Findings of 14 studies (4, 13, 15–26) included five hypotheses regarding the relationship between hormones and the presence of PPD (Table 1): (H-I) changes in dopamine receptors, (H-II) increase in cortisol, (H-III) changes in estrogen levels, (H-IV) presence of thyroid dysfunction, and (H-V) decrease in oxytocin.

An increase in dopaminergic response is associated with the presence of psychosis and schizophrenia in men and women (27). This first hypothesis is explored in three studies (15–17) with the application of subcutaneous doses of the dopamine agonist apomorphine and the measurement of its effect on postpartum growth hormone (GH). Two studies analyzed whether the secretion of GH in greater quantities was related to effective binding of apomorphine in the dopaminergic receptors in the perinatal period. The first study (15) evaluated pregnant women: 29 at risk of postpartum psychosis (RPP) and a control group of 47. Fifteen participants from each group were selected for hormone testing and a dose of apomorphine on the fourth day postpartum. Eight of the 15 with RPP presented a relapse and a greater response of GH to apomorphine than the control group and the other seven women with RPP, suggesting that PPD is associated with greater sensitivity of the hypothalamic receptors and possible other regions of the brain.

A second study (16) evaluated ten mothers with high RPP (histories of manic psychosis) and compared them to mothers not at risk. They received three doses of apomorphine in the postpartum, and their progesterone, estradiol, and prolactin were measured. In contrast to the first study (15), no marker of GH response or any hormone was identified for the development of PPD. A third study (17) explored the role of the hypothalamic dopaminergic receptors by evaluating the GH response to apomorphine in the follicular and luteal phases of the menstrual cycle, comparing the effects in eight women who had recovered from postpartum bipolar psychotic episodes with those in nine women not at risk for PPD. This study identified a significant increase in the area under the curve of GH concentration after luteal phase administration of apomorphine in women with bipolar psychoses, suggesting a possible biological marker for the disorder, but one that must be considered with caution given the small sample size.

The second hypothesis posits the influence of cortisol in the appearance of psychosis (11) with a stress-vulnerability model analyzed by three studies (18–20). Buckwalter et al. (18) studied the relationship of mood, cortisol, and various hormones in 19 healthy women in pregnancy and postpartum. During pregnancy, dehydroepiandrosterone (DHEA), progesterone and testosterone were correlated with psychotic symptoms but not blood cortisol, though this was correlated with other aspects of mood. After childbirth, neither estradiol, progesterone, nor cortisol were correlated with any of the tests of mood. The study concluded that steroid hormones appear to play a role in psychotic symptoms during and after pregnancy, although it did not include participants with diagnosed mental disorders.

Aas et al. (19) studied the effect of cortisol and immunological markers associated with psychotic episodes in 14 women with RPP and with postpartum psychosis (PP), as compared with 26 healthy women. They found that women with PP had higher levels of cortisol 30 min after waking in the morning than those in the control group. The level of high-sensitivity C-reactive protein (hsCRP) was greater in women with PP, and they presented hypothalamic-hypophyseal-adrenal hyperactivity associated with a greater frequency of recent stress events, greater perceived stress, higher diurnal levels of cortisol and hsCRP than the control group. Hazelgrove et al. (20) had similar findings with a group of 51 women with RPP, of whom 22 had relapses. They measured cortisol in saliva, inflammatory blood markers, and psychosocial stress (severe child abuse and stressful life events). The group with PP relapses showed diurnal cortisol levels greater than those in the healthy group, and there were no differences between groups in inflammatory markers (IL-1β, IL-2, IL-6, IL-8, TNFα, VEGF, EGF, and hsCRP). Severe child abuse and high cortisol level in the third trimester of pregnancy predicted relapse at 4 weeks postpartum in women with RPP, after adjustment for clinical and sociodemographic covariables. Both studies (19, 20) supported the hypothesis of a relationship between elevated cortisol levels and the appearance of PP, and their association of stressful life events during postpartum also allows for direct inferences between variations in cortisol and RPP.

