Ovarian Hormones as a Source of Fluctuating Biological Vulnerability in Borderline Personality Disorder

Abstract

Purpose of Review:

To examine the potential role of ovarian hormones in biological vulnerability to borderline personality disorder (BPD). The review focuses primarily on research examining the menstrual cycle as a source of short-term lability of BPD symptom expression, while discussing the currently understudied possibility of ovarian hormone influence in the developmental course of BPD.

Findings:

Several patterns of menstrual cycle effects on BPD symptoms and relevant features in non-clinical samples have been observed in empirical studies. Most symptoms demonstrated patterns consistent with peri-menstrual exacerbation; however, timing varied between high and low arousal symptoms, potentially reflecting differing mechanisms. Symptoms are typically lowest around ovulation, with an exception for proactive aggression and some forms of impulsive behaviors.

Summary:

Preliminary evidence suggests ovarian hormones may exert strong effects on BPD symptom expression, and further research is warranted examining mechanisms and developing interventions. Recommendations for researchers and clinicians working with BPD are provided.

Introduction

Borderline personality disorder (BPD) is characterized by rapidly shifting emotional, interpersonal, and behavioral symptoms, including intense affect, identity disturbance, impulsive and self-destructive behavior, aggression, and chaotic relationships [1]. The lability of these symptoms makes them hard to predict and therefore more challenging to manage and treat. Identifying detectable factors contributing to symptom variability would improve the ability of researchers, patients, and providers to predict symptom changes and maximize effects of interventions. Cyclical biological systems with the potential to influence psychiatric symptoms may provide key insights into explaining components of both short-term lability of and longer-term developmental changes in BPD symptom expression. Female reproductive hormones are a prime candidate to consider as a biological source of variability, given the changes occurring both across the lifespan (puberty, pregnancy, menopause) and within the monthly menstrual cycle.

The menstrual cycle prototypically occurs over 28 days, divided into two sections (see Figure 1). The follicular phase (beginning at onset of menses, continuing through ovulation) is characterized by continuous low levels of progesterone (P4) combined with increasing levels of estradiol (E2), peaking just prior to ovulation. The luteal phase begins following ovulation and continuing through the day before onset of the following menses. In the week after ovulation, both E2 and P4 rise and peak, and then both fall in the week before menses. In conjunction with their effects on the reproductive system, both P4 and E2 exert direct effects on the central nervous system, including the majority of neural systems implicated in emotional disorders [2]. Although most females do not experience clinically significant changes in their mood and behavior across the menstrual cycle [3], a growing body of work indicates strong individual differences in sensitivity to normal hormone changes (“hormone sensitivity”) that cause some women to experience severe symptoms (or worsening of pre-existing symptoms) in the luteal phase and especially around the onset of menses [4]. Similar individual differences in hormone sensitivity have been identified in pregnancy [5] and the menopause transition [6–8], and likely also exist during puberty [9]; however, given the dearth of literature on BPD during these other reproductive transitions, the present review will focus on the menstrual cycle.

Figure 1.

Proposed Model of Menstrual Cycle Symptom Exacerbation of High and Low Arousal Symptoms in Borderline Personality Disorder

Relatively little work to date directly examines the role of female reproductive hormones in BPD; however, multiple components of the extant literature on BPD suggest the possibility that hormone sensitivity across the menstrual cycle and the reproductive lifespan may play a role in biological vulnerability to BPD symptom expression. Within female individuals, BPD symptoms tend to emerge in adolescence, peaking shortly after puberty at approximately age 15 years [10]. Experiences of childhood abuse are prevalent among individuals with BPD [11], and early stressful and invalidating experiences are theorized to be key components of BPD’s etiology [12]; childhood adversity and stressors also appear to exacerbate hormone sensitivity [13–16], suggesting the potential of heightened hormone sensitivity in females as a mechanism through which these early life experiences may contribute to the emotional vulnerability characteristic of BPD.

