The reproductive function of women with bipolar disorder (BD) is of increasing interest to clinicians and patients alike. Previous studies have suggested that some medications used to treat BD, particularly anti-epileptic drugs (AEDs) such as valproate (VPA), are associated with a higher incidence of menstrual abnormalities (MAs) and polycystic ovary syndrome (PCOS) [1-6], a poorly understood endocrine disorder characterized by chronic anovulation and hyperandrogenism . However, there has been some disagreement in the literature regarding this association [8-12]. In women with epilepsy treated with AEDs, it has been proposed that reproductive dysfunction may be influenced both by medications and the neuroendocrine effects of epilepsy itself [13, 14]. A similar question has been raised in the population of women with BD, given that the neuroendocrine systems are central to both reproductive function and mood disorders . In addition, individuals with BD are often treated with combinations of psychotropic medications including atypical antipsychotic agents (AAPs), a class of medication that has been associated with weight gain, central adiposity, and the development of insulin resistance and type 2 diabetes [16-18].
PCOS is one of the most common endocrine disorders in women, with an estimated incidence between 4% and 6% [19-21]. In addition to being one of the most common causes of anovulatory infertility, PCOS is associated with an increased risk for type 2 diabetes, impaired glucose utilization, and cardiovascular disease [22-24]. The standardized definition of PCOS has evolved over time and varied in the literature before the development of the standardized Rotterdam criteria (last revised in 2003), which proposed that PCOS be diagnosed based on fulfilling two of the following three criteria: oligomenorrhea or amenorrhea, clinical and/or biochemical hyperandrogenism, and polycystic ovaries . This has led to some disagreement as it has been argued that these criteria are not robust enough to support discerning clinical research [26-29]. For example, the Androgen Excess Society regards hyperandrogenism as necessary for a diagnosis, along with either anovulation or polycystic ovary morphology .
We have previously reported high rates of MAs in women with BD, around 40% of whom report a current or previous MA [31-33], similar to rates published by others [5, 34]. We have also reported cases of MA associated with VPA use in a cross-sectional study design , with length of VPA exposure being significantly associated with free testosterone levels and VPA use being associated with an increase in total testosterone over time in a longitudinal study setting . However, no studies to our knowledge have been able to compare these rates to a control population. Similarly, no studies have integrated an objective, longitudinal determination of ovulation simultaneously with self-report measures of menstrual cycle length. The present study aimed to bring together cross-sectional biochemical assessments, self-report questionnaire data regarding reproductive functioning, and longitudinal, prospective ovulation tracking in women with BD and compare these measures of reproductive function to those in a control group of healthy women with no psychiatric history.
- Top of page
The reproductive function of women with BD has increasingly become an area of interest, driven by the need to clarify the risks and benefits of medication regimens in order to improve treatment guidelines. However, much remains to be elucidated; for example, few studies describe even baseline levels of reproductive hormones or self-reported MAs in this population, and virtually no studies have compared reproductive indices in women with BD to those in active controls. In addition, it is poorly understood whether reproductive function may be impacted by the pathophysiology of BD itself, perhaps through modulations of the hypothalamic-pituitary-adrenal (HPA) or hypothalamic-pituitary-gonadal (HPG) axes, or through the potential effects of medications used to treat BD, such as VPA, Li, and increasingly AAPs. Pharmacological effects that have been implicated include the possible effect of VPA on testosterone and insulin resistance in the development of PCOS [1, 3-5] and the metabolic effects of AAPs, which may also mediate the development of anovulatory syndromes such as PCOS, although studies remain divided [8-12].
To our knowledge, this is the first published study to have examined reproductive function in women with BD in comparison to a control population. In this clinical sample, the self-reported rate of oligomenorrhea or amenorrhea was close to 40%, consistent with our previously reported rates in a longitudinal evaluation of reproductive function in 25 women with BD (which represented a different sample but used similar enrollment criteria)  as well as rates published by others [5, 34]. Also consistent with our 2005 findings, we found that rates of oligomenorrhea and amenorrhea did not differ between treatment groups, including those who had been treated for less than three months or not at all, and that many patients noted an MA that preceded the diagnosis, and therefore the treatment, of BD .
