Background

Healthy cognitive development in early childhood has a positive impact on an individual’s long-term life chances. Children with cognitive delays or cognitive impairments are more susceptible to enter negative developmental pathways, such as experiencing severe academic failure [1], depression and anxiety [2], low social development [3], and suicidal ideation [4].

The placenta, as the most important target organ connecting maternal–fetal communication, serves a key function in transmitting oxygen, exchanging nutrients and waste, and releasing hormones [5]. Optimal placental function is crucial for the trajectory of intrauterine fetal development, especially in the brain [6]. Placental inflammation is strongly implicated in fetal growth restriction, pre-eclampsia, miscarriage, and preterm delivery [7, 8]. Furthermore, placental inflammation may impact on fetal brain and neurological development, including early brain injury, pediatric stroke, and ventriculomegaly [9,10,11,12]. The pathway by which the placenta affects fetal brain development is known as the placenta–brain axis [13].

The elevated levels of circulating cytokines and their correlation with declined cognitive abilities in both individuals with schizophrenia and healthy controls suggest a possible contribution of inflammation to the exacerbation of neurocognitive dysfunction [14, 15]. Studies of the relationship between inflammatory biomarkers and cognitive developmental outcomes in children are receiving more attention [16,17,18]. Inflammation is a natural defense mechanism that allows body tissues to respond to potentially harmful stimuli [19]. During pregnancy, the placenta plays a crucial role in regulating the inflammatory process through immunomodulation and cell-to-cell communication between maternal and fetal tissues [20]. Animal models show that maternal immune activation-induced inflammatory processes have important negative effects on neuroplasticity and neurogenesis and are sufficient to impart lifelong psychiatric and neurologic disorders in offspring [21, 22]. Among the multiple inflammatory biomarkers of interest in the context of possible relationships with neurodevelopmental processes, the most frequently measured are cytokines [17]. Cytokines have a dual role the uterine immune response [23] and brain function [24, 25]. Cytokine imbalances can disrupt fetal development and chronically impair brain function, thereby may have a lasting effect on the neurological function of the offspring [26].

To our knowledge, there are studies that have focused on the links of maternal inflammatory cytokines during pregnancy with offspring’s neurodevelopmental outcomes, including cognitive performance [18, 27], neurodevelopmental delay [28], autism [29, 30], psychomotor development [31], and depression [32]. Although the findings of these studies are not consistent, it suggests that maternal biomarkers of inflammation such as IL-8, IL-6, IFN-γ, TNF-α, and CRP may be important biomarkers that affect children’s neurodevelopment [18, 27, 29, 31, 33]. However, only a few case series and case–control studies suggest that placental inflammation is related to early childhood neurological abnormalities, mainly manifesting as early brain damage [34]. Human data on the effects of inflammatory cytokines in utero on offspring’s brain and cognitive development remain scarce [34]. Moreover, the balance between pro- and anti-inflammatory cytokines has been identified as essential for individual’s normal cognitive function [35], and elevated production of inflammatory cytokines may be particular key mediators in modifying fetal brain development [36]. The integrated expression profile and interaction of pro/anti-inflammatory cytokines are critical for the normal function of the maternal and fetal immune systems [37].

Therefore, based on a Chinese birth cohort, we sought to test the association of placental inflammatory cytokines transcript expression and integrated index of pro- and anti-inflammatory cytokines with preschool children’s cognitive development and aimed to provide new clues to the impact of an adverse intrauterine immune environment on long-term cognitive development in children.

Methods

Study participants

This investigation was nested within the Ma’anshan Birth Cohort (MABC), which is a regional birth cohort study from Anhui, China. From May 2013 to September 2014, 3474 pregnant women who were first attending antenatal care at the Ma’anshan Maternal and Child Health Center in China were recruited.

The criteria for women’s inclusion included: ≥ 18 years of maternal age, within 14 weeks of gestation, living in Ma’anshan city for over 6 months, without mental illness, being able to understand and complete the questionnaires, and being willing to take part in follow-up assessments throughout the offspring’s childhood.

We have set up the following exclusion criteria: (1) adverse pregnancy outcomes, including embryonic arrest, stillbirth, spontaneous abortion, therapeutic abortion, and ectopic pregnancy (n = 162); (2) multiple pregnancies (n = 39); (3) no placenta samples after delivery (n = 754); (4) missing data on preschoolers’ cognitive developmental assessments (n = 850); and (5) maternal smoking during pregnancy (n = 4). Finally, 1665 mother–child pairs were included in current analysis. A detailed flow chart of the inclusion of mother–child pairs is shown in Fig. 1.

