Introduction

Depression, one of the most frequent mental disorders after childbirth, is considered a severe and current health problem due to its high prevalence and the negative impact that it has on both the mother's health and the development of her child [1]. The postpartum period, considered one of the most complex experiences of human life [2], is characterized by an emotional vulnerability that, associated with physiological, psychological, social, and cultural changes, promotes the appearance of maternal mental disorders. PPD is defined as a mood disorder that usually manifests itself in the first four weeks after delivery and usually reaches its maximum intensity in the first six months after delivery.

Epidemiological studies on PPD suggest a multiplicity of risk factors involved in its genesis, among which are endocrinopathies [3, 4]. Thyroid disorders are the endocrine diseases that have been most researched to identify an association with postpartum depression. During pregnancy, the maternal immune system undergoes many changes to accommodate the development of the fetus [5] and tries to return to its prepregnancy state in the postpartum period. These modifications include changes in the production of autoantibodies that target thyroid antigens, such as thyroid peroxidase (anti-TPO) [6]. Anti-TPO is the most common type of thyroid autoantibody found in euthyroid individuals and is associated with various forms of thyroid dysfunction [7].

Several studies have shown a positive association between the thyroid peroxidase antibody and the development of mood disorders, with a high prevalence of positive anti-TPO antibodies among patients with bipolar and unipolar disorders. However, this association between positive thyroid antibody and postpartum depression has not been replicated in other studies. Menna et al. [8] and Le Donne et al. [9] concluded that there is insufficient evidence to confirm an association between PPD and postpartum thyroiditis or positive anti-TPO antibodies (in euthyroid women who did not develop postpartum thyroiditis).

The relationship between thyroid dysfunction and PPD cannot yet be considered consistent [10, 11] and requires further study. Therefore, the present study aims to assess the potential of thyroid biomarkers as predictors for the development of PPD through a systematic review. Our working hypothesis is that one or more thyroid markers can predict the risk of developing PPD in pregnant/puerperal women, and these markers may be related to worse disease.

Methods

Review question

Can thyroid markers predict the risk of developing PPD in pregnant/puerperal women?

Inclusion and exclusion criteria

As an inclusion criterion, the publication should contain original data, and the research must be carried out with humans, not including literature reviews, editorials, perspectives, letters, commentaries, and abstracts from meetings. No other exclusion criteria, such as language, year limit, sample sizes, or diagnostic tools for PPD, were used.

Search strategy

This study was based on a systematic review of scientific articles published in indexed journals until the date of January 24, 2022. The PubMed, Lilacs (Latin American and Caribbean Literature in Health Sciences) SciELO (Scientific Electronic Library Online), and Scopus databases were searched, according to Table 1. For the outcome (thyroid changes), the terms “thyroid dysfunction”, “thyroid hormones”, “TPO protein, human”, “thyroid peroxidase antibody”, “hypothyroidism” and “hyperthyroidism” were used. As terms of exposure (PPD), the terms “depression, postpartum” and “perinatal depression” were used. In the Lilacs and SciELO databases, the terms “postpartum depression” and “postnatal depression” were used to determine if they were present in the articles, according to Health Sciences Descriptors (DeCS).

Table 1 Structured search strategy carried out in databases. The search followed the structure of each database
Full size table

All references were managed in EndNote X7 software (Thomson Reuters, New York, NY, US). Initially, duplicate references were excluded. Titles and abstracts were independently screened based on the aforementioned inclusion and exclusion criteria by two reviewers (PMSS and AL). The screened lists were compared, and in case of disagreement, a consensus was reached by discussion. When a consensus was not achieved, a third reviewer decided if the article should be included (AMA). After the initial screening of titles and abstracts, full articles were evaluated by the same two reviewers. In addition to an electronic search, the reviewers also performed a hand search in the reference lists of all included studies. Predefined data collection worksheets were used for the data extraction of each selected publication. This systematic review followed the PRISMA statements, with some adjustments [12].

Data extraction

The titles and abstracts of the studies were initially analyzed. For the full evaluation, publications with original results on the prevalence or incidence of PPD in women/pregnant women with previous thyroid evaluation were selected (Fig. 1). A secondary search was also carried out in the bibliographic list of the articles initially evaluated to identify other important references not captured by the initial search.

Fig. 1
figure 1

Flow chart analysis of the articles selected in the PubMed, LILACS, Scielo and Scopus databases (January 24, 2022)

Full size image

In addition to data on thyroid assessment and PPD, information was collected on methodological aspects that could have some influence on the results of suspected or diagnosed cases of PPD, such as the study design, sociodemographic profile of the population evaluated, gestational and postpartum period used and blood analysis method and criteria.

