The effect of chorioamnionitis on acute respiratory disease in the
premature infant is unclear. It has been reported to either increase,
decrease, or make no contribution to the incidence or severity of acute
respiratory distress in this population.1 These
conflicting reports may be due in part to differing definitions (eg,
clinical signs of maternal infection versus pathologic examination).
Chorioamnionitis and prolonged rupture of membranes (PROM) frequently
coexist.5 Longer duration of membrane rupture is associated
with increasing infiltration of inflammatory cells into the fetal
membranes.6 Additionally, the presence of chorioamnionitis may itself cause rupture of the membranes.7 Several studies have suggested that PROM confers protection against respiratory distress syndrome (RDS).8 A larger review did not
confirm this protective effect.12
We postulated that the protective effect attributed to PROM in the
earlier studies resulted from the presence of chorioamnionitis. Chorioamnionitis is associated with an increased placental production of interleukin-1
(IL-1
) and other inflammatory
mediators.13,14 IL-1
, in turn, stimulates the release of
corticotropin-releasing factor and corticotropin.15,16 We
hypothesized that this process would result in increased secretion of
cortisol, with resultant lung maturation. To test this hypothesis, we
prospectively studied the relationship of chorioamnionitis to serum
cortisol concentrations and to acute respiratory distress in very low
birth weight (VLBW) infants.
METHODS
This study was conducted at the Hershey Medical Center of the
Pennsylvania State University School of Medicine, and was approved by
its institutional review board. Infants admitted to the newborn intensive care unit were eligible for this study if they (1) were 501 to 1500 g at birth, with weight appropriate for gestational age;
(2) had no apparent major congenital anomaly; and (3) did not undergo a
major surgical procedure during the first week of life. Infants were
enrolled after parental consent was obtained.
Surfactant administration was used as a marker for respiratory
distress, since it is not administered prophylactically at the
Hershey Medical Center. No attempt was made to distinguish whether
respiratory distress was caused by pneumonia versus RDS. Congenital
sepsis was defined as a positive blood or cerebrospinal fluid culture.
Degree of respiratory support was quantified with a respiratory acuity
score (RAS) shown to have good correlation with measures incorporating
blood gas data17: 1 point/cm H2O inspiratory
pressure + 1 point/cm H2O end expiratory pressure + 1 point/mechanical breath/minute + 1 point/%O2
>21%. This score allowed comparison of respiratory support for all
infants, whether or not blood gases were obtained. RAS data were
recorded daily at 6:00 AM and 6:00 PM The
scores closest to 12 and 24 hours of life were used for analysis. White
blood cell counts obtained for clinical indications before 6 hours of
life were also recorded. All clinical care was at the discretion of the
attending physician, who was unaware of the study results.
Day of birth was defined as day of life 0. Blood samples were drawn for
cortisol determinations on the afternoon of (a) day 2, (b) day 3 or 4, and (c) day 5, 6, or 7 of life. On either day (b) or (c), the baseline
sample was followed by administration of 3.5 µg cosyntropin (
1-24
corticotropin, an adrenocorticotrophic hormone [ACTH] analog).
Cosyntropin was permitted to be given intravenously by slow push, or
intramuscularly; however, all patients in this study received the
medication intravenously. After 30 minutes, another blood sample was
drawn for cortisol analysis.
Cortisol concentrations were measured in one laboratory, in duplicate,
by radioimmunoassay (Diagnostics Products Corporation, Los Angeles CA).
The intraassay and the interassay variabilities for the assay were
5.0% and 7.5%, respectively. Cross-reaction for all other naturally
occurring adrenal steroids was less than .1%. All pathologic
evaluations of the placentas and fetal membranes were performed by one
author (R.L.N.), who was unaware of the clinical care or course of the
study patients.
Criteria for diagnosis of chorioamnionitis were those previously
described by the author.6 Briefly, the diagnosis of acute chorioamnionitis was made when polymorphonuclear leukocytes accumulated beneath the chorionic (fetal) plate of the placenta. This finding is
classified as stage 1 and studies have shown that it usually persists
for the first 3 days after the infecting agents have gained access to
the amniotic cavity.6 During the next 3 to 4 days the
leukocytes invade and slowly pass through the chorionic plate (stage
2). When the leukocytes reach the basement membrane just beneath the
amnion, the process is at stage 3. It usually takes at least a week for
the process to reach stage 3.6
Statistical Analysis
Population data were compared by unpaired Student's
t test. Linear regression was used to evaluate the
relationship of chorioamnionitis to gestational age. After that, the
relationship of chorioamnionitis to respiratory distress was analyzed.
