© 2002 American Thoracic Society
Association of Rhinovirus Infection with Increased Disease Severity in Acute BronchiolitisResearch Laboratories, Second Department of Pediatrics, University of Athens School of Medicine, Athens, Greece Correspondence and requests for reprints should be addressed to Nikolaos G. Papadopoulos, M.D., Ph.D., Research Laboratories, Second Department of Pediatrics, University of Athens, 13 Levadias Street, 11527 Goudi, Athens, Greece. E-mail: ngp{at}clubmember.org
Respiratory syncytial virus (RSV) is the major pathogen responsible for acute bronchiolitis in infancy. However, evaluation of the relative importance of rhinovirus or multiple viral infections has been hampered by the lack of sensitive diagnostic methodologies. Therefore, in this study we used the reverse transcription-polymerase chain reaction for 11 respiratory pathogens to assess the etiology in infants with acute bronchiolitis and correlate it with clinical characteristics of the disease. Viruses were detected in 73.7% of patients. RSV was identified in 72.4% of virologically confirmed cases, rhinovirus in 29%, whereas multiple infections represented 19.5% of cases, most of which (69%) were combinations of rhinovirus with RSV. In a logistic regression model controlling for age, sex, birth weight, presence of fever, and day of disease on admission, the presence of rhinovirus was found to increase by approximately five-fold, the risk for severe disease. Multiple pathogens had a similar trend in the univariate analysis, which was eliminated in the multivariate model. Multiple virus cases were admitted to the hospital later in the course of their disease than unique pathogen cases, suggesting successive infections. In conclusion, rhinovirus is second only to RSV as a causative agent of bronchiolitis and is associated with more severe disease. The presence of more than one pathogen may influence the natural history of acute bronchiolitis.
Key Words: bronchiolitis • polymerase chain reaction • rhinovirus
Bronchiolitis is an acute, inflammatory respiratory illness of children in the first two years of life and a major cause of hospitalization in that age group. Bronchiolitis occurs in a typical seasonal pattern, with yearly epidemic peaks during winter, closely following the activity of respiratory syncytial virus (RSV). Clearly, RSV is the agent most commonly identified from infants with the disease, with detection rates ranging from 40 to 100% in various studies (1, 2). Partly because of the great preponderance of RSV, the role of other respiratory viruses in acute bronchiolitis is less well studied. Although some studies have supported a milder clinical course of non-RSV bronchiolitis (3), this has not been generally confirmed (1). Furthermore, in many instances, virological testing in bronchiolitis has been performed only for RSV, overlooking the possibility of infection with multiple agents. This is especially relevant in the case of human rhinoviruses (RV), the most prevalent agents of upper airway infections in older children and adults. Until more recently, the methodologies for RV detection were considerably less sensitive than those for RSV or other respiratory viruses (4). In addition, the ability of RV to reach and replicate in the lower airways of immunocompetent individuals had been disputed. However, we have shown that RV is able to infect the bronchial epithelium of immunocompetent adults, after induction of an experimental upper airway infection (5). It is therefore likely that this could occur regularly in infants as well. In addition to its epidemiological importance, the understanding of possible differential effects of RSV versus RV or other respiratory viruses in bronchiolitis has some significant implications. It is well documented that children with severe bronchiolitis have an increased possibility of developing recurrent wheezing later in life (6). It is conceivable that viruses other than RSV, or multiple virus infections, may have an effect on the severity of bronchiolitis, which may in turn influence the development of reactive airway disease. Most importantly, however, the imminent availability of new therapeutic modalities against both RSV and RV could soon pose therapeutic dilemmas in the clinical setting. We hypothesized that the incidence of RV infections, and therefore their importance, has been underestimated in acute bronchiolitis. In this study, therefore, we used polymerase chain reaction (PCR)-based methodologies, which offer increased sensitivity for most respiratory viruses, to evaluate the frequency of detection of these agents in acute bronchiolitis and their relationship to the clinical characteristics of the disease.
