Maternal Serum Activin A, Inhibin A and Follistatin-Related Proteins across Preeclampsia: Insights into Their Role in Pathogenesis and Prediction

The initial screening process was carried out based on the following inclusion criteria: (a) non-randomised observational research (case-control and cohort studies) published in the last ten years (2012–2022); (b) studies conducted on a population of pregnant women with singleton pregnancies and diagnosed with any subtype of preeclampsia and; (c) studies that reported the difference in serum levels of activin, inhibin or follistatin isoforms compared to pregnant women without preeclampsia and/or the predictive value of each protein. In addition, to ensure accuracy, literature review studies focused on measuring activin, inhibin, and follistatin isoforms in fluids other than the pregnant women's serum, twin pregnancies and pregnant women with additional comorbidities to preeclampsia were excluded.

On 1 May 2022, three reviewers (L.M.V.C, J.N.I and K.A.V) conducted a literature search on consulted databases. The titles and abstracts of the retrieved records were compiled and managed using Microsoft Excel 2020. Duplicates were removed in the same program, and two reviewers (J.A.B and J.N.I) independently screened all titles and abstracts to exclude articles irrelevant to the review's objective. Then, studies selected by title and abstract were assessed by full text by four reviewers (J.A.B, L.M.V.C, J.N.I, and K.A.V) to determine their final inclusion based on the previously discussed inclusion criteria. Disagreements in the study selection process were resolved by consensus, and when necessary, an additional reviewer (J.A.R.R) was consulted.

For each included study, the following data were extracted: author, year of publication, sample size, protein analysed, weeks of gestation in which the analyte was measured, the subtype of preeclampsia and primary outcomes regarding variation in maternal serum levels. Additionally, depending on the results of each study, the predictive value for preeclampsia of the biomarker studied was extracted, as well as the difference in maternal serum levels of the protein in pregnant women with preeclampsia versus pregnant women with uncomplicated pregnancies. The information was gathered in Table 1.

Inhibins are heterodimeric glycoproteins made up of α and β subunits that assemble two main isoforms: inhibin A (α-βA) and inhibin B (α-βB). While these hormones are mainly expressed by the granulosa cells upon luteinizing hormone and follicle-stimulating hormone influx, from the fourth week of gestation, inhibin A is predominantly secreted by the fetoplacental unit [46] rather than by ovarian follicles or the corpus luteum [47]. On the other hand, activins are homodimeric glycoproteins with three isolated and classified isoforms according to the combination of β subunits: activin A (βAβA), activin AB (βAβB) and activin B (βBβB). Activin A is the most predominant isoform in humans [48] and exhibits an increase in maternal serum throughout gestation that reflects the fetoplacental unit secretion [49]. Interestingly, within the maternal-foetal interface, activins elicit multiple functions in a paracrine, endocrine and autocrine fashion that hinders its classification solely as a classic hormone [48].

While not recognised as hormones per se, follistatin-related proteins are still tightly involved in multiple signalling pathways in the female reproductive system [50]. Follistatin is a glycoprotein encoded by the FST gene, which following alternative splicing, expresses a series of follistatin-related proteins that bind to peptides of the TGF-β superfamily, thereby inhibiting ligand-receptor complex formation [51]. Similar to both inhibin A and activin A, placental follistatin secretion is responsible for the rise in serum levels during pregnancy [52]. Investigations in animal models have shown that follistatin might be involved in decidualisation [53] and ovarian suppression throughout gestation [54].

Activins exert their functions through binding to heterooligomeric transmembrane receptors with intrinsic Ser/Thr kinase activity made up of two type I subunits and two type II subunits [55]. Signal transduction and transcriptional response are mediated by the Sma- and Mad-related proteins (SMADs), particularly SMAD2 and SMAD3 [56]. In contrast, inhibin A and follistatin-related proteins downregulate activin A signalling by distinct mechanisms. A more detailed explanation of the canonical signalling pathways regulated by activin A, inhibin A, follistatin and follistatin-like 3 in the maternal-foetal interface is presented in Figure 2. The activin-inhibin-follistatin system plays a crucial physiological role in normal placentation and may also serve as a predictive biomarker. Hence, hypotheses about their role in preeclampsia pathogenesis are discussed below based on the variation in maternal serum across preeclamptic pregnancies.