First, logistic regression was performed to evaluate the effect of
gestational age and chorioamnionitis on the administration of
surfactant, as a marker for the clinical diagnosis of RDS. Secondly,
the effect of gestational age and chorioamnionitis on fraction of
inspired oxygen and respiratory acuity score (RAS) at 12 and 24 hours
of life was analyzed using multiple regression analysis.
Because study findings might be influenced through enrollment bias, a
log was kept of infants otherwise eligible for the study whose parents
were not approached because of death before enrollment, or because of
clinical instability. Eleven patients were identified as "met
criteria, not approached for clinical reasons," and had placental
pathology performed. Seven of these infants died. Clinical and
pathologic data from this group of patients was added to the analyses
to evaluate for enrollment bias. These analyses resulted in no change
in statistical significance or trend from that found in the study
population.
Serum cortisol concentrations were not normally distributed; therefore,
log transformation was performed, which yielded normally distributed
data for analysis. Cortisol concentrations are written as
nanomoles/liter (= micrograms/dL × 27.6). The effect of exposure to
maternal steroids on postnatal cortisol concentrations was evaluated;
because day 2 concentrations were significantly lower in these
patients, all cortisol values from these patients were excluded from
further analysis. No infants received postnatal steroids during the
first week of life. The relationship of chorioamnionitis to serum
cortisol concentrations was then evaluated with a general linear models
analysis of variance, in a model that included gestational age as a
cofactor.
RESULTS
Eighty-seven infants were enrolled in the study. Three were
excluded from analysis due to: (1) parental withdrawal of consent, (2)
diagnosis of congenital adrenal hyperplasia, and (3) a small for
gestation patient inadvertently enrolled. Placental pathology was
performed on placentas from 79 of the 84 patients (94%).
Characteristics of the population are shown in Table 1.
The presence of chorioamnionitis correlated inversely with gestational
age (F = 15.56, P < .001). Because of this very strong correlation, gestational age was included as a cofactor in all analyses.
Chorioamnionitis was associated with a decrease in the need for acute
respiratory support in this population (Table 2). First, infants with chorioamnionitis were significantly less likely to receive
surfactant. In a logistic regression with surfactant administration as
the outcome variable, both gestational age and chorioamnionitis contributed significantly to a
2 of 22.73 (gestation,
14.05; chorioamnionitis 9.37; both P < .005). Adding
the factor "exposure to prenatal steroids" (16 infants) produced no
significant effect in this analysis (addition to
2 = .52, P = .47). Secondly, infants exposed to
chorioamnionitis required significantly less supplemental oxygen and
received less ventilator support at 12 and 24 hours than infants
without chorioamnionitis exposure (Table 2).
Nineteen of the infants had rupture of membranes for longer than 24 hours (PROM), and more of these had chorioamnionitis (Table 1). A
single factor
2 analysis of the relationship of PROM to
surfactant administration would have shown a significant relationship (P = .04); however, when gestational age was
added, and a logistic regression performed, this relationship was no
longer significant (P = .16). When
chorioamnionitis was added to the regression, there was no independent
relationship between PROM and surfactant administration
(P = .81).
Before analyzing the relationship of chorioamnionitis to serum cortisol
concentrations, we examined the effect of prenatal exposure to
corticosteroids. Sixteen infants had received prenatal steroids (Table
1). Cortisol values were significantly decreased on day 2 in these
infants (geometric mean 119 nmol/L vs 213 nmol/L; P < .001). Although day 4 and 6 values were not significantly different, we
excluded all cortisol data from these infants from analysis.
Infants exposed to chorioamnionitis had significantly higher serum
cortisol concentrations during the first week of life
(F = 8.88, P < .005) (Fig.
1). Gestational age was not a significant predictor of
baseline cortisol concentration in this analysis (F < .01, P = .97), nor was day
of life (F = 2.97, P = .054). Infants exposed to chorioamnionitis also had higher cortisol
concentrations after stimulation with cosyntropin (poststimulation,
chorioamnionitis group (n = 31) vs nonexposed group (30):
geometric mean 463 vs 377 nmol/L, F = 11.07, P < .005), and a greater increase over their baseline
values (stimulated-baseline value, or delta cortisol: geometric mean
252 vs. 210 nmol/L, F = 10.50, P < .005) (Fig. 2). The day of life that the test was
performed was not significant in this analysis. We did not find a
significant relationship between stage of chorioamnionitis and either
cortisol concentrations or indices of respiratory support; however, the
number of patients in each group was small (Table 1). Gestation was
significant in these analyses, such that the response to cosyntropin
stimulation increased significantly with increasing gestational age
(poststimulation, F = 9.85, P < .005;
delta cortisol, F = 20.84, P < .001).