Patients The study was performed in the Second Pediatric Clinic, University of Athens (Athens, Greece), at the P&A Kyriakou Children's Hospital, and was approved by the hospital's Ethics Committee. Recruitment took place between October 1, 1999 and September 30, 2000. On admission patients were evaluated for enrollment by one of the study physicians (M.M., M.T., E.A., or A.P.), using a standardized questionnaire and clinical evaluation. Initial history included age; sex; birth history (duration of pregnancy and birth weight); family history of atopy, defined as a history of medical diagnosis of asthma, allergic rhinitis, atopic dermatitis, or anaphylaxis in first-degree relatives; exposure to tobacco smoke, considered positive if either parent smoked any number of cigarettes in the house; and history of previous wheezing episodes or other chronic illnesses, according to medical diagnosis. Inclusion criteria were clinical diagnosis of bronchiolitis and age less than 18 months. Bronchiolitis was defined as an acute infection of the lower airway characterized by increased respiratory effort (tachypnea [more than 50 respirations/minute] and/or use of accessory respiratory muscles) and expiratory wheezing and/or crackles. Exclusion criteria were underlying chronic diseases (e.g., cystic fibrosis, bronchopulmonary dysplasia, congenital heart disease, immunodeficiency) and recurrent (more than two) wheezing episodes. One hundred and nineteen infants (73 male) were included in the study. The mean age was 3.9 ± 2.8 months (0.5–12.5 months). There were 111 full-term (birth weight, 3,183 ± 41 g) and 8 premature (birth weight, 2,040 ± 208 g) infants, 23.5% had a positive family history of atopy, 66.4% were regularly exposed to tobacco smoke, and 26.1% of the mothers smoked during pregnancy. For 10 patients (8.4%) this was their second wheezing episode. Clinical severity score (Table 1), based on heart rate, respiratory rate, wheezing, evidence of cyanosis or difficulty of feeding, and oxygen saturation was assessed on admission (7). A chest X-ray, venous blood sample, and nasopharyngeal wash fluid (2 ml normal saline in each nostril) were also obtained at this time (8). The nasopharyngeal wash fluid was diluted in 5 ml of virus transport medium, aliquoted, and frozen at –70° C until assayed for the presence of respiratory viruses. Infants were managed and discharged according to routine clinical practice.*
Virus Detection The presence of viral RNA in nasopharyngeal wash samples was examined by reverse transcription-PCR (RT-PCR). To extract RNA, 10 µl of sample was diluted 1:10 in ultra-high-quality water followed by the addition of an equal volume of TRIzol reagent (GIBCO, Grand Island, NY) and chloroform (1:5, vol/vol). RNA was precipitated with isopropyl alcohol, washed with 80% ethanol, vacuum dried, and resuspended in ultra-high-quality water. Reverse transcription was performed in buffer (50 mM Tris-HCl, 75 mM KCl, and 3 mM MgCl2) with 10 mM dithiothreitol, 0.4 mM deoxytrinucleoside triphosphates, 0.5 µg of random hexamer primers, and 200 U of reverse transcriptase (SuperScript; GIBCO). The mix was incubated at 37° C for 60 minutes to yield single-stranded cDNA and was thereafter kept at –20° C until assayed further. PCRs were performed with primer sets and conditions specific for RSV (8), RV (4), coronaviruses OC43 and 229E (9), influenzaviruses A and B (10), parainfluenza viruses 1, 2, and 3 (11), adenoviruses (12), and Chlamydia pneumoniae (13), according to published protocols. Primers were obtained from MWC-Biotech (Munich, Germany). Amplicons were visualized by ethidium bromide staining after electrophoresis on a 1.5% agarose gel (GIBCO). Reference viral strains were always included as positive controls; several negative controls (virus transport medium) were also included in each run.
Statistical Analysis Logistic regression analysis was used to evaluate the risk factors for the severity of the disease as this was indicated by the clinical severity score recorded on admission, ranging from 5 to 12 (mean, 8.5; SD, 1.5; median, 9). For the purpose of this analysis, the study population was classified in two groups: low-risk, including all infants with a clinical severity score on admission at or below the 50th centile, and high-risk, including infants with a score higher than the 50th centile. We initially performed univariate models introducing as independent variables age, sex, birth weight, personal and family history of atopy, exposure to tobacco smoke, presence of fever, day of disease on admission, and type of detected virus. Day of disease at admission was defined as the number of days from the initial presentation of respiratory symptoms, as reported by the parents. Subsequently, these variables were introduced in a multivariate model, to eliminate the possibility of mutual confounding.
