Perinatal Depression – the Fourth Inflammatory Morbidity of Pregnancy? Theory and Literature Review

2.1 Inflammation and Major Depression

Early research on the role of inflammation in major depression centered on the observation that some people with depression show elevated levels of pro-inflammatory cytokines, and in particular those released by activated macrophages (including IL-1, IL-6, and TNF-α) (Smith 1991). (See Table 1 for a review of the origin and function of cytokines studied in this paper.) [Table 1 about here] Further research, however, found that some depressed individuals showed elevated cytokines, but others did not, and still others who did not have depression had cytokine levels in the same range as those with depression. Recent reviews and pooled analyses have found general immune dysregulation (Miller et al. 2009) and/or positive associations between depression and various cytokines (IL-1, IL-6, and C-reactive protein (CRP) in Howren et al. 2009, TNF-α and IL-6 in Dowlati et al. 2010). Some researchers have speculated that the association may exist only for certain subsets of depressed people, those who have a particular sensitivity to pro-inflammatory signals (Raison et al. 2010).

Most recently, an extensive review by Leonard and Maes indicates that many different factors, both peripheral and central, that concern cell-mediated immunity, inflammation, and oxidative and nitrosative stress, are related to the onset of depressive symptoms (Leonard & Maes 2012). These factors include but are not limited to increased production of the Th1 cytokines IFN-γ and IL-2 and the pro-inflammatory cytokines produced by activated macrophages (IL-1β, IL-6, TNF-α), as well as an increased acute phase response. Leonard and Maes review the various mechanisms through which inflammation might contribute to depressive symptomatology, including cytokine-induced changes in 5-HT receptors and cytokine-induced activation of indoleamine 2,3-dioxygenase (IDO). IDO is the enzyme that activates the catabolism of tryptophan, and its activation is associated with lowered plasma tryptophan (a crucial precursor of monoamines) as well as with elevation of various tryptophan catabolites that may have depression-inducing effects in and of themselves (Leonard & Maes 2012). Moreover, although the connection is clearly bidirectional, there is compelling evidence of causality as well. Several trials have documented that the therapeutic administration of cytokines (for example, administration of interferon-α during treatment of cancer or infectious diseases) causes a high proportion of patients to experience symptoms that are responsive to standard treatment and are associated with changes in serotonin metabolism (Raison et al. 2006, Miller et al. 2009).

2.2 Inflammation and Immunity in Normal Pregnancy

It is increasingly clear that inflammation plays some role in depression, though perhaps only for certain subsets of people. Could pregnant women be one of those subsets? Before considering this question, we first must consider the state of inflammation in normal pregnancy. Researchers initially believed that pregnancy must be a state of immune suppression – a state generated to avoid maternal attack on the foreign body growing within her, but a state that left her vulnerable to pathogenic attack herself (Kraus et al. 2010, Chen et al. 2012, Kraus et al. 2012, Pazos et al. 2012). Early pregnancy offered evidence of a proinflammatory Th1 environment, essential to implantation, whereas later pregnancy seemed to be characterized by a decrease in Th1 activity and an increase in Th2 cells, which are both tolerant of paternal antigens and supported by progesterone (Larocca et al. 2008). The improvement of Th1 autoimmune diseases such as rheumatoid arthritis during pregnancy and the worsening of those characterized by Th2 dominance (such as lupus) was seen as evidence of this theory (Larocca et al. 2008, Groer et al. 2011). More recently, a more complex model has emerged, with macrophages as the central mediators of complex endocrine, neural, and immune mechanisms. Macrophages produce the major proinflammatory cytokines and contain receptors for both estrogen and serotonin. Their presence is essential at various key points of gestation (implantation, parturition), but an excess of macrophages at other points in pregnancy is a sign of excess inflammation that may lead to pregnancy morbidity (Larocca et al. 2008, Kraus et al. 2012).

2.3 Inflammation and Pregnancy Morbidities

While normal pregnancy is thus a time of altered immune and inflammatory responses, there are also a number of pregnancy complications that are associated with additionally modified inflammatory responses. The major pregnancy morbidities known to be associated with inflammation are preeclampsia, preterm birth, and gestational diabetes. Because these disorders may be related to perinatal depression via the common mechanism of inflammation, each will be briefly reviewed.

