INTRODUCTION

Postpartum depression (PPD) occurs in 14.5% of women within the first 3 months of postpartum (Gaynes et al, 2005; Wisner et al, 2006), and is associated with adverse consequences for the mother (England, 1994; Kendler et al, 1993), child (Goodman and Gotlib, 1999; Murray, 1992), and family (Goodman, 2004). Despite its high prevalence and pernicious effects on future generations, little is known about the neurobiological mechanisms of PPD and whether they are distinct from non-PPD mechanisms. Greater understanding of PPD neurobiology can improve nosological clarity as well as facilitate development of more effective treatments. In this study, we evaluated dopaminergic mechanisms of PPD based upon alterations in dopamine (DA) system function in postpartum rodents and in major depressive disorder.

Preclinical rodent studies converge upon heightened DA system function during the postpartum period. Increased postpartum dopaminergic activity is essential for maternal–pup caregiving (Hansen et al, 1991; Stolzenberg and Numan, 2011), with positive correlations between suckling and maternal striatal DA (Champagne et al, 2004). Higher striatal DA concentrations were detected both at 4 days postpartum compared with estrous controls (Glaser et al, 1990) and in parous compared with nulliparous rodents (Byrnes et al, 2001). Striatal D2 receptor density was lower in late pregnancy relative to diestrus and early pregnancy (Bakowska and Morrell, 1995), but postpartum, D2 receptor agonists induced greater stereotypy and disruption of prepulse inhibition in parous compared with nulliparous rodents (Byrnes et al, 2001). The perinatal hormones estradiol, progesterone, cortisol, prolactin and oxytocin are all potential modulators of DA and D2 receptor function (Lammers et al, 1999; Tonnaer et al, 1989). Evidence of striatal dopaminergic changes in postpartum humans is limited, but has been a subject of speculation given the 24-fold elevated risk for postpartum psychosis in the first postpartum month (Kendell et al, 1987a; Munk-Olsen et al, 2006). Whether this is related to postpartum DA concentration increases or to increased D2 receptor density or affinity (Petraglia et al, 1987) has not been studied directly. Greater monoamine catabolic enzyme availability reported in early postpartum women (Meyer J, 2008, personal communication) may be a compensatory mechanism for monoamine elevations.

Deficits in striatal dopaminergic function in major depression are well described (Dunlop and Nemeroff, 2007; Nestler and Carlezon, 2006; Willner et al, 2005) and highlighted by mood elevation with DA agonists (Cassano et al, 2004; Post et al, 1978), mood depression with DA antagonists or DA depleting drugs (Willner, 1983; Willner et al, 2005), lower CSF and serum concentrations of DA and DA metabolites (Sher et al, 2006), lower DA transporter and D1 receptor binding (Cannon et al, 2009; Nutt, 2006), lower pursuit of reward in laboratory simulations, and lower striatal activity to rewards (Forbes and Dahl, 2005; Forbes et al, 2009; Pizzagalli et al, 2009). D2/3 receptors are of particular interest in major depression because of their dense concentration in striatum, their role in antidepressant response (Willner et al, 2005; Zarate et al, 2004), and their regulatory role within the larger striatal DA system (Grace, 1991).

The positron emission tomography (PET) radioligand [11C]raclopride, a D2/3 receptor antagonist that is sensitive to endogenous DA transmission, binds primarily to D2 receptors (Narendran et al, 2006), but also to D3 receptors, which are concentrated in ventral striatum (Gurevich and Joyce, 1999). Depressed compared to healthy individuals revealed alternatively higher (D’haenen and Bossuyt, 1994; Meyer et al, 2006; Shah et al, 1997) or equivalent (Ebert et al, 1996; Parsey et al, 2001, Schneier F, 2012, personal communication, 2012) D2/3 receptor binding. Inconsistencies may stem from lack of discrimination between ventral and dorsal striatal D2/3 receptor binding, failure to control for concurrent medication use and comorbid psychiatric diagnoses, lack of power to detect gender-specific differences in D2/3 receptor binding, and lack of control for menstrual cycle phase (Wong et al, 1988). In studies of homogenous subject groups, such as psychomotorically slowed or hospitalized subjects (Ebert et al, 1996; Meyer et al, 2006; D’haenen and Bossuyt, 1994), depressed individuals had higher D2/3 receptor binding relative to controls, and psychomotor slowing was positively correlated with D2/3 receptor binding (Meyer et al, 2006; Shah et al, 1997) measured using techniques sensitive to intrasynaptic DA levels.

