Introduction

Becoming a mother is usually regarded as one of life’s most emotionally positive and rewarding experiences. However, for a significant number of women, the postpartum period can instead be a difficult time accompanied by mental illness. Indeed, recent analyses indicate that at least 15% of new mothers worldwide each year are affected by postpartum depression (PPD) making it the most common complication of childbirth (Wisner et al. 2013). PPD is detrimental to maternal well-being and is one of the leading causes of maternal mortality resulting from suicide (Lindahl 2005; Osborne and Monk 2013). Further, PPD can compromise mother-infant interactions and as a result, negatively impact the development of the offspring (Grace et al. 2003; Letourneau et al. 2013; Verbeek et al. 2012; Hoffman et al. 2017) which carries significant economic and social long-term costs to society (Baurer et al. 2015).

Although PPD has been deemed a major public health concern due to its prevalence and the risks it poses to mothers and their children (Wisner et al. 2006; Chesney et al. 2014; Meaney 2018), our current understanding of the underlying neurobiology of PPD remains limited. However, interest in PPD has been growing (Fig. 1) with an increasing number of preclinical and clinical studies turning their attention to addressing this important issue. To date, findings from this emerging body of work have revealed that PPD is accompanied by dysregulation of mood-related neural circuits that have also been implicated in maternal caregiving (Pawluski et al. 2017). One such circuit is the mesolimbic reward system, which is the focus of this review (for other circuits, see reviews by Moses-Kolko et al. 2014; Duan et al. 2017; Pawluski et al. 2017). Specifically, we discuss normative changes in the mesolimbic system during motherhood in both rodents and humans and how these are impacted in PPD. We also consider modulation of mesolimbic dopamine (DA) by the hypothalamic neuropeptide oxytocin (OT) and how OT-DA interactions regulate mood and mothering during the postpartum period. Aside from providing an overview of reward mechanisms in PPD, we highlight areas where further research is necessary.

Fig. 1
figure 1

Representation of the number of studies published on postpartum depression (PPD) from 1960 to 2017. Numbers were generated using “postpartum depression” as a search term in PubMed. The red bar indicates when PPD was added to the DSM in 1994 as a form of major depression (MD). The gray bar represents when functional magnetic resonance imaging (fMRI) was first utilized to study the mesolimbic reward system in PPD (Silverman et al. 2007). In 2013 (green bar), a peripartum onset specifier was added to the DSM-V. As of 2017 (blue bar), there have only been approximately twenty fMRI studies completed on mothers suffering from PPD (Pawluski et al. 2017). Of these, nine showed reward system changes with PPD. Adapted from Li and Chou 2016

Full size image

Defining PPD

After birth, mild and transient disruptions in mood are normative and characteristic of “postpartum blues”. PPD however is a clinical condition that is more severe and which, if left untreated, can be long-lasting persisting for many months or even longer as some women with PPD continue to experience elevated levels of depressive symptoms years after childbirth (Vliegen et al. 2014; Netsi et al. 2018). PPD was formally recognized in 1994 when the fourth edition of the Diagnostic and Statistical Manual of Mental Health Disorders (DSM-IV) classified PPD as major depression (MD) with postpartum onset, defined as within four weeks after delivery. The diagnostic classification of PPD did not change until 2013 when the DSM-V introduced a peripartum onset specifier to account for antenatal onset of depression during pregnancy as well as after birth. Although generally regarded as an improvement, recent data suggests that peripartum depression beginning during pregnancy may be a distinguishable subtype from PPD that manifests postnatally as there are differences in their symptomology and severity (Altemus et al. 2012; Putnam et al. 2017). Further, many researchers and clinicians consider the four week postpartum onset specifier to be too conservative because depression that begins later than four weeks after delivery may still negatively impact mothers and their children (Murray et al. 2011; Verbeek et al. 2012; Letourneau et al. 2013; Stein et al. 2014; Hoffman et al. 2017; Meaney 2018). As such, in spite of current DSM-V guidelines, time frames that range up to one year postpartum are commonly used in research studies and clinical practice (Gaynes et al. 2005; Wisner et al. 2010; O’Hara and McCabe 2013).

