Main

Since the initial description of bronchopulmonary dysplasia (BPD) in premature newborns by Northway et al. (1) in 1967, many investigators have been searching for a causative etiology. Despite stricter control of delivered oxygen, improved ventilatory techniques, and the use of exogenous surfactant, BPD continues to occur in 20%–50% of the sickest, most immature infants (2).

There is increasing evidence supporting the role of congenital infection in the development of BPD. Congenital infections may be especially important in very small infants who develop chronic lung disease after receiving ventilatory support because of poor respiratory effort rather than because of severe underlying disease. Thus, prenatally acquired infections of the lung and/or trachea may play a role in the development of BPD (3).

During the last decade there has been increased interest in possible pathogenic role of organisms that colonize the airways of preterm infants and cause disease processes that could be devastating in this age group. In particular, Chlamydia trachomatis (4–7), Mycoplasma hominis, and Ureaplasma urealyticum have been studied in the context of the development of BPD (6–10). While serological evidence has associated C. trachomatis with BPD (4), cultures for C. trachomatis have failed to confirm these data (6, 7). Several studies have also suggested an association between infection with U. urealyticum and the development of BPD (7–9). A meta-analysis of 17 studies suggested that colonization with U. urealyticum was associated with BPD with an overall relative risk of 1.72 (9). However, other studies have failed to demonstrate a link between this agent and BPD (6, 10), and it has been suggested that the prior studies demonstrating a link between the agent and BPD failed to take into account the gestational age of the infants (10).

Viral agents have also been reported to play a role in the development of BPD and other diseases in preterm infants. Colonization of the lungs with cytomegalovirus (CMV) has been reported to be associated with an increased risk of developing BPD (11, 12). However, CMV infections in infants that developed BPD were believed to have occurred postnatally. Notably, we recently reported that CMV is commonly detected in the amniotic fluid of abnormal pregnancies, particularly in association with fetal hydrops (13). Adenoviruses and enteroviruses have also been detected in amniotic fluid associated with abnormal pregnancies (13, 14), as well as cardiac samples of stillborn or newborn infants with myocarditis (15) or severe respiratory failure (16, 17). Intrauterine infection with parvovirus B19 has been linked to serious fetal disease (18), as well. However, in the case of adenovirus, enterovirus, and parvovirus, the association of congenital infection with the development of BPD has not been investigated.

Polymerase chain reaction (PCR) is a versatile, sensitive, and rapid technique that has been used by many investigators to detect viral and bacterial nucleic acid in clinical specimens (6, 13, 15, 19–34). For infectious agents that are difficult to culture, such as U. urealyticum, studies indicate that PCR is at least as sensitive as culture for its detection (6, 30, 31). Similar results have been reported using PCR for detection of M. hominis (31, 32) and C. trachomatis (33, 34) in tracheal aspirate samples.

We hypothesized that PCR detection of microorganisms in the tracheal aspirates of preterm infants would be associated with the development of BPD. We prospectively studied infants less than 30 wk of gestation admitted to the Neonatal Intensive Care Unit over an 11-mo period. Tracheal aspirate samples were obtained within the 1st week of life and screened by PCR for adenovirus, enterovirus, CMV, parvovirus, U. urealyticum, M. hominis, M. pneumoniae, and Chlamydia species (trachomatis, psittaci, and pneumoniae).

METHODS

Study design.

Infants delivered at less than 30 wk of gestation and admitted to the Neonatal Intensive Care Unit of Texas Children's Hospital, Houston, Texas, were studied. Tracheal aspirate samples were collected as approved by the Institutional Review Board of Baylor College of Medicine, nucleic acid isolated from the samples and analyzed for the presence of the genomic sequences of a range of microorganisms.

Patients and controls.

