Bronchopulmonary dysplasia (BPD) is a respiratory syndrome that causes both morbidity and mortality among premature infants. Although the aetiology of BPD is multifactorial, there is increasing evidence of a link with intrauterine infection.1^–4 Several reviews suggested that antenatal infection or its consequences lead to an excessive pulmonary inflammatory response that causes lung injury.5 6 The presence of Ureaplasma species in the maternal upper genital tract or neonatal airway is significantly associated with adverse perinatal outcomes including preterm delivery and antenatal infection, as well as adverse postnatal outcomes that include BPD.7^–13 Experimental studies using immature baboons have supported the concept that intrauterine infection with Ureaplasma species contributes to lung injury after birth.14 15 There is also evidence that infection with other microbial species may be associated with the development of BPD.16^–18

Recently, Miralles et al19 reported that the presence of microbes in the gastric fluid at birth is strongly correlated with the occurrence of chorioamnionitis (CAM). Assuming that gastric fluid is in contact with both the amnionitic fluid and lung fluid, the presence of gastric fluid microbes could reflect antenatal exposure of the fetal lungs to microbes. Such exposure might influence the fetal inflammatory response in a manner that leads to the development of BPD after birth. However, it remains unclear whether or not the presence of gastric fluid microbes is actually associated with the development of BPD.

Analysis of bacterial 16S ribosomal RNA (rRNA) gene is a useful technique for detecting microbes that are difficult to isolate by standard culture methods.20 21 Recently, this method was successfully applied to identify microbes in placental or neonatal samples.19 22 23 In the present study, we employed this technique in combination with polymerase chain reaction (PCR) amplification of the urease gene of Ureaplasma species.

In addition, KL-6 was employed to evaluate lung injury and BPD. KL-6 is a mucinous glycoprotein that is expressed by alveolar type II cells and bronchiolar cells,24 and the plasma KL-6 level is specifically increased in patients with interstitial lung disease.25 26 This marker was recently shown to be specifically correlated with the severity of respiratory distress in BPD infants, as well as being useful to predict the pulmonary outcome of BPD.27 28 Because the prevalence and severity of clinically defined BPD depend on the oxygen therapy protocol of each centre, this marker is presumed to be useful in evaluating lung injury more objectively.

Finally, we investigated the following: (1) which microbes are present in the gastric fluid specimens of preterm infants at birth; (2) whether the presence of microbes is associated with development of BPD diagnosed from clinical features and by the lung injury marker KL-6; and (3) whether the presence of Ureaplasma species in gastric fluid is associated with respiratory colonisation up to 28 days of life.

METHODS

Subjects

Preterm infants with a gestational age (GA) of less than 32 weeks who were admitted to the neonatal intensive care unit (NICU) of Osaka Medical College Hospital between April 2005 and March 2007 were eligible for this study. Exclusion criteria were as follows: (1) complicated congenital heart disease; (2) multiple malformations; and (3) a documented chromosomal anomaly. Infants who died of non-respiratory cause, were transferred before reaching a postmenstrual age (PMA) of 36 weeks, or from whom gastric fluid specimens were not obtained, were also excluded from analysis. The enrolled infants were categorised into a microbe-positive group and a microbe-negative group, depending on the presence of microbes in gastric fluid.

