Chorioamnionitis and Early Childhood Outcomes among Extremely Low-Gestational-Age Neonates

Abstract

Importance

Chorioamnionitis is strongly linked to preterm birth and to neonatal infection. The association between histological and clinical chorioamnionitis and cognitive, behavioral and neurodevelopmental outcomes among extremely preterm neonates is less clear. We evaluated the impact of chorioamnionitis on 18-22 month neurodevelopmental outcomes in a contemporary cohort of extremely preterm neonates.

Objective

To compare the neonatal and neurodevelopmental outcomes of three groups of extremely-low-gestational-age infants with increasing exposure to perinatal inflammation: no chorioamnionitis, histological chorioamnionitis alone, or histological plus clinical chorioamnionitis.

Design

Longitudinal observational study.

Setting

Sixteen centers of the Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network.

Participants

2390 extremely preterm infants born <27 weeks' gestational age between January 1, 2006 and December 31, 2008 with placental histopathology and 18-22 months' corrected age follow-up data were eligible.

Main exposure

Chorioamnionitis

Main Outcome Measures

Outcomes included cerebral palsy, gross motor functional limitation, behavioral scores (according to the Brief Infant-Toddler Social and Emotional Assessment), cognitive and language scores (according to the Bayley Scales of Infant Development, 3^rd-Edition) and composite measures of death/neurodevelopmental impairment. Multivariable logistic and linear regression models were developed to assess the association between chorioamnionitis and outcomes while controlling for important variables known at birth.

Results

Neonates exposed to chorioamnionitis had a lower gestational age (GA) and had higher rates of early-onset sepsis and severe periventricular-intraventricular hemorrhage as compared with unexposed neonates. In multivariable models evaluating death and neurodevelopmental outcomes, inclusion of gestational age in the model diminished the association between chorioamnionitis and adverse outcomes. Still, histological+clinical chorioamnionitis was associated with increased risk of cognitive impairment as compared with no chorioamnionitis (Adjusted OR 2.4, [1.3- 4.3] without GA; Adjusted OR 2.0, [1.1-3.6] with GA as a covariate). Histological chorioamnionitis alone was associated with lower odds of death/neurodevelopmental impairment as compared with histological+clinical chorioamnionitis (Adjusted OR 0.68, [0.52-0.89] without GA; 0.66, [0.49-0.89] with GA). Risk of behavioral problems did not differ statistically between groups.

Conclusions and Relevance

Antenatal exposure to chorioamnionitis is associated with altered odds of cognitive impairment and death/neurodevelopmental impairment in extremely preterm infants.

Introduction

The increased survival of extremely-low-gestational-age neonates (ELGAN) has focused attention on the importance of assessing and improving long-term neurodevelopmental outcomes. It is estimated that as many as 11-15% of extremely preterm infants develop cerebral palsy (CP) and approximately half have other adverse outcomes including abnormalities in cognition, language development and behavior.^1-3 Chorioamnionitis and perinatal inflammation are thought to play a causal role in inciting preterm birth and white matter injury.^4-6 The association of chorioamnionitis and cognitive and neurobehavioral deficits remains unclear. Published studies in extremely preterm neonates offer inconclusive findings with reports suggesting associations with cerebral palsy and PVL^7-9, lower mental developmental index (MDI) scores¹⁰ and, alternatively, reports of no association between chorioamnionitis and neurodevelopmental outcomes.^11-14 Disparate results may be related to differences in the definitions of chorioamnionitis (histological vs. clinical), the gestational age range studied, the age at follow-up, the outcomes selected (e.g., brain lesions, neurological outcomes or composite death/impairment), and to issues of sample size and single versus multiple-center recruitment. GA is particularly important as it not only conveys maturity, but also conveys information on the immune response^15-18 and other developmentally regulated phenomena¹⁹; some studies have grouped ELGAN with moderate or late preterm infants.

In the present study, we evaluate whether histological and clinical chorioamnionitis are associated with increased risk of in-hospital morbidities, mortality and neurodevelopmental impairment (NDI) at 18-22 months' corrected age among extremely preterm neonates <27 weeks. At early gestations (prior to CNS myelination), inflammatory injury may be diffuse and result in unique pathological and clinical manifestations. Beyond periventricular leukomalacia and CP, we aimed to report on neurocognitive and behavioral outcomes. We hypothesized that (1) extremely preterm infants exposed to chorioamnionitis would have a higher risk for neonatal morbidities, mortality and poor neurodevelopmental outcomes compared to infants with no exposure, and (2) more severe, clinically manifest chorioamnionitis would be associated with worse outcomes.

