Blood Cytokines during the Perinatal Period in Very Preterm Infants: Relationship of Inflammatory Response and Bronchopulmonary Dysplasia

Study Population 

The VLGA infants were born alive at Oulu University Central Hospital between November 1998 and November 2002. The parents signed written informed consent, and the study design was approved by the hospital's Ethics Committee. Antenatal and neonatal risk factors were prospectively recorded. The diagnosis of BPD was made at 36 weeks' postmenstrual age. The infants with moderate or severe BPD received either supplemental O2 to maintain oxygen saturation at 88% to 93% or continuous distending pressure to the airways. Severe BPD was defined as the need for > 30% of O2 or mechanical ventilation. RDS was defined as typical chest x-ray findings or the need for either ventilation and supplemental oxygen for at least 48 hours or surfactant therapy for respiratory distress after the chest x-ray. The severity of respiratory distress was expressed as oxygenation index (OI): fraction of inspired O2 × 100 × mean airway pressure/arterial O2 tension. OI during the first day of life was defined as the mean of the prospectively collected recordings.

Diagnosis of Intrauterine Inflammation 

All placentas were fixed in neutral buffered formalin. The rim of membrane was taken from the site of membrane rupture. The umbilical cord specimens were taken from the fetal and placental sides of the umbilical cord and from midway between the sides of insertion. A full-thickness placental specimen was taken from midway between the umbilical cord insertion and the placental margin. Sections were stained with hematoxylin and eosin, and a single investigator (R.H.) blindly assessed the histology. CA was defined as the presence of polymorphonuclear leukocytes in the amnion and chorion decidua. The umbilical cord and the chorionic plate of the placenta also were studied. Each of the 3 tissues was graded for the degree of severity of inflammation based of the density of leukocytes.20 Salafia grade 1 to 2 was defined as “mild” inflammation; grade 3, as “moderate” inflammation; and grade 4, as “severe” inflammation. The severity of CA was classified based the the most severe histological grade found in the umbilical cord, fetal membranes, or chorionic plate of the placenta.

Analysis of Cytokines 

Cord blood was collected from a clamped umbilical cord artery, and arterial blood was collected at 24 hours and at 168 hours after birth. EDTA-plasma samples were separated by centrifugation and stored at −70°C until analysis. Concentrations of the following cytokines were analyzed using the Cytometric Bead Array Kit (BD Biosciences, San Diego, California): interleukin (IL)-12p70, tumor necrosis factor (TNF)-α, IL-6, IL-8, IL-10, granulocyte colony-stimulating factor (G-CSF), granulocyte-macrophage colony-stimulating factor, eotaxin, IL-4, IL-3, and IL-1β. Bead populations with distinct fluorescence intensities for specific soluble proteins were measured with flow cytometry. The interassay variation coefficients ranged from 8% to 20% (1 to 10 pg/mL), 3% to 10% (70 to 80 pg/mL), and 2% to 6% (527 to 630 pg/mL).

Statistical Analysis 

Statistical analysis was performed using SPSS 16.0 for Windows (SPSS Inc, Chicago, Illinois). The infants with and without exposure to CA were analyzed together and also as 2 separate populations. In the initial analyses, the correlations among cytokines, risk factors (Table I), and outcome variables were assessed using Spearman's rank correlation. The trends for cytokine concentrations during the perinatal transition in infants with and without BPD were evaluated using gestational age at birth as the continuous independent variable in the covariance analysis. Cytokines and early neonatal risk factors were evaluated for predicting the risk of BPD by defining the receiver operating characteristic (ROC). Stepwise logistic regression analysis was subsequently used to test a single cytokine for predicting the risk of BPD. The other risk factors evaluated were gestational age at birth, intrauterine growth retardation, CA, and severity of respiratory distress during the first day of life. All tests were 2-tailed.

Table I. Characteristics of the 128 VLGA infants included in the study

	Variable	Median (min, max) or number of infants (%)
	Gestational age at birth, weeks	29.4(24.1,31.9)
	Birth weight, g	1170(370,2210)
	Male sex	68/128(53%)
	Small for gestation	29/128(22.6%)
	Antenatal steroid use	108/124(87.1%)
	Histological CA⁎	48/123(39.0%)
	Rupture of fetal membranes	32/128(25%)
	Preeclampsia	36/128(28.1%)
	Surfactant therapy	77/128(60.1%)
	Mechanical ventilation, days	3.2(0.0,95.5)
	RDS	84/128(65.6%)
	BPD†	32/128(25.0%)
	Moderate at 36 postmenstrual weeks	27/128(21.1%)
	Severe at 36 postmenstrual weeks	5/128(3.9%)

⁎ One case revealed mild inflammation in the chorionic plate of the placenta without detectable CA in reflecting membranes or funisitis. † Eight infants did not have RDS.