Interpretation of [11C]raclopride binding potential (BPND) presents a challenge to researchers because it can reflect disparate physiological processes. [11C]raclopride BPND is a measure determined both by density and affinity of D2/3 receptors, as well as by the availability of D2/3 receptors to bind [11C]raclopride, when not already bound by intrasynaptic DA (Laruelle, 2000). Conditions of high phasic DA release, such as during the rewarding stimulus of monetary receipt, will lead to competitive binding of intrasynaptic DA to D2/3 receptors, thus leading to reduced [11C]raclopride BPND (Zald et al, 2004).

Using [11C]raclopride–PET in reproductive-aged women varied for depression and postpartum status, we hypothesized that, relative to non-postpartum healthy women, striatal D2/3 receptor binding would be lower in postpartum women due to higher levels of phasic DA release, associated with mother–infant caregiving behaviors, such as nursing, that would compete for [11C]raclopride binding. Given HPA axis alterations through 12 weeks postpartum (Magiakou et al, 1996), we further hypothesized that a corticosteroid-related reduction of D2/3 receptor expression could contribute to lower [11C]raclopride BPND in postpartum women. We considered two competing hypotheses for depressed women: (1) Depression would be associated with increased D2/3 receptor BPND on the basis of such findings in the majority of prior PET studies of depression and lower cerebrospinal fluid DA concentrations in depression (Sher et al, 2006), conceivably leading to D2/3 receptor upregulation (Grace, 1991). (2) Depression would be associated with decreased D2/3 receptor BPND due to the important role of chronic stress in depression in reproductive-aged women and lack of psychomotoric slowing in the selected sample (Dziedzicka-Wasylewska et al, 1997). We hypothesized lower D2/3 receptor BPND in PPD relative to healthy controls given the combined contributions of increased phasic DA release of motherhood and decreased expression of D2/3 receptors due to hypercortisolemia in postpartum women.

MATERIALS AND METHODS

Subjects

Sixty-three women were enrolled into four groups: non-postpartum healthy (n=13), postpartum healthy (n=13), non-postpartum depressed (n=17 total; 10 unipolar and 7 bipolar), and postpartum depressed (n=20 total; 13 unipolar and 7 bipolar) women. The structured clinical interview for DSM-IV (First et al, 1998) was used to assess psychiatric status. Healthy subjects had no personal history of an axis I disorder, no family history of a mood or psychotic disorder, and a 17-item Hamilton Rating Scale for Depression score (HAM17) ⩽7 or Beck Depression Inventory (BDI) score ⩽9. All depressed subjects met DSM-IV criteria for a current major depressive episode and had a HAM17 ⩾14 or HAM25 ⩾18 in the past month and were scanned during the depressive episode, at which time mean HAM25 was 19.8±7.6 and 20.1±7.5 for unipolar and bipolar subjects, respectively. Individuals with bipolar depression also met DSM-IV criteria for past manic or hypomanic episodes. The psychomotor retardation item of the HAM-D on the scan day was 0 (none) or 1 (slight) for 91% of unipolar and 86% of bipolar subjects. Prevalent rather than incident cases of PPD were included to maximize the generalizability of the research, as PPD commonly begins antenatally (Stowe et al, 2005). We assessed maternal–infant attachment in postpartum women with a 19-item self-report scale (Condon and Corkindale, 1998) completed on the scan day in order to characterize the sample and to explore associations with D2/3 receptor binding. The psychometric properties of this scale were established by its authors in a sample of 260 perinatal women with a factor analysis that showed clustering of items onto three factors: quality of attachment, pleasure in interaction, and absence of hostility.

Postpartum subjects (both breast and bottle feeders) were included if they delivered a healthy, term infant in the preceding 16 weeks and were not using hormonal contraception. We acquired image data early postpartum or in the early follicular phase (day 3–9 after onset of menses) in order to minimize the potential influence of circulating ovarian hormones on D2/3 receptor BPND (Wong et al, 1988). Subjects were excluded if they had medical or neurological illnesses likely to affect cerebral physiology or anatomy, gross abnormalities of brain structure evident by magnetic resonance imaging, suicidal intent, substance abuse within the past year, lifetime history of substance dependence (other than nicotine), or exposure to psychotropic or other medications likely to alter cerebral physiology or monoamine function within the 3 weeks (5 weeks for fluoxetine) before scanning.