Of note, while the possibility of fluctuating biological vulnerability to BPD symptoms across the menstrual cycle may be useful for improving predictability and treatment of symptoms in females with BPD, they do not necessarily imply sex differences in the overall occurrence or severity of BPD. Findings on sex ratios in BPD are mixed, with some indicating higher rates of diagnosis in clinical settings in females [17], perhaps due to higher female treatment utilization rates [18], but most recent studies find similar rates of BPD in female and male participants [17, 19]. Within individuals meeting criteria for BPD, findings also vary, with some studies more symptomology overall in female patients [20], including greater affective instability, hostility, and emptiness than males [20–22], but another recent study found more impulsive and aggressive behavior in male patients, with similar rates of suicidal behavior across sexes but more lethal attempts for males [23]. In another sample, typical general population sex differences in aggressive behavior and suicidality were attenuated in BPD patients, due to higher than usual levels of female aggression and higher than usual levels of male suicidal ideation [20]. Given these mixed findings, it remains unclear exactly what sex differences may exist in BPD, perhaps in part due to the heterogenous nature of the disorder. A key methodological limitation is that most of these studies utilize cross-sectional diagnostic interviews and self-report measures. Modeling BPD symptom variability longitudinally may be essential to understand sex differences in symptom expression, given that if reproductive hormones differentially affect vulnerability, these effects might be observed only if females’ symptoms are measured during phases of hormone-related vulnerability. It is also possible, and worthy of further research, that similar within-person steroid mechanisms (e.g., sensitivity to random or stress-related testosterone changes) play a role in some hormone-sensitive males with BPD, but are simply less predictable due to lack of monthly cycling [24].

PME of Mood Disorders and Non-Clinical BPD features

Findings within the broader literature on premenstrual dysphoric disorder (PMDD) and premenstrual exacerbation (PME) of underlying psychiatric disorders also suggest potential relevance for BPD. The most common symptoms of PMDD and PME are increased anger and irritability and heightened sensitivity to rejection [25], both hallmarks of BPD [26, 27]. Furthermore, PME appears to be common in mood disorders broadly, occurring in rates as high as 60% of depressed women [28]. PME of bipolar disorder symptoms appears to occur at similar rates [29], and its presence predicts greater severity in course of illness.

Emerging work suggests a greater-than-average hormone sensitivity in those with BPD symptomology. Higher overall levels of emotion-related impulsivity (a core component of BPD [30, 31]) has been linked to increased severity of PME of mood symptoms in a clinical sample of women with clinically significant PMDD symptoms (N = 54) [32]. In a sample of undergraduate females, those with higher-than-average BPD features at baseline reported increased symptoms after starting oral contraceptives, compared to no changes or reductions in symptoms for those with lower BPD features taking contraceptives and control group participants of all levels of BPD features [33] (N =46). In another undergraduate sample, BPD features were at highest levels during cyclical steroid changes [33] (N = 57). In an undergraduate sample specifically recruited to include females with representation across a full range of BPD features (N = 40), those with generally higher levels of BPD features demonstrated peaks in their symptoms when their E2 levels were lower than average and P4 levels were higher than average, a pattern characteristic of the mid-luteal and perimenstrual phases [34].

PME of Symptoms in Clinical Samples with BPD

A small but emerging area of research specifically examines whether individuals diagnosed with BPD are also at elevated risk for PME and other forms of hormone sensitivity. One challenge in conducting this research is factoring in the role of psychotropic medication, given that selective-serotonin reuptake inhibitors (SSRIs) are both commonly prescribed to individuals with BPD and a first-line treatment for PMDD [35]. An early study found no evidence of elevated symptoms in the premenstrual relative to postmenstrual cycle phases in patients with BPD [36] (N = 14); however, participants were taking psychotropic medications, which could have reduced PME of symptoms in the study [36]. However, a recent study examining menstrual cycle effects on BPD symptom expression in an unmedicated, naturally cycling clinical sample [37] (N=15) found strong evidence for hormone sensitivity in the sample. In prospective daily assessments, several patterns of cycle-based symptom exacerbation emerged that were clinically significant for all but one participant. Of note, retrospective self-reports of at baseline indicated that the participants in the sample were not aware (i.e., did not report) that they were experiencing significant cyclical changes in their symptoms.