PPD is characterized by low mood and sadness accompanied by anhedonia, impaired concentration, disrupted sleep and appetite, psychomotor disturbance, feelings of worthlessness or guilt, social withdrawal, and recurrent suicidal ideation (Meltzer-Brody et al. 2018). Because these symptoms mimic those of MD, whether PPD is a distinct disorder remains controversial (O’Hara and McCabe 2013; DiFloro and Metlzer-Brody 2015; Pawluski et al. 2017). Several features of PPD however do suggest a certain degree of distinctiveness. First, PPD occurs during a unique time physiologically when there are dramatic endocrine alterations involving steroid and peptide hormones (i.e., estrogens, progesterone, glucocorticoids, OT) as well as significant shifts in the immune profile (Robinson and Klein 2012; Schiller et al. 2015; Brummelte and Galea 2016). Second, PPD presents with greater comorbid anxiety with postpartum anxiety often preceding the onset of depression (Prenoveau et al. 2013; Wisner et al. 2013; Farr et al. 2014; Fox et al. 2018). Lastly, and perhaps most importantly, PPD strikes at a critical time when there is the added responsibility of caring for an infant. This can be challenging for depressed mothers who are more likely to partake in unhealthy feeding and sleep practices as compared to mothers without PPD (Field 2010). Along with such compromised caregiving activities, maternal depression can be damaging to mother-infant interactions. Thus, while non-depressed mothers exhibit positive, warm, and sensitive caregiving, those with PPD interact with their infant in a way that is either withdrawn, passive, and under-stimulating or intrusive, controlling, and over-stimulating (Malphurs et al. 1996). Depressed mothers also tend to be more irritable and hostile, less affectionate, and less sensitively attuned to their infants (Lovejoy et al. 2000). Maternal interactions in PPD are further characterized by reduced vocal and visual communication, less touch, and less smiling (Righetti-Veltema et al. 2003; Herrera et al. 2004; Granat et al. 2017) which likely contributes to difficulties bonding and disrupted synchrony (i.e., a mother’s capacity to coordinate her behavior with infant signals) (Feldman 2007). Given that these disturbances occur when attachment processes and the mother-infant relationship shape the cognitive, emotional, and social development of the offspring, the children of depressed mothers are at high risk for experiencing negative outcomes in these domains and these can extend beyond infancy into childhood and late adolescence (Murray et al. 2011; Verbeek et al. 2012; Letourneau et al. 2013; Stein et al. 2014; Hoffman et al. 2017; Meaney 2018). The detrimental effects of PPD on offspring have been well studied (Drury et al. 2016), but far less research has investigated the neurobiological sequelae of PPD in the mother.

Animal models of PPD

Animal models represent a valuable translational tool that have been widely used to investigate the neurobiology of psychiatric disorders, including PPD (Perani and Slattery 2014; Li and Chou 2016). One approach to modeling mental illness in rodents is to incorporate known biological, psychosocial, and/or other (i.e., environmental, genetic) risk factors. For PPD, endocrine events occurring during the perinatal period are considered to be among the biological factors that contribute to increased susceptibility in some women. These include alterations in the ovarian hormones, estrogen and progesterone, as well as dysregulation of the hypothalamic-pituitary-adrenal (HPA) axis (Glynn et al. 2013; Schiller et al. 2014). As such, there are models of PPD which employ peripartum-related hormonal manipulations involving either withdrawal from ovarian steroids (Galea et al. 2001; Green and Galea 2008; Suda et al. 2008; Navarre et al. 2010; Schiller et al. 2013; Baka et al. 2017) or administration of high dose corticosterone postpartum (Brummelte and Galea 2010) as well as other strategies to interfere with HPA axis function (Melón et al. 2018).