Methods

This is a retrospective analysis of prospectively collected data on a cohort of extremely preterm neonates with and without exposure to histological or clinical chorioamnionitis. Preterm infants <27 weeks born between January 1, 2006 and December 31, 2008 at centers of the Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network (NRN) were included in the study if they had placental histopathology data and follow-up to 18-22 months' corrected age. Infants with congenital or chromosomal anomalies were excluded. Routine placental pathology is recommended for premature and high-risk deliveries by the Placental Pathology Task Force of the College of American Pathologists²⁰; and this recommendation is generally followed by the NRN centers (performed in 82% of inborn infants during the study period). Neonates were divided into three exposure groups: no chorioamnionitis, histological chorioamnionitis alone and histological+clinical chorioamnionitis. Clinically apparent chorioamnionitis was deemed the highest exposure group; a strong association with fetal inflammatory response syndrome has been reported by other investigators.^8,21 Data from the NICHD NRN Generic data base²² and Follow-up studies²³ were utilized. The studies were approved by the institutional review board (IRB) of each participating NRN center and written informed consent was obtained from the parents or guardians for longitudinal follow-up per requirement of the individual IRBs.

Study definitions

Neonatal and maternal data were collected prospectively for all live-born infants from birth until death, hospital discharge, transfer or 120 days, whichever occurred first.²² Infant follow-up data were collected during a comprehensive follow-up evaluation at 18-22 months' corrected age. Maternal and neonatal demographic characteristics included: age, race, prenatal care, maternal complications, preterm premature rupture of membranes (PPROM)>18 hours, antenatal antibiotics, antenatal corticosteroids, chorioamnionitis, Cesarean-delivery, birth weight, GA, gender, small for gestational age status²⁴, delivery room resuscitation, Apgar scores, cord pH and base deficit.

Histopathological chorioamnionitis was recorded if chorioamnionitis was noted on the placental pathology report or if acute, subacute chorioamnionitis or chronic chorionitis (as defined by the Stillbirth Collaborative Research Network Pathology Protocol²⁵) were documented. Assessment of histopathological chorioamnionitis was based on local reports made by individual site pathologists. Clinical chorioamnionitis (typically characterized by two or more of: maternal fever, uterine tenderness, malodorous amniotic fluid, maternal or fetal tachycardia or evidence of inflammation) also was noted if recorded in the mother's medical record by the treating clinicians and confirmed histopathologically. Cases of clinical chorioamninonitis without histopathological confirmation were excluded to avoid misclassification.

In-hospital morbidities included: mortality, early (EOS) and late-onset sepsis (LOS), bronchopulmonary dysplasia (BPD)²⁶, periventricular-intraventricular hemorrhage (PIVH)²⁷, ventriculomegaly, cystic periventricular leukomalacia (PVL), necrotizing enterocolitis (NEC)²⁸ and retinopathy of prematurity (ROP) stage 3 or greater as defined in prior publications.²²

As part of the NICHD NRN Follow-up Study, surviving infants underwent a comprehensive follow-up assessment at 18-22 months' corrected age performed by certified examiners trained to reliability on an annual basis.²³ Psychometric testing was performed using the cognitive and language subscales of the Bayley Scales of Infant and Toddler Development-III (Bayley-III).²⁹ A score of 100+15 on the Bayley III represents the mean +1SD; a score <70 represents 2SD below the mean. Children who were so severely developmentally delayed that they could not be assessed were assigned scores (54 for severe cognitive delay and 46 for severe language delay). Behavioral screening was performed using the Brief Infant-Toddler Social and Emotional Assessment (BITSEA) administered to the primary caregiver in the form of a structured interview.³⁰ The BITSEA is a nationally standardized behavioral screener used to determine the need for diagnostic assessment for socioemotional and behavioral problems. Total Problem Scores can be compared with specific percentile rankings of normative populations; low percentile rankings (≤25%) are associated with higher problem scores (externalizing, internalizing, dysregulation problems and behaviors seen in autism spectrum disorders).

CP was defined as a non-progressive CNS disorder with abnormal muscle tone in at least one extremity and abnormal control of movement and posture that interfered with age-appropriate activities.³ Disabling CP was classified based on the modified Gross Motor Function Classification System (≥Level II).³¹ Neurodevelopmental impairment was defined by one or more of: disabling CP, Bayley-III cognitive score <70, Gross Motor Function Level≥ II, blindness or permanent hearing loss that did not permit the child to understand or communicate despite amplification.