Fig. 1
figure 1

Flow chart of participants

Full size image

Collection and storage of placenta samples

Obstetricians/midwives collected placental tissues within half an hour after delivery (n = 2519). After washing the placenta with normal saline, an intact placental lobule free of maternal decidua, calcification, and fascia was extracted vertically from the full-thickness placentas in the position of 5-cm periumbilical and cut longitudinally into tissues smaller than ≤ 0.5 cm in size (each piece contained both maternal and fetal sides). The extracted sample was then immersed in RNAlater and saved at 4 °C overnight. The supernatant was drained the next morning, and the sample was transferred to a lyophilization tube and later stored at − 80 °C for testing.

Assay of placental inflammatory cytokine mRNA expression

We used real-time quantitative polymerase chain reaction (RT-qPCR) to measure the expression of placental cytokine mRNA, including interleukin-8 (IL-8), IL-6, IL-1β, tumor necrosis factor-α (TNF-α), interferon-γ (IFN-γ), C-reactive protein (CRP), IL-4, and IL-10. The detailed detection approach was described in our prior study [38].

The TRIzol reagent (MRC Inc., Cincinnati, OH, USA) was used to isolate total RNA. A Nanodrop® ND-1000 (Nano Drop, USA) was used to evaluate all RNA quality, and the AMV Reverse Transcription System (Promega, USA) was used to reverse-transcribe total RNA (1.0 g) into cDNA with a value of 260/280 ≥ 1.8 in accordance with the manufacturer’s instructions. A 480 SYBR Green I kit (Roche Diagnostics GmbH, Mannheim, Germany) was used for real-time PCR. The amplification reactions were carried out on a Light Cycler® 480II instrument (Roche, Germany).

Ninety-five degrees Celsius for 10 min and 45 three-step cycles (95 °C for 15 s, 60 °C for 15 s, and 72 °C for 20 s) were procedures for the real-time PCR. Although we used 45 cycles, test wells with a CT > 40 were disregarded (per MIQE guidelines). Primer 5.0 was used to create an RT-qPCR primer. Sequences of the oligonucleotides utilized in qRT-PCR are displayed in Additional file 1: Table S1.

The 18S rRNA, used as an endogenous reference RNA, was used to normalize all RT-qPCR data. Delta cycle threshold (ΔCt) refers to the difference between the target mRNA and the normalized RNA (ΔCt = Ct mRNA-Ct normalized RNA). The detailed QC measures for this study are shown in Additional file 1: Table S2.

Assessment of children’s cognitive development

The Wechsler Preschool and Primary School Scale of Intelligence, Fourth Edition of Chinese Version (WPPSI-IV CN) [39], which is extensively used in the field of cognitive function evaluation, was used to assess children’s cognitive development [40]. The following five subscales make up the WPPSI-IV: verbal comprehension index (VCI), visual spatial index (VSI), fluid reasoning index (FRI), working memory index (WMI), and processing speed index (PSI). The five domain subscales were used to determine the full-scale intelligence quotient (FSIQ).

Cognitive tests were administered to children aged 3.0–6.0 years by two professionally trained investigators in a quiet private room at the Ma’anshan Maternal and Child Health Center from June 2018 to January 2020. The age-standardized WPPSI-IV Chinese criteria were used to determine each child’s cognitive score. It is worth noting that two professionals taking the WPPSI were blinded to the placental cytokine data and important covariates.

Covariates

We identified the confounders using literature review [38, 41, 42] and directed acyclic graph (DAG) [43] (Additional file 1: Fig. S1). The confounders in the current analysis included maternal age, pre-pregnancy BMI, maternal intelligence quotient (IQ), parity, family monthly income per capita, maternal metabolic dysfunctions during pregnancy, maternal fever during pregnancy, maternal infection or inflammation conditions during pregnancy, maternal alcohol use during pregnancy, father’s education level, placental efficiency, and children’s sex. Detailed information on confounders is presented in Additional file 1: Table S3. Early childhood information, including exclusive breastfeeding in the first 6 months, main caregivers before 3 years, time spent in outdoor activities, and screening was collected by questionnaires during the childhood follow-up survey. These variables were used in the sensitivity analyses.