In addition to hormones, numerous psychosocial risk factors for PPD have been identified such as prior history of depression, low socioeconomic status, lack of social support/social isolation, history of trauma, and other adverse life events (Robertson et al. 2004; Milgrom et al. 2008; Yim et al. 2015; Biaggi et al. 2016). Common to all of these is the psychological experience of stress and thus, stress-based PPD models have also been developed. Such models apply various types of stressors (i.e., restraint; restraint and overcrowding or bright light; chronic variable stress), typically for 7–14 days, during pregnancy (Smith et al. 2004; Champagne and Meaney 2006; O’Mahony et al. 2006; Hillerer et al. 2011; Haim et al. 2014; Leuner et al. 2014, 2016; Vanmierlo et al. 2018) or during the postpartum period in the form of repeated maternal separation (Boccia et al. 2007) or social stress (Nephew and Bridges 2011). Recent attempts have also been made to develop models of maternal depression based on early life stress (Nephew et al. 2017a) and other factors, such as high-fat diet/obesity (Perani et al. 2015; Bolton et al. 2017), which increase risk for PPD. Each of these models induces one or more critical aspects of postpartum depressive-like symptomology including behavioral despair, anhedonia, anxiety-like behavior, and/or impaired maternal care which like the human condition can negatively impact offspring neurodevelopment (Smith et al. 2004; Champagne and Meaney 2006; Brummelte et al. 2006; Babb et al. 2014). Other rodent PPD models, like the Flinders Sensitive Line (FSL) of rats, do not rely on risk factors but instead these animals are bred for depressive-like behavior which for postpartum females is accompanied by deficits in maternal care and impaired maternal motivation (Lavi-Avnon et al. 2005, 2008). Lastly, different inbred mouse strains, which exhibit variations in their emotional and maternal phenotype, have been employed to investigate PPD (Avraham et al. 2017).

Like models for other complex psychiatric disorders, those for PPD are not necessarily intended to recapitulate all possible risk factors and the entire symptomology. Nonetheless, they can be used to study certain aspects of the disorder, particularly at cellular, neurochemical, and molecular levels of analyses that may not be as readily feasible in humans. Due to the conservation of major neural, neurotransmitter, and neuromodulatory systems between rodents and humans, it is expected that novel mechanistic insights gained from animal models of PPD will be relevant to human mothers.

The mesolimbic reward system

The brain undergoes dramatic alterations during pregnancy and the postpartum period that are essential for optimizing emotional well-being and caregiving abilities (Barrett and Fleming 2011; Rilling 2013; Kim et al. 2016; Swain and Ho 2017). Many brain regions and systems exhibit modifications, although there is considerable overlap in the neural circuits which regulate mood and various aspects of mothering (Pawluski et al. 2017). Thus, a possible way both depression and maternal disturbances might arise during the postpartum period would be if the normative, adaptive changes in these overlapping circuits failed to occur (Hillerer et al. 2012; Duan et al. 2017). Few studies have explicitly tested this hypothesis by performing direct comparisons of non-mothers to mothers with and without PPD. Even so, the available evidence points to key neural changes during motherhood and in PPD with both human and animal studies implicating the mesolimbic system (Pechtel et al. 2013; Moses-Kolko et al. 2014; Duan et al. 2017; Pawluski et al. 2017).

The mesolimbic system (Fig. 2) consists primarily of dopaminergic neurons in the ventral tegmental area (VTA) of the midbrain that project to the nucleus accumbens (NAc), part of the ventral striatum (Sesack and Grace 2010). The VTA-NAc circuit plays a well-established role in processing rewards and motivated behavior (Salamone et al. 2016). DA neurons in the VTA also innervate cortical areas, the amygdala, and hippocampus and in doing so link reward processes to cognitive and emotional function (Sesack and Grace 2010; Russo and Nestler 2013). The mesolimbic reward system has been a target in neurobiological investigations of PPD mechanisms based on known dopaminergic/reward changes which occur after parturition, the role of this system in healthy caregiving (Stolzenberg and Numan 2011; Moses-Kolko et al. 2014; Swain and Ho 2017) and also because of convergent findings of reward system dysfunction and diminished DA in MD (Dunlop and Nemeroff 2007; Russo and Nestler 2013; Admon and Pizzagalli 2015).

The reward system in PPD: humans

The neural correlates of PPD have been investigated using functional magnetic resonance imaging (fMRI) approaches. The first fMRI study of PPD was done by Silverman et al. in 2007 and since that time, approximately twenty others have been published (Fiorelli et al. 2015; Duan et al. 2017; Pawluski et al. 2017). Although not all of these were designed to assess the reward system, nine studies to date have identified mesolimbic dysfunction in depressed mothers (Table 1). Consistent with the anhedonia features of PPD, some of this work has provided evidence for a blunted striatal response to positive, rewarding stimuli. For example, positive words (Silverman et al. 2007) were shown to elicit less activation of the striatum in mothers with PPD relative to healthy mothers. Similarly, in response to positive faces, mothers with higher levels of depressive symptoms exhibit reduced striatal activity (Morgan et al. 2017). In other work using a monetary reward task, Moses-Kolko et al. (2011) found that although initial activation of the ventral striatum was similar in depressed and healthy mothers, depressed mothers’ responses rapidly attenuated to baseline while healthy mothers had a sustained response to reward. This rapid attenuation suggests blunted reward function which could contribute to decreased motivation in PPD. However, this may not be the case for all mothers with PPD as a more recent study found that decreased reward-related ventral striatal activity does not generalize to young mothers with less severe depressive symptoms (Moses-Kolko et al. 2016).

