Endocrine Reviews 18 (6): 774-800
Copyright © 1997 by The Endocrine Society
Insulin Resistance and the Polycystic Ovary Syndrome: Mechanism and Implications for Pathogenesis1
Andrea Dunaif
Pennsylvania State University College of Medicine, Hershey,
Pennsylvania 17033
 |
Abstract
|
|---|
- I. Introduction
- A. Background and historical perspective
- B. Definition of PCOS
- II. Insulin Action in PCOS
- A. Glucose tolerance
- B. Insulin action in vivo in PCOS
- C. Insulin secretion in PCOS
- D. Insulin clearance in PCOS
- E. Cellular and molecular mechanisms of insulin resistance
- F. Constraints of insulin action studies in PCOS
- G. PCOS as a unique NIDDM subphenotype
- III. Hypotheses Explaining the Association of Insulin Resistance and PCOS
- A. Causal association
- B. Possible genetic association of PCOS and insulin resistance
- IV. Clinical Implications of Insulin Resistance in PCOS
- A. Clinical diagnosis of insulin resistance
- B. Other metabolic disorders in PCOS
- C. Therapeutic considerations
- V. Summary
|
I. Introduction
|
|---|
A. Background and historical perspective
POLYCYSTIC ovary syndrome (PCOS) is an exceptionally common
disorder of premenopausal women characterized by hyperandrogenism and
chronic anovulation (1, 2). Its etiology remains unknown. Although
there have been no specific population-based studies, a 5–10%
prevalence of this disorder in women of reproductive age is probably a
reasonable conservative estimate. This is based as an upper limit on
studies of the prevalence of polycystic ovaries, which found that
20% of self-selected normal women had polycystic ovary morphology
on ovarian ultrasound (3). Many of these women had subtle endocrine
abnormalities (3). The lower estimate is based on the reported 3%
prevalence rate of secondary amenorrhea for 3 or more months (4) and
the fact that up to 75% of women with secondary amenorrhea will
fulfill diagnostic criteria for PCOS (5). PCOS women can also have less
profound disturbances in menstrual function (1, 3, 6).
Since the report by Burghen et al. (7) in 1980 that PCOS was
associated with hyperinsulinemia, it has become clear that the syndrome
has major metabolic as well as reproductive morbidities. The
recognition of this association has also instigated extensive
investigation of the relationship between insulin and gonadal function
(1, 8, 9, 10, 11). This review will summarize our current understanding of
insulin action in PCOS, address areas of controversy, and propose
several hypotheses for this association. Abnormalities of
steroidogenesis and gonadotropin release will not be discussed in
detail; these changes have been reviewed recently by Erhmann and
colleagues (12) and by Crowley (13), respectively.
The association between a disorder of carbohydrate metabolism and
hyperandrogenism was first described in 1921 by Achard and Thiers (14)
and was called "the diabetes of bearded women (diabete des femmes a
barbe)." The skin lesion, acanthosis nigricans, was reported to occur
frequently in women with hyperandrogenism and diabetes mellitus by
Kierland et al. (15) in 1947. Brown and Winkelmann (16)
noted in 1968 that it was insulin-resistant diabetes mellitus, and a
genetic basis was suggested by reports of affected sisters (17),
including a pair of identical twins who also had acromegaloid features
(18). Several additional syndromes with distinctive phenotypic
features, acanthosis nigricans, hyperandrogenism, and insulin-resistant
diabetes mellitus have been identified (Table 1
). These include the lipoatrophic (total
and partial) diabetes syndromes, leprechaunism (intrauterine growth
retardation, gonadal enlargement, elfin facies, and failure to thrive),
and Rabson-Mendenhall syndrome (unusual facies, pineal hypertrophy,
dental precocity, thickened nails, and ovarian enlargement) (8, 19, 20).
Attention was focused on the association of hyperandrogenism, insulin
resistance, and acanthosis nigricans in 1976 when Kahn and colleagues
(21) described a distinct disorder affecting adolescent girls, which
they designated the type A syndrome. These girls were virilized
(i.e., increased muscle bulk, clitoromegaly, temporal
balding, deepening of the voice) and had extreme insulin resistance
with diabetes mellitus as well as striking acanthosis nigricans. This
group identified a second distinct extreme insulin resistance syndrome
in postmenopausal women with acanthosis nigricans and features of
autoimmune disease, which they termed the type B syndrome and
determined that it was caused by endogenous antiinsulin receptor
antibodies (22, 23). Subsequent studies have identified insulin
receptor mutations as the cause of leprechaunism, Rabson-Mendenhall
Syndrome, and some cases of type A syndrome (19, 23).
In 1980 Burghen and colleagues (7) reported that women with the common
hyperandrogenic disorder, PCOS, had basal and glucose-stimulated
hyperinsulinemia compared with weight-matched control women, suggesting
the presence of insulin resistance. They noted significant positive
linear correlations between insulin and androgen levels and suggested
that this might have etiological significance. In the mid-1980s several
groups noted that acanthosis nigricans occurred frequently in obese
hyperandrogenic women (24, 25, 26, 27) (Fig. 1).
These women had hyperinsulinemia basally and during an oral glucose
tolerance test, compared with appropriately age- and weight-matched
control women. The presence of hyperinsulinemia in PCOS women,
independent of obesity, was confirmed by a number of groups worldwide
(28, 29, 30).
View larger version (91K):
[in this window]
[in a new window]
|
Figure 1. A woman with PCOS who has acanthosis nigricans, a
cutaneous marker of insulin resistance (panel A). She also has severe
hirsutism on her face and chest (panels B and C). [Reproduced from A.
Dunaif et al.: Obstet Gynecol
66:545–552, 1985 (25) with permission from The American College of
Obstetricians and Gynecologists.]
|
|
Our study (25) suggested that these women had typical PCOS, except for
increased ovarian stromal hyperthecosis, which is diagnosed by finding
islands of luteinized theca cells within the ovarian stroma (25). When
this is very extensive, it is called hyperthecosis and is associated
with more profound hyperandrogenism (31). Hughesdon (32) reported,
however, that upon careful examination of ovaries from PCOS women,
small islands of hyperthecosis were usually present. This morphological
change was more extensive in insulin-resistant PCOS women, suggesting
that hyperinsulinemia had an impact on ovarian morphology as well as on
function (25) (Fig. 2). This hypothesis
has been further supported by the finding, in a subsequent study (33),
of a positive correlation between hyperinsulinemia and ovarian stromal
hyperthecosis.