Subjects provided written informed consent as approved by the University of Pittsburgh Biomedical Institutional Review Board. The sample of healthy control non-postpartum women was supplemented by a concurrently run study of D2/3 receptor binding where healthy women served as comparators for women with eating disorders. Inclusion criteria were nearly identical to those of our study, with the exception that BDI rather than HAM17 score was used to measure depressive symptoms. Of note, the Radiation Safety Committee of the University of Pittsburgh approved no interruption in lactation after completion of the PET scan on the basis of <1 μSv radioactivity detected in the breast milk samples of the first five participants (Moses-Kolko et al, 2005).

PET Imaging and Analyses

All subjects underwent single-acquisition PET imaging on an ECAT HR+ PET scanner (Siemens, Erlangen, Germany) in three-dimensional (3D) mode (63 transaxial planes (2.4 mm in thickness; in-plane resolution=4.1 mm full-width at half-maximum over a 15.2-cm field of view; Drevets et al, 2001). Radiosynthesis of [11C]raclopride was performed as previously described (Halldin et al, 1991). A transmission scan was obtained to correct the PET data for attenuation effects. A dynamic emission scan (22 frames of increasing length over 60 min) was then initiated following IV bolus administration of 7.8 to 11.2 mCi (mean±SD=10.1±0.9) of high specific activity [11C]raclopride (1.60±0.43 mCi/nmol at time of injection). Arterial blood was sampled during scanning and corrected for radiolabeled metabolites to compute the plasma input function of [11C]raclopride in a subset of subjects (n=26 of 63; see Table 2 for number of subjects per group).

To provide an anatomical framework for analysis of the PET data, magnetic resonance images were obtained using a 1.5 T Signa Scanner (GE Healthcare, Milwaukee, WI) and a 3D spoiled gradient recalled sequence (TE=5, TR=25, flip angle=40°, NEX=1, section thickness=1.5 mm with no intersection gap). PET images were aligned to MR images using Automated Image Registration (Woods et al, 1993). The BPND was examined in anatomically and functionally distinct (Drevets et al, 2001) subregions of the striatum. Regions of interest (ROIs) in the anteroventral striatum (AVS), dorsal caudate (DCA), dorsal putamen (DPU), and ventral putamen (VPU) were manually traced on the MR image using a modified version of the IDL-based (Interactive Data Language, Boulder, CO) computer program, ROITOOL, of CTI PET Systems (Knoxville, TN) according to previously published guidelines (Drevets et al, 2001). A reference region for assessing the volume of distribution of nondisplaceable (VND) uptake was defined in the non-vermis cerebellum (CER), which is devoid of D2/3 receptors (Hall et al, 1994).

Regional tissue time–activity concentrations curves were obtained from the dynamic PET data for each ROI. Regional [11C]raclopride BPND values were determined using a simplified reference tissue method (SRTM; Gunn et al, 1998; Lammertsma et al, 1996). In subjects for whom arterial blood was sampled (n=26), Logan graphical analysis (Logan et al, 2001) additionally was applied to the arterial input function and regional tissue time–activity concentrations to derive the [11C]raclopride distribution volume (VT) measure that was used to compute regional BPND values as ((VT−ROI/VND)−1; Lammertsma et al, 1996; Mintun et al, 1984). Regional BPND values were determined for all subjects using the SRTM as k3/k4=(Bavail/Kd)fND, which is equivalent to (VT−ROI/VND)−1, where k3 is the association rate of [11C]raclopride to D2/3 receptor, k4 is the dissociation rate of [11C]raclopride from D2/3 receptors, Bavail is the available D2/3 receptor density, Kd is the equilibrium dissociation constant, and fND is the free fraction of [11C]raclopride in tissue (Lammertsma et al, 1996; Mintun et al, 1984).

