UC Davis

The studies’ objectives were to identify cow-level and environmental factors associated with metritis cure in animals receiving ceftiofur treatment and metritis spontaneous cure in animals remaining untreated for metritis, to predict metritis cure using traditional statistics and machine learning algorithms. Additionally, to characterize the changes in the uterine metabolome in cows with metritis development and cure for cows treated with antimicrobials due to metritis. The dataset used for study one is from a previous study comparing the efficacy of different therapies and self-cure for metritis. Metritis was defined as fetid, watery, reddish-brownish discharge, with or without fever. Cure was defined as an absence of metritis signs 12 days after diagnosis. Cows were randomly allocated to receive ceftiofur treatment or (CEF, n = 275) remain untreated (CON, n = 275). For study two focusing on spontaneous metritis cure study, a subset of 438 primiparous and multiparous lactating Holstein cows having metritis and remaining untreated was used. Environmental and cow-related variables were offered for univariate logistic regression and selected for the multivariable logistic regression model according to their P-value (P < 0.05). In the first study, variables were offered to the model to assess their association with metritis cure, and in the second study to assess their association with spontaneous metritis cure. Additionally, machine learning algorithms analysis was performed in both studies. Multivariable logistic regression and ROC analysis indicate that cows developing metritis after the first week postpartum, having an increase in milk production starting on the day before metritis diagnosis day, and no fever at the time of metritis diagnosis have an increased odds of cure of metritis. Additionally, the machine learning algorithms achieved satisfactory accuracy for the prediction of metritis cure. For the metabolome study, vaginal discharge samples were collected from 86 cows within 6 hours after parturition, at 4 and 7 DIM, at metritis diagnosis, and at 4 and 7 days after metritis diagnosis. Cows with metritis (MET; n = 17) were paired with counterparts without metritis (NoMET) of similar DIM and parity (n = 49). Metabolomic data were analyzed using the MetaboAnalyst software. Untargeted GC-TOF-MS metabolomic analysis highlighted changes in the uterine metabolome in the first week postpartum in cows developing metritis compared to healthy animals. For the metritic group there are significant changes in the uterine metabolome associated to cure. In all scenarios, the metabolites lignoceric, malic, and maleic acids, ornithine, and hypotaurine, which are associated with arginine/aminoacyl-tRNA biosynthesis and taurine/purine metabolism, were upregulated in NoMET group and in cows curing from metritis. Also, cows not curing from metritis had significant changes in the uterus metabolome independent of receiving ceftiofur or remaining untreated. Improving spontaneous metritis cure prediction is an important way to contribute toward antimicrobial stewardship due to promoting selective treatment for metritis, reducing the use of antibiotics, potentially reducing the dissemination of antimicrobial resistance, reducing the cost of the disease, and improving animal welfare. Metabolome analysis may be an important tool to understand changes in the uterus during the postpartum and the dynamic of metritis development.