View larger version (212K):
[in this window]
[in a new window]
|
Figure 2. Section of a polycystic ovary with multiple
subscapular follicular cysts and stromal hypertrophy (left
panel). At higher power (x100) islands of luteinized theca
cells are visible in the stroma (right panel). This
morphological change is called stromal hyperthecosis and appears to be
directly correlated with circulating insulin levels. [Figure is used
with permission from A. Dunaif.]
|
|
B. Definition of PCOS
The current recommended diagnostic criteria for PCOS are
hyperandrogenism and ovulatory dysfunction with the exclusion of
specific disorders, such as nonclassic adrenal 21-hydroxylase
deficiency, hyperprolactinemia, or androgen-secreting neoplasms (1)
(Table 2). The polycystic ovary
morphology is consistent with, but not essential for, the diagnosis of
the syndrome (1, 3). Polycystic ovaries are defined on
ultrasound by the presence of eight or more subcapsular follicular
cysts 
10 mm and increased ovarian stroma (2, 3). These changes,
however, can be present in women who are entirely endocrinologically
normal (2, 3). Thus, the ovarian morphological change must be
distinguished from the endocrine syndrome of
hyperandrogenism and anovulation.
Gonadotropin-secretory changes, with a characteristic increase in LH
relative to FSH release, have long been appreciated in PCOS (34, 35).
Frequent (e.g., every 10 min), prolonged (12–24 h) serial
blood sampling studies have revealed that there is a significant
increase in the frequency and the amplitude of LH release with normal
FSH release in PCOS (36, 37). The increased LH pulse frequency reflects
an increase in GnRH release and suggests the presence of a
hypothalamic defect in PCOS (13, 37). Other causes of hyperandrogenism,
however, can result in similar gonadotropin-secretory changes, such as
androgen-secreting neoplasms (38) or adrenal hyperandrogenism resulting
from nonclassic 21-hydroxylase deficiency (39). Ovulatory women with
the polycystic ovary morphology can have increased LH/FSH ratios (2).
Because of the pulsatile nature of gonadotropin release, a single blood
sample can fail to detect an increased LH/FSH ratio (40). This, as well
as its lack of specificity, has led to the recommendation that LH/FSH
ratios not be included in the diagnostic criteria for PCOS (1).
Other nomenclature has been proposed for the syndrome,
e.g., chronic hyperandrogenic anovulation (CHA) (1). Many
hyperandrogenic anovulatory women have significantly increased ovarian
steroidogenic responses to stimulation with GnRH analogs that
Rosenfield and colleagues (41) have termed functional ovarian
hyperandrogenism (FOH). They have proposed this as an alternative name
for PCOS (12). The majority of women who have hyperandrogenemia and
chronic anovulation will have polycystic ovary (PCO) on ultrasound and
will have responses to GnRH analogs consistent with FOH (1, 2, 12)
(Fig. 3). Thus, the terms PCOS, FOH, and
CHA define similar groups of women (Fig. 3).
View larger version (70K):
[in this window]
[in a new window]
|
Figure 3. The majority of women with CHA will also have
polycystic ovary morphology (PCO) and responses to GnRH analogs
consistent with FOH. [Figure is used with permission from A.
Dunaif.]
|
|
PCOS often has a menarchal age of onset characterized by a failure to
establish a regular pattern of menses (42). Hirsutism may develop
peripubertally or during adolescence (42) or it may be absent until the
third decade of life (43). Seborrhea, acne, and alopecia are other
common clinical signs of hyperandrogenism (44, 45). Some women never
develop signs of androgen excess because of genetic differences in
target tissue number and/or sensitivity to androgens (46). The clinical
consequence of chronic anovulation is some form of menstrual
irregularity ranging from oligomenorrhea (menses every 6 weeks to 6
months), amenorrhea, or dysfunctional uterine bleeding (2, 5, 6).
Infertility may be the presenting symptom of the anovulation. Depending
on the population studied, 16–80% of PCOS women are obese (47, 48, 49).
Mild to moderate acanthosis nigricans is commonly present in obese PCOS
women (25, 26, 27, 49, 50). A rapid progression of androgenic symptoms
and/or true virilization (increased muscle bulk, clitoromegaly,
temporal balding, and/or deepening of the voice) are rare in PCOS (2, 6, 42). PCOS women can occasionally have acromegaloid features (44).
It is important to recognize that there is an inherent bias of
ascertainment in studies of PCOS that constrains the assessment of the
frequency of associated clinical and biochemical findings. Obviously,
all women will have polycystic ovaries when this feature is an
essential diagnostic criterion. Studies that use an increased LH/FSH
ratio as a selection criterion will be biased toward finding increased
pulsatile LH release when gonadotropin secretion is examined. The
appropriate study would be a population-based one in which clinical and
biochemical features were systematically examined in a defined
population of women. Until such a study is performed, the prevalence of
PCOS and frequency of associated findings will remain subject to
debate.
|
II. Insulin Action in PCOS
|
|---|
A. Glucose tolerance
Insulin resistance is an important defect in the pathogenesis of
noninsulin-dependent diabetes mellitus (NIDDM) (51). Despite the fact
that hyperinsulinemia, reflecting some degree of peripheral insulin
resistance, was well recognized in PCOS by the mid-1980s (Fig. 4), glucose tolerance was not
systematically investigated until our study in 1987 (49). We found that
obese PCOS women had significantly increased glucose levels during an
oral glucose tolerance test compared with age- and weight-matched
ovulatory hyperandrogenic (i.e., elevated plasma androgen
levels) and control women (Fig. 4). Twenty percent of the obese PCOS
women had impaired glucose tolerance or frank NIDDM by National
Diabetes Data Group Criteria (49, 52) (Fig. 4). The women studied
ranged in age from 18–36 yr with a mean age of 27 yr for the obese
PCOS women. There were no significant differences, however, in glucose
levels during the oral glucose tolerance test in the nonobese PCOS
women compared with age- and weight-matched control women (Fig. 4).
View larger version (21K):
[in this window]
[in a new window]
|
Figure 4. Insulin (panels A and C) and glucose (panels B and
D) responses basally and after a 40 g/m2 oral glucose load
in obese and lean PCOS women, ovulatory hyperandrogenic women (HA)
women, and age- and weight-matched ovulatory control women. Insulin
responses are significantly increased only in PCOS women, suggesting
that hyperinsulinemia is a unique feature of PCOS and not
hyperandrogenic states in general (panels A and B). Glucose responses
are significantly increased only in obese PCOS women (C), and 20%
of obese PCOS women have impaired glucose tolerance or NIDDM using
National Diabetes Data Group Criteria (52). [Derived from Ref. 49.]
|
|
A subsequent study in postmenopausal women with a history of PCOS found
a significantly increased prevalence of NIDDM as well as of
hypertension (see below) (53). We have continued to find prevalence
rates of glucose intolerance as high as 40% in obese PCOS women
when the less stringent World Health Organization (WHO) criteria are
used (49, 52, 54, 55, 56, 57). The majority of affected women are in their
third and fourth decade of life, but we and others (58) have
encountered PCOS adolescents with impaired glucose tolerance or NIDDM.
These prevalence rates of 20–40% are substantially above prevalence
rates for glucose intolerance reported in population-based studies in
women of this age (5.3% by National Diabetes Data Group criteria and
10.3% by WHO criteria in women aged 20–44 yr) (59). We have found
that the prevalence of glucose intolerance is significantly higher in
obese PCOS women (30%) than in concurrently studied age-,
ethnicity-, and weight-matched ovulatory control women (10%) (48).
