LIMITING THE EXPOSURE TO INTRA-UTERINE INFECTION/INFLAMMATION OF SELECT FETUSES IMPROVES SHORT-TERM NEONATAL OUTCOMES IN PRETERM PREMATURE RUPTURE OF MEMBRANES

Abstract

Introduction

To improve neonatal outcomes in pregnancies at heightened risk for early-onset neonatal sepsis (EONS), there is a need to identify fetuses that benefit from expectant management as opposed to early delivery. Detectable haptoglobin and haptoglobin-related protein (Hp&HpRP “switch-on” status) in cord blood has been proposed as biomarker of antenatal exposure to infection and/or inflammation (IAI), an important determinant of EONS.

Materials and Methods

We analyzed 185 singleton newborns delivered secondary to preterm premature rupture of membranes (PPROM). In 123 cases, amniocentesis was performed to exclude amniotic fluid (AF) infection. Delivery was indicated for 61 cases with confirmed infection. Women without AF infection (n=62) and those without amniocentesis (n=62) were managed expectantly. IL-6 and Hp&HpRP switch-on status were evaluated by ELISA and Western blot. Newborns were followed prospectively for short-term outcomes until hospital discharge or death.

Results

Newborns exposed antenatally to IAI had an increased risk of adverse neonatal outcome (OR: 3.0 [95% CI 1.15–7.59]). Increasing gestational age (OR: 0.61 [95%CI: 0.52–0.70]) and management with amniocentesis (OR: 0.37 [0.14–0.95]) lowered the newborn’s risk of developing adverse outcomes.

Discussion

In the setting of PPROM and IAI, early delivery benefits a select subgroup of fetuses that have not yet progressed to Hp&HpRP switch-on status.

INTRODUCTION

The prevalence of early-onset neonatal sepsis (EONS) is estimated at approximately 11.0 of 1,000 very low birth weight (400g-1,500g) and 1.4 of 1,000 low birth weight newborns (1,501g-2,500g) [1]. EONS is associated with increased risk of neonatal morbidity including respiratory distress syndrome (RDS), necrotizing enterocolitis (NEC), bronchopulmonary dysplasia (BPD), retinopathy of prematurity (ROP), intraventricular hemorrhage (IVH) and death [2]. Clinically symptomatic and asymptomatic intra-amniotic inflammation and/or infection (IAI) is an important risk factor for development of EONS [2,3,4] and is a more frequent occurrence in women when preterm birth debuts as preterm premature rupture of membranes (PPROM) [5]. Because of the risk of maternal and fetal infectious morbidity and mortality, immediate delivery is recommended for women with either clinical chorioamnionitis or with a gestational age (GA) greater than 34 weeks [6]. In women with PPROM at less than 34 weeks, without clinical signs of chorioamnionitis, the risk of prematurity resulting from provider-initiated delivery must be balanced against the risk that exposure to subclinical IAI results in worse outcomes than prematurity alone, including heightened danger for EONS [7,8].

Amniocentesis can provide direct evaluation of bacterial and inflammatory status of the gestational sac [9]. Amniotic fluid (AF) cultures and Gram stain are referred to as diagnostic test of AF infection. Despite their use as the gold standard, AF cultures underestimate the prevalence of AF infection as they do not capture uncultivated or difficult to cultivate species [10]. In addition, the turnaround time for cultures results (~48 hours) limits the ability to intervene in a timely manner. Thus, quite often, elevated AF white blood cell (WBC) count, decreased glucose, and elevated lactate dehydrogenase (LDH) are used in the clinical setting to make decisions regarding timing of delivery [9]. Nevertheless, despite the diagnostic benefits of amniocentesis, technical difficulties and absent randomized clinical trials measuring risk versus benefit have resulted in debate on the role of this procedure in the routine clinical practice [11].

Haptoglobin (Hp) is an acute phase reactant produced by the liver in response to transcriptional activation by interleukin (IL)-6 [12]. Hp has important antioxidant, antimicrobial, pro-angiogenic and immunomodulatory functions which are genotype and phenotype determined [13]. Interestingly, expression of Hp is silenced in utero resulting in normal term newborns having cord blood (CB) Hp levels below the limit of detection for the methods validated for adults (Hp “switch-off” status, ahaptoglobinemia or Hp phenotype 0–0) [14, 15]. Through proteomics analysis of CB, our previous investigation found elevated levels of Hp (~60 fold) in newborns exposed antenatally to IAI independent of GA and birth weight [15]. Neither proteomics nor immunoreactive validation methods could discriminate Hp from Hp-related protein (HpRP), a smaller abundance protein that is >90% identical in aminoacid sequence with Hp [16]. Thus, from a biomarker perspective we refer to them together as Hp&HpRP. CB Hp&HpRP “switch-on” status (detectable Hp&HpRP by Western blot) was found to be a biomarker of EONS with additional ability to predict poor short-term neonatal outcome including IVH and death, independent of GA at birth and clinical diagnosis of EONS [15].

