Intra-amniotic inflammation in the mid-trimester of pregnancy is a risk factor for neuropsychological disorders in childhood

Abstract

Objectives

Intra-amniotic inflammation is a subclinical condition frequently caused by either microbial invasion of the amniotic cavity or sterile inflammatory stimuli, e.g., alarmins. An accumulating body of evidence supports a role for maternal immune activation in the genesis of fetal neuroinflammation and the occurrence of neurodevelopmental disorders such as cerebral palsy, schizophrenia, and autism. The objective of this study was to determine whether fetal exposure to mid-trimester intra-amniotic inflammation is associated with neurodevelopmental disorders in children eight to 12 years of age.

Methods

This is a retrospective case-control study comprising 20 children with evidence of prenatal exposure to intra-amniotic inflammation in the mid-trimester and 20 controls matched for gestational age at amniocentesis and at delivery. Amniotic fluid samples were tested for concentrations of interleukin-6 and C-X-C motif chemokine ligand 10, for bacteria by culture and molecular microbiologic methods as well as by polymerase chain reaction for eight viruses. Neuropsychological testing of children, performed by two experienced psychologists, assessed cognitive and behavioral domains. Neuropsychological dysfunction was defined as the presence of an abnormal score (<2 standard deviations) on at least two cognitive tasks.

Results

Neuropsychological dysfunction was present in 45% (9/20) of children exposed to intra-amniotic inflammation but in only 10% (2/20) of those in the control group (p=0.03). The relative risk (RR) of neuropsychological dysfunction conferred by amniotic fluid inflammation remained significant after adjusting for gestational age at delivery [aRR=4.5 (1.07–16.7)]. Of the 11 children diagnosed with neuropsychological dysfunction, nine were delivered at term and eight of them had mothers with intra-amniotic inflammation. Children exposed to intra-amniotic inflammation were found to have abnormalities in neuropsychological tasks evaluating complex skills, e.g., auditory attention, executive functions, and social skills, whereas the domains of reasoning, language, and memory were not affected in the cases and controls.

Conclusions

Asymptomatic sterile intra-amniotic inflammation in the mid-trimester of pregnancy, followed by a term birth, can still confer to the offspring a substantial risk for neurodevelopmental disorders in childhood. Early recognition and treatment of maternal immune activation in pregnancy may be a strategy for the prevention of subsequent neurodevelopmental disorders in offspring.

Introduction

Intra-amniotic inflammation, i.e., the presence of an inflammatory signature in the amniotic fluid, is recognized by either an excessive number of immune cells [1, 2], such as neutrophils [3], [4], [5], [6], [7], [8], [9], [10] or lymphocytes [11], [12], [13], [14], [15], [16], [17], or inflammatory cytokines [14, 18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28], [29]. The inflammatory process can be caused by microorganisms (bacteria, viruses, fungi, and protozoa) [30], [31], [32], [33], [34] or by alarmins [35], [36], [37], [38], [39] released by host cells under conditions of stress or cell death. Intra-amniotic inflammation has been recognized in patients who have obstetrical complications such as preterm labor [1, 2, 8, 18, 30, 40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50], [51], [52], [53], [54], [55], [56], [57], [58], [59], preterm prelabor rupture of the membranes [60], [61], [62], [63], [64], [65], [66], [67], [68], [69], [70], [71], cervical insufficiency [72], [73], [74], [75], idiopathic vaginal bleeding [76], and a sonographic short cervix [77, 78]; however, in some cases, intra-amniotic inflammation is present in the absence of any complication of pregnancy and can be incidentally detected at the time of mid-trimester amniocentesis by measuring inflammatory cytokines [79].