Detection of Respiratory Viruses by PCR One patient provided insufficient material for evaluation of all examined viruses and was therefore excluded from the analysis. Overall, PCRs revealed the presence of a virus in 87 of 118 patients (73.7%). Figure 1 shows the distribution of cases by time and virus. As expected, the most common causative agent was RSV, which was found in 63 of 87 positive cases (72.4%). Nevertheless, RV was also identified in a considerable number of cases (25 of 87, 29%). Other respiratory viruses were occasionally found (adenovirus in nine cases, influenzavirus in three, parainfluenza virus in three, and coronavirus in three). More than one type of virus coexisted in 16 of 87 patients (19.5%). In one infant, a triple infection consisting of RV, adenovirus, and coronavirus was detected. The majority of multiple infections (10 of 16 cases, 69%) consisted of RSV detected simultaneously with RV. The remaining double-positive samples included two with RSV and adenovirus (12.5%) and one each with RV and coronavirus, adenovirus and coronavirus, and RSV and parainfluenza virus. RSV was a unique agent in 50 cases, and RV was a unique agent in 12 cases.
Clinical Characteristics of Multiple-Virus Infections Table 2 depicts the patients' demographic and clinical characteristics by number of identified viruses. When groups defined by virus identification (no virus, one or more viruses detected) were compared, infants in which the causative agent could not be detected were found to have significantly higher birth weight (p = 0.01). The severity score was lower in the former group (8.3 ± 1.1 versus 8.8 ± 1.7), but this was not significant (p = 0.1). There were no significant differences with respect to age, sex, exposure to tobacco smoke, family history of atopy, presence of fever, or day of disease at admission.
To address the question of whether infants with multiple viruses have a different natural history than those infected with a single virus, these groups were compared. No differences were found in relation to age, sex, birth weight, family history of atopy, or presence of fever (Table 2). However, infants with multiple viruses were admitted later in the course of their disease (4.0 ± 2.4 versus 2.8 ± 1.6 days, p = 0.025). There was also a trend toward more cases of exposure to tobacco smoke at home in that group, which did not reach significance (75 versus 47%, p = 0.058).
Clinical Characteristics of RSV and RV Infections
Logistic Regression of Clinical Severity by Type of Virus Logistic regression analysis revealed that the presence of RV indicated an excess risk for having a clinical severity score on admission above the 50th centile (odds ratio [OR], 5.4; 95% confidence interval [CI], 1.7–17.2; p = 0.004). The presence of RV still corresponded to a substantial and significant increase in the odds for high-risk group placement when it was tested in the multivariate model after controlling for age, sex, birth weight, presence of fever, and day of disease on admission (OR, 4.9; 95% CI, 1.2–18.7; p = 0.022). The presence of more than one virus also increased the possibility of belonging to the high-risk group (OR, 3.7; 95% CI, 0.95–14.5) but this increase did not reach significance (p = 0.059) and it was eliminated when multiple infections were tested in a multivariate model including the type of identified virus (RSV or RV). This elimination is probably attributed to the effect of colinearity as almost half (10 of 22) of the patients with RV had more than one virus detected. When entered in the multivariate model, RSV also tended to increase the severity of the disease, without, however, reaching significance (OR, 3.2; 95% CI, 0.98–11.5; p = 0.053).