The process of searching, extracting, and tabulating the selected articles was performed independently by two reviewers (PMSS and AL) to be submitted to descriptive analysis. Cases of disagreement were handled as described above. Both reviewers also manually searched the reference section of each relevant article and independently assessed and collected all other articles identified as eligible. A meta-analysis of the data was not feasible, given the absence of agreement in the literature to date.

Statistical analysis

We chose not to meta-analyze data from eligible studies because significant heterogeneity exists across studies in terms of the assays used to measure anti-TPO and the cutoffs used to determine anti-TPO (i.e., normal or abnormal levels), the methods used to define depression, and the timing of assessment for both TPO-AB status and depression. Consistent timing in the measurement of TPO-AB and thyroid hormones during pregnancy is required for valid meta-analysis given the changing immune milieu of this period of life [13, 14].

Results

Our initial searches yielded 1768, 272, 38, and 54 studies in the PubMed, Lilacs, SciELO, and Scopus databases, respectively (Fig. 1). Twenty-one full text articles were examined as they were thought to contain data that could address our objective (associations between anti-TPO during pregnancy or the puerperium or postnatal depression). However, only 15 original articles suggested the presence, in the full text, of data on the frequency of PPD and thyroid assessment [9, 15,16,17,18,19,20,21,22,23,24,25,26,27,28]. The selected articles were published between 1992 and 2019.

The characteristics and results of the studies included in this review can be found in Table 2. The sample sizes in these studies ranged from 31 [15] to 1075 [16]. One study used a case–control design [17], four used a cross-sectional design [9, 18,19,20], and ten utilized prospective cohort designs [15, 16, 21,22,23,24,25,26,27,28].

Table 2 Summary of selected studies on maternal thyroid disorder and postpartum depression
Full size table

Of these studies, seven reported on the relationships between anti-TPO during pregnancy and postnatal depression [16, 20,21,22,23,24,25], and five investigated the links between anti-TPO during the postpartum period and postnatal depression [9, 17,18,19, 26].

Of these, one study used a cutoff value of 19.6 kIU/l [23], two studies each utilized cutoffs of 20 IU/ml [16, 25], and one utilized 27 IU/ml [17], 34 IU/ml [9], 49 kIU/l [18], and 50 U/ml [24]. One study used a cutoff value of 525 U/ml of microsomal antibodies [21], and one study identified women as anti-TPO (TPO-AB +) if they detected any level of microsomal antibodies at 32 weeks of gestation [22]. Two studies also examined TPO-AB as a continuous measure [9, 17].

Estimates of the prevalence of PPD in pregnant women with thyroid disorders ranged from 8.3% [17] to 59.0% [24]. For follow-up studies, the cumulative incidence of self-reported depression from the first episode in the first postpartum year was 6.3% in a high-city survey (n = 1075) [16]. The incidence of depression at 3 days postpartum and 4 weeks postpartum was 14.58% and 7.29%, respectively [28]. Among other findings, women in the postpartum period (24–48 h) with positive anti-TPO had a five times greater risk of hypothyroidism [17]. Clinical thyroid dysfunction occurred in 19% of patients with postpartum psychosis compared to only 3% of the control group [15].

Anti-TPO and anti-TG levels were significantly higher in women at risk for PPD [9]. In a study by Groer and Vaughan [25], anti-TPO positive women had significantly higher scores for depression, anger, and total scores of mood disturbance postpartum than anti-TPO negative women, regardless of the development of postpartum thyroiditis (n = 25).

In another study, after adjustment for previous psychiatric contact, smoking during pregnancy, prepregnancy BMI and sleep, TSH levels above 4.0 mU/L were associated with an increased risk of depressive symptoms at six months postpartum [26]. Lower concentrations of TGB at the end of pregnancy also proved to be a strong predictor for perinatal syndromic depression, as well as a history of trauma [27].

The postnatal period in which maternal depressive symptoms were assessed varied between the selected studies (Table 3). Although all studies were restricted to up to 1 year postpartum, eight studies (53,3%) covered some period between the immediate postpartum period and 6 months postpartum [9, 18,19,20, 25,26,27,28], and seven publications (46,7%) carried out a maternal mental health assessment after 6 months [15,16,17, 21,22,23,24]. A follow-up after 1 year of the postpartum period was not performed by any of the studies.

