Perimenopause and First-Onset Mood Disorders: A Closer Look

A narrative review of literature was performed using PsychInfo Ovid (Figure 1). The search included articles from 2005 to 2020 in the English language. The search terms were “menopause and mental disorders/or affective disorders/or anxiety disorders/or bipolar disorder.” The search produced 831 articles. Abstracts were screened, and duplicates were eliminated. Studies pertaining to first-onset or new-onset mood disorder were further analyzed. Literature was deemed eligible if it also provided background information on the hormonal implications of perimenopause and on other causes known to contribute to mood disturbances during the transition to menopause. Additional relevant articles were found by manual search of article reference lists from PubMed and Google Scholar.

As recognized by the DSM-5, individuals with premenstrual dysphoric disorder may experience worsening symptoms when approaching menopause. Although the DSM-5 statement highlights the well-recognized influence of hormones on mood, the influence of the menopausal transition is likely more complex. Those experiencing perimenopausal symptoms across a longer period (e.g., 27 months) may have increased risk for depression, which suggests that independent of personal history of mood disorders, components of perimenopausal physiology contribute to a disturbance in mood (14). The specific hormones involved in perimenopausal mood changes are numerous. Recent investigations have suggested the influence of cortisol, hypothalamic-pituitary-adrenal (HPA) axis and inflammation (15), and testosterone (16). The hormonal influence of estrogen and progesterone, which have been extensively studied, will be the main focus of this section.

From premenopausal levels of 5 ng/dL to 35 ng/dL, estrogen declines to 1.3 ng/dL postmenopause, and the transition can be marked by mood disturbance, especially for those with a lengthy transition, personal or family history of mental illness, poor sleep, certain racial backgrounds, perimenstrual mood disturbance, vasomotor symptoms (VMS), social stressors, or impaired health (14, 15). Perhaps more important than perimenopausal estrogen decrease is the perimenopausal variance of anovulatory cycle frequency with increased variability of follicle-stimulating hormone (FSH), estrogen, and luteal progesterone, which can last up to 5 years (15). From a functional neuroanatomy perspective, the menopausal transition may alter the activity of ventral limbic regions, including the medial temporal lobe, which may influence consolidation of negative emotional information, possibly because of the presence of estrogen receptors in these regions (17).

Within the medial temporal lobe, estrogen may be linked to brain-derived neurotrophic factors (BDNF) expressed in the hippocampus. This process appears to be more sensitive to estrogen in acutely ovariectomized versus chronic ovariectomized mice (14). Furthermore, ovariectomized mice have been found to increase serotonin production after estradiol administration. Research (18) has illustrated that this effect could be related to a downregulation of monoamine oxidase A (MAO-A), and/or targeted expression of the serotonin transporter. MAO-A, an enzyme found in the brain, is responsible for apoptosis and oxidative stress. Elevated MAO-A levels occur in major depressive disorder in the prefrontal cortex, and greater MAO-A activity has been found in cortical and subcortical regions in animals with depression (19). MAO-A levels increase in the brain after a decrease in estrogen, which can be seen significantly in perimenopause. MAO-A V_T has been shown to be increased by 34% in all regions of the brain among perimenopausal women compared with women of reproductive age, and by 16% compared with menopausal women (p<0.05) (19).

Preliminary findings suggest that estrogen receptor (ER) subtypes, ER alpha and ER beta, are expressed in serotonin (5-HT) neurons, including the dorsal raphe nucleus, which is a site of serotonin volume transmission. ER beta is thought to interact with serotonin through the serotonin transporter and to increase 5-HT2aR binding, with dysfunction thought to increase 5-HT1A in the amygdala (18). The effects of estrogen can thus affect the ventral neuroanatomical circuit of mood disturbance (15, 17).

