Main

An early severe inflammatory reaction has been found in the lungs of preterm infants with the IRDS and appears to be involved in the development of BPD. This inflammatory reaction is characterized by accumulation and activation of leukocytes(1–5) and local release of inflammatory mediators(1, 2, 4–10). Activated leukocytes are predominantly PMN and alveolar macrophages(1–5). Release of elastase by activated PMN in the presence of low levels of the main proteinase inhibitorα1-PI has been associated with lung connective tissue destruction and the development of BPD(1, 2, 4, 5, 7). These destructive effects of elastase are aggravated by pneumonia or hyperoxic exposure(11–13). Activated alveolar macrophages release more TNF-α in infants with IRDS, who develop BPD, compared with those with an uncomplicated course(9, 10). Furthermore, inflammatory mediators such as PAF(6) leukotriene B4(4, 5), IL-1β(5), IL-8(4), and complement component C5-derived anaphylatoxin(C5a)(4, 8) have been found in tracheal aspirates of infants at risk for BPD. These mediators likely contribute to respiratory insufficiency by several mechanisms including attraction and activation of PMN and breakdown of pulmonary vascular endothelium with subsequent protein leakage into the small airways(4–6, 14, 15). This pulmonary protein-rich edema is thought to play an important role in the pathogenesis of BPD(16).

Similar to infants with IRDS, accumulation of activated PMN in the lungs is found in children and adults with severe clinical disorders that result in the ARDS. This influx of PMN into the lungs is accompanied by a decrease of the circulating PMN count and massive injury of lung tissue due to local release of oxygen radicals and proteolytic enzymes (e.g. elastase)(17–19). Several inflammatory mediators are thought to play a role in PMN activation and accumulation in the lungs of ARDS patients(17, 19, 20). However, some of these mediators do not always act exclusively in the airways. Increase of the TNF-α plasma concentration and systemic activation of complement have been described in ARDS patients(19–22). Furthermore, systemically administered PAF mediates lung tissue injury in animal experimental studies(23).

We questioned whether accumulation of PMN in the lungs of preterm infants with IRDS is accompanied by activation of circulating PMN with subsequent systemic release of elastase. Circulating inflammatory mediators including TNF-α, activated complement compounds, and PAF might be involved in this activation process. Therefore, this study was undertaken to obtain more insight in the process of PMN activation in preterm infants with IRDS. We measured 1) the total leukocyte count, 2) the PMN and monocyte count, 3) the plasma concentration of elastase-α1-PI complex (indicator of PMN activation),4) TNF-α (indicator of monocyte activation), 5) PAF-IC in plasma (indicator of PAF release), and 6) the plasma C3a des-Arg concentration (indicator of complement activation) simultaneously during the first 5 d of life in preterm infants with severe IRDS and in preterm infants without respiratory problems.

METHODS

Patients. Thirty preterm infants consecutively admitted to the neonatal intensive care unit of the Beatrix Childrens Hospital, University Hospital Groningen, were included in this study. These infants fitted the following criteria for study enrolment: 1) no maternal infection, amnionitis, or prolonged rupture of membranes (>24 h before birth);2) gestational age between 27 and 33 wk; 3) birth weight appropriate for gestational age; and 4) no major congenital malformations and/or evidence of infection. Fifteen infants showed clinical and radiologic signs of severe IRDS (IRDS group). Severe IRDS was defined as oxygen requirement of more than 30% for adequate oxygenation, artificial ventilation dependency because of respiratory insufficiency, and grade 3 or 4 abnormalities on a chest x-ray according to the Giedion score(24). Fifteen preterm infants without IRDS or other significant problems served as a reference group. The study was approved by the hospital ethical committee, and informed consent was obtained from the parents of all infants.