Previous studies indicated that intra-amniotic inflammation is a heterogeneous condition and that perturbations of the cytokine network can occur in intra-amniotic infection as well as in sterile intra-amniotic inflammation [28, 80], [81], [82], [83]. In the mid-trimester of pregnancy, two subtypes of intra-amniotic inflammation have been described: one characterized by an elevation of interleukin (IL)-6 and another by an elevated concentration of C-X-C motif chemokine ligand 10 (CXCL-10), also known as interferon-γ-inducible protein 10 (IP-10) [79]. Of interest, patients with high amniotic fluid IL-6 concentrations are at risk for early preterm delivery, and their placentas typically show histological evidence of acute histologic chorioamnionitis [20, 24]. By contrast, those with elevated CXCL-10 are at risk for late spontaneous preterm delivery and show evidence of chronic placental inflammatory lesions, e.g., chronic chorioamnionitis, villitis of unknown etiology, and deciduitis of the basal plate [14, 84].

Based upon epidemiologic studies, fetal exposure to inflammation and infection during pregnancy has been implicated in the origin of mental illnesses, including cerebral palsy [85], [86], [87], [88], [89], [90], [91], [92], [93], [94], [95], [96], [97], [98], [99], [100], [101], autism spectrum disorder [102], [103], [104], [105], [106], [107], [108], [109], [110], schizophrenia [102, 103, 111], [112], [113], [114], [115], [116], [117], bipolar disorder [109], and depression [109, 118], [119], [120], [121]. A body of experimental evidence also supports the concept that infection or inflammation during pregnancy can induce fetal neuroinflammation, impair neurodevelopment, and predispose to adverse outcomes such as neurologic and mental disorders [110, 122, 123]. Although the literature sometimes refers to exposure to a maternal inflammatory response, it is likely that maternal inflammation operates through the induction of fetal neuroinflammation.

Intra-amniotic inflammation can be diagnosed as early as the mid-trimester, often occurs in the absence of a detectable clinical or subclinical maternal inflammatory response [79], and is frequently associated with delivery at term [79]. It is unknown if this condition increases the risk for adverse neurodevelopmental outcome. Therefore, the purpose of this study was to determine whether fetuses exposed to intra-amniotic inflammation in the mid-trimester of pregnancy are at risk for long-term neuropsychological disorders. We addressed this question by examining the outcomes of children whose mothers underwent a mid-trimester amniocentesis for clinical indications and in whom amniotic fluid analysis permitted a diagnosis of subclinical intra-amniotic inflammation.

Materials and methods

Study design

This retrospective nested case-control study was conducted to determine the likelihood of neurodevelopmental disorders in children exposed to intra-amniotic inflammation in the mid-trimester of pregnancy. The study is based on a cohort of 847 consecutive mothers who underwent a diagnostic mid-trimester amniocentesis for clinical indications. Exclusion criteria for this study were chromosomal anomalies and confirmed fetal anomalies. Amniotic fluid cultures for aerobic and anaerobic microorganisms, polymerase chain reaction (PCR) testing for eight viruses, white blood cell counts, IL-6 and CXCL-10 concentrations, and obstetrical outcomes of this cohort were previously reported and the specific assays for IL-6 and CXCL-10 described [79, 124].

Women had previously consented to donate samples of amniotic fluid and maternal blood for research purposes and extended their consent to the performance of neuropsychological testing in their children under a protocol approved by the Institutional Review Boards of the participating institutions (Azienda Ospedaliera Treviso, Azienda Ospedale/Universita’ Padova, Veneto Region, Italy). All patients provided written informed consent prior to the collection of samples. The hospitals have a Federal Wide Assurance with the Department of Health and Human Services of the United States.

Definition of intra-amniotic inflammation

Intra-amniotic inflammation was defined as either an IL-6 or a CXCL-10 amniotic fluid concentration above the 95th percentile of the normal values for a study population with uncomplicated pregnancies and term deliveries [79]. The cut-off values for IL-6 and CXCL-10 concentrations were 2,935 and 2,200 pg/mL, respectively. In the parent cohort, the prevalence of elevated amniotic fluid IL-6 was 6.3% (50/796) and the prevalence of elevated amniotic fluid CXCL-10 was 5.8% (45/770).

