TREATMENT OF BIPOLAR DISORDER IN CHILDREN AND ADOLESCENTS - PART 9 doc

PART III
OTHER ISSUES

Other IssuesPolycystic Ovary Syndrome
CHAPTER 17
Polycystic Ovary Syndrome
PARAMJIT T. JOSHI and ADELAIDE S. ROBB
Polycystic ovary syndrome (PCOS) is a common, complex, and se-
rious endocrine disorder that affects women in their reproductive years.
The disorder was first recognized in 1935 by two gynecologists, Stein and
Leventhal (1935). They described a group of women who had a constella-
tion of infertility and several menstrual irregularities and were obese. They
were also found to have enlarged ovaries with multiple cysts at laparotomy
(Zawadski & Dunaif, 1992). Although it has come to be known as PCOS,
the ovarian morphology is a nonspecific finding. Approximately 20% of
normal women can have classic polycystic ovarian morphology on ultra
-
sound examination (Dunaif, 1997). Subsequently, others have reported
PCOS to be a disorder characterized by ovulatory dysfunction and hyper
-
androgenism that is thought to have a higher prevalence in women with ep
-
ilepsy and, perhaps, bipolar disorder.
Unlike the earlier reports by Stein and Leventhal (1935), others have
noted that women with PCOS can be lean, and the symptoms of androgen
excess may be absent. PCOS is the leading cause of hormonally related in
-
fertility and hirsutism and has been associated with multiple reproductive
and metabolic disorders. Approximately 80% of women with oligomen
-
orrhea, the clinical consequence of chronic anovulation, have PCOS (Dunaif,

1997). PCOS is also a major risk factor for type 2 diabetes mellitus (DM) in
women (Legro et al., 1999).
333
An association between the development of PCOS and the use of
antiepileptic drugs was first suggested by Isojärvi et al. (1993). This sugges
-
tion was based on clinical observations that women with epilepsy who re
-
ceived an antiepileptic drug experienced an increased rate of menstrual ab
-
normalities. Although there are no reports of PCOS in teenage girls, use of
mood stabilizers such as valproate is increasing for the treatment of bipolar
disorders in children and adolescents. Three letters to the editor of the Jour
-
nal of the American Academy of Child and Adolescent Psychiatry (Garland
& Behr, 1996; Eberle, 1998; Johnston, 1999) draw attention to this poten
-
tial association between use of valproate and development of PCOS in teenage
girls. They stimulate awareness for both clinicians and researchers and ask
for prospective studies to be conducted to shed light on this possible associ
-
ation between the use of mood stabilizers (specifically valproate), other
antiepileptic medications, and PCOS.
PCOS is associated with several reproductive, metabolic, and general
health disorders, including increased risk of miscarriage, insulin resistance,
hyperlipidemia, and cardiovascular disease. Endometrial, ovarian, and
breast cancer have all been reported to be more common in women with
PCOS (Hardiman, Pillay, & Atiomo, 2003; Balen, 2001; Coulam, Annegers,
& Kranz, 1983; Schildkraut et al., 1996). Elevated levels of circulating es-
trogen and the lack of cycling shedding of the endometrium are considered

to be the likely etiology of the increased risk for endometrial carcinoma in
women with PCOS (Rasgon, 2004; Hardiman et al., 2003; Siiteri, 1987). In
a large case control study examining the relationship between endogenous
steroid hormones and endometrial cancer, Potischman and colleagues
(1996) found increased risk of endometrial cancer in women with de-
creased sex-hormone-binding globulin (SHBG) and increased androgen lev-
els.
Although obesity is not always present in women with PCOS, it is a
common finding, with reports of up to 50% of women with PCOS being
obese, described primarily as the android-type obesity, with an increase in
the waist–hip ratio (Lobo & Carmina, 2000). It has been postulated that
obesity within itself may promote the development of PCOS through pe
-
ripheral aromatization of androgen to estrogen within adipose tissue
(Franks, 1995; Siiteri, 1987). Subsequently this obesity contributes to the
high rates of type 2 diabetes and hyperlipidemia and increases the risk of
cardiovascular disease in women with PCOS (Rasgon, 2004).
DEFINITION
PCOS is characterized by both hormonal and metabolic abnormalities. Dis
-
parate definitions of this syndrome have been proposed. A contemporary
working definition is hyperandrogenism and chronic anovulation (i.e.,
334 OTHER ISSUES
menstrual abnormalities and reproductive morbidity) in the absence of
identifiable pituitary or adrenal pathology. Interestingly, polycystic ovaries
are not necessary for the diagnosis to be made (Dunaif & Thomas, 2001;
Lobo & Carmina, 2000).
Many of these women have endocrine abnormalities such as elevated
testosterone and/or luteinizing hormone (LH ) levels (Franks, 1995). How
-

ever, some women with polycystic ovaries can be entirely endocrinologi
-
cally normal. Moreover, approximately 10% of women with all the features
of the endocrine syndrome have normal-appearing ovaries by ultrasound
examination (Ehrmann et al., 1995). Accordingly, the recommended diag
-
nostic criteria for PCOS at the 1990 National Institutes of Health (NIH)
conference on polycystic ovary syndrome (Zawadski & Dunaif, 1992) were
hyperandrogenism and chronic anovulation in the absence of specific dis
-
eases of the ovaries, adrenals, or pituitary. It is important to differentiate
this endocrine syndrome from the ovarian morphological change of poly
-
cystic ovaries. Outside the United States, it is still typical to diagnose
women by the appearance of their ovaries on ultrasound examination.
These differing diagnostic criteria for PCOS account for many of the dis-
crepant findings in the literature. It appears, however, that polycystic ova-
ries function abnormally, even in the absence of the peripheral endocrine
syndrome, both in the steroidogenic activity of the theca interna and in the
follicular responses to exogenous follicle-stimulating hormone (FSH; Franks,
1995).
Therefore, the definition of PCOS differs in the United States and in
Europe in the following ways:
• In the United States, PCOS is defined as a metabolic syndrome, and
anatomical changes need not be present to establish diagnosis.
•
In Europe, on the other hand, PCOS is defined as polycystic ovaries
in the presence of one or more clinical signs of endocrine dysfunc
-
tion, such as menstrual irregularities, hirsutism, or infertility.

The diagnostic criteria for PCOS on which participants agreed at the 1990
NIH–PCOS consensus conference (Zawadski & Dunaif, 1992; Duncan,
2001; Ernst & Goldberg, 2002), are as follows:
•
The presence of ovulatory dysfunction (polymenorrhea, oligomenor
-
rhea, or amenorrhea).
•
Clinical evidence of hyperandrogenism and/or hyperandrogenemia.
•
Exclusion of other endocrinopathies affecting adrenal or thyroid
function, such as hyperprolactinemia, hypothyroidism, adrenal hy
-
perplasia, or Cushing’s syndrome.
•
Exclusion of anatomical findings of polycystic ovaries, multifollicu
-
lar ovaries, or hyperandrogenism in isolation.
Polycystic Ovary Syndrome 335
However, as suggested by Chappell, Markowitz, and Jackson (1999),
the diagnosis of PCOS is generally made through a combination of clinical,
biochemical, and ultrasonographic findings.
DIFFERENCES BETWEEN PCOS
AND POLYCYSTIC OVARIES
Whereas PCOS is a complex endocrine disorder characterized by metabolic
and endocrine abnormalities that affects women in their reproductive
years, polycystic ovaries are a common but not intrinsically pathological
occurrence in 22–30% of the general female population (Luef, Abraham,
Haslinger, et al., 2002; Genton et al., 2001). The accepted definition of
polycystic ovaries by ultrasonographic and anatomical criteria is the pres