Table 1 Findings from functional magnetic resonance imaging (fMRI) and positron emission tomography (PET) studies on the mesolimbic reward system in PPD
Full size table

Another strategy to evaluate the reward system in PPD has been to examine neural responses to positive infant cues—a more motivationally relevant stimulus for mothers. A number of postpartum neuroimaging studies have shown that healthy mothers display increased activation to the smiling face of their own infant as compared to an unknown infant in reward areas including the VTA, ventral striatum, as well as the orbitofrontal cortex (OFC) (Bartels and Zeki 2004; Nitschke et al. 2004; Strathearn et al. 2008; Rilling 2013). In contrast, mothers meeting the diagnostic criteria for PPD with ongoing high depressive symptoms were shown to have reduced responses to their own infant’s joy faces in striatal areas and the OFC (Laurent and Ablow 2013). Depressed mothers have difficulty identifying happy affect in their infant’s facial expression which may be a contributing factor (Arteche et al. 2011). Accordingly, mothers struggling with PPD may be less able to respond to their infant’s joy because they experience it as less rewarding which could underlie diminished maternal responsiveness (Laurnet and Ablow 2013). Some evidence suggests that healthy and depressed mothers also process negative infant cues differently. For example, healthy mothers show activation of reward pathways in response to their own infant’s cry (Lorberbaum et al. 2002; Noriuchi et al. 2008), while mothers with PPD exhibit attenuated responses to their own infant’s cry in the NAc, striatum, and OFC (Laurent and Ablow 2012). As such, depressed mothers may also have a blunted motivational response to approach their crying infants which would further derail mothering.

Given the findings above, it seems plausible that PPD would be accompanied by dopaminergic dysregulation. Using positron emission tomography (PET), Moses-Kolko et al. (2012) investigated striatal DA functioning in PPD reporting lower D2/3 receptor binding with postpartum state but no differences between depressed and healthy postpartum women. The striatum was however identified in another PPD PET study as among the sites of increased monoamine oxidase-A (MAO-A) levels which, given the role of this enzyme in monoamine catabolism, could lead to a deficiency in DA (Sacher et al. 2015). Genetic studies of PPD have further implicated MAO-A, as well as cathechol-O-methyltransferase (COMT), an enzyme that like MAO-A inactivates DA (Doornbos et al. 2009; Comasco et al. 2011; Alvim-Soares et al. 2013). More work is needed to further explore whether other aspects of DA signaling are impacted in PPD, perhaps in ways that differ from MD given the hormonal transition and behavioral adaptations unique to the postpartum period (Zsido et al. 2017).

Other studies have sought to establish functional significance of striatal/DA responses to infant stimuli by linking them with observations of mother-infant interactions. Studying non-depressed mothers, Atzil et al. (2011) found that synchronous mothers who coordinate their behavior with infant signals showed greater NAc activation when viewing video vignettes of their own infant. A similar pattern was not seen in intrusive mothers. In a subsequent PET study, high synchronous mothers were shown to have a stronger DA response in the NAc when watching a film of their own infant while low synchronous mothers did not (Atzil et al. 2017). These data may suggest that synchronous mothers experience the mother-infant interaction as more rewarding than intrusive/low synchronous mothers. Corroborating this work are findings from mothers with attachment disturbances who showed reduced striatal responses to infant stimuli (Strathearn et al. 2009). Recent evidence further indicates that mothers with greater reward circuitry function are those who are able to establish and maintain warm and nurturing relationships with their infant despite psychiatric symptoms (Morgan et al. 2017). Taken together, these findings underscore the importance of dopaminergic reward regions in positive maternal caregiving and maternal attachment while also providing support for the possibility that impaired mothering in PPD results from lower striatal activity/DA function.