Statistical analysis

SPSS 23.0 software (IBM) and R software (version 4.2.1, R Core Team) were used for all statistical analyses, and P-values < 0.05 were defined as statistical significance.

To satisfy the requirements of normality and homogeneity of variance, the cytokine levels were translated to natural log-transform, which generated acceptable data skewness and kurtosis for further statistical analysis. For children’s and women’s Wechsler scale scores, the frequency distributions were moderately asymmetric, with skewness between − 1 and 1 for all variables and kurtosis between − 3 and 3 for all variables, indicating that the extreme values were not very different from those expected based on the normal data distribution. We used t-tests to compare the differences in cognitive developmental dimension scores between those with and without placental inflammatory cytokine data. Bivariate correlation using the Spearman correlation coefficient (r) was conducted to test the relationships between these placental cytokine expression measurements.

To fully explore the association between mRNA expression of inflammatory cytokines in the placenta and cognitive development in preschool children, we performed the following analyses.

First, the relationship between cytokines (IL-8, IL-1β, IL-6, TNF-α, CRP, IFN-γ, IL-10, and IL-4) and children’s cognitive score in each dimension (VCI, VSI, FRI, WMI, PSI, and FSIQ) was investigated separately using multiple linear regression models. The Benjamini–Hochberg procedure was used to adjust the false discovery rate (FDR) for multiple testing on eight inflammatory cytokines (IL-8, IL-1β, IL-6, TNF-α, CRP, IFN-γ, IL-10, and IL-4) [44].

Second, a significant interaction between inflammatory cytokines and sex was verified prior, and we further performed sex-stratified analyses in all adjusted models.

Third, we fitted two new indices, including an integrated index of anti-inflammatory cytokines (Index 1: IL-10 and IL-4) and an integrated index of pro-inflammatory cytokines (Index 2: IL-8, IL-6, TNF-α, IL-1β, CRP, and IFN-γ), to examine the association between integrated cytokines mRNA expression level and cognitive development. We summarized levels weighted by the number of cytokines in a given category based on the natural log-transformed values of each cytokine [the weights were 1/2 for cytokines Index 1 (n = 2) and 1/6 for cytokines Index 2 (n = 6). Then, multiple linear regression models were used to analyze the association between the two indices and children’s cognitive development scores.

Fourth, to examine the potential nonlinear relationship between placental inflammatory cytokines and children’s cognitive scores, a restricted cubic spline model was used to observe the association between each cytokine mRNA expression (IL-8, IL-1β, IL-6, TNF-α, CRP, IFN-γ, IL-10, and IL-4) and children’s cognitive development scores (VCI, VSI, FRI, WMI, PSI, and FSIQ).

Three sensitivity analyses were performed in all multiple linear regressions. First, considering that cognitive test age may affect children’s cognitive assessment performance, we further controlled for children’s age at cognitive testing to test the stability of the results. Second, breastfeeding duration [45], outdoor activities [46], screening [47], and main caregivers [48] may be important covariates that influence children’s cognitive development. Indeed, these factors will not confound the relation between exposure and outcome, but they are regarded as very important factors that related with children’s cognition function. Thus, we adjusted for these factors as precision variables to further explore the precision of the findings. Third, although existing studies of mRNA transcription support the use of 40 cycles or 45 cycles in human placental samples [49,50,51], we conducted further analyses of all data for Ct < 35 in each inflammatory cytokine, given the potential issues that might be posed by PCR Ct values all above 35 [IFN-γ was excluded because of its overall low expression in the placentas of this study, with high mean Ct (> 35)].

Results

Basic characteristics of included participants

In total, 1665 mother–child pairs were included in the analysis. Table 1 lists the basic characteristics of the included participants. The average maternal age was 26.4 years. Most of the pregnant women were primiparous (90.5%) and did not drink alcohol during pregnancy (92.4%), and the average maternal pre-pregnancy BMI was 20.9 kg/m2. The average gestational age was 39.1 weeks, and the average birth weight was 3383.5 g. The proportion of children born preterm was 3.2%. The mean age of the children’s cognitive assessment was 55.4 (± 6.8) months. Comparisons of the basic demographic characteristics of the included and excluded populations are shown in Additional file 1: Table S4.