Fig. 1.
Serum cortisol concentrations, plotted as log cortisol (nmol/L),
mean ± SEM, in infants exposed to chorioamnionitis (chorio,
)
versus those not exposed (no chorio,
). Values are different between
the two groups (P < .005). Geometric mean
cortisol concentrations (nmol/L), chorio vs no chorio, were as follows:
235 vs 175, 192 vs 145, 195 vs 121 (nmol/L = (µg/dL) × (27.6)).
None of these infants received prenatal or postnatal glucocorticoid.
[View Larger Version of this Image (11K GIF file)]
Fig. 2.
Increase in serum cortisol after cosyntropin (ACTH analog) stimulation,
plotted as log (
cortisol(nmol/L)) versus birth weight, where
cortisol = [stimulated cortisol]-[baseline cortisol]. Patients exposed to chorioamnionitis (
) had significantly higher values for
cortisol than those not exposed (
), (geometric mean 252 vs 210 nmol/L, P < .005). Data are plotted using
birth weight to better visualize individual data points; however, all
analyses were performed using gestational age as a measure of
maturation, which was a significant factor in the relationship,
P < .001. None of these infants received prenatal or
postnatal glucocorticoid.
[View Larger Version of this Image (14K GIF file)]
The presence of chorioamnionitis was also associated with significantly
higher peripheral polymorphonuclear (PMN) neutrophil counts, immature
PMN counts and immature:total neutrophil (I:T) ratios in these infants
immediately after birth (Table 3). Two factors that
might affect these numbers are pregnancy-induced hypertension (PIH) and
labor. Analyzing patients without PIH, the immature PMN and I:T ratio
were still significantly higher in infants with chorioamnionitis, and
the total PMN count trended higher. In those patients delivered after
labor, total PMNs, immature PMNs, and I:T ratios were significantly
higher in patients exposed to chorioamnionitis.
DISCUSSION
In this study, we investigated the relationship of
chorioamnionitis to acute respiratory distress in VLBW infants. We
found that the presence of chorioamnionitis was associated with a
decreased incidence and severity of respiratory disease, measured
qualitatively by exogenous surfactant administration, and
quantitatively with a respiratory acuity score. In addition, we found
that infants exposed to chorioamnionitis had significantly higher serum
cortisol concentrations, providing evidence of a possible mechanism for this effect: acceleration of lung maturity through glucocorticoid action.
Previous studies have provided conflicting information about the
relationship of chorioamnionitis to acute respiratory disease in the
premature infant.1 This disagreement may be due to
differing definitions of chorioamnionitis and differing study designs.
The strong inverse correlation of chorioamnionitis with gestational age
that we and others17 have found may also have influenced the evaluation of this relationship. In our study, the diagnosis of
chorioamnionitis was based solely on placental pathology, and was
determined by one experienced investigator. An additional strength of
this study is that placental pathology was available for 79 of the 84 patients enrolled.
PROM has also been observed to decrease the incidence of RDS in several
studies.8 In addition, one of those studies found that
infants with PROM had increased serum cortisol concentrations immediately after birth.8 A large retrospective review
failed to confirm a protective effect of PROM against
RDS.12 Because the incidence of chorioamnionitis increases
with increasing duration of membrane rupture,6 PROM may
have served as a marker for chorioamnionitis in the previous studies.
In our study, we found that an isolated
2 analysis would
have shown a significant relationship between PROM and a decreased
incidence of RDS (as defined by exogenous surfactant administration);
however, when gestational age was added to the model, that relationship
no longer achieved significance. When chorioamnionitis was added, PROM
contributed no significance to the model; however, chorioamnionitis was
significantly inversely related to the incidence of RDS.
To investigate the etiology of this inverse relationship, we measured
cortisol concentrations in these infants during the first week of life,
and found them to be significantly higher in infants exposed to
chorioamnionitis. Additionally, these infants responded to cosyntropin
stimulation with an increased release of cortisol, compared with
infants without chorioamnionitis. These effects were significant
through the end of the first week of life. Exogenous corticosteroids
have clearly been shown to accelerate lung maturation in the
fetus19 and to decrease the incidence of respiratory
distress syndrome in premature infants.20,21 Together with
other studies that have shown lower serum cortisol concentrations in
infants with RDS,22 these data provide evidence that an
increase in endogenous corticosteroids has a similar effect. We did not
find a significant relationship between the duration of
chorioamnionitis, assessed by stage, and either cortisol concentrations or measures of respiratory support. This may have been because stage 1 chorioamnionitis persists for several days before
progressing6; however, the numbers in each group were
small.
