In the present study, we used suspension array to simultaneously detect multiple viral and bacterial pathogens in paired BALF and OPS specimens from symptomatic patients hospitalized with respiratory illness. To validate the reliability and accuracy of the multipathogen testing platform, we have compared the yield of suspension array with that of metagenomic next-generation sequencing for microbiological findings and highlighted the high concordance of 13 targets between the two methods (manuscript in preparation). These results showed that our suspension array can reliably identify pathogens in patients with RTI. The fast turnaround time (within 4–5 h) makes it possible to be a valuable tool in clinical settings.
The reliability of the specimens taken via oropharyngeal/nasopharyngeal swab or wash compared to the deep samples such as BALF is a matter of debate and it would be interesting to investigate if they are really useful. Here, BALF and OPS had similar microbiological yields (75% vs. 65%, p > 0.05). The differences in the frequency distribution and sensitivity for most targeted pathogens except Moraxella catarrhalis of two sampling methods were not statistically significant. Selection of a sampling method for detecting respiratory pathogens must take into account its sensitivity, feasibility and costs. The collection of OPS is relatively simple, quick and less invasive compared to other sampling methods. For these reasons, we consider the OPS as the preferred method of respiratory tract sampling for pathogen detection.
For Moraxella catarrhalis, we detected 4 cases in BALF and none in paired OPS specimens. In general, Moraxella catarrhalis causes mainly upper respiratory tract infection (otitis media) in children and lower respiratory tract infection in adults with previously compromised airways such as chronic obstructive pulmonary disease [17, 18]. However, some reports have demonstrated that Moraxella catarrhalis may be involved in lower respiratory tract infections in children [19], which is consistent with our results. More studies are needed to investigate its role in respiratory illness in hospitalized children.
Among the 76 patients, 25 (32.9%) had the same pathogens identified from the OPS and BALF suspension array, and 14 (18.4%) had concordant negative results. For the 14 cases, they might be infected by some rare pathogens that are not covered by our suspension array. The top 5 pathogens were RSV, SP, Hi, MP and Adv, accounting for 74.2% of etiological fraction. Our data are in agreement with other recent findings in multi-country case-control studies that found RSV was the most common cause of severe childhood pneumonia [20, 21]. Thus, RSV could be a primary target for children hospitalized with respiratory illness. In general, RSV is most commonly found in lower respiratory tract infections particularly in infants [22]. However, the sensitivities of RSV between OPS and BALF was not significantly different in our testing. RSV infection and replication initiates in the nasopharynx. The virus could be found in both upper and lower airway via high-sensitivity molecular techniques when it spreads from the upper respiratory tract to the lower in individuals with compromised immunity.
Bordetella pertussis was rarely identified in infants perhaps due to high vaccination rates. In this study, we detected Bordetella pertussis in 5 young children. Identification of Bordetella pertussis in BALF or OPS specimens may provide predictive value for the outcome of respiratory illness at the individual case level [23]. In terms of some pathogens such as influenza A, adenovirus, Mycoplasma pneumoniae and Streptococcus pneumoniae, relatively low concordance between BALF and OPS specimens for them may reflect different cell tropisms for different parts of the respiratory tract. Notably, none of the 76 cases was tested positive for HRV in this work, which is somewhat surprising given that this virus is often associated with upper respiratory infection. However, another project in our group showed that the detection rate of HRV was ~ 2% in OPS among 2895 pediatric outpatients with respiratory illness. Here, the included children were inpatients. Although HRV infections are frequent, they are mostly limited to the upper respiratory tract and generally cause relatively mild symptoms [24, 25]. The contribution of HRV may vary by disease severity of included patients and other factors.
Cultivation is regarded as the gold standard in etiological identification. As shown in Table 1, we have performed sputum culture in 42 of 76 cases. Compared with PCR-based methods, the detection rate of it was significantly lower. One of possible explanations is the empirical antibiotic therapy in patients before sampling. Here, we focused on comparison of OPS and paired BALF in detecting respiratory pathogens, rather than the sensitivity of nucleic acid-based array compared to the gold standard, i.e. cultivation. Based on the same consideration, a healthy control group was not tested for ruling out false positives in this work because of our specific interest and aim. Similarly, a prior study, 86 patients enrolled and no healthy controls included, has also applied this strategy to compare the yields of bronchoalveolar lavage samples with that of nasopharyngeal swabs by using FilmArray respiratory panel [8].
This study has several limitations. First, our sample size was relatively small (n = 76) because we focused on paired BALF and OPS specimens collected from hospitalized children. As a result, there was limited power to compare the sensitivities of BALF and OPS for specific respiratory pathogens in patients. Further studies in a larger cohort may generate a relatively high degree of precision when performing comparative statistical analysis. Second, although BALF are regarded lower respiratory tract samples, oropharyngeal intubation for BALF might result in potential contamination by upper respiratory tract “contaminants”, particularly for bacteria/viruses known to colonize the oropharynx. Thus, the BALF specimen might be actually both an upper and a lower airway combined sample. However, it does not affect our primary purpose that focuses on the microbiological findings of OPS sampling. Third, empirical antibiotic use in clinical practice may reduce the sensitivity of assays, particularly for bacteria. Collectively, identification of a pathogen does not necessarily equate to the etiological agent, particularly in a multipathogen testing and laboratory results require further interpretation by experienced clinicians. In addition, we here focused on a subset of potential pathogens because the 13 agents (8 viruses and 5 bacteria) are key respiratory pathogens in children based on previous epidemiological investigations in China. In fact, fungi are also important pathogens causing severe infections of the respiratory system. Another project in our group by using metagenomic next-generation sequencing found that the infection rates of Candida albicans, Pneumocystis jiroveci and Aspergillus fumigatus in young children admitted to PICU with respiratory illness were 3.52, 1.68 and 1.23%, respectively. Given the high prevalence and importance of the airborne fungal pathogens in respiratory infections, we plan to add fungal species to our upgraded in-house array, which would be a separate study due to many experiments and large undertakings.