Ventilator settings and oxygen therapy were standardised to maintain a pH ⩾7.20, the arterial oxygen saturation measured by pulse oximetry (SPO₂) between 88% and 95%, and the partial pressure of carbon dioxide (PaCO₂) between 40 and 60 mmHg. Supplemental oxygen was defined as the requirement of FiO₂ >0.21 to maintain SPO₂ ⩾88. Conventional mechanical ventilation was initiated in infants who had respiratory distress and was changed to high-frequency oscillation if it failed to maintain the above criteria. Intubated infants who needed a fraction of inspired oxygen (FiO₂) ⩾0.4 to maintain an SPO₂ ⩾90% and had typical radiographic features of respiratory distress syndrome (RDS) were diagnosed as having RDS. Chest radiographs were reviewed independently by two paediatric radiologists, blind to clinical features. BPD was defined according to the National Insitutes of Health consensus definition for infants with a GA of less than 32 weeks (administration of oxygen >21% for at least 28 days). Then the infants were classified into the following three subgroups at a PMA of 36 weeks: mild BPD (FiO₂  =  0.21), moderate BPD (0.21 <FiO₂ <0.30), and severe BPD (FiO₂ ⩾0.30 and/or positive pressure assistance).29 CAM was diagnosed histologically if polymorphonuclear leucocytes were seen in the fetal membranes, and its severity was graded according to Blanc.30 Premature rupture of membranes (PROM) was defined as the rupture of the chorioamniotic membrane before the onset of labour.31 Infants who had patent ductus arteriosus (PDA) accompanied by symptoms consistent with a major ductal shunt were treated with indomethacin, and infants who were refractory to indomethacin therapy underwent duct ligation. Those who were treated with indomethacin or underwent ligation were diagnosed as having PDA.

Preparation of samples

For the detection of microbes, we collected 1 ml of gastric fluid via a sterile nasogastric tube within 3 h after birth. For the detection of Ureaplasma species, tracheal aspirate fluid (TAF) was collected as reported previously.32 Briefly, a suction catheter was inserted slightly beyond the distal tip of the endotracheal tube, 2×0.5 ml of 0.9% saline was instilled, and TAF was collected into a sterile specimen trap after five ventilator breaths.

The TAF specimens were obtained from ventilated preterm infants on days 1, 3 and 7 and then once every week thereafter until 28 days of life. After centrifugation of gastric fluid or TAF specimens at 20 000 × g at 4°C for 10 minutes, we discarded the supernatant and stored the pellet at –80°C until analysis. DNA was directly extracted from the pellet using a QIAamp DNA Mini Kit (Qiagen K.K., Tokyo, Japan).

Detection of microbes

For the detection of bacterial 16S rRNA genes, the extracted DNA was amplified by PCR using the FD1 primer (AGA GTT TGA TCC TGG CTC AG) and the rP1 primer (ACG G(T/A/C)T ACC TTG TTA CGA CTT), as reported previously.19 21 22 After PCR amplification, 1 μl of the reaction product was re-amplified using another pair of primers: FD1 and rD2 (G(T/A)A TTA CCG CGG C(G/T)G CTG).22 The detailed PCR protocols were reported previously.19 22 After each step, PCR products were purified with a QIAquick PCR purification kit (Qiagen K.K., Tokyo, Japan). The purified 400−500 bp DNA fragments were cloned into a plasmid vector by using a TOPO TA cloning kit (Invitrogen, California, USA). Transformed competent cells were then screened and 10 clones for each specimen were selected. Plasmids were extracted with a QIAprep Spin Miniprep Kit (Qiagen K.K., Tokyo, Japan) and the inserted DNA sequences were analysed by employing a DYEnamic ET Terminator Cycle Sequencing Kit (GE Healthcare UK Ltd., Buckinghamshire, UK). The closest known sequences were determined by BLAST database searches.

For the detection of Ureaplasma species, extracted DNA was amplified by PCR with a pair of primers: U5 (CAA TCT GCT CGT GAA GTA TTA C) and U4 (ACG ACG TCC ATA AGC AAC T), as reported previously.33 The two Ureaplasma species (Ureaplasma urealyticum and Ureaplasma parvum) were then differentiated by employing another pair of primers: UMS125 (GTA TTT GCA ATC TTT ATA TGT TTT CG) and UMA226 (CAG CTG ATG TAA GTG CAG CAT TAA ATT C), as reported previously.34

Measurement of KL-6

We collected cord blood samples used for routine cord blood gas analysis at birth and also obtained blood samples by venipuncture at 1, 2 and 4 weeks of life after parental consent was obtained. The blood samples were immediately centrifuged at 3000 × g for 10 minutes at 4°C to obtain plasma, which was stored at –80°C until analysis. The plasma KL-6 level was measured with a sandwich ELISA kit (Eitest KL-6, Eisai Co. Ltd., Tokyo, Japan), as reported previously.27 28