Statistical analysis

Children with and without exposure to chorioamnionitis defined histologically or clinically were compared with respect to maternal and neonatal baseline characteristics and outcome measures. Demographic characteristics were analyzed for the entire sample, as well as separately for infants excluded or lost to follow-up. For outcome measures, three exposure groups were compared: no chorioamnionitis, histological chorioamnionitis alone, and histological+clinical chorioamnionitis. Multivariable logistic and linear regression models were developed to assess the primary (death/NDI) and secondary outcomes adjusting for important covariates available at the time of birth that were selected a priori (maternal age, multiple birth, parity, antenatal steroids, maternal hypertension, antepartum hemorrhage, gender, GA, SGA status, insurance, race and center). Models including GA at delivery were compared to those without any GA at delivery variable considering the fact that while GA at delivery lies in the causal pathway between chorioamnionitis and outcomes, it also provides important information about the risks of adverse outcomes in extremely preterm newborns.

Results

3082 infants were assessed for eligibility and the following groups were excluded: infants with chromosomal or congenital anomalies (n=55), infants with ‘clinical chorioamnionitis’ with normal histopathology (n=82) and infants lacking placental histopathology (n=555). The final study cohort consisted of 2390 inborn infants.(Figure 1) Infants with and without placental histopathology had similar baseline characteristics except for 5-minute-Apgar score (Apgar score <3 occurred in 20.8% among those with placental histopathology vs. 26.4% without) and base deficit (4.99 mmol/L among those with histopathology vs. 4.28 mmol/L without). Mothers whose placentas were examined were more likely to have prepartum hemorrhage (22.9% vs. 17.6%), to receive antibiotics (66.9% vs. 58%) or antenatal steroids (75.4% vs. 66.9%), and to deliver by C-section (54% vs. 46.8%).(Table 1) Of the eligible participants, 38% were born to mothers with histological chorioamnionitis alone and 19% were born to mothers with histological+clinical chorioamnionitis. Primary outcome (death/NDI) was available for 2235 (93.5%) of all eligible infants.

Figure 1. Participants.

Table 1. Maternal and neonatal characteristics among neonates with and without placental pathology data.

Characteristics	Neonates with placental pathology	Neonates with missing placental pathology (excluded)	P-value
Maternal	N = 2390	N = 555
Mother's age, years	27.2 (6.42)	26.8 (6.39)	0.20
Mean (SD), Median (Q1, Q3)	27 (22, 32)	26 (22, 32)
Maternal Race/ethnicity, %
Black	39.5	36.3
White	35.7	37.8	0.44
Hispanic	19.8	21.5
Other	5.0	4.4
≥ 1 Prenatal visit, %	93.8	94.0	0.95
Diabetes mellitus, %	3.9	5.1	0.23
Parity, %
0 or 1	38.6	38.9	0.089
2 or 3	46.0	49.2
> 3	15.4	11.9
Hypertension, %
Chronic	8.3	8.0	0.54
Pregnancy-induced	9.8	11.3
None	81.9	80.7
Prepartum hemorrhage, %	22.9	17.6	0.0077**
PPROM >18h, %	25.7	25.6	0.10
Duration of PPROM, hrs	34.4 (103.2)	43.6 (152.7)
Mean (SD), Median (Q1, Q3)	0 (0, 8.33)	0 (0, 7.95)	0.67
Antenatal antibiotics, %	66.9	58.0	<.0001**
Antenatal steroids, %	75.4	66.9	<.0001**
Multiple birth, %	25.9	21.8	0.054
Cesarean delivery, %	54.0	46.8	0.0025**
Mother's education: high school graduate, %	74.2	72.0	0.41
Insurance, %
Medicaid	46.9	43.6	0.21
Private	39.5	40.4
Self/Other/Unins./Uknown.	13.6	16.0
Neonatal
Birth weight, %
401-500 g	12.9	113.9	0.34
501-750 g	54.1	50.6
751-1000 g	33.1	35.5
Gestational age, wk	24.3 (1.35)	24.3 (1.42)
Mean (SD), Median (Q1, Q3)	25 (23, 25)	24 (23, 26)	0.74
Male, %	51.4	54.6	0.19
SGA at birth, %	5.8	5.8	1.00
Delivery room resuscitation, %	71.4	67.2	0.053
Intubation, %	71.2	67.2	0.067
Epinephrine, %	5.8	6.9	0.41
Chest compression, %	10.0	9.2	0.62
Apgar score < 3, %
At 1 minute	43.5	47.2	0.13
At 5 minutes	20.8	26.4	0.0050**
Cord pH1	7.28 (0.12)	7.29 (0.12)
Mean (SD), Median (Q1, Q3)	7.31 (7.24, 7.36)	7.31 (7.24, 7.36)	0.47
Base deficit1, mmol/L
Mean (SD)	4.99 (4.98)	4.28 (4.34)
Median (Q1, Q3)	4 (2, 7)	3 (1, 5)	0.020**

Percentages were tested with a chi-square test. Medians were tested with a Wilcoxon rank sum test.