All infants less than 30 weeks of gestation were electively intubated at birth. Because of the variations in the literature for the definition of BPD (35, 36), we compared the frequency of microorganism detection between BPD patients and controls using two definitions of BPD:1) oxygen dependency at 28 d of life, or 2) oxygen dependency at 36 wk postconceptional age (PCA). The gestational age of each infant was estimated by ultrasound scan, early in pregnancy when possible, or by the date of the last menstrual period. Infants were defined as control patients when they did not develop BPD, according to either definition. Infants that expired before the time point necessary to meet diagnostic criteria for BPD were excluded from statistical analysis.

Specimens.

Tracheal aspirate samples were collected within 7 days of birth from all infants that were intubated at birth for surfactant administration. Most samples were collected within the first 4 d of life. Samples were collected through the endotracheal tube by gentle suction in a closed sterile system (Sherwood Medical, St. Louis, MO), after installation of 0.5 mL of sterile normal saline. This was repeated two times. Mean volume of aspirate collected was 0.6 mL (range 0.2 to 1.0 mL). The specimens were placed on ice and processed immediately or stored at −80°C.

Nucleic acid isolation.

All chemical reagents were purchased from GIBCO-BRL (Gaithersburg, MD) unless otherwise indicated. Total RNA and DNA were isolated simultaneously from specimens using a modification of the RNAzol method, as previously described (13, 15). In brief, aliquots of the tracheal aspirate samples (0.25 mL) were first homogenized in RNAzol (0.75 mL) using disposable RNase-free pestles (purchased from PGC Scientific, Gaithersburg, MD). The nucleic acids isolated were precipitated and then resuspended in 30 μL of diethyl pyrocarbonate (DEPC) (Sigma Chemical Co., St. Louis, MO)-treated water.

Positive and Negative controls.

Nucleic acid was isolated from bacterial cultures (U. urealyticum) or tissue culture cells (American Type Culture Collection (ATCC), Manassas, VA) infected with adenovirus types 2 and 5, Coxsackievirus B3 (Nancy), CMV, M. hominis, M. pneumoniae, or C. pneumoniae, using the modified RNAzol method (13, 14, 15). These nucleic acid samples were used as positive controls in the virus- and bacteria-specific PCR reactions.

Nucleic acids isolated from noninfected cultured human cells (HeLa: ATCC) served as positive controls for the β-actin-specific PCR reactions and negative controls for the virus- and bacteria-specific PCR reactions. Additional negative controls consisted of water instead of nucleic acid in the reaction mix.

Reverse transcription and polymerase chain reaction.

Primer pairs for the detection of bacterial DNA were based upon sequences (37–41), whereas primers for the detection of viral nucleic acid or the β-actin gene were based upon sequences in GenBank (National Institutes of Health). All primers were synthesized and purified by GIBCO-BRL (Gaithersburg, MD), and their sequences are shown in Table 1.

Table 1 Sequences of PCR primers used for the detection of viral, bacterial and β-actin sequences (Y = C + T, K = T + G, M = A + C, B = T + C + G, H = A + T + C, R = A + G)
Full size table

For the detection of enteroviral genomic nucleic acid, reverse transcription-PCR (RT-PCR) was used. For the synthesis of cDNA, 3 μL of extracted total nucleic acid was mixed with 2 μL of random primers (3 mg/mL) and 6.2 μL of DEPC-treated water, in the presence of 20 units (0.5 μL) of Prime RNase Inhibitor (5 Prime-3 Prime, Inc, Boulder, CO). To anneal the primers to the RNA, the samples were heated to 95°C for 5 min, then cooled on ice. After the annealing step, 4 μL of 5× reverse transcriptase buffer (GIBCO-BRL), 2 μL of 100 mM DTT, 0.8 μL of 25 mM dideoxynucleotide triphosphates (dNTPs), another 0.5 μL of Prime RNase Inhibitor, and 200 U (1 μL) of Moloney-murine leukemia virus reverse transcriptase were added. The samples were incubated at 37°C for 1 h, followed by 5 min at 95°C to inactivate the reverse transcriptase. Two microliters of the cDNA were subjected to PCR amplification to detect β-actin cDNA or enteroviral cDNA.