There were no differences between the IRDS and the reference infants for prenatal factors that could influence activation of circulating PMN, release of TNF-α PAF, or activation of complement (Table 1). The mothers of four infants in the IRDS group and of four infants in the reference group showed HELLP and/or toxicosis before delivery. Pregnant women, who develop HELLP syndrome show complement activation at delivery(25). However, complement components do not cross the placenta(26). About half of the newborn infants whose mothers have HELLP and/or toxicosis during pregnancy are known to have diminished PMN counts at birth(27). However, PMN counts in the four reference infants could not be distinguished from the other reference infants, whereas PMN counts in the four IRDS infants were lower than those of the four reference infants. Dexamethasone was given more than 2 d before delivery to the mothers of five infants in both the IRDS and the reference groups. Dexamethasone does not clearly affect plasma complement activation, but decreases leukocyte activation and release of bioactive mediators(28) and increases activity of PAF-acetylhydrolase, the natural occurring inhibitor of PAF(29). However, elastase-α1-PI and TNF-α plasma concentrations and the PAF-IC in the five IRDS and the five reference infants did not differ from the other infants in each group.

Table 1 Prenatal factors that may influence the activation pattern of the studied parameters, and postnatal characteristics of the IRDS and reference infants*
Full size table

Postnatal characteristics of the IRDS and the reference group are presented in Table 1. All IRDS infants had significantly lower 1- and 5-min Apgar scores and significantly lower arterial umbilical pH values than the reference infants. Despite of these signs of birth asphyxia the IRDS infants showed no other organ failure but respiratory insufficiency from birth. All IRDS infants in this study required endotracheal intubation at birth and were artificially ventilated throughout the study period.

Patient management. Infants admitted with hypoxia and respiratory insufficiency were artificially ventilated with supplemental oxygen to maintain arterial Po2 between 7.5 and 10.0 kPa, and Pco2 between 5.5 and 6.5 kPa. When infants showed clinical signs of respiratory distress, required more than 30% oxygen for adequate oxygenation and artificial ventilation because of respiratory insufficiency, they were eligible for the Dutch Multicenter study “Nedsurf” for surfactant replacement therapy (Alvofact, Boehringer Ingelheim, Germany). In the Nedsurf study, IRDS infants are prospectively classified according to the roentgenologic severity of the disease using the Giedion score(24). Infants showing Giedion 1 or 2 abnormalities on a chest x-ray were considered to have moderate IRDS, whereas infants with a Gideon 3 or 4 chest x-ray were considered to have severe IRDS. In our study all infants of the IRDS group had severe IRDS. According to the criteria of the Nedsurf study, 11 of the 15 infants of the IRDS group received surfactant(100 mg/kg body weight endotracheally). The remaining four infants were not given surfactant because of bloody tracheal secretions.

In all study infants superficial cultures were obtained immediately after admission and on d 4. None of these infants showed colonization with pathogenic bacteria, especially those, that are known to be predominant causes of early onset septicemia in our neonatal intensive care unit. In 12 IRDS infants, cultures of blood, tracheal effluent, and urine were taken because of a decrease of the leukocyte count. The blood, tracheal effluent, and urine cultures remained negative in 10 of these 12 infants. These 10 infants were included in the present report and did not show other clinical signs of an infection but respiratory failure, which was thought to be caused by IRDS. Two other infants were excluded from the study. One of these infants showed a pneumonia/sepsis with group B streptococci, whereas another infant showed anEscherichia coli sepsis.

According to the local protocol, infants of less than 34 wk of gestational age with respiratory insufficiency requiring continuous positive airway pressure or artificial ventilation were screened with echocardiography for a patent ductus arteriosus on the 2nd to 4th d of life. When the ductus arteriosus was open, these infants were treated with indomethacin (Indocid, Merck Sharpe & Dohme, Haarlem, The Netherlands) i.v. Seven infants of the IRDS group were treated for an open ductus arteriosus.

Infants received transfusions of cryoglobulin-poor plasma or packed red blood cells with cryoglobulin-poor plasma for replacement of blood taken for routine laboratory investigations. The transfusion volume for each infant in the study did not exceed 10% of the calculated blood volume in any 24-h period. Furthermore, cryoglobulin-poor plasma is stored frozen and contains only native inactive plasma proteins, whereas packed red blood cells contain only 10 to 20% plasma, including plasma proteins, which are slightly activated by the processing procedure(30). Therefore, replacement transfusions are not considered to influence the findings of this study in the IRDS infants nor the reference infants.