It is interesting to note that reported rates of MA in our current study did not necessarily correlate with anovulation as measured by three consecutive monthly luteal-phase progesterone levels, and did not reflect the relatively low rate of PCOS, at around 5%, which matches the normal population rate of 4–6% [20, 21], although few studies reflect the most updated Rotterdam criteria. The majority of patients (almost 90%) had at least two positive ovulations during the consecutive three months, and this did not differ between women with BD and controls, and did not differ significantly by MA status. Notably, only 57% of our participants with BD completed even one month of ovulation tracking, which reflects the feasibility constraints of tracking, liaising, and coordinating with the necessary precision in this population. Although the exact reasons why participants were unable to complete ovulation tracking were not assessed, many of the women with BD in our study were either students or worked part- or full time, presenting significant scheduling difficulties. However, this was similarly true for the majority of our control participants, of whom 92% completed at least one ovulation tracking cycle, which may reflect the unique motivations of the largely self-selected control population. Additionally, there may have been self-selection bias within the control population, given that recruitment efforts made reference to assessment of reproductive function, but the direction of this bias is unclear as reasons to participate in such a study could vary. Although the percentage of women with BD who obtained ovulation tracking data was low, we were able to retain most of these subjects over the three-month tracking period, with an attrition rate of 19%.
It is unclear why self-reported rates of MA – specifically, oligomenorrhea and amenorrhea – do not match up with objectively measured ovulation rates. This may reflect a perceived experience of variability in the menstrual cycle, with a retrospective reporting bias that appears to be similar both in women with BD and in controls, but does not necessarily reflect current ovulatory functioning. This finding points to a future need to assess closely for objective measures of reproductive functioning, such as biochemical ovulation tracking rather than relying on self-reported menstrual cycle questionnaires.
Almost 10% of women with BD self-reported a diagnosis of bulimia nervosa, with 5% reporting a diagnosis of anorexia nervosa. Given that the present study selected a control group without a psychiatric diagnosis, we cannot compare these two populations in our study; however, these data add to a sparse but growing literature on the prevalence of eating disorders in women with BD [38-40]. A recent study by McElroy et al.  reported a rate of binge eating disorder in women with BD of 9%, compared to 5% for bulimia nervosa and 3% for anorexia nervosa, with an overall rate of 14.3% for all eating disorders. In the present study, we did not specify between binge-eating disorder and bulimia, but our preliminary rates appear to be similar to these reported rates. They are also similar to rates reported in a subgroup analysis comparing men and women with BD in the Systematic Treatment Enhancement Program for Bipolar Disorder (STEP-BD) study, which found significantly higher rates of reported bulimia in women (12%) compared to men (2%) . Further exploration of the intersection of BD and eating disorders is needed, particularly in light of the growing consensus that certain eating disorders may share neurocircuitry with dysregulated impulse control .
Twenty-two percent of women with BD indicated that they had experienced a period of time during which they stopped having periods due to stress or exercise, versus 8% of controls, although this represents retrospective self-reported data. This incidence of what is sometimes called hypothalamic or central amenorrhea due to the surmised central role of HPA dysregulation  has not, to our knowledge, been reported in the BD population to date. However, these preliminary data suggest that it should be further investigated, particularly as we continue to seek to understand how the HPA and HPG axes function in the context of BD and its pharmacotherapy.
In fact, these rates of self-reported eating disorders and central amenorrhea may help to explain the high rates of reported past MA in the BD population. In particular, if the eating disorder or period of exercise or stress was in the past, resolution of the MA would occur with resolution of the disorder and this may explain why past MA was far more commonly reported than current MA, and even why current MAs were, at times, accompanied by normal ovulation tracking results. However, these aspects of the potential inter-relatedness of eating disorders, central amenorrhea, and BD deserve further careful longitudinal study for clarification.
Interestingly, DHEAS and 17-OH-progesterone were both significantly higher in controls compared to patients with BD. Increased DHEAS and 17-OH-progesterone values, as general indicators of adrenal function, have been associated with states of HPA and HPG dysregulation, such as PCOS , although rates of PCOS and MA were not higher in controls. However, given the higher reporting of past central amenorrhea in women with BD than in controls, the lower DHEAS and 17-OH-progresterone values could represent an adrenal ‘burn-out’ scenario in which continued dysregulation at the level of the HPA axis eventually leads to lower adrenal hormone production. This theory remains speculative at this point and further studies are needed to test it.
There has been considerable discussion in the literature regarding whether psychotropic medications, in particular those used as MSs, cause reproductive dysfunction or MAs such as PCOS. The use of VPA has been a particular point of interest, although the metabolic effects of AAPs have also caused growing concern. As reviewed by Joffe and Hayes , although many studies have shown an association between VPA use and MAs or biochemical abnormalities, not all have upheld this association. Similar to the results of our previous studies, we did not find any biochemical differences, including in free or total testosterone, between women receiving or not receiving VPA. However, in our previous studies we found that the duration of VPA use was positively correlated with free testosterone levels, something we could not corroborate in the current study. Notably, in the present study we found that a higher percentage of women currently taking an AAP indicated a past or present MA compared to those not on an AAP, suggesting a future area for research as little as been published regarding the use of AAPs and reproductive dysfunction.