Hormone Analyses

Reproductive status was assessed through self-reported menstruation charting and scan day reproductive hormone concentrations, measured between 6:00 am and 12:15 pm on the scan day. Specimens were analyzed in duplicate and were assayed as part of a single batch to reduce variability. Estradiol and progesterone concentrations were measured by radioimmunoassay (Coat-A-Count, DPC, Los Angeles, CA). Intra- and inter-assay coefficients of variation for each of these assays are <10% and <5%, respectively. Prolactin concentration was measured using time-resolved immunofluorescence (Delfia, Finland), as described previously (Berga et al, 1997). Between- and within-assay CVs were <10%. In the subgroup of non-postpartum controls from a concurrently run study of D2/3 receptor binding in eating disorders, estradiol and progesterone concentrations were measured by Chemiluminescent Immunoassay System (Advia Centaur, Walpole, Massachusetts). Intra- and inter-assay coefficients of variation for this assay were <10% for estradiol and <13 and 6%, respectively, for progesterone. Owing to assay variability for measurement of reproductive hormones, and low power to determine statistical significance of many independent variables, hormone concentrations were not included in the statistical model for D2/3 receptor binding.

Statistical Analyses

Statistical analyses were performed with STATA software, version 10 (Stata Corp, College Station, TX) and SPSS software version 17. Subject characteristics were compared with Pearson's χ2 for categorical, and t-tests and ANOVAs for continuous variables. To be complete, we examined differences by postpartum status, depression status, and group (six groups varied for postpartum, unipolar, and bipolar status, and three groups varied for depressive status: no depression, unipolar, and bipolar depression). Pearson correlations were used to measure the agreement between SRTM and Logan-derived D2/3 receptor BPND in the subset of subjects with arterial data.

Mixed-effects regression modeling was used to estimate the fixed effects of age, region, depression, postpartum status, and two-way interactions on D2/3 receptor BPND. We conducted a separate regression each for ventral (AVS and VPU) and dorsal (DPU and DCA) striatal regions based upon their functional differences (Haber et al, 2000). On the basis of differences between unipolar and bipolar diagnosis relative to D2/3 receptor BPND in our analyses, we conducted separate models for unipolar and bipolar depression (supplementary data contain regression results for unipolar and bipolar groups combined). ROI was the repeated-measure variable. Exploratory univariate mixed-effect regressions on D2/3 BPND were conducted with the following independent variables: body mass index (BMI), breastfeeding status, race, parity, weeks post birth, smoking, mother–infant attachment, Hamilton anxiety, HAM25 scores, and EPDS scores. Variables that were significantly associated with the dependent measure at p⩽0.15 were then added to the mixed-effect model. We used −2Δlog likelihood to test for the significance of an independent variable or interaction.

RESULTS

Subject Characteristics

Sixty-three [11C]raclopride–PET scans were acquired and analyzed according to published methods (Drevets et al, 2001). Demographic and clinical data for all subjects are presented in Table 1. Non-postpartum women (31%) were more likely to smoke than postpartum women (6%; p=0.02). There were no smokers in the postpartum healthy group. BMI was not statistically different by group, but trended toward being higher in the postpartum relative to non-postpartum women (p=0.06). When examined by subject group, although not statistically significant, the unipolar depressed group's BMI was remarkably similar between postpartum (27.8±5.0) and non-postpartum women (27.3±7.1), whereas BMI was higher in the postpartum group (26.9±3.8 and 26.9±3.3) for both healthy control women and women, with bipolar depression, relative to the non-postpartum group (23.6±3.7 and 23.9±5.6; Figure 2). All healthy and 61% of depressed women were antidepressant naıve. Women with bipolar disorder were significantly more likely to have prior psychotropic exposure compared with women with unipolar disorder (p<0.001). No subject had prior antipsychotic exposure. All subjects were free of psychotropic drug exposure for a minimum of 3 weeks before scanning. Depressed women were mild-to-moderately depressed on the scan day (HAM25=19.9±7.4) and had significantly higher depression and anxiety ratings that did healthy women (p<0.001), as expected. There was no difference between HAM25 ratings for unipolar (19.8±7.6) and bipolar women (20.1±7.5) on the scan day. Quality of mother–infant attachment and absence of hostility were significantly greater in healthy compared with unipolar depressed postpartum women (p=0.02 and p=0.01).