Conversion of the fetal Hp&HpRP to “switch-on” status is consistent with ability to sense and react to environmental stimuli, in this case IAI. Characterization of exact temporal relationships between IAI and CB Hp&HpRP “switch-on” status will require simultaneous and repeated access to CB, an impossible clinical undertaking. Based on our knowledge so far, it is reasonable to assume that, neonatal outcomes are determined by a complex interaction of factors including exposome (sum of interactions between genetic make-up and environment), and for cases complicated by IAI, the time fetus remains exposed in utero to the hostile environment [17]. We hypothesized that newborns that are born prior to fetal conversion to Hp&HpRP “switch-on” status have improved neonatal outcomes compared to those born after conversion. The objective of our study was to compare short-term outcomes (sepsis, NEC, ROP, IVH, BPD, death) of newborns born prematurely in the context of PPROM when mothers were managed based on amniocentesis results versus those who were managed expectantly without amniocentesis.

MATERIALS AND METHODS

Study population.

We performed a prospective cohort study of 185 women pregnant with singletons pregnancies consecutively enrolled between May 2004 to July 2012. All women were admitted at Yale New Haven Hospital (YNHH) in the setting of PPROM. Exclusion criteria included fetuses with congenital anomalies and known maternal viral infections (hepatitis B, hepatitis C or HIV). The study protocol was approved by Yale University Human Investigational Committee and all women signed informed consent. A flowchart depicting of the study population grouped by AF and CB biomarkers is presented in Figure 1. In 123/185 (68%) cases an amniocentesis was performed to evaluate for presence or absence of microbial invasion of the amniotic cavity and to guide timing of delivery. Indications for amniocentesis included advanced cervical dilatation (>3.0 cm), persistent uterine contractions, maternal leukocytosis (WBC>15,000 cells/µL), signs and symptoms of clinical chorioamnionitis (abdominal tenderness, maternal fever >100.4 F) and PPROM. In 62/185 (32%) cases an amniocentesis was not performed. Reasons for not performing an amniocentesis included: patient declined, amniocentesis was not offered, or technical non-feasibility. In women who did not have an amniocentesis or in whom IAI was excluded by amniocentesis, timing of delivery of the fetus was based on clinical criteria for maternal chorioamnionitis, onset of spontaneous labor, or non-reassuring status of the fetus. All fetuses were live born and were admitted to YNHH Newborn Special Care Unit (NBSCU). The decision to perform a diagnostic amniocentesis, interpretation of the clinical AF analysis results (glucose, LDH activity, WBC count, Gram stain and cultures) and the decision for expectant management or delivery occurred independent of our study protocol.

Figure 1. Study population and groups.

Representation of the study population based on performance of amniocentesis, intra-amniotic infection and neonatal “exposure” to intra-amniotic inflammation (IAI) as determined by cord blood (CB) Hp&HpRP and IL-6 levels. All neonates were admitted in Newborn Special Care unit (NBSCU) and short-term outcomes used for data analysis. *P=0.004; ** P=0.005. ns, no statistical significance.

Clinical diagnosis of AF infection and inflammation.

Following amniocentesis, diagnosis of IAI was established based on recognized clinical, biochemical and microbiological AF test results for glucose (cut-off of ≤15 mg/dL), LDH (cut-off ≥419 U/L), AF WBC (cut-off ≥50 cells/mm3), Gram stain and bacterial cultures for aerobes, anaerobes, Ureaplasma and Mycoplasma species [18]. AF infection was considered in the presence of positive AF culture result and/or Gram positive stain

Histologic examination of the placenta.

In all cases, hematoxylin & eosin-stained sections of extraplacental membranes (amnion and chorio-decidua), chorionic plate and umbilical cord were examined systematically for inflammation. Three histological stages of chorioamnionitis (stage I: intervillositis, stage II: chorionic inflammation, and stage III: full-thickness inflammation of both chorion and amnion) were complemented by a previously described histological grading system that grades maternal versus fetal inflammation based on 4 grades of inflammation of the amnion, chorio-decidua and umbilical cord [19].

Cord blood Hp and IL-6 by ELISA.

CB was collected at birth using sterile conditions for all cases. For quantification of Hp&HpRP, CB samples were diluted 100-fold. A mixed phenotype of Hp standard (Hp1–2, Sigma-Aldrich, St. Louis, MO) was used to prepare the standard curve (250–3.9 ng/ml). CB IL-6 was measured using an ELISA assay (Pierce-Endogen, Rockford, IL) with a reported minimum detection level of 0.039 pg/mL. Immunoassays for both Hp&HpRP and IL-6 were performed in duplicate by investigators unaware of sample origin or outcome as previously described [15].

Assessment of CB Hp&HpRP switch status by Western blotting.