The best studied condition is preeclampsia (PE), defined as a hypertensive disorder of pregnancy that is accompanied by proteinuria and has its onset after 20 weeks of gestation. Preeclampsia occurs in up to 14% of pregnancies and is responsible for 15-20% of maternal and fetal morbidity and mortality in developed countries (Dolea & AbouZahr, 2000, Cudihy & Lee 2009, Ahn et al. 2011). The etiology of this disorder has been elusive. There is general agreement that the immediate spur for the disorder is a generalized activation of the vascular endothelium – but it is not entirely clear what factors lead to that activation. Considerable research has focused on abnormal cytotrophoblast invasion of the maternal spiral arteries, leading to improper remodeling and placental hypoxia, which in turn leads to the release of placental debris into the bloodstream. Others propose that placentation may occur normally, but the placenta may then deteriorate (for a variety of reasons), again leading to the release of placental debris. Recently, research has focused on the possible role of the immune and inflammatory systems in achieving this vascular endothelial activation. A recent review by Ahn et al. (2011) sums up the literature on this topic, covering the involvement of toll-like receptors, maternal cytokines, apoptotic bodies, macrophages, NK cells, T cells, thrombophilias, and autoimmunity. According to the Ahn et al. review, numerous studies show that in preeclamptic pregnancies, compared to normal pregnancies, there is an increased presence of pro-inflammatory cytokines. These include those produced by macrophages and NK cells, as well as those produced by Th1 cells. In fact, the ratio of Th1 to Th2 cells is altered (favoring Th1, the opposite of the case in normal pregnancy) (Ahn et al. 2011).

There is also considerable research devoted to the connection between preterm birth and inflammation (often but not always in the context of infection). Preterm birth (i.e., birth prior to 37 weeks of gestation) is even more widespread than preeclampsia and is the leading cause of perinatal morbidity and mortality in the developed world, with 9.6% of births worldwide occurring prior to term (10.6% in North America) (Beck et al. 2010). It leads to three-fourths of all neonatal deaths and about half of all long-term neurologic disability, and it is alarmingly on the rise in developed countries (Wei et al 2010). This rise is due both to an increase in indications for iatrogenic preterm birth (i.e., preeclampsia and intrauterine growth retardation) as well as an increase in factors that lead to spontaneous preterm birth, including infection or inflammation, vascular disease, very low or very high BMI, and uterine overdistention (Goldenberg et al. 2008, Torloni et al. 2009). The increase in multiple pregnancies associated with assisted reproductive technologies contributes to both types of preterm birth (Goldenberg et al. 2008). It is increasingly clear that inflammation may play an important role. This inflammation can sometimes be shown to be associated with infection (clinical or subclinical), but may also be associated with other trends that are increasing in developed countries (for example, sleep disturbance and obesity). Wei and colleagues recently reviewed the literature on inflammatory cytokines and spontaneous preterm birth. They performed a systematic review of 17 studies comprising 6,270 participants, and concluded that spontaneous preterm birth was strongly associated with increased markers of local but not systemic inflammation (specifically, IL-6 in midtrimester cervicovaginal fluid and amniotic fluid and CRP in midtrimester amniotic fluid). There was no association for these markers in plasma specimens, and there were insufficient data on other inflammatory cytokines.

There is also evidence of a link between inflammation and gestational diabetes, which occurs in 2-10% of all pregnancies and is associated with risks to both mother and fetus and with a 35-60% chance of developing Type II diabetes at some point after pregnancy (National Diabetes Statistics, 2011). Adipocytes produce both pro and anti-inflammatory cytokines. Recent research indicates that lean adipocytes activate macrophages that preferentially produce anti-inflammatory cytokines, while the adipocytes of obese people instead favor the production of pro-inflammatory cytokines (Olefksy & Glass 2010). This shift to a pro-inflammatory environment is one cause of increased insulin resistance, and provides a link between obesity and Type 2 diabetes. Even normal pregnancy is characterized by some insulin resistance, and for those who develop gestational diabetes the degree of insulin resistance is greater than that in Type 2 diabetes not associated with pregnancy. (Lowe et al. 2010). This insulin resistance in pregnancy has been shown to be related to inflammation, independent of BMI (Wolf et al. 2004, Bo et al. 2005, Kinalski et al. 2005). Moreover, as the recent Hyperglycemia and Adverse Pregnancy Outcome Study showed, levels of inflammatory mediators (CRP and plasminogen activator inhibitor type 1) were associated with glucose levels in pregnant women regardless of whether they developed frank diabetes (Lowe et al. 2010).