In contrast, we have found that nonobese PCOS women have impaired
glucose tolerance only occasionally, consistent with the synergistic
negative effect of obesity and PCOS on glucose tolerance (54, 55).
Finally, based on the prevalence of glucose intolerance in women (59),
the prevalence of glucose intolerance in PCOS (49), and on a
conservative estimate of the prevalence of PCOS (5%), it can be
extrapolated that PCOS-related insulin resistance contributes to
approximately 10% of cases of glucose intolerance in premenopausal
women. The study in postmenopausal women with a history of PCOS found a
15% prevalence of NIDDM (53), consistent with our extrapolated
prevalence estimates. It is thus clear that PCOS is a major risk factor
for NIDDM in women, regardless of age.
B. Insulin action in vivo in PCOS
Although insulin has a number of actions, in addition to those
regulating glucose metabolism, such as inhibition of lipolysis and
stimulation of amino acid transport (51), the effects of insulin on
glucose metabolism are usually examined in studies of insulin
resistance (60). This can be studied quantitatively in humans with the
euglycemic glucose clamp technique: a desired dose of insulin is
administered and euglycemia is maintained by a simultaneous variable
glucose infusion whose rate is adjusted based on frequent arterialized
blood glucose determinations and a negative feedback principle
(60, 61, 62). At steady state, the amount of glucose that is infused equals
the amount of glucose taken up by the peripheral tissues and can be
used as a measure of peripheral sensitivity to insulin, known as
insulin-mediated glucose disposal (IMGD) or M (61, 62). The suppression
of hepatic glucose production by insulin can be assessed by the use of
a simultaneous infusion of isotopically labeled glucose.
Insulin-mediated glucose disposal occurs only in muscle (skeletal
and cardiac) and in fat; muscle accounts for about 85% of this (60).
Euglycemic glucose clamp studies have demonstrated significant and
substantial decreases in insulin-mediated glucose disposal in PCOS (54, 55) (Fig. 5). This decrease (35–40%)
is of a similar magnitude to that seen in NIDDM (Fig. 5). Obesity (fat
mass per se), body fat location (upper vs. lower
body, e.g., waist to hip girth ratio), and muscle mass all
have important independent effects on insulin sensitivity (63, 64, 65, 66).
Alterations in any of these parameters could potentially contribute to
insulin resistance in PCOS. PCOS women have an increased prevalence of
obesity (6, 47), and women with upper, as opposed to lower body,
obesity have an increased frequency of hyperandrogenism (66). Since
muscle is the major site of insulin-mediated glucose use (60) and
androgens can increase muscle mass (67), potential
androgen-mediated changes in lean body (primarily muscle) mass must
also be controlled for in PCOS (54, 55). Studies in which body
composition, assessed by the most precise available method (hydrostatic
weighing), has been matched to normal control women, and in which lean
PCOS women, who had body composition and waist to hip girth ratios
similar to controls, were studied, have confirmed that PCOS women are
insulin resistant, independent of those potentially confounding
parameters (1, 55, 68). The impact of hyperandrogenism on insulin
sensitivity is discussed below, but studies in cultured cells have
confirmed the impression from these in vivo studies that an
intrinsic defect in insulin action is present in PCOS (69).
Basal hepatic glucose production and the ED50 value of
insulin for suppression of hepatic glucose production are significantly
increased only in obese PCOS women (54, 55) (Fig. 6). This synergistic negative effect of
obesity and PCOS on hepatic glucose production is an important factor
in the pathogenesis of glucose intolerance (49, 54, 55, 70). This is
analogous to NIDDM in general where defects in insulin action,
presumably genetic, synergize with environmentally induced insulin
resistance, primarily obesity-related, to produce glucose intolerance
(51, 60). Sequential multiple-insulin-dose euglycemic clamp studies
have indicated that the ED50 insulin for glucose uptake is
significantly increased, and that maximal rates of glucose disposal are
significantly decreased in lean and in obese PCOS women (55) (Fig. 6).
It appears, however, that body fat has a more pronounced negative
effect on insulin sensitivity in women with PCOS (68, 71).
C. Insulin secretion in PCOS
In the presence of peripheral insulin resistance, pancreatic
ß-cell insulin secretion increases in a compensatory fashion. NIDDM
develops when the compensatory increase in insulin levels is no longer
sufficient to maintain euglycemia (72, 73). It is essential, therefore,
to examine ß-cell function in the context of peripheral insulin
sensitivity. Under normal circumstances, this relationship is constant
(72, 74) (Fig. 7). ß-Cell dysfunction
is felt to be present for values falling below this hyperbolic curve
(73, 74). This relationship can be quantitated as the product of
insulin sensitivity and first-phase insulin release known as the
disposition index (72).
View larger version (22K):
[in this window]
[in a new window]
|
Figure 7. The relationship between insulin sensitivity (SI)
determined by frequently sampled intravenous glucose tolerance test and
first-phase insulin secretion to an intravenous glucose load (AIRg).
The majority of PCOS women fall below the normal curve determined in
concurrently studied age- and weight-matched control women as well as
normative data in the literature. [Derived from Ref. 57.]
|
|
Fasting hyperinsulinemia is present in obese PCOS women and this is, in
part, secondary to increased basal insulin secretion rates (Fig. 4
and
Ref.75). Insulin responses to an oral glucose load are increased in
lean and obese PCOS women (Fig. 4), but acute insulin responses to an
intravenous glucose load (AIRg), first-phase insulin secretion, are
similar to weight-matched control women (49, 57). When the relationship
between insulin secretion and sensitivity is examined, lean and obese
PCOS women fall below the relationship in weight-matched control women,
and the disposition index is significantly decreased by PCOS as well as
by obesity (57) (Fig. 7). Further evidence for ß-cell dysfunction in
PCOS is provided by the elegant studies of Erhmann et al.
(76), who have demonstrated defects in ß-cell entrainment to an
oscillatory glucose infusion and decreased meal-related insulin
secretory responses (75). These defects are much more pronounced in
PCOS women who have a first-degree relative with NIDDM, suggesting that
such women may be at particularly high risk to develop glucose
intolerance (76). There are reports of increased insulin secretion in
PCOS, but these studies have not examined insulin secretion in the
context of insulin sensitivity and/or have included women in whom the
diagnosis was made on the basis of ovarian morphological changes rather
than endocrine criteria (71, 77). In summary, the most compelling
evidence suggests that ß-cell dysfunction, in addition to insulin
resistance, is a feature of PCOS. The ability to diagnose PCOS at the
time of puberty will make possible prospective longitudinal studies of
the ontogeny of these defects.
D. Insulin clearance in PCOS
Hyperinsulinemia can result from decreases in insulin
clearance as well as from increased insulin secretion. Indeed,
decreased insulin clearance is usually present in insulin-resistant
states since insulin clearance is receptor-mediated, and acquired
decreases in receptor number and/or function are often present in
insulin resistance secondary to hyperinsulinemia and/or hyperglycemia
(78, 79). Thus, PCOS would be expected to be associated with decreases
in insulin clearance; however, relatively few studies have examined
this question. Direct measurement of posthepatic insulin clearance
during euglycemic clamp studies has not been abnormal in PCOS (54, 56).