-
ence of at least 10 subcapsular follicular cysts, measuring 2–8 mm in diam
-
eter, arranged around or within thickened ovarian stroma (Adams et al.,
1985; Adams, Polson, & Franks, 1986; Duncan, 2001). As many as 25%
of women with radiological findings of polycystic ovaries have no endo-
crine or menstrual irregularities, suggesting that an isolated finding of
polycystic ovaries may be a normal variation and may not necessarily imply
altered fertility (Genton et al., 2001). It is therefore important to distin-
guish between these two conditions when interpreting clinical studies
(Ernst & Goldberg, 2002). A case control study of 258 women with and
without other hormonal or metabolic symptoms of PCOS, showed that
there was no significant effect on fertility (Hassan & Killick, 2003). There-
fore, the finding of polycystic ovaries in otherwise healthy women may not
necessarily predict reproductive dysfunction (Rasgon, 2004).
Other disorders that need to be considered in the differential for PCOS in
-
clude: nonclassic adrenal 21-hydroxylase deficiency (prevalence 1–5%), hyper
-
prolactinemia and Cushing’s syndrome (occasional occurrence), surreptitious
androgen use (rare), extreme insulin resistance syndromes, for example, type A
(rare), and ovarian and adrenal androgen-secreting neoplasms (very rare).
PREVALENCE
Only recently have there been studies of the prevalence of the classic endo
-
crine syndrome of hyperandrogenism and chronic anovulation. The preva
-
lence of PCOS in the general population of reproductive-age women hasbeen
estimated to be between 4 and12%,withoutany differences in prevalence be
-

tween Caucasian and African American women (Dunaif & Thomas, 2001;
Lobo & Carmina, 2000; Knochenhauer et al., 1998). However, most reports
show a higher prevalence (10.5–26%) of PCOS in women with epilepsy than
in the general population (Bauer et al., 2000; Bilo et al., 2001). Franks (1995)
reported that 37% of women with amenorrhea and 90% with oliomenorrhea
336 OTHER ISSUES
had PCOS. Bauer et al. (2000) studied 93 women with epilepsy and found
that the incidence of PCOS was 10.5% in an untreated group, 11.1% in a
valproate-treated group, and 10% in a carbamazepine-treated group. There
are no prevalence studies in females under the age of 18 years.
Valproate is an approved treatment for epilepsy syndrome. Bipolar
treatment guidelines from Canada and the United States recommend
valproate as a first-line strategy in the acute treatment of bipolar disorder
(O’Donovan et al., 2002). Most persons with bipolar disorder require
maintenance treatment, which necessitates the need for careful appraisal
of long-term tolerability and safety issues. There have been reports of
valproic acid inducing PCOS in females with epilepsy (Franks, 1995).
These observations have initiated preliminary investigation in bipolar dis
-
order (Dunaif & Thomas, 2001; Knochenhauer et al., 1998; Yen, 1991).
Recently, O’Donovan et al. (2002) reported that valproate-treated fe
-
males with bipolar disorder exhibited a high prevalence of menstrual ir
-
regularities and exhibited ultrasonographically confirmed polycystic ova
-
ries (41%). A study of ambulatory females with DSM-IV-defined bipolar
disorder between the ages of 18 and 45 (10 receiving valproate mono-
therapy) failed to identify any biochemical or ultrasonographic evidence
of PCOS in females receiving valproate or lithium (Rasgon et al., 2000).

It was noted by both groups that bipolar females exhibited a higher prev-
alence of menstrual disturbances than the general population. Others
have described the potential associations between PCOS and valproate
(Herzog, 1996; Post et al., 2001).
Although it awaits to be established whether females with bipolar dis-
order manifest a higher prevalence of primary reproductive endocrine dis-
orders, they appear to be more overweight or obese than the general popu-
lation (Suppes, Leverich, & Keck, 2001). Valproate and several other
psychotropic agents impart substantial weight gain (Ferriman & Gallwey,
1961; Roste et al., 2001). Excess weight gain may independently predispose
and portend risk for subsequent reproductive endocrine and metabolic dis
-
orders. Various theories have been offered to explain this higher prevalence
of PCOS and other reproductive disorders in these patient populations, in
-
cluding the effects of the disease itself and of antiepileptic drugs, especially
valproate, which may directly cause PCOS or indirectly lead to the disorder
by causing weight gain that triggers insulin resistance, increased testoster
-
one levels, and other reproductive abnormalities.
CLINICAL FEATURES
Hyperandrogenism and anovulation are the key features of PCOS, as de
-
fined by the National Institutes of Health (NIH) consensus diagnostic crite
-
ria. The common clinical manifestations of these abnormalities, therefore,
are as follows:
Polycystic Ovary Syndrome 337
Menstrual Irregularities
These may manifest themselves at puberty either with delayed menarche

followed by the onset of irregular periods or as the breakdown of a previ
-
ously regular menstrual cycle within a few years. Chronic anovulation in
PCOS is associated with disordered gonadotropin secretion and presents as
oligomenorrhea (8–10 menstrual cycles/year) or amenorrhea (the absence
of menstrual cycles) before menopausal onset (Lobo & Carmina, 2000).
Women with PCOS often are infertile, and for the few PCOS patients who
become pregnant, there are increased risks of miscarriage, gestational dia
-
betes, and pregnancy-induced hypertension (Duncan, 2001; Lobo & Car
-
mina, 2000; Ernst, 2002).
These menstrual irregularities are also associated with weight gain,
and it is reported that approximately half of the women with PCOS are
obese and that 20% of them will have either impaired glucose tolerance or
type 2 diabetes by the time they reach 40 years of age (Duncan, 2001).
Other risks associated with PCOS are endometrial hyperplasia or malig
-
nancy, hypertension, coronary heart disease, and unhealthy lipid profiles,
that is, elevated levels of triglycerides and low-density lipoproteins (LDLs).
Hyperandrogenism
Hyperandrogenism may appear clinically as hirsutism, acne, male pattern
balding, and/or male distribution of body hair or alopecia (Lobo &
Carmina, 2000). The virilizing features of this illness are due to the elevated
androgens (testosterone and androstenedione) and their precursors dehydro-
epiandrosterone (DHEA) and dehydroepiandrosterone-sulfate (DHEAS;
Herzog, 1996). The excess androgens are associated with subtle hyperes
-
trogenism (Lobo & Carmina, 2000; Dahlgren et al., 1992).
It is thought that the disorder may be caused by increased steroido

-
genic activity that is an intrinsic defect in the ovary (Dunaif & Thomas,
2001). In vitro studies show that women with PCOS secrete increased
amounts of androstenedione (an androgen) and increased amounts of 17-
hydroxyprogesterone (a steroid that is an intermediate in the androgen and
glucocorticoid biosynthetic pathway) from thecal cells (the androgen-
producing cells of the ovary). This increased secretion by thecal cells may
be a result of dysregulation of the rate-limiting enzyme in androgen
biosynthesis, cytochrome P-450c17α.
Reproductive Endocrine Abnormalities
Reproductive endocrine abnormalities are often present, but none are path
-
onomonic or found in all women with the disorder. The common endocrine
abnormalities include: elevation of LH in urine and serum and low normal
338 OTHER ISSUES
plasma FSH, leading to an increased LH/FSH ratio (Duncan, 2001). This go
-
nadotropin hormonal imbalance leads to an increase in LH-stimulated ovar
-
ian steroidogenesis and a decrease in follicle maturation (Franks, Mason, &
Willis, 2000; Rasgon, 2004). This incomplete follicle maturation in turn is
thought to lead to the formation of a larger number of small, immature folli
-
cles and, subsequently, the formation of follicular cysts (Rasgon, 2004).
Furthermore, decreases in SHBG as a result of hyperinsulinemia and
hyperandrogenism, are commonly seen in this syndrome. The decreased
SHBG concentration increases the bioavailable fraction of androgens and
estrogens, which may increase free testosterone levels. In general, however,
estrogen and FSH levels remain in the normal range (Isojärvi et al., 1995;
Bauer et al., 2000; Herzog, 1996).