2.4 Pregnancy Morbidities and Perinatal Depression

The three pregnancy morbidities discussed above also may be associated with perinatal depression, although the evidence is mixed. For preeclampsia, some studies showed no relationship between preeclampsia and antenatal depressive symptoms (Sikkema et al. 2001, Andersson et al. 2004), but one study showed an odds ratio of 2.5 for developing preeclampsia among women with symptoms of anxiety and depression (Kurki et al. 2000). Another very recent study also found an increased risk of preeclampsia among women with depression (Kharaghani et al. 2012), while Palmsten et al. (2012) found an elevated risk for preeclampsia in women treated with antidepressants but not among women with untreated depression. Similarly, for preterm birth (as summarized in Alder 2007), some studies have found an association between preterm birth and antenatal depressive and anxiety symptoms (Steer et al. 1992, Hoffman and Hatch 2000, Orr et al. 2002) whereas several others have not (Andersson et al. 2004a, Berle et al. 2005). For both morbidities, most studies have looked at antenatal depressive symptoms rather than at the risk of developing postpartum depression.

The state of the research for gestational diabetes is similar. Several small studies have shown no relationship between diabetes during pregnancy and mood (Langer & Langeret 2000, Kim et al. 2005, Mautner et al. 2009). Katon et al. 2011, in a very large cross-sectional study, also found no relationship between antenatal depression and diabetes. However, several studies that included postpartum as well as antenatal depression did find an association (a two-fold increase in Kozhimannil et al. 2009, and an increase in severity of depressive symptoms in Dalfra et al. 2012). These results are not specific to gestational diabetes, but rather to all diabetes in pregnancy. Finally, Bansil et al. 2010, in a large nationwide sample of over 32,000,000 women hospitalized for childbirth from 1998-2005, found an increased incidence of all three morbidities (along with numerous others not discussed here) among women with a diagnosis of depression.

2.5 Background Summary

Inflammation is associated with the onset of major depression outside the context of pregnancy and childbirth. Pregnancy is an altered state of immune functioning and it is against this altered background that inflammation also plays a role in several disorders that contribute to maternal and fetal morbidity, and that themselves may be associated with perinatal depression. What role, then, might inflammation play in perinatal depression? What follows is a comprehensive review of the literature on this important question.

Antenatal Depression

Four studies examined cytokines in antenatal depression. Three found an association between depressed mood and at least one marker of inflammation; one did not. In Christian et al. 2009, 60 pregnant women from a low-income, high psychosocial risk group were enrolled at an average gestational age of 15 weeks. At the single study visit, subjects completed the Center for Epidemiologic Studies Depression Scale (CES-D) and the Perceived Stress Scale (PSS) and gave blood to measure IL-6 and TNF-α. IL-6 is produced in the periphery by macrophages and T cells, and in the central nervous system (CNS) by astrocytes and microglia. It is capable of crossing the blood-brain barrier through both passive and active transport. Once there, it acts as a neuromodulator, with both pro and anti-inflammatory properties, and may also be involved in the induction of IDO, which is associated with a decrease in CNS serotonin (Leonard & Maes 2012). TNF-α is a pro-inflammatory cytokine produced by activated macrophages. It activates the vascular endothelium and increases vascular permeability. After controlling for pre-pregnancy BMI, IL-6 levels were found to be significant and positively associated with CES-D score (p = .05). There was no significant association with TNF-α. Neither stress nor any of a number of health behaviors was associated with either marker of inflammation. There was also a small effect size relationship between both inflammatory markers and severity of symptoms.

In Christian et al. 2010, the authors extended the work reported in their prior study to a comparison of the inflammatory response to an in vivo antigen challenge among depressed and non-depressed women. Mood was again measured using the CES-D, at the baseline visit only (conducted at an average of 17 weeks gestation). Macrophage migration inhibitory factor, a wide-ranging cytokine that inhibits the antinflammatory effects of glucocorticoids in a dose-response fashion and is known to be implicated in preeclampsia, was measured at the baseline visit and one week later, following vaccination against the influenza virus. Women in the highest quartile of CES-D scores were compared with those in the lowest quartile, and were found to have a significantly higher level of MIF one week after vaccination (p < .04). The groups did not differ in demographic characteristics, including BMI, nor in health behaviors.