Insertion or removal of a venous or arterial umbilical catheter or a radial artery catheter was approved by the attending neonatologist. The patency of arterial umbilical catheters and arterial radial catheters was maintained by continuous infusion of a 0.9% NaCl solution containing 3 U/mL heparin at a rate of 0.5 to 0.7 mL/h.

Heparin is known to inhibit the complement system(31). We have tried to minimize effect of heparin on complement activation in the samples of our IRDS infants. In these infants, blood samples were obtained from the venous umbilical catheter, which was not heparinized nor flushed with heparin before collecting the actual blood sample to be used for analysis. Although we did not measure the heparin concentration in the blood sample that was analyzed, this should be very low. It has been demonstrated that complement inhibition occurs especially with heparin concentrations similar to those needed for anticoagulation. However, despite heparinization, complement activation has still been demonstrated in humans and animals undergoing extracorporeal life support(32, 33). Furthermore, it has been shown that the heparin infusion that we used does not influence clotting in preterm infants(34). We, therefore, do not consider heparin as used in our infants of great influence on the results of our study. All infants of the IRDS group had a venous umbilical catheter, whereas 10 of them also had an arterial umbilical catheter (n = 7) or an arterial radial catheter(n = 3). The infants of the reference group did not receive an indwelling catheter.

Study protocol. Blood samples were taken from the venous umbilical catheter (IRDS infants) or were obtained by venipuncture (reference infants) at 6 to 12 h from birth (1st d of life), and on the 2nd, 3rd, and 4th to 6th d of life. The first blood sample was taken before treatment with surfactant, indomethacin, or blood products was performed. Each blood sample was taken during routine blood sampling and used for the determination of the total leukocyte count, the PMN and monocyte count, the plasma concentration of elastase-α1-PI and TNF-α, the plasma PAF-IC, and the activation of complement (C3a des-Arg).

At each sampling 0.3 mL blood was anticoagulated with EDTA (0.01 M) for determination of the total leukocyte count and differential cell counts. Another 0.5 mL of blood was anticoagulated with citrate (0.3%). This citrated blood sample was centrifuged at room temperature immediately after collection. After centrifugation the plasma sample was immediately stored at -80°C. Because blood sampling had to be as limited as possible and only one blood sample per study infant could be centrifuged and stored immediately after collection, we decided to determine the concentration of elastase-α1-PI, C3a des-Arg, and PAF-IC in this sample. With this technique even better results were obtained with regard to elastase-α1-PI concentrations and PAF-IC than with centrifuging EDTA samples after cell counting was performed. The latter, however, would be a better alternative for determination of C3a des-Arg regarding in vitro activation of complement, which was now limited by rapid processing to storage of the citrated samples. The elastase-α1-PI/PMN ratio was calculated to exclude the influence of the circulating PMN count on the systemic release of elastase. Also the TNF-α/monocyte ratio was calculated to exclude the influence of the circulating monocyte count on the systemic release of TNF-α. Both ratios are expressed in nanograms/106 PMN or monocytes

Assays. The total nucleated cell count was determined using a cell counter (Hemolog, Coulter Electronics, UK). The total leukocyte count was determined after correction of the total nucleated cell count for the presence of nucleated red cells (erythroblasts). The leukocytes were differentiated by morphologic classification of 100 cells in blood films, which were stained by the May-Grünwald-Giemsa method.

Plasma elastase-α1-PI concentrations were quantitated by an ELISA (Merck, Darmstadt, Germany). TNF-α was determined with a RIA(TNF-α IRMA, Medgenix, Brussels, Belgium). The C3a des-Arg plasma concentration was determined by RIA (Upjohn Co., Kalamazoo, MI).