Statistical analysis

Differences between two groups were assessed by the Mann–Whitney U test for continuous variables or by Fisher’s exact test for categorical data. Plasma KL-6 levels were compared between groups by analysis of Wilcoxon rank-sum test. Differences of continuous variables or categorical data among groups were assessed with StatView version 5.0 statistical software (SAS Institute, Inc.). In all analyses, p<0.05 was considered to be significant.

RESULTS

During the study period, 124 infants with a GA of less than 32 weeks were admitted to our unit (fig 1). Among them, nine infants did not qualify for enrolment at birth (six were stillbirth and three had congenital anomalies). In addition, 12 infants were excluded from the final analysis; 6 died of non-respiratory causes before a PMA of 36 weeks (no infant died of respiratory causes before this PMA); 2 were transferred for surgical procedures; and gastric fluid or blood specimens were not obtained from 4 infants. The remaining 103 infants (83% of all eligible infants) were enrolled in this study. The 12 infants excluded from analysis and the 103 analysed infants were similar with regard to mean GA (29.0 vs 28.9) and mean birth weight (1150 vs 1094). Placental histology was examined in 92 of the 103 infants (89%). The infants examined were similar to those in whom the occurrence of chorioamnionitis was unknown with regard to mean birth weight (1066 vs 1090), mean GA (28.9 vs 29.1), RDS (53% vs 45%) and BPD (47% vs 45%).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Summary of eligible, enrolled and analysed subjects. GA, gestational age; IVH, intraventricular haemorrhage; NEC, necrotising enterocolitis; PDA, patent ductus arteriosus; PMA, postmenstrual age.

The microbe-positive group consisted of 42 infants who had at least one microbe detected in their gastric fluid. Clinical characteristics and symptoms were compared between the microbe-positive group and the microbe-negative group (table 1). The occurrence and severity of CAM, as well as the occurrence and duration of PROM, were significantly greater in the microbe-positive group than in the microbe-negative group. Although the occurrence of BPD was similar between the two groups, severe BPD was significantly more frequent in the microbe-positive group compared with the microbe-negative group (14% vs 1.6%, p = 0.018). The percentage of infants who required supplemental oxygen at PMA of 36 weeks was slightly higher in the microbe-positive group than in the microbe-negative group, but the difference was not significant (21.4% vs 9.8%, p = 0.154).

The microbes identified from the gastric fluid specimens of 42 infants and the number of infants positive for each organism are shown in table 2. Ureaplasma species were isolated from 23 of the 42 (55%) gastric fluid specimens, including 16 (70%) Ureaplasma parvum isolates and 7 (30%) Ureaplasma urealyticum isolates. Among the 23 infants positive for Ureaplasma species, 6 (26%) were only exclusively positive for Ureaplasma species and the remaining 17 (74%) were also positive for other microbes. At least 22 species of microbes other than Ureaplasma species were isolated from 35 infants, accounting for 83% of the microbe-positive group.

The paired samples of both gastric fluid and TAF specimens were obtained from 57 intubated infants and the presence of Ureaplasma species was examined in these samples. Sixteen of the 57 (28%) infants were positive for Ureaplasma species in either gastric fluid or TAF specimens and the detected patterns during 28 days of life were presented (table 3). Thirteen of the 16 infants were positive for Ureaplasma species in TAF specimens, and all infants who had Ureaplasma species in TAF specimens were also positive in gastric fluid specimens. The other three infants who had Ureaplasma species in gastric fluid specimens but not in TAF specimens were all ventilated for less than 7 days.