Blackmore et al. (2011) observed 145 low-income women and assessed mood and inflammatory markers (IL-6 and TNF-α) in detail at approximately 18 and 32 weeks gestation. Mood was assessed using the Edinburgh Postnatal Depression Scale (EPDS), the State-Trait Anxiety Inventory (STAI), and the Structured Clinical Interview for DSM Disorders (SCID). No significant association was found between depressed mood or anxiety symptoms and either cytokine. There was, however, a significant association between elevated TNF-α and trauma exposure (p < .03).

Cassidy-Bushrow et al. (2012) recruited 187 African American women to examine the association between depressive symptoms and four pro-inflammatory biomarkers (hs-CRP, IL-6, IL-1β, TNF-α) and one anti-inflammatory biomarker (IL-10). CRP is an acute phase protein produced by hepatocytes and stimulated by TNF-α, IL-1, and IL-6. CRP in turn enhances the fixation of complement to pathogens. IL-1β is the most potent known mediator of inflammation in humans. It is one of the culprits in the constellation of symptoms known as “sickness behaviors” (Dantzer et al. 2008) that accompany many acute and chronic illnesses. These symptoms include most significantly fatigue and fever, as well as depression, anorexia, and myalgias. IL-1β is produced by activated macrophages and is known to be elevated in normal pregnancy. It is also associated with the synthesis of prostaglandins, so may affect uterine contractility and the onset of labor. IL-10 is an anti-inflammatory cytokine produced by Th2 CD4 cells. It is involved in the maintenance of pregnancy and progesterone production, and low levels of IL-10 have been found to increase the risk of preeclampsia (Redman 1991, Greer et al. 1994, Wadhwa et al. 2001).

Mood was measured with the CES-D. Women were recruited during the second trimester of pregnancy, and both mood assessment and blood draws were performed once. Both IL-6 and IL-1β were significantly and positively associated with depressive symptomatology (p < .04 and .03, respectively). The other markers were not significantly associated with CES-D score, nor was BMI. BMI did moderate the relationship between depressive symptoms and IL-6, however; IL-6 was significantly associated with depressive symptoms in lean women but that association disappeared in women of high BMI. In addition, IL-10 was higher in lean depressed women, but lower in heavy depressed women, likely reflecting a feedback loop (higher IL-10 in lean women in response to the increased proinflammatory burden, whereas those with depressive symptoms added to the already proinflammatory state of obesity are less likely to be able to mount an anti-inflammatory response).

In sum, two of these studies found significant associations between depression and IL-6 while one did not. Two found no association with TNF-α, one found an association with MIF, and one found an association with IL-1β. The three positive studies all measured mood with the CES-D, a screening tool that is not used to diagnose depression clinically, whereas the one negative study used a clinical tool, the EPDS.

Mixed Antenatal and Postpartum

There are 5 studies examining mood and inflammatory measures in both the antepartum and postpartum periods. All five found a significant association between mood and at least one inflammatory marker. Schmeelk et al. (1999) studied 58 pregnant adolescents. Blood samples for IL-1β and IL-1ra were drawn between 9 and 21 weeks of gestation. IL-1ra is an endogenous receptor antagonist for IL-1β. Within 24 hours of childbirth subjects were assessed for labor and delivery complications as well as for maternal depression. (No blood was drawn at the time of delivery, nor was mood measured during pregnancy.) Maternal depression was determined by scores on the Diagnostic Interview Schedule for Children (DISC) 2.1. The study authors do not comment on the reliability and validity of this scale for maternal depression, and it is notable that it was administered within 24 hours postpartum only. In addition, the authors do not mention whether they used a cutoff score or examined depressive symptoms as a continuous variable. The authors report that IL-1β levels were measured, but do not compare these levels to depression scores. It is possible that they declined to make this comparison because detectable IL-1β was found in only 4 of the 58 participants, but the authors do not directly address this. The authors also measured IL-1ra. They did not directly examine the link between depression and IL-1ra, but instead used a hierarchical multiple regression analysis to test paths in a model linking depression, CRH, IL-1ra, and postpartum complications. They state that “CRH and depression were found to predict significant levels of IL-1ra” (p=.05), and state elsewhere that lower CRH predicted lower IL-1ra (no significance level stated).

There are three relevant studies by Maes et al., all variations in the reporting of data collected at the same time on the same group of participants (up to 98 pregnant Belgian women with up to 22 nonpregnant controls). Of note, there are several other papers by the same group concerning the same set of data, but those papers were excluded because they did not directly compare mood and inflammatory measures.

