Distinguishing between some benign lipomas (BLs), atypical lipomatous tumors (ALTs), and dedifferentiated liposarcomas (DDLs) can be challenging due to overlapping magnetic resonance imaging characteristics, and poorly understood molecular mechanisms underlying the malignant transformation of liposarcomas.
To identify metabolic biomarkers of the lipomatous tumor spectrum by examining human tissue specimens using high-resolution 1H magnetic resonance spectroscopy (MRS).
In this prospective study, human tissue specimens were obtained from participants who underwent surgical resection for radiologically-indeterminate lipomatous tumors between November 2016 and May 2019. Tissue specimens were obtained from normal subcutaneous fat (n=9), BLs (n=10), ALTs (n=7) and DDLs (n=8). Extracts from specimens were examined with high-resolution MRS at 17.6T. Computational modeling of pattern recognition-based cluster analysis was utilized to identify significant differences in metabolic signatures between the lipomatous tumor types.
Significant differences between BLs and ALTs were observed for multiple metabolites, including leucine, valine, branched chain amino acids, alanine, acetate, glutamine, and formate. DDLs were distinguished from ALTs by increased glucose and lactate, and increased phosphatidylcholine. Multivariate principal component analysis showed clear clustering identifying distinct metabolic signatures of the tissue types.
Metabolic signatures identified in 1H MR spectra of lipomatous tumors provide new insights into malignant progression and metabolic targeting. The metabolic patterns identified provide the foundation of developing noninvasive MRS or PET imaging biomarkers to distinguish between BLs, ALTs, and DDLs.