^**P-value is significant at alpha = 0.05 level of significance.

¹Cord pH and base deficit were found for the 1,648 that had cord blood gas (1,360 with placental pathology information, 288 without placental pathology information).

Mothers with histological or histological+clinical chorioamnionitis were more likely to identify with Black race, to have PPROM >18 hours, to deliver vaginally, to receive intrapartum antibiotics, and to be normotensive as compared with mothers without chorioamnionitis.(Table 2) Neonates exposed to antenatal chorioamnionitis had a lower GA and had lower Apgar scores as compared with unexposed neonates (despite a higher cord pH). A smaller proportion of neonates exposed to chorioamnionitis was SGA as compared to infants with no exposure.

Table 2. Baseline maternal and neonatal characteristics by exposure to chorioamnionitis.

Characteristics	No chorioamnionitis	Histological chorioamnionitis alone	Histological and clinical chorioamnionitis	P-value
Maternal	N = 1,014	N = 910	N = 466
Mother's age, years
Mean (SD), Median	27.2 (6.48)	26.9 (6.29)	27.8 (6.53)
(Q1, Q3)	27 (22,32)	26.5 (22,32)	27 (23,32)	0.12
Race/ethnicity, %
Black	35.4	42.2	43.2
White	39.4	34.7	29.9	0.0042*
Hispanic	20.6	18.4	20.9
Other	4.7	4.7	6.0
Education, % high school graduate	72.3	75.2	76.5	0.24
≥ 1 Prenatal visit, %	93.1	93.9	95.3	0.26
Diabetes mellitus, %	4.7	2.9	3.9	0.10
Hypertension, % Chronic	13.7	4.1	4.9
Pregnancy-induced	20.5	2.4	0.6	<.0001*
None	65.8	93.5	94.4
PPROM >18h, %	9.3	30.2	53.2	<.0001*
Duration of PPROM, hrs
Mean (SD),	18.5 (102)	36.4 (95)	70.8 (114)
Median (Q1, Q3)	0 (0, 0.05)	0.03 (0, 17.7)	12.9 (0.02, 94.4)	<.0001*
Antenatal antibiotics, %	50.6	73.4	89.7	<.0001*
Antenatal steroids, %	75.4	75.9	74.6	0.88
Cesarean delivery, %	67.9	45.1	41.2	<.0001*
Neonatal
Birth weight, %
401-500 g	14.0	11.1	13.7
501-750 g	54.6	54.4	52.4	0.25
751-1000 g	31.4	34.5	33.9
Gestational age, wk
Mean (SD)	24.6 (1.29)	24.2 (1.36)	24.1 (1.39)
Median (Q1, Q3)	25 (24, 26)	24 (23, 25)	24 (23, 25)	<.0001*
Male, %	53.5	49.0	51.7	0.15
SGA at birth, %	10.9	2.0	2.2	<.0001*
Delivery room resuscitation, %	72.0	71.0	71.0	0.85
Intubation	71.6	70.9	71.0	0.92
Epinephrine	5.3	5.3	8.0	0.09
Chest compression	9.5	9.8	11.6	0.43
Apgar score < 3, %
At 1 minute	40.5	41.8	53.1	<.0001*
At 5 minutes	17.3	21.1	27.9	<.0001*
Cord pH
Mean (SD),	7.26 (0.13)	7.30 (0.11)	7.29 (0.10)
Median (Q1, Q3)	7.29 (7.21, 7.35)	7.33 (7.26, 7.37)	7.31 (7.24, 7.36)	<.0001*
Base deficit, mmol/L
Mean (SD),	5.35 (5.56)	4.48 (4.18)	5.13 (4.88)
Median (Q1, Q3)	4.0 (2.0, 7.0)	3.0 (1.0, 6.0)	4.0 (2.0, 7.0)	0.08

Percentages were tested with a chi-square test. Medians were tested with a Wilcoxon rank sum test.

^*P-value is significant at alpha = 0.05 level of significance.

Of the 1376 infants exposed to any chorioamnionitis, 44.8% with histological chorioamnionitis died by 18-22 months' corrected age compared with 51.2% with histological+clinical chorioamnionitis. In comparison, 46.1% with no exposure to chorioamnionitis died. A greater proportion of infants born to mothers with histological or clinical chorioamnionitis had early onset sepsis (3.6% and 5.9% respectively vs. 1.3% without chorioamnionitis) and a greater proportion received IV antibiotics (55% and 76.5% vs. 45.3%). Adjusted risks of sepsis and antibiotic administration were elevated for the histological+clinical chorioamnionitis group as compared with the other two groups. Among infants with histological+clinical chorioamnionitis, adjusted risk of severe PIVH also was increased (p=0.033)(Table 3). No statistically significant differences in the risks of late-onset sepsis, BPD, central nervous system ventriculomegaly, cystic PVL, NEC or ROP were observed between groups.