One of the main limitations to neuroimaging research has been the focus on seed regions, although network-based approaches are becoming more prevalent. Functional connectivity analyses have been used to show that maternal bonding behavior relies on the synchronous firing of the NAc, amygdala, and prefrontal cortex as a network (Atzil et al. 2011). Further, by combining fMRI and PET, stronger connectivity within this network was found to be associated with greater in-network DA responses (Atzil et al. 2017). Whether disruptions in this network contribute to bonding difficulties in PPD was not examined. However, another study using motivationally salient baby cry stimuli found diminished functional connectivity between the NAc and the extended amygdala in depressed mothers while healthy mothers showed results in the opposite direction (Ho and Swain 2017). Given the roles of extended amygdala for threat processing and the NAc for reward processing, this could represent a biological mechanism underlying the difficulties of depressed mothers to integrate baby-cry distress signal processing with the reward processing needed for sensitive parenting behavior. Healthy mothers, on the other hand, may be better able to activate their NAc during baby-cry distress signaling to motivate caring behaviors for their baby. More studies are needed to better understand the connectivity of the reward network with other networks in PPD (Moses-Kolko et al. 2014; Duan et al. 2017).

Human neuroimaging studies of PPD underscore the intricate interplay among maternal mental health, the mother-infant relationship, and the reward system. It should be noted that depressive symptoms in MD are also related to blunted mesolimbic reward function in response to rewarding stimuli and low motivation and pleasure for positive events and interactions (Surguladze et al. 2005; Epstein et al. 2006; Admon and Pizzagalli 2015). On the surface, this may suggest that at least within the reward system, the neurobiology of MD and PPD are similar. However, more subtle but meaningful, differences may exist. For example, currently available brain imaging methods do not have the resolution to differentiate cell types within reward regions showing activational changes and cannot distinguish between alterations in excitation or inhibition. Thus, the possibility remains that a differential mechanism might underlie reward system function in PPD.

The reward system in PPD: animal models

Few studies using animal models of PPD have focused on the reward system. In some of the only work to date, Haim et al. (2014) found that depressive-like behavior in mothers exposed to chronic gestational stress was associated with compromised neuroplasticity in the NAc including reduced dendritic length, branching, and spine density. These results point to a potential mechanism underlying attenuated neuronal activation and changes in functional connectivity reported in the striatum of mothers with PPD discussed above. Importantly, the neuroplastic changes seen in gestationally stressed mothers differ from what has been reported in the NAc of males (more dendritic spines) and females (no postsynaptic effects) after stress and thus may be a unique feature associated with depressive behavior during the postpartum period (Christoffel et al. 2011; Bessa et al. 2013; Brancato et al. 2017). It was further shown that antidepressant treatment reversed the stress-induced behavioral and morphological alterations (Haim et al. 2014) suggesting that structural modifications in the NAc following gestational stress may contribute to the pathophysiology of PPD as well as its pharmacologically induced recovery.