Chorioamnionitis and Cytokines in Cord Blood 

The infants exposed to CA did not have an increased risk for BPD compared with the infants without exposure to CA (31.3% vs 22.7%; P = .25). After adjustment for gestational age at birth, the odds ratio (OR) was 1.20 (95% confidence interval [CI] = 0.40 to 2.80). Cord concentrations of IL-6, IL-8, IL-10, and G-CSF were higher in infants with severe CA than in those with mild to moderate CA (P = .01) (Table II). In severe CA, those infants with cord IL-8 concentrations above the median value (> 1 682 pg/ml) tended to be at greater risk for BPD than those with values below the median (14% vs 67%; P = .07). Cytokine concentrations tended to be higher in infants with funisitis (n = 22; BPD, 45.5%) or inflammation of the chorionic plate (n = 33; BPD, 33.3%) than in those with isolated inflammation in reflecting membranes (n = 14; BPD, 21.4%).

Table II. Cord blood cytokine concentrations in VLGA infants with and without BPD based on CA⁎ exposure

	No BPD
	Median (IQ), pg/mL	No CA (n = 58) GA 29.9 ± 1.6 weeks	Mild CA (n = 12) GA 28.6 ± 1.9 weeks	Moderate CA (n = 13) GA 29.0 ± 1.7 weeks	Severe CA (n = 8) GA 29.0 ± 1.2 weeks
	IL-6	20(4.4,78)	6.9(1.0,65)	329(7.0,4135)	458(4.2,2757)
	IL-8	53(22,133)	160(27,2911)	246(50,1100)	428(18,2118)
	IL-10	3.1(1.9,4.7)	2.2(1.8,4.8)	3.6(2.2,7.0)	5.0(1.8,14)
	G-CSF	29(11,78)	27(11.8,1800)	46(0,9974)	2169(136,7652)

	BPD
	Median (IQ), pg/mL	No CA (n = 17) GA 28.0 ± 2.0 weeks	Mild CA (n = 4) GA 27.3 ± 2.1 weeks	Moderate CA (n = 6) GA 27.1 ± 1.9 weeks	Severe CA (n = 5) GA 27.0 ± 1.9 weeks
	IL-6	13(5.5,25)	8.2(3.7,53)	25.3(0,37)	2412(4.2,2757)
	IL-8	55(28,88)	96(65,548)	90(43,2450)	3504(2203,11643)
	IL-10	2.4(1.8,2.8)	2.2(1.2,3.1)	3.3(2.8,23)	10(4.3,73)
	G-CSF	2(11,115)	84(3.3,2222)	441(319,3945)	9540(431,10725)

GA, gestational age, mean±SD; IQ, interquartiles.

⁎ Funisitis complicated CA as follows: mild CA (n = 1), moderate CA (n = 10), severe CA (n = 11). The grades of funisitis were as follows: 7 mild (3 associated with BPD), 10 moderate (5 associated with BPD), 5 severe (2 associated with BPD).

Cytokines During the Perinatal Period 

In the infants born with CA, the high concentration of the inflammatory cytokines in cord blood decreased during day 1, whereas in the infants not exposed to CA, the cytokine concentrations increased. On day 1, the concentrations of IL-6, IL-8, and IL-10 were higher in the infants without CA than in those with CA (Table III; available at www.jpeds.com).

Table III. Cytokine concentrations during the perinatal period in the VLGA infants (n = 48 CA; n = 75 no CA)