Selection of cases and controls

We selected as “cases” the children of the first 20 consecutive women with evidence of exposure to intra-amniotic inflammation in the mid-trimester according to the criteria outlined above. Twenty children of mothers without evidence of intra-amniotic inflammation, matched for gestational age at amniocentesis and at delivery (≤3 weeks), were selected as controls. Matching did not address the sex of the child. The sample size was set arbitrarily and therefore represents a sample of convenience. All amniotic fluid samples were considered negative for bacteria after testing by culture and molecular microbiologic methods (16 s broad primers). Moreover, samples were PCR-negative for eight viruses (adenoviruses, herpes simplex virus, varicella-zoster virus, human herpesvirus 6, parvovirus B19, human cytomegalovirus, enteroviruses, and Epstein-Barr virus).

Neurodevelopmental testing

Neurodevelopmental performance was assessed with a battery of tests that analyzed the cognitive and behavioral domains (Table 1). Neuropsychological testing was performed by two experienced psychologists who were masked to the results of the amniocenteses. Tests took approximately 90 min per child and involved both psychologists for each child. Neuropsychological dysfunction was defined as the presence of an abnormal score (<2 standard deviations) on at least two cognitive tasks. This approach provides dichotomous results, and it is considered useful to assess a wide range of impairments, which are then summarized in a single score. Table 1 provides a list of the neuropsychological tests [125], [126], [127], [128], [129], [130], [131], [132] and behavioral questionnaires [133], [134], [135] administered. Neuropsychological scores were transformed into age-corrected standardized scores by using published normative data.

Table 1:

The battery of tests performed to assess neuropsychological status and behavior.

Domain tested	Specific instrument used for evaluation
Abstract reasoning	Raven’s Coloured Progressive Matrices™ [125]
Language	Naming test [126]
Attention	Visual and auditory attention tests of the NEPSY-IIa [127, 128]
Memory
Short-term verbal and visual-spatial memory	Digit span test [126]
	Corsi block-tapping test [126]
Long-term verbal memory	Word’s list [126]
	List recall [126]
Executive function	Phonemic verbal fluency test [126]
	Tower of London test [129]
	Coding test of the WISC-IV [130]
	Stroop test [131]
	Backward digit span test [126]
Visual-motor function	Visual-motor integration test [132]
Social skills	Theory of mind and emotion, a sub-test of NEPSY-IIa [127, 128]
Behavior
	Child behavior checklist (CBCL) [133]
	Strengths and difficulties questionnaire (SDQ) [134]
	Conners’ rating scales-revised (CRS-R) [135]

^aA developmental NEuroPSYchological assessment. WISC-IV, Wechsler Intelligence Score for Children.

Statistical analysis

Descriptive statistics for continuous variables are displayed as medians and quartiles as well as mean and standard deviations (Table 2). Results of the neuropsychological and behavioral tests were age-corrected and converted to the following scores: z, equivalent, T, and standard as appropriate, by using published normative data [125, 126, 128], [129], [130], [131], [132]. The z score indicates the deviation from the mean population score, which is set to 0, standard deviation 1. A z score of −2 (or less) comprises 2.5% of the normal distribution and is considered significantly lower than average. Equivalent scores are a 5-point scale standardized after adjustment for age and education. An equivalent score of 0 is considered significantly lower than average. A T score indicates the deviation from the mean population score, which is set to 50, standard deviation 10. A T score of 70 (or more) is considered significantly higher than average. Scaled scores indicate the deviation from the mean population score, which is set to 10 (standard deviation 3). A scaled score of 4 (or less) is considered significantly lower than average. Scores were also converted in dichotomized outcomes (0: normal, 1: altered). Abnormal scores were those <5th percentile or 2 standard deviations. Pearson’s chi-square (X²) and Fisher’s exact tests were used for the dichotomous variable. All tests were 2-tailed, and a p-value <0.05 was considered statistically significant. Statistical analysis was performed with SPSS 20.0 (Chicago, IL). In addition, the effect of the amniotic fluid inflammation on the rate of neuropsychological dysfunction was also assessed by utilizing robust Poisson regression models and the geepack package in the R statistical language and environment (www.r-project.org).

Table 2:

Demographic, clinical, and laboratory characteristics of the study population.

