Infants hospitalized for Bordetella pertussis infection commonly have respiratory viral coinfections

BACKGROUND
Whether viral coinfections cause more severe disease than Bordetella pertussis (B. pertussis) alone remains
unclear. We compared clinical disease severity and sought clinical and demographic differences between infants with B. pertussis infection alone and those with respiratory viral coinfections. We also analyzed how respiratory infections were distributed during the 2 years study.


METHODS
We enrolled 53 infants with pertussis younger than 180 days (median age 58 days, range 17–109 days, 64.
1% boys), hospitalized in the Pediatric Departments at “Sapienza” University Rome and Bambino Gesù Children’s Hospital from August 2012 to November 2014. We tested in naso-pharyngeal washings B. pertussis and 14 respiratory viruses with real-time reverse-transcriptase-polymerase chain reaction. Clinical data were obtained from hospital records and demographic characteristics collected using a structured questionnaire.


RESULTS
28/53 infants had B. pertussis alone and 25 viral coinfection: 10 human rhinovirus (9 alone and 1 in coinfection
with parainfluenza virus), 3 human coronavirus, 2 respiratory syncytial virus. No differences were observed in clinical disease severity between infants with B. pertussis infection alone and those with coinfections. Infants with B. pertussis alone were younger than infants with coinfections, and less often breastfeed at admission.


CONCLUSIONS
In this descriptive study, no associations between clinical severity and pertussis with or without
co-infections were found.


TRIAL REGISTRATION
Policlinico Umberto I: protocol 213/14, 3085/13.02.2014, retrospectively registered.
Bambino Gesù Children’s Hospital: protocol n. RF-2010-2317709.


Background
Pertussis (whooping cough) is a highly contagious, respiratory disease caused by Bordetella pertussis (B. pertussis). The clinical symptoms of pertussis change with age, previous exposure to B. pertussis and immunization status. In newborns clinical manifestations may be severe. Most infants have a typical paroxysmal cough which can last more than two months [1].
Pertussis is a major cause of morbidity worldwide and of mortality in infants in developing countries. Pertussis continues as a public health concern threat given its re-emergence despite high vaccination coverage [2]. Epidemic cycles reoccur every 2 to 5 years and 2015 has witnessed the worst outbreak in the past 70 years [3].
We designed this study to compare clinical disease severity in infants with B. pertussis infection alone and those with B. pertussis and viral coinfections hospitalized in two Italian centers over two years. We also analyzed how respiratory infections and pertussis cases were distributed during the two years study. As primary outcome measures we assigned each infant a clinical severity score and assessed length of hospitalization. As an experimental approach to provide reliable data on lower respiratory virus infections we used an extended respiratory virus panel that can detect 14 respiratory viral targets with real-time reverse-transcriptase-polymerase chain reaction (RT-PCR) assay.

Patients
In a longitudinal double-center study, we enrolled 53 consecutive infants with pertussis younger than 180 days hospitalized from August 2012 to November 2014 at the Pediatric Departments "Sapienza" University and Bambino Gesù Children's Hospital Rome. Eligible children had to have a nasopharyngeal specimen that tested positive for B. pertussis, be younger than 180 days, be unvaccinated against pertussis and have at least one of the following symptoms: cough lasting more than 5 days, paroxysmal cough, apnea or cyanosis and post-cough vomit.
At admission all infants underwent a nasopharyngeal washing obtained by instilling 3 ml of sterile saline into each nostril and collected with a syringe. All samples were delivered within two hours to the Department of Infectious, Parasitic & Immune-mediated Diseases at the Istituto Superiore di Sanità (Rome) for B. pertussis detection and to the Molecular Medicine Department ("Sapienza" University Rome Virology Laboratory) for virus detection. We considered children with B. pertussis infection, those with PCR positive results.
Before infants were enrolled, all children's parents agreed and gave written informed consent to participate in the study, which was approved by institutional review boards at both hospitals (Policlinico Umberto I: protocol 213/14, 3085/13.02.2014; Bambino Gesù Children's Hospital: protocol n. RF-2010-2317709).
Bordetella pertussis detection by RT-PCR and culture B. pertussis DNA was extracted with QIAamp DNA minikit (QiaGEM, Hilden, Germania) and amplified with the "Bordetella Real-Time PCR" kit (Diagenode Diagnostics, Liège, Belgio). The assay gave binary results. For RT-PCR the SYBR Green Detection assay we used the LightCycler 2.0 system (Roche Diagnostic). Data were analyzed with LightCycler software (version 4.0, Roche Diagnostic). Only the positive samples for B. pertussis were cultured on charcoal agar plates (Oxoid England) containing defibrinated sheep blood at 10% and incubated at 35°C up to 7 days and inspected daily, as previously described [13].