Figure 2.

During the first trimester of gestation, activin A is expressed by the syncytiotrophoblast, cytotrophoblast and decidual cells [57,58]. The type II subunit receptor ALK4 is expressed in the extravillous trophoblast and holds the highest affinity for activin A [59], indicating the presence of paracrine signalling within the maternal-foetal interface. Nonetheless, activin A reaches the systemic circulation, as evidenced by Silver et al. [60], who found that elevated levels of activin A in pregnancy result more accurately from increased placental secretion rather than higher levels of inhibin A. As illustrated in Figure 2, the transcriptional response to activin A augments the expression of N-cadherin [61], integrin β1 [62] and matrix metalloproteinase 2 (MMP2) via canonical ALK4-SMAD2/3 pathways [59]. In normal placentation, N-cadherin allows extravillous trophoblast adhesion, thereby conferring the proliferative, migratory and invasiveness capacity [63] aided by MMP2, which promotes decidual stromal invasion by extracellular matrix degradation and regulation of vascular remodelling [64]. More recently, integrin β1 has been recognised to enhance extravillous trophoblast migration and invasion coupled with reduced cellular apoptosis via PI3K and Akt signalling pathways [65].

Based on the evidence gathered in this review, in preeclamptic pregnancies activin A levels rise significantly across the third and late second trimesters, from 25 weeks of gestation [15], between 30⁺⁰–37⁺⁰ weeks [17,20] and until delivery [44]. However, two studies found that while levels increased, there was no significant difference with healthy pregnancies [19,36]. In contrast, changes in activin A during the first trimester are inconclusive, as recent evidence shows both a substantial increase in preeclampsia [16] between 11⁺⁰–13⁺⁰ weeks of gestation [43] and no significant difference in early gestational periods (11⁺⁰–25⁺⁰ weeks) [15,17,19]. While activin A stimulates trophoblast invasiveness [13], its upregulation in early or late gestation as a possible pathogenic mechanism in preeclampsia onset has been explored to a lesser extent.

A recent study conducted by Brkić et al. [66] revealed that overexpression of SMAD2, a canonical second messenger activated by activin A through ALK4 signalling, significantly suppressed the ability of trophoblasts to form endothelial-like networks conveying a potential link between elevated activin A levels and preeclampsia pathogenesis. Furthermore, as activin A is secreted into the systemic circulation, it plays an essential role in preeclampsia pathophysiology by prompting endothelial oxidative stress [67]. The underlying mechanism lies in SMAD2/3 phosphorylation, which increases the production of NADPH oxidase 2 (NOX2), a primary source of reactive oxygen species in the endothelial cells resulting in endothelial dysfunction [67]. Moreover, NOX2 expression seems inversely proportional to endothelial nitric oxide synthase (eNOS) expression; hence, indirect decrease of eNOS due to activin A elevation reduces vasodilator stimuli, favouring the onset of hypertension [67].

Other hypotheses suggest that activin A involvement in preeclampsia pathogenesis rely on endothelin-1, ICAM-1 and VCAM-1 overexpression, favouring hypertension and oedema [68]. Furthermore, antepartum activin A elevation has been directly associated with abnormal longitudinal strain, indicating myocardial dysfunction in women with hypertensive pregnancy disorders [69]. Interestingly, recent evidence suggests that prophylactic aspirin administration reduces maternal serum activin A, significantly improving global longitudinal strain and decreasing postpartum cardiac dysfunction in preeclamptic women [70]. Results from these studies suggest that elevated activin A levels might be one of the factors underlying preeclampsia pathogenesis and further cardiovascular risk later in life.