Characteristics	Controls (n=20)	Intraamniotic inflammation (n=20)	p-Value
Maternal age, years	36 (35–37.2)	35.5 (34–37)	0.262
Nulliparity	35% (n=7)	30% (n=6)	1.0
Body mass index, kg/m2	22.8 (20.9–25.3)	23.5 (21.1–25.2)	0.818
Gestational age at amniocentesis, weeks	16.7 (16.2–17.2)	16.2 (15.9–18.1)	0.416
Amniotic fluid interleukin-6, pg/mL	465.3 (362.7–1182.2)	5690.4 (4707.5–8106.4)	<0.001
Amniotic fluid CXCL-10, pg/mL	616.1 (430.9–978.7)	3536.9 (2411.1–5194.3)	<0.001
Maternal blood interleukin-6, pg/mL	0.69 (0.56–0.89)	0.69 (0.64–0.90)	0.23
Maternal blood CXCL-10, pg/mL	88.31 (65.4–123)	92.39 (72.1–117.1)	0.69
Gestational age at test, weeks	16.5 (16.1–17.2)	16.15 (15.6–18.1)	0.768
Gestational age at delivery, weeks	39.6 (38.1–40.3)	38.6 (37.5–39.3)	0.123
Cesarean delivery	20% (n=4)	50% (n=10)	0.096
Birthweight, grams	3377.5 (3,135–3472.5)	3,145 (2,575–3,685)	0.417
Birthweight percentile for gestational age	55 (50–75)	50 (25–75)	0.509
Small for gestational age (birthweight below 10th centile)	0% (n=0)	5% (n=1)	1.0
Male neonate	40% (n=8)	70% (n=14)	0.111
Apgar score at 1 min<7	9 (9–9)	9 (9–9)	1.0
Apgar score at 5 min<7	10 (10–10)	10 (10–10)	0.979
Duration of hospitalization of the newborn, days	3 (3–4)	3 (3–4)	0.371

Data presented as median (interquartile range) and percentage (n/N); CXCL-10, C-X-C motif chemokine ligand 10.

Results

Characteristics of the study population

Demographic, clinical, and laboratory characteristics of the study population are displayed in Table 2. There were no significant differences between cases and controls for all of the examined variables except for IL-6 and CXCL-10, which, by design, were used to define cases and controls and for age at evaluation, which was solved by converting neuropsychological scores into age-correlated standardized scores by using published normative data. The adequacy of matching was evaluated with the standardized mean differences between the intra-amniotic inflammation and control groups, and there were no differences between them. Of the 20 cases with intra-amniotic inflammation, only four delivered preterm (<37 weeks of gestation). Apgar scores were above 6 in all cases and controls except for one child in the control group who was born at 28 weeks, had a birthweight of 850 g, and had Apgar scores of 2 (1 min) and 6 (5 min).

Intra-amniotic inflammation and neuropsychological dysfunction

Table 3 describes the clinical and laboratory characteristics of the women included in the study as well as pregnancy outcomes and results of neuropsychological tests of their children. Neuropsychological dysfunction was present in 45% (9/20) of children exposed to intra-amniotic inflammation but in only 10% (2/20) in the control group (p=0.03) (Table 4). Of the 11 children with neuropsychological dysfunction, nine were term deliveries, and eight of them had mothers with intra-amniotic inflammation. The relative risk (RR) of neuropsychological dysfunction conferred by amniotic fluid inflammation remained significant [aRR=4.5 (1.07–16.7)] after adjusting for gestational age at delivery. Some cognitive domains, such as reasoning, language, and memory, were not affected in the cases and controls. Neuropsychological tasks evaluating complex skills such as auditory attention, executive functions, and social skills were abnormal in children exposed to intra-amniotic inflammation.

Table 3:

Clinical and laboratory characteristics of the study participants, pregnancy outcomes, and neuropsychological tests results.

