficiency, severe infection, alcohol abuse, history of lactic acidosis, pregnancy, use of intravenous radiographic contrast; reviewed in Refs. 213, 216) should minimize the risk of metformin-induced lactic acidosis.
5. Guidelines for the clinical use of metformin. As recently reviewed (231) metformin or sulfonylurea therapy can be initiated when patients with NIDDM continue to have hyperglycemia despite diet and exercise. Metformin appears to be the drug of choice to start pharmacological treatment in insulin-resistant and overweight/obese diabetic subjects (239, 322). However, since the antihyperglycemic effects of metformin are similar in lean and obese subjects, it can also be recommended as first-line treatment in the absence of obesity. Addition of metformin to sulfonylureas in patients with secondary sulfonylurea failure appears reasonable in view of their synergistic mechanisms of action and has been shown to improve glycemic control. Furthermore, especially in overweight/obese patients, the addition of metformin to insulin is advantageous compared with insulin alone (507). Finally, metformin is not recommended for patients with type 1 diabetes, or in insulin-resistant states in the absence of overt type 2 diabetes. However, metformin is currently under investigation as an agent to prevent type 2 diabetes in subjects with IGT as one of the three arms (vs. diet and intensive life-style modification) of the Diabetes Prevention Program (322), but it is not yet approved for use in subjects with IGT.
C. Thiazolidinediones
1. Introduction. The thiazolidinediones are a new class of
hypoglycemic agents that were originally developed in the early 1980s
in Japan as antioxidants (323). Soon after the synthesis of the first
thiazolidinedione, ciglitazone, the blood glucose-lowering potential of
these compounds was observed in animals, with particularly pronounced
effects in animals with genetic insulin resistance such as the
KK, db/db, and ob/ob mice, and
fa/fa rats (324, 325, 326). The observation that glycemia
improved in the absence of increasing insulin and the lack of effect in
insulin-deficient animals (327) led to the conclusion that
thiazolidinediones improved insulin resistance and resulted in the
nickname "insulin sensitizers." However, due to an unacceptable
side effect profile, ciglitazone and, later, englitazone never
proceeded to human studies. Troglitazone became the first
thiazolidinedione available for clinical use and was released in 1997
in the United States and Japan followed by rosiglitazone and
pioglitazone (both marketed in 1999 in the United States). In Europe,
except for the United Kingdom, where it was available for a few months,
troglitazone has not been approved and, due to an untoward risk-benefit
ratio (hepatotoxic side effects), was withdrawn from the US market by
the Food and Drug Administration (FDA) in March 2000. Thus, at the
present time rosiglitazone and pioglitazone are the two members of the
thiazolidinedione class available for clinical use in some countries
including the United States, Japan, and Europe. Since the majority of
clinical data originate from studies using troglitazone, however, this
substance will be included in this review.
2. Mechanism of action. The cellular mechanism of action of
the thiazolidinediones is not precisely understood. However, the body
of evidence indicating that a subtype of the PPAR is the principal
receptor mediating the antidiabetic activity of the thiazolidinediones
is substantial and has recently been reviewed (328, 329). Expression
levels of the nuclear receptor PPAR are highest in adipocytes,
intestinal cells, and macrophages but very low in most other tissues
including muscle. PPAR activated by specific agonists, including
thiazolidinediones, heterodimerizes with the retinoid X receptor to
bind to specific DNA repeats, resulting in transcription of
thiazolidinedione-responsive genes. In various cell models
(preadipocytes, fibrobl