Regarding our population sample, clinical and demographic characteristics were similar in many ways to those reported in other large, multi-site studies in the BD population , although only one of these large studies completed a subset analysis comparing data by gender . In the latter study, investigators found that the incidence of BD type II was higher in women compared to men (30% versus 16%, respectively), a finding that had been noted in two previous studies but had not reached statistical significance [48, 49]. Rates of BD NOS were not reported . Notably, our percentages of women diagnosed with BD type II and BD NOS were higher than those previously cited, at 39% and 23%, respectively. It is unclear why BD type II and BD NOS patients were more highly represented in our sample, although, given that women with BD generally self-selected to contact the study coordinator for enrollment in our study, it may reflect an increased drive toward study participation in these populations, for unknown reasons. It is important to note that our sample population did not include women with uncontrolled medical conditions or illicit substance use, so the results may not be generalizable to these populations. It is also interesting to note that the stability of our population is both a strength of the study, as it avoids the potential confounding effects of acute affective episodes (either depression or mania), and a weakness, in that we were unable to explore whether changes in our biochemical variables are reflected in acute affective states.
Although the present study was the first to compare reproductive function data in women with BD against a control group of healthy women, it had some limitations, particularly in the size and sampling of the control group, which could have contributed to the lack of differences between groups for some variables. At the same time, the present study included a fairly large sample of women with BD (n = 103). The reason for the unequal numbers of women with BD and control women is that the original proposal was designed with a 2:1 ratio of women with BD to healthy controls for three-month ovulation tracking data as part of the overall metabolic and reproductive aims of the study. Given that a high dropout was projected for women with BD, a much larger number of women with BD underwent baseline function. As noted earlier, the control population was solicited through community advertisements, which led to a self-selected group of women pursuing enrollment, and it is possible that these women represent a unique population with a vested interest in obtaining information about their reproductive functioning, thus posing potential difficulties in generalizability. However, the two groups were notably similar in terms of demographic information.
As in almost all studies of BD, a significant portion of patients with BD were receiving more than one psychotropic medication, and the absence of uniform treatment regimens represents a limitation of the results. Missing data and small cell sizes prevented analysis of some measures, such as AAP dosing and the use of AAPs known to cause prolactinemia. At the same time, the present study is a reflection of the real-life treatment of BD, as no single treatment approach provides adequate outcomes for many of the diverse patients with this illness. It would also be relevant and important to include a measure of trauma history, which was not included in our assessment, given the prevalence of trauma in this population and its potential impact on neuroendocrine axes.
In summary, women with BD exhibit a high rate of current and past MA, but this rate does not appear to differ from the rate in control populations. Further studies investigating MAs should take into consideration the fact that many subjects who subjectively reported oligomenorrhea or amenorrhea had two out of three, if not three out of three, positive ovulations during tracking, suggesting the need for more objective means of measuring menstrual functioning. Medications did not show associations with biochemical markers or objective evaluation of ovulation tracking; however, current AAP use was associated with a higher rate of current or past MA. History of and/or current incidence of central amenorrhea due to HPA dysregulation should be further investigated in this population.
- Top of page
PW has received grant/research support and/or has been a consultant and/or received lecture honoraria from Abbott Laboratories, AstraZeneca, Bristol-Myers Squibb, Cephalon, Eli Lilly & Co., GlaxoSmithKline, Janssen, Jazz Pharmaceuticals, Novartis, Organon, Otsuka, Pfizer, Repligen, Solvay, Valeant Pharmaceuticals, and Vanda Pharmaceuticals. TAK has received grant/research support and/or has been a consultant and/or received lecture honoraria from Abbott Laboratories, AstraZeneca, Bristol-Myers Squibb, Cephalon, Eli Lilly & Co., GlaxoSmithKline, Janssen, Jazz Pharmaceuticals, Johnson & Johnson, Novartis, Organon, Otsuka, Pfizer, Repligen, Solvay, Valeant Pharmaceuticals, and Vanda Pharmaceuticals. NLR has received grant/research support and/or has been a consultant from Bayer HealthCare, Bristol-Myers Squibb, Forest Laboratories, GlaxoSmithKline, and Wyeth-Ayerst. MFR-M, HAK, WM, and PGS do not have any financial disclosures to report.