Additionally, variability in progesterone and the downstream metabolite allopregnanolone (ALLO) from menstruation to perimenopause could imply variability in negative feedback on the HPA, as mediated by variance in ALLO action on gamma aminobutyric acid receptor subunit dynamics (15, 17). Exogenous estrogens may also play a role. A cross-sectional study (20) of women ages 45–55 has suggested that the concentration of urinary lignan phytoestrogen is inversely associated with depression as measured by the Patient Health Questionnaire–9. Each of these mechanisms in theory could contribute to a heightened sensitivity to stress during low or variable estrogen and/or progesterone states and to increased consolidation of fear-based stimuli. Additionally, links between estrogen and mood extend across the mood spectrum, potentially having implications for bipolar disorder and postpartum psychosis (21).

The association between perimenopause and mood disorders has been established (22). The complexity of this association is far from understood, but a myriad of causes can contribute to this correlation. Hormonal changes, inflammatory markers, diet, and structural brain changes have been studied as potential contributors.

Hormonal changes occur during perimenopause. Biological changes include “programmed aging” of the HPA, with changes in sex hormones playing a role in perimenopausal depression. The “estrogen withdrawal hypothesis” suggests that estrogen insufficiency leads to depression (22). The above section on the hormonal implications of perimenopause provided details on the role of hormones, including estrogen, that are central to perimenopausal mood. Androgens also have some effect on mood. Testosterone may have anxiolytic and antidepressant effects, and lower levels have been found among women with depression (23). Mitogen-activated protein kinase 1 (MKP-1) has also been associated with depression, with a study (23) showing increased MKP-1 among perimenopausal women with depression. Higher cortisol levels are also found in severe depression, especially in the morning and evening. A study (24) investigated the association of depression and cortisol levels at awakening, 6 p.m., and 9 p.m. among midlife women. Women with more depressive symptoms had higher levels of cortisol at 6 p.m., but the finding was not significant after adjustment for smoking and education. A flatter diurnal slope of salivary cortisol levels has been associated with depressive symptoms among perimenopausal women (p<0.05).

Inflammatory processes also play a role in mood disorders among perimenopausal women. As previously mentioned, MAO-A activity has been found to be increased in certain areas of the brain in depression (19). Another study (25) noted a trend of increasing tumor necrosis factor alpha (TNFα) and decreasing interleukin 6 (IL-6) among perimenopausal women with depression, compared with those without depression and regardless of hot flashes, although the findings were not significant (p=0.53 for TNFα and p=0.90 for IL-6). However, in the same study, hot flashes were strongly associated with perimenopausal depression (p<0.05). The relationship between peripheral neurotrophic, oxidative, and inflammatory markers in new-onset depression among women during the menopause transition has also been researched (26). It has been suggested that 3-nitrotyrosine (p<0.05) and BDNF (p<0.05) are more likely to be related to first-onset depression among midlife women. Serum BDNF levels were decreased 12 months prior to development of depression and stayed low throughout the time leading to the episode (26). A different study (23) showed no difference in BDNF between perimenopausal women and nonperimenopausal women. Furthermore, heat shock protein (HSP) 70 is a member of the superfamily of HSPs. These proteins are upregulated in cells undergoing stressful stimuli, including inflammation and oxidative stress. HSP70 levels are higher at baseline among perimenopausal women who develop first-onset depression compared with those who do not (p<0.05). Therefore, it is speculated that HSP may be associated with protecting against free radicals because of an increase in 3-nitrotyrosine (26).

Diet plays an important role in mood disorders (27). High levels of lipids are present in the central nervous system, which change in response to dietary fat content (28). For example, saturated fatty acids have been shown to affect mood regulation by impairing the number of brain circuits (28). A study (28) of dietary saturated fatty acid intake of perimenopausal women at baseline and 4 years later showed a positive association between depression and saturated fatty acid intake. Similarly, trans fatty acids are linked to proinflammatory processes and gut dysbiosis, which is thought to negatively affect mental health (29). Trans fatty acid intake among women with depressive symptoms has been found to be higher than that of women without these symptoms (29). In contrast, omega-3 polyunsaturated fatty acid (n-3 PUFA) has been shown to be protective against mood changes (27). That study (27) showed that n-3 PUFA intake was inversely associated with depressive symptoms among perimenopausal women in a dose-response inverse relationship. N-3 PUFA supplementation is suggested to help alleviate some of the mood symptoms associated with depression (27).