Metabolic Abnormalities
Metabolic abnormalities such as hyperinsulinemia and insulin resistance
occur at greater frequency and intensity in women with PCOS. Approxi-
mately 40% of women with PCOS have been shown to have impaired glu-
cose tolerance tests (Ehrmann et al., 1999; Legro et al., 1999). The rates of
impaired glucose tolerance vary from 31 to 35%, versus 7.8% when com-
pared with the general U.S. female population (Ehrmann et al., 1999; Legro
et al., 1999). Consequently, up to 20% of obese women may exhibit type 2
diabetes by age 40 years (Dunaif, 1995). Insulin resistance is independent
of the effect of obesity and may occur regardless of whether the women are
lean or obese compared with normal women (Franks, 1995). Further, Lobo
and Carmina (2000) have shown that insulin resistance has been found to
be more pronounced in women with chronic anovulation than in those
who have ovulatory cycles.
In women with PCOS, insulin resistance is characterized by decreased
sensitivity to insulin in peripheral tissues but not hepatic resistance, unlike
insulin resistance in type 2 diabetes. Hopkinson and colleagues (1998) re
-
ported that there was support for suggesting that decreased peripheral insu
-
lin sensitivity and, consequently, hyperinsulinemia were pivotal to the
pathogenesis of PCOS. Hopkinson hypothesized that insulin acts in the
liver to inhibit the production of insulin-like growth factor 1 (IGF-1) binding
protein and SHBG, with the latter leading to an increase in free testoster
-
one. Therefore, according to Hopkinson, insulin resistance not only in
-
creases the secretion of ovarian androgen but also promotes an increase in
the proportion of free (biologically active) hormone.
Lipid and Lipoprotein Abnormalities

These abnormalities include elevated LDLs and triglycerides, decreased lev
-
els of high-density lipoproteins (HDL), and apolipoproteins A-1 (Legro,
Polycystic Ovary Syndrome 339
Kunselman, & Dunaif, 2001; Lobo & Carmina, 2000). Additionally, im
-
paired fibrinolytic activity has also been reported, as assessed by measure
-
ments of elevated levels of circulating plasminogen activator inhibitor levels
(a potent inhibitor of fibrinolysis), which has been shown to be a risk factor
for the occurrence of hypertension and myocardial infarction (Dahlgren et
al., 1992; Hopkinson et al., 1998). However, decreased levels of HDL is
considered to be the most characteristic lipid abnormality in women with
PCOS (Hopkinson et al., 1998).
Reproductive Abnormalities
These abnormalities can often develop shortly after menarche in many
women and can last most of their reproductive lives. In others it may ap
-
pear as a breakdown of a previously regular menstrual cycle, which is often
associated with weight gain (Duncan, 2001). However, the most pressing
concern is the occurrence of varying degrees of infertility, with PCOS iden
-
tified in 75% of women with anovulatory infertility (Legro et al., 2001).
Women have also been reported to be at much greater risk of having multi-
ple pregnancies through ovulation induction or after in vitro fertilization
(Legro et al., 2001).
ETIOLOGY
The etiology of PCOS is not fully understood, though several authors have
suggested that PCOS is caused by interactions between a variety of genetic,
neuroendocrine, metabolic, and environmental factors (Rasgon, 2004;

Dunaif & Thomas, 2001).
Genetic Factors
Familial aggregation of PCOS has been clearly established, suggesting ge
-
netic susceptibility (Franks, 1995; Legro & Strauss et al., 1998). Various
modes of transmission have been discussed, including an autosomal domi
-
nant inheritance pattern based on familial aggregation of hyperandro
-
genism in first-degree relatives of patients with PCOS (Legro, Strauss, et al.,
1998; Ernst & Goldberg, 2002). In addition, brothers of women with
PCOS often show evidence of insulin resistance and elevated dehydroepian
-
drosterone levels, findings that might suggest their reproductive and meta
-
bolic phenotypes resemble those of their sisters with PCOS (Dunaif &
Thomas, 2001). There is also evidence that there may be a genetic defect in
ovarian and adrenal androgen biosynthesis that may synergize with a meta
-
bolic abnormality (Rasgon, 2004; Lobo & Carmina, 2000). Studies on
women with PCOS suggest that the disorder may be caused by increased
340 OTHER ISSUES
steroidogenic activity that is intrinsic, presumably a genetic defect in the
ovary (Dunaif & Thomas, 2001). In vitro studies have shown that women
with PCOS secrete increased amounts of androstenedione (an adrogen) and
17-hydroxyprogesterone (a steroid that is an intermediate in the androgen
and glucocorticoid biosynthetic pathway) from thecal cells (the androgen
producing cells of the ovary). This increased secretion by thecal cells may
be a result of dysregulation of the rate-limiting enzyme in androgen
biosynthesis, cytochrome P-450c17a (Franks, 1995).

In order to determine whether there was a biochemical reproductive
endocrine phenotype, Legro, Spielman, et al. (1998) studied the sisters of
women with PCOS and found that there was familial aggregation of
hyperandrogenemia in PCOS kindreds, with 46% of 115 sisters thus af
-
fected. Only one-half of these sisters fulfilled diagnostic criteria for PCOS
with chronic anovulation and hyperandrogenemia. The remaining affected
sisters had hyperandrogenemia with regular menses. The affected sisters
also had a significant elevation of DHEAS levels, suggesting that there was
an adrenal component to the hyperandrogenemia. The distribution of tes-
tosterone levels in the sisters appeared to be bimodal. Although the sample
size was relatively small, this suggested that testosterone levels in PCOS
families reflected a monogenic trait controlled by two alleles at an auto-
somal locus.
A genetic defect may be responsible for the insulin resistance found in
women with PCOS (Ernst & Goldberg, 2002). Defects in insulin receptors
have been reported in up to half of women with PCOS who also may have
a decrease in tyrosine phosphorylation and an increase in serine phosphor-
ylation. These factors can all contribute to impaired insulin activity
(Dunaif, 1995).
Neurological Factors
The incidence of menstrual irregularities and PCOS both appear to be more
common among women with epilepsy than among women without epi
-
lepsy (Bilo et al., 1988). PCOS has been reported to occur in 20% of
women with temporal lobe epilepsy and 25% of women with complex par
-
tial seizures (Herzog et al., 1986). Another study described PCOS in 15%
of women with primary generalized epilepsy (Bilo et al., 2001). In general,
it has been reported that PCOS occurs in 10.5–26% of women with epi