Maternal activin A levels were evaluated for their ability to predict preeclampsia in all trimesters of gestation. In the ROC curve analysis, activin A yielded an AUC that appeared to increase according to gestational weeks: in the first trimester an AUC of 0.670 was reported [16], between 16⁺⁰–30⁺⁰ weeks the AUC was 0.710 [19], and between 30⁺⁰–33⁺⁰ weeks an AUC of 0.722 was reported [17]. In addition, one study assessed maternal activin A levels near delivery (35⁺⁰–37⁺⁰ weeks), and an AUC of 0.710 was obtained [20]. Although no subtype of preeclampsia was clearly defined in the studies mentioned before, one study demonstrated that maternal serum activin A yielded an AUC of 0.890 in early-onset preeclampsia prediction at 22⁺¹–28⁺⁰ weeks of gestation. As such, maternal activin A could convey a potential complementary biomarker for preeclampsia prediction, particularly early-onset preeclampsia based on the obtained AUC value, when measured in the second and third trimesters. However, determining its predictive value with greater accuracy requires future investigation.

Inhibin A is expressed in both decidua [71] and trophoblast-derived tissues: syncytiotrophoblast [72] and cytotrophoblast [73]. During pregnancy, placental secretions are the primary source of maternal serum inhibin A [74], and as shown by Taché et al. [75], pregnancies with elevated inhibin A levels exhibit a higher risk of developing early-onset (RR: 5.0) and late-onset (RR: 2.3) severe preeclampsia. Supporting this evidence, the studies gathered in this review reported an increase of inhibin A in maternal serum levels between 10⁺⁰–22⁺⁰ weeks [22,24,25,26,31,34], and from week 25⁺⁰ until delivery [27,28,32,34] in pregnancies complicated with preeclampsia. While 12 out of 14 investigations disclose elevated inhibin A levels in maternal serum, Belovic et al. [30] and Kim et al. [33] reported no statistically significant difference between women who developed preeclampsia and those with healthy pregnancies. Overall, maternal inhibin A levels seem to be significantly augmented across the first, second and third trimester in preeclampsia.

As such, early elevations in inhibin A levels interfere with placentation in response to activin A signalling downregulation [76] (Figure 2), while later elevations are associated with an accelerated cytotrophoblast differentiation that results in villous cytotrophoblast depletion potentially leading to preeclampsia onset [77]. Disease pathogenesis and systemic manifestations in preeclamptic women could also be prompted by inhibin A disruption of maternal blood vessels diminishing the adaptability of the cardiovascular system to pregnancy [31] and reduction of placental irrigation aggravating placental ischemia [31]. Inhibin A is associated with placental hypoxia, an important trigger in preeclampsia pathogenesis, as shown by Depoix et al. [78] who found that hypoxia-inducible factors and endothelial PAS domain protein 1 (EPAS1) upregulate inhibin α subunit in human trophoblast cells.

Furthermore, inhibin A has been widely investigated as a possible biomarker for preeclampsia prediction due to its inclusion in non-invasive Down syndrome screening [79]. In the ROC analysis for preeclampsia prediction, inhibin A yielded an AUC of 0.780 at 11⁺⁰–13⁺⁶ weeks [25], an AUC of 0.590 at 12⁺⁰–18⁺⁰ weeks and 0.620 at 24⁺⁰–26⁺⁰ weeks [23]. Between 26⁺⁰–28⁺⁰ weeks of gestation, Keikkala et al. [34] showed an overall AUC of 0.717 for preeclampsia prediction, and when addressing preeclampsia subtypes, they found an AUC of 0.889 and 0.674 in early-onset preeclampsia and late-onset preeclampsia prediction, respectively. Notably, current guidelines advise preeclampsia prophylaxis with aspirin to be initiated before 16 weeks of gestation [80]. Hence, inhibin A elevation between 11⁺⁰–13⁺⁶ weeks [25], unlike activin A, could aid in the early identification of at-risk women, potentially enabling timely prophylactic therapy initiation.

The elevation of inhibin A in maternal serum has been identified as an independent risk factor for preeclampsia (OR: 7.45; CI 95%: 4.69–11.85) [31] and appeared to be associated both with late-onset preeclampsia (OR: 3.08; CI 95%: 2.15–4.42) and early-onset preeclampsia (OR: 4.17; CI 95%: 2.11–8.27) [21]. The likelihood of preeclampsia was also found to increase by 10% for every 0.1 ng/mL rise in maternal serum inhibin A [28]. Broumand et al. [29] assessed the predictive value of inhibin A levels ≥1.25 MoM in preeclampsia prediction at 16⁺⁰–20⁺⁰ weeks of gestation. They reported a sensitivity of 83.83%, a positive predictive value of 4.67%, a negative predictive value of 99.48%, a positive likelihood of 2.41, a negative probability of 0.24 and a diagnostic accuracy of 65.66%. As such, the authors stated that inhibin A levels were directly associated with the incidence of preeclampsia and its severity [29].