To verify the isolation of cellular RNA or genomic DNA from the tracheal aspirate samples, 2 μL of cDNA or total nucleic acid, respectively, were PCR amplified using the β-actin primers (listed in Table 1). The DNA or cDNA template was amplified in a 20 μL reaction containing 1× PCR buffer, 2.5 mM magnesium chloride, 0.25 mM dNTPs, 0.5 μM β-actin primers, and 2.5 U Taq DNA polymerase. After an initial 5 min of incubation at 94°C, 35 rounds of amplification were performed using a Stratagene Robocycler (La Jolla, CA), under the following conditions: 94°C for 45 s, 64°C for 45 s, and 72°C for 45 s. This was followed by a 72°C incubation for 5 min.

For the detection of viral genomic sequences, 2 μL of either cDNA (for the enteroviruses) or nucleic acid isolated from the tracheal aspirate samples (for the DNA viruses) were subjected to nested PCR, using the primers shown in Table 1. The primary reaction was performed under the same conditions as described for the amplification of β-actin using 35 rounds of amplification. For the secondary amplification, 2 μL of the primary reaction product was diluted in 98 μL of TE (10 mM Tris, pH 7.5, 1 mM EDTA) and then 2 μL of the diluted product was subjected to 30 cycles of PCR amplification, as described for the primary amplification (above).

For the detection of the bacterial DNA using the universal bacterial primers (UBP), 2 μL of isolated nucleic acid was subjected to a single PCR using the conditions described for the amplification of β-actin, except that the annealing temperature was 60°C rather than 64°C. The detection of U. urealyticum, M. hominis, M. pneumoniae, and Chlamydia species was performed by a single PCR using the same conditions as described for the detection of the β-actin gene.

Analysis of PCR products.

The products of each reaction were analyzed by agarose gel electrophoresis using 1.75% agarose containing 0.5 μg/mL of ethidium bromide (Sigma Chemical Co., St. Louis, MO) and the DNA product was visualized by UV transillumination. In all cases, positive and multiple negative control reactions were performed simultaneously with the test samples. All samples were analyzed without prior knowledge of the clinical data for each patient and all PCR-positive samples were repeated to ensure that they were positive. Any sample not giving a signal with the β-actin primers was excluded.

DNA sequencing.

The adenovirus, enterovirus and UBP-specific amplimers were re-amplified and the PCR products were purified using a PCR purification kit (Qiagen, Inc., Valencia, CA), according to the manufacturer's instructions. The DNA sequence was determined by automated cycle sequencing, according to the kit manufacturer's instructions (Perkin Elmer Corp, Foster City, CA).

Statistics.

The frequency of detection of viral and bacterial species was compared between BPD and control patients using χ2 and Fisher exact test for the statistical analysis (42). Comparison of means was performed by the Student t test (42). A p value of ≤0.05 was considered statistically significant. To control for variables such as gestational age, logistic regression analyses were used for the analysis of data that were statistically significant by the above methods.

RESULTS

Patients.

During the 11-mo study period (1997–1998), 89 preterm newborn infants delivered before 30 wk of gestation were enrolled. At 28 d of life, 13 infants had expired, 45 had a supplemental oxygen requirement (BPD28d), and 31 had no supplemental oxygen requirement. At 36 wk of PCA, 15 infants expired, 39 remained oxygen dependent (BPD36wk), and 35 did not need any supplemental oxygen.

There were significant differences between the BPD and no BPD groups for gestational age, birth weight, days requiring supplemental oxygen, and duration of hospital stay. As part of the routine monitoring of premature infants, blood cultures were performed when clinically indicated. In the BPD28d group, there was a statistically significant increase in the frequency of culture-positive bacterial agents by comparison with the no BPD28d group (p ≤ 0.01): this was not statistically significant by 36 wk of PCA, however (p = 0.11). There were no differences between any of the groups with respect to gender, the incidence of prolonged rupture of membranes, the administration of prenatal steroids, or the frequency of the development of necrotizing enterocolitis.

