Differential diagnosis of benign and malignant ovarian tumors based on the blood metabolome

Background. The high mortality rate from ovarian cancer is largely due to the asymptomatic course of the disease. The signs of malignant and borderline ovarian tumors are similar to the manifestations of benign lesions, which determines the relevance of developing additional examination procedures and searching for new cancer markers that will distinguish benign and malignant processes. Aim. To build stable blood lipid panels for differentiation of healthy women, patients with benign (BOT) and malignant (MOT) ovarian tumors. Materials and methods. The search for markers for clustering of molecular profiles of blood samples of patients of the Kulakov National Medical Research Center for Obstetrics, Gynecology and Perinatology with BOT (cystadenoma – n=30, endometrioid cyst – n=56, teratoma – n=21), with MOT (borderline tumor – n=28, ovarian cancer of low malignancy – n=16, ovarian cancer of high malignancy – n=59) and volunteers of the group control (n=19) using discriminant analysis of orthogonal projections to hidden structures with an established threshold of importance of the variable VIP1 (OPLS) and the method of projections to hidden structures (PLS-PLS – it is a technology of multidimensional statistical analysis used to reduce the dimension of the feature space with minimal loss of useful information; VIP importance threshold 1) and other statistical tools. Samples’ molecular profile was complete by species, which were identificated by nuclear magnetic resonance and high-perfomance liquid chromatography-mass spectrometry. The analysis of the involvement of compounds that are potential markers of malignant processes in metabolic pathways was carried out. Results. Based on the OPLS and PLS methods, as a result of pairwise and multiclass comparisons, respectively, sets of lipids were identified that can be considered as markers of malignant and benign neoplasms. The overlap of the obtained panels with databases of metabolic pathways was studied, in particular, it was shown that all markers (except glucose) obtained by PLS for differentiation of healthy patients, patients with BOT or with MOT are involved in the transport of small molecules, glucose and lactate are involved in the “TCA Cycle” pathway “Nutrient Utilization and Invasiveness of Ovarian Cancer”. Triglycerides TG 16:0_16:0_18:1, TG 16:0_18:0_18:1, TG 16:0_18:1_18:1, TG 18:0_18:1_18:1, TG 18:0_18:1_18:2 and lactate are involved in the “HIF1A and PPARG regulation of glycolysis” pathway, and The HIF1A and PPARG genes are associated with the development of tumors. Metabolites CE 20:4, TG 16:0_16:0_18:1, TG 16:0_18:0_18:1, TG 16:0_18:1_18:1, TG 18:0_18:1_18:1, TG 18:0_18:1_18:2 are included in the pathways of energy metabolism, and LPC 16:0, PC 16:0_20:3, PC 16:0_20:4 is involved in the path of “Choline metabolism in cancer”. Graphs of the correlation interaction of markers that allow solving classification problems with an unambiguous interpretation of the results are constructed, which makes it possible to assert the prospects of using these panels for further creation of classification models. Conclusion. It is shown that lipids from the developed panels are involved in metabolic pathways associated with the development of tumor diseases and can be used for further validation of diagnostic models based on advanced machine learning methods. The introduction of the achievements of postgenomic research has the potential to increase the diagnostic value of the applied methods of differentiation of benign and malignant proliferative processes, as well as to supplement the available data on the processes of carcinogenesis in the ovaries. Thus, the analysis of the molecular profile of blood by mass spectrometry is a minimally invasive potentially effective diagnostic method.