In the first study, in 1999, Maes et al. examined Clara Cell Protein 16 (CC16), an antinflammatory protein produced by cells of the bronchiolar epithelium and urogenital system that suppresses IL-6, IL-1, and IFN-γ. Mood was assessed with the State-Trait Anxiety Inventory (STAI), the Zung Depression Rating Scale (ZDS), and the SCID (true for all the Maes studies). For the STAI and ZDS, women whose scores increased in the postpartum were deemed STAI or ZDS “reactors,” defined as those with a residualized score >q75. The SCID was administered 6-10 months after delivery by telephone. Serum samples were collected 3-5 days before expected delivery and at 1 and 3 days postpartum. The researchers decided to use 3 months as the cut-off for postpartum onset of depression (rather than the 4 weeks defined by DSM-IV). Women with postpartum depression (as recalled 6-10 months later) had significantly lower levels of CC16 than did those without (p=.002) at all three time points. There were no significant differences in CC16 concentration between STAI and ZDS reactors and nonreactors.

In 2000, the group examined the correlation between IL-6 (measured at the same time points as above), LIF-R (a versatile cytokine that works along with IL-6 to induce the acute phase response), IL-1ra, and IL-6r (a soluble receptor that forms a receptor-ligand complex with IL-6 and helps to regulate and coordinate inflammatory events mediated by IL-6) with STAI and ZDS reactor status. STAI reactor status was correlated with increased IL-6 level (p = .0001) at 1 day postpartum, as was ZDS reactor status (p = .03), though timing was not specified for ZDS. For LIF-R, prepartum serum LIF-R was significantly lower in STAI reactors than in non-reactors (p=.004). No significant relationship was found for ZDS reactors. IL-1ra was significantly higher in STAI reactors than in nonreactors one day after delivery (p =.006). ANOVA analysis showed a significant effect of time x STAI reactor status x parity interaction (p=.02). No significant differences were found for ZDS reactors. IL-6r was also found to be significantly increased in ZDS reactors when measured at 3-5 days before EDD (p = .001), at 1 day postpartum (p = .001), and at 3 days postpartum (p = .002); no relationship was found for STAI reactors. These correlations did not become more positive with higher scores – in other words, reactor status predicted whether inflammatory variables were elevated, but severity of symptoms did not.

In 2001, Maes et al. looked at a subset of the same sample to ask whether women with a lifetime history of major depression had greater increases in inflammatory markers in the early puerperium than women without such a history. Within 6-10 months after delivery, women had telephone interviews using the Structured Clinical Interview of the DSM to assess past history of depression. Sixteen women had a past history and 50 did not. Women with a history of depression had significantly higher levels of serum IL-6 at 1 and 3 days after delivery (p = .007 and .02, respectively) and of IL-1ra at the same time points (p<10^-3 for both). IL-6r was found to be significantly increased in those with a history of major depression (no time point was given) when compared to those without such a history. The authors also looked at current ZDS scores and found no difference between women with and without a history of depression. They therefore conclude that current symptoms cannot be responsible for the difference in inflammatory response between the two groups. Of note, they did not look at differences in STAI scores; and as their prior study demonstrates, these were associated with IL-1ra levels but ZDS scores were not.

Scrandis et al. (2008) studied 27 women at high risk for developing postpartum depression. They measured mood using the Hamilton Depression Rating Scale-Seasonal Affective Disorder (SIGH-SAD), and collected scores at three points: 35-38 weeks gestation, 1-5 days postpartum, and 5-6 weeks postpartum. No information is provided about why this scale was chosen or whether it is reliable or validated for the purpose of screening for postpartum depression. Blood was drawn to measure IL-6 and CRP at the same time points. There were no significant relationships between IL-6 levels and total or atypical depression scores at any of the time points measured. CRP, however, was found to be positively correlated with total (p = .03) and atypical (p=.002) depression scores in the prepartum. In contrast, CRP was negatively correlated with atypical depression scores in the early postpartum (p=.03), and no relationship was found in the late postpartum or with total depression scores in any postpartum period.

In sum, two of these studies of mixed prepartum and postpartum measures found an association between mood and IL-1ra. One study found a positive association for CC16. One study found an association between mood and IL-6, while another study did not. One found a relationship between mood and IL-6r. One found a relationship between mood and CRP. While the balance thus favors an association between mood and inflammation, the markers used ranged widely and represented different elements of the inflammatory response. The same is true of mood measures, which include the DISC 2.1, a combination of the SCID with the STAI and the ZDS, and the SIGH-SAD. No studies used the EPDS, the most widely used depression screen in the perinatal period.