Recent studies (30) have suggested that menopause may be associated with structural changes in the brain. Gray matter shrinking has been shown in various areas of the brain, including the left putamen, right pallidum, right inferior parietal gyrus, right superior frontal gyrus (orbital part), and right postcentral gyrus of perimenopausal women compared with premenopausal women (p<0.05) (30). These structures have different functions that are affected in perimenopause, including autonomic involvement, regulations of hormones such as FSH and luteinizing hormone (LH), higher cognitive functions, motor control, and sexual functions. Several studies (30) have illustrated that estrogen therapy may slow gray matter loss. White matter changes may also occur during perimenopause (22). Studies (22) have reported decreased fractional anisotropy, which assesses white matter integrity of the brain, in depression. Perimenopausal women with subsyndromal symptoms of depression have shown decreased fractional anisotropy values in the insula and increased values in the thalamus and midbrain, compared with nonperimenopausal age-matched controls (p<0.05) (22). The Wang et al. study (22) has suggested that subsyndromal depressive symptoms are a predictive factor of future depression and has shown that the altered microstructures in the above-mentioned areas make perimenopausal women more susceptible to depression.

With its changing hormonal milieu, the transition to menopause has been associated with new onset of depressive symptoms and actual diagnosed depression among women with no history of the disorder (31). The literature on first-episode depression among perimenopausal women has identified many contributing factors, in addition to hormonal changes. Among women with no history of depression, depressive symptoms were four times more likely when a woman was in the menopausal transition compared with the premenopausal state, and an actual diagnosis of major depressive disorder was twice as likely (31).

Many prior risk factors contribute to first-onset depression during perimenopause. When assessed by Bromberger et al. (32), 15.8% of perimenopausal and menopausal women with no past major depressive disorder met criteria for first onset of an MDE. Women more likely to develop first onset of a MDE were found to have low role functioning that was due to physical health limitation, low social functioning, or history of anxiety. Furthermore, women with negative life events had significantly increased risk of developing a first MDE as they transitioned to perimenopause compared with women remaining in premenopause (odds ratio [OR]:2.4, 95% confidence interval [CI]: 1.2–4.6) (33). Medical comorbidities, including cardiovascular disease (hazard ratio [HR]: 1.35, CI: 1.15–1.57), cerebrovascular disease (HR: 1.77, CI: 1.52–2.07), and liver disease (HR: 1.19, CI: 1.03–1.36), have also shown increased contribution to the development of depression during perimenopause (34). Mild increased risk for depression has been observed among women who report VMS (OR: 2.2, CI: 1.1–4.2) (33). Symptomatic menopausal transition has been identified as an independent risk factor for new depression (34). When symptomatic women during the menopausal transition were compared with those without symptoms, 11.4% of the symptomatic women were diagnosed with a psychiatric disorder, compared with 5.6% of those without symptoms (p<0.05) (35). Major depressive disorder (risk ratio [RR]:  2.15, CI : 1.91–2.42) was the most common psychiatric disorder identified; however bipolar disorder (RR : 1.71, CI: 1.01–2.96), anxiety disorder (RR : 2.07, CI: 1.81–2.37), and sleep disorder (RR : 2.01, CI : 1.73–2.34) were also associated with symptomatic perimenopause (35). VMS should be given more attention among those who develop first-onset depression during the menopausal transition to guide treatment options (36). VMS frequently lead to sleep disturbance, which can be severe enough to negatively affect functioning and quality of life (33).

