‘Untargeted’ analytical workflow, Kendrick mass defect filtering and chemometric evaluation of dysregulated acylcarnitine patterns discriminating neoplastic vs. benign tissue samples among intermediate/high-risk localized prostate cancer patients

Background: Metabolomics is emerging as cancer-related metabolic dysregulations applications. Untargeted and subsequent Chemometric analysis of Acylcarnitines (ACs), essential for energy metabolism and fatty acid β-oxidation, could be pivotal in Prostate Cancer (PCa) research. ACs may serve as diagnostic biomarkers, prognostic tools, or targets for future therapies. Methods: Histologically-confirmed Benign Prostatic Hyperplasia (BPH) and PCa tissue samples from clinically significant (cs) localized intermediate/high-risk patients undergoing Robotic-assisted Radical Prostatectomy (2021–2023) were analyzed. ACs were characterized using high-resolution mass spectrometry, Kendrick mass defect filtering, and retention time prediction in reversed-phase chromatography. Data processing was performed with Compound Discoverer software, and a PLS-DA model was developed for annotated ACs. Adjusted regression modeling assessed the role of individual metabolites. Results: N=50 matched-pair csPCa and adjacent BPH tissue samples were analyzed. Using Compound Discoverer software for Kendrick mass defect filtering, ≥90% of initial features were removed. PLS-DA on n=74 annotated ACs achieved over 93% classification accuracy for both groups. LogP values for ACs ranged from -7.24 (malonylcarnitine) to 5.86 (hexacosylcarnitine). ACs with >25% SD in quality control samples were excluded, resulting in a final dataset of n=62 normalized Variance analysis highlighted ACs contributing significantly to tissue discrimination, including AC 5:0_1, AC 5:0_2, AC 3-OH 6:0_1, AC 3-OH 6:0_2, AC 3-OH 16:0, AC 18:2, AC 19:0, AC 20:4, AC 20:3, and AC 24:0. Short- and medium-chain ACs (C2–C12) showed negative values in non-oxidized forms, while oxidized ACs, such as malonylcarnitine and 3-hydroxybutyrylcarnitine, had positive values. Multivariate models highlighted two isomers of 3-hydroxyhexanoylcarnitine consistently higher in PCa samples. Pearson correlation heatmaps showed distinct ACs clusters. Three isomers of 3-hydroxybutyrylcarnitine, 3-hydroxyoctanoylcarnitine, 3-hydroxyheptanoylcarnitine, and malonylcarnitine also demonstrated significant correlations. Conclusion: PCa tissue is linked to oxidative changes in short- and medium-chain ACs, unlike longer-chain compounds, supporting their anti-angiogenic roles. Increased hydroxylated short-chain ACs and decreased non-hydroxylated forms in PCa suggest oxidative modifications are associated with neoplastic growth. These metabolic differences may aid early PCa detection through biological fluid analysis or targeting AC hydroxylation for preventive or therapeutic approaches.