Circulating insulin to C-peptide molar ratios are increased in PCOS,
suggesting decreased hepatic extraction of insulin, but such ratios
also reflect insulin secretion (28, 80). Direct measurement of hepatic
insulin clearance in non-PCOS hyperandrogenic women has found it to be
decreased (81). The one study of this question in PCOS found decreased
hepatic insulin extraction by model analysis of C-peptide levels (75).
Therefore, in PCOS, hyperinsulinemia is probably the result of a
combination of increased basal insulin secretion and decreased hepatic
insulin clearance.
E. Cellular and molecular mechanisms of insulin resistance
1. Molecular mechanisms of insulin action (Figs. 8 and 9). Insulin acts on cells by binding
to its cell surface receptor (51, 82, 83). The insulin receptor
is a heterotetramer made up of two 
,ß- dimers linked by disulfide
bonds (84) (Fig. 8). Each ,ß-dimer is the product of one gene (85, 86). The -subunit is extracellular and contains the ligand-binding
domain whereas the ß-subunit spans the membrane, and the cytoplasmic
portion contains intrinsic protein tyrosine kinase activity, which is
activated further by ligand-mediated autophosphorylation on specific
tyrosine residues (87) (Fig. 8). The insulin receptor belongs to a
family of protein tyrosine kinase receptors that includes the
insulin-like growth factor-I (IGF-I) receptor, with which it shares
substantial sequence and structural homology, as well as the epidermal
growth factor (EGF), fibroblast growth factor, platelet-derived growth
factor, and colony-stimulating factor-1 receptors (88). A number of
oncogene products are also protein tyrosine kinases (85, 89).
View larger version (34K):
[in this window]
[in a new window]
|
Figure 9. The tyrosine-phosphorylated insulin receptor
phosphorylates intracellular substrates, such as insulin receptor
substrate (IRS)-1 and IRS-2, initiating signal transduction and the
plieotropic actions of insulin. The activation of PI3-K (PI3-kinase) by
tyrosine-phosphorylated IRS-1 appears to be the pathway for
insulin-mediated glucose transport. The Ras-MAP kinase pathway appears
to regulate cell growth and glycogen synthesis. [Adapted with
permission from C. R. Kahn: Diabetes 43:1066–1084,
1994 (51).]
|
|
Ligand binding induces, probably via conformational changes,
autophosphorylation of the insulin receptor on specific tyrosine
residues and further activation of its intrinsic kinase activity (Fig. 8) (90, 91, 92). The activated insulin receptor then tyrosine
phosphorylates intracellular substrates to initiate signal transduction
(Fig. 9) (82). Over the last few years a number of these substrates
have been characterized. The first was insulin receptor substrate-1
(IRS-1), which serves as a docking molecule for signaling and adaptor
molecules (93, 94). The tyrosine-phosphorylated insulin receptor
tyrosine phosphorylates IRS-1 on specific motifs, and these
phosphorylated sites then bind signaling molecules, such as the SH2
domain of phosphatidylinositol 3-kinase (PI3-K), or the adaptor
molecule, Nck (51, 82, 94). This leads to activation of downstream
signaling pathways, such as that leading to insulin-mediated glucose
transport, which appears to be modulated through the PI3-K signal
cascade (82). More recently, insulin receptor substrate-2 (IRS-2),
another substrate for the insulin receptor, has been identified (95, 96). Shc (an adaptor molecule) can also bind directly to the insulin
receptor initiating signal transduction (82, 97).
Insulin has numerous target tissue actions, such as stimulation of
glucose uptake, gene regulation, DNA synthesis, and amino acid uptake
(51, 82). The mechanisms of insulin receptor signal specificity are
currently a subject of intense investigation. It now appears that the
Ras-Raf-MEK pathway is involved in the regulation of cell growth and
metabolism whereas the PI3-K pathway is involved in glucose uptake
(98, 99, 100, 101). The mechanisms by which the insulin signal is terminated
remain incompletely understood. Receptor-mediated endocytosis and
recycling are well known to occur and may be important to signal
termination (83, 102). Serine phosphorylation has been shown to
terminate signaling by the EGF receptor (103, 104), another tyrosine
kinase growth factor receptor, and it can be shown under a variety of
experimental conditions that insulin receptor serine phosphorylation
decreases its tyrosine kinase activity (105, 106, 107, 108). It has been
postulated that protein kinase C (PKC)-mediated serine phosphorylation
of the insulin receptor is important in the pathogenesis of
hyperglycemia-induced insulin resistance (102, 109). Recent evidence
suggests that tumor necrosis factor- (TNF-)-mediated serine
phosphorylation of IRS-1 inhibits insulin receptor signaling and is the
mechanism of TNF--induced insulin resistance (110). Studies
addressing this important question have been constrained by a lack of
sensitive anti-phosphoserine antibodies. Identification of
phosphoserine residues usually requires painstaking phosphoamino acid
analysis of 32P-labeled receptors (111). The use of
fluorophore labeling of phosphoserine promises to provide a sensitive
methodology for examining in vivo serine phosphorylation
events (112).
In summary, insulin action is mediated through a ligand-activated
tyrosine kinase receptor, similar to a number of other growth factors.
A variety of phosphorylation-dephosphorylation signaling cascades are
then activated, leading to the pleiotropic actions of insulin. The
mechanisms of signal specificity and termination require further
investigation.
2. Molecular insulin action defects in PCOS. Studies in
adipocytes, a classic insulin target tissue, have failed to confirm
earlier reports in blood cells of decreases in insulin receptor number
and/or receptor affinity in PCOS (25, 26, 27, 113) when appropriately
weight-matched controls have been included. The one adipocyte study
reporting a decrease in insulin receptor number used a control group
consisting primarily of lean individuals (114). Studies of insulin
action in isolated PCOS adipocytes have revealed marked decreases in
insulin sensitivity together with less striking, but significant,
decreases in maximal rates of insulin-stimulated glucose transport (55, 115) (Fig. 10). There is evidence for
decreases in adipocyte levels of adenosine in PCOS (116), but whether
this is a primary defect or secondary to hyperinsulinemia is unclear.
The decrease in maximal rates of adipocyte glucose uptake is secondary
to a significant decrease in the abundance of GLUT4 glucose
transporters (117). Similar defects are present in NIDDM and in obesity
but are ameliorated by control of hyperglycemia and hyperinsulinemia as
well as by weight reduction, suggesting acquired rather than intrinsic
defects (65, 118, 119, 120). In contrast, in PCOS such defects can occur in
the absence of obesity, glucose intolerance, or changes in waist to hip
girth ratios (55, 117). Moreover, these abnormalities are not
significantly correlated with sex hormone levels, suggesting that
abnormalities of insulin action in PCOS may be intrinsic (55, 117).