Hp is a tetrameric protein with two α and two β-chains. Human Hp occurs in two co-dominant allelic forms, Hp1 and Hp2, differentiated by the α-chain [20]. The human population has 3 major Hp phenotypes (Hp1–1, Hp2–2 and the heterozygous Hp1–2), derived from variations in the α-chain with identical β-chains. Absence of Hp at protein level denotes Hp0–0 phenotype (ahaptoglobinemia) [20]. Therefore, differences in Hp phenotypes and “switch-on” or “switch-off” pattern can be distinguished by Western blotting. We have previously shown that a cut-off in CB Hp&HpRP immunoreactivity of 3,370 ng/mL in our ELISA assay has almost 100% sensitivity and specificity of segregating between samples with non-detectable immunoreactive Hp β-chain at Western blotting (switch-off) and switch-on samples (detectable β-chain with or without α chains).

SDS-PAGE gels (10–20%, InVitrogen, Carlsbad, CA) were loaded with equal amounts of CB protein (2 µg/lane) mixed 1∶2 with reducing sample buffer (Bio-Rad, Hercules, CA) and boiled for 5 min. After electrophoretic transfer, nitrocellulose membranes (Bio-Rad) were blocked with 5% milk and then incubated overnight at 4°C rabbit anti-Hp polyclonal antibody (1∶3,000, Sigma-Aldrich). Detection was performed using biotinylated goat anti-rabbit secondary antibody (1∶5,000, Jackson Immunoresearc, West Grove, PA) followed by streptavidin-linked horseradish peroxidase, (1∶8,000, Amersham Biosystems, Piscataway, NJ), chemiluminescence (ECL-Plus, Amersham) and a timed 3 min. exposure to film (Kodak Biomax, PerkinElmer Life Sciences, Boston, MA). Purified Hp from blood of adults with known phenotypes (Hp1–1, Hp1–2 and Hp2–2, Sigma) was used as positive control.

Neonatal outcome.

All newborns were followed prospectively until death or discharge from NBSCU. Primary study outcomes were IVH, NEC, BPD, ROP, PVL, and death. Neonatal hematological indices were assessed from blood specimens obtained within 2-hours post-delivery. The diagnosis of EONS was “presumed” in the presence of signs and symptoms suggestive of sepsis (i.e lethargy, apnea, respiratory distress, hypoperfusion, shock) and at least two of the following hematological criteria: absolute neutrophil count (ANC) of <7,500 or >14,500 cells/mm³; absolute band count (ABC) >1,500 cells/mm³; immature/total neutrophil (I:T) ratio >16% or platelet count <150,000 cells/mm³ as previously described [21, 22]. EONS was termed “confirmed” when neonatal blood cultures returned a positive result. All newborns with presumed and confirmed EONS received intravenous antibiotic therapy.

Clinical neonatal evaluation was done in the NBSCU. In addition to individual adverse neonatal outcomes, a composite outcome was created to include only severe outcomes (Grades 3 or 4) in NEC, IVH or ROP; and if present BPD or death. Evaluation for IVH was done per institutional protocol using serial cranial ultrasounds on days 3, 7–10 and 30 of life with additional scans if clinically indicated [23, 24]. The diagnosis and grading of IVH was established by experienced pediatric radiologists: grade 1, germinal matrix hemorrhage; grade 2, intraventricular blood without distension of the ventricular system; grade 3, blood filling and distending the ventricular system and grade 4, parenchymal involvement of hemorrhage, also known as periventricular venous infarction [23]. The ophthalmologist classified ROP in each eye according to the international classification [25]. Clinical, metabolic, hematologic and abdominal x-ray abnormalities (i.e. pneumatosis intestinalis, portal venous gas) criteria were used to diagnose NEC [26]. BPD was defined as need of receiving supplemental oxygen at 36 weeks’ corrected postmenstrual age [27].

Data analysis and statistical considerations.

The variable “antenatal exposure to IAI” was determined based on the two CB biomarkers (Hp&HpRP and IL-6) and the algorithm derived reported in our earlier study (16). Briefly, cases were considered as being “antenatally exposed to IAI” in the setting of either Hp&HpRP “switch-on” pattern (irrespective of CB IL-6 level) or Hp&HpRP “switch-off” plus IL-6 levels ≥100 pg/mL. Cases were considered as “non-exposed antenatally to IAI” if they had cord blood Hp “switch-off” pattern and IL-6 levels <100 pg/mL (Figure 1).

Data distribution was tested using the Shapiro-Wilk test, and data reported as median and interquartile [IQR] range. Immunoassay results for IL-6 and Hp&HpRP immunoreactivity were logarithmically transformed prior to statistical analysis. Data were compared Mann Whitney Ranks Sum test or Kruskal–Wallis on ranks followed by Dunn’s tests (non-parametric). Proportions were compared with Chi-square test. Relationships between variables and occurrence of poor neonatal outcomes were explored using multiple linear or logistic regression analysis. Odds Ratio (OR) and Confidence Intervals (CI) were reported as appropriate. A probability level of <0.05 was considered statistically significant. MedCalc (Mariakerke, Belgium), SigmaStat 12 (RockWare, Golden, CO) and GraphPad Prism (San Diego, CA) statistical software were used for data analysis.