Statistical analysis
Statistical significance was analyzed with SPSS version 23.0 (SPSS Inc., Chicago, IL, USA). Data included percentages for discrete variables, median and range for continuous variable. Differences among groups were compared using non-parametric test for median comparison, Mann-Whitney test. The χ-square test was applied to analyze categorical independent variables qualitatively. P-values <0.05 were considered to indicate statistical significance.
Clinical severity score and days hospitalization were similar in children with B. pertussis infection alone and those with B. pertussis and viral coinfection ( Table 1).
The questionnaire indicated that infants with B. pertussis alone were younger than infants with coinfections (p = 0.044 by Mann-Whitney test). Questionnaire answers also showed that infants with B. pertussis alone were less often breastfeed at admission than infants with coinfections (p = 0.048 by χ-square test) and for a shorter time (p = 0.004 by Mann-Whitney test). Finally, infants with coinfections more frequently had a higher number of cohabitants though not significantly (p = 0.076 by χ-square test).
The two groups had no significant differences for other demographic, clinical, laboratory and radiological data (Tables 2, 3 and 4).

Discussion
In this descriptive study investigating the clinical data for 53 infants younger than 180 days hospitalized with pertussis, no associations between clinical severity and pertussis with or without co-infections were found.
As many as 25/53 (47%) of the infants hospitalized with pertussis over the two years were coinfected with other respiratory viruses. A distinctive point is that because we analyzed a wide virus battery and did so in infants hospitalized continuously in two centers over two years we feel confident that our study provides reliable data on the distribution of respiratory infections. Few studies found B. pertussis cases in infants with respiratory pathogens. For example, Piedra et al., identified by RT-PCR only 4 B. pertussis cases in 2068 patients with respiratory pathogens and all these infants were younger than 6 months. In 3 of the 4 infants with B. pertussis RT-PCR identified a second respiratory pathogen: 2 had an hRV and 1 child an hCoV coinfection [8]. In a similar study, Korppi et al., showed in a nasopharyngeal aspirate by RT-PCR B. pertussis infection in 7 on 9 of patients hospitalized for RSV infection [15]. Our lower percentage of coinfections reflects the analysis of a larger series of children with B. pertussis than the other two studies.
As well as extending current epidemiological knowledge on respiratory pathogens, we confirmed that pertussis in Italy arises mainly during the summer months, whereas respiratory virus coinfections are equally distributed over the year [10]. Virological analysis showed that the most  frequent virus observed was hRV (36%) and only 2 infants had RSV (8%). We presume that the distribution of coinfections reflects the observation that hRV epidemiology maintains a steady epidemic curve throughout the year and hRV can be detected also in asymptomatic children. In fact in a Netherlands study conducted in recent years Wildenbeest JG et al., found hRV in 25% of asymptomatic children [16].
A new finding in this study is the lack of differences in clinical disease severity (measured as clinical severity score and days hospitalization) in infants with B. pertussis infection alone and those with coinfections. Although no other published studies have tested 14 respiratory viruses in infants with pertussis, Nuolivirta et al., studied 142 infants aged less than 180 days hospitalized for bronchiolitis who underwent a nasopharyngeal aspirate to detect only 7 respiratory viruses and B. pertussis by RT-PCR. B. Pertussis involvement was found in 12 of 142 (8.5%) infants hospitalized for bronchiolitis and, of these, 8 were in coinfection with RSV. They found no differences in clinical findings, days hospitalization and breastfeeding at admission in patients with respiratory viruses alone than in patients with respiratory viruses in coinfections with B. pertussis [17]. Schnoeller et al. showed that respiratory infection of neonatal mice with an attenuated B. pertussis can protect against RSVinduced disease in adult life [18,19]. In a human model Schiavoni et al. demonstrated that an attenuated B. pertussis rescues the immune functions of RSV infected human dendritic cells by promoting a protective Th1/ Th17 responses. Self-limiting respiratory infections with attenuated bacteria or commensal microbes may have a beneficial effect limiting potentially lethal diseases caused by respiratory viruses in infants [20].
When we evaluated the secondary endpoints in our study, infants with B. pertussis alone were younger, less often breastfeed and breastfeed for a shorter time than infants with coinfections. Similarly, in a previous study from our group, comparing infants with B. pertussis and those with bronchiolitis, we found that the percentage of breastfed infants at hospitalization was lower in infants with B. pertussis than in those with RSV bronchiolitis [13]. This finding is difficult to explain. We always strongly encourage breastfeeding in infants because it has well-known immunological and nutritional advantages [21]. In a recent large multicenter study, on pertussisassociated pneumonia in children from low-and middleincome countries, pertussis-positive cases were more likely to have never been breastfed compared with controls [22]. Our finding might reflect mother-to-child transmission of respiratory infections during breastfeeding.
Our study has limitations: small sample size and our failure to compare B. pertussis infection with viral coinfections by single respiratory viruses.

Conclusions
In conclusion, about one third of infants with B. pertussis may have a respiratory viral coinfection. These undetected coinfections seem to leave the clinical severity of pertussis in infants unchanged.