Within the family of follistatin-like proteins, follistatin-like 3 exhibits the highest homology to follistatin, differing only in length sequence and the presence of heparin-binding and Kazal-like domains [51]. In healthy pregnancies, follistatin levels increase significantly in all trimesters compared to non-pregnant women [54]. Additionally, placental tissues have shown an upregulated expression of follistatin-related proteins and β-glycans that serve as binding proteins for activins and inhibins, respectively [81] (Figure 2). According to the results derived from this study, follistatin decreases between the gestational weeks 25⁺⁰–40⁺⁰ [36,37] and 34⁺²–35⁺⁴ weeks of gestation [35], while follistatin-like 3 showed an increase in the late second (24⁺⁰–28⁺⁰ weeks) [39,40] and third trimester [42] but no significant changes were evidenced between 9⁺⁰–14⁺⁰ weeks [38,41].

Recent evidence shows that follistatin elicits a significant inhibitory effect on both trophoblast invasion and activin A secretion at 10⁺⁰–12⁺⁰ weeks of gestation [73], though the results of Nevalainen et al. [38] and Purut et al. [41] show no significant increase in preeclampsia within this period. The downregulation of TGF-β signalling by follistatin-related proteins, particularly activin A inhibition by follistatin and follistatin-like 3, requires a 2:1 ratio between the binding protein and the corresponding ligand [82]. Hence, in preeclamptic women, a reduction of follistatin-related proteins during the second and third trimester is observed, resulting in an increase in free activin A, leading to oxidative stress and endothelial dysfunction [82]. Placentation dysfunction mediated by follistatin-like 3 upregulation [83] has shown to be associated not only with activin A inhibition [84,85] but also with binding of GDF8/myostatin, which enhances extravillous trophoblast cells invasiveness [86]. In addition, follistatin-like 3 regulates trophoblast proliferation, apoptosis and lipid metabolism, suggesting that late elevations in follistatin-like 3 will favour pathogenic processes related to placental dysfunction; however, the signalling pathways involved remain uncertain [83].

Only one study has evaluated follistatin variation in mild preeclampsia prediction in the late second and third trimesters in the past decade, with an AUC of 0.771 when measured between the 25⁺⁰–40⁺⁰ weeks of gestation [37]. Follistatin-like 3, on the other hand, yielded lower values in the ROC analysis for preeclampsia prediction; AUC of 0.650–0.750 between 9⁺⁰–13⁺⁶ weeks of gestation [38], an AUC of 0.522 between 11⁺⁰–14⁺⁰ weeks of gestation [41] and an AUC of 0.706 between 24⁺⁰–28⁺⁰ weeks of gestation [39]. Follistatin-related proteins exhibit significant variation in maternal serum across pregnancies, though based on the results of this review, a substantial difference between preeclamptic and uncomplicated gestations is evident from the late second trimester. As such, early prediction might be hindered given the late upregulation and downregulation of follistatin-like 3 and follistatin, respectively. Moreover, the scarcity of longitudinal studies assessing their predictive value limits their utility as a biomarker, and further investigations are required to establish their role in preeclampsia prediction accurately.

As the studies included in this review evaluated different subtypes of preeclampsia, the heterogeneity of the population in each investigation undermines the generalisability of the results regarding the change in maternal serum levels of activin A, inhibin A and follistatin-related proteins in preeclampsia. In addition, although weeks of gestation are indicated in most studies, the wide intervals limit the accuracy in determining the period in which a significant change in maternal serum levels of these proteins occurs. Therefore, further studies with greater homogeneity and narrower intervals of weeks of gestation are recommended. Nevertheless, the results of this review allow us to estimate the variation of maternal serum activin A, inhibin A and follistatin-related proteins in preeclamptic pregnancies.