Table 1 Scan Day Sample Characteristics
Full size table

Variability of within-group hormone concentrations and use of two different laboratories limited the ability to detect between-group differences. As expected, postpartum relative to non-postpartum women had higher prolactin concentrations (p=0.002), although it is noteworthy that prolactin data were missing for eight non-postpartum healthy subjects and one postpartum depressed subject. Low-range estradiol and progesterone concentrations in 54 of 62 subjects suggested that the scan was obtained during the early follicular menstrual cycle phase or during postpartum anovulation, as planned. Eight subjects were scanned in their mid-cycle or luteal phase despite efforts to obtain the scan during low levels of reproductive hormones.

PET Data (Table 2, Figure 1)

Consistent findings across all models were as follows: D2/3 receptor binding decreased across all ROI with increasing age, as reported previously, at an approximate rate of 0.2–0.3 BPND units/decade (p<0.001). ROI was a significant covariate in all models, signifying the expected lower D2/3 receptor BPND in AVS and DCA relative to VPU and DPU (p<0.001). PET data were examined for potential confounds as follows: CER VT (VND) was not statistically different between the six groups for those cases in whom an arterial plasma input function was obtained (Table 2). There was no significant group difference in radiotracer mass or R1 (rate of radiotracer delivery to the ROI relative to radiotracer delivery to CER) in any ROI. Logan and SRTM BPND values were highly correlated (Pearson's r) in all regions (AVS 0.89, DCA 0.99, DPU 0.97, VPU 0.91; p<0.001; n=26). The agreement between Logan and SRTM BPND values supports the use of simpler SRTM BPND values in the maximal data set for statistical modeling of D2/3 receptor binding.

Table 2 Observed D2/3 Receptor BPND by Group (SRTM)
Full size table

In exploratory univariate models, both BMI and weeks post birth were inversely related with D2/3 BPND (p<0.15). All other independent variables tested were not significantly associated with D2/3 BPND (p>0.15), including injected [11C]raclopride mass, breastfeeding status, race, parity, smoking, mother–infant attachment, Hamilton anxiety, and HAM25 and EPDS scores. Weeks post birth proved nonsignificant in the full models.

In the unipolar depression model (Table 3, model 1; Figure 1), for the whole striatum and ventral striatum, both unipolar depression (b=−0.15, p=0.02; b=−0.16, p=0.07) and postpartum status (b=−0.12, p=0.05; b=−0.20, p=0.02) were associated with 7–8% lower D2/3 receptor BPND. BMI was not significant in any of the unipolar depression models. There was a region × age interaction for D2/3 receptor BPND in the ventral striatum model, such that the slope of the relationship between BPND and age was steeper for VPU than for AVS. There were no other two- or three-way interactions among age, region, subject group, and BMI. In the dorsal striatum, unipolar depression was associated with a 0.15 reduction in BPND (p=0.06; 5% lower), similar to the models for whole and ventral striatum, but postpartum status was not a significant predictor variable.

Table 3 Model for D2/3 Receptor BPND
Full size table
Figure 1
figure 1

In ventral striatum, healthy, non-postpartum healthy women (**) had increased D2/3 receptor BPND relative to postpartum healthy women (b=−0.20, p=0.02) and all women with unipolar depression (b=−0.16; p=0.07; χ2=750.01, p<0.0001). Stippled boxes representing bipolar depressed women are included to show dissimilarity of D2/3 receptor BPND between unipolar and bipolar cohorts. Note: The length of the box is the interquartile range (IQR). AVS=anteroventral striatum.

PowerPoint slide

Full size image

In the bipolar depression model, for both the whole striatum and ventral striatum, postpartum status (model 2a: b=−0.13, p=0.03; b=−0.20, p=0.02) and BMI (model 2b: b=−0.02, p=0.03) were equally good predictors of D2/3 receptor BPND (Table 3, models 2a and 2b). Bipolar depression was not associated with D2/3 receptor BPND in whole or ventral striatum. Neither depression, postpartum status, nor BMI was a significant predictor of dorsal striatal D2/3 receptor BPND. There were no other two- or three-way interactions among age, region, subject group, and BMI. D2/3 receptor BPND was higher in bipolar relative to unipolar depressed women in DCA and DPU (Table 2, t=−2.3, p=0.03; t=−2.6, p=0.01).