Table 3 Instruments for diagnosis/screening of postpartum depression
Full size table

Concerning the detection instruments used for PPD screening or diagnosis (Table 3), nine studies (60,0%) used the Edinburgh Postnatal Depression Scale (EPDS), of which they made exclusive use of the EPDS 5 surveys (33,3%) [16, 17, 19, 26, 28]. The Research Diagnostic Criteria, RDC, was exclusively used in two publications (13,3%) [22, 24], and in only one publication (6,7%), the clinical interview (SCI) was used to characterize PPD [15].

The cutoff values for the EPDS also varied between studies from 9 to more than 12 points (Table 3). Only one publication used different cutoff points for the gestational trimesters and the postpartum period: 1st trimester ≥ 11, 2nd and 3rd trimester ≥ 10 and postpartum ≥ 13 [16].

Discussion

From this systematic review of the literature to clarify the relationship between maternal thyroid changes and postpartum depression, based on our search criteria, it was observed that the studies on the subject are heterogeneous in terms of study size, population studied, design (prospective, case–control, transversal), psychometric scale, and the evaluation of thyroid hormones (different analysis methods and different cutoff points). However, for some authors, the status of thyroid peroxidase antibodies has become considered a marker of vulnerability to depression. It is observed that studies have been concerned with assessing PPD and thyroid changes throughout the gestational period and in the postpartum period through longitudinal studies. Most studies followed the participants over a certain period, with periodic measurements of postnatal depressive symptoms, thus obtaining an estimate of the incidence of the condition. The studies that showed a relationship between PPD and thyroid function suggested that thyroperoxidase antibodies (anti-TPO) may be a possible target in the search for a biomarker to predict the development of emotional disorders, including PPD [16, 17, 21, 23,24,25].

Ruschi et al. [19] and Kuijpens et al. [24] showed that the frequency of PPD was high, without an association between PPD and thyroid alterations. Multiple studies examining associations between thyroid hormones and depression during the perinatal period have suggested a link [20, 26, 27, 29, 32,33,34]. However, a consensus does not exist as to whether clinical syndromes of thyroid dysfunction (e.g., hyper- and/or hypothyroidism) are linked to depression in the perinatal period [23, 35,36,37,38].

Regarding TSH, our research shows few studies directly correlating TSH levels and PPD [26, 28]. Zhang et al. [28] found no significant difference in the occurrence of PPD between the TSH groups > 2.5 mUI/L and TSH ≤ 2.5 mUI/L. However, for serum T4, a meta-analysis article with low heterogeneity conducted with population-based studies showed that serum T4 was positively correlated with depressed mood, while TSH was negatively associated with depressed mood [39]. A study by Sylvén et al. [26] suggested that there was no significant association between PPD and TSH levels at five days or six weeks after delivery. However, after adjustment for previous psychiatric contact, smoking during pregnancy, prepregnancy BMI and sleep, TSH levels above 4.0 mU/L were associated with an increased risk of depressive symptoms at six months postpartum.

The findings of our study showed heterogeneity in the methods used to investigate both thyroid alterations and PPD (Tables 2 and 3). According to Lewandowski et al. [33], when they evaluated baseline concentrations of free T4, free T3, and TSH at 30-min intervals in 110 healthy pregnant women, in a significant number of patients, the diagnosis of subclinical thyroid dysfunction could be misdiagnosed, not as a result of "disease", but as a result of physiological variation in TSH concentrations. Additionally, in 2021, Xing et al. [29] found that the TSH reference range was significantly influenced by sex, age, iodine intake, sample size, region and test methods and manufacturers. Therefore, for the reliability of the thyroid alteration in a sample, each laboratory must validate an appropriate TSH reference interval based on local conditions and based on the physiological variations of pregnant women, postpartum women and the postpregnancy period.

Lambrinoudaki et al. [22] investigated whether thyroid function within the normal range affects the incidence of postpartum mood disorders. The findings indicated an association between the occurrence of postpartum mood disorders and prenatal thyroid function. Within normal limits, lower levels of free T3 and free T4 were associated with an increased incidence of mood disorders in the first postpartum week.

However, Albacar et al. [27] did not observe any association between thyroid function and PPD. All women who scored 9/10 on the EPDS at 8 weeks and 32 weeks postpartum were defined as likely cases of major depression. Among the 1053 women evaluated in the study, 8.3% were depressed. Although 14.4% had high levels of anti-TPO and slightly elevated TSH concentrations with normal free T4, no association was found between thyroid function and PPD. It was concluded that thyroid function at 48 h after delivery does not predict the risk of PPD. However, it may be that the observed anti-TPO positivity worsens thyroid function and that subsequent anti-TPO positivity may be associated with PPD, requiring additional investigations at multiple postpartum intervals. For Le Donne et al. [18], the risk of PPD and alexithymia is directly associated with thyroid autoimmunity. The risk of postpartum depression and alexithymia had a significant correlation with positive anti-TPO, suggesting that these mood disorders may have neurobehavioral consequences of an autoimmune attack (because of the anti-TPO circulation in the CSF and its possible cross-reactivity with brain autoantigens) [18, 32].

