Preview

PDF, English Download (6MB) | Terms of use

Abstract

As two leading female cancers, breast cancer, especially metastatic breast cancer, and ovarian cancer, have brought an increasing health and economic burden globally. Biomarkers could improve patient outcomes and quality of life because they play vital roles in cancer screening, diagnosis, prognosis, and prediction. Metabolites are promising cancer biomarkers, as they represent the ultimate phenotypic alteration of the organism and are closely related to cancer. Plasma metabolites can be accessed with minimally invasive procedures. Using plasma metabolites as biomarkers for cancer and other diseases has been widely explored because of the possibility of repeated sampling and periodic monitoring of blood samples. However, metabolic studies are still in their infancy, and only a few studies with large sample sizes are available so far. In this thesis project, we explored the potential of metabolites as putative diagnostic and prognostic markers in breast and ovarian cancer. Plasma metabolite profiling and subsequent validation in primary breast cancer patients and healthy controls identified 18 metabolites that were significantly differentially represented (FDR < 0.05). Multivariate logistic regression analysis selected a panel of seven metabolites to discriminate primary breast cancer patients from healthy controls with an AUC of 0.80. If this panel of metabolites identified here could be verified in large prospective study cohorts, this panel, including Glu, Orn, Thr, Trp, Met-SO, C2, and C3, might add value to multi-molecular diagnostic marker sets for breast cancer early detection. The association of plasma metabolites with metastatic breast cancer was investigated as well. Metastatic breast cancer patients with high numbers of circulating tumor cells (termed CTC-positive) and those with low numbers or without CTCs (termed CTC-negative) were analyzed and compared to healthy controls as well as primary breast cancer patients. Lists of 19 and 12 metabolites were identified to significantly distinguish CTC-positive and CTC-negative samples from healthy controls, respectively. A panel comprising His, C4:0, C18:1, lysoPC a C18:2, PC aa C40:6, and PC ae C42:3 for CTC-positive patients with AUC = 0.92, and a combination of Asn, Glu, His, Thr, Trp, C16:0, C18:0, C18:1, C18:2, lysoPC a C18:2, and PC aa C40:6 for CTC-negative patients with AUC = 0.89 were selected to distinguish from healthy controls. Significantly different metabolites between CTC-positive/CTC-negative and primary breast cancer patients exhibited significant overlaps with those between CTC-positive/CTC-negative patients and healthy controls. We also investigated the prognostic value of metabolites in metastatic breast cancer patients. After successive analysis of the discovery and validation cohorts, four metabolites were found to be significantly negatively correlated with progression-free survival, while 12 metabolites were negatively correlated with overall survival. Amongst these metabolites associated with survival, LASSO Cox regression analysis selected a combination of PC ae C36:1 and PC ae C38:3 to predict progression-free survival, and a combination of lysoPC a C20:3, lysoPC a C20:4, PC aa C38:5, PC ae C38:3, and SM (OH) C22:2 to predict overall survival. Even though