The third hypothesis proposes a relationship between variations in estrogen levels and the appearance of PPD, with six studies presenting contrasting evidence (13, 18, 21–24). Estradiol was not associated with postpartum psychotic symptoms in healthy mothers (18). Another study (21) examined the frequency of previously reported (28) four genetic variations of the estrogen receptor alpha (ERa) implicated in the appearance of PP. A mapping of 231 persons with bipolar disorder included 112 mothers who had a postpartum psychotic episode and 50 healthy mothers. The results showed no evidence that a variation in ERa was involved in the etiology of bipolar disorder or in the precipitation of perinatal psychotic bipolar episodes. Three studies analyzed the effect of estrogen treatment in preventing or reducing PPD episodes (13, 22, 23). One study (22) evaluated whether 4-week sublingual estradiol treatment reduced symptoms of depression and psychosis in ten women with postpartum depression and four with PP, and whether it affected levels of allopregnanolone and progesterone. Contrary to expectations, levels of allopregnanolone were reduced during treatment. None of the steroids investigated was associated with psychiatric symptoms before or after the women's mood was stabilized. A second study (13) presented evidence of a satisfactory response to treatment with sublingual 17β-estradiol in 10 women with PP, in whom estrogen deficiency was documented before treatment. The increase in blood estradiol levels in the first week of treatment was accompanied by a rapid and significant decrease in psychiatric symptoms, which practically disappeared in week 2, except for one woman who discontinued treatment. A third study (23) reported that treatment with transdermal 17β-estradiol at three different doses was not sufficient to achieve a prophylactic or therapeutic effect in 29 pregnant women with a history of psychosis and a high risk of relapse in postpartum, 12 of whom relapsed despite starting treatment within 48 h of birth. More recently, Khedr et al. (24) identified significantly lower levels of estrogen in 60 women with PP. The estrogen levels correlated negatively with the total score on the Psychiatric Rating Scale.

The fourth hypothesis relates to the influence of thyroid hormones on the appearance of PP and is reported in three studies (4, 24, 25). Khedr et al. (24) besides estrogen, also evaluated progesterone, and the thyroid hormones T3, T4, and TSH as factors in the appearance of PP in 60 women at 4 weeks postpartum. They found a greater proportion of thyroid dysfunction (18%) and the presence of transient thyroid dysfunction in women with PP. Bergink et al. (25) analyzed thyroid-stimulating hormone (TSH) and free thyroxine in 31 primiparous women with PP, comparing them with a control group of 177 healthy women. At 4 weeks postpartum they found that 19% of the women with PP developed autoimmune thyroid disease (AITD), vs. 5% of the control group, suggesting that this disease could be a risk factor for developing PP. A previous study (4) evaluated the thyroid hypothesis and the adrenal glands hypothesis (cortisol) about the etiology of PP. They recorded levels of sex hormones [prolactin, follicle-stimulating hormone, luteinizing hormone, and dehydroepiandrosterone (DHEA) sulfate] in the blood of 23 women with PP and 30 healthy postpartum women. Only the free T3 hormone showed significantly lower levels in the group with PP, but within normal levels in both groups. The authors suggest that there is a secondary thyroid effect in psychotic episodes, strengthening the hypothesis of a relationship between PP and thyroid alteration, vs. a possible causal association. These findings support the idea that psychotic episodes may interfere with the functioning of the hypothalamus-pituitary-thyroid axis, provoking a non-thyroid illness as a response to the underlying acute psychopathological process (4, 24).

