SPME-LC/MS-based serum metabolomic phenotyping for distinguishing ovarian cancer histologic subtypes: a pilot study

Abstract

AbstractEpithelial ovarian cancer (EOC) is the most common cause of death from gynecological cancer. The outcomes of EOC are complicated, as it is often diagnosed late and comprises several heterogenous subtypes. As such, upfront treatment can be highly challenging. Although many significant advances in EOC management have been made over the past several decades, further work must be done to develop early detection tools capable of distinguishing between the various EOC subtypes. In this paper, we present a sophisticated analytical pipeline based on solid-phase microextraction (SPME) and three orthogonal LC/MS acquisition modes that facilitates the comprehensive mapping of a wide range of analytes in serum samples from patients with EOC. PLS-DA multivariate analysis of the metabolomic data was able to provide clear discrimination between all four main EOC subtypes: serous, endometrioid, clear cell, and mucinous carcinomas. The prognostic performance of discriminative metabolites and lipids was confirmed via multivariate receiver operating characteristic (ROC) analysis (AUC value > 88% with 20 features). Further pathway analysis using the top 57 dysregulated metabolic features showed distinct differences in amino acid, lipid, and steroids metabolism among the four EOC subtypes. Thus, metabolomic profiling can serve as a powerful tool for complementing histology in classifying EOC subtypes.