164 Chapter 9 Optimization of the detection of AmpC-mediated resistance in E. coli To ensure effective surveillance of antimicrobial resistance, accurate diagnostic tools play a crucial role. Detecting plasmid-encoded AmpC involves both phenotypic and genotypic testing. However, distinguishing between chromosome-encoded and plasmidencoded ampC genes can be challenging due to their coexistence, particular in E. coli. While molecular confirmation tests are commonly used for screening plasmid-encoded AmpC-producing E. coli, they can be time-consuming and costly, making them less feasible in certain settings, especially those with limited resources. In low-resource settings, it would be beneficial to have a practical algorithm that can distinguish between ampC genotypes in E. coli based on phenotype alone. This approach would help identify which isolates should undergo further confirmation with molecular testing. By quickly identifying plasmid-encoded AmpC isolates, infection control practices can be improved, unnecessary and costly isolation measures can be minimized, and appropriate treatment strategies can be implemented. Chapter 4 provides a detailed description of the development of such a model, utilizing machine learning techniques. The algorithm utilizes the minimum inhibitory concentration (MIC) of cefotaxime to predict the presence of plasmid-encoded AmpC in cefoxitinresistant (>8 mg/L) and ESBL-negative E. coli. Although the algorithm demonstrated high accuracy, it is important to note that the data used for model development and training primarily consisted of blaCMY-2-containing E. coli and a limited number of other plasmid-encoded AmpC genotypes. While this reflects the Dutch epidemiological situation, it is worth considering that geographical variations may exist in other regions. Nevertheless, the application of machine learning techniques holds immense potential in optimizing screening algorithms within the field of microbiology. With well-curated datasets based on standardized phenotypical and genotypical testing, the possibilities for advancements in laboratory development and infection control methods are boundless. The screening of antimicrobial resistance, particularly in Enterobacterales obtained from intestinal samples with a multitude of other gut bacteria, poses significant challenges. To improve detection rates, selective and differential media like MacConkey or Drigalski lactose agars are commonly employed, as they effectively suppress grampositive bacteria. Furthermore, the use of highly selective media tailored for specific ESBL- or carbapenemase-producing Enterobacterales has shown enhanced detection capabilities for resistant isolates (Glupczynski et al. 2007; T. D. Huang et al. 2010; Göttig et al. 2020). However, limited attention has been given to the development of
RkJQdWJsaXNoZXIy MTk4NDMw