To further evaluate the postbinding defect in insulin action in PCOS,
we examined insulin receptor function in receptors isolated from
cultured skin fibroblasts. Because fibroblasts are removed from the
in vivo environment for several generations, they provide a
constant source of insulin receptors that are not influenced by the
hormonal imbalance of PCOS. Consistent with our earlier results from
the adipocyte studies, fibroblasts from PCOS women showed no change in
insulin binding or receptor affinity (69). However, in approximately
50% of PCOS fibroblasts (PCOS-ser), we observed decreased insulin
receptor autophosphorylation (69). This was secondary to markedly
increased basal autophosphorylation with minimal further
insulin-stimulated autophosphorylation (Fig. 11). Phosphoamino acid analysis
revealed decreased insulin-dependent receptor tyrosine phosphorylation
and increased insulin-independent receptor serine phosphorylation (69)
(Fig. 11). The ability of the PCOS-ser insulin receptors to
phosphorylate an artificial substrate was also significantly reduced
(Fig. 12).
View larger version (26K):
[in this window]
[in a new window]
|
Figure 11. Representative autoradiograms of
autophosphorylated skin fibroblast insulin receptor ß-subunits
(top) and phosphoamino acid analysis
(bottom) ± 1 µM insulin from a normal
(control), a PCOS woman with normal insulin-stimulated tyrosine
phosphorylation (PCOS-nl) and a PCOS woman with high basal
autophosphorylation on serine residues (PCOS-ser); S-serine,
Y-tyrosine. Basal autophosphorylation is increased and there is minimal
further insulin-stimulated phosphorylation in the PCOS-ser
ß-subunits. The high basal phosphorylation represents phosphoserine,
and phosphotyrosine content does not increase in response to insulin in
the PCOS-ser ß-subunits. [Reproduced from A. Dunaif et
al.: J Clin Invest 96:801–810, 1995 (69)
by copyright permission of The American Society for Clinical
Investigation.]
|
|
Serine phosphorylation of the insulin receptor has been shown in
cell-free systems and in vivo to inhibit the receptor’s
tyrosine kinase activity, analogous to our findings in the PCOS-ser
insulin receptors (69, 105, 106, 107, 108). Thus, this defect in the early steps
of the insulin-signaling pathway may cause the insulin resistance in
PCOS-ser women. Increased insulin-independent serine phosphorylation in
PCOS-ser insulin receptors appears to be a unique disorder of insulin
action since other insulin-resistant states, such as obesity, NIDDM,
type A syndrome, and leprechaunism, do not exhibit this abnormality (1, 51, 65, 69) (Table 1). The PCOS-ser phosphorylation abnormality appears
to be physiologically relevant because it is present in insulin
receptors partially purified from skeletal muscle, a classic insulin
target tissue, and because the same pattern of abnormal phosphorylation
occurs in insulin receptors phosphorylated in intact cells (69).
Fibroblasts from approximately 50% of PCOS women (PCOS-nl) have no
detectable abnormality in insulin receptor phosphorylation (69) (Figs. 11 and 12). Although these women demonstrate the same PCOS phenotype
and the same degree of insulin resistance as the PCOS-ser women with
abnormal phosphorylation, insulin receptor phosphorylation in
fibroblasts and skeletal muscle from these women is similar to that of
control women (69). This observation suggests that a defect downstream
of insulin receptor signaling, such as phosphorylation of IRS-1 or
activation of PI3-K, is responsible for insulin resistance in PCOS-nl
women (51, 69, 102). Indeed, our recent human studies demonstrate a
significant decrease in muscle PI3-K activation during insulin infusion
in PCOS women (121), consistent with a physiologically relevant defect
in the early steps of insulin receptor signaling.
We found no insulin receptor mutations in two PCOS-ser women by direct
sequencing of genomic DNA (120), and sequence analysis of the tyrosine
kinase domain in the ß-subunit of an additional eight PCOS-ser
women also revealed no mutations (69). This finding has recently been
confirmed by other investigators (122). Immunoprecipitation and mixing
experiments suggest that a factor extrinsic to the insulin receptor is
responsible for the excessive serine phosphorylation (69). PCOS-ser
insulin receptors autophosphorylate normally, if they are first
immunoprecipitated from wheat-germ agglutinin (WGA) lectin eluates.
Furthermore, mixing control human insulin receptors and WGA eluates
from PCOS-ser fibroblasts results in increased insulin-independent
serine phosphorylation and decreased insulin-stimulated tyrosine
phosphorylation of the normal receptors (69) (Fig. 13). Both experiments suggest that a
factor present in WGA eluates is responsible for the abnormal
phosphorylation.
View larger version (34K):
[in this window]
[in a new window]
|
Figure 13. Phosphoamino acid analysis of immunopurified
human insulin receptors (hIR) ß-subunits basally and mixed with
WGA-Sepharose eluates from control or PCOS-ser fibroblasts. hIRs were
immunopurified from WGA-Sepharose eluates, mixed in a ratio of 10 fmol
hIR:1 fmol PCOS-ser or control lectin eluate insulin-binding activity,
and autophosphorylation ± 1 µM insulin was
examined. Phosphoamino acid analysis revealed a striking increase in
phosphoserine content and a marked decrease in insulin-stimulated
phosphotyrosine content after mixing hIR with PCOS-ser lectin eluates
as compared with mixing hIR with control lectin eluates or in the
absence of mixing. [Reproduced from A. Dunaif et al.:
J Clin Invest 96: 801–810, 1995 (69) by copyright
permission of The American Society for Clinical Investigation.]
|
|
The serine/threonine kinase, PKC, is a candidate for the putative
serine phosphorylation factor (108). However, evidence against this
possibility includes the observation that no phosphothreonine is
detected in the PCOS-ser insulin receptors, and PKC has been shown to
phosphorylate threonine 1336 of the insulin receptor (123).
Furthermore, the IGF-I receptor, which is a known substrate of PKC
under certain conditions, phosphorylates normally in PCOS-ser women
(69, 124). Finally, preliminary Western blot analyses showed no
significant differences in the abundance of PKC isoforms in PCOS-ser
fibroblasts compared with controls (A. Dunaif, unpublished
observations).
Other serine/threonine kinases that might cause the increased serine
phosphorylation of PCOS-ser insulin receptors include a casein kinase
I-like enzyme and cAMP-dependent protein kinase (125, 126).
However, the casein kinase I-like enzyme has been shown to
phosphorylate insulin-stimulated insulin receptors twice as well as
unstimulated insulin receptors (125). This phosphorylation pattern
differs from what we observe with PCOS-ser insulin receptors, namely
excessive serine phosphorylation in the absence of insulin.
cAMP-dependent protein kinase is a candidate because increases in cAMP
cause serine phosphorylation of insulin receptors in cultured
lymphocytes (127). However, insulin receptor phosphorylation by
cAMP-dependent protein kinase is probably indirect because the human
insulin receptor ß-subunit does not contain the amino acid sequences
classically recognized by this kinase (128).