Postpartum

Nine studies have examined the link between mood and inflammatory state in the postpartum period. Seven studies have found an association between depressed mood and at least one measure of inflammation (though in some of these studies the association holds true only for a subset of women). Two studies did not find an association.

Hucklebridge et al. (1994) examined the relationship between dysphoria and circulating immune cell populations in 75 postpartum women. On day 4 postpartum, blood was drawn and participants completed three mood questionnaires – the Stein Blues Scale (SBS), the EPDS, and the Hospital Anxiety and Depression Scale (HADS), with subscales for depression and for anxiety. Blood was examined for the presence of T helper cells, cytotoxic T cells, and natural killer (NK) cells. HADS Anxiety, HADS Depression, and the EPDS were all negatively correlated with total lymphocyte count (p<.05, one-tailed, for all). Because the four mood measurements were highly intercorrelated, the authors combined the results into a single factor they termed “dysphoria.” Dysphoria was significantly negatively correlated with total lymphocyte count (p<.03 one tailed). When divided into T cells and NK cells, the T cell count remained significantly negatively correlated with dysphoria (p<.03 one-tailed), but NK cell count did not.

In 2006, Groer and Davis sought to analyze the relationships among stress, immunity, and mood in breast-feeding as compared to formula-feeding mothers. They examined 181 healthy mothers at 4-6 weeks postpartum. Mood was measured using the Profile of Mood States (POMS) and a Total Mood Disturbance Score (TMDS) was calculated. This instrument (along with stress measures) was mailed to the mothers for self-administration at home prior to blood draws for IL-10 and IFN-γ. IFN-γ is the characteristic pro-inflammatory cytokine produced by CD4 Th1 cells. It activates macrophages and is crucial to the body's response to intravesicular pathogens. The authors note that the validity of the POMS is excellent but its reliability and consistency less so; they make no comment on validity or reliability of the TMDS. Neither breastfeeding women nor formula feeders showed any correlation between IL-10 and any of the mood measures. No regression model provided a significant source of variance in IL-10. The POMS depression score had a Pearson correlation of -0.24 (p<.05) with IFN-γ but was nonsignificant with the ratio. TMDS score correlation coefficient was -0.24 with IFN-γ and -0.27 with the ratio (p<.05 for both).

Groer and Morgan (2007) sought to compare the demographic, health, stress, immune, and endocrine factors in depressed postpartum women compared to their non-depressed counterparts. The POMS was administered to 200 women, and those in the top-scoring decile (n=25, mean score 32.1, range 21-47) were deemed to be depressed. The remaining 175 women had a mean score of 5.7 (range 0-20). This study is a secondary analysis using the same data collected in Groer's 2006 study. Depressed mothers had lower IFN-γ concentrations in serum (p<.001). In addition, they had a lower IFN-γ/IL-10 ratio in both serum (p<.04) and whole blood culture supernatants (p<.009).

Corwin et al. (2008) examined 26 women recruited on day 0 postpartum for a larger study on fatigue. Women provided clean-catch urine samples on postpartum days 0, 7, 14, and 28, and completed the CES-D on day 28. There were no differences in urinary IL-6 levels based on the level of depressive symptoms measured on day 28. IL-6 was found to fall over time in women without, but not in women with, depression (p = .001), but the authors themselves question this finding because IL-6 was undetectable in most women in days 14 and 28. Women with depressive symptoms on day 28 as measured by the CES-D had elevated levels of urinary IL-1β on day 14, but there was no significant change over time in the depressed group or the control group.

Boufidou et al. (2009) studied cerebrospinal fluid (CSF) and plasma cytokines in 56 Greek parturients. Mood was measured using the Posptartum Blues Questionnaire (PPBQ) at 1-4 days postpartum and the EPDS at 1 and 6 weeks postpartum. Those with PPBQ scores >8.2 or EPDS scores >11 were deemed to have depressed mood. Of note, the EPDS scoring instructions indicate that >10 is possible depression and >13 is likely depression; for the PPBQ, 3-8 indicates mild-moderate blues and >9 severe blues. There was no significant relationship between mood and serum IL-6 levels, though the authors do report a trend result in the immediate postpartum (p = 0.09). The authors report this result as significant by using a cutoff of p <.1. There was a positive association between IL-6 measured in the CSF at postpartum days 1-4 (p < .04), postpartum week 1 (p < .04), and postpartum week 6 (p <.01). CSF TNF-α at delivery was significantly correlated with a high score on the PPBQ in days 1-4 (p=.03), as was serum TNF-α at delivery (p = .02). There was also a significant association between increased CSF TNF-α at delivery and EPDS score at 1 week postpartum (p = .009); no other significant associations with TNF-α were found.