Models of PPD can also be used to investigate the extent to which reward system dysfunction may contribute to maternal care deficits in PPD (Nephew et al. 2015). A large rodent literature indicates that maternal care is a highly rewarding, motivated behavior. Mother rats will bar press and develop a place preference for pups as they would other rewarding stimuli (Lee et al. 2000; Mattson et al. 2001). For mothers, pups are so reinforcing that they are preferred over addictive drugs such as cocaine which have high reward value (Mattson et al. 2001; Ferris et al. 2005). Not surprisingly, reward circuit alterations enable maternal females to respond to offspring as rewarding (Lonstein and Morrell 2007; Pereira and Morrell 2011; Stolzenberg and Numan 2011). Indeed, numerous studies have shown that the mesolimbic system is activated in postpartum rats during maternal interactions (Fleming and Walsh 1994; Hernandez-Gonzalez et al. 2005; Febo 2011; Fang et al. 2018) and within the NAc, altered expression of various reward-related genes occurs postpartum (Zhao et al. 2014). Furthermore, in response to pup stimuli or the expression of maternal behavior, DA levels in the postpartum NAc increase and are correlated with the quality of maternal care displayed (Hansen et al. 1993; Champagne et al. 2004; Afonso et al. 2013; Shnitko et al. 2017). Conversely, maternal care is impaired following NAc or VTA lesions and after VTA inactivation (Hansen et al. 1991a, b; Seip and Morrell 2009; Numan et al. 2009). Additional evidence that mesolimbic DA is essential for maternal responsiveness comes from work demonstrating that DA antagonists administered into the NAc disrupt maternal behavior (Numan et al. 2005). Since place preference for pups is also disrupted by blocking DA systemically or following VTA inactivation (Fleming et al. 1994; Seip and Morrell 2009), a major way DA within the reward system is thought to facilitate maternal behavior is by enhancing the incentive value of pups. Together, these results suggest that disrupted maternal care seen in various PPD models may be due to mesolimbic dysregulation and a deficient pup-reward mechanism. Consistent with this are data from FSL mothers which show a lower DA response in the NAc while interacting with pups along with a failure to express a place preference for pups (Lavi-Avnon et al. 2008). Decreased striatal DA has also been observed in BALB/c mice, “poor mothers” that develop depressive-like behavior following pregnancy and delivery (Avraham et al. 2017). Work using the gestational stress (Leuner et al. 2016) and high-fat diet (Bolton et al. 2017) models of PPD further point to altered DA signaling in the NAc and other VTA targets. DA dysfunction would also be predicted in endocrine models of PPD since sex steroid hormones and stress hormones modulate the reward system and influence reward behavior (Brummelte and Galea 2010; Montoya et al. 2014; Macoveanu et al. 2016), but this has not been examined.

Like human research, most of the animal studies using PPD models focus on discrete brain regions. However, functional connectivity imaging methods have been recently applied to maternal rodents displaying caregiving deficits as a result of early life stress to investigate brain circuits in PPD. These results show reward-related connectivity changes in maternal females following exposure to early life stress (Nephew et al. 2018). Notably, early life stress produced different functional connectivity changes in virgin females providing additional support for the possibility of distinct neurobiological features in PPD (Nephew et al. 2017b).

Overall, findings from animal models complement the human literature and suggest that the reward circuit is impacted in PPD. Animal models of PPD further implicate changes in neuronal connectivity, altered neuroplasticity, and dopaminergic dysfunction, and thus are beginning to shed light on more specific mechanisms that may be involved, some of which appear to be unique to postpartum females.

Oxytocin-dopamine interactions in PPD

The activity of the reward circuitry is regulated by a number of modulators including the neuropeptide OT (Fig. 2) which has several crucial functions during the peripartum period (Rilling and Young 2014). OT is synthesized in the hypothalamus and released from the pituitary into the periphery where it acts as a hormone at the end of pregnancy to stimulate parturition as well as during the postpartum period to promote lactation. Besides peripheral actions, OT is released intracerebrally (Leng et al. 2008) and via interactions with mesolimbic DA system regulates maternal care by making offspring rewarding (Stolzenberg and Numan 2011; Love 2014; Olazábal 2018). This is accomplished via projections from hypothalamic OT neurons to the VTA which expresses OT receptors (OTR) and where OT enhances the activity of DA neurons to stimulate DA release (Shahrokh et al. 2010; Beier et al. 2015; Song et al. 2016; Hung et al. 2017; Peris et al. 2017). Behavioral studies have provided a functional link among OT, the DA system, and maternal care—mother rats displaying greater levels of maternal care have more OT projections to the VTA whereas blockade of OTR in the VTA disrupts maternal behavior (Pedersen et al. 1994; Shahrokh et al. 2010).

Fig. 2
figure 2

Simplified schematic diagram of dopamine (DA) and oxytocin (OT) neurocircuitry in the rodent brain

Full size image

Parallel investigations in humans have associated higher peripheral OT levels with better mothering (Levine et al. 2007; Galbally et al. 2011; Feldman and Bakermans-Kranenburg 2017; Kohlhoff et al. 2017) and genetic studies have linked variations in the OTR gene to maternal caregiving (Bakermans-Kranenburg and van Ijzendoorn 2008; Feldman et al. 2012; Mileva-Seitz et al. 2013; Tombeau Cost et al. 2017). Several neuroimaging studies also suggest interactions between OT and the DA reward system in the regulation of mothering. For example, mothers that deliver vaginally, which involves a substantial OT surge and enhanced bonding, display increased neural activation in the striatum compared to mothers who deliver by Cesarean-section (Swain et al. 2008). Breastfeeding also increases OT levels and elevates striatal activity when mothers hear the cry of their own infant (Kim et al. 2011). Moreover, in synchronous mothers and those with greater maternal attachment, there is a positive correlation between levels of circulating OT and NAc activation when viewing infant stimuli (Strathearn et al. 2009; Atzil et al. 2011). OT administration has also been shown to increase activation of the VTA in response to infant stimuli (Gregory et al. 2015). Collectively, the human data align with the rodent findings by suggesting that OT via actions in the reward system supports maternal caregiving.