Detection of viral genomes.

Of the 45 infants with BPD28d, 12 were PCR-positive for adenovirus (27%) while out of the 31 patients without BPD, only 1 patient was adenovirus positive (3%) (Fig. 1 and Table 2). In all cases, the DNA sequence of the amplimer was identical to the published sequence of the hexon gene of adenovirus type 5 (data not shown). Using the BPD36wk definition, 39 infants had BPD and, of these, 10 were positive for adenovirus (29%) compared with 2 (6%) of the 35 infants with no supplemental oxygen requirement (Table 2). One of the adenovirus-positive infants (with BPD) died before the 36-wk time point. Using either definition of BPD, there is a statistically significant increase in the prevalence of adenoviral infection in BPD patients when compared with control patients (Fisher exact test:p ≤ 0.01 for BPD28d and p = 0.013 for BPD36wk). Logistic regression analysis using BPD as the dependent variable and adenovirus status as the independent variable, while controlling for other covariables (gestational age, birth weight, U. urealyticum status, patent ductus arteriosus, bacterial infection proven by blood cultures, prolonged rupture of membranes, and prenatal steroid administration to the mother), produced p values of 0.02 and 0.04 for BPD28d and BPD36wk, respectively.

Figure 1
figure 1

The detection of adenoviral genomic DNA in tracheal aspirate samples. Adenovirus DNA was detected by nested PCR analysis of nucleic acid isolated from tracheal aspirate samples of preterm infants. The products of the secondary reaction, using the inner primers, were detected by ethidium bromide staining of the 1.75% agarose gel. A 100-bp size ladder (M) is shown in the first lane. The adenovirus positive control PCR (+) is seen as a 330-bp amplimer in the last lane (from left to right). In the preceding lanes are patient samples (indicated by the sample number) and negative controls (−).

Full size image
Table 2 The frequency of detection of microorganisms in tracheal aspirates of infants with or without BPD (diagnosed at 28 d postgestation or 36 wk postconception)
Full size table

One patient who had a supplemental oxygen requirement both at 28 d of life and 36 wk of PCA age was PCR-positive for CMV DNA (Fig. 2 and Table 2). Among the infants not requiring supplemental oxygen, 1 patient was positive for CMV and one for enterovirus (Table 2). Sequencing of the enterovirus amplimer indicated a Coxsackievirus B3 isolate (data not shown); the sequence shared 93% homology with the closest published sequence, determined by BLAST search of GenBank databases.

Figure 2
figure 2

The detection of CMV genomic DNA in tracheal aspirate samples. CMV DNA was detected by nested PCR analysis of nucleic acid isolated from tracheal aspirate samples of preterm infants. See Figure 1 for details. The CMV positive control is detected as a 391-bp amplimer in the last lane (from left to right).

Full size image

Detection of bacterial DNA.

Initially bacterial DNA was detected by PCR using the UBP. Thirty-eight of the first 42 patients (90%) screened yielded a band approximately 520bp in size: in some samples multiple bands were detected. DNA sequencing of the first six UBP amplimers produced sequencing outputs of the type shown in Figure 3. We interpreted these data to indicate the presence of multiple bacterial types in each sample because there are also regions of considerably lower homology between bacterial 16S ribosomal RNA genes, which could account for both the multiple bands detected by PCR and the sequence data. Therefore, it was apparent that this approach was unlikely to generate effectively and unequivocally data linking a particular bacterial strain with BPD. It was decided that future samples would be screened for the specific bacterial types previously implicated in BPD (Table 2). Only one sample was positive for M. hominis (an infant not requiring supplemental oxygen) while no cases were positive for M. pneumoniae, or any Chlamydia species. However, 10 (22%) of the 45 BPD28d patients and 8 (25%) of the 31 infants without BPD were PCR-positive for U. urealyticum (Fig. 4 and Table 2). Similarly, 10 (29%) of the 39 BPD36wk patients and 7 (20%) of the 35 infants without BPD were positive for U. urealyticum (Table 2). The p values for U. urealyticum were 0.72 and 0.56 for BPD28d and BPD36wk, respectively.