Table 1 The basic characteristics of the included participants
Full size table

Distribution of placental cytokine expression and children’s cognitive performance

The Ct for 18S, IL-8, IL-1β, IL-6, TNF-α, CRP, IFN-γ, IL-4, and IL-10 are shown in Additional file 1: Table S5. All cytokines had a strong positive correlation between each other. The Spearman correlation coefficients for each inflammatory cytokine are shown in Additional file 1: Fig. S2.

The mean age of children was 55.4 months. The average scores (SD) of children in this study on VCI, VSI, FRI, WMI, PSI, and FSIQ were 109.0(12.7), 106.6(12.8), 104.8(11.4), 103.4(11.9), 101.9(10.9), and 107.8(11.2), respectively (Additional file 1: Table S6). There were no significant differences in FSIQ between children without placental data and children with placental data. The sub-dimensions of children’s cognitive scores in both groups are shown in Additional file 1: Table S6.

Associations between placental inflammatory cytokines expression and children’s cognitive performance

The associations between placental cytokine mRNA expression and cognitive test scores among preschool children are shown in Fig. 2. After adjusting for confounders, we observed significant associations between multiple placental cytokines mRNA expression and children’s cognitive scores (Table 2). For the pro-inflammatory chemokine IL-8, it was found to be negatively associated with preschoolers’ WPPSI scores on VSI (β =  − 0.57; 95% CI, − 1.01 to − 0.13), FRI (β =  − 0.63; 95% CI, − 1.05 to − 0.21), FSIQ (β =  − 0.53; 95% CI, − 0.92 to − 0.15). In pro-inflammatory cytokines, for IL-6, it was negatively associated with scores on FRI (β =  − 0.54; 95% CI, − 0.97 to − 0.11), WMI (β =  − 0.48; 95% CI, − 0.91 to − 0.05), PSI (β =  − 0.46; 95% CI, − 0.88 to − 0.04), and FSIQ (β =  − 0.58; 95% CI, − 0.97 to − 0.19). For TNF-α, it was negatively associated with scores on VCI (β =  − 0.44; 95% CI, − 0.82 to − 0.05), WMI (β =  − 0.39; 95% CI, − 0.76 to − 0.01), and FSIQ (β =  − 0.37; 95% CI, − 0.71 to − 0.02). For IFN-γ, it was negatively associated with scores on VSI (β =  − 0.35; 95% CI, − 0.69 to − 0.01), WMI (β =  − 0.44; 95% CI, − 0.75 to − 0.12), PSI (β =  − 0.32; 95% CI, − 0.63 to 0.00), and FSIQ (β =  − 0.31; 95% CI, − 0.61 to − 0.03). We did not observe any significant linear association between anti-inflammatory cytokines (IL-4 and IL-10) and preschoolers’ cognitive performance.

Fig. 2
figure 2

The associations between placental inflammatory cytokines mRNA expression (IL-8, IL-1β, IL-6, TNF-α, CRP, IFN-γ, IL-10 and IL-4) and children’s cognitive performance (VCI, VSI, FRI, WMI, PSI and FSIQ) by linear regression model

Full size image
Table 2 Multiple linear regression analysis of the relationship between placental cytokine mRNA expression and children’s cognitive performance
Full size table

When p-values were adjusted for multiple testing corrections, we still observed significant negative correlations of IL-8 (P-FDR = 0.02), IL-6 (P-FDR = 0.02), TNF-α (P-FDR = 0.04), and IFN-γ (P-FDR = 0.05) all with FSIQ. Furthermore, IL-8 was negatively associated with FRI (P-FDR = 0.02), IL-6 was negatively associated with FRI (P-FDR < 0.05), and IFN-γ was negatively associated with WMI (P-FDR < 0.05) (Fig. 2 and Table 2).

In addition, we added a model that forced all cytokines to be included in one model and found that IL-8, IL-6, and IFN-γ were all significantly associated with certain dimensions of cognitive development scores (Additional file 1: Table S7).

