We sought to correlate maternal and cord blood cytokine and intercellular adhesion molecule-1 levels with antibiotic exposure and perinatal outcomes after conservatively managed preterm premature rupture of the membranes.
Conservatively managed women with preterm premature rupture of the membranes at 24-32 weeks had blood sampling at randomization (n = 222) and delivery (n = 121). Plasma from these, and umbilical cord blood (n = 196), was stored at –70°C. Interleukin (IL)-6, IL-10, granulocyte colony-stimulating factor (G-CSF), tumor necrosis factor-α, and intercellular adhesion molecule-1 levels were assessed for associations with antibiotic treatment, latency, amnionitis, neonatal sepsis, pneumonia, and composite neonatal morbidity.
Cord blood IL-6 and G-CSF were higher than maternal levels. Antibiotic treatment lowered only maternal G-CSF ( P = .01). Elevated maternal cytokine levels were associated with delivery within 7 days and with development of chorioamnionitis. All umbilical cord blood markers were increased with amnionitis ( P ≤ .01 for each). No maternal marker was associated with neonatal morbidities. Cord G-CSF and IL-6 were increased with neonatal sepsis within 72 hours of birth ( P = .004 for both), and with composite neonatal morbidity ( P = .001 and .002, respectively). Maternal and umbilical cord cytokine levels demonstrated low predictive values for perinatal outcomes.
Umbilical cord blood cytokine values are higher than maternal levels, suggesting significant fetal/placental contribution. Maternal and umbilical cord cytokine levels are not adequately predictive to be used clinically.
Preterm premature rupture of the membranes (PPROM) is associated with brief latency from membrane rupture to delivery, infectious complications, and adverse neonatal outcomes related to preterm birth. The availability of markers that could predict these adverse outcomes would be helpful in guiding management when PPROM occurs remote from term. Prior knowledge of these outcomes could heighten awareness, and might lead to a decision to proceed with early delivery if conservative management was destined to fail.
Elevated maternal plasma and amniotic fluid cytokines in the second trimester and at delivery have been associated with adverse pregnancy and neonatal outcomes in symptomatic and asymptomatic gravidas. Elevated cytokine levels in neonatal cord blood have been associated with perinatal infectious complications in the mother and newborn. Adhesion molecules play an important role in the interaction between neutrophils and vascular endothelium in response to inflammatory stimuli. Circulating intercellular adhesion molecule (ICAM)-1 levels are also increased in both adults and neonates with sepsis. However, studies of inflammatory markers with respect to PPROM have generally included women already experiencing labor and/or amnionitis, or have studied heterogeneous groups with both PPROM and preterm labor with intact membranes. Further, the impact of antibiotic therapy on maternal cytokine levels has not been prospectively studied in this setting.
To be useful clinically, predictive markers of perinatal complications should discriminate between those who will and will not have the adverse outcome, and these differences should be evident before the outcome occurs. Optimally, such markers would be obtained noninvasively.
Herein, we evaluate maternal and umbilical cord plasma cytokine (granulocyte colony-stimulating factor [G-CSF], interleukin [IL]-6, IL-10, tumor necrosis factor [TNF]-α) and ICAM-1 levels during conservative management of PPROM remote from term to evaluate their change during conservative management, the relative distributions of these in the fetal and maternal circulation, the impact of antibiotic treatment on their levels, and their predictive value for adverse perinatal outcomes.
Materials and Methods
This study includes women, with singleton gestations, who participated in a multicenter double-masked trial of antibiotic treatment of PPROM at 24-32 weeks’ gestation. Details of the study have previously been reported. Briefly, this trial was performed at 11 clinical centers from 1992 through 1995. A total of 614 nonlaboring gravidas with PPROM were randomized to antibiotic therapy or a matching placebo regimen during conservative management. The antibiotic regimen consisted of intravenous ampicillin (2 g every 6 hours) and erythromycin (250 mg every 6 hours) for 48 hours, followed by oral amoxicillin (250 mg every 8 hours) and erythromycin base (333 mg every 8 hours) for 5 days unless delivery occurred. Group B streptococcus (GBS) carriers were identified and treated prior to and during labor. Consistent with prevalent practice at that time, tocolysis and corticosteroid administration were prohibited subsequent to randomization. Elective labor induction <34 weeks was prohibited. Neonatal management was left to the discretion of the neonatal caregiver. Head sonograms were performed routinely on newborns weighing <1750 g, and on larger infants when clinically indicated. The protocol was reviewed and approved by the human research committees of all participating institutions prior to initiation of randomization.
Seven of 11 participating clinical centers undertook to collect maternal blood at randomization and at the time of delivery, and umbilical cord blood at the time of delivery. Specimens were collected into prechilled 10-mL EDTA-treated glass tubes, and underwent refrigerated centrifugation (4°C) at 200 g . The plasma supernatant was then stored in aliquots at –70°C until they were transported in dry ice to a central laboratory at the University of Tennessee, Memphis, for analysis (D.T.C.). IL-6 (sensitivity: 1 pg/mL; Endogen Inc, Rockford, IL), IL-10 (sensitivity: 3 pg/mL; Endogen Inc), G-CSF (sensitivity: 39 pg/mL; R&D Systems, Minneapolis, MN), TNF-α (sensitivity: 5 pg/mL; Endogen Inc), and circulating ICAM-1 (sensitivity: 7 ng/mL; R&D Systems) levels were determined in batch fashion by personnel masked to study group assignment and clinical course.