		Gestational age at amniocentesis, weeks	Amniotic fluid IL-6, pg/mL	Amniotic fluid CXCL-10, pg/mL	Spontaneous preterm labor with intact membranes or PPROM	Birth weight, grams	Birthweight percentile	Gestational age at birth, weeks	Gestational age at evaluation, years	Neuropsychological index	Neuropsychological domains affected
1	Case	18.2	4930.42	612.592	No	3,120	25	40 + 5	9.2	0
	Control	20.3	402.671	1584.97	No	3,350	25	41 + 3	9.5	0
2	Case	16.2	841.313	9100.3	No	2,600	25	36 + 2	11.3	0
	Control	16.5	125.62	366.846	No	2,800	10	41 + 0	11.5	0
3	Case	22.3	984.547	2303.03	No	3,800	75	40 + 4	9.5	0
	Control	22.1	456.444	1933.06	No	3,440	50	39 + 3	10.3	0
4	Case	15.5	4855.84	447.826	No	2,300	<3	38 + 6	9.1	1	Executive and social skills, attention
	Control	16.6	70.844	815.295	No	3,140	50	38 + 1	9.4	0
5	Case	15.2	6360.99	307.433	No	3,105	50	37 + 1	10.7	0
	Control	15.4	482.061	652.659	No	3,120	50	38 + 1	10.8	0
6	Case	15.6	6909.96	334.995	No	980	30	28 + 0	10.7	1	Executive functions, attention
	Control	17.6	172.11	534.117	No	850	25	28 + 4	8.8	0
7	Case	16.5	13142.7	1570.45	No	3,110	25	39 + 2	8.8	0
	Control	16.4	421.461	1233.48	No	3,330	30	40 + 5	9.2	0
8	Case	20.2	414.103	5152.2	No	3,170	50	38 + 2	9.1	1	Executive functions, reasoning
	Control	15.5	576.301	212.972	No	3,400	75	38 + 4	11.2	0
9	Case	15.6	3163.88	541.281	No	4,240	>95	39 + 2	10.4	1	Attention
	Control	16.2	242.731	404.357	No	3,390	75	38 + 6	11.4	0
10	Case	15.6	439.111	4105.66	No	3,680	75	38 + 2	9.11	0
	Control	15.5	425.053	538.109	No	3,300	50	39 + 5	10.9	0
11	Case	16.1	3183.68	667.341	No	3,560	75	40 + 0	10.8	0
	Control	16.5	1458.75	436.925	No	3,365	75	37 + 4	10.11	0
12	Case	15.3	110.547	2413.12	No	3,270	40	39 + 0	10	1	Executive and social skills
	Control	16.1	709.132	693.992	No	3,560	60	39 + 6	10.2	0
13	Case	15.6	7806.19	1057.8	No	2,610	10	39 + 1	11.2	0
	Control	16.2	2287.26	518.54	No	3,480	50	40 + 4	12	0
14	Case	18	528.102	2404.89	PPROM	2,360	25	35 + 4	9	0
	Control	17.1	690.765	1278.08	No	2,800	50	36 + 1	9.7	0
15	Case	16	5019.86	622.452	No	3,870	>90	38 + 0	11	1	Executive functions and attention
	Control	17.3	448.5	1649.09	No	3,470	60	39 + 6	11.7	0
16	Case	15.5	9007.19	N/A	No	2,410	10	37 + 4	11.3	0
	Control	16.4	2605.82	376.591	No	3,860	90	40 + 3	12.2	0
17	Case	16.6	4262.65	872.118	No	3,720	75	39 + 3	10.1	0
	Control	16.5	761.038	719.676	No	4,265	>90	40 + 2	10.5	0
18	Case	25.5	522.934	5320.62	No	3,700	90	39 + 3	11.2	1	Executive functions and attention
	Control	20.1	1423.45	878.888	No	3,680	75	40 + 1	10.8	1	Social skills
19	Case	19.4	293.863	2968.17	Spontaneous preterm labor with intact membranes	2,500	70	34 + 0	9.6	1	Social skills, abnormal lateralization
	Control	16.6	1745.63	452.248	PPROM	2000	75	31 + 6	9.4	1	Executive and social skills
20	Case	16.3	14008.8	2146.28	No	3,225	75	37 + 5	12.11	1	Executive functions
	Control	16	1089.98	387.847	No	3,450	75	39 + 0	12.6	0

IL-6, interleukin-6; CXCL-10, C-X-C motif chemokine ligand 10; PPROM, preterm prelabor rupture of the membranes; Neuropsychological index; 0, absence of neuropsychological dysfunction; 1, presence of neuropsychological dysfunction.

