Abstract
Background
MALDI-TOF MS allows a rapid and cheap identification of bacteria in the clinical laboratory. Antibiotic susceptibility (AS) prediction and typing of bacteria require at least an extra day. Furthermore, typing if often not feasible to be performed at all. Here, we explore the use of MT for AS prediction and typing using Escherichia coli (EC) as a proof-of-concept.
Methods
A total of 50 EC strains, isolated from different clinical samples, were whole genome sequenced (WGS) by means of nanopore sequencing. AS patterns were assessed by means of the disc diffusion method (for amikacin, amoxicillin-clavulanic acid, ampicillin, cefotaxime, ceftazidime, cefuroxime, colistin, fosfomycin, levofloxacin, meropenem, nitrofurantoin, piperacillin-tazobactam, temocillin, tigecyclin, tobramycin, sulfomethoxazole-trimethoprim) and read by the Adagio reader. Culture EC (McConkey agar and TSA agar) were used for protein extraction and MT peptide spectra were generated using the FlexControl platform (Bruker). MT spectra, WGS and AS data were imported and analyzed by BioNumerics. As a reference for typing, multi-locus sequence types (MLST) were assessed in BioNumerics using the allele sequences from the online PubMLST database. Peptide MT spectra were compared and clustered using the Pearson correlation and UPGMA algorithm.
Results
MacConkey grown isolates clearly yielded more defined MT clusters compared to TSA grown isolates. WGS and MLST analysis showed the presence of 18 MLST types. MT defined 9 clusters that correlated well with MLST based typing (Figure 1). Furthermore, MT clustering correlated with AS patterns (Figure 1).
Conclusion
MT typing holds potential as a rapid and cheap tool to type bacterial isolates at the time of identification. Furthermore, it allows, at the time of identification, to make predictions of the AS. Further machine learning algorithms will allow to improve current predictions.
MALDI-TOF MS allows a rapid and cheap identification of bacteria in the clinical laboratory. Antibiotic susceptibility (AS) prediction and typing of bacteria require at least an extra day. Furthermore, typing if often not feasible to be performed at all. Here, we explore the use of MT for AS prediction and typing using Escherichia coli (EC) as a proof-of-concept.
Methods
A total of 50 EC strains, isolated from different clinical samples, were whole genome sequenced (WGS) by means of nanopore sequencing. AS patterns were assessed by means of the disc diffusion method (for amikacin, amoxicillin-clavulanic acid, ampicillin, cefotaxime, ceftazidime, cefuroxime, colistin, fosfomycin, levofloxacin, meropenem, nitrofurantoin, piperacillin-tazobactam, temocillin, tigecyclin, tobramycin, sulfomethoxazole-trimethoprim) and read by the Adagio reader. Culture EC (McConkey agar and TSA agar) were used for protein extraction and MT peptide spectra were generated using the FlexControl platform (Bruker). MT spectra, WGS and AS data were imported and analyzed by BioNumerics. As a reference for typing, multi-locus sequence types (MLST) were assessed in BioNumerics using the allele sequences from the online PubMLST database. Peptide MT spectra were compared and clustered using the Pearson correlation and UPGMA algorithm.
Results
MacConkey grown isolates clearly yielded more defined MT clusters compared to TSA grown isolates. WGS and MLST analysis showed the presence of 18 MLST types. MT defined 9 clusters that correlated well with MLST based typing (Figure 1). Furthermore, MT clustering correlated with AS patterns (Figure 1).
Conclusion
MT typing holds potential as a rapid and cheap tool to type bacterial isolates at the time of identification. Furthermore, it allows, at the time of identification, to make predictions of the AS. Further machine learning algorithms will allow to improve current predictions.
| Original language | English |
|---|---|
| Publication status | Published - 2022 |
| Event | 32nd European Congress of Clinical Microbiology & Infectious Diseases (ECCMID) - Lisbon, Portugal Duration: 23 Apr 2022 → 26 Apr 2022 https://www.eccmid.org/ |
Conference
| Conference | 32nd European Congress of Clinical Microbiology & Infectious Diseases (ECCMID) |
|---|---|
| Country/Territory | Portugal |
| City | Lisbon |
| Period | 23/04/22 → 26/04/22 |
| Internet address |