?Mass spectrometry-based metabolomics for renal cell carcinoma biomarker discovery?

MARÍA EUGENIA MONGE; MALENA MANZI; NICOLÁS ZABALEGUI; MARÍA ELENA KNOTT; MARTÍN PALAZZO; PATRICIO YANKILEVICH; PIERRE BEAUSEROY; MARÍA ISABEL GIMÉNEZ

Acapulco

Congreso; 8th Symposium of the Mexican Proteomics Society, 3rd PanAmerican-Human Proteome Organization (Pan-HUPO) Meeting, and 2nd Ibero-American Symposium on Mass Spectrometry; 2019

Sociedad Mexicana de Proteómica

Kidney cancer is accepted to be a metabolic disease. More than 50% of cases of renal cell carcinoma (RCC) are incidentally diagnosed; being clear cell RCC (ccRCC) the most common histological subtype (75% of cases), and considered a glycolytic and lipogenic tumor. Surgery is the most promising treatment for curation when the disease is detected at earlier stages since it is inherently resistant to chemotherapy and radiotherapy. Our research group has recently developed a protocol for harvesting and extracting extracellular metabolites from an in vitro model of human renal cell lines to profile the exometabolome by means of a discovery-based metabolomics approach using ultraperformance liquid chromatography coupled to quadrupole-time-of-flight mass spectrometry (UPLC-QTOF-MS) (1). The exometabolome from a ccRCC cell line (n=22) was differentiated from a non-tumor renal cell line (n=22) through a 21-metabolite panel with 100% specificity, sensitivity, and accuracy. In addition, 9 discriminant compounds detected in the exometabolome, including isoleucine/leucine, phenylalanine, N-lactoyl-leucine, and N-acetyl-phenylalanine, provided a metabolic footprint capable of differentiating human serum samples from stage IV ccRCC patients from controls (n=10). In a parallel study, serum samples from a cohort that included patients with different ccRCC stages (I, II, III, IV; n=112) and controls (n=52) were interrogated with a discovery-based lipidomics approach using UPLC-QTOF-MS and machine learning methods. Support vector machine models and the Lasso variable selection method yielded two discriminant panels for ccRCC detection and early diagnosis. A 18-feature panel allowed discriminating ccRCC patients from controls with 81% accuracy in an independent test set. A second model was trained to discriminate early stages (I-II) from late stages (III-IV) ccRCC and provided a 26-feature panel that allowed classification of stage I ccRCC patients from an independent test set with 82% accuracy. Current work involves lipid identification. Results are auspicious for early ccRCC diagnosis after validation of the panels in larger and different cohorts. References1 Knott M. E., Manzi M. et al (2018) J. Proteome Res. 17: 3877-88.This Project was supported by the Argentine National Agency of Scientific and Technological Promotion (PRH-PICT-2015-0022 project), and the National Research Council (CONICET) (PUE 055 project).