Table 3. Adjusted neonatal and 18-22 months' corrected age outcomes by exposure to chorioamnionitis.

Outcomes	No chorioamnionitis	Histological chorioamnionitis alone	Histological and clinical chorioamnionitis	Adjusted4 P-value Histological chorioamnionitis alone vs None	Adjusted4 P-value Histological+Clinical Chorioamnionitis vs None	Adjusted4 P-value Histological alone vs. Histological+Clinical Chorioamnionitis
Neonatal/early death	N = 1,014	N = 910	N = 466
Death within 12 hrs, %	15.7	20.6	27.0	0.47	0.0097*	0.044*
Death prior to discharge, %	41.7	41.4	46.1	0.0023*	0.16	0.23
In-hospital outcome	N = 855	N = 723	N = 340
Early-onset sepsis, %	1.3	3.6	5.9	0.0035*	<.0001*	0.067
Physiologic BPD, %	56.0	54.1	56.8	0.43	0.99	0.50
PIVH grade 3 or 4, %	20.9	22.0	29.3	0.67	0.033*	0.0097*
(Grade 3)	(7.0)	(10.1)	(12.5)
(Grade 4)	(13.8)	(11.9)	(16.8)
Cystic PVL, %	4.6	4.8	6.1	0.77	0.53	0.37
Ventriculomegaly, %	6.0	4.4	5.8	0.24	0.90	0.39
Late-onset sepsis, %	43.7	47.3	41.3	0.15	0.50	0.057
Neonatal antibiotics, %	45.3	55.0	76.5	0.0049*	<.0001*	<.0001*
NEC, %	11.1	15.1	15.3	0.082	0.18	0.92
ROP stage 3 or greater, %	24.7	28.3	30.9	0.88	0.41	0.45
18-month outcome	N = 512	N = 473	N = 209
Cognitive score1 < 85	27.5	30.5	35.4	0.42	0.055	0.17
Cognitive score1 < 70	7.1	8.1	14.4	0.81	0.024*	0.029*
Cognitive score1 < 55	2.0	3.0	3.8	0.48	0.36	0.73
Cognitive score-
Mean (SD),	89.4 (14.4)	87.9 (13.6)	86.1 (16.2)
Median (Q1, Q3)	90 (80, 100)	90 (80, 95)	90 (75, 95)	0.45	0.0060*	0.027*
Language score1 < 85	50.2	52.6	52.4	0.50	0.92	0.65
Language score1 < 70	18.5	18.1	21.8	0.96	0.30	0.29
Language score1 < 55	4.6	6.0	7.3	0.34	0.26	0.73
Language score-
Mean (SD),	84.7 (17.4)	83.4 (16.7)	82.7 (18.7)
Median (Q1, Q3)	83 (74, 97)	83 (74, 94)	83 (71, 97)	0.70	0.38	0.22
GMF Level > II, %	6.3	6.6	9.6	0.58	0.52	0.24
Cerebral palsy, moderate to severe, %	4.5	4.9	8.1	0.50	0.38	0.12
Spastic diplegia, moderate to severe, %	0.6	1.1	1.4	0.70	0.53	0.33
Behavioral problems2
Rating < 26th pctile, %	39.3	43.6	36.9	0.44	0.39	0.12
Parental worry ≥ 3, %	7.9	10.8	6.7	0.11	0.70	0.10
Hearing impairment, %	1.2	2.1	1.9	NA	NA	NA
Vision impairment, %	1.8	1.3	0.5	0.32	0.15	0.36
NDI3, %	10.9	11.6	18.2	0.64	0.14	0.041*
	N = 950	N = 857	N = 428
Death prior to 18 months, %	46.1	44.8	51.2	0.011*	0.71	0.075
Composite outcome	N = 950	N = 857	N = 428
Death or NDI, %	52.0	51.2	60.1	0.011*	0.57	0.0061*

¹Scores from Bayley Scales of Infant Development, Third Edition.

²Scores from the Brief Infant Toddler Socioemotional Adjustment Scale (BITSEA). Parental worry ≥ 3 means the parent is either “worried” or “very worried” about the child's behaviors, emotions and relationships.

³Neurodevelopmental impairment.