RESULTS

Figure 1 illustrates the grouping of the cases in our study. In 123/185 (68%) cases, an amniocentesis was performed to evaluate for presence or absence of microbial invasion of the amniotic cavity based on clinically standard results of AF cultures and/or Gram stain which were used to guide timing of delivery. In 62/185 (32%) cases an amniocentesis was not performed. Cases in the 3 groups (positive AF infection, negative AF infection and unknown AF infection status) were further grouped based on our previously reported CB algorithm combining Hp&HpRP switch status and IL-6 (above or below the 100 pg/mL cut-off) [15]. The demographic, clinical, and pregnancy outcome characteristics of mothers and newborns in the 3 study groups are described in Table 1. African American women were more likely to have an amniocentesis confirming AF infection compared to their Caucasian and Hispanic counterparts (P=0.039). Women who had amniocentesis and evidence of IAI were admitted and delivered at an earlier GA than women where infection was ruled out at amniocentesis or women who were expectantly managed without amniocentesis (P<0.001). Compared to the other two groups, women with PPROM and AF infection had a higher degree of cervical dilation at admission (P=0.006), higher rate of steroid exposure prior to delivery, shorter amniocentesis-to-delivery interval (P<0.001), lower GA at delivery (P<0.001), lower birth weight (P<0.001) and lower newborn Apgar scores (P<0.05).

Table 1. Demographic and clinical characteristics of women and their newborns.

Variable	MATERNAL GROUPS (N=185)			P value
	PPROM (+) AF infection n=61	PPROM (−) AF infection n=62	PPROM No Amniocentesis n=62
Maternal characteristics at enrollment and hospital course
Age, years ‡	30 [25–35]	29 [24–33]	31 [23–36]	0.791
Gravidity ‡	3 [1–5]	2 [1–3]	3 [2–4]	0.396
Parity ‡	1 [0–2]	0 [0–1]	1 [0–2]	0.239
Race §  Caucasian  African-American  Hispanic  Other	21 (34) 26 (43) 11 (18) 3 (5)	34 (55) 13 (21) 8 (13) 7 (11)	27 (44) 18 (29) 15 (24) 2 (3)	0.039
Body Mass Index ‡	27 [24–34]	27 [25–33]	31 [25–36]	0.216
Gestational age, weeks ‡	27.3 [24.8–30.1]	31.3 [29.4–32.4]	30.8 [25.6–33.1]	<0.001
Uterine contractions §	15 (25)	16 (26)	26 (42)	0.066
Clinical chorioamnionitis §	9 (15)	3 (5)	4 (6)	0.111
History of preterm birth §	16 (26)	15 (24)	11 (18)	0.501
Cervical dilatation >2cm §	16 (26)	10 (16)	3 (5)	0.006
Steroid exposure during pregnancy §	60 (98)	54 (87)	52 (84)	0.021
Prenatal antibiotic treatment §	55 (90)	58 (94)	57 (92)	0.789
Antenatal magnesium sulfate §	10 (16)	17 (27)	11 (18)	0.254
Outcome characteristics
PPROM-to-delivery (latency), hours ‡	30 (18–114)	91 (27–284)	93 (22–271)	0.006
Amniocentesis-to-delivery, hours ‡	12 [5–22]	60 [20–192]	NA	<0.001
Delivery within 72 hours from rupture §	55 (90)	32 (52)	NA	<0.001
Gestational age at delivery, weeks †	27.4 [25.2–30.8]	32.2 [30.1–33.4]	31.9 [27.3–33.6]	<0.001
Birthweight, grams ‡	990 [780–1535]	1848 [1451–2137]	1755 [985–2146]	<0.001
Cesarean delivery §	30 (49)	23 (37)	31 (50)	0.272
Apgar score at 1 minute ‡	7 [3–8]	8 [6–9]	8 [5–9]	0.038
Apgar score at 5 minutes ‡	8 [6–9]	9 [8–9]	9 [8–9]	0.009

^‡Data presented as median [interquartile range] and analyzed by Kruskal-Wallis ANOVA;

^§Data presented as n (%) and analyzed by Chi square tests;

P values in bold font are considered significant based on P<0.05.

Abbreviations: PPROM, preterm premature rupture of membranes; IAI, intra-amniotic inflammation

As presented in Table 2, women with PPROM and positive AF infection were more often anemic (P=0.006) and had clinical biochemical AF test results consistent with IAI (P<0.001 for all).

Table 2. Laboratory analyses available for clinical management.