For the determination of the PAF-IC, platelets were isolated from citrated platelet-rich plasma of healthy adult volunteers by filtration through Sepharose CL-2B (Pharmacia Biotech Inc., Stockholm, Sweden). These isolated platelets were resuspended in PBS to a final platelet concentration of 50× 109/L. A 700-μL aliquot of this platelet suspension was dispensed into a cuvette, incubated at 37°C, and stirred at 900 rpm with a metallic rod within an aggregometer (Chronolog Corp., Havertown, PA) which recorded the light transmission pattern of the platelet suspension. Then a 50-μL aliquot of plasma of the study infants or normal human adults was then added to the cuvette after 5 min followed by 5 μL of the PAF C16 suspension (10 mg/mL, Cayman Chemical, Ann Arbor, MI). Platelet aggregation after addition of PAF was allowed to continue until maximal increase in light transmission. The percentage of platelet aggregation after addition of PAF is inversely correlated with the PAF-IC of plasma. The PAF-IC is expressed as the percentage of maximal PAF inhibition in normal human plasma.

Statistical analysis. Data are presented as mean ± SD or as median with 25th and 75th percentiles as appropriate. The χ2 test with Yates's correction for continuity was used for comparison of nominal data between the IRDS and the reference group. Comparison of gestational age, birth weight, 1- and 5-min Apgar scores and arterial umbilical pH values between the two groups was carried out using the unpaired t test.

For total leukocyte counts and TNF-α, two-way analysis of variance for repeated measures was used to test the effects of time and group-time interactions. In addition, an unpaired t test was performed to compare the values of these parameters between the two groups on d 1, 2, 3, and 4 to 6. A paired t test was used to compare total leukocyte count and TNF-α on d 2, 3, and 4 to 6 with the values of these parameters on d 1.

For leukocyte differential cell counts, plasma elastase-α1-PI and C3a des-Arg concentrations, plasma elastase-α1-PI/PMN and TNF-α/monocyte ratio, and the PAF-IC, means of five observations were calculated for each individual infant and compared between the IRDS and the reference group, thus testing intergroup difference for this parameter. The Mann-Whitney U test and the Wilcoxon signed rank test were used to determine specific differences between and within the two groups for the PMN and monocyte count, the plasma elastase-α1-PI and C3a des-Arg concentration, the plasma elastase-α1-PI/PMN and TNF-α/monocyte ratio, and the PAF-IC.

The significance level was adjusted for multiple comparisons by means of the Bonferroni correction as appropriate. Statistical significance was assumed when the p value was less than 0.05.

RESULTS

Ventilatory support. The mean Fio2 value in the IRDS infants decreased significantly from 0.72 ± 0.14 on d 1 to 0.46± 0.25 on d 5 (p < 0.05). Peak inspiratory pressure values did not change significantly between d 1 and 5 (24 ± 3 cm H2Oversus 23 ± 8 cm H2O) in these infants. Two reference infants showed apnea and bradycardia and required nasal continuous positive airway pressure without supplemental oxygen for 1 and 4 d, respectively. One reference infant showed signs of a wet lung shortly after birth and needed endotracheal intubation and artificial ventilation during the 1st d of life with maximal Fio2 and peak inspiratory pressure values of 0.27 and 18 cm H2O, respectively. The remaining 12 reference infants showed no respiratory problems.

Total leukocyte count. The total leukocyte count (mean ± SD) of the IRDS infants was significantly lower than that of the reference infants (p = 0.001, analysis of variance for repeated measures). The difference between both study groups was significant at each sampling point(Fig. 1). In the IRDS group, the total leukocyte count decreased from 5.4 ± 1.9 × 109/L on d 1 to 3.5 ± 2.0 × 109/L on d 2 (p < 0.05 versus d 1) and increased afterward to 5.1 ± 2.2 × 109/L on d 4 to 6. In the reference group, the total leukocyte count did not change significantly between d 1 and d 4 to 6.

Figure 1
figure 1

Total leukocyte count in preterm infants with(open circles) and without IRDS (closed circles) during the first 5 d of life. Data are expressed as mean ± SD. *p< 0.05, **p < 0.01, ***p < 0.001 for IRDS group compared with reference group.

Full size image

PMN count. The median PMN count (25th to 75th percentile) of the IRDS group was significantly lower than that of the reference group from the 2nd d of life (Fig. 2). The median PMN count did not change significantly in both study groups throughout the study period.