Sleep disturbances, such as insomnia, are also common during the menopausal transition (37). Perimenopausal women are 40% to 56% more likely to report sleep issues, compared with 31% of premenopausal women (37). Nighttime awakening has been significantly associated with late menopausal transition stage and early postmenopause, as well as with a history of depression, with increased severity of depressed mood resulting in increased difficulty in falling asleep (38). In a cross-sectional study (39) controlled for specific menopausal stage, worsening depressive symptoms were associated with difficulties in sleeping more than five times a week (OR: 4.22, CI: 1.53–11.64) and with experiencing insomnia once a week or more (OR: 3.01, CI: 2.02–4.49). Specifically among women with newly diagnosed psychiatric conditions, new sleep disorders have been significantly associated with symptomatic menopausal transition (35). Conflicting research, however, has indicated that increased poor sleep is not significantly associated with newly diagnosed depression among perimenopausal women (OR: 3.68; CI: 1.22–11.15) (31) and that poor sleep is not a predictor of new onset major depressive disorder (HR: 1.23, CI: 0.64–2.35) (32). Although there is a lack of research in this area, awareness of this topic continues to be important.

Moreover, it has been identified that certain social determinants of health are protective against depression. These protective factors are high education level, high income, nonsmoker status, and social support, whereas psychosocial risk factors include low income and education level, nulliparity, and lack of social support (40). Depression is multifactorial, and many risks need to be considered.

A woman’s estradiol variability around her own mean level has been found to be significantly greater among women reporting depressive symptoms (31). Freeman et al. (31) found that women were 9.33 times more likely to have had higher FSH levels and 4.47 times more likely to have had higher LH levels than before the diagnosis of depression, offering support for a salient role of estradiol levels during perimenopause. That study also found that hot flashes, high body mass index, chronic smoking, and premenstrual syndrome were significant risk factors for depressive symptoms. In a study (40) assessing the relationship between menopause and depressive symptoms among midlife women, 18.8% of the women were classified as having depression, with both premature menopause (OR: 1.45, CI: 1.07–1.97) and use of hormone therapy (HT) (OR: 1.21, CI: 1.02–1.44) identified as risk factors for increased depression. Decreased exposure to endogenous estrogen, as occurs in premature menopause, can contribute to onset of depression among this population. Establishing this correlation may shed light on estrogen therapy as a possible treatment. It is interesting that use of HT was associated with increased depression as mentioned above. Furthermore, when comparing perimenopausal women with a hysterectomy to women without a hysterectomy, risk of new onset depressive symptoms was higher among women with hysterectomy (for hysterectomy only, RR: 1.20; CI: 1.06–1.36, and for hysterectomy–bilateral oophorectomy, RR: 1.44; CI: 1.22–1.68) (41). The role of exogenous hormones (HT) and association with depressive symptoms also was compared among women with hysterectomy and ovarian conservation versus those with hysterectomy and bilateral oophorectomy. Women with ovarian conservation who were using HT had a higher risk of developing depressive symptoms than women with hysterectomy and ovarian conservation who did not use HT (RR: 1.57; CI: 1.31–1.88). Interestingly, women with hysterectomy and bilateral oophorectomy who used menopausal HT did experience increased risk (41).

By recognizing menopausal transition as an independent risk factor for new onset depression, we can provide focused screening and early intervention (31). Psychiatric symptoms can present before physical symptoms of perimenopause. We know there is an impact on the neurotransmitters in the central nervous system years before peripheral serum assays detect hormonal changes in menopause (42).

The association between perimenopause and first-onset of bipolar disorder has been demonstrated in a few studies; however, evidence remains low and mainly in the form of case reports. We mention these studies because the women in these studies had no previously diagnosed mood disorders before the onset of perimenopausal symptoms (43). It is understood that the transition to menopause can increase the risk of exacerbation of already existing psychiatric disorders, especially among those experiencing menopausal symptoms (35). A case report of a 46-year-old woman with a previous diagnosis of type 1 bipolar disorder showed an increase in frequency of hospitalizations and manic episodes following bilateral salpingo-oophorectomy for endometritis (44). Understanding how surgical or natural menopause plays a role in mood disorders is critical for proper psychiatric care. The authors of the case study (44) speculated that, like the postpartum period, surgically induced perimenopause may be a time of increased risk for onset or exacerbation of bipolar disorder, because of the rapid decline of estrogen and progesterone. Another case report (45) followed a 51-year-old woman with type 1 bipolar disorder posthysterectomy. Initially, the patient denied any mood concerns until 2 years postsurgery. Despite various psychopharmacological and lifestyle adjustments, complaints of mood lability, insomnia, and anhedonia remained. The patient subsequently experienced menopausal VMS, which were suggested as the underlying cause of her worsening moods. The patient was placed on HT by her general practitioner and the aforementioned symptoms subsided (45).