Variable	MATERNAL GROUPS (N=185)			P value
	PPROM (+) AF infection n=61	PPROM (−) AF infection n=62	PPROM No Amniocentesis n=62
Maternal hematology
Hematocrit, g/dL ‡	32.9 [30.1–34.7]	34.6 [32.0–36.8]	33.7 [31.6–36.1]	0.006
Hemoglobin, g/dL ‡	11.3 [10.1–11.9]	11.8 [11.2–12.5]	11.8 [10.8–12.4]	0.003
WBC count, x1,000 cells/μL ‡	13.1 [10.7–17.0]	11.8 [9.8–14.5]	12.4 [10.8–15.3]	0.084
Neutrophils, % ‡	78.0 [71.5–85.5]	75.0 [71.0–81.5]	75.5 [69.0–82.0]	0.181
Lymphocytes, % ‡	13.0 [7.0–19.0]	16.0 [11.0–20.0]	16.5 [12.0–20.2]	0.050
Platelets, x1,000 cells/μL ‡	260 [194.5–302.5]	264.5 [198.0–306.0]	267.5 [218.8–318.0]	0.550
Amniotic fluid analysis results
Glucose, mg/dL ‡	6 [2–15]	26 [19–37]	NA	<0.001
LDH activity, U/L ‡	571 [277–1,175]	164 [110–328]	NA	<0.001
WBC, cells/mm3 ‡	387 [50–1,940]	13 [4–85]	NA	<0.001
Positive Gram stain §	39 (64)	0 (0)	NA	<0.001
Positive cultures §	58 (95)	0 (0)	NA	<0.001
Intra-amniotic infection (positive culture or positive Gram stain) §	61 (100)	0 (0)	NA	<0.001

^‡Data presented as median [interquartile range] and analyzed by Mann Whitney Ranks Sum test;

^§Data presented as n (%) and analyzed by Chi square tests;

P values in bold font are considered significant based on P<0.05.

Abbreviations: PPROM, preterm premature rupture of membranes; IAI, intra-amniotic inflammation; WBC, white blood cell; LDH, lactate dehydrogenase; MR score: mass restricted score

CB acid-base analysis and placental histology are presented in Table 3. A higher degree of maternal or fetal histological chorioamnionitis was noted in women with PPROM and positive AF infection. There were no differences in cord pH or base deficit among the 3 groups. Only one case in our study registered a cord pH<7.

Table 3. Cord blood gas analyses and placental pathology results.

Variable	MATERNAL GROUPS (N=185)			P value
	PPROM (+) AF infection n=61	PPROM (−) AF infection n=62	PPROM No Amniocentesis n=62
Cord blood gas analysis	n=40	n=34	n=44
Arterial pH ‡	7.32 [7.27–7.33]	7.30 [7.23–7.33]	7.28 [7.24–7.34]	0.281
Arterial base deficit, mmols/L ‡	4.9 [3.3–6.4]	4.2[2.7–7.5]	6.0[3.0–8.0]	0.791
Venous pH ‡	7.36 [7.31–7.39]	7.36 [7.32–7.39]	7.35 [7.31–7.38]	0.437
Venous base deficit, mmols/L ‡	4.2 [2.5–5.9]	3.5 [2.2–5.6]	4.0 [2.0–7.0]	0.746
Any cord pH < 7 §	0 (0)	1 (3)	0 (0)	0.288
Placental pathology	n=61	n=62	n=60
Amniochorion, grade ‡	2 [1–3]	0 [0–1]	0 [0–2]	<0.001
Amniochorion (grade 2–4) §	25 (74)	14 (23)	26 (43)	<0.001
Chorionic plate, stage ‡	3 [2–3]	0 [0–2]	2 [0–3]	<0.001
Chorionic plate (stage II-III) §	50 (82)	22(35)	35 (57)	<0.001
Funisitis, grade ‡	2 [0–4]	0 [0–1]	0 [0–4]	<0.001
Funisitis (grade 1–4) §	41 (67)	17 (27)	30 (49)	<0.001
Maternal HCA §  absent  mild  severe	12(20) 19(31) 30(49)	45(73) 9(15) 8(12)	34(56) 13(21) 14(23)	<0.001
Fetal HCA §  absent  mild  severe	5(8) 6(10) 50(82)	35(56) 5(8) 22(36)	24(39) 0(0) 37(61)	<0.001
Evidence of decidual hemorrhage or abruption §	5(8)	11(18)	8(13)	0.291

^‡Data presented as median [interquartile range] and analyzed by Mann Whitney Ranks Sum test;

^§Data presented as n (%) and analyzed by Chi square tests;

P values in bold font are considered significant based on P<0.05.

Abbreviations: PPROM, preterm premature rupture of membranes; IAI, intra-amniotic inflammation; HCA, histologic chorioamnionits.

The results of the neonatal hematologic indices at birth are presented in Table 4. Neonates in the positive AF infection group were more likely to be anemic (P=0.007), have leukocytosis (P=0.040), have higher number of nucleated red blood cells (P<0.001). Newborns in this group were more often diagnosed with presumed EONS by hematologic indices compared to newborns from the groups of women where AF infection was ruled-out and those managed expectantly without amniocentesis (P=0.004).

Table 4. Results of newborn hematologic indices used clinically for neonatal sepsis workup.