-
lepsy (Bauer et al., 2000; Bilo et al., 2001). Some authors have suggested
that epilepsy may play an intrinsic role in the development of PCOS
(Herzog et al., 1986; Ernst & Goldberg, 2002) and postulate that epileptic
discharges from the amygdala to the hippocampus may affect the secretion
of gonadotropin-releasing hormone (GnRH). Increased GnRH pulse fre
-
quency in turn promotes LH secretion over FSH secretion and leads to an
elevated LH/FSH ratio (Knobil, 1980).
Polycystic Ovary Syndrome 341
Furthermore, studies have demonstrated higher LH pulse frequencies
with left-sided than with right-sided temporal foci (Drislane et al., 1994),
and PCOS may be more common with left temporolimbic epileptiform dis
-
charges than with right temporolimbic epileptiform discharges. In turn,
anovulatory cycles can trigger limbic seizure discharges (Herzog, 1993).
Levels of progesterone, a hormone that can raise the seizure threshold, are
low in anovulatory women, including those with PCOS (Herzog et al.,
1986). Therefore, it is postulated that temporolimbic structures in anovula
-
tory women are primarily exposed to estrogen, which has a known
proseizure effect. In addition, limbic seizure discharges may also reduce lev
-
els of serum dopamine, leading to increased LH and prolactin secretion by
the pituitary (Herzog et al., 1986; Ernst & Goldberg, 2002). Alternatively,
a dysfunction in neurotransmission or genetic vulnerability common to
both epilepsy and reproductive endocrine disorders may account for the
link between PCOS and epilepsy (Herzog et al., 1986; Ernst & Goldberg,
2002).
Endocrine Factors

The possible endocrine factors that may contribute to the development of
PCOS include an increased LH/FSH ratio and increased insulin and andro-
gen concentrations. Increased levels of 17-hydroxyprogesterone levels in
thecal cells have also been implicated in playing a role in the development
of PCOS. Abbott, Dumesic, and Franks (2002) propose a developmental
theory in understanding some of these endocrine etiological factors. Ac-
cordingly, they postulated that during gestation, placental human chorionic
gonadotrophin (hCG), fetal pituitary LH, and genes regulating folliculo-
genesis and steroidogenesis, individually or together, result in fetal ovarian
hyperandrogenemia that leads to prenatal, and possibly prepubertal, expo
-
sure to excess androgen. Postpubertally, this early exposure to excess
androgen diminishes steroid hormone negative feedback on pituitary LH,
resulting in abnormal LH secretion and predisposing women to accumula
-
tion of abdominal adiposity that exaggerates insulin resistance. The result
-
ing hyperinsulinemia interacts with LH hypersecretion to augment ovarian
steroidogenesis and to induce premature arrest of the follicle development
and anovulation.
Metabolic Factors
The insulin resistance that has been described in patients with PCOS de
-
creases the release of SHBG in the liver, which in turn increases free andro
-
gen levels that have been implicated in PCOS (Duncan, 2001). When pres
-
ent, obesity worsens insulin resistance (Dunaif & Thomas, 2001) further
making the patients more vulnerable to develop PCOS by the mechanism
342 OTHER ISSUES

described herein, as obesity independently is associated with decreased lev
-
els of SHBG and elevated estrogen levels (Ernst & Goldberg, 2002).
Research has shown that experimentally raising insulin levels can directly
stimulate ovarian androgen production in women with PCOS (Dunaif &
Thomas, 2001). Insulin can also stimulate steroidogenesis by enhancing the
sensitivity to adrenocorticotropic hormone and thereby increasing pituitary
LH release. These reproductive effects of insulin appear to be limited to
women with PCOS. It is important to note that insulin-lowering therapies
can restore menstrual cycles in some chronically anovulatory women with
PCOS (Dunaif & Thomas, 2001).
Environmental Factors
A number of investigators have recognized that PCOS-like symptoms may
be manifested in response to environmental cues, such as prenatal exposure
to androgens and weight gain (Adams et al., 1985; Adams et al., 1986).
Further, anabolic steroids and antiepileptic drugs (AEDs) have also been
implicated in the development of PCOS. Investigators have reported an in-
creased frequency of reproductive disorders in patients with epilepsy
(Franks, 1995; Wang, Davies, & Norman, 2001). Thus it is also possible
that this population of patients is more likely to be treated with valproate,
and, as a consequence, investigators observed a higher incidence of PCOS-
like symptoms. However, the finding that valproate increases steroid
biosynthesis in theca cells isolated from both normal-cycling and PCOS pa-
tients suggests that valproate treatment could independently induce PCOS-
like symptoms in the absence of a genetic predisposition for PCOS (Herzog
& Schachter, 2001; Ernst & Goldberg, 2002).
CORRELATION BETWEEN BIPOLAR
DISORDER AND PCOS
Reproductive disorders have also been reported to have an increased
prevalence in women with bipolar disorder. Similar to epilepsy, contro

-
versy exists as to whether these abnormalities are caused by the bipolar
disorder or by treatment (Rasgon et al., 2000; O’Donovan et al., 2002).
The high rate of reported menstrual disturbances may indicate a preexist
-
ing compromise in reproductive endocrine function in women with bipo
-
lar disorder. This preexisting potential compromise, in turn, may be a
marker for dysregulation of the hypothalamic–pituitary–gonadal (HPG)
axis. Matsunaga and Sarai (1993) evaluated the HPG axis in 12 women
with bipolar disorder. They reported elevated basal LH in 8 women and
decreased basal FSH in 6 women. Polycystic ovaries were observed by
ultrasonography in 8 of 12 cases, suggesting that a relationship might ex
-
Polycystic Ovary Syndrome 343
ist between bipolar disorder and the PCOS-associated hormonal abnor
-
malities in these cases. In a study by Rasgon et al. (2005), 80 women
ages 18–45 years being treated for bipolar disorder and not taking ste
-
roid contraceptives were recruited to complete questionnaires about their
menstrual cycles and to provide blood samples for measurement for a
range of reproductive endocrine and metabolic hormone levels. All women
received antimanic medications for bipolar disorder. The investigators re
-
ported that 50% of women reported current menstrual abnormalities that
preceded the diagnosis of bipolar disorder. Fifteen percent reported devel
-
oping menstrual abnormalities since treatment for bipolar disorder, of
which 80% reported changes in menstrual flow (heavy or prolonged

bleeding) and 33% reported changes in cycle frequency. These results
were consistent with an earlier report by Rasgon and colleagues (2000)
indicating that menstrual abnormalities are common in women with bi
-
polar disorder and that the HPG axis may be compromised in some
women with bipolar disorder. In the latter study the participants received
antimanic medications, and 35% also were taking oral contraceptives.
Although it is possible that long menstrual cycles resulted from the phar-
macological treatment of bipolar disorder, the observations that men-
strual abnormalities precede the onset of bipolar symptoms and that si-
multaneous oral contraceptive use is associated with long menstrual
cycles suggest that women with bipolar disorder may have an underlying
predisposition to long or abnormal menstrual cycles (Matsunaga & Sarai,
1993; Rasgon et al., 2000). A recent study examined 300 women ages
18–45 years with bipolar disorder who were evaluated for PCOS. A compari-
son was made between the incidence of hyperandrogenism and oligomen-
orrhea that developed while taking valproate versus other anticonvulsants
(lamotrigine, topiramate, gabapentin, carbamaxepine, and oxcarbaze
-
pine) and also lithium. Of the 230 women who completed the evalua
-
tion, results showed that hyperandrogenism with oligomenorrhea developed
in 9 (10.5%) of 86 women on valproate and 2 (1.4%) of 144 women on
a nonvalproate anticonvulsant or lithium (relative risk 7.5%; p = .002).
Oligomenorrhea always began within 12 months of valproate use (Joffe
et al., 2006).
Others have suggested screening for bipolar disorder in women with
PCOS. In a pilot study of 78 women identified with PCOS, Klipstein and
Goldberg (2006) reported that 28% had either been previously diagnosed
with bipolar disorder or had met Mood Disorder Questionnaire (MDQ)