The actual effects of peri- and/or postmenopause on first-onset mood disorders are debatable. Although the correlation can seem distinct, it does not necessarily imply causation (11). A study (43) of midlife women diagnosed with major depressive disorder and no history of bipolar disorder divided the women into two subgroups on the basis of the presence or absence of symptomatic menopausal transition. In that study (43), symptomatic menopausal transition was associated with increased risk of developing bipolar disorder (HR: 1.14, CI: 1.07–1.23). The findings (43) highlighted that the effects of perimenopause can extend past already established depression and can contribute to a new diagnosis. However, the absence of appropriate hormonal follow-up has been a factor possibly negatively affecting the validity of studies (45). The risk of bipolar disorder in symptomatic transition to menopause was found to be increased when follow up was longer than 1 year. This finding suggests the importance of extended follow-up to assess the risk of bipolar disorder (35). Further research is needed on first-onset bipolar disorder during perimenopause and its relationship with other psychiatric disorders as well as with antidepressant and hormone use.

The relationship between first-onset depression and perimenopause highlights the importance of adequate treatment for women during this time. The Canadian Network for Mood and Anxiety Treatments (CANMAT) has established guidelines for the treatment of major depressive disorder among perimenopausal women, identifying desvenlafaxine and cognitive-behavioral therapy as first-line treatments. Second-line treatments include transdermal estradiol, citalopram, duloxetine, escitalopram, mirtazapine, quetiapine extended release, and venlafaxine extended release (46). Medication options need to be weighed with side effect profiles and other symptoms of perimenopause besides major depressive disorder, such as appetite, sleep, and VMS. The CANMAT guidelines (47) for bipolar disorder do not specifically comment on treatment for new-onset bipolar disorder during perimenopause, because of the paucity of research. Further research is needed for treatment of this specific population. Table 1 summarizes the effects of various interventions for depression (46, 48–53).

In a randomized placebo-controlled study (48), desvenlafaxine was found to be effective and safe in treating depressive symptoms of perimenopausal women (p<0.05). It is important to note that the study (48) permitted HT treatment among the patient population, although this only affected 6% of the sample. The study authors speculated that serotonin-norepinephrine reuptake inhibitors (SNRIs) may alleviate 5-HT dysregulation during perimenopause and can potentially explain why selective serotonin reuptake inhibitors (SSRIs) are not always effective. To date it appears that desvenlafaxine is considered a first-line treatment, because it is the only SNRI that has been studied with randomized, placebo-controlled trials for antidepressant efficacy in perimenopause.

Despite CANMAT listing estradiol as a second-line agent, a meta-analysis (54) found that bioidentical estrogen (mainly estradiol) had no benefit on depression among menopausal women. However, the study did not specifically isolate for perimenopausal women and included postmenopausal women and thus concluded that for mild-to-moderate depression in perimenopause, a clinical benefit of estradiol could not be excluded.

Antidepressant effects versus those of HT have been studied. In an open-label study (49), both escitalopram and estrogen plus progesterone (EPT) significantly decreased depressive symptoms among perimenopausal women; however, the decrease was greater among the escitalopram group (p<0.05). The efficacy of escitalopram versus EPT for the improvement of depressive symptoms, menopause-related symptoms, sleep, and quality of life among peri- and postmenopausal women was studied. Interestingly, hot flashes were also significantly decreased in both groups, but only escitalopram decreased other menopause-related symptoms, including psychological, physical, and other VMS. The improvement in other menopause-related symptoms may have resulted from the improvement in depressive symptoms. Women treated with escitalopram or EPT demonstrated significant improvement in sleep characteristics (p<0.05), with no significant differences between treatment groups (49).