⁴P-values for the significance of chorioamnionitis type were obtained from either logistic regression or linear regression models of the outcome in each row, adjusting for covariates for gender, antenatal steroids, small for gestational age (SGA), hypertension, and gestational age. A covariate for centers was added to models for death within 12 hours and death prior to discharge. The SGA covariate was excluded from the model for convergence reasons for early on-set sepsis. P-values for NDI, NDI or death, and death prior to 18 months were taken from the full models as described in the odds ratio tables for those outcomes. Because of sparseness in the outcome data, it was not possible to model hearing impairment or spastic diplegia. The p-value for hearing impairment is from a chi-square test. The p-value for diplegia is from an exact test.

^*P-value is significant at alpha = 0.05 level of significance.

Survivors to 18-22 months' corrected age included 512 infants whose mothers had no chorioamnionitis and 682 whose mothers had chorioamnionitis. Children with neurodevelopmental assessments at 18-22 months' corrected age and children lost to follow-up were comparable for all baseline characteristics except for mode of delivery and maternal age and race (proportionally more children born vaginally and proportionally more children born to younger mothers and mothers of Hispanic race were lost to follow-up).(Supplemental eTable 1) Of infants followed to 18-22 months' corrected age, infants exposed to histological+clinical chorioamnionitis had a significantly higher risk of cognitive impairment (defined as a cognitive score <70) as compared with the no chorioamnionitis (p=0.02) and histological chorioamnionitis alone (p=0.03) groups.(Table 2) Adjusted risks of CP were not significantly different despite increasing rates with increasing exposure to inflammation: 4.5% among unexposed infants, 4.9% among infants exposed to histological chorioamnionitis alone and 8.1% among those exposed to histological+clinical chorioamnionitis. Similarly, there were no significant differences in the risks of spastic diplegia, gross motor functional limitation, language delay or behavioral problems between groups.

On multivariable analyses, the association between chorioamnionitis and outcome was significantly impacted by inclusion of GA in the model. With GA as a covariate (along with other important covariates impacting outcome), the association between chorioamnionitis and adverse outcomes was diminished.(Table 4) Sequential analysis of each variable in the logistic regression models revealed that only GA impacted the association between chorioamnionitis and outcome in this manner. Compared with no chorioamnionitis, histological+clinical chorioamnionitis was associated with a heightened risk of low cognitive score <70 (Adjusted OR 2.38, [1.32-4.28] without GA; Adjusted OR 2.00, [1.10-3.64] with GA) and a decreased continuous Cognitive Score (β, -3.87 [-6.19 to -1.54] without GA and β, -3.25 [-5.56 to -0.93] with GA). Compared with histological+clinical chorioamnionitis, histological chorioamnionitis alone was associated with reduced odds of death/NDI (Adjusted OR=0.68, [0.52-0.89] without GA; Adjusted OR=0.66, [0.49-0.89] with GA) and NDI (Adjusted OR=0.57, [0.35-0.93] without GA; Adjusted OR=0.59, [0.36-0.98] with GA). Histological chorioamnionitis also was associated with reduced risk of death/NDI as compared with the no chorioamnionitis group, but only in models that included GA as a covariate (Adjusted OR 1.00, [0.80-1.26] without GA; Adjusted OR 0.72, [0.56-0.92] with GA). Direct comparison of the unadjusted rates of death and neurodevelopmental outcomes by week of gestation revealed the same patterns of association.(Supplemental eTable 2) The rate of death was lower in the histological chorioamnionitis group than in the no chorioamnionitis or histological+clinical chorioamnionitis groups for each week <27. The rate of cognitive impairment was highest in the clinical plus histological chorioamnionitis group for each gestational week at delivery and nearly two-fold higher than the other exposure groups overall. We further tested for an interaction between histological or clinical chorioamnionitis and GA in models for NDI or death, death (alone), and NDI (alone), and there was no significant interaction effect in any of the models. (The smallest interaction p-value was 0.19 and it occurred in the NDI model).

Table 4. Comparison of multivariable models including gestational age at delivery to those without any gestational age at delivery variable.