threshold criteria for bipolar disorder. These authors concluded that there
is likely a higher rate of women with PCOS who screen positive for bipolar
disorder than is expected in the general population. Further, they postu
-
lated that there could be a link between PCOS and bipolar disorder second
-
ary to a possible shared HPG axis abnormality.
344 OTHER ISSUES
CORRELATION BETWEEN ANTIEPILEPTIC
DRUGS AND PCOS
In Patients with Epilepsy
It has been suggested that use of anticonvulsants in general and sodium
valproate in particular leads to an increased incidence of polycystic ova
-
ries and PCOS (Isojärvi et al., 1993; Isojärvi et al., 1995; Isojärvi et al.,
1996; Isojärvi et al., 1998). In their initial study of 238 women with epi
-
lepsy, Isojärvi and colleagues (1993) reported some symptoms of PCOS
in 96 women, such as irregular menstrual cycles and hyperandrogenism,
although the criteria used were not as defined by the NIH. Twenty-nine
(12%) were treated with valproate alone, 120 (50%) with carbamazepine
alone, and 12 (5%) with a combination of valproate and carbamazepine;
62 (26%) were treated with other AEDs; and 15 (6%) received no medi
-
cation. Menstrual irregularities were reported by 45% of women receiv
-
ing valproate, by 19% of those receiving carbamazepine, by 25% of
those receiving a combination of both, by 13% receiving other medica-
tions, and by none who were untreated. Results of vaginal ultrasono-
graphy completed on the 96 women with histories of menstrual irregular-

ities and epilepsy revealed polycystic ovaries in 43% of those treated with
valproate, in 22% of those treated with carbamazepine, and in 50% of
the combination group.
In a subsequent study, Isojärvi et al. (1995) studied 8 women with epi-
lepsy before and after 1 and 5 years of carbamazepine treatment. All of the
women had regular menstrual cycles before the study. Of these 8 women, 2
reported menstrual irregularities, and 3, including these 2 women, had ele-
vated SHBG levels after 5 years of carbamazepine therapy.
In a 1996 study, Isojärvi et al. compared 22 women with epilepsy re
-
ceiving valproate monotherapy, 43 women receiving carbamazepine mono
-
therapy, and a normal control group without epilepsy. Polycystic ovaries,
hyperandrogenism, or both were found in 64% of the valproate group,
21% of the carbamazepine group, and 19% of the control participants.
Further, polycystic ovaries and hyperandrogenism occurred more often in
obese women on valproate (about 40%) than in lean women on valproate
(about 20%). These women also had slightly higher fasting insulin levels
(valproate-treated group, 16.9 ± 10.5; carbamazepine group, 15.4 ± 10.5;
control group, 9.6 ± 5.1). The investigators also reported that half of the
valproate-treated group had progressive weight gain (mean = 21 kg) and
lower insulin-like growth factor binding protein levels (Duncan, 2001).
This weight gain is thought to perhaps lead to the development of the meta
-
bolic syndrome, including hyperinsulinemia, which has been shown to
stimulate polycystic ovaries and androgen synthesis. The diagnostic criteria
used in the preceding two studies did not meet NIH criteria, and no distinc
-
Polycystic Ovary Syndrome 345
tion was made by the authors between polycystic ovaries and PCOS. No in

-
formation was available about ovarian structure and function prior to
treatment with AEDs.
To examine whether discontinuing valproate would reverse its appar
-
ent effects on weight, menstrual irregularities, frequency of polycystic ova
-
ries, testosterone levels, HDL-C, and triglycerides, Isojärvi et al. (1998)
studied 16 women with epilepsy who had polycystic ovaries or hyperandro
-
genism and were obese and who were switched to lamotrigine. Over the
next year, 12 women who remained in the study lost weight, decreased
their waist and hip circumference, and showed a decrease in their body
mass index (BMI). In addition, insulin and testosterone levels decreased,
and the lipid profiles improved. Ultrasonography revealed that the number
of follicles per ovary decreased and the number of women who had men
-
strual abnormalities decreased. The small sample size, lack of control
group, selection of only obese women, and nonrandomization were among
the limitations of this study.
In a larger cohort of women with epilepsy being treated with either
valproate or carbamazepine, Isojärvi et al. (2001) further assessed the fre-
quency of metabolic and reproductive endocrine disorders. The authors
studied 72 women and noted that neither the duration of epilepsy, treat-
ment with valproate, nor dose was associated with polycystic ovaries or
hyperandrogenism. Similar to their previous study (Isojärvi et al., 1996),
polycystic ovaries and hyperandrogenism were seen in 79% of obese
women and 65% of lean women treated with valproate. This information
was not reported for the carbamazapine group or control group. The fre-
quency of polycystic ovaries and hyperandrogenism in the carbamazepine-

treated group was 20% and, among the normal controls, 19%. Menstrual
disorders were seen in 79% of the obese valproate-treated women and in
48% of the lean valproate-treated women, in comparison with 13% of the
obese control participants and 17% of the lean control participants who
had menstrual irregularities. This information was not reported for the
carbamazepine-treated women, though this group had a slightly lower fre
-
quency of menstrual disorders than controls. The authors also did not re
-
port separate hormone values for lean or obese carbamazepine-treated or
control-group participants.
Compared with controls in the same study, women on valproate had
higher testosterone levels and lower ratios of HDL-C to total cholesterol.
As expected, obese valproate-treated women had higher BMI, higher insu
-
lin, and higher lipid profiles and lower HDL-C to total cholesterol ratios
than the lean valproate-treated women. However, the contribution of
valproate versus obesity to metabolic and hormonal abnormalities in this
sample was unclear as only 14 (38%) of 37 women taking valproate were
obese. Furthermore, LH levels in valproate-treated women were not signifi
-
cantly different from those of the control group. Although at first glance
346 OTHER ISSUES
these data appear to support the hypothesis that the metabolic abnormali
-
ties induced by valproate may contribute to the development of the symp
-
toms of PCOS, it is difficult to make generalizations because of the study
design and unreported data sets. The effects of obesity versus those of
AEDs on the development of polycystic ovaries and hyperandrogenism

need to be studied independently.
Several other investigators have conducted studies to replicate the find
-
ings reported by Isojärvi and colleagues. Pylvanen et al. (2002) demon
-
strated that the rates of obesity were the same in patients with epilepsy tak
-
ing valproate as in normal control participants. However, both lean and
obese patients receiving valproate had higher insulin levels than controls,
but there was no difference in leptin levels between any of the groups. In re
-
viewing this study, Rasgon (2004) concluded that these data support a
causative role for valproate rather than obesity in the development of
hyperinsulinemia and suggest that valproate may be directly linked to insulin-
stimulated hyperandrogenism in women with epilepsy. However, it is im
-
portant to note that no untreated patients with epilepsy were included in
the study, and the effect of epilepsy on hyperinsulinemia could not be inde-
pendently assessed (Rasgon, 2004).
It can be argued that early exposure to valproate may increase the risk
of PCOS secondary to earlier changes in cellular mechanisms that are
thought to be the potential mechanisms for valproate dependent changes
that can have an effect on follicular development in the young ovaries of
developing girls. Because increased androgen production is a stable pheno-
type of PCOS theca cells, it is possible that the earlier these biochemical
and perhaps structural changes occur, the more irreversible they may be,
leading to a greater susceptibility and risk for the development of PCOS
secondary to valproate treatment.
In another study examining the effects of AEDs in women with and
without epilepsy, Betts, Yarrow, Dutton, Greenhill, and Rolfe (2003) stud