Sudden estradiol withdrawal, and its role in depressive symptom precipitation among women with and without a history of perimenopausal depression, has been studied (50). Participants were given 3 weeks of estradiol treatment and were then randomly assigned to receive placebo or continued estradiol (50). Interestingly, none of the women reported any depressive symptoms during the open-label phase of estradiol use; however, women with a history of perimenopausal depression who were subsequently assigned to the placebo group, experienced significantly increased depressive symptoms (P < 0.05). Women with past perimenopausal depression who continued with estradiol therapy, along with women in the control group who had no history of depression, had no change in their depressive symptoms. It is important to note that among the perimenopausal women with a history of depression, only 4 of 26 women had depression before perimenopause, and the remainder had first-onset depression at perimenopause. The study (50) highlighted the important relationship between perimenopausal depression and estrogen treatment and how caution regarding depression relapse should be exercised when discontinuing HT among women who have responded to estrogen therapy.

Prevention of depressive symptoms among perimenopausal and early postmenopausal women, rather than treatment of an established diagnosis, is also crucial. Treatment with transdermal estradiol plus intermittent micronized progesterone was found to prevent development of depressive symptoms among perimenopausal and early postmenopausal women who were euthymic at the start of the study, with 17.3% of the women taking treatment versus 32.3% of the women taking placebo developing significant depressive symptoms (OR: 2.5; CI: 1.1–5.7) (51). The study (51) included women with a history of depression, although this history was not associated specifically with benefits of the treatment on mood, suggesting that women with first-onset depression during perimenopause may benefit more. Transdermal estradiol and intermittent progesterone may be beneficial for perimenopausal women and for those with a history of stressful life events. By reducing estradiol variability, estradiol and progesterone can exert a prophylactic effect on depression among women with previous stressful life events (51).

In addition to HT acting as monotherapy or adjunctive therapy, research (52) has illustrated that nontraditional medications can be useful adjuncts. Tibilone is classified as a selective tissue estrogenic activity regulator and has estrogen-, androgen-, and progesterone-activating properties. The addition of tibilone to traditional SSRI or SNRI treatment resulted in less depressive symptoms compared with placebo among perimenopausal women with new-onset and recurrent cases of major depressive disorder, with no significant side effects (p<0.05). Vortioxetine, a newer antidepressant demonstrating multimodal serotonin activity, significantly decreased depressive symptoms among women with perimenopausal major depressive disorder (p<0.05) (53). Unfortunately, the study did not comment on first-episode versus recurrent major depressive disorder. Overall, treatment modalities vary for this population, and clear recommendations for specific first-onset mood disorders have not been defined, although these conditions appear to respond to established methods, with adjunctive treatments remaining as options.

Sleep difficulties during the menopausal transition can be multifactorial and, with specific guidelines lacking, treatment may vary on an individual basis. The reasons for sleep disturbances need to be investigated. Treatment options include HT, SSRI and/or SNRIs, and behavioral interventions such as cognitive-behavioral therapy, both as monotherapy and in combinations (37).

The research has indicated a relationship between perimenopause and first-onset mood disorders. In addition to variability of hormones, such as estrogen and progesterone, inflammatory markers, diet, and structural changes in the brain have been studied as potential contributors. Major depressive disorder can be attributed to the changing hormonal milieu during this vulnerable period among women with no history of previous psychiatric illness. However, other factors, such as VMS, prior negative life events, and socioeconomic status need to be considered. These associations were much stronger for first-onset depression; research on first-onset bipolar disorder among midlife women is lacking. Treatment options for this population need to consider psychosocial stressors and other perimenopausal symptoms, such as VMS and sleep disturbance, when establishing the best option for the patient in terms of antidepressant, HT, and psychotherapy use. Certain patients may merit adjunctive treatment and stronger focus on psychosocial factors and diet. Further research is needed in identifying a clear association between perimenopause and first-onset mood disorders, specifically bipolar disorder, because research in this area is limited. However, relationships have been established and remain multifactorial.