Alternatively, a novel serine/threonine kinase or an inhibitor of a
serine/threonine phosphatase may be responsible for the abnormal
phosphorylation of PCOS-ser insulin receptors (69, 129). Because it is
present in WGA eluates, the PCOS-ser factor is either a membrane
glycoprotein or a protein associated with a glycoprotein. In some
respects, our putative serine phosphorylation factor is similar to a
recently identified inhibitor of insulin receptor tyrosine kinase, the
membrane glycoprotein PC-1 (130) (Fig. 14). Both factors are extrinsic to the
insulin receptor, both are present in WGA eluates from human skin
fibroblasts, and both appear to inhibit insulin receptor tyrosine
kinase activity. This represents an important new mechanism for human
insulin resistance related to factors that modulate the tyrosine kinase
activity of the insulin receptor (51) (Fig. 14). The major difference
between the two factors is that PC-1 is not associated with increased
insulin-independent serine phosphorylation characteristic of the
PCOS-ser insulin receptors (69, 130, 131). Recent studies suggest that
TNF- produces insulin resistance by a related mechanism: serine
phosphorylation of IRS-1, which then inhibits insulin receptor tyrosine
kinase activity (Fig. 7). Isolation and characterization of the factor
in PCOS-ser fibroblasts are now in progress, as is the mapping of
phosphorylated serine residues in PCOS-ser insulin receptors.
Although fibroblasts are not classic insulin target cells, defects
identified in insulin receptor number and/or kinase activity in them
have reflected insulin receptor mutations (19). Thus, the presence of
the putative serine phosphorylation factor in cultured cells of
PCOS-ser women suggests that the abnormal insulin receptor
phosphorylation is genetically programmed. In addition, we have found
that some first degree relatives of PCOS women are insulin resistant,
including brothers, consistent with a genetic defect (132). Recent twin
(133) and family studies (134) have also suggested that insulin
resistance is a genetic defect in PCOS. Our putative serine
phosphorylation factor is a candidate gene for a mutation producing the
insulin resistance associated with PCOS (see below).
F. Constraints of insulin action studies in PCOS
There is general consensus in the literature that obese PCOS women
are insulin resistant. Controversy remains as to the pathogenesis of
the insulin resistance, and there are studies that suggest that obesity
per se or increased central adiposity are responsible for
the associated defects in insulin action (135, 136). Many of the
conflicting studies can be explained by differing diagnostic criteria
for PCOS and by the inclusion of both lean and obese women in the
experimental sample. Our studies (49) and those in the United Kingdom
(137, 138) strongly suggest that anovulation is associated with insulin
resistance. We found insulin resistance only in women with
hyperandrogenism and anovulation (Fig. 4). Studies using ovarian
morphology to ascertain women have found that only anovulatory women
with PCO morphology are insulin resistant (137, 138). Women with
regular ovulatory menses and hyperandrogenism [elevated plasma
androgen levels (49)] (Fig. 4) or with PCO detected by ovarian
ultrasound (137, 138) are not insulin resistant. Therefore, studies
that have defined PCOS by PCO morphology without further assessment of
ovulation could have included women who were not insulin resistant.
Similarly, studies that have included ovulatory hyperandrogenic women
will bias the sample with insulin-sensitive subjects.
One reason for the general acceptance of the diagnostic criteria for
PCOS of hyperandrogenism and anovulation (1) (Table 2, see above) is
that they define the insulin-resistant subset. Even with subjects so
identified, not all are insulin resistant, despite using the relatively
lenient criterion of 1 SD below the control mean value for
insulin action. Moreover, the occasional PCOS woman can have insulin
sensitivity more than 2 SDs (95% confidence interval)
above the control mean (117). There is clearly heterogeneity in this
feature of the syndrome. Obesity is another important factor, and it
appears that it has a more pronounced effect on insulin action in PCOS
than in control women (71). Ideally, lean and obese PCOS women should
be studied separately (30, 49, 54, 55, 68). If groups are pooled, PCOS
women should be matched to controls so that the spectrum of body
weights are equally represented. This is often not the case so that,
although mean body mass may be similar, the PCOS group often contains
more obese individuals, thereby skewing the results (114). Moreover,
there are very few studies in the literature in which lean PCOS woman
have been separately studied (30, 54, 55, 68, 135). There are also
major ethnic variations in insulin sensitivity, and this is another
less well appreciated potential confounding factor (56). Recent studies
from Denmark suggest that adiposity accounts for insulin resistance
in their PCOS population in contrast to our US population (135, 136).
We have consistently found significant decreases in insulin-mediated
glucose disposal in both lean and obese PCOS women (54, 55, 56). Similarly,
our group (57) as well as Yen’s group (68) have found significant
decreases in insulin sensitivity (SI) determined by modified frequently
sampled intravenous glucose tolerance test with minimal model analysis
in such PCOS women (57). Insulin resistance has been found in PCOS
women of many racial and ethnic groups including Japanese, Caribbean
and Mexican Hispanics, non-Hispanic Whites, and African Americans (55, 56, 139, 140).
G. PCOS as a unique NIDDM subphenotype (Table 3)
Our studies in premenopausal women, extrapolated data based on
prevalence estimates of PCOS and glucose intolerance, and studies in
postmenopausal women with a history of PCOS all suggest that
PCOS-related insulin resistance confers a significantly increased risk
for NIDDM (see above). Familial clustering of affected individuals as
well as studies in monozygotic twins indicate that NIDDM has an
important genetic component (51, 102, 141, 142, 143, 144). Insulin resistance is
a major inherited abnormality, but studies in which insulin secretion
has been examined in the context of insulin sensitivity demonstrate
that ß-cell dysfunction may also be an important contributing factor
to the ultimate development of the NIDDM phenotype (51, 145, 146).
There is clearly genetic heterogeneity with insulin resistance being
absent in some affected individuals (146, 147).
The underlying genetic defects have been identified in fewer than 5%
of NIDDM individuals and consist of mutations in genes such as the
insulin receptor gene, mitochondrial DNA, or the glucokinase gene
(Table 3) (19, 51, 102, 144, 148, 149). Defects in a number of
candidate genes, such as GLUT4, GLUT2, and hexokinase, have been
excluded (102, 150). The major cause of insulin resistance in typical
NIDDM is reduced insulin-stimulated muscle glycogen synthesis. Defects
found in NIDDM in insulin receptor number and/or phosphorylation or
glucose transport, however, are reversible with the control of
hyperglycemia (51, 65, 102, 151), elevated free fatty acid levels
(152), and/or hyperinsulinemia (119). Only one study has shown an
intrinsic abnormality in NIDDM-cultured cells (153): decreased
insulin-stimulated glycogen synthesis. Studies in NIDDM first-degree
relatives, who are normoglycemic but insulin resistant, suggest that
there is an inherited decrease in both insulin-stimulated muscle
glucose transport/phosphorylation and glycogen synthase activity that
results in the reduced glycogen synthesis (154, 155, 156). In contrast, in
PCOS, intrinsic abnormalities in the early steps of insulin receptor
signaling are present, making this the first common NIDDM subphenotype
in which such defects have been identified (69, 102, 151). Moreover,
the defective pattern of insulin receptor phosphorylation is unique,
suggesting it should be possible to distinguish PCOS-related insulin
resistance from that related to other NIDDM genotypes. This should make
it possible to assign affected status accurately for linkage studies of
the genetics of PCOS-related insulin resistance (157).