Sex-stratified associations between placental inflammatory cytokines expression and children’s cognitive performance

The association between placental inflammatory cytokines mRNA expression*sex and children’s cognitive performance is presented in Additional file 1: Table S8. The results of the analysis stratified by sex are shown in Table 2. After adjusting for confounders and performing multiple testing corrections, we observed that IL-6 was negatively associated with WMI in boys (β =  − 0.83; 95% CI, − 1.45 to − 0.20; P-FDR < 0.05) and VCI in girls (β =  − 0.86; 95% CI, − 1.44 to − 0.28; P-FDR < 0.05). IL-8 was negatively associated with VSI (β =  − 0.97; 95% CI, − 1.64 to − 0.31; P-FDR < 0.05) and WMI (β =  − 0.75; 95% CI, − 1.37 to − 0.14; P-FDR < 0.05) in boys. TNF-α was negatively associated with VCI in girls (β =  − 0.84; 95% CI, − 1.34 to − 0.33; P-FDR < 0.01). IFN-γ was negatively associated with WMI in boys (β =  − 0.57; 95% CI, − 1.03 to − 0.11; P-FDR < 0.05).

Associations between summary index of cytokines and children’s cognitive performance

Linear regression analyses of the association between the summary index of cytokines and preschoolers’ cognitive scores are shown in Table 3. It was observed that index 2 (integrated index of six pro-inflammatory cytokines) was negatively correlated with FSIQ (β =  − 0.65; 95% CI, − 1.16 to − 0.15; P = 0.01).

Table 3 Associations between placental summary index of cytokines and children’s cognitive performance
Full size table

Stratified analysis by sex showed that index 2 was mainly negatively associated with boys’ cognitive development (FSIQ, WMI, and PSI) (Table 3).

Test for non-linear relationship between placental cytokines mRNA expression and preschool children’s total cognitive performance

After adjusting for the confounders, we did not find any significant non-linear association of placental inflammatory cytokines mRNA expression (IL-8, IL-1β, IL-6, TNF-α, IFN-γ, IL-10, and IL-4) with FSIQ (Fig. 3), suggesting that these four cytokines (IL-1β, CRP, IL-10, and IL-4) did not have any significant associations (linear and nonlinear) with FSIQ in preschool children. In further exploring the nonlinear associations of placental inflammatory cytokine mRNA expression with cognitive subdimensions (VCI, VSI, FRI, WMI, PSI) in children, we only found significant nonlinear associations between CRP and PSI (P for nonlinearity = 0.011) (Additional file 1: Fig. S3).

Fig. 3
figure 3

Restricted cubic spline analysis of the association between placental inflammatory cytokine mRNA expression (IL-8, IL-1β, IL-6, TNF-α, CRP, IFN-γ, IL-10 and IL-4) and children’s FSIQ

Full size image

Sensitivity analyses

For sensitivity analysis 1, we found that index 2 was significantly and negatively correlated with both the FSIQ and each sub-dimension (VCI, VSI, FRI, WMI, PSI) (Additional file 1: Table S9). After adjusting for confounders and performing multiple comparisons for correction, in addition to the results of the main model remaining stable, we further found that IL-6 was negatively associated with VCI and WMI, and TNF-α was negatively associated with VCI and WMI (Additional file 1: Table S10). After analyses stratified by sex, we observed that IL-6 was negatively associated with FSIQ, VSI, WMI, and FRI in boys and FSIQ and VCI in girls. TNF-α had a negative correlation with WMI in boys and VCI in girls. IL-8 was found to be negatively associated with FSIQ, VSI, and WMI in boys. IFN-γ showed a negative association with WMI in boys (Additional file 1: Table S11).

For sensitivity analysis 2, stable results were observed compared to the primary analysis results (Additional file 1: Table S9-S11).

Sensitivity analyses 3 shows that when only data from Ct < 35 in the inflammatory cytokine assay were considered the results obtained remained stable after adjusting for confounding and multiple corrections (Additional file 1: Fig. S4).

Discussion

The primary purpose of the current study was to investigate the relationship between markers of placental inflammation and children’s cognitive development. The results of the study support a possible linear association between placental immune activity and offspring cognitive development. Negative correlation between preschooler’s FSIQ and placental mRNA expression of four inflammatory cytokines (IL-8, IL-6, IFN-γ, TNF-α) was observed.