Variable	MATERNAL GROUPS (N=185)			P value
	PPROM (+) AF infection n=61	PPROM (−) AF infection n=62	PPROM No Amniocentesis n=62
Hematologic indices
Hematocrit, % ‡	45 [41–48]	48 [44–53]	47 [41–51]	0.007
Hemoglobin, g/dL ‡	14 [13–15]	16 [14–17]	15 [14–17]	0.002
WBC count, cells x 1,000/mm3 ‡	13 [8–18]	10 [8–14]	12 [9–16]	0.047
Platelet count, cells x 1,000/mm3 ‡	260 [219–308]	286 [214–331]	252 [189–306]	0.172
Segmented neutrophils, % ‡	36 [22–43]	35 [26–42]	40 [22–49]	0.292
Lymphocytes, % ‡	38 [25–54]	45 [33–59]	38 [31–55]	0.094
Nucleated RBC, cells per 100 WBC ‡	17 [9–43]	8 [4–17]	11 [5–25]	<0.001
ANC, cells/mm3 ‡	3913 [2005–7762]	3157 [2378–5863]	5236 [1995–8112]	0.286
ABC, cells/mm3 ‡	682 [252–2388]	132 [0–592]	270 [0–976]	<0.001
I:T ratio, % ‡	8 [2–13]	1 [0–5]	2 [0–8]	<0.001
Presumed EONS §	22 (37)	8 (13)	10 (17)	0.004
Culture confirmed EONS §	2 (3)	1 (2)	1 (2)	0.779

^‡Data presented as median [interquartile range] and analyzed by Kruskal-Wallis ANOVA.

^§Data presented as n (%) and analyzed by Chi square tests.

Abbreviations: PPROM, preterm premature rupture of membranes; IAI, intra-amniotic inflammation; WBC, white blood cell; RBC, red blood cells; ANC= absolute neutrophil count; ABC= absolute band count; I:T, immature/total neutrophil count; EONS: early-onset neonatal sepsis.

We first compared CB IL-6 (Figure 2, panel A) and Hp&HpRP immunoreactivity (Figure 2, panel B) measured by ELISA between the group that had an amniocentesis to guide management (n=123) and the group managed expectantly without amniocentesis. After correction for GA, the group managed without amniocentesis had lower levels for both biomarkers. Hp&HpRP immunoreactivity is known to be impacted by Hp phenotype [15], hence the value of assessing Hp&HpRP switch pattern (on or off) instead. After correction for GA at birth, there was no difference in the proportion of newborns classified as having Hp&HpRP switch-on pattern between the groups managed with and without amniocentesis (Figure 2, panel C). Next, we evaluated the IAI “exposed” vs. “non-exposed” status based on our previously published algorithm [15]. Cluster assignment was driven in 96% (178/185) of the cases by Hp&HpRP switch pattern (independent of CB IL-6) with only the remaining minority (7/185) assigned as “exposed” based on elevated CB IL-6 (Hp&HpRP switched-off). After correction for GA at birth, there was no significant difference in the proportion of newborns antenatally “exposed to IAI”, between the group who had [53% (65/123)] or not [37% (23/62)] an aminocentesis (Figure 2, panel D).

Figure 2. Analysis of cord blood biomarkers and exposed status based on performance of amniocentesis.

(A) IL-6 and (B) Hp&HpRP measured by ELISA. Horizontal lines represent the group median; (C) Proportion of newborns with Hp&HpRP switch-on status at birth (hashed bar); (D) Proportion of newborns clustered as antenatlly exposed to IAI (checkered bar). Statistical analysis: Mann-Whitney tests (A&B) or chi-square tests (C&D). P values are shown after correction for gestational age at birth. *P=<0.001; ** P=0.021. ns, no statistical significance.

After correction for GA, the levels of CB IL-6 of newborns delivered by women with AF infection were rendered non-significant (Figure 3, panel A). In contrast, the elevated Hp&HpRP ELISA immunoreactivity remained significantly higher in newborns with IAI (Figure 3, panel B). The Hp&HpRP switch-on pattern was more frequently encountered in newborns of women with AF infection [75% (46/61)] compared to the women testing negative [23% (14/62)] (Figure 3, panel C). The number of newborns identified as having been exposed to IAI were significantly higher in the group of women with positive AF infection [75% (46/61)] compared with negative AF infection [31% (19/62)] (Figure 3, panel D).

Figure 3. Analysis of cord blood biomarkers and exposed status based on amniotic fluid (AF) positive (Pos) or negative (Neg) infection status.

(A) IL-6 and (B) Hp&HpRP measured by ELISA. Horizontal lines represent the group median; (C) Proportion of newborns with Hp&HpRP switch-on status at birth (hashed bar); (D) Proportion of newborns identified as antenatally exposed to IAI (checkered bar). Statistical analysis: Mann-Whitney tests (A&B) or chi-square tests (C&D). P values are shown after correction for gestational age at birth.