|
III. Hypotheses Explaining the Association of Insulin Resistance
and PCOS
|
|---|
A. Causal association
1. Do androgens cause insulin resistance? If glucose
utilization is expressed as a function of muscle mass rather than total
body mass, women do appear to be more insulin sensitive than men (158, 159). Moreover, when isolated fat cells are compared, female adipocytes
are more sensitive than male adipocytes to insulin-mediated glucose
uptake (160). These are subtle differences, however, and do not
approach the degree of impairment in insulin sensitivity observed in
PCOS (54, 55). Finally, in the rare syndromes of extreme insulin
resistance and hyperandrogenism, specific molecular defects in insulin
action have been clearly identified as the cause of insulin resistance
(19, 161).
It is possible, however, that androgens may produce mild insulin
resistance. Women receiving oral contraceptives containing
"androgenic" progestins can experience decompensations in glucose
tolerance, as can individuals receiving synthetic anabolic steroids
(162, 163). Prolonged testosterone administration to female-to-male
transsexuals, which produced circulating testosterone levels in the
normal male range, resulted in significant decreases in
insulin-mediated glucoses uptake in euglycemic clamp studies (164).
These decreases were largest at lower doses of insulin (25% at
300 pM steady-state levels), not significant at moderate
insulin doses (1,000 pM steady-state levels), and
minimal at higher doses (7% at 5,000 pM steady-state
levels) (164) (Fig. 15). Studies in
testosterone-treated castrated female rats have suggested that
androgen-mediated insulin resistance may be the result of an increase
in the number of less insulin-sensitive type II b skeletal muscle
fibers (165) and an inhibition of muscle glycogen synthase activity
(166).
View larger version (25K):
[in this window]
[in a new window]
|
Figure 15. Hyperinsulinemic euglycemic clamp studies basally
and during treatment with virilizing doses of testosterone in 13
female-to-male transsexuals. Insulin-mediated glucose disposal
decreased significantly at low and at high doses of insulin.
[Reproduced with permission from K. H. Polderman et
al.: J Clin Endocrinol Metab 79:265–271,
1994 (164). © The Endocrine Society.]
|
|
It has been more difficult to demonstrate that decreasing
androgen levels improve insulin sensitivity in PCOS. We found no
significant changes in peripheral or hepatic insulin action in
profoundly insulin-resistant obese PCOS women by single-insulin dose
(steady-state insulin levels 600 pM) glucose clamp
studies after prolonged androgen suppression produced by the
administration of an agonist analog of GnRH (167). Diamanti-Kandarakis
and colleagues (168) reported that antiandrogen therapy did not alter
insulin sensitivity in PCOS. Other investigators have found modest
improvements in insulin sensitivity in PCOS during androgen suppression
or antiandrogen therapy (169, 170) (Fig. 16). Such changes were apparent in less
insulin-resistant, less obese, or nonobese PCOS women (169, 170).
Moreover, insulin resistance was improved but not abolished (170) (Fig. 16). It is of considerable interest that the effects of sex steroids on
insulin sensitivity appear to be sexually dimorphic. Testosterone
administration to obese males improves insulin sensitivity (171), and
synthetic estrogen administration to male-to-female transsexuals
produces insulin resistance (164).
View larger version (33K):
[in this window]
[in a new window]
|
Figure 16. Basal and insulin-mediated glucose disposal in 43
hyperandrogenic women and 12 control women. The hyperandrogenic women
were studied before and after 3–4 months of antiandrogen therapy with
spironolactone, flutamide, or Buserelin. Insulin-mediated glucose
disposal increased significantly during treatment
(P < 0.01). [Adapted with permission from P.
Moghetti et al.: J Clin Endocrinol
Metab 81:952–960, 1996 (170). © The Endocrine Society.]
|
|
Givens and colleagues (172) have proposed that androgens have
differential effects on insulin action, with testosterone worsening
insulin sensitivity and the adrenal androgen, dehydroepiandrosterone
(DHEA), improving it. This hypothesis is based on differing
correlations of these steroids with insulin-binding studies in blood
cells and on their observation that women with elevated
dehydroepiandrosterone sulfate (DHEAS) levels have normal insulin
sensitivity (172). The one direct in vitro study
supporting this hypothesis was constrained by a small sample size
(n = 3), and the examination of testosterone and DHEA effects on
insulin binding using blood cells rather than a more relevant insulin
target tissue (172). Studies in which DHEA or DHEAS have been
administered to humans have failed to support this hypothesis.
Administration of supraphysiological amounts of DHEA (which also result
in testosterone elevations since DHEA is a testosterone prehormone) has
produced mild hyperinsulinemia in women, but had no effects on insulin
sensitivity in men, as would be expected given the sexually dimorphic
effects of androgens on insulin action (173, 174). Moreover, PCOS women
with elevated DHEAS levels similar to those in ovulatory
hyperandrogenic women are significantly more insulin resistant, arguing
against an insulin-sensitizing action of DHEA (49, 175).
In summary, the modest hyperandrogenism characteristic of PCOS may
contribute to the associated insulin resistance. Additional factors are
necessary to explain the insulin resistance, since suppressing androgen
levels does not completely restore normal insulin sensitivity (167, 170). Further, androgen administration does not produce insulin
resistance of the same magnitude as that seen in PCOS (54, 55, 164).
Finally, there are clearly defects in insulin action that persist in
cultured PCOS skin fibroblasts removed from the hormonal milieu for
generations (see above) (69).
2. Does hyperinsulinemia cause hyperandrogenism? The syndromes
of extreme insulin resistance are commonly associated with
hyperandrogenism when they occur in premenopausal women (19, 20) (Table 1). The cellular mechanisms of insulin resistance in these conditions
range from antibodies that block insulin binding to its receptor (type
B syndrome) to genetic defects in the receptor resulting in decreased
numbers and/or depressed function of the receptor (type A syndrome,
leprechaunism); the common biochemical feature is profound
hyperinsulinemia (19, 20) (Table 1). Accordingly, it has been proposed
that hyperinsulinemia causes hyperandrogenism. Insulin can be shown
experimentally to have a variety of direct actions on steroidogenesis
in humans (1, 9, 20). Insulin can stimulate ovarian estrogen, androgen,
and progesterone secretion in vitro (1, 20, 176).
Although some of these actions have been observed at physiological
insulin concentrations, most actions have been observed at higher
concentrations (1, 20).