The absolute KL-6 level and the change of KL-6 from the cord blood level at each time were significantly greater in the microbe-positive group than in the microbe-negative group within 4 weeks of birth (fig 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 Comparison of the median absolute plasma KL-6 level at each time (A) or the median change of plasma KL-6 from cord blood levels at 1, 2 and 4 weeks (B) after birth between the two groups. *p<0.05 vs microbe-negative group, or †p<0.005 vs microbe-negative group, respectively. Box or whisker plot shows the median with interquartile range (box) or the 10th and 90th percentiles (whiskers) of plasma KL-6, respectively.

DISCUSSION

In this study, we examined the presence of microbes in gastric fluid specimens recovered shortly after birth from preterm infants and investigated the association of such microbes with the development of BPD through analysis of 16S rRNA gene and the urease genes of Ureaplasma species. Our analysis revealed that microbes were present in 42 out of 103 (41%) gastric fluid specimens. Although Ureaplasma species were the most frequently isolated microbes and were present in 23 of the 42 (55%) specimens, only six infants were exclusively positive for Ureaplasma species and the remaining 17 were also positive for other microbes. Twenty-two species of other microbes were isolated from 35 of the 42 gastric fluid specimens, accounting for 83% of the microbe-positive group. This finding suggested that it would be difficult to assess the effect of Ureaplasma species alone and that the other microbes were also of importance.

We found that the presence of gastric fluid microbes was associated with the severity of CAM and the duration of PROM, indicating that the detection of such microbes probably reflects antenatal infection. Furthermore, the occurrence of severe BPD was increased in the microbe-positive group compared with the microbe-negative group (table 1). Our results support the recent proposal that antenatal exposure to microbes might induce lung injury and thus contribute to the development of BPD.5 6 Consistent with the clinical features, the presence of gastric fluid microbes at birth was related to the increased KL-6 level at each time (fig 2A). In addition, the increase of KL-6 in the microbe-positive group was further enhanced after birth, compared to the microbe-negative group (fig 2B). Because the increased KL-6 level at 1 or 2 weeks correlates well with the severity of pulmonary injury and predicts severe BPD,27 28 these results indicate that the presence of gastric fluid microbes at birth is associated with subsequent continuing lung injury during the first 4 weeks of life that may lead to BPD.

It remains unclear exactly how gastric fluid microbes are related to lung injury that leads to BPD. Because respiratory colonisation by microbes including Ureaplasma species is a risk factor for the development of BPD,7–13 16–18 it is possible that the microbes detected in gastric fluid may also colonise the airways. Therefore, we investigated the relationship between the presence of Ureaplasma species in gastric fluid and subsequent respiratory colonisation in intubated infants (table 3). Ureaplasma species were isolated from gastric fluid specimens of 16 infants, 13 (81%) of whom also had Ureaplasma species in their TAF specimens. All 13 infants who were ventilated for more than 1 week showed respiratory colonisation with Ureaplasma species, indicating that gastric fluid microbes could be associated with subsequent respiratory colonisation in preterm infants ventilated for more than 1 week.

Contrary to a previous study,2 the presence of gastric fluid microbes reflecting antenatal infection was not associated with a reduced incidence of RDS in this study. However, we could not claim that our result was valid because this study was not designed to diagnose RDS accurately. In addition, it has limitations for assessing the bacterial load and the relevance of the detected microbes in our study, because we did not perform a quantitative PCR approach. Although Ureaplasma species may have a great influence on the pathogenesis of BPD, it is difficult to assess its single effect and needs further investigations about the effect of other microbes that share a considerable part of the microbe-positive group. If the presence of the gastric fluid microbes reflects antenatal exposure of the fetal lungs to infection, such microbes or other signals generated by these organisms could prime vulnerable fetal lungs to trigger an excessive inflammatory response that leads to the development of BPD. Other risk factors for BPD, such as mechanical ventilation, oxygen therapy, and postnatal infection, may act synergistically to amplify the progression of BPD after it occurs. Although we cannot draw any definitive conclusions because this was a preliminary study, our results should be useful for investigating the relationship between antenatal colonisation/infection and the role that chorioamnionitis plays in the development of BPD.