Table 4:

Frequency of neuropsychological dysfunction in children exposed and unexposed to intra-amniotic inflammation.

	Neuropsychological dysfunction
	Absent	Present	Total
Controls	18	2	20
Intra-amniotic inflammation	11	9	20
Total	29	11	40

Fisher’s exact test, p=0.03; Intra-amniotic inflammation was defined as an elevated IL-6 or CXCL-10 concentration in amniotic fluid.

Of the cases with intra-amniotic inflammation, 12 (12/20) had elevated amniotic fluid IL-6 and eight (8/19) had elevated amniotic fluid CXCL-10 (Table 3). Of the cases with neuropsychological dysfunction, 5/11 had elevated amniotic fluid IL-6, 4/11 had elevated CXCL-10, and 2/11 had both IL-6 and CXCL-10 within normal range at the time of mid-trimester amniocentesis (Table 3). Of note, of the 11 children who tested positive for a neuropsychological dysfunction, nine were male, and seven were delivered by cesarean section. The outcome of behavioral testing, as reflected by interviews of the parents, showed that behavioral modifications were observed in 30% (6/20) of the cases and in 20% (4/20) of controls (p-value>0.05).

Discussion

Principal findings of the study

The main conclusion of our study is that exposure to subclinical mid-trimester intra-amniotic inflammation is a risk factor for abnormal neurodevelopmental outcomes in children 8–12 years of age and that most pregnancies affected by intra-amniotic inflammation were delivered at term.

Results in the context of what is known

Intra-amniotic inflammation is observed in 5–6% of asymptomatic patients in the mid-trimester of pregnancy, and it is a risk factor for spontaneous preterm delivery [79]. However, most patients with this condition deliver at term after a pregnancy without known complications [79, 136, 137]. Herein, we report that exposure to sterile intra-amniotic inflammation during fetal life confers risk for abnormal neurodevelopmental outcomes in childhood.

Maternal immune activation due to infection or sterile inflammation has been implicated in the genesis of neurodevelopmental disorders [110,138], [139], [140], [141], [142], [143], [144], [145], [146], [147], [148], [149], [150], [151], [152], e.g., cerebral palsy [85], [86], [87], [88], [89], [90], [91], [92], [93], [94], [95], [96], [97], [98], [99], [100], [101], schizophrenia [102, 103, 111], [112], [113], [114], [115], [116], [117], depression [109, 118], [119], [120], [121], autism spectrum disorder [102], [103], [104], [105], [106], [107], [108], [109], [110], attention deficit hyperactivity disorder [110, 153], [154], [155], and Tourette syndrome [110]. Moreover, the spectrum of disorders may extend to learning disabilities [156], [157], [158] without demonstrable neurologic deficits.

The initial observations linking maternal infection and the subsequent occurrence of neurodevelopmental disorders in the offspring were made by Mednick et al. who reported that individuals diagnosed with schizophrenia were more likely to have been exposed to influenza during fetal life (ascertained by the diagnosis of maternal infection) [159]. This association was subsequently confirmed in prospective studies, and Brown et al. estimated that 14–21% of cases with schizophrenia were attributable to maternal infection [115, 160, 161]. The association was not specific to influenza and was observed with other viral infections such as rubella [162], [163], [164], [165], [166]. Since that time, a large body of epidemiological evidence has accumulated, indicating that maternal bacterial and viral infections during pregnancy can be associated with neurodevelopmental disorders [108].