-
ied 105 women, 54 of whom had been treated only with valproate and 51
of whom had either been treated with lamotrigine or carbamazepine for at
least 1 year. They were compared with 50 women without epilepsy. Mea
-
surements of FSH, LH, testosterone, and prolactin were obtained from
days 2 to 6 of their menstrual cycle, along with magnetic resonance image
(MRI) scans of their pelvises. Women with epilepsy in general were signifi
-
cantly more likely to exhibit polycystic ovaries on their scans. Women tak
-
ing valproate but not an oral contraceptive were also significantly more
likely to have clinical biochemical evidence of PCOS with raised LH and/or
testosterone than women who did not have epilepsy. This was not the case
with women on either lamotrigine or carbamazepine. The investigators
concluded that women with epilepsy—particularly if they are not taking
oral contraceptives—are more likely than those without epilepsy to have
polycystic ovaries by the European definition. Second, they concluded that
Polycystic Ovary Syndrome 347
women with epilepsy who are not taking oral contraceptives are signifi
-
cantly more likely to have PCOS if they have ever taken valproate (but not
lamotrigine or carbamazepine; p = .003). Based on this finding, they sug
-
gested that perhaps oral contraceptives protect against polycystic ovaries in
women who take valproate as an anticonvulsant. They reason that this in
-
creased risk with valproate is due to valproate being the only anticonvul
-
sant to be associated with an increase in insulin resistance, which is one of

the many factors to be associated with polycystic ovaries. Betts et al. (2003)
recommended that valproate should be avoided in women of childbearing
age.
Murialdo et al. (1997) studied 101 women with epilepsy between the
ages of 16 and 50 years who were treated with a number of different AEDs
that included phenobarbitol, phenytoin, and primidone, in addition to
valproate and carbamazepine. They reported the occurrence of polycystic
ovaries in 12% of the phenobarbitol group; 21% of the carbamazepine
group; 0% of the valproate group; 40% of those receiving polytherapy that
included valproate; and 13% of the group receiving polytherapy that did
not include valproate. None of the women with polycystic ovaries exhib-
ited any PCOS. A subsequent study by the same investigators (Murialdo et
al., 1998) of 65 women with epilepsy being treated with valproate,
carbamazepine, or phenobarbitol reported that the rates of polycystic ova-
ries, ovary volume, and hirsutism did not differ significantly among the
three treatment groups.
In a study of 50 women with various forms of epilepsy, Bilo et al.
(2001) found no significant association between reproductive endocrine
disorders such as PCOS, hypothalamic amenorrhea, and luteal phase
deficiency based on epilepsy type or AED treatment with or without
valproate. They concluded that PCOS preceded the use of any AEDs and
was increased in women with epilepsy independent of any drug treatment
effects.
In an international multicenter study, 222 women with epilepsy of re
-
productive age were examined for the presence of menstrual irregularities
and endocrine changes related to treatment with AEDs. Their results
showed that although testosterone levels were within normal range for
both groups (valproate or lamotrigine monotherapy), they were somewhat
higher in the valproate group. On the other hand, total cholesterol and

LDL levels were lower in the valproate group, and there was no difference
in the insulin levels (Taylor, 2001).
Similarly, Bauer et al. (2000), in a prospective study, found no associa
-
tion between PCOS and valproate or carbamazapine treatment in a study
of 93 women with epilepsy, refuting the earlier results of Isojärvi and col
-
leagues (1993; Isojärvi et al., 1995; Isojärvi et al., 1996; Isojärvi et al.,
1998). In this study the diagnosis of PCOS was made using the NIH criteria
and hence distinguished between polycystic ovaries and PCOS. Compari
-
348 OTHER ISSUES
sons were made between four groups of women: untreated, valproate
treated, carbamazepine treated, and polytherapy treated.
In a longer term study following 43 women with epilepsy for 3 years
who were being treated with valproate (n = 22), lamotrigine, or carbama
-
zepine (n = 21), Luef, Abraham, Trink, et al. (2002) also found no associa
-
tion between valproate treatment and the frequency of menstrual disorders,
polycystic ovaries, or both. However, they did report increased androgen
levels in women on valproate. A larger study of 105 women with epilepsy,
completed by the same authors (Luef, Abraham, Trinka, et al., 2002), also
reaffirmed the findings reported in the earlier study. Patients in this study
treated with valproate (n = 52) had a slightly lower incidence of menstrual
disturbance (11%) and polycystic ovaries (12%) than those treated with
carbamazepine (n = 53; 16% and 14%, respectively). These differences
were not found to be significant. The rate of incidence of polycystic ovaries
in the patients was 27%, compared with the rate of 20–30% in the general
population, indicating no increase in the incidence of polycystic ovaries

with valproate treatment.
Examining the relationship between ovulatory function and treatment
with AEDs, Morrell et al. (2002) conducted a study on 94 women with epi-
lepsy and 23 controls. The AEDs used were carbamazepine, valproate,
phenytoin, phenobarbitol, lamotrigine, or gabapentine as monotherapy for
6 months or more. There were no statistically significant differences in the
frequency of polycystic ovaries between any of the AED treatment groups.
However, women with epilepsy were more likely to be obese than the con-
trols, and obesity was higher in patients receiving valproate or lamotrigine.
The authors acknowledged, however, that they were unable to control for
previous use of AEDs prior to 6 months or for medication change and that
the numbers of participants in any of the AED groups were variable, be
-
cause the AED was selected based on the type of seizure.
A year later Morrell and colleagues (2003) published the results of an
open-label, cross-sectional multicenter study of 198 women under the age
of 35 years with a history of epilepsy who had been menstruating for at
least 4 years and had been treated with lamotrigine or valproate for at least
8 months but no more than 60 months and who were not taking oral con
-
traceptives (lamotrigine, n = 106; valproate, n = 92). Their results showed
that compared with lamotrigine monotherapy, valproate monotherapy was
associated with weight gain and higher androgen levels, hyperandrogenism,
and longer menstrual cycles that were less likely to be regular, suggesting
that these endocrine changes observed in some women using valproate for
epilepsy may be secondary to drug therapy.
There is a paucity of studies examining the long-term reproductive en
-
docrine health of young women with epilepsy during puberty. Little is
known about the long-term effects of the use of AEDs during childhood