The placenta acts as the main barrier between the fetus and the mother and serves a key role in maintaining immune homeostasis at the mother–fetus interface and in utero [52]. The source of placental inflammatory cytokines may be either from the placental tissue itself or from the mother [53, 54]. Firstly, the placenta creates an immune environment that supports pregnancy, containing immune cells that promote fetoplacental growth and protect both the mother and fetus from external challenges [55]. Increased placental cytokine release may help to activate particular inflammatory pathways necessary to cause maternal insulin resistance, which is necessary for the normal development of pregnancy [56] and may also be a protective mechanism in the face of several adverse factors (trauma, hypoxia/ischemia, metabolism/obesity/diet/diabetes, stress, and pollution) [57, 58]. Numerous cytokines expressed in the placenta can play a significant role in modulating the maternal–fetal immune interface to control the mother from rejecting the fetus [59]. Secondly, maternal factors such as infection or disease are the most common causes of placental inflammation include maternal infections that activate the maternal immune system to release pro-inflammatory cytokines such as IL-6 and TNF-α, which can cross the placenta directly into fetal circulation [30]. Placental inflammation has been proven in animal models to stimulate fetal endogenous cytokine production [52, 60]. Peripheral cytokines can interact with the brain and central nervous system in several ways, such as through blood–brain barrier cells that release immunoreactive chemicals, diffusion from circumventricular organs, cytokine transporters, or cranial nerves [17]. It has been shown that cytokines and chemokines are present in human fetal forebrain cells as early as 5 weeks of gestation [61]. Although it is not clear whether cytokines in fetal brain are endogenously released, or diffusion of peripheral cytokines through blood–brain barrier, cytokines in the fetal brain may have direct effects on offspring’s later cognitive development.

We found that placental IL-6 mRNA expression was negatively associated with total cognitive scores and multiple sub-dimensional scores in preschoolers. The association remained significant at the FSIQ and FRI, even after multiple corrections were made. When further adjusted for cognitive test age, IL-6 was found to present in negative correlation with FSIQ, VCI, FRI, and WMI. Studies have shown that the pro-inflammatory cytokine IL-6 has the ability to penetrate both the placental and blood–brain barriers [62, 63]. Elevated levels of maternal IL-6 have been shown to trigger inflammatory processes in the fetal brain [64] through direct placental transfer to the fetal compartment [65] or indirectly via placental inflammation [60]. According to a longitudinal study conducted at the University of California, there exists a negative correlation between maternal IL-6 concentrations during pregnancy and FRI in children [42], which further identified pars triangularis volume as an important mediator of the relationship between maternal IL-6 concentrations and offspring FRI [42]. Other studies by the group also demonstrated that maternal IL-6 concentrations were negatively associated with working memory performance and impulse control at age 2 years, as well as cognitive ability in offspring at 12 years of age [66,67,68]. Other epidemiological studies have also suggested that increased maternal IL-6 concentrations are correlated with poorer cognitive ability in infants [27]. Thus, the effects of IL-6 on cognitive development are not only implicated in working memory and fluid reasoning but may also have important implications for verbal comprehension.

IL-8 is a well-known circulating inflammatory cytokine and a mediator of the systemic inflammatory response that plays multiple roles in the brain, including roles in neurogenesis, synaptic plasticity, and neurotransmission [69, 70]. Several epidemiological studies have found that increased maternal prenatal IL-8 is related to altered brain structure and elevated risk of psychiatric symptoms in offspring [71,72,73]. Dozmorov et al. [41] found that maternal prenatal plasma IL-8 was negatively associated with child spatial abilities but positively associated with verbal abilities. Some studies also suggest that IL-8 also has positive effects on child development. For example, higher child self-regulation is found to be associated with the concentration of IL-8 in maternal serum [74]. Also, elevated gestational levels of IL-8 were linked with enhanced performance on both the Drawing task and Tactile Finger Recognition Task [18]. Gilman found that reduced maternal IL-8 was correlated with the occurrence of neurologic abnormalities [75]. In the current study, IL-8 was found to be strongly correlated with FSIQ and FRI, even after multiple corrected comparisons. It was inconsistent with the current study, as probably because the neuroprotective or neurotoxic effects of IL-8 in the brain depend on the levels, conditioning, and co-occurrence of other cells or molecules.

Meanwhile, placental TNF-α and IFN-γ mRNA expression was found to be negatively associated with preschool children’s FSIQ. Current epidemiological studies also support that these two cytokines may play an important role in children’s cognitive development, whether during pregnancy or at birth or in childhood [16, 18, 27, 76]. The Collaborative Perinatal Project study found that children with higher maternal TNF-α levels at the second and third trimesters of pregnancy had lower IQ, reduced cognitive performance, and increased problem scores at age 7 years [18]. Another prospective cohort study found that IFN-γ in cord blood was associated with a reduced risk of having a low performance intelligence quotient at age 5 years and that TNF-a was protective against low verbal intelligence quotient in preterm infants [76]. Rhea study finds that children with high serum TNF-α and IFN-γ have decreased score on memory performance at preschool age [16]. A study based on children with autism showed that IFN-γ levels in children were negatively associated with WMI [77]. After further adjusted for children’s age at cognitive testing, we similarly found that TNF-α was negatively both VCI and WMI, as well as IFN-γ was negatively correlated with WMI, which supports the important role of these two cytokines in memory development or verbal comprehension in children.

