Tracheostomy in children is associated with neutrophilic airway inflammation

Background Tracheostomies in children are associated with significant morbidity, poor quality of life, excess healthcare costs and excess mortality. The underlying mechanisms facilitating adverse respiratory outcomes in tracheostomised children are poorly understood. We aimed to characterise airway host defence in tracheostomised children using serial molecular analyses. Methods Tracheal aspirates, tracheal cytology brushings and nasal swabs were prospectively collected from children with a tracheostomy and controls. Transcriptomic, proteomic and metabolomic methods were applied to characterise the impact of tracheostomy on host immune response and the airway microbiome. Results Children followed up serially from the time of tracheostomy up to 3 months postprocedure (n=9) were studied. A further cohort of children with a long-term tracheostomy were also enrolled (n=24). Controls (n=13) comprised children without a tracheostomy undergoing bronchoscopy. Long-term tracheostomy was associated with airway neutrophilic inflammation, superoxide production and evidence of proteolysis when compared with controls. Reduced airway microbial diversity was established pre-tracheostomy and sustained thereafter. Conclusions Long-term childhood tracheostomy is associated with a inflammatory tracheal phenotype characterised by neutrophilic inflammation and the ongoing presence of potential respiratory pathogens. These findings suggest neutrophil recruitment and activation as potential exploratory targets in seeking to prevent recurrent airway complications in this vulnerable group of patients.


Patients and volunteers
Tracheostomy cohorts included children enrolled at the time of tracheostomy and followed for up to 3 months (serial cohort).Samples were collected at the time of tracheostomy and subsequently at approximately one week, one month and three months post-procedure.A second cohort comprised children with an established tracheostomy in place for over 6 months (long-term cohort).Exclusion criteria for both tracheostomy cohorts comprised; age >15 years or a diagnosis of hereditary disorder associated with recurrent airway infections, such as cystic fibrosis or primary immunodeficiency.Controls were a convenience sample of children recruited from elective operating lists of patients undergoing airway examination under general anesthesia for investigation of structural (non-infective) airway problems.Exclusion criteria comprised; age >15 years or a diagnosis of a hereditary disorder associated with recurrent airway infections.

Collection of tracheal secretions, airway swabs and tracheal brushings
Eligible patients had tracheal secretions collected by an experienced operator.Briefly, in nontracheostomised patients secretions were suctioned under direct vision from the trachea via a bronchoscope or laryngoscope.A nasal swab was also collected for microbiome analysis.It was not possible to collect every sample at each visit due to patient compliance with sampling (for example cytology brushings in awake participants) or failure to provide samples (for example no aspiration within the allocated timeframe).All samples were transported to the laboratory fresh and immediately frozen at -80°C until further processing.
BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s)

RNA sequencing
RNA was extracted from tracheal wall brushings using the RNeasy mini kit (Qiagen, Hilden, Germany) and stranded total RNA sequencing libraries were prepared using the SMART-Seq Stranded Kit (Takara Bio, CA) and SMARTer RNA Unique Dual Index Kit (Takara Bio, CA) following the manufacturer's protocol.Libraries were quantified using a TapeStation 4200 (Agilent Technologies, CA) and Qubit 4 (Thermo Fisher, MA) and equimolar samples pooled.
The pooled library was sequenced at ~80 million (2 x 100 base pair) reads per sample on a NovaSeq 6000 using an S2 200 cycle flow cell (Illumina, CA).Data for individual samples were demultiplexed into separate FASTQ files using Illumina's bcl2fastq software (version 2.20.0.422) and quality checks on the raw data were performed using FastQC (version 0.11.50) and Fastq Screen (version 0.14.1).Alignment of the RNA-seq paired-end reads was to the GRCh38.104version of the human genome, and annotation performed using Hisat2 (version 2.2.1).Expression levels were determined and statistically analysed by a workflow combining HTSeq (version 0.6.1), the R environment, utilising packages from the Bioconductor data analysis suite and differential gene expression analysis based on the negative binomial distribution using the DESeq2 package.All FDRs were calculated using the DESeq2 package default method.Non-protein-coding genes were excluded from downstream analysis.Further data analysis and visualisation used R (environment version 4.0.3) and Bioconductor packages.