Figure 3
figure 3

DNA sequence analysis of the amplification products of the universal bacterial primers. The amplification products using the UBP primers were analyzed by automated DNA sequencing. This is the output from one of the amplimers but is typical of the sequence data generated. Note the presence of regions of readable sequence (nucleotides 0–100 and 270–370) that probably correspond to regions of homology between the bacterial strains detected in this sample, interspersed by illegible sequence.

Full size image
Figure 4
figure 4

The detection of U. urealyticum DNA in tracheal aspirate samples. U. urealyticum DNA was detected by single PCR analysis of nucleic acid isolated from tracheal aspirate samples of preterm infants. See Figure 1 for details. The positive control is detected as a 426-bp amplimer in the last lane (from left to right). Note that samples 52, 56, 60, and 62 are positive.

Full size image

DISCUSSION

The aim of this study was to examine the possible association of the presence of nucleic acid sequences of viral or bacterial agents in the upper airways of preterm infants during the 1st week of life with the development of BPD. Previously such an association has been reported for U. urealyticum (7–9); however, this has not been shown by all studies (6, 10), and it has been speculated that when corrections for gestational age are made that no statistically significant association is apparent in any of the studies (10). Additionally, C. trachomatis, M. hominis, and CMV have been implicated as etiologic agents for the development of BPD (4–8, 11, 12), which was not supported by our findings.

The high rate of bacterial detection by PCR using the universal bacterial primers (90%) could reflect colonization of the upper airways of newborn infants with common flora within the 1st few days of life. However, it is possible that the detection of bacterial DNA by PCR represents contamination of the sample during sample recovery, nucleic acid isolation, or analysis. Subsequent PCR analyses were performed using bacterial-specific primers to detect the organisms previously reported to be associated with BPD. Contamination was not apparent using these primers.

The detection of U. urealyticum in many infants suggests that infection or colonization with this agent is relatively common in preterm infants but is not associated with the development of BPD. However, the frequency of detection of U. urealyticum (24%) in our study was somewhat lower than previously reported (45%; seeRef. 6). This difference most likely reflects a lower infection rate among our study group, but could indicate lower sensitivity.

With the exception of adenovirus, other microorganisms were rarely detected in tracheal aspirate samples and were not associated with the development of BPD (Table 2). The lack of detection or very low prevalence of the bacterial agents is similar to other reports (6, 7). A previous report linking CMV with the development of BPD investigated 32 CMV-positive infants; of these 32 patients, 24 (75%) developed BPD (12). Because we detected CMV in just 2 infants, we cannot conclude that CMV infection of the lungs does not lead to the development of BPD, only that CMV is not commonly associated with the development of BPD in preterm infants. This interpretation is also relevant for C. trachomatis, M. hominis, M. pneumoniae, enterovirus, and parvovirus.

Tracheal aspirate samples from 13 infants were positive for adenovirus DNA. Of these 13 infants, 12 had developed BPD at 28 d of life and 10 continued to be oxygen-dependent at 36 wk of PCA (one of the adenovirus infected infants died before 36 wk of PCA). Compared with the patients without BPD, a statistically significant increase in the prevalence of adenovirus infection in BPD patients was observed and suggests that congenital adenoviral infection is associated with the development of BPD. The statistical significance was maintained even after correcting for other variables, including gestational age and nosocomial infection. Of note was that there was a significant difference (p ≤ 0.01) between the mean gestational age of BPD patients that were adenovirus positive (27.0 ± 1.1 wk) and adenovirus negative (25.7 ± 1.5 wk). These data suggest that the major factor in the development of BPD is immaturity unless the patient is exposed to adenovirus, in which case BPD develops in the more mature patient.