DISCUSSION

This is the first study of D2/3 receptor imaging in PPD. Strengths of our study were the inclusion of unmedicated women and the use of control groups for both postpartum and depression status. Because unipolar vs bipolar depression diagnoses diverged in their relationship to D2/3 receptor BPND, separate models enabled us to make inferences about each diagnostic category. Our main study finding when we restricted the sample to women with unipolar depression was lower striatal D2/3 receptor BPND in association with both depression and postpartum status in ventral striatum (7–8%) and depression alone in dorsal striatum (5%). Because both ventral and dorsal striatal percent differences in D2/3 receptor BPND observed are similar to test–retest reliability rates of 8.6 and 4.4%, respectively (Mawlawi et al, 2001), the magnitude of change related to depression and postpartum status is considered small, but similar to the magnitude of group differences in other PET studies. Combined depressed and postpartum status did not produce additive lowering of D2/3 receptor binding; therefore, there were no differences in D2/3 receptor BPND between depressed and healthy postpartum women. In the model for bipolar depression, depression was not associated with D2/3 receptor BPND. Instead, postpartum status and BMI were equally good predictors of D2/3 receptor BPND. Aside from the association of D2/3 receptor BPND decreases with increasing age (Volkow et al, 1996; Wong et al, 1984) and increasing BMI (Wang et al, 2001b), no other demographic or clinical characteristics were significantly associated with D2/3 receptor BPND in our final models.

It was unexpected that depression was not associated with any change in D2/3 receptor BPND in bipolar depressed relative to healthy control subjects. We may have been underpowered to detect a significant association for bipolar depression, but it is more likely the case that the DA system is functionally different in unipolar relative to bipolar depressed, reproductive-aged women. Indeed, D2/3 receptor BPND was higher in bipolar relative to unipolar depressed women (Table 2, t=−2.3, p=0.03; t=−2.6, p=0.01). That significantly more bipolar than unipolar depressed women had prior psychotropic exposure might also relate to neurobiological differences between the groups related to prior drug exposure or severity of illness. It is also noteworthy that in the bipolar depression regression, postpartum status and BMI were interchangeable variables in prediction of ventral striatal D2/3 receptor BPND. We note (Figure 2) the similarity of BMI for healthy and bipolar women, and higher BMI in the postpartum compared with the non-postpartum group; therefore, we can speculate that BMI might be a mediator of the postpartum association with lower D2/3 receptor BPND in this model (applicable to healthy women and women with bipolar depression). Indeed, a study of morbidly obese individuals (BMI>40) revealed lower D2/3 receptor BPND in obese relative to nonobese subjects, which was attributed to low number of D2/3 receptors common among addictive behaviors including overeating (Wang et al, 2001a). Because the BMI for unipolar, non-postpartum women was higher and more variable (Figure 2) compared with healthy controls, this might have limited the ability to discern a relationship between BMI and D2/3 receptor BPND in the regression restricted to unipolar women. Because we had limited power to detect significant interactions, it remains conceivable that high BMI might have contributed to the observed reduction in D2/3 receptor BPND observed in depressives in the unipolar regression.

Figure 2
figure 2

BMI was higher in postpartum relative to non-postpartum women (p=0.06) and very similar for healthy and bipolar depressed women. Because the BMI for unipolar, non-postpartum women was higher and more variable compared with healthy controls, this might have limited the ability to discern a relationship between BMI and D2/3 receptor BPND in the regression restricted to unipolar women.