Figure 2
figure 2

PMN count in preterm infants with (open boxes) and without IRDS (gray boxes) during the first 5 d of life. Data are presented as box graphs showing the median values(horizontal plane line) ranges of 50% around the median value(boxes), and the 10th and 90th percentile (error bars).**p < 0.01, ***p < 0.001 for IRDS group compared with reference group.

Full size image

Elastase -α1-PI complex and elastase -α1-PI/PMN ratio. From d 2, the median plasma elastase-α1-PI complex concentration (25th to 75th percentile) of the IRDS infants was significantly lower than that of the reference infants (Fig. 3). The median plasma elastase-α1-PI complex concentration did not change significantly during the first 5 d of life in both groups.

Figure 3
figure 3

Plasma concentrations of elastase-α1-PI complex in preterm infants with (open boxes) and without IRDS(gray boxes) during the first 5 d of life. Data are presented as box graphs showing the median values (horizontal plane line), ranges of 50% around the median value (boxes), and the 10th and 90th percentile (error bars). *p < 0.05 for IRDS group compared with reference group.

Full size image

From d 2, the median elastase-α1-PI/PMN ratio (25th to 75th percentile) of the IRDS group was significantly higher than that of the reference group (Fig. 4). The median elastase-α1-PI/PMN ratio did not show significantly changes within each study group throughout the study period.

Figure 4
figure 4

Elastase-α1-PI/PMN ratio in preterm infants with (open boxes) and without IRDS (gray boxes) during the first 5 d of life. Data are presented as box graphs showing the median values (horizontal plane line), ranges of 50% around the median value (boxes), and the 10th and 90th percentile (error bars). *p < 0.05 for IRDS group compared with reference group.

Full size image

Monocyte count. The median monocyte count (25th to 75th percentile) of the IRDS group was significantly lower than that of the reference group from the 3rd day of life (0.3 (0.08-0.4) × 109/Lversus 0.8 (0.5-0.9) × 109/L on d 3, p < 0.01; 0.4 (0.2-0.9) × 109/L versus 1.1 (1.0-1.6)× 109/L on d 4 to 6, p < 0.05). The median monocyte count within each study group did not change during the first 5 d of life.

TNF -α and TNF -α/ monocyte ratio. The plasma concentration of TNF-α (mean ± SD) of the IRDS group did not differ from that of the reference group (p = 0.244, analysis of variance for repeated measures). The TNF-α concentration did not change significantly in the IRDS group (8.0 ± 7.5 ng/mL on d 1; 13.7 ± 11.5 ng/mL on d 4 to 6) and in the reference group (13.2 ± 9.1 ng/mL on d 1; 17.1 ± 9.2 ng/mL on d 4 to 6) throughout the study period. During the first 5 d of life the median TNF-α/monocyte ratio of both groups was not different and did not change significantly.

PAF-IC. The median PAF-IC (25th to 75th percentile) of the IRDS infants was significantly lower than that of the reference group on d 1, 2, and 4 to 6 (Fig. 5). The median PAF-IC of both groups did not change significantly during the first 5 d of life.

Figure 5
figure 5

Plasma PAF-IC in preterm infants with (open boxes) and without IRDS (gray boxes) during the first 5 d of life. Data are presented as box graphs showing the median values(horizontal plane line), ranges of 50% around the median value(boxes), and the 10th and 90th percentile (error bars).*p < 0.05, **p < 0.01 for IRDS group compared with reference group.

Full size image

C3a des-Arg. During the first 5 d of life the median C3a des-Arg plasma concentration (25th to 75th percentile) of the IRDS infants was higher than that of the reference infants (Fig. 6). The median C3a des-Arg plasma concentration decreased in the IRDS group without being significant. The median C3a des-Arg plasma concentration in the reference group remained low during the first 5 d of life and did not change significantly.

Figure 6
figure 6

Plasma concentrations of C3a des-Arg in preterm infants with (open boxes) and without IRDS (gray boxes) during the first 5 d of life. Data are presented as box graphs showing the median values(horizontal plane line), ranges of 50% around the median value(boxes), and the 10th and 90th percentile (error bars).**p < 0.01, ***p < 0.001 for IRDS group compared with reference group.