and adolescence on reproductive endocrine health. Mikkonen et al. (2004)
Polycystic Ovary Syndrome 349
studied 69 patients and 51 control participants over a period of time. At
entry to the study, the age range was 8–18.5 years, and at follow-up, 12.5–
25.8 years. Initially 35 patients were taking valproate, 17, carbamazepine,
and 17, oxcarbazepine, as monotherapy. At follow-up only 42 of the 69 pa
-
tients were on medications. All the participants were examined clinically,
medical and menstrual histories were obtained, ovarian ultrasonography
was examined, and serum reproductive hormone concentrations were ana
-
lyzed. The results revealed no significant differences in laboratory or clini
-
cal findings between patients off medication and the controls. Postpubertal
patients still on medication had higher serum testosterone and androsten
-
edione levels than patients off medication, All the patients still on valproate
had elevated serum androstenedione levels. PCOS was more common in
38% of the patients on medication (63% on valproate, 25% on other
AEDs) than in patients off medication (6%) or in controls (11%; p =
.0005). The investigators concluded that epilepsy during pubertal matura
-
tion does not affect reproductive endocrine health in participants who dis
-
continue the medication before adulthood. However, an increased preva-
lence of endocrine disorders is detected if the patients remain on AEDs,
especially valproate, until adulthood.
In an attempt to resolve the controversy and difficulty in separating
menstrual and metabolic disturbances in women with epilepsy from effects
of the AEDs, Ferin et al. (2003) studied the endocrine and metabolic re-

sponse to long-term therapy with valproate on the normally cycling rhesus
monkey. They compared two groups of 7 monkeys in each group, one free
of medication and the other receiving valproate for a duration of 12.7–15.7
months. The results showed no difference in testosterone or LH levels be-
tween the two groups; both groups also had similar glucose and insulin re-
sponses to a glucose tolerance test. Examination of all 14 ovaries showed
no histological evidence of PCOS. The investigators concluded that their re
-
sults “did not support the hypothesis that treatment with valproate per se is
responsible for the induction of PCOS” (Ferin et al., 2003, p. 2915).
In contrast, Nelson-DeGrave et al. (2004) published a study reporting
that valproate potentiates androgen biosynthesis in human ovarian theca
cells. The theca cells were isolated from follicles of normal-cycling women.
The cells were treated for 72 hours with sodium valproate. Whereas low
doses (i.e., 30–300 ug) had no effect on basal and forskolin-stimulated pro
-
gesterone production, higher doses (1000–3000 ug) inhibited progesterone
production. The most pronounced effect of valproate on androgen biosyn
-
thesis was observed in the dose range of 300–3000 ug, which represents
therapeutic levels in the treatment of epilepsy and bipolar disorder. The
investigators reported that valproate increased both basal and forskolin-
stimulated P459c17 and P450scc protein levels, whereas the amount of
steroidogenic acute regulatory protein was unaffected. Consistent with the
ability of valproate to act as a histone deacetylase (HDAC) inhibitor in
350 OTHER ISSUES
other cell systems, valproate (500 µg) treatment was observed to increase
histone H3 acetylation and P450 17 alpha-hyroxylase mRNA accumula
-
tion. The HDAC inhibitor butyric acid (500 µM) similarly increased

histone H3 acetylation and DHE biosynthesis, whereas valproate derivative
valpromide (500 µM), which lacks HDAC inhibitory activity, has no effect
on histone acetylation of DHEA biosythesis. The authors of this study con
-
cluded that these findings suggest that valproate-induced ovarian androgen
biosynthesis results from changes in chromatin modifications (histone
acetylation) that augment transcription of steroidogenic genes. This is the
first study of its kind to provide biochemical evidence to support the role of
valproate in the genesis of PCOS-like symptoms and to establish a direct
link between valproate and increased ovarian androgen biosynthesis.
On the other hand, animal experiments have confirmed that in pa
-
tients with unilateral amygdaloid seizures, the amygdala ipsilaterally acti
-
vates neurons in the medial preoptic area, ventrolateral part of the
ventromedial area, and premammillary nuclei of the hypothalamus, areas
that are specifically involved in reproductive neuroendocrine function
(Silveira et al., 2000). Furthermore, induction of seizures in the amygdala
leads to a decrease in ipsilateral GnRH neuron fiber number (Friedman et
al., 2002). These findings further support the hypothesis that unilateral
limbic seizures may modulate reproductive endocrine function in a laterally
asymmetric manner.
In Patients with Bipolar Disorder
Several investigators have also examined the occurrence of PCOS in pa-
tients with bipolar disorder (See Table 17.1). As is the case with patients
with epilepsy, with whom there is controversy about whether PCOS occurs
independent of the treatment with AEDs, so is the case in patients with bi
-
polar disorders, with whom AEDs are often used as mood stabilizers to
treat bipolar disorder. Rasgon et al. (2000) conducted a pilot study to de

-
termine whether PCOS is associated with valproate use in the treatment of
bipolar disorder. The study evaluated the clinical and hormonal character
-
istics of PCOS in 22 women with DSM-IV diagnoses of bipolar disorder
between the ages of 18 and 45 years. None of the patients met the NIH cri
-
teria of PCOS at the beginning of the study. Ten patients were receiving
lithium monotherapy, another 10, valproate monotherapy, and 2, lithium-
valproate combination therapy. Patients had a mean exposure of 5 years to
lithium, 3 years to valproate, and 1 year to combination therapy.
All patients on lithium monotherapy or combination therapy and 60%
of patients on valproate monotherapy reported menstrual disturbances pre
-
ceding the start of medication, again suggesting that, as in the case of pa
-
tients with epilepsy, some women with bipolar disorder may have compro
-
mised HPG axis independent of therapeutic agents used. There were no
Polycystic Ovary Syndrome 351
significant differences in BMI or hirsutism, and hormone levels were within
normative values in the three treatment groups. Ovarian ultrasound re
-
vealed more follicles in one patient receiving lithium monotherapy and
none in the patients on valproate monotherapy. The authors concluded that
there was no significant association between PCOS and valproate or lith
-
ium therapy in women with bipolar disorder (Rasgon et al., 2000).
Another study in women with bipolar disorder, conducted by O’Donovan
et al. (2002), studied 32 women who were either receiving valproate ther

-
apy (n = 17) or no drug therapy (n = 15) for bipolar mood disorder. The
control group consisted of 22 women who had never been diagnosed with
or treated for a psychiatric illness. At the start of treatment women receiv
-
ing valproate had a significantly greater rate of menstrual abnormalities
(47%) than women receiving no drug therapy (13%) and the control
women (0%). Because the information about menstrual irregularities relied
on a mailed questionnaire, there may have been some recall bias, and hence
the authors suggested that the interpretation of data must be done with
caution. Seven of the 8 women receiving valproate currently had menstrual
problems, 2 had high BMI, and 5 had hirsutism. Two of the latter reported
that the hirsutism began after valproate treatment was initiated. None of
the women in the study had abnormal FSH or dehydro-epiandrosterone
levels, although all the women had hyperandrogenism, 4 women had ele-
vated LH/FSH ratio, and 5 women had polycystic ovaries, as determined by
ultrasound. The authors concluded that these 7 women had clinical PCOS
and estimated that PCOS was present in 41% of all valproate-treated
women in the study. However, the authors did not assess the clinical char-
acteristics of PCOS in all women in the study, making it difficult to justify
their claim of a higher prevalence of PCOS in women receiving valproate
than in women receiving no drug treatment or normal women. In looking
at obesity in the three groups (BMI > 25), the rates were 43% in the
valproate group, 57% in the nonmedication group, and 46% in the control
group. There was no statistical difference between the three groups. The
authors concluded that “this finding argues against a causal link between
valproate, obesity and the presence of PCOS features” (O’Donovan et al.,
2002, p. 328).
McIntyre et al. (2003) conducted a cross-sectional study of 38 women
with bipolar disorder who had been receiving valproate or lithium mono