PowerPoint slide

Full size image

The association of unipolar depression with lower D2/3 receptor BPND concurs with Klimke et al (Klimke et al, 1999), in which lower pretreatment D2/3 receptor BPND was reported for depressed individuals who later proved to be SSRI treatment responders. Our findings of lower D2/3 receptor BPND in depression contrast, however, with reports of higher (D’haenen and Bossuyt, 1994; Meyer et al, 2006; Shah et al, 1997) or equivalent (Ebert et al, 1996; Parsey et al, 2001) striatal D2/3 receptor BPND in the whole striatum of depressed subjects relative to controls. It remains unclear how these data compare with ours; however, these studies did not discriminate the AVS from the remainder of the striatum, and the differences we identified were maximal in the ventral striatum. Moreover, these studies lacked sufficient power to conduct gender-based analyses, and did not control for menstrual cycle phase, so it remains possible that the differences we report herein may not generalize to males. In addition, the results of some of these previous studies were confounded by recent exposure to psychotropic drugs. Finally, several prior studies were in hospitalized or psychomotorically retarded depressives (Ebert et al, 1996; Meyer et al, 2006; D’haenen and Bossuyt, 1994), and several reported positive correlations between D2/3 receptor BPND and ratings of psychomotor slowing (Meyer et al, 2006; Shah et al, 1997). Such patients may have had lower dopaminergic tone or lower intrasynaptic DA concentrations in the dorsal striatal regions that subserve motor processing, which putatively may result in compensatory increases in D2/3 receptor expression or affinity, or in reduced competition for binding to [11C]raclopride or [123I]- iodobenzamide, which are sensitive to endogenous DA levels (Laruelle, 2000). The depressed sample we studied, in contrast, did not include subjects who overtly manifested psychomotor slowing. Therefore, it is conceivable that homogeneous subgroups of depressed subjects (ie, psychomotorically slowed, treatment responders, women in early follicular phase) have distinct patterns of dopaminergic system function, suggesting that a single, specific underlying dopaminergic deficit is not universal in all individuals with depressive disorders.

We posit that stress or hypercortisolemia may be a common mechanism that explains the reduction of D2/3 receptor BPND in postpartum and unipolar depressed women in this sample. Although the current study cannot clarify whether lower D2/3 receptor binding is a mechanism or consequence of postpartum status or depression, it is noteworthy that animal models of depression and chronic stress similarly revealed lower D2/3 receptor mRNA in ventral striatum in rodents (Dziedzicka-Wasylewska et al, 1997) and lower [11C]raclopride–PET measurements in nonhuman primates (Morgan et al, 2002; Shively et al, 1997). In rodents, chronic (Moore et al, 2001) and acute (Valenti et al, 2011) stressors were associated with increased subpopulations or total DA neuron burst firing. On this basis, in depressed and postpartum subjects in this cohort, the combination of chronic (depression or childcare stress) and acute stress (arterial cannulation/PET scanning procedure; Drevets et al, 2002) could also conceivably increase phasic DA release that could compete with [11C]raclopride binding. It is thus conceivable that lower striatal D2/3 receptor BPND in postpartum (regardless of depressive status) and unipolar depressed women (regardless of postpartum status) may be a result of stress-related effects on reducing D2/3 receptor expression and increasing intrasynaptic DA concentrations. Likewise, the healthy women may have experienced the PET scan as less stressful, thus having less phasic DA release to compete with [11C]raclopride binding and higher measured D2/3 receptor BPND. If the postulated stress mechanism for lower D2/3 receptor BPND is indeed similar for postpartum and unipolar depressed women, D2/3 receptor BPND does not appear to be a specific biomarker for depression among reproductive-aged women. Because the anatomical extent of striatal D2/3 receptor reductions differed between depressed and postpartum women (Table 3), there remains the possibly of mechanistic differences. Furthermore, because of the hormonal excursions and maternal behavioral adaptations unique to perinatal women and due to challenges in the interpretation of D2/3 receptor BPND, we suspect that other experimental designs could more precisely distinguish between DA system functional alterations that accompany depression vs those of the postpartum period.

It is of interest whether the postpartum-related differences in prolactin concentration could inform the lower D2/3 receptor BPND in postpartum women. It is conceivable that higher prolactin concentrations might result from lower brain DA concentration overall, as DA inhibits prolactin secretion. Prior studies in rodents also describe that prolactin not only modulated tuberoinfundibular DA, but also was associated with increased striatal DA release (Perkins and Westfall, 1978) and striatal D2 receptor density (Di Paolo et al, 1982). As human peripheral prolactin concentration is highly variable and easily altered by behaviors, such as motor activity and food intake, it is unlikely our measure was precise enough to use as a guide for interpretation of D2/3 receptor BPND.

In conclusion, this study provides evidence for a dopaminergic mechanism for unipolar depression in reproductive-aged women, which may provide greater rationale for DA-modifying treatments in this population. This study also reveals postpartum modifications of the striatal DA system, which may contribute to the high relative risk of depression, psychosis, and mania in this reproductive period (Kendell et al, 1987b; Munk-Olsen et al, 2006). Studies of presynaptic DA, use of D2/3 agonist radioligands, use of DA challenge paradigms, and examination of the broader reward circuitry in this population are needed to shed further light on these observations.