These results, while preliminary given the small sample, suggest that hormone sensitivity could play a role in the etiology and maintenance of BPD. Below, we highlight several key findings from the most recent study in this area [37], and note how these preliminary findings, along with extant research on hormones and related behaviors, point to hypotheses about the causes and consequences of hormone sensitivity in BPD.

High Arousal Symptoms

For females with BPD, preliminary work indicates that high arousal symptoms and those characterized by increased interpersonal reactivity—such as irritability, anger, reactive aggression, and rejection sensitivity—may rise in the luteal phase and peak in the perimenstrual phase, with lowest levels in the ovulatory phase [37] (see Figure 1). This pattern is generally consistent with findings from studies on PMDD demonstrating similarly timed elevations in symptoms of irritability and rejection sensitivity [25, 38], suggesting the possibility of a shared mechanism underlying PME of BPD and similar PMDD symptoms. The most plausible mechanism of these midluteal-onset, high-arousal symptoms is an altered plasticity of the GABA-A receptor that reverses the typical beneficial effects of GABAergic progesterone metabolites (e.g., allopregnanolone). This mechanism has been demonstrated in PMDD [39, 40] but has yet to be studied in midluteal-onset PME. Of note, while individuals with PMDD experience a drastic clearance of their symptoms to a non-clinical level during the follicular phase, individuals with BPD maintained significant, if lower, levels of symptoms across the cycle in the largest study to date [37]; therefore, it is unclear whether the pathophysiology of these midluteal-onset exacerbations is shared with PMDD. Of course, the potential for misdiagnosis in either direction exists, with either individuals with PMDD (i.e., those with luteal-phase-only symptoms) being misdiagnosed with BPD, or individuals with BPD (i.e., those with chronic symptoms, with or without PME) diagnosed with PMDD. Of note, a forthcoming study has identified temporal subtypes of PMDD characterized by early-onset (midluteal) and late-onset (premenstrual week) PMDD symptoms, and the former PMDD subgroup may be most likely to exhibit shared mechanisms with midluteal-onset PME of high arousal symptoms [41]. Experimental and longitudinal studies are needed to test possible mechanisms of shared luteal risk for high-arousal symptoms in various subtypes of PMDD and BPD.

Low Arousal Symptoms

Preliminary work suggests that lower-arousal symptoms of BPD, such as depression, shame, and hopelessness, while similarly at lowest points during ovulation, demonstrated PME with later onset and more prolonged elevation, rising in the perimenstrual phase and extending further into the follicular phase [37] (see Figure 1). This is inconsistent with the rapid follicular clearance of symptoms typically observed in PMDD with the onset of menses, suggesting the possibility of differing physiological mechanisms underlying shifts in these lower arousal symptoms among those with PME of BPD. Some have theorized that cycling individuals with BPD may also have greater sensitivity of serotonergic systems to E2 fluctuations, and it may be that this or other estrogen mechanisms underlie PME of low-arousal symptoms [34, 42]. This premenstrual-to-early-follicular phase pattern is also more similar to that observed in studies of cycle-based shifts in suicidality, a common symptom and criterion of BPD [43]. A greater proportion of suicide attempts and more lethal suicide attempts in cycling, at-risk individuals occur in the early follicular (i.e., menstrual) phase, when ovarian steroids are at lowest levels [44–46]. As noted above, a forthcoming study has identified temporal subtypes of PMDD characterized broadly by early-onset (midluteal) and late-onset (premenstrual week) symptoms; depression symptoms were generally characterized by a later onset than other symptoms, and the late-onset PMDD subgroup may be most likely to exhibit shared mechanisms with PME of low arousal symptoms [41]. Preliminary evidence from an experimental study suggest that E2 and P4 stabilization during typical hormone-withdrawal weeks of the cycle eliminates PME of suicidal ideation and planning around menses onset [47]. If increases in symptoms such as shame, depression, and hopelessness result from similar mechanisms, cyclical E2 or P4 stabilization may have promise as a treatment for menstrual cycle exacerbation of this broader class of symptoms in BPD as well. More research is needed to replicate these findings, to define whether it is E2 or P4 stabilization that may be most therapeutic and to determine the safest methods of administration. Finally, given that rumination has been found to prolong the duration of low-arousal symptoms in PMDD [32], it is possible that the higher trait levels of rumination and other negatively-valenced repetitive thought common in BPD [48–50] prolong PME of depression and other low-arousal symptoms.