The presence of insulin receptors in crude ovarian membranes does not
necessarily indicate a physiological role for insulin in the regulation
of steroidogenesis since such receptors are widely distributed through
the body (51, 83). Insulin is present in human follicular fluid but in
concentrations most likely representing an ultrafiltrate of plasma
rather than local production (177). In contrast, IGF-I is produced by
human ovarian tissue, and IGF-I receptors are present in the ovary
(178, 179). IGF-I and its receptor share considerable sequence,
structural, and functional homology with insulin and its receptor,
respectively (180). The IGF-I receptor is a heterotetramer with two
,ß-dimers assembled analogous to the insulin receptor (85, 88, 181, 182, 183) (see above). Insulin can bind to the ligand-binding domain of
the IGF-I receptor and activate the tyrosine kinase activity of the
ß-subunit and the intracellular events normally mediated by IGF-I
(85, 88, 180, 181). IGF-I can bind to and activate the insulin
receptor, resulting in rapid effects on glucose metabolism (85, 88, 181). In general, the affinity of the IGF-I receptor for insulin is
considerably less than it is for IGF-I and vice versa
(181). However, this varies by tissue; thus data on receptor affinity
cannot be extrapolated from one tissue to another. There are also
so-called "atypical" IGF-I receptors that bind IGF-I and insulin
with similar affinity (184, 185). ,ß-Dimers of the insulin and
IGF-I receptor can assemble together to form hybrid heterotetramers
(11, 182, 186, 187).
Insulin-like growth factor-binding proteins (IGFBPs) are major
regulators of IGF action. IGFBPs can specifically bind IGF-I and
modulate its cellular actions by altering its bioavailability (182, 188). Insulin decreases hepatic production of IGFBP-1 and may, thus,
make IGF-I more biologically available (182). Growth factor regulation
of ovarian steroidogenesis appears to be primarily a paracrine system
with locally produced IGF-I and IGFBPs acting on neighboring cells in
concert with gonadotropins (1, 178, 179, 189). A number of other growth
factors, including IGF-II, EGF, and transforming growth factor- and
-ß, appear to have a role in the regulation (both stimulatory and
inhibitory) of ovarian steroidogenesis (1, 188, 190). Insulin cannot
interact directly with the receptors for these hormones (84, 88, 181, 182). However, the receptors for some of these growth factors, such as
the EGF receptor (which binds both EGF and transforming growth
factor-), are also protein kinases (1, 84, 88). Thus the potential
exists for communication between the insulin-IGF-I system and the other
protein kinase growth factor systems through receptor "cross-talk"
and/or by shared kinases or phosphatases that may regulate all of these
receptors (51, 191). For example, serine phosphorylation of the EGF
receptor also decreases its tyrosine kinase activity (103, 104). In
rodents, hyperinsulinemia can result in up-regulation of ovarian
IGF-I-binding sites, and this may provide yet another mechanism by
which insulin can modulate growth factor action (192).
Insulin in high concentrations can mimic IGF-I actions by occupancy of
the IGF-I receptor (1, 181, 182), and this has been a proposed
mechanism for insulin-mediated hyperandrogenism (8, 9, 10). However, it
has recently been shown that insulin has specific actions on
steroidogenesis acting through its own receptor (193). Moreover, these
actions appear to be preserved in insulin-resistant states (193, 194),
presumably because of differences in receptor sensitivity to this
insulin action or because of differential regulation of the receptor in
this tissue. Our studies in cultured skin fibroblasts suggest that a
mechanism for this may be selective defects in insulin action. Both
insulin- and IGF-I-stimulated glycogen synthesis are significantly
decreased in PCOS fibroblasts whereas thymidine incorporation is
similar to control fibroblasts (Fig. 17) (195). Thus only the signaling
pathways regulating carbohydrate metabolism may be impaired in PCOS,
while those involved in steroidogenesis are preserved. This would
explain the paradox of persistent insulin-stimulated androgen
production in insulin-resistant PCOS women. Insulin decreases hepatic
IGFBP-1 production, the major circulating IGF-I-binding protein (183).
Thus, bioavailable IGF-I levels are increased in insulin-resistant PCOS
women, and this may contribute to the ovarian steroidogenic
abnormalities via activation of the IGF-I receptor (68). In lean PCOS
women, increases in GH release may also affect ovarian steroidogenesis
(68).
View larger version (12K):
[in this window]
[in a new window]
|
Figure 17. Dose-response curves for insulin-stimulated
glycogen synthesis (left panel) and thymidine
incorporation (right panel) in confluent skin
fibroblasts from PCOS (•) and control ( , NL) women. Maximal
responses for insulin-stimulated glycogen synthesis were significantly
decreased (P < 0.001). There were no significant
differences in thymidine incorporation in the PCOS fibroblasts
(right panel). The dose-response curves for IGF-I were
similar to those for insulin (data not shown). [Reproduced with
permission from A. Dunaif (195).]
|
|
It has been more difficult to demonstrate insulin actions on
steroidogenesis in humans in vivo because it is not
feasible to administer insulin to nondiabetics for prolonged periods
(1, 196, 197, 198). Relatively physiological levels of insulin (100 µU/ml
or 600 pM), when infused over approximately 2 h, can
slightly increase plasma androstenedione levels in normal women (1).
However, these increases are minor and are not in the range seen in
women with hyperandrogenism. Moreover, it is arguable whether insulin
contributes to androgen production in normal women since insulin levels
in the 100 µU/ml (600 pM) range are generally seen
only after meals (1, 196). Furthermore, such transient meal-related
increases in insulin do not result in increased androgen levels,
whereas the more sustained increases produced by continuous insulin
infusion can slightly increase androgen levels (196).
Studies in which insulin levels have been lowered for prolonged periods
have been much more informative. This has been accomplished for 7 days
to 3 months with agents that either decrease insulin secretion,
diazoxide (199) or somatostatin (200), or that improve insulin
sensitivity, metformin (201) or troglitazone (202). Circulating
androgen levels have decreased significantly in women with PCOS in
these studies. Sex hormone binding globulin (SHBG) levels have
increased (199, 202), compatible with a major role for insulin in
regulating hepatic production of this protein (203, 204). Abnormalities
in apparent 17,20-lyase activity have improved in parallel with reduced
circulating insulin levels consistent with insulin-mediated stimulation
of this enzyme (205). However, estrogen levels also decreased
significantly, suggesting that insulin has diffuse effects on
steroidogenesis (202). Changes in estrogen levels were seen only when
insulin levels were lowered with troglitazone and thus, alternatively,
these changes might be the result of troglitazone-mediated increases in
sex steroid metabolism, a recently reported action of this agent
(Rezulin Package Insert, Parke-Davis, Morris Plains, NJ). It is also
possible that troglitazone has direct effects on steroidogenesis.
Indeed, the thiazolidinediones have been shown to have such effects on
granulosa cell steroidogenesis (206).
Most of the reported actions of insulin on steroidogenesis are observed
only in women with PCOS (197, 198) and are greatly enhanced by the
addition of gonadotropins when measured in in vitro
experiments (
Top
Abstract
I. Introduction
, Nob PCOS); nonobese normal women (
, Nob
NL); obese PCOS women (•, Ob PCOS); and obese normal women (
), and four control (