Chronic kidney disease (CKD) affects approximately 12% of the global population, posing a significant health threat. Inflammation plays a crucial role in the uremic phenotype of non-dialysis-dependent (NDD) stage 5 CKD, contributing to elevated cardiovascular and overall mortality in affected individuals. This study aimed to explore novel metabolic pathways in this population using semi-targeted metabolomics, which allowed us to quantify numerous metabolites with known identities before data acquisition through an in-house polar compound library. In a prospective observational design with 50 patients, blood samples collected before the initial hemodialysis session underwent liquid chromatography and high-resolution mass spectrometer analysis. Univariate (Mann–Whitney test) and multivariate (logistic regression with LASSO regularization) methods identified metabolomic variables associated with inflammation. Notably, adenosine-5′-phosphosulfate (APS), dimethylglycine, pyruvate, lactate, and 2-ketobutyric acid exhibited significant differences in the presence of inflammation. Cholic acid, homogentisic acid, and 2-phenylpropionic acid displayed opposing patterns. Multivariate analysis indicated increased inflammation risk with certain metabolites (N-Butyrylglycine, dimethylglycine, 2-Oxoisopentanoic acid, and pyruvate), while others (homogentisic acid, 2-Phenylpropionic acid, and 2-Methylglutaric acid) suggested decreased probability. These findings unveil potential inflammation-associated biomarkers related to defective mitochondrial fatty acid beta oxidation and branched-chain amino acid breakdown in NDD stage 5 CKD, shedding light on cellular energy production and offering insights for further clinical validation.