Analysis of plasmid transmissions in a single hospital using a real-time plasmid transmission detection pipeline

Introduction: Horizontal plasmid transmissions play a major role in the dissemination of antimicrobial resistance. Therefore, a new real-time plasmid transmission detection pipeline was implemented in SeqSphere+ (Ridom GmbH, Muenster, Germany). We evaluated this pipeline using published data of carbapenemase-producing isolates from a UK hospital (Roberts et al. Microb. Genomics 2023).

Goals: Our aim was to evaluate different pipeline parameters and to quantify the frequency of transmission events.

Methods: Mash databases for all plasmids were created with sketch sizes 1,000 and 10,000. Clustering was performed with a distance threshold of 0.001. Additional clustering with a size correction was applied for both sketch sizes. Here, the Mash distance was lowered by 0.0003 per 1% size difference. For size differences >40%, the uncorrected value was taken to account for multimer formation. Discrepancies between plasmid clusters resulting from these four approaches were assessed by analysing the genomic differences using Quast. Transmissions of integrative mobile genetic elements (iMGE) were analysed using CGE MobileElementFinder. Potential transmission events were counted based on the clusters found with sketch size 10,000 and size correction.

Results: Only one additional plasmid pair was found with sketch size 10,000, which was also found when size correction was applied to sketch size 1,000. With size correction (sketch size 10,000), another 3 clusters contained additional plasmids. According to Quast, these had 13, 5, and 2 genomic differences, respectively. In total, 44 potential transmission events were counted. Of these, 12 were clonal, 18 single plasmid transmissions, and 14 co-transmissions. Fifteen samples transferred more than one plasmid, 9 to the same and 6 to different receiving samples. Two samples were involved in a clonal and an additional plasmid transmission. No transmissions of iMGE were detected.

Conclusions: For analyses of plasmid transmissions using Mash, a sketch size of 1,000 is sufficient if the distance is corrected for size differences. Plasmid transmissions in hospitals could be twice as common as clonal transmissions.