		Cord blood at birth		Day 1		Day 7
		CA	No CA	CA	No CA	CA	No CA
	IL-6, pg/mL
	Median (IQ)	37.0(7.0,458)	13.8(4.9,62.7)	19.0(9.0,14.4)	52.6(14.1,174)	8.3(2.4,22.1)	5.6(2.7,11.7)
	Mean	701	253	54.4	361	17.1	32.5
	P, CA vs no CA⁎	.001		.001		NS
	IL-8, pg/mL
	Median (IQ)	216(58.7,1683)	53.9(24.8,103)	76.4(38.9,151)	104(53.6,211)	50.4(38.9,151)	24.7(15.7,45.4)
	Mean	1108	216	108	174	69.8	46.0
	P, CA vs no CA	<.0001		.001		NS
	IL-10, pg/mL
	Median (IQ)	3.3(2.2,8.2)	2.6(1.9,4.3)	3.2(1.5,4.8)	5.0(2.9,11.1)	2.5(1.2,4.2)	2.0(0.5,3.1)
	Mean	7.5	4.4	3.7	13.4	11.8	2.4
	P, CA vs no CA	NS		.006		NS
	G-CSF, pg/mL
	Median (IQ)	319(50.9,2223)	28.2(10.4,89.9)	37.1(11.1,281)	138.4(38.8,813)	9.5(3.6,24.9)	7.0(3.3,13.4)
	Mean	1589	603	416	650	23.5	39.3
	P, CA vs no CA	<.0001		NS		NS

IQ, interquartiles; NS, not significant.

⁎ Covariance analysis, using gestational age at birth as continuous independent variable.

Plasma concentrations of cytokines in cord blood were not increased in the infants who developed BPD. However, concentrations of IL-8, IL-10, and G-CSF on day 1 of life were higher in these infants than in the infants without BPD (Table IV; available at www.jpeds.com). Table V (available at www.jpeds.com) shows the concentrations of the 4 cytokines in infants with moderate (n = 27) and severe (n = 5) BPD. In infants with severe BPD, IL-10 concentration peaked on day 1 (median [interquartiles] = 14.3 [7.4, 255] pg/mL) and tended to be higher than that in infants with moderate BPD (5.4 [3.5, 10.4] pg/mL).

Table IV. Cytokine concentrations during the perinatal period in the VLGA infants (n =32 BPD; n = 96 no BPD)

		Cord blood at birth		Day 1		Day 7
		BPD	No BPD	BPD	No BPD	BPD	No BPD
	IL-6, pg/mL
	Median (IQ)	13.2(4.8,42.0)	24.3(6.9,235)	44.8(13.0,276)	30.0(11.4,81.8)	7.9(3.5,21.0)	4.9(2.4,14.2)
	Mean	560	396	275	222	16.9	28.5
	P, BPD vs no BPD⁎	NS		.03		NS
	IL-8, pg/mL
	Median (IQ)	89.6(47.2,962)	61.4(27.4,315)	173.3(80.3,369)	72.0(37.3,127)	50.7(27.4,96.2)	25.5(13.7,51.0)
	Mean	1146	362	257	107	86.9	45.6
	P, BPD vs no BPD	NS		.001		NS
	IL-10, pg/mL
	Median (IQ)	2.6(2.0,8.4)	3.5(1.9,5.9)	5.9(3.6,11.7)	3.4(1.9,5.9)	3.5(1.5,4.8)	2.0(0.5,2.8)
	Mean	8.0	4.7	24.4	4.9	11.5	4.5
	P, BPD vs no BPD	NS		.001		NS
	G-CSF, pg/mL
	Median (IQ)	131(14.3,886)	33.9(17.8,217)	280(65.1,1661)	58.3(21.4,238)	11.3(6.1,23.9)	7.0(3.1,13.4)
	Mean	1751	702	918	421	18.2	39.6
	P, BPD vs no BPD	NS		.005		NS

IQ, interquartiles; NS, not significant.

⁎ Covariance analysis, using gestational age at birth as continuous independent variable.

Table V. Cytokine concentrations during the perinatal period for 27 VLGA infants who developed moderate BPD and 5 VLGA infants who developed severe BPD⁎

		Cord blood at birth		Day 1		Day 7
		Moderate BPD	Severe BPD	Moderate BPD	Severe BPD	Moderate BPD	Severe BPD
	IL-6, pg/mL
	Median(IQ)	12.6(5.1,52.0)	13.7(3.7,38.7)	33.1(10.0,184)	113(20.1,1821)	7.9(3.5,21.9)	7.9(3.2,25.4)
	IL-8, pg/mL
	Median(IQ)	89.6(48.5,11623)	89.6(0.5,93.1)	165(18.4,365.7)	203.5(114,1025)	52.1(30.6,104)	31.7(20.8,54.0)
	IL-10, pg/mL
	Median (IQ)	2.8(2.0,11.2)	2.3(1.8,2.4)	5.4.(3.5,10.4)	14.3(7.4,255)	3.6.(1.5,4.9)	1.7(0.8,4.6)
	G-CSF, pg/mL
	Median (IQ)	146(14.8,2223)	35.6(3.3,420)	283(89.4,1752)	215(33.8,1537)	107(5.8,23.0)	13.0(3.7,37.1)

⁎ Moderate BPD was defined as the requirement for 22% to 30% of oxygen in inspired gas to maintain oxygen saturation between 88% and 93% or requirement of continuous distending airway pressures at postmenstrual age 36 weeks. Severe BPD was defined as requiring > 30% of inspired oxygen and/or mechanical ventilation at 36 weeks.

