Abstract #1988

Geha R, Singer S, OConnor R, DeCarolis P, Wu Y, Chen J

Necrosis, the major cell death mechanism in tumors, is an important prognostic factor for cancer metastasis and recurrence. New methods for the quantitative analysis of necrosis are needed. A metabolomics study of necrosis in sarcomas from 17 patients was carried out using 1H hr-MAS NMR spectroscopy, supervised Support Vector Machine (SVM) learning, conventional pathologic analysis and Oil-Red-O staining. Compared to viable tumors, the necrotic tumors were characterized by a 1.5-fold increase in the observed lipid signals and a more than 90% loss of small metabolites and protein amide proton signals. SVM learning confirmed these observations and accurately discriminates necrotic from viable tumors. The substantial loss in the protein amide proton signals in necrotic tissue provides an explanation for the increased water relaxation times detected in necrotic compared to viable tissue.