Experimental studies provide strong evidence for the hypothesis that infections during pregnancy cause maternal and/or fetal immune activation, and this, in turn, can damage the developing brain [102, 142, 167], [168], [169], [170], [171], [172], [173]. During pregnancy, maternal exposure to bacterial endotoxin or lipopolysaccharide (LPS) or polyinosinic:polycytidylic acid (Poly I:C), a synthetic analog of double-stranded RNA that mimics some viral infections [172, 174, 175], can induce brain injury. These effects are mediated through an inflammatory response elicited by these microbial products and microorganisms. Further, there is evidence that fetal exposure to maternal sterile inflammation in autoimmune disorders [176], [177], [178], [179], [180], [181], [182], asthma [183], [184], [185], and obesity [186] can increase the risk of neurodevelopmental disorders.

Since maternal immune activation is associated with elevation in the concentrations of proinflammatory cytokines, these molecules have been studied and found to mediate the effects of maternal immune activation on fetal brain development [151, 161, 168, 187], [188], [189]. Specifically, IL-6 has been identified as a mediator of fetal brain injury, based on observations in pregnant mice in whom a single dose of IL-6 administered to the mother (on gestational day 12.5) led to behavioral deficits in the offspring, which can be prevented with the administration of anti-IL-6 [190]. Similar observations have been made with other cytokines [145, 172, 187, 188, 191], [192], [193], [194], [195], [196], [197], [198]. Recent studies in humans indicated a relationship between maternal IL-6 concentrations during pregnancy and neonatal functional connectivity [199], [200], [201], as assessed by magnetic resonance imaging (MRI), as well as infant cognitive development [199, 202].

Although a major emphasis has been the relationship between cytokine concentrations in the maternal compartment and neurodevelopmental outcome, it is possible and indeed likely that these effects are mediated through cytokines in the fetus. Indeed, amniotic fluid cytokine concentrations are thought to reflect fetal rather than purely maternal immune activation [203]. A series of studies using the Danish Historical Birth Cohort reported that individuals diagnosed with autism spectrum disorder had a higher mid-trimester amniotic fluid concentration of several analytes, e.g., monocyte chemoattractant protein-1 (MCP-1) [204], IL-4 [104], IL-10 [104], tumor necrosis factor-alpha (TNF-α) [104], tumor necrosis factor-beta [104], and matrix metalloproteinase-9 [205]. These observations are consistent with our report that individuals exposed to an inflammatory state during fetal life have a greater risk for abnormal neuropsychological testing in the future.

The observations in the present study are based on the results of the analysis of amniotic fluid obtained in the mid-trimester when genetic amniocenteses were performed. However, similar observations have been made between amniotic fluid concentrations of cytokines in the third trimester and the subsequent development of neurodevelopmental disorders. We reported that infants who develop white matter injury, diagnosed by ultrasound in the neonatal period, had higher amniotic fluid concentrations of IL-6, IL-1β, and TNF-α, and a subset of these infants developed cerebral palsy [86]. The same association was found between the umbilical cord concentrations of cytokines (reflecting fetal systemic inflammation) and the occurrence of white matter lesions [85]. Similar observations were reported by other investigators [93, 206], [207], [208]. Indeed, Lu et al. studied a group of patients who had an episode of premature labor and underwent amniocentesis <34 weeks of gestation. The amniotic fluid concentrations of IL-1β, IL-6, IL-8, TNF-α, granulocyte-colony stimulating factor, MCP-1, soluble intercellular adhesion molecule-1, and activin were significantly higher in cases subsequently diagnosed with brain injury by neurosonography or MRI. Similar findings were observed when examining umbilical cord blood, and this argues for the importance of fetal inflammation [209]. Indeed, there was high correlation between amniotic fluid and umbilical cord concentrations of IL-1β, IL-6, and TNF-α [209].