We did not find significant relationships between IL-1β, CRP, IL-4, and IL-10 and preschoolers’ cognitive functions after multiple corrected comparisons. Some epidemiological studies likewise did not find any association [16, 18, 27, 41, 76, 77]. However, Dozmorov et al. [41] found that high maternal plasma IL-1β in the first two trimesters of pregnancy was associated with low general conceptual, nonverbal, and spatial ability. Nazzari et al. [78] found that serum CRP levels at 30–33 weeks of gestation were associated with low infant cognitive performance. Krakowiak et al. [79] found that neonatal IL-4 was associated with increased odds of mild/moderate autism ASD and that high IL-4 was negatively related to nonverbal cognitive abilities in male subjects with ASD. In a cohort study of preterm infants, top quartile concentrations of IL-4 in the first month of life were associated with an elevated risk of low IQ and low processing speed at age 10 years [80]. More prospective studies are necessary to support evidence on the association of maternal prenatal IL-1β, CRP, and IL-4 levels with offspring’s cognitive development and the possible sensitivity period of the effect. It should be noted that the transcripts corresponding to the primers for IL-10 in this study contain both coding (NM_001382624.1, NM_000572.3) and non-coding RNAs (NR_168466.1), and non-coding transcripts have the potential to have a significant impact on many facets of RNA biology, including splicing and modification. Although this study did not find a significant correlation between children’s cognitive development and the transcript levels of IL-10, the association with each of the transcripts of IL-10 as well as the protein expression need to be further explored.

The inflammatory response is a complex physiological process involving the interaction of multiple inflammatory cells, molecules, and pathways. Armstrong-Wells et al. [11] found that placental inflammation on the fetal side was associated with increased maternal IL-6 and IL-8 at delivery, as well as increased fetal IL-6, IL-8, and TNF-α. They suggested that IL-8 and IL-6—with funisitis—appear to be particularly important in the fetal inflammatory response and children’s neurological outcome [11]. Based on the function of inflammatory cytokines in the inflammatory response, we exploratively fitted the integrated anti-inflammatory index (index 1) and integrated pro-inflammatory index (index 2). After adjusting for potential confounders and children’s age at cognitive testing, we found that the index of integrated six typical pro-inflammatory cytokines (IL-8, IL-6, TNF-α, IFN-γ, IL-1β, CRP) was negatively associated with all cognitive developmental dimensions of WPPSI, and this association was stronger than for individual cytokines. This may be due to the synergistic effects among inflammatory cytokines on cognitive function. The findings would highlight the importance of comprehensive assessment of multiple classes of cytokines and the possible pathways in influencing individual’s cognition.

During inflammation process, both mRNA expression and protein expression of inflammatory cytokines change, and the expression levels of transcriptional biomarkers can indirectly reflect protein expression to some extent. However, it should be acknowledged that the association between mRNA and protein expression levels of inflammatory cytokines is not invariable and may be influenced by various factors, such as post-translational modifications that lead to translation inefficiency. However, mRNA assays are more feasible than protein levels in large-sample, multicytokine assays. In addition, at the transcriptional level, differences based on sex have been observed in the human placenta, involving genes that are associated with placental development, maintenance of pregnancy, and maternal immune tolerance for the fetus [81]. After stratifying by sex, we observed that IL-8 and IFN-γ were negatively associated with cognitive scores mainly in boys. The pro-inflammatory index similarly showed a significant association with cognitive scores (FSIQ, WMI, PSI) in boys, suggesting that there may be sex differences in the effects of different inflammatory cytokines on children’s cognitive development. The findings of the present study further add to the evidence that sex-specific susceptibility to poorer neurodevelopmental outcomes originates in the fetus. Existing studies similarly support the important role of the placenta as an organ in producing a sexually dimorphic response to intrauterine pressure exposure [82, 83]. Trophoblast cells of placenta or chorion from male fetuses in pregnancy produced higher levels of TNF-α after LPS stimulation, and produced lower levels of IL-10 and G-CSF than those from female fetuses, suggesting that in the presence of a male fetus, trophoblast cells have likely to produce a more pro-inflammatory environment [84]. It should be clarified that the internal reference gene 18S selected for this study covers a sequence on the Y chromosome that was updated for the latest study. However, by comparing the Ct amplified by 18S, we found that the Ct was essentially the same for males and females, suggesting that the amplification of this primer sequence is stable between the sexes (Additional file 1: Table S12).