We next regrouped the cases in the study using the combination of AF infection status (negative, positive or unknown) and antenatal exposure to IAI as identified from our CB biomarker algorithm (Figure 4, panel A) [16]. Within each of the AF infection groups, “exposed” newborns had significantly higher CB IL-6 levels compared to their “non-exposed” counterparts (positive AF infection group P<0.001; negative AF infection group P<0.001; unknown AF infection group P=0.001). Notably, CB IL-6 levels of “exposed” neonates delivered by women managed expectantly without amniocentesis (?I+E), were lower than those of “exposed” newborns born to women who had an amniocentesis confirming AF infection (+I+E) (P=0.023).

Figure 4. Analysis of cord blood IL-6 (A) and neonatal outcome (B) by the combination of amniotic fluid infection status (I: positive, negative or unknown) at amniocentesis and exposed status (E: positive or negative) at birth.

(A) Scatterplot of cord blood IL-6 in logarithmic format. The bar represents the median level for the group. (B) Proportion of newborns displaying at least one major adverse outcome represented by the black portion of the bar. Data analyzed by one-way ANOVA followed by Holm Sidak multiple comparison method (A) or chi-square test (B). Groups sharing one common letter are not statistically different at P>0.05.

In Table 5 we present the frequency of neonatal complications using the same combination of AF infection status and exposure to IAI as presented above for CB IL-6. Figure 4, panel B is a graphic representation of the percent of neonates with at least one severe adverse neonatal outcome. “Exposed” neonates of women with an amniocentesis negative for IAI and of mothers who did not have an amniocentesis to rule-out infection had longer latency periods compared to newborns of mothers who had positive IAI at amniocentesis (P<0.001 for both). Neonates in the “exposed” groups based on CB biomarkers had higher rates of NEC, ROP, IVH and death individually and as a composite outcome. Neonates born to mothers who did not have an amniocentesis, yet were deemed as “exposed” based on CB biomarkers, had the highest rate of short-term adverse outcomes despite the longer latency (P<0.001).

Table 5. Adverse outcomes of newborns grouped based on amniotic fluid and/or cord blood biomarkers.

Variable	PPROM (+) AF infection n=61		PPROM (−) AF infection n=62		PPROM No Amnio n=62		P value
	Exposed n=46	Non-exposed n=15	Exposed n=19	Non-exposed n=43	Exposed n=23	Non-exposed n=39
GA at birth, weeks ‡	26 [25–30]	29 [26–32]	32 [29–33]	32 [30–34]	30 [25–33]	33 [30–34]	0.020
Latency, days ‡	1 [0–5]	1 [0–5]	7 [2–18]	3 [1–8]	6 [1–8]	2 [1–12]	<0.001
Number of newborns diagnosed with adverse outcome
Confirmed EONS §	2 (4)	0 (0)	1 (5)	0 (0)	0 (0)	0 (0)	0.336
NEC §	7 (15)	0 (0)	3 (16)	3 (7)	7 (30)	3 (8)	0.043
ROP §	15 (33)	2 (13)	0 (0)	2 (5)	3 (13)	5 (13)	0.002
IVH §	11 (24)	1 (6)	3 (16)	1 (2)	4 (17)	3 (8)	0.041
BPD §	8 (18)	0 (0)	1 (5)	1 (2)	5 (22)	4 (10)	0.052
Death §	7 (15)	0 (0)	4 (21)	0 (0)	8 (35)	1 (3)	<0.001
At least one severe adverse outcome §*	23 (50)	1 (6)	5 (26)	3 (7)	15 (65)	9 (23)	<0.001

^‡Data presented as median [interquartile range] and analyzed by Kruskal-Wallis ANOVA

^§Data presented as n (%) and analyzed by Chi square tests.

^*Includes newborns with at least one of the following outcomes: IVH grade 3–4, ROP grade 2–4, NEC grade 2–4, BPD and/or death.

Abbreviations: PPROM, preterm premature rupture of membranes; GA, gestational age; EONS, early-onset neonatal sepsis; NEC, necrotizing enterocolitis; ROP, rethinopathy of prematurity; IVH, intra-ventricular hemorrhage; BPD: bronhopulmonary dysplasia; Hp&HpRP: haptoglobin and haptoglobin-related protein.

Stepwise logistic regression determined that pregnancy management with performance of amniocentesis (OR: 0.3 [95%CI: 0.2–0.80]), and a more advanced GA at delivery (OR: 0.6 [CI: 0.5–0.74]) were protective against the newborn from the respective pregnancy having a diagnosis of at least one adverse neonatal outcome. Conversely, newborn’s “exposed” status based on our CB biomarker algorithm at birth was associated with an increased risk of severe outcome or death (OR: 2.9 [CI: 1.16–7.40]). Variables excluded from the model included exposure to antenatal steroids, exposure to antenatal magnesium, CB IL-6 level and presumed EONS based on clinical symptoms corroborated with neonatal hematology results.