Ovulatory Effects

Consistent with broader findings that for most cycling individuals, the ovulatory phase is characterized by less emotional and cognitive vulnerability [2, 51], the ovulatory phase was characterized by lowest levels of most BPD symptoms in the largest clinical study to date [37]. However, in contrast to findings for most other symptoms, individuals with BPD demonstrated highest levels of proactive aggression (aggression with the purpose to meet one’s needs [52]) during the ovulatory phase [53]. This ovulatory peak in proactive aggression aligns with non-clinical findings of ovulatory increases in rewarding and risky behavior, such as sex drive [54, 55], binge eating [56–58], binge drinking [59], and gambling behaviors [60], as well as ovulatory increases in assertive behavior [61]. Animal studies have demonstrated effects of E2 in upregulated dopaminergic reward processing systems [62–65], and cyclical increases of E2 in humans has been similarly shown to upregulate neural reward processing [66, 67]. For cycling individuals with BPD, the combined enhancement of cognitive resources [68], assertiveness, and reward drive at ovulation occurs within the context of negative cognitive biases that the world is hostile, dangerous, and untrustworthy [48]. This may be expressed as higher levels of proactive aggression in what is seen as the most viable way to achieve goals effectively. Further work should also examine whether the impulsive, appetitive behaviors common in BPD, such as substance abuse or risky sex [1], show similar ovulatory increases for individuals with the disorder.

Assessment of Menstrual Cycle Effects

Given that retrospective measures of cyclical symptom changes are prone to recall bias that produces a high rate of false positives, single-time-point self-report measures of premenstrual symptoms are not acceptable as evidence of such symptoms in research or for accurate clinical assessment. In order to conclude that a female has cyclical symptom change, symptoms must at least be measured at weekly (and preferably daily) intervals. In order to minimize false positives, repeated (daily) ratings must be used. The Carolina Premenstrual Assessment (C-PASS [69]) is a scoring system for two months of daily symptom ratings on the daily record of severity of problems [70] that can be used to measure both categorical diagnosis (of PME or PMDD) and dimensional degree of cyclical syptom change. A worksheet with detailed instructions, an excel macro, and a SAS macro are available for facilitating use of the C-PASS system [69].

Even for researchers studying BPD who are not primarily interested in ovarian hormones, it may be important to include or consider the menstrual cycle in research design and statistical models. For these purposes, menstrual cycle phase can be measured with acceptable reliability and validity using self-reported menstrual cycle start dates. Using cycle counting methods to establish menstrual cycle day (and categorical cycle phase) is an inexpensive and powerful way to understand or covary the effects of the cycle on a repeated outcome of interest in a longitudinal study. For cross-sectional studies, researchers may control for cycle phase or recruit cycling individuals to participate at the same cycle phase; power may be maximized if cycle phase is chosen where the potential symptoms of interest is most pronounced. In order to generate both cycle day and cycle phase variables for use in models, three dates are needed: the date of the observation, the date of the prior menses onset, and the date of the subsequent menses onset. Backward-counting from the day before menses onset (day −1) to day −15 is recommended for delineating the luteal phase and its sub-phases; forward-counting from the day of menses onset (day 1) to day 10 is recommended for delineating the follicular phase. This results in a “cycle day” variable that can be used to graph the impact of the entire cycle on the repeated outcome of interest. For more information about the many options for coding menstrual cycle phases from this cycle day variable in order to test hypotheses about the impact of cycle phases on an outcome of interest, see a recent methodological paper on this topic [71].