What is the relationship between a maternal immune response and a concentration of cytokines in amniotic fluid? Although the amniotic fluid and the maternal circulatory compartment are considered to be separate, further evidence indicated that systemic administration of LPS to pregnant rats results in an elevation in amniotic fluid concentrations of cytokines [210, 211] and also in a fetal inflammatory response, as reflected by the cytokine response in the brain as well as by changes in the fetal brain metabolome [212]. These observations are important because potential interventions aimed at maternal immune activation may target fetal inflammation and prevent brain injury. For example, the administration of N-acetyl-cysteine to pregnant rats prevents amniotic fluid [213, 214] and placental inflammation [213] and reduces cytokines in the maternal blood [213, 215], fetal blood [215, 216], and fetal brain [217]. Magnesium sulfate, currently used for neuroprotection, appears to exert its effects by downregulation of fetal inflammation [211], as demonstrated by the decrease in cytokine concentrations in the fetal brain after maternal exposure to endotoxin [218, 219]. After birth, treatment with minocycline [220], [221], [222], a microglia modulator, cyclooxygenase-2 (COX-2) inhibitors [223, 224], or N-acetyl-cysteine in dendrimers [225] can downregulate brain inflammation and improve neurologic or behavioral responses, suggesting that early treatment of neuroinflammation is possible.

Neurodevelopmental abnormalities

Among the cognitive domains, executive functions, attention, and social skills were the most affected in our cases. These high-order cognitive functions are strictly interrelated and overlapping. Executive functions have been strongly associated with academic functioning in childhood [226] and, in many cases, they are a better predictor of adult outcomes than intelligence quotient (IQ) and socio-economic-status [227]. Executive functions in childhood also predict the successful transition to kindergarten [228], school performance, and social competence in adolescence [229], and fewer drug-related problems and criminal convictions in adulthood [227]. Furthermore, abnormal social skills, attention, and executive functions are thought to be a core problem in several neurodevelopmental disorders, e.g., autism spectrum disorder [230], obsessive-compulsive disorder [231], and attention-deficit hyperactivity disorder (ADHD) [232].

These cognitive functions and some neurodevelopmental disorders, such as autism spectrum disorder, are considered disturbances of neuro-connectivity and are highly correlated to the integrity of the white matter [233], [234], [235]. The pathophysiologic mechanisms underlying white matter’s contribution to abnormal cognition and behavior are only partially understood but most probably involve the generation of reactive oxygen species that disrupt oligodendrocytes’ maturation and, ultimately, myelin formation [95, 167, 202, 236]. Given the importance of white matter in coordinating neuronal assemblies firing through myelination, even mild alterations in white matter structure may interfere with the cognitive performance [236, 237].

Clinical and research implications

The main clinical implication of this study is that a fraction of neurodevelopmental disorders appears to result from exposure to subclinical intra-amniotic inflammation in early pregnancy. The etiology, natural history, and consequences of mid-trimester intra-amniotic inflammation are unknown. Microbial invasion of the intra-amniotic cavity is a cause of intra-amniotic inflammation; however, it was not demonstrated in any of our cases. Sterile intra-amniotic inflammation has been attributed to alarmins, which are released under conditions of cellular stress [35], [36], [37], [38], [39, 71, 238]. Further work is required to determine if alarmins are elevated in the subset of patients with sterile intra-amniotic inflammation. Important clinical questions include whether it is possible to develop non-invasive methods to identify women with subclinical intra-amniotic inflammation as well as a method of treatment. Recent observations suggest that antimicrobial agents can be used to eradicate intra-amniotic inflammation even in the absence of intra-amniotic infection [239, 240].

Strengths and limitations

The strengths of this study include the ascertainment of intra-amniotic inflammation, which was objectively defined by the presence of an elevated concentration of inflammatory markers and by the performance of bacterial cultures and molecular tests to exclude bacterial infection as well as a subset of viruses. We also assessed two cytokines, IL-6, a biomarker of acute intra-amniotic inflammation, and CXCL-10, which reflects chronic inflammatory processes associated with maternal anti-fetal rejection. Another strength of this study is the long-term follow-up of children (8–12 years of age). The weakness is the limited sample size; however, this type of study is extremely difficult to conduct given that amniocentesis is less frequently performed now because of the preference for non-invasive prenatal testing for antenatal diagnosis.

Conclusions

Fetal exposure to mid-trimester intra-amniotic inflammation increases the risk of neurodevelopmental disorders even in infants born at term.