In Figure 1 (available at www.jpeds.com), cytokine concentrations are shown separately for the 48 infants born with CA and the 75 infants not exposed to CA. The cytokine concentrations were analyzed in both infants who developed BPD and those who did not develop BPD. In the infants exposed to CA, the concentrations of inflammatory cytokines decreased during the first week of life in both the infants with BPD and those without BPD; however, the decrease appeared more modest during the first day in the infants with BPD. On day 1, IL-8 concentration was higher in the infants with BPD than in those without BPD. Among those infants without CA who developed BPD, cytokine concentrations increased between birth and day 1, but then decreased between day 1 and day 7. On day 1, concentrations of IL-8, IL-10, and G-CSF were higher in the infants with BPD than in those without BPD.

View Large Image Download to PowerPoint

Figure 1. Concentration of plasma cytokines during the perinatal transition of VLGA infants born with CA (n = 48) and without CA (no CA; n = 75). The results for infants developing BPD (n = 32) and for infants with no BPD (n = 91) are shown. Plasma concentrations (pg/mL) are shown as box blots: medians, interquartile ranges, 10th and 90th percentiles, and means. *Significant (P < .05) and **very significant (P < .01) differences between the 2 groups (covariance analysis, using gestation at birth as covariate).

Cytokines as Risk Factors of BPD 

Using ROC analysis, the concentrations of IL-8, IL-10, and G-CSF were found to predict the risk of BPD in 1-day-old infants. IL-8 predicted the risk of BPD for both 1-day-old infants exposed to CA (sensitivity, 0.611; specificity, 0.689 at 103.3 pg/mL) and those not exposed to CA (sensitivity, 0.833; specificity, 0.725 at 119 pg/mL) (Figure 2A, B). IL-10 predicted the risk of BPD in infants exposed to CA (P = .041; sensitivity, 0.556; specificity, 0.690 at 3.7 pg/mL). IL-10 and G-CSF predicted the risk of BPD in infants not exposed to CA (P = .002; sensitivity, 0.728; specificity, 0.650 at 5.25 pg/mL and P = .022; sensitivity, 0.722; specificity, 0.647 at 144 pg/mL, respectively). The degree of respiratory distress, expressed as the mean OI during the first day of life, predicted the risk of BPD in ROC analysis (Figure 2C, D).

View Large Image Download to PowerPoint

Figure 2. ROC analysis of the risk of BPD for infants born with CA and those without CA (no CA). Mean plasma IL-8 concentration at age 1 day (A, 83.3 [41.2 to 151] pg/ml; B, 98.9 [51.8 to 211]) and mean oxidation index during the first day (C, 26.6 [19.7 to 50.5 cmH2O/kPa]; D, 31.0 [16.4 to 48.0]) were evaluated. The areas under the ROC curves (mean ± standard error) were as follows: A, 0.694 ± 0.076, P = .034; B, 0.829 ± 0.053, P < .0001; C, 0.705 ± 0.083, P = .028; D, 0.788 ± 0.058, P = .0001.

For all 128 infants, the risk of BPD was evaluated using stepwise logistic regression analysis. Of the independent risk factors, degree of prematurity and severity of respiratory distress during day 1 of life significantly explained the risk. Cytokine concentrations in the 1-day-old infants were categorized according to median values and entered separately into the equation. With the gestational age as an independent variable, IL-8 (OR = 3.0; 95% CI = 1.2 to 7.5; P = .018), IL-10 (OR = 3.7, 95% CI = 1.5 to 9.3; P = .005), and G-CSF (OR = 4.5; 95% CI = 1.7 to 11.8; P = .002) predicted the risk of BPD. With both gestational age and the mean OI during the day 1 as independent variables, IL-8 (OR = 2.5; 95% CI = 1.0 to 6.2; P = .048), IL-10 (OR = 3.6; 95% CI = 1.4 to 8.8; P = .006), and G-CSF (OR = 3.2; 95% CI = 1.3 to 8.1; P = .013) predicted the risk of BPD.