Death and neurodevelopmental outcomes by chorioamnionitis including gestational age
Outcome	Histo Chorio vs. No	Histo+Clinical vs. No	Histo vs. Histo+Clinical
Death/NDIa	0.72 (0.56, 0.93)	1.09 (0.80, 1.49)	0.66 (0.49, 0.89)
Deatha	0.72 (0.56, 0.93)	0.94 (0.69, 1.28)	0.76 (0.57, 1.03)
NDIa	0.89 (0.56, 1.42)	1.51 (0.88, 2.59)	0.59 (0.36, 0.98)
CPa	0.80 (0.42, 1.53)	1.39 (0.67, 2.87)	0.58 (0.29, 1.16)
Cognitive Scoreb	-0.72 (-2.59, 1.14)	-3.25 (-5.56, -0.93)	2.52 (0.29, 4.76)
<70a	1.07 (0.62, 1.85)	2.00 (1.10, 3.64)	0.53 (0.31, 0.94)
<85a	1.15 (0.82, 1.60)	1.50 (0.99, 2.28)	0.76 (0.52, 1.12)
Death and neurodevelopmental outcomes by chorioamnionitis without gestational age
Outcome	Histo Chorio vs. No	Histo+Clinical vs. No	Histo vs. Histo+Clinical
Death/NDIa	1.00 (0.80, 1.26)	1.47 (1.11, 1.93)	0.68 (0.52, 0.89)
Deatha	1.01 (0.81, 1.27)	1.32 (1.00, 1.74)	0.77 (0.59, 1.00)
NDIa	1.02 (0.65, 1.61)	1.79 (1.05, 3.04)	0.57 (0.35, 0.94)
CPa	0.89 (0.47, 1.69)	1.60 (0.78, 3.28)	0.56 (0.28, 1.11)
Cognitive Scoreb	-1.30 (-3.17, 0.57)	-3.87 (-6.19, -1.54)	2.57 (0.31, 4.82)
<70a	1.21 (0.71, 2.08)	2.38 (1.32, 4.28)	0.51 (0.29, 0.89)
<85a	1.25 (0.90, 1.74)	1.66 (1.10, 2.51)	0.75 (0.51, 1.10)

Abbreviations: NDI, neurodevelopmental impairment; CP, cerebral palsy

^aFor categorical outcomes, the adjusted OR (95% CI) from logistic regression models is presented

^bFor continuous outcomes (e.g., Cognitive score), the regression coefficient is presented (β, 95% CI) from linear regression models. This reflects the adjusted difference in score with exposure to chorioamnionitis. A positive parameter estimate indicates that the average score of the treatment in the first position is relatively higher compared to the average score of the treatment in the second position. A negative parameter estimate indicates that the average score of the treatment in the first position is relatively lower than the average score of the treatment in the second position.

Note: Death/NDI, Death, NDI, and Cognitive Score <85 were adjusted by the covariates used for finding adjusted odds ratios of the primary outcome. The remaining outcomes were adjusted by reduced models which contained covariates for center, gender, antenatal steroids, small for gestational age, and hypertension. Because of sparseness of some outcomes, it became necessary to combine some centers.

Discussion

Conflicting reports on neonatal outcomes following exposure to chorioamnionitis among extremely premature neonates makes prediction of outcomes difficult. This study evaluated the neonatal and neurodevelopmental outcomes of <27 week GA neonates exposed to no chorioamnionitis, histological chorioamnionitis alone and histological+clinical chorioamnionitis. Compared to unexposed neonates, neonates exposed to any chorioamnionitis had a lower GA and had higher rates of EOS and severe PIVH. In multivariable models evaluating death and neurodevelopmental outcomes, histological+clinical chorioamnionitis was associated with increased risk of cognitive impairment as compared with no chorioamnionitis. Histological chorioamnionitis alone was associated with lower odds of death/neurodevelopmental impairment as compared with histological+clinical chorioamnionitis. Histological chorioamnionitis alone also was associated with lower odds of death/NDI as compared with the no chorioamnionitis group, but only when GA at delivery was included in the multivariable model. Both composite and non-composite outcomes are presented as non-composite outcomes ignore censoring from infant deaths.

Our findings are consistent with the results of Hendson et al who reported an association between chorioamnionitis and lower Bayley-II mental-developmental-index scores.¹⁰ We reported on cognitive impairment specifically using Bayley-III cognitive scores. Histological chorioamnionitis was found to be protective for death/NDI with GA included as a covariate. As GA may lie in the causal pathway between chorioamnionitis and death and neurodevelopmental outcomes, one may argue whether adjustment for GA is appropriate. Regardless, other causes of prematurity may impact survival more profoundly (e.g., aberrant placentation, hypoxia-ischemia). Additionally, a regulated maternal or fetal immune response that triggers parturition and removes the fetus from an inflammatory intra-uterine environment may be protective.^32,33 Histological chorioamnionitis is consistently less detrimental than histological+clinical chorioamnionitis. In a study evaluating all fetal outcomes, Lahra et al reported a survival advantage for neonates with histological chorioamnionitis; surviving premature neonates <30 weeks were more likely to have histological chorioamnionitis or histological evidence of a fetal response as compared to fetuses who were stillborn or preterm neonates who died within 28 days.³⁴ We collected no data on fetal deaths.