After analysis, we found that higher prevalence and incidence rates of depression and/or more severe complaints of depression have been reported in antithyroperoxidase antibody (TPOAb)-positive women by some authors [16, 21, 23,24,25,26], while others could not demonstrate such an association [17,18,19,20, 26]. Sylven et al. [26] evaluated depressive symptoms during each trimester and at four postpartum moments (6 weeks, 4 months, 8 months, and 12 months). A positive anti-TPO state was associated with an increased risk of self-reported depression of a single onset at four months postpartum but not at other postpartum time points. The longitudinal pattern of self-reported postpartum depression in the positive anti-TPO group was similar to the typical course of anti-TPO in the postpartum period. This suggests an overlap in the etiology of PPD and autoimmune thyroid dysfunction. By analyzing these results, we can assess thyroid function in women with PPD. Similarly, the results of Groer and Vaughan [25] suggest that the presence of positive anti-TPO in pregnant women and euthyroid mothers increases the possibility of a negative dysphoric mood, especially of depressive symptoms that cannot be explained by stress or demographic factors.

It is noteworthy that among the studies included in this work, the reference value for anti-TPO positivity varied; for example, levels above 19.6 KIU/L [23] and serum levels above 50 IU/mL [24]. For the reference values ​​for TSH, the most indicated for the reliability of anti-TPO positivity is to compare the patient’s anti-TPO value with that of the local laboratory reference. However, the studies found do not refer to adjustment for this possible confounding factor.

For the PPD detection or screening instruments, we found that the most used scale was the Edinburgh Postnatal Depression Scale (EPDS), with cutoff values between 9 and more than 12 points (Table 3); nevertheless, in some studies, there was the use of other associated scales, such as the Montgomery-Asberg Depression Rating Scale, the Toronto Alexithymia Scale, the Hamilton Anxiety Scale, the Modified MINI Plus 5.0.0., the Perceived Stress Scale and the Mood States Profile Questionnaire [17,18,19, 21, 22], and even when used alone, the cutoff point differed between studies. The standardization of a gold diagnostic method is essential to guarantee the validity of the diagnosis of PPD in a population sample. The absence of this standardization in the studies found should be considered a limitation for the correct diagnosis of PPD.

Another limitation that should be considered in the studies addressed by this systematic review is the influence of the social conditions of pregnant women with PPD, since the development of PPD is also influenced by social factors, and these factors can have a more significant impact than a slight increase in anti-TPO. According to Zhang et al. [28], the most cited risk factors for PPD in the literature over the last 5 (five) years were lack of family or partner support, unplanned pregnancy, family or personal history of psychiatric illness, low education and being a minor. The research also concludes that social and emotional factors have more influence on the prevalence of PPD than economic factors.

Indeed, dysregulation of various endocrine systems has been implicated in the pathophysiology of both antenatal and postnatal depression [40, 41]. Research suggests that the etiology of perinatal depression involves a combination of social [42, 43], psychological [44, 45], and biological factors [40, 41, 46].

In summary, the studies comprising this systematic review suggest that associations may exist between anti-TPO-positive status during gestation and postpartum depression. However, further studies are needed that consider the aforementioned limitations and analyze different stages of the period because Anti-TPO fluctuates throughout gestation and the postpartum period [13, 14].

Conclusion

After analyzing these results, it is clear that the association between anti-TPO antibodies and PPD was previously examined with contradictory results. The studies carried out are heterogeneous in terms of study size, population studied, design (prospective, case–control, transversal), psychometric scale, and anti-TPO measures (different analysis methods and different cutoff points). Although some authors consider the status of positive anti-TPO antibodies to be a possible marker of vulnerability to depression (Fig. 2), it is not yet possible to conclude which are the mechanisms of thyroid function involved in the pregnancy-puerperal cycle and PPD.

Fig. 2
figure 2

Our main finding is that there is no definition in the literature about the function of the thyroid gland and the development of PPD in pregnant/puerperal women. However, most studies that showed a relationship between PPD and thyroid function suggest that thyroperoxidase antibodies (anti-TPO) may be a possible target in the search for a biomarker to predict the development of PPD

Full size image