Clinical Interventions

Given the prevalence of PME in mood disorders and the emerging evidence for PME of BPD symptoms, it is reasonable to routinely screen all naturally-cycling patients for PME or PMDD. Evidence-based treatment of PMDD has been reviewed extensively elsewhere [72, 73].

While no study to date has attempted pharmacological intervention to address cyclical BPD symptom exacerbation, selective-serotonin reuptake inhibitors (SSRIs) administered either continuously or intermittently during following ovulation until menstruation have been demonstrated to reduce similar cycle effects especially on irritability in PMDD [35]. RCTs examining efficacy of these medications in reducing high arousal symptoms in the mid-to-late luteal phase among individuals with BPD are warranted. In addition, there is preliminary evidence that stabilization of perimenstrual hormones (prevention of withdrawal) reduces PME of many low-arousal symptoms in a transdiagnostic sample of women with suicidal ideation [47]. Therefore, it should also be a high priority to conduct experiments that replicate, extend, and refine such treatment models in order to understand and treat PME of these BPD symptoms.

Although certain oral contraceptives have demonstrated benefit relative to placebo for PMDD [74], the same oral contraceptives have been found to be ineffective as an adjunct to SSRI for women with PME of depression [75], and have been found to acutely increase symptoms among women high BPD features [33]; therefore, studies are needed to understand the risks vs. benefits of hormonal contraception in BPD, and to confirm a lack of psychiatric benefit in this population. In PMDD, the final lines of treatment involve suppression of ovarian function using GnRH agonist (with stable addback; [76]) or oophorectomy, both of which are effective in ending cycling and cyclical symptoms [77, 78]. Clinical trials are needed to determine whether such treatments are appropriate in severe cases of PME in BPD.

In addition, psychosocial interventions, such as cognitive-behavioral therapy (CBT), may have utility for individuals with PME but are understudied. A meta-analysis based on the limited work in this area suggests that CBT is currently only evidence-based for reducing impairment—and not symptom severity—in PMDD [79], but CBT has not been examined specifically as a treatment for PME. Nevertheless, given the substantial interpersonal impairment observed in many with PMDD and PME [25], reduction of impairment may be a worthwhile target for psychotherapy, particularly in conjunction with medical management of symptoms. Existing evidence-based interventions for BPD, such dialectical behavior therapy [80] that instruct patients in emotion regulation skills may be particularly promising for lessening the impact of PME, given findings that trait-level emotion regulation deficits are linked to more severe effects of the cycle on symptoms [32].

Conclusions

These findings suggest the importance of considering effects of the menstrual cycle—and ovarian hormones across the lifespan—when conducting research and practicing clinical work with individuals with BPD. Ovarian hormone effects on symptom expression may be common [37], and individuals may be unaware of these hormone-related changes in their symptoms. Further research is needed to clarify effects of cyclical changes in ovarian hormones on BPD symptom expression, given that work to date suggests the potential for large effects through multiple mechanisms. Additionally, very little is currently known about how reproductive hormone changes across the lifespan may confer risk, with the potential for puberty, pregnancy, and perimenopause in particular to increase biological vulnerability. Understanding the complex and potentially interactive effects of ovarian hormones on BPD symptoms is also important in order to provide more informed guidance to patients for decisions about hormonal birth control methods and, for transgender individuals, hormone replacement therapy.