Evaluated perinatal outcomes included: delivery within 7 days after randomization, clinical amnionitis, confirmed neonatal sepsis within 72 hours of birth or before discharge, neonatal pneumonia before discharge, and composite neonatal morbidity (≥1 of: fetal/neonatal death, neonatal sepsis within 72 hours of birth, grade 3-4 intraventricular hemorrhage, stage 2-3 necrotizing enterocolitis, or respiratory distress syndrome). The diagnosis of clinical amnionitis required at least 2 of: antepartum temperature ≥100.4°F, uterine tenderness, foul-smelling vaginal discharge or amniotic fluid, maternal tachycardia (>100 beats/min), fetal tachycardia (>160 beats/min), or maternal white blood cell count >20,000. Respiratory distress syndrome was diagnosed in symptomatic infants who required ventilatory support for at least 24 hours and had radiographically confirmed hyaline membrane disease or a diagnosis of respiratory insufficiency of prematurity. Confirmed sepsis was diagnosed when an infant had suspicious clinical findings and a positive blood or cerebrospinal fluid culture or evidence of cardiovascular collapse requiring volume expansion or pressor agents. Pneumonia was diagnosed if there were compatible symptoms with diagnostic x-ray findings and positive blood cultures, or persistent leukopenia (<4000 per microliter) or a band count >15%.
Statistical analyses were performed using software (SAS; SAS Institute Inc, Cary, NC). Cytokine (pg/mL) and ICAM-1 (ng/mL) levels are presented as medians (25th-75th percentile). Multivariable logistic regression controlled for gestational age at randomization, GBS culture status, and antibiotic study group assignment. Biomarker values for patients with all 3 blood samples available were compared in a pairwise fashion using Wilcoxon matched pairs signed rank test. For this, the Bonferroni corrected P value of .017 was used to declare significance. Biomarker values for groups with and without outcomes were compared using the Wilcoxon test. Where significant associations were identified, receiver operator characteristic analyses were performed to determine the area under the curve (AUC), optimal cutoff points, and predictive values of maternal and umbilical cord markers for brief latency, amnionitis, and adverse neonatal outcomes.
The 7 participating centers contributed 417 patients to the overall study (68% of eligible patients), and blood samples were collected from 277 of these patients (66%; range, 10–75 subjects per center) ( Figure 1 ) . IL-6, IL-10, G-CSF, TNF-α, and ICAM-1 levels were analyzed from 222 maternal samples obtained at randomization and 121 at delivery, and from 196 umbilical cord blood samples. In all, 82 patients had all 3 samples collected. Clinical characteristics of the pregnancies evaluated in this analysis were similar for those with samples available at randomization, at delivery, and from umbilical cord blood ( Table 1 ). The study population was 61% African American with a mean gestation of 29 weeks at randomization. Perinatal outcomes included: clinical amnionitis (31%), neonatal sepsis (14%), pneumonia (4%), and composite morbidity (48%).
|n||Maternal plasma at randomization 222||Maternal plasma at delivery 121||Umbilical cord plasma 196|
|Maternal age, y a||24.8 (5.9)||25.2 (6.1)||25.1 (5.8)|
|African American race, %||61||60||61|
|Prior preterm delivery, %||36||37||36|
|Gestation at randomization, wk a||28.9 (2.3)||29.1 (2.2)||29.2 (2.2)|
|Cervical dilation, median cm||0||0.5||0|
|Delivery within 7 d, %||66||61||63|
|Clinical amnionitis, %||31||28||27|
|Confirmed neonatal sepsis within 72 h, %||8||10||8|
|Neonatal sepsis before discharge, %||14||16||14|
|Neonatal pneumonia before discharge, %||4||4||3|
|Composite neonatal morbidity, % b||52||53||45|
Plasma marker distribution in maternal and umbilical cord blood
Evaluation of subjects with all 3 samples available, and who were assigned to placebo therapy (n = 42), revealed umbilical cord blood levels to be higher than maternal levels obtained at randomization and at delivery for IL-6 (median [25th-75th percentile]: 185 [41-908] vs 4 [2-8] and 46 [25-87] pg/mL, respectively, P ≤ .0001 for each), and for G-CSF (1359 [545-5353] vs 69 [54-97] and 461 [280-818] pg/mL, respectively, P ≤ .0001 for each). Maternal ICAM-1 levels from blood obtained at randomization and at delivery were higher than those obtained from umbilical cord blood (214 [150-269] and 271 [206-341] vs 163 [117-227] ng/mL, P < .0001 for each). IL-10 levels were similar among umbilical cord blood (10 [7-20] pg/mL), maternal randomization (3 [3-5] pg/mL), and maternal delivery (27 [16-62] pg/mL) samples. TNF-α levels were rarely positive, with >99%, 95%, and 90% of samples having undetectable levels at randomization, at delivery, and in cord blood.
Impact of antibiotic administration on plasma markers
No differences were identified in initial maternal systemic markers between the antibiotic and placebo groups, other than for ICAM-1, which was lower in the placebo group at randomization than the antibiotic group ( P = .02) ( Table 2 ). After controlling for gestational age at entry and GBS carriage, antibiotic treatment resulted in significantly lower maternal G-CSF levels at delivery than in the placebo group (284 vs 439 pg/mL, P = .01), and statistically insignificant trend toward lower IL-6 levels (28 vs 35 pg/mL, P = .08), but no other differences in maternal or umbilical cord blood were seen.