Abstract
Characterization of transmission routes of Salmonella among various food-animal reservoirs and their antibiogram is crucial for appropriate intervention and medical treatment. Here, we analyzed 3728 Salmonella enterica serovar Newport (S. Newport) isolates collected from various food-animals, retail meats and humans in the United States between 1996 and 2015, based on their minimum inhibitory concentration (MIC) toward 27 antibiotics. Random Forest and Hierarchical Clustering statistic was used to group the isolates according to their MICs. Classification and Regression Tree (CART) analysis was used to identify the appropriate antibiotic and its cut-off value between human- and animal-population. Two distinct populations were revealed based on the MICs of individual strain by both methods, with the animal population having significantly higher MICs which correlates to antibiotic-resistance (AR) phenotype. Only ∼9.7% (267/2763) human isolates could be attributed to food-animal origins. Furthermore, the isolates of animal origin had less diverse antibiogram than human isolates (P < 0.001), suggesting multiple sources involved in human infections. CART identified trimethoprim-sulfamethoxazole to be the best classifier for differentiating the animal and human isolates. Additionally, two typical AR patterns, MDR-Amp and Tet-SDR dominant in bovine- or turkey-population, were identified, indicating that distinct food-animal sources could be involved in human infections. The AR analysis suggested fluoroquinolones (i.e., ciprofloxacin), but not extended-spectrum cephalosporins (i.e., ceftriaxone, cefoxitin), is the adaptive choice for empirical therapy. Antibiotic-resistant S. Newport from humans has multiple origins, with distinct food-animal-borne route contributing to a significant proportion of heterogeneous isolates.