ACCURATE determination of antimicrobial resistance (AMR) in Streptococcus pneumoniae is critical for both clinical management and surveillance efforts.
Traditional antimicrobial susceptibility testing (AST) methods for pneumococci, such as disk diffusion and broth microdilution, are essential but time-consuming. Additionally, determining exact minimal inhibitory concentrations (MIC) for beta-lactam antibiotics remains a challenge.
Recently, next-generation sequencing (NGS) tools such as Pathogenwatch and AREScloud have integrated AMR prediction, including beta-lactam MICs, into their analysis workflows. However, the accuracy of these tools in comparison to phenotypic AST based on The European Committee on Antimicrobial Susceptibility Testing (EUCAST) guidelines has not been thoroughly assessed.
A recent study evaluated the performance of Pathogenwatch and AREScloud in predicting AMR in a cohort of 538 S. pneumoniae isolates. These isolates were tested against penicillin, amoxicillin, cefotaxime/ceftriaxone, erythromycin, trimethoprim-sulfamethoxazole, and tetracycline. Disk diffusion was performed on all isolates, and broth microdilution testing was carried out for isolates with reduced beta-lactam susceptibility. Demultiplexed FASTQ files generated from Illumina sequencing were used as input for the NGS tools, which were then compared to the results from traditional phenotypic AST.
The results showed that both NGS tools performed well, particularly for beta-lactam antibiotics. Categorical agreement (CA) was above 94% for both tools in predicting resistance to penicillin, amoxicillin, and cefotaxime/ceftriaxone, with very low major error (ME) and very major error (VME) rates (<1%). For erythromycin and tetracycline, AREScloud achieved a CA greater than 93%, while Pathogenwatch showed a slightly lower CA around 88%.
However, both tools struggled with trimethoprim-sulfamethoxazole, where the CA was below 86%. Importantly, the study found high VME rates for erythromycin and tetracycline, with Pathogenwatch showing significantly higher rates (53.6% and 47.0%, respectively) compared to AREScloud (14.3% and 19.1%, respectively).
In conclusion, both Pathogenwatch and AREScloud demonstrated strong performance in predicting beta-lactam resistance, making them valuable tools for AMR surveillance and clinical decision-making. However, further optimisation and validation are needed, especially for non-beta-lactam antibiotics, where higher VME rates were observed.
Ada Enesco, EMJ
Reference
Sanchez GJ et al. Prediction of antimicrobial susceptibility of pneumococci based on whole-genome sequencing data: a direct comparison of two genomic tools to conventional antimicrobial susceptibility testing. J Clin Microbiol. 2024; DOI: 10.1128/jcm.01079-24.