Sequence analysis of the PCR amplification products identified sequences most consistent with adenovirus type 5 in all cases, based upon comparison with published sequences. However, it should be noted that many adenovirus serotypes have not been sequenced so that definitive subtyping is difficult. In most cases of pediatric and neonatal myocarditis that we have studied, the major serotype of adenovirus detected has been type 2 (a group C adenovirus, like type 5) with the remainder being type 5 (15, 19, 43). It is interesting to note that in young children with upper respiratory tract illness the group C, adenoviruses are most commonly detected (44). In contrast, in adults other adenovirus groups are associated with respiratory disease, and many adults remain seropositive for the group C adenoviruses. These data could suggest that most mothers would be resistant to infection with the group C viruses and that the detection of these viruses represent direct infection of the neonate. In this regard there has been a case report of neonatal adenoviral infection resulting from ascending intrauterine infection in a preterm infant (45) and another report of fatal adenovirus pneumonia where the virus was believed to be contracted from the birth canal during delivery (46). However, it has also been reported that many healthy adults have lymphocytes persistently infected with group C adenoviruses (47), which could be a source of virus to infect the susceptible fetus. Furthermore, there are cases of adult respiratory tract infection with group C viruses. Trans-placental transmission of adenovirus has been reported to result in neonatal infection (48), as well as used as a tool for the delivery of recombinant adenoviral vectors (49). That the group C adenoviruses require the expression of a unique receptor (the common Coxsackievirus B-adenovirus group C receptor (CAR, Refs. (50) and (51) for cell entry may be crucial for determining the link between this group of viruses and the development of specific diseases.

One of the major concerns with the use of PCR is that, because of the sensitive nature of the technique, there exists the possibility of contamination between samples and from the positive controls. It should be noted that in each case in which a positive result was obtained, the analysis was repeated on at least one occasion to confirm the result. In addition, positive results were repeated in the presence of samples known to be negative and numerous negative control reactions (Fig. 1). The low rate of positivity for the detection of Mycoplasma and Chlamydia species, as well as CMV, enterovirus and parvovirus, suggests that routine contamination is not a problem. In addition, all viral determinations were performed before diagnosis. Most of the bacterial screening was similarly blinded with the exception of the first third of the samples (due to the change in methodology during the period). No difference in the frequency of U. urealyticum detection was apparent between the blinded and nonblinded samples.

Adenovirus gene expression or viral replication could lead to the development of early lung inflammation. In a cotton rat model of adenovirus-induced pneumonia, it is thought that the lung pathology is due to host immune response rather than direct damage to the cells resulting from virus replication (52). Several years ago it was reported that a latent adenovirus infection of the lung results in chronic obstructive pulmonary disease in humans (53). In those cases it was suggested that the adenoviral genome integrated in a linear fashion with subsequent rearrangement and amplification of the early regions, particularly E1A (53, 54). Persistent adenoviral infection of the lungs has also been associated with bronchiolar obstruction in children with asthma who have poor responses to steroid therapy (55). We have demonstrated that adenovirus infection of the lung is common in lung transplant recipients that develop obliterative bronchiolitis (OB), a condition that is believed to result from a host immune response and which leads to rejection of the transplanted organ in many cases (56). The E1A region has been implicated in the sensitization of infected cells to destruction by cytokines (57), as well as the induction of apoptosis (58), which could be important components in the development of inflammatory responses against infected cells.

In summary, the present study demonstrated that colonization of the trachea of preterm infants with U. urealyticum does not correlate with the development of BPD, nor does cytomegalovirus infection. This is the first study to demonstrate a significant association between adenovirus infection and the development of BPD. These data, together with the demonstration of adenovirus infection of the fetus in abnormal pregnancies (13), suggest that congenital adenovirus infection is important in the development of fetal and neonatal diseases and that studies to determine the role of adenovirus in the development of BPD, particularly in neonates born after 26 wk, are warranted.