Proteome testing
Tracheal aspirate samples were mixed with sodium dodecyl sulfate solution, heated at 95 °C and sonicated to lyse cells and remove DNA/RNA.The dried samples were then dissolved in tetraethylammonium bromide, reduced with tris(2-carboxyethyl)phosphine and alkylated with iodoacetamide.Each sample was then acidified and loaded onto S-Trap cartridges (Protifi, NY) following the manufacturer's instructions.Retained proteins were digested with trypsin.
BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s) Peptides released from the cartridge were frozen and dried, reconstituted and loaded onto a liquid chromatography-mass spectrometer.Peptides were separated with a 70 min non-linear gradient (3-40% B, 0.1% formic acid (Line A) and 80% acetonitrile, 0.1% formic acid (LineB)) using a UltiMate 3000 RSLCnano high-performance liquid chromatographer (Thermo Fisher, MA).Samples were first loaded/desalted onto a 300μm x 5mm C18 PepMap C18 trap cartridge in 0.1% formic acid at 10 µL/min for 3 min and then further separated on a 75μmx50cm C18 column (EasySpray -C18 2 µm) with integrated emitter at 250nl/min.The eluent was directed to an Orbitrap Exploris 480 mass spectrometer (Thermo Fisher, MA) through the EasySpray source at a temperature of 320 °C, spray voltage 1,900V.Orbitrap full scan resolution was 60,000, re-imagined focus lens 50%, normalised ACG Target 300%.Precursors for tandem mass spectrometry were selected via a top 20 method.Intensity threshold 5.0 e3, charge state 2-7 and dynamic exclusion after 1 times for 35 seconds 10 parts per million (ppm) mass tolerance.ddMS2 scans were performed at 15,000 resolution, higher-energy collisional dissociation energy 27%, first mass 110 m/z, automatic gain control target standard.The acquired data were searched against the human proteome sequence database (https://www.uniprot.org/uniprot/)concatenated to the Common Repository for Adventitious Proteins v.2012.01.01 (ftp://ftp.thegpm.org/fasta/cRAP),using MaxQuant v1.6.43 (https://maxquant.net/perseus/).Parameters used: cysteine alkylation: iodoacetamide, digestion enzyme: trypsin, Parent Mass Error of 5ppm, fragment mass error of 10ppm.The confidence cut-off representative to false discovery rate (FDR)<0.01 was applied to the search result file.The report was generated to include protein groups, then further processed to exclude reversed sequences, common contaminants, proteins with only <2 unique peptides.
Each protein was required to have more than 50% valid values (protein intensity>0) in at least one experimental group.Low quality samples were excluded from the analysis based on visual assessment of the spectral data and minimum number valid quantifiable proteins (at least 900

Metabolome testing
Non-targeted metabolomics was performed by Metabolon (Morrisville, NC).Briefly, recovery standards (quality control) were added to tracheal aspirates and proteins precipitated with methanol followed by centrifugation.The resulting extract was divided into five fractions: two (early and late eluting compounds) for analysis by ultra-high-performance liquid chromatography-tandem mass spectrometry (UHPLC-MS/MS) (positive ionisation), one for BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s)

BMJ
Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s) Protein intensity values were normalized by Log2 transformation, followed by median subtraction within each sample to account for unequal loading.The remaining missing values were then imputed by random selection from the left tail of the valid values distribution (1.8 standard deviations from the mean, within a 0.3 standard deviation range).T-test was used to identify proteins that were significantly different between groups.Permutation based FDR was calculated to account for multiple comparisons (250 randomisations) and FDR<0.05 was used as a threshold.All the analysis and visualisation of the proteomic data were conducted using Perseus 1.6.15.0.The most significantly differentially abundant proteins (top 250 most abundant and 250 least abundant) between the control and long-term tracheostomised patients generated by this analysis were uploaded to Ingenuity® Pathway Analysis (IPA®) software, Qiagen (Hilden, Germany).IPA® uses a knowledge base derived from the scientific literature to relate genes or proteins based on their interactions and functions.Based on the uploaded dataset, the program algorithmically generates biological networks and functions and is able to assign activation (z-scores), where positive predicts activation and negative inhibition of a gene/pathway.

Table S1 .
Extended clinical details of the serial tracheostomy cohort.

Table S2 .
Extended clinical details of the long-term tracheostomy cohort.

Table S4 .
Metabolomic dipeptide sub-pathway findings from tracheal aspirates.Metabolite fold change is demonstrated in the post-tracheostomy group relative to the time of tracheostomy.Significant values (p<0.05) are in bold and marked with an asterisk.Statistical analysis was by One-Way Repeated Measure ANOVA: time of tracheostomy n=9, 1 week n=9, 1 month n=6, 3 months n=5.BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s)