The Effect of Inhaled Corticosteroid Withdrawal on the Airway Metagenome and Antimicrobial Resistance: Data From the INCOGNITO Trial

Introduction: Previous studies using 16S rRNA sequencing have identified changes in the airway
microbiome associated with inhaled corticosteroid treatment (ICS). Recent data has shown that studying microbial interactions (the interactome) may provide additive clinical insight. This study aimed to evaluate the effect of ICS withdrawal on the airway metagenome in patients with COPD.
Methods: Stable patients with moderate to severe COPD currently treated with ICS and with a
blood eosinophil count <300cell/ul were enrolled. Patients were randomised to receive either
Tiotropium and Olodaterol (ICS withdrawal) or Fluticasone Furoate/Vilanterol (ICS continuation) for 6 months. Sputum samples were taken at baseline and months 1,2,3 and 6 and during exacerbation. The primary outcome was sputum bacterial load using 16S qPCR and the secondary outcome was microbiome characterisation by 16S rRNA sequencing. In an exploratory substudy, shotgun metagenomic sequencing was performed in sputum. Here we present preliminary data from 270 sputum samples from 54 participants. Interactome analysis was performed by compositionality-corrected generalized boosted linear models (GBLM). Results: 80 patients with COPD were randomized, mean age 69.3, 51.2% male, 32.5% current smokers. 38 patients were randomised to T/O, 42 to FF/VI, over a six-month treatment period. There were no differences in the primary outcome of bacterial load between groups (mean difference -0.03 log units 95%CI-0.15 to 0.09, p=0.61). In the sputum metagenome, the most abundant taxa were Streptococcus parasanguinis, Rothia mucilaginosa, Streptococcus mitis, salivarious and pneumoniae, Haemophilus influenzae and Gemella sanguinis. The resistome showed a high abundance of genes related to fluoroquinolone, tetracycline, beta-lactam, macrolide and multidrug resistance. Metagenomic sequencing supported the results of 16S rRNA sequencing showing an increase in Streptococcus and relative reduction in Haemophilus following ICS withdrawal. Interactome analysis showed that following ICS withdrawal, commensal taxa such as Streptococcus, Prevotella, Campylobacter and Veillonella were the most busy, critical and influential within the microbial network. In patients in the ICS continuation arm, microbial interaction networks demonstrated a lesser role of commensals and greater prominence of Haemophilus, Moraxella and lactobacillus (figure 1). In keeping with this, longitudinal differential analysis of the interactome over the study period suggested ICS withdrawal was associated with largest changes in Streptococcus, Fusobacterium and Mogibacterium, while ICS continuation primarily affected Veillonella and Parvimonas. Conclusion: ICS withdrawal is associated with changes in microbial interactions in the
airway metagenome. Figure 1. Microbial interaction networks (from GBLM) between microbes with positive (green) and negative (red) interactions. Data from 6-month treatment visit).