Full size image

DISCUSSION

In this study, artificially ventilated preterm infants with respiratory failure from birth and clinical and radiologic signs of severe IRDS (IRDS group) showed a lower total leukocyte count than preterm infants without respiratory problems (reference group). This lower total leukocyte count was due to a lower PMN and monocyte count. In the IRDS infants the lower PMN count was accompanied by a lower plasma elastase-α1-PI concentration but a higher elastase-α1-PI/PMN ratio than in the reference infants. The latter suggests systemic activation of circulating PMN in the IRDS infants. Simultaneously, indications were obtained for systemic PAF release and complement activation but not for systemic TNF-α release in these infants.

From the 2nd d of life, the PMN count was significantly lower in the IRDS group than in the reference group. This lower PMN count probably reflects influx of PMN into the lungs, which has been reported in preterm infants with severe IRDS(1–5). Simultaneously, we observed lower elastase-α1-PI concentrations in the IRDS infants than in the reference infants in accordance with findings of Ogden et al.(1). The plasma concentrations of elastase-α1-PI of the reference infants did not differ from those of cord blood of healthy term newborn infants(35). The low plasma concentrations of elastase-α1-PI in the IRDS infants might be explained by the lower circulating PMN count that will reduce the absolute amount of elastase to be released. To exclude this influence of the circulating PMN count on the plasma elastase-α1-PI concentration, we have calculated the plasma elastase-α1-PI/PMN ratio. This ratio was significantly higher in the IRDS group than in the reference group from the 3rd d of life, indicating increased release of elastase by circulating PMN in the IRDS infants. This increased release may be caused by circulating inflammatory mediators such as TNF-α, PAF, and complement split products(19–23).

Our data do not indicate systemic release of TNF-α in the IRDS infants. Actually, plasma TNF-α concentrations of our IRDS infants were slightly but not significantly lower than those of our reference infants, similarly to the findings of Murch et al.(9). TNF-α is mainly produced and released by activated monocytes and macrophages(36). Circulating monocytes are known to produce less TNF-α than stimulated alveolar macrophages(37). From d 3, we have observed a significantly lower monocyte count in the IRDS infants than in the reference infants, which might explain the lower TNF-α concentrations in the IRDS infants. In addition, we could not observe increased TNF-α release by circulating monocytes by using the TNF-α/monocyte ratio, which confirms the limited TNF-α producing capacity of circulating monocytes. Thus, release of TNF-α seems to be a localized phenomenon as demonstrated by high TNF-α concentrations found in bronchopulmonary secretions of infants with IRDS, who subsequently develop BPD(9, 10).

Throughout the study period the inhibition of PAF by plasma of the IRDS infants was lower than by plasma of the reference infants. PAF inhibition is represented by PAF-acetylhydrolase, which is consumed by complexing PAF(38). PAF inhibiting capacity is decreased in newborn preterm infants due to preexisting low plasma activity of PAF-acetylhydrolase(39). However, PAF inhibition was lower in the IRDS group than in the reference group, being significant on d 1, 2, and 4 to 6. Therefore, the lower inhibition of PAF in plasma of our IRDS infants can be considered as an indication of PAF-acetylhydrolase consumption due to increased PAF release in the circulation. Release of PAF in blood has been demonstrated in preterm ventilated infants during the 1st wk of life(40) and confirms our findings in the IRDS infants. Systemic release of PAF is likely to contribute to (further) respiratory insufficiency in the IRDS infants by systemic activation of PMN and complement, thus mediating increased pulmonary vascular permeability with formation of protein-rich edema(15, 16, 41).