The fifth and most recent hypothesis suggests a reduction in levels of oxytocin, in comparison to normal values in women with PPD, which was based on previous evidence from non-perinatal samples (29, 30). Only one study (26) met the criteria in our search. A non-perinatal study (29) found low levels of oxytocin in patients with a first psychotic episode, highlighting that increasing it can lessen alterations of mood and social cognition. Another study (30) reported that during menopause, lower levels of oxytocin in women with schizophrenia are associated with more severe psychiatric symptoms. However, the only study identified that focused on the perinatal period (26) did not support this hypothesis, which analyzed the expression of genes, proteins, oxytocin, arginine vasopressin (AVP), and their respective receptors in the placental tissue of 22 pregnant women after a first psychotic episode, in comparison with 20 healthy pregnant women. Those who had experienced the psychotic episode showed greater gene and protein expression of oxytocin and AVP and their receptors.

Only one study reported that hormones were related with cognition and perinatal psychotic symptoms in a healthy sample (18). Buckwalter et al. (18) measured serum levels of estradiol, progesterone, testosterone, cortisol, and DHEA, as well as cognitive processes, mood, and psychiatric symptoms in 19 healthy women 2 months before and after childbirth. During pregnancy, high levels of DHEA were associated with better visuospatial performance, verbal episodic memory, attention, better mood, and fewer psychotic symptoms. High cortisol was related to less perseverance in verbal learning. High progesterone correlated with psychotic symptoms and confusion. In the postpartum, high DHEA and cortisol were related with better cognitive performance. The decrease in DHEA after childbirth was associated with an increase in affective disorders and the increase of testosterone contributed to a worse mood, depression, and hostility. Cognitive deficits were greater during pregnancy regardless of mood, suggesting that extremely high levels of steroid hormones negatively affect cognition and mood, in addition to other perinatal factors. Although it has not been documented during pregnancy, cholinergic dysfunction would be consistent with the findings of this study.

Results from two studies located outside the initial search can indirectly suggest a relationship between hormones, cognition and PPD (31, 32) that requires further exploration. One study (31) in non-perinatal women with (n = 29) and without psychosis (n = 31) proposes that sex hormones have an activating effect on brain function during the menstrual cycle, and that circulating levels of estrogen in the brain influence cognition. Another study (32) identified that pregnant women who developed PP showed greater connectivity in the right dorsolateral prefrontal cortex and the ipsilateral middle temporal gyrus than the control group although significant differences between the groups were not identified.

The findings of the studies reviewed suggest the involvement of hormones in PPD and a relationship with cognitive deficits in healthy and non-perinatal samples, where exposure to psychosocial factors increases risk (19, 20). Interventions should seek improvement in mood and cognition to optimize decision making, learning, social cognition, and self-care, through promoting social support, sleep quality and strategies to reduce stress associated with affective and behavioral tasks related to the mother-child relationship.

There is evidence that during pregnancy a reduction in the gray matter of areas in the brain that involve social cognition relates with adaptive demands of motherhood (33). The findings are controversial as to whether pregnancy has negative consequences on brain physiology for middle-aged and older women by affecting the brain's response (plasticity or inflammation) (34). In postpartum depression, inflammation and oxidative stress can negatively affect neurons and promote degeneration (35). Therefore, cognitive assessments throughout the perinatal period could be useful for mothers at risk of PPD and severe perinatal mental health disorders.

The hypotheses analyzed provide some evidence on the effectiveness of hormonal treatments coupled with customary pharmaceutical approaches (antipsychotics, mood stabilizers), electroshock therapy (36–38), psychological and psychoeducational therapy, and family and institutional support (3). Nevertheless, caution is suggested since the evidence for them is not conclusive (36). The findings show a positive mood and cognitive impact of oxytocin (39, 40) and estradiol (13), but with emotional side effects (39).

Bolton (41) proposes the 4P model for care, which considers predisposing, precipitating, perpetuating, and protective factors where hormonal follow-up during the perinatal period in women at risk could help early identification of drastic changes related with the development of PPD and reduce its severity and/or relapses. Such follow-up would also allow for early intervention to minimize cognitive negative symptoms of the PPD and promote the affective and cognitive recovery of women with diverse clinical profiles of the perinatal psychotic disorders.