Studies on chorioamnionitis in older GA neonates report a strong association between chorioamnionitis and white matter injury and CP^6,35 alluding to possible differences in critical GA windows or immune maturity.^15,18 Studies in very preterm infants report mixed results^{7,8,11-13,36,37} and are limited by retrospective assessment of outcomes, incongruent definitions of chorioamnionitis, single center design (necessitating study over many years, when perinatal care may have changed) and small or inadequate sample sizes over broader GA ranges. Few prospective studies have addressed the potential association of histological chorioamnionitis with the outcomes of extremely preterm neonates. One noteworthy exception is the study by Leviton et al that evaluated <28 week neonates (n=899 with placental data and neurological assessment at 24 months' corrected age).⁹ Histological chorioamnionitis, defined using stricter criteria than in our study, was associated with an increased risk of ventriculomegaly and diparetic CP. The risk of ventriculomegaly was further increased by recovery of a single or multiple placental organisms, whereas the risk of CP was not. We found no association of ventriculomegaly with chorioamnionitis, but measurements of ventricular size were not ascertained. Our study also found no statistically significant association with cerebral palsy, though the percentage of infants with CP increased with increasing severity of perinatal inflammation (4.5% with no chorioamnionitis vs. 8.1% with histological+clinical chorioamnionitis). Strunk et al. reported a reduced risk of late-onset sepsis following chorioamnionitis.³⁸ We found no difference in the rate of LOS among neonates with and without exposure to chorioamnionitis. Our study also evaluated behavioral outcomes based on screening using the BITSEA. Though we found no significant association with chorioamnionitis, this does not rule out possible effects that would be detected later in childhood or with other behavioral instruments.

Studies that assess chorioamnionitis without assessing fetal inflammatory response (including our own) lack important data in that fetal involvement is likely a more important predictor of neonatal outcomes than isolated maternal or intrauterine inflammation. Further, inflammatory biomarkers (e.g., measurements of cytokines, chemokines, damage markers) or amniotic or placental microbiological studies are an important omission. Early biomarkers of clinically significant disease may provide more rapid and accurate risk characterization. Though the NRN centers used the Stillbirth Collaborative Research Network Pathology Protocol to classify chorioamnionitis, the NRN registry provided no information on the details of these assessments. Additionally, histopathological assessments were performed by multiple pathologists, raising the possibility of inter-observer and center variability; we had no ratings of inter-rater reliability. Center variation also may have impacted the decision to obtain placental histopathology and the care provided. Ideally, placental histopathology would have been performed for all neonates born during the study period. Additionally, as in other studies targeting preterm infants, a healthy comparison group is lacking. Prematurity in the absence of histological chorioamnionitis is likely a result of other pathological processes that may alter the risk of death/impairment. Evaluation of inflammation over time (not only throughout gestation but in the postnatal period) is important as well.³⁹ Preclinical models have shown that the timing of an inflammatory exposure may alter the course of injury (e.g., preconditioning or sensitization).^40-43 Subacute exposure to lipopolysaccharide (LPS) 24h prior to a second noxious exposure lessened brain injury in a rodent model compared with exposure to the second insult alone. In contrast, very acute (6h prior) or remote exposures (72h prior) intensified injury following the second exposure.⁴⁰ It is possible that histological chorioamnionitis may correspond to this subacute exposure or that combining all forms of histological chorioamnionitis blurs the distinction between acute, sub-acute and chronic intrauterine inflammation. We have not considered the impact of postnatal inflammation.

The strengths of our study include the large sample size of 2390 neonates born prior to 27 weeks (including 1376 exposed to chorioamnionitis), the multicenter study design over a three-year period and the detailed follow-up and careful neurodevelopmental assessments that included cognitive and behavioral outcomes. Studying a narrow GA cohort of extremely preterm neonates allowed us to focus on the impact of chorioamnionitis on very immature neonates, while much of the literature has combined neonates of broader gestational ranges.

Conclusions

Antenatal exposure to chorioamnionitis is associated with altered odds of cognitive impairment and death/neurodevelopmental impairment in extremely preterm infants.

List of abbreviations

Bayley-III

Bayley Scales of Infant and Toddler Development-III

BITSEA

Brief Infant-Toddler Social and Emotional Assessment

BPD

bronchopulmonary dysplasia

BW

birth weight

CI

confidence interval

CNS

central nervous system

ELGAN

extremely low-gestational-age neonate

EOS

early-onset sepsis

FIRS

fetal inflammatory response syndrome

GA

gestational age

Histo

histological chorioamnionitis

PIVH

periventricular-intraventricular hemorrhage

LOS

late-onset sepsis

LPS

lipopolysaccharide

MDI

mental developmental index

NDI

neurodevelopmental impairment

NEC

necrotizing enterocolitis

NICHD

National Institute of Child Health and Human Development

NRN

Neonatal Research Network

OR

odds ratio

PDA

patent ductus arteriosus

PROM

premature rupture of membranes

PVL

periventricular leukomalacia

ROP

retinopathy of prematurity