Results from restricted cubic spline model demonstrated a similar linear or monotonic association of inflammatory cytokine mRNA expression (IL-6, IL-8, TNF-α, IFN-γ) with cognitive development levels, suggesting that this association exhibits a relatively stable dose–response relationship. It is important to note that although this study was set up with 45 cycles of the real-time PCR, test wells with a Ct > 40 were disregarded (per MIQE guidelines) [85]. Existing studies also support the use of 40 or even 45 cycles in qPCR assays in human placental samples [49,50,51, 86]. Meanwhile, we separately analyzed the associations of samples with Ct < 35 for IL-1β, IL-8, IL-6, TNF-α, CRP, and IL-4 with cognitive development in children and found that the results were consistent with those of the main analyses. We note that of the eight inflammatory cytokines only IFN-γ had a mean Ct > 35, which also suggests that the mRNA expression corresponding to this target gene was low overall in the placental samples studied. During early pregnancy, IFN-γ production is mainly limited to chorionic villous [87], but both IFN-γR1 and IFN-γR2 proteins are expressed in placental trophoblasts throughout human pregnancy [88]. Once the pregnancy is established and enters the second and third trimesters, IFN-γ production decreases; Thus, reduced IFN-γ secretion by late placental trophoblasts is physiologically required to avoid uncontrolled invasion [88].

Placental inflammation may mediate the effects of prenatal adverse pregnancy environment on cognitive development. Maternal environment during pregnancy such as infection, metabolic dysfunction, high maternal BMI, family economic status, and alcohol consumption can have potential effects on the developing fetal brain and long-term cognition [89,90,91,92,93]. These prenatal adverse conditions affect the function of the placenta in regulating nutrient transport, endocrine function, and immune tolerance [94, 95] and thus its involvement in fetal growth restriction, hypoxia, and associated cognitive and brain development [96,97,98,99]. It was found that the relationship between maternal obesity and offspring visuomotor ability may be influenced in part by maternal inflammation [100]. The current study has adjusted these important factors to reduce the potential confounding effect of the association of placental inflammatory factors with cognitive development.

It is important to recognize that there exist certain limitations that need to be acknowledged. Firstly, we only assayed placental cytokines at the mRNA expression and did not validate them at the level of corresponding protein expression. The results found in this study may not be fully applicable directly to proteomics. Second, the placental samples tested in the present study included both maternal and fetal surfaces and could not further explore whether the association of inflammatory burden on different surfaces of the placenta with cognitive development in children differs and the possible mechanisms of bidirectional transfer. Third, although our participants did not include pregnant women with mental disorders, we did not collect the detailed information on children’s family history of neurodevelopmental disorders. However, it was shown that the point prevalence of mental disorders in China was 1.1% in 2013 [101]. Fourth, although rigorous quality control and FDR correction for multiple testing were implemented, there is still the possibility of spurious association estimates due to possible measurement errors introduced by not performing repeated measurements of qPCR. We encourage future studies to perform more stringent quality control and to perform at least two repeated measurements to validate our conclusions. Fifth, this present study is only an exploratory study conducted in China. We used WPPSI to assess children’s cognitive development. The findings would be further validated by different cognitive assessment instruments and repeated tests at later multiple ages are encouraged to be performed to observe the long-lasting effect of placental cytokines. In addition, there are some drivers in early childhood that may influence cognitive development have not been collected, such as in-home nurturance, stimulation, peer contact, the quality and duration of preschool and kindergarten programs. Therefore, we cannot further explore the potential modifying effects of these important early childhood covariates.

Conclusions

In conclusion, this prospective study suggests that placental pro-inflammatory milieu may have long-term effects on children’s cognitive development. High mRNA expression of pro-inflammatory cytokines (IL-8, IL-6, TNF-α, IFN-γ) may be early predictive markers of low cognitive function in children in utero. Future studies are needed to validate our findings.