comparison of MMS, OPS and BAL samples in metagenomic and clinical microbiological surveillance. SAINTS-CF-BAL-MMS-OPS

Background Surveillance for airway infection is standard practice in clinical cystic fibrosis (CF) management. In young children who do not produce sputum, oropharyngeal swabs (OPS) are the most widely used method but lack specificity. Broncho-Alveolar Lavage (BAL) is the standard reference method of detecting infection, but is invasive and requires general anaesthesia. A non-invasive alternative to BAL and OPS is a nasal swab. Swabbing of the middle meatus (MMS) has been shown to be an effective non-invasive method of reliably detecting pathogens in children with acute and chronic rhinosinusitis but is little studied in CF. Methods In a stable Irish preschool CF cohort undergoing surveillance BAL in a single centre we compared BAL, MMS, and OPS samples using both standard microbiological culture-based surveillance and high-throughput metagenomic sequencing. Standard culture was performed in hospital microbiology laboratories in line with international recommendations. Metagenomic sequencing was carried out on an Illumina HiSeq 550 platform. Results Matched BAL, MMS, and OPS were collected from 30 preschool children, producing 88 samples of which 86 were available for analysis. High-throughput metagenomic sequencing demonstrated that the microbiome in OPS was significantly more taxonomically diverse than in BAL or MMS (richness, Shannon, and Inverse Simpson indices; Wilcox test: W = 6-574, p =0.02-<0.0001) although it was lower in nucleotide sequence diversity. Microbiome data also revealed that the pathogens Escherichia, Staphylococcus, and Streptococcus were prevalent across the cohort and in all sample sites (BAL, MMS, OPS), while other pathogens (Achromobacter, Chryseobacterium, Haemophilus, Moraxella, Pseudomonas) were detected infrequently from all sample sites. On examining agreement between culture and metagenomic methods at identifying pathogens, we found corroboration between culture and metagenomic methods to be highly variable for all sample types, and found diagnostic performance to be highly dependent on the pathogen under consideration. When comparing microbiome and culture data from OPS and MMS samples with BAL samples, MMS more accurately represented BAL than OPS, particularly for Streptococcus, Staphylococcus, and Escherichia. Conclusions Metagenomic sequencing on airway samples of preschool children with CF demonstrated that BAL and MMS samples are highly similar in their total microbial composition, and as a result performed equally well as proxies for culture-based surveillance. These results suggest that sampling of the middle meatus is a more accurate surrogate for lower airway samples when compared with oropharyngeal swabs and should be considered as an alternative airway sampling method in young children with CF.