-
therapy for at least 2 years. Metabolic, hormonal, and reproductive effects
of treatment were assessed. They measured FSH, LH, and SHBG, which are
often abnormal in women with PCOS. Among the valproate-treated pa
-
tients, menstrual irregularities were reported by 50% of the patients com
-
pared with 15% of the lithium-treated patients, and they were more preva
-
lent in obese patients than in lean patients. The lipid profiles were within
reference ranges for both groups. Free testosterone and androstenedione
levels were significantly higher in the valproate-treated patients, and LH
352 OTHER ISSUES
was elevated in both groups. The data suggested that valproate may have
an adverse impact and result in some aspects of the metabolic syndrome in
some women with bipolar disorder. However, among the limitations of this
study were the relatively small sample size, the cross-sectional design, and
the lack of a control group.
Insulin resistance and hyperinsulinemia are hallmarks of the metabolic
disorder and have been associated with the development of polycystic ova
-
ries and hyperandrogenism (Glueck et al., 2003). In yet another study,
Rasgon et al. (2002) presented data that assessed the association of insulin
resistance and bipolar disorder in 39 women who were receiving lithium,
valproate, other anticonvulsants, or antipsychotics. Seventeen women had
BMIs over 27, and increased insulin resistance occurred in 19 of them. The
distribution of insulin resistance was not associated with the type of mood
stabilizer used. However, the authors did not report on the type of medica
-
tion being taken by the women who showed insulin resistance. The authors

concluded that their data suggest that insulin resistance is common among
women with bipolar disorder, regardless of treatment.
Thirty women with bipolar disorder who were receiving valproate
monotherapy, lithium monotherapy, or valproate-lithium combination ther-
apy were compared with 15 women with idiopathic generalized epilepsy
(IGE) receiving valproate monotherapy by Akdeniz et al. (2003). The mean
duration of treatment in the valproate monotherapy group was less than 3
years in bipolar women and less than 5 years in women with IGE. No men-
strual abnormalities were reported in the lithium group, which had had no
previous exposure to valproate. In contrast, menstrual disturbances were
reported in 20% of bipolar patients on valproate therapy and in 46.7% of
patients with IGE on valproate therapy. Hirsutism was reported in 1 pa-
tient with bipolar disorder treated with lithium and in 4 patients with IGE
receiving valproate. Total testosterone levels were significantly higher in pa
-
tients with IGE receiving valproate compared with patients with bipolar
disorder receiving lithium, and the LH/FSH ratio was significantly elevated
in patients with IGE receiving valproate. There were no significant differ
-
ences in the BMI, waist/hip ratio, and fasting lipid levels among the three
study groups. However, fasting glucose was significantly elevated in pa
-
tients with bipolar disorder and IGE receiving valproate. Other characteris
-
tics of PCOS, including hirsutism and an elevated LH/FSH ratio, were more
prevalent in women with IGE receiving valproate than in women with bi
-
polar disorder receiving valproate. Again, the results of this study suggest
that valproate is more likely to be associated with certain symptoms of
PCOS than is lithium. Furthermore, menstrual disorders in women with

IGE receiving valproate had a higher incidence compared with women with
bipolar disorder also receiving valproate. This suggests that perhaps neuro
-
logical disorders may differentially contribute to the development of PCOS
symptoms, independent of treatment.
Polycystic Ovary Syndrome 353
354
TABLE 17.1. Studies of Women with Bipolar Disorder
Study Population Type
Number of
participants
Drugs Results
Rasgon et al.
(2000)
BP OL 22 10: Li (5 years)
10: VPA (3 years)
2: Li + VPA (1 year)
- BMI & hirsutism: no significant difference in three
groups
- Hormone levels WNL’s in three groups
- Increased ovarian follicles in 1 patient on Li & none on
VPA monotherapy
O’Donovan
et al. (2002)
BP CG 32 active;
22 control
17/32: VPA
15/32: no drug
- No difference in obesity between three groups
- No causal link between VPA, obesity, & presence of

PCOS features
Rasgon et al.
(2002)
BP OL 39 Li, VPA, other
AC&AP
- 17 had BMI > 27 - 19 had increased insulin resistance
- No difference between drugs
McIntyre et al.
(2003)
BP CS 38 VPA or Li Menstrual irregularities:
- 50% with VPA
- 15% with Li
Akdeniz et al.
(2003)
BP: 30
IGE: 15
OL 45 VPA, Li, or
combination
- VPA more likely to be associated with PCOS Sxs
(Symptoms) than Li
- IGE patients had higher incidence than BP patients
(46.7% vs. 20%)
Note. AC, anti-convulsants; AP, antipsychotics; BP, bipolar; CG, control group; CS, cross-sectional; IGE, idiopathic generalized epilepsy; Li, lithium; OL, open label; VPA,
valproic acid; WNL’s, within normal limits.
SUMMARY
The studies reviewed in this chapter present several separate viewpoints re
-
garding the association of AEDs (valproate in particular), epilepsy, bipolar
disorder, and PCOS. These are summarized in the following list.
•

AEDs such as valproate directly cause polycystic ovaries and/or
hyperandrogenism.
•
Although the evidence regarding AED-induced weight gain and its
effects on reproductive function is mixed, several studies concluded
that the weight gain from AEDs (in particular valproate) leads to in
-
sulin resistance, hyperandrogenism, and other reproductive abnor
-
malities of PCOS.
•
Based on the observation that there is overrepresentation of PCOS
in women with temporal lobe epilepsy and bipolar disorders, an al
-
ternative hypothesis is that the underlying disorder (epilepsy or
mood disorder) independently leads to the development of PCOS.
• Women with epilepsy and bipolar disorder may react differently to
valproate treatment in terms of weight gain and thus be more or less
predisposed to developing the metabolic syndrome, hyperinsulin-
emia, polycystic ovaries, or hyperandrogenism.
• Alternatively, women with epilepsy may have an underlying repro-
ductive abnormality caused by factors independent of valproate
treatment that is aggravated by weight gain. However, because the
prevalence of polycystic ovaries and hyperandrogenism was not con-
sistently reported for lean or obese women treated with other AEDs
or for normal women, these studies did not demonstrate whether
obese control participants or obese women with epilepsy taking
other AEDs also had a higher or lower prevalence of polycystic ova
-
ries or hyperandrogenism.

CONCLUSIONS
As increasing numbers of patients with bipolar disorder are in long-term
treatment with valproate, there is an important need for further research
that clarifies the relationship between long-term administration of valpro
-
ate and other AEDs used as mood stabilizers and the potential development
of reproductive endocrinological abnormalities. There is need for increased
awareness among clinicians and patients of the unknown potential for
these worrisome side effects. Some authors have suggested that if clinicians
are concerned about the literature on PCOS and valproate, they can use
valproate with caution, especially in young women with family histories
suggestive of PCOS. Further, in women with epilepsy who are overweight
Polycystic Ovary Syndrome 355