Detection of antibiotic susceptibility in adult people with cystic fibrosis using a metagenomics approach

Routine antibiotic susceptibility (ABS) testing for bacterial respiratory infections in people with cystic fibrosis (pwCF) relies on phenotypic techniques. These take several days whilst timely intervention is key. To bridge this time gap, physicians in CF-care use historical ABS data to start ABs, with risk of resistance to it. This pilot study explores a time-saving alternative to detect ABS: the possibility to detect antibiotic resistance genes (ARGs) using shotgun metagenomics. Adult CF patients were recruited in a one-year longitudinal study [Neerincx, A.H. et al. ERJ Open Res 2021; 7(2)]. Collected sputum samples were shotgun sequenced and analyzed for presence of ARGs by deep learning tool deepARG [Arango-Argoty, G. et al. Microbiome 2018; 6:23]. The deepARG analysis was compared to results from routine ABS testing using areas under receiver operating characteristic curves (AUROCCs) and % sensitivity. Twenty CF patients (median age 25 years, 40% females) provided 68 sputum samples at different study visits that were suitable for routine ABS testing and metagenomic sequencing. Significant results were obtained for the following antibiotic classes: aminoglycoside (AUROCC = 0.81 (95% CI: 0.67-0.95, sensitivity = 73%)), cephalosporin (AUROCC = 0.70 (95% CI: 0.54-0.86, sensitivity = 95%)), and fluoroquinolone (AUROCC = 0.73 (95% CI: 0.56-0.89, sensitivity = 88%)). For other antibiotic classes results were not significant. These results suggest that a metagenomic approach and using bioinformatic tools is promising for timely detection of ABS for commonly used antibiotics. In pwCF, these results may lead to improved treatment of exacerbations and increased antibiotic stewardship.