In addition to maternal behavior, OT has also been implicated in mood regulation (Neumann and Landgraf 2012). In rodents, OT has antidepressant and anxiolytic properties, although the relationship between OT and emotion in humans remains unclear as discrepant findings have been reported (Slattery and Neumann 2010; McQuaid et al. 2014; Massey et al. 2016). For PPD, some studies have shown an inverse relationship between peripheral OT levels and maternal depression such that mothers with higher OT levels during pregnancy or postpartum present less depressive symptoms (Skrundz et al. 2011; Apter-Levy et al. 2013; Stuebe et al. 2013; Eapen et al. 2014; Jobst et al. 2016). Depressed mothers often have difficulties nursing and are more likely to stop nursing earlier, which may indicate a common pathogenesis involving diminished OT (Stuebe et al. 2013). In other research, however, no association was found between OT and maternal depressive symptoms, although higher OT was associated with lower depressive symptomology exclusively in mothers with greater psychosocial stress, suggesting that OT may protect women in stressful situations from developing depression (Garfield et al. 2015; Zelkowitz et al. 2014). Adding to the studies measuring peripheral OT levels, complementary studies have associated PPD with genetic and epigenetic variations in the genes for OT as well as the OTR (Apter-Levy et al. 2013; Jonas et al. 2013; Mileva-Seitz et al. 2013; Bell et al. 2015; Kimmel et al. 2016; King et al. 2017).

The limitation of all human OT studies is that analyses are done on peripheral samples which may not accurately reflect changes that occur in the brain. Non-invasive measurement of central OT is not feasible in humans and despite some recent advances, OTR currently cannot be quantified in the living human brain (Smith et al. 2016). With animal models however, such analyses are possible and, as previously discussed, have provided much insight into the role of the central OT system in maternal care. Animal models are also beginning to shed light on OT mechanisms in PPD (Table 2). In both gestational stress and chronic social stress rodent models of PPD, OT gene expression is reduced in the hypothalamus (Hillerer et al. 2011; Murgatroyd and Nephew 2013; Wang et al. 2018) suggestive of less OT availability in areas receiving OT input including the reward system. Fewer OT fibers and lower OTR expression has also been found in the VTA of gestationally stressed mothers (Leuner et al. 2016), another indicator of diminished OT signaling. Further research is necessary to more fully characterize how the OT system is affected in PPD (Moura et al. 2016) and how this in turn impacts reward system functioning to ultimately influence depressive and maternal behavior.

Table 2 Findings from animal models investigating the role of dopamine (DA) and oxytocin (OT) in PPD
Full size table

Conclusions

Converging evidence from both clinical and preclinical studies implicates the mesolimbic dopaminergic system as a critical node of dysfunction in PPD. Consequently, treatments augmenting the reward system may be effective in improving mood and maternal functioning in mothers suffering with PPD and in doing so prevent the detrimental effects on the offspring. In this regard, OT has potential but its role as a therapeutic tool for the treatment of PPD is still unclear and requires further study (Kim et al. 2014; Mah 2016; Moura et al. 2016; Wang et al. 2018). Another potential strategy involves reward-based psychotherapy as behavioral therapy explicitly encouraging patients to engage in rewarding activities during treatment has been found to be effective in MD, potentially by affecting striatal response to reward (Dichter et al. 2009). Whether this treatment, like some other non-pharmacological approaches, would be beneficial to women suffering from PPD has not been examined but warrants investigation (Daley et al. 2007; Pawluski et al. 2017; Swain et al. 2017). It is also expected that as our understanding of the neurobiological underpinnings of PPD continues to grow, novel targets for intervention within the reward system will likely be uncovered.