In the IRDS group, activation of the complement system was represented by higher plasma concentrations of C3a des-Arg, the inactive split product of C3. Increased C3a des-Arg plasma concentrations are described in preterm and term infants with respiratory insufficiency, who had birth asphyxia or circulatory insufficiency(42), and in term neonates with severe respiratory failure requiring extracorporeal life support(35). Recently, Groneck et al.(4, 8) have found that the complement anaphylatoxin C5a is present in plasma and tracheobronchial aspirate fluid of preterm infants with IRDS at risk for chronic lung disease. These reports support our findings that the complement system is activated in newborn infants with severe respiratory distress. C3a and C5a are known to increase vascular permeability(43) and might contribute to pulmonary edema formation in IRDS infants. Furthermore, C5a contributes to PMN chemotaxis, aggregation, and local sequestration in the lungs(20).

We found higher C3a des-Arg plasma concentrations in both our reference and IRDS infants than was demonstrated by others(44). This might be explained by differences in methods. We sampled blood with citrate instead of EDTA. Citrate probably allows more in vitro activation of the complement system than EDTA, but this was limited by rapid processing of the samples to storage at -80°C. Most importantly, we used a commercially available RIA. In this assay polyclonal antibodies to C3a also measures determinants present on the native C3 molecule. Burger et al.(45) used a highly sensitive C3a assay that measures C3a without being affected by the presence of C3.

The C3a des-Arg values found in our IRDS infants were higher than those found in our reference infants. We did not find evidence of infections in these infants as demonstrated by negative cultures of blood, tracheal effluent, and urine. The low TNF-α plasma concentrations further confirm the absence of infections in our IRDS infants because very high TNF-α plasma concentrations have been demonstrated in newborn infants with sepsis(46). However, apart from infections, several explanations can be proposed for the high C3a des-Arg values in our IRDS infants. First, all of these infants showed low Apgar scores and low arterial umbilical pH values at birth, indicating perinatal asphyxia (hypoxemia and acidosis). Hypoxemia and tissue hypoperfusion are accompanied by release of activated complement components(47), which might explain the high C3a des-Arg plasma concentrations shortly after birth. Second, all IRDS infants required artificial ventilation with high peak inspiratory pressures and high Fio2 values throughout the study period. Tissue trauma is accompanied by local complement activation(48). The aforementioned C5a in tracheobronchial aspirate fluid of preterm infants with IRDS at risk for BPD(4, 8) probably leaks into the circulation. Third, in ventilated infants with severe IRDS, increased plasma factor XII and kallikrein activity has been described(49, 50). Both activated factor XII and kallikrein are able to induce complement activation(51). Finally, low plasma concentrations of C1 esterase inhibitor has been described in preterm infants(52). C1 esterase inhibitor is one of the main inhibitors of the complement system. Release of activated complement products out of the lungs together with activated factor XII and kallikrein and low plasma concentrations of C1 esterase inhibitor might explain the high C3a des-Arg plasma concentrations in our IRDS infants throughout the study period.

The cause of the lower PAF-IC values in our IRDS infants also is not yet clarified. Similar to activation of complement, activation of PAF may be caused by hypoxemia due to birth asphyxia(53) and tissue injury due to artificial ventilation(6). The production and release of inflammatory mediators such as PAF and activated complement factors may become self-perpetuating. Systemic release of PAF causes vasoconstriction, thus contributing to tissue ischemia in several organs(41) and subsequent activation of complement(47). In turn, complement activation causes production and release of more PAF by activating PMN(54).

In summary, we have found that activation of circulating PMN represented by increased elastase-α1-PI/PMN ratios occurs simultaneously with a low circulating PMN count in preterm infants with severe IRDS during the first five days of life. We suggest that these activated PMN have more localized effects because no other organ failure but respiratory insufficiency occurred in the IRDS infants. Furthermore, we have found indications for systemic PAF release and complement activation. Hypoxemia, tissue ischemia due to perinatal asphyxia, and lung tissue injury due to artificial ventilation may all have induced PAF release and complement activation. We did not find evidence of any infection in our IRDS infants. In preterm infants, systemic release of PAF and activated complement factors may play a role in the pathogenesis of severe IRDS by chemotaxis and sequestration of activated PMN into the lungs. However, further studies in preterm infants with mild, moderate and severe IRDS are required to determine whether systemic PAF release and complement activation in these infants is related to IRDS severity and local inflammation in the lung.