The Skalkidou et al. (2009) study involved 347 women, with study visits both at delivery and in the postpartum. Mood was measured by a structured questionnaire including the EPDS at five days, six weeks, and six months after delivery. Women were deemed to have postpartum depression if they scored 14 or greater at 5 days or 12 or greater on the EPDS at either of the other two time points. Sixty-seven women were thus deemed to have postpartum depression. Serum was drawn at delivery only. There was no difference in mean IL-6 levels between cases and controls.

Okun et al. (2011) looked at 56 pregnant women with a past history of either MDD or PPD, but not depressed at time of enrollment. At eight points during the first 17 weeks postpartum, blood was drawn to measure various hormones and the women were evaluated for depressive symptoms (Hamilton Depression Rating Scale, or HAM-D). The primary aim of the study was to assess the relationship among sleep, depressive symptoms, and hormone levels, but a post-hoc analysis was also done of a single measure of IL-6 (unspecified when). There was no significant relationship between IL-6 level and recurrence of depression (though a trend was noted for the interaction among these two factors and sleep quality).

Kondo et al. (2011) enrolled 139 lactating Japanese mothers, who filled out a questionnaire and provided samples of breast milk at their children's 3-month checkups. Breast milk was assayed for TGF-β, a central cytokine in the regulation of the immune system. TGF-β has an important role in developing the mucosal immune system in infants, and seems to play a dual role: decreased TGF-β is thought to render the individual more susceptible to immune dysregulation and the rise of a pro-inflammatory state, and that dysregulation can then lead to increased levels of TGF-β and other anti-inflammatory cytokines (Saxena et al. 2008). Mood in the lactating mothers was assessed using the EPDS, using 9 as the cut-off for defining postpartum depression. Depressed mothers were found to have a 4.4-fold (CI 1.5-13) higher likelihood to have increased levels of TGF-β in their breast milk.

Finally, Fransson et al. (2012) measured the relationship between maternal emotions and cytokines at birth in both maternal serum and cord blood, and whether those parameters differed between women who delivered preterm and those who delivered at term. The sample included 27 women who delivered preterm and 37 who delivered at term. Emotions were measured within 5 days of birth using the Positive and Negative Affect Scale (PANAS) and a structured clinical interview with questions about depression. Blood was drawn for cytokines during labor, and cord blood was sampled immediately after delivery. IL-1a, IL-1β, IL-2, IL-4, IL-6, IL-8, IL-10, IL-12 p70, IL-13, IL-17, IL-18, and IFN-γ were all measured. Significant associations were found for the preterm group only between depressive symptoms and maternal IL-8 (p = .013) and cord IL-18 (p = .038) and between negative affect and maternal IL-6 (p = .016), cord IL-6 (p = .001) and cord IL-18 (p = .023). IL-8 is produced by macrophages and other cells with toll-like receptors. It is an important mediator of the innate immune response, and its primary purpose is the induction of chemotaxis in neutrophils and other target cells. IL-18 is a pro-inflammatory cytokine produced by macrophages and epithelial cells, among others; it induces the production of IFN-γ.

In sum, one study found an association between total lymphocyte count and depressed mood. One found a correlation between IFN-γ and the IFN-γ/IL-10 ratio, but no correlation with IL-10 alone. One study found a correlation between IL-1β and mood. One study found an association between negative affect and maternal and cord IL-6, but four studies failed to find an association between IL- 6 (three in serum, one in urine) and mood. One study found an association between depressive symptoms and maternal IL-8 and between depressive symptoms and negative affect and maternal and cord IL-18. One study found an association between CSF IL-6 and mood, and between CSF TNF-α and mood. One study found an association between depression and increased TGF-β in breast milk. One study found an association between urinary IL-1β and mood. One study each used the SBS, the PANAS, the HADS, the HAM-D, and the PPBQ, and one used a self-created “dysphoria score.” Two each used the POMS and TMDS, and three used the EPDS.