Acute bronchiolitis is almost synonymous with RSV disease in infancy. Nevertheless, only a small proportion of infants contracting RSV develop serious lower respiratory disease, whereas there are cases of clinical disease identical to bronchiolitis in which other respiratory viruses are the only identified agents (14). PCR-based methodology for the detection of respiratory viruses offers improved sensitivity for some viruses, especially RV. We therefore investigated the role of such agents in acute bronchiolitis. In addition to its epidemiological value, this approach could offer evidence regarding the mechanisms of the disease as well the induction of postviral wheezing later on in life. This is the first study to observe a significant correlation between RV infection and increased disease severity in acute bronchiolitis, in a logistic regression model controlling for age, sex, birth weight, the day of disease on admission, and the presence of fever. A trend toward a similar correlation noted for multiple respiratory virus infection-related disease was eliminated when the type of virus was introduced in the multivariate model. However, the overwhelming majority of multiple virus infections were a combination of RSV and RV; hence it is difficult to distinguish between the effect of RV per se and that of multiple infection. In a study of children with bronchopulmonary dysplasia, RV and RSV were the causes of significant and comparable lower respiratory complications (15). Furthermore, the severity of lower respiratory symptoms caused by RSV and RV in 9- to 11-year–old children in the community did not differ (16). RV is able to produce a wide range of clinical pictures, ranging from the common cold to life-threatening pneumonia. The pattern and severity of the clinical presentation depend on the host, as shown in immunocompromised patients, in whom life-threatening disease may develop (17). In older children and adults, RSV and RV usually produce common cold symptoms, without any differences in disease characteristics reported to date (18). It is therefore possible that variation in disease severity could be partly attributed to differences in populations affected by these viruses. Indeed, in this study, infants infected with RV were older, but had a lower birth weight than those infected with RSV. This finding suggests that non-RSV bronchiolitis cases might be defined as a different disease entity, that is, an acute viral illness resembling an asthma exacerbation; however, both the common clinical presentation and the considerable number of double infections render this suggestion open to further investigation. Furthermore, although PCR might offer considerable advantages as a diagnostic methodology, its use in acute bronchiolitis has not been fully validated. In agreement with the data on increased severity, our data show that RV infection led the patients to the hospital considerably faster than RSV. Interestingly, and in contrast to this, in the case of combined RV and RSV infection, admission occurred much later in the course of the disease, suggesting that these infants had contracted these viruses successively. This was also the case in the analysis of single versus multiple infections, irrespective of the type of virus. As multiple infections seem not to be infrequent, it could be proposed that successive infections may play an important role in the natural history of bronchiolitis development. Furthermore, in a prolonged disease course, aggravation factors such as exposure to tobacco smoke could be more injurious, a theory that is in agreement with our observations of increased tobacco exposure in children with multiple infections. The high percentage of infants exposed to tobacco that was found in this study (70%) is close to that reported in epidemiological surveys for the general population in Greece (75%) (19). Several studies in the past have acknowledged the possibility of multiple infections in acute bronchiolitis. In one study, increased morbidity was attributed to unique cases of Pneumococcus, cytomegalovirus, and Pneumocystis carinii infection (20), suggesting that comprehensive viral and bacterial evaluation should be considered in the disease. In studies in which the presence of RV was assessed by virus culture, multiple infections did not affect disease characteristics (21). This can probably be attributed to the low detection rate for RV owing to the insensitivity of the culture techniques. In a study assessing the role of RV and multiple viruses in bronchiolitis, using sensitive PCR methodology, incidences of RSV and RV were similar to those reported herein (22). However, the severity of disease on admission was not assessed in that study. In contrast, duration of hospitalization was reported, which did not differ between single and multiple virus-infected infants. Although an indirect index of severity, duration of hospitalization depends considerably on a number of clinical decisions, which are more difficult to control. In evaluating the findings of the present study, a possibility that must be taken into account is the between-year variation in the pattern and/or severity of respiratory viruses in the community. RV includes more than 100 serotypes, whereas RSV subtypes A and B could be present at different ratios, possibly differing in severity in different years. Nevertheless, whereas some studies have shown differences in disease severity between RSV subtypes A and B (23), similar differences between RV serotypes have not been documented. Furthermore, in this and other studies, several cases remain undiagnosed, suggesting that technical issues, such as sample handling, may be further optimized. It has been previously hypothesized that in concomitant respiratory virus infections, picornaviruses or adenoviruses could serve as clinical illness promotion factors functioning additively or synergistically in bronchiolitis (24). Our data support this hypothesis. Taking into account the high prevalence of RV upper respiratory infections in the community, it is not surprising that many infants are infected with this virus as well. This cohort is now being followed up for evaluation of the effects of different viral etiologies in the development of reactive airway disease.
* All subjects received supplemental oxygen; often, nebulized ipratropium bromide was prescribed, and antibiotics were administered to children who appeared toxic or when radiological evidence of consolidation was present. On occasion, salbutamol (nebulized) and methylprednisolone (intravenous) were also given.  Received in original form December 10, 2001; accepted in final form February 13, 2002
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
INTRODUCTION
2 test was performed for the analysis of ordinal or categorical data. Results of the analysis of continuous data are expressed as means ± the standard error of mean. Probability values of less than 0.05 were considered significant. 