DISCUSSION

Our work demonstrates that regardless of antepartum management, neonatal outcomes of premature newborns born in the context of PPROM, are worse once antenatal exposure to IAI extends to a point when it results in conversion to Hp&HpRP “switch-on” status. However, newborns born to mothers managed expectantly without knowledge of inflammatory status of the AF cavity who had converted to Hp&HpRP “switch-on” state had the worst outcomes. As this group had a significantly longer membrane rupture-to-delivery interval, our findings underscore the potential impact of the duration of exposure to IAI in determining neonatal outcomes. In the subgroup of neonates who had evidence of infection based on AF analysis but who had not yet converted to “switch-on” Hp&HpRP status, early delivery resulted in improved outcomes.

The current study has a pragmatic character because our observations reflect the state of current clinical practice where in women with PPROM an amniocentesis procedure is either performed or not (i.e. attending preference, technically impossible, patient refusal). Our findings suggest that in the setting of PPROM and IAI, the ability to identify the subgroup of fetuses with Hp&HpRP “switch-off” status and deliver them prior or shortly after inflammation initiates expression of Hp may lead to a more personalized approach to improving neonatal outcomes. At present this is not an easy task as more research is required to develop non-invasive methods able to monitor fetal Hp&HpRP status prior to birth. So far, the validation of Hp&HpRP as biomarker of antenatal exposure to IAI has only been performed in CB available at birth, a timeframe which is far too late for clinical decisions by physicians. In a previous study Dulay et al. have shown that the “compartmentalization” process characteristic to human pregnancy is a significant obstacle to using levels of maternal blood or urine inflammatory cytokines to infer information on the same analytes in AF or CB (fetal) compartment [28]. Although the study by Dulay et al. did not include the evaluation of Hp&HpRP as biomarker, it is known that maternal blood Hp&HpRP expression or levels are clinically irrelevant in this context since Hp is normally expressed in the adult at high levels [14,15].

In the current study the information on the infectious and inflammatory status of the amniotic cavity was taken at presentation. The women with clinically manifest IAI (clinical chorioamnionitis, n=12) and those with amniocentesis results suggestive of IAI were all delivered within 24 hours. However, the duration of fetal exposure to IAI in women without evidence of infection or inflammation, at the time of amniocentesis, or among women who did not have an amniocentesis could not be determined. Although an amniocentesis is an option to diagnose infection in-utero, frequently this procedure cannot be performed despite clinical suspicion due to objective (oligohydramnios) or subjective factors (such as physicians’ insufficiently trained or not comfortable to perform invasive diagnostic procedures), as well as to the associated risks. Development of new non-invasive technologies for longitudinal assessment of fetal infectious and/or inflammatory status in utero is critical because increasing latency time between PPROM and delivery impacts the risk of neonatal infectious morbidity [29]. Future studies uniquely applicable to PPROM, could incorporate serial examination of vaginal AF with the goal of identifying a critical threshold of various vaginal AF biomarkers before a fetus converts to a “switch-on” Hp&HpRP status, at which point the fetus could be delivered. Consistent with the notions of precision medicine, strategies that identify fetuses that may benefit from early delivery, prior to conversion to Hp&HpRP “switch-on” status, may decrease the rate of complications associated with EONS and improve outcomes. The next best practical solution is development of novel technologies to allow for rapid identification of CB Hp&HpRP “switch-on” status immediately after birth which will help those caring for the neonate [30, 31].

As seen from our findings, exposed newborns of women managed expectantly without amniocentesis had the worst outcomes despite a longer latency, and interestingly, only mildly elevated CB IL-6 levels in this group. The most likely explanation lies in the non-linear and distinct time courses of CB IL-6 and CB Hp&HpRP immunoreactivity in utero post-infectious trigger. Since IL-6 is the trigger for initiating Hp&HpRP transcription in the liver the spike in CB IL-6 likely precedes the CB Hp&HpRP switch-on event [12]. We propose that following the initial elevation, CB IL-6 decreases while Hp&HpRP continues to increase and thus is found elevated at delivery.

In the current study we had the ability to analyze a relatively large sample size of neonates where an amniocentesis procedure was clinically indicated to rule-out infection. We also benefited from a consistent evaluation of neonatal outcomes by staff neonatologists. We do not exclude the possibility that women with an AF pocket that could not be easily accessed had lower AF volumes which may have contributed to worse outcomes in the “no amniocentesis” group [32]. However, studies assessing the inflammatory status of the AF of women with PPROM with and without oligohydramnios found no difference in AF and CB IL-6 levels at birth and no association between oligohydramnios and neonatal morbidity [33]. We therefore think this possible study bias is unlikely to have influenced our findings. Finally, hypothetically, the findings of our study could be strengthened by randomization (i.e. delivery vs. non-delivery) in women where a diagnosis of AF infection was established by amniocentesis. Yet, this type of study design will be challenging based on clinical practice standards and ethical grounds.

In summary, our findings lays important groundwork supporting development of clinical tools based on new or existing biomarkers that are aimed to identify in the antenatal period fetuses in which the risks of continued exposure to IAI are greater than the risks of prematurity alone.