The increased cortisol concentrations found in infants exposed to
chorioamnionitis may result from prenatal stimulation of the
hypothalamic-pituitary-adrenal axis by inflammation. Naeye et
al2 demonstrated on autopsy that evidence of antenatal
infection was associated with an increase in adrenal weight, as well as with a decrease in hyaline membrane formation in the lung. The increased adrenal weight was due to a greater number of cortical cells
in both the fetal and adult (definitive) zones, and to a greater mass
of cytoplasm per cell in the adult zone.
Inflammatory mediators such as IL-1
have been shown to stimulate
secretion of corticotropin releasing factor and ACTH.15,16 Placental production of IL-1
is greatly increased in the presence of
preterm labor and chorioamnionitis.13 We have previously found increased IL-1
concentrations in the tracheal lavage fluid of
infants exposed to chorioamnionitis.4 An additional
indicator of increased prenatal inflammation in this study is the
significant elevation of polymorphonuclear leukocytes in peripheral
blood samples.
In these VLBW infants, chorioamnionitis may also underlie the
previously observed phenomenon that cortisol concentrations during the
first week of life are higher in infants born at very young
gestations.22,25,26 In this study, that phenomenon was explained by the higher incidence of chorioamnionitis at lower gestational age. An analysis examining only the relationship of gestational age to cortisol concentration would have shown a
significant correlation (F = 2.81, P = .028). However, chorioamnionitis, added to the
analysis, accounted for more of the variance (F = 10.4, P = .002), although gestation did not
significantly add to the model (F <.01,
P = .94). This study was limited to VLBW infants; therefore, these data do not address the effect of increasing gestation
on cortisol concentrations throughout the range of prematurity, an
effect that may well be nonlinear in nature.
In contrast, gestation was a significant factor in the ability of these
infants to secrete cortisol in response to stress, as we have
previously reported.27 Additionally, we found that infants
exposed to chorioamnionitis secreted significantly more cortisol in
response to cosyntropin stimulation. The prenatal stimulus of
infection, with resultant increase in cell number and cytoplasm in the
definitive zone of the adrenal cortex,2 may increase the
adrenal gland's ability to produce more cortisol in response to
postnatal stimuli. This more robust response may result in the observed
increase in basal and stimulated cortisol values, as well as the
finding that these infants require less initial respiratory support.
The prevalence of chorioamnionitis in VLBW infants and the apparent
magnitude of its effect on cortisol values in this study would suggest
that this factor should be included in future studies of adrenal
function in this population. Because infants exposed to
chorioamnionitis have less severe acute respiratory illness, studies
that aim to evaluate the normal development of adrenal function in
premature infants by focusing on relatively well premature infants may
be particularly influenced by this factor.
In summary, we have demonstrated an inverse relationship between
chorioamnionitis and acute respiratory distress in the VLBW infant, and
have suggested a mechanism for this observation: exposure to
inflammation leading to increased cortisol production, resulting in
accelerated lung maturation. This finding should not, however, dissuade
clinicians from administering exogenous corticosteroids to women at
risk for preterm delivery, a practice with proven benefit to this
population of infants.20,21 Chorioamnionitis cannot be
reliably diagnosed prenatally, and the study population exposed to
chorioamnionitis still had a significant incidence of respiratory
distress.
Received for publication Apr 22, 1996; accepted Jun 28, 1996.
Reprint requests to (K.L.W.) Department of Pediatrics, Milton
S. Hershey Medical Center, PO Box 850, Hershey, PA 17033-0850.
Supported by grant MCJ-420627 from the Maternal and Child Health
Bureau (Title V, Social Security Act), Health Resources and Service Administration, Department of Health and Human Services.
PROM, prolonged rupture of membranes.
RDS, respiratory distress syndrome.
IL-1
, interleukin-1
.
VLBW, very
low birth weight.
ACTH, adrenocorticotrophic hormone.
RAS, respiratory
acuity score.
PMN, polymorphonuclear leukocyte.
I:T ratio, immature to
total neutrophil ratio.
PIH, pregnancy-induced hypertension.