Objectives: CARB-ES-19 is a comprehensive, multicenter, nationwide study integrating whole-genome sequencing (WGS) in the surveillance of carbapenemase-producing K. pneumoniae (CP-Kpn) and E. coli (CP-Eco) to determine their incidence, geographical distribution, phylogeny, and resistance mechanisms in Spain.
Methods: In total, 71 hospitals, representing all 50 Spanish provinces, collected the first 10 isolates per hospital (February to May 2019); CPE isolates were first identified according to EUCAST (meropenem MIC > 0.12 mg/L with immunochromatography, colorimetric tests, carbapenem inactivation, or carbapenem hydrolysis with MALDI-TOF). Prevalence and incidence were calculated according to population denominators. Antibiotic susceptibility testing was performed using the microdilution method (EUCAST). All 403 isolates collected were sequenced for high-resolution single-nucleotide polymorphism (SNP) typing, core genome multilocus sequence typing (cgMLST), and resistome analysis.
Results: In total, 377 (93.5%) CP-Kpn and 26 (6.5%) CP-Eco isolates were collected from 62 (87.3%) hospitals in 46 (92%) provinces. CP-Kpn was more prevalent in the blood (5.8%, 50/853) than in the urine (1.4%, 201/14,464). The cumulative incidence for both CP-Kpn and CP-Eco was 0.05 per 100 admitted patients. The main carbapenemase genes identified in CP-Kpn were blaOXA–48 (263/377), blaKPC–3 (62/377), blaVIM–1 (28/377), and blaNDM–1 (12/377). All isolates were susceptible to at least two antibiotics. Interregional dissemination of eight high-risk CP-Kpn clones was detected, mainly ST307/OXA-48 (16.4%), ST11/OXA-48 (16.4%), and ST512-ST258/KPC (13.8%). ST512/KPC and ST15/OXA-48 were the most frequent bacteremia-causative clones. The average number of acquired resistance genes was higher in CP-Kpn (7.9) than in CP-Eco (5.5).
Conclusion: This study serves as a first step toward WGS integration in the surveillance of carbapenemase-producing Enterobacterales in Spain. We detected important epidemiological changes, including increased CP-Kpn and CP-Eco prevalence and incidence compared to previous studies, wide interregional dissemination, and increased dissemination of high-risk clones, such as ST307/OXA-48 and ST512/KPC-3.
Rapid bacterial identification and antimicrobial resistance gene (ARG) detection are crucial for fast optimization of antibiotic treatment, especially for septic patients where each hour of delayed antibiotic prescription might have lethal consequences. This work investigates whether the Oxford Nanopore Technology’s (ONT) Flongle sequencing platform is suitable for real-time sequencing directly from blood cultures to identify bacteria and detect resistance-encoding genes. For the analysis, we used pure bacterial cultures of four clinical isolates of Escherichia coli and Klebsiella pneumoniae and two blood samples spiked with either E. coli or K. pneumoniae that had been cultured overnight. We sequenced both the whole genome and plasmids isolated from these bacteria using two different sequencing kits. Generally, Flongle data allow rapid bacterial ID and resistome detection based on the first 1,000–3,000 generated sequences (10 min to 3 h from the sequencing start), albeit ARG variant identification did not always correspond to ONT MinION and Illumina sequencing-based data. Flongle data are sufficient for 99.9% genome coverage within at most 20,000 (clinical isolates) or 50,000 (positive blood cultures) sequences generated. The SQK-LSK110 Ligation kit resulted in higher genome coverage and more accurate bacterial identification than the SQK-RBK004 Rapid Barcode kit.
With the reduction in sequencing price and acceleration of sequencing speed, it is particularly important to directly link the genotype and phenotype of bacteria. Here, we firstly predicted the minimum inhibitory concentrations of ten antimicrobial agents for Staphylococcus aureus using 466 isolates by directly extracting k-mer from whole genome sequencing data combined with three machine learning algorithms: random forest, support vector machine, and XGBoost. Considering one two-fold dilution, the essential agreement and the category agreement could reach >85% and >90% for most antimicrobial agents. For clindamycin, cefoxitin and trimethoprim-sulfamethoxazole, the essential agreement and the category agreement could reach >91% and >93%, providing important information for clinical treatment. The successful prediction of cefoxitin resistance showed that the model could identify methicillin-resistant S. aureus. The results suggest that small datasets available in large hospitals could bypass the existing basic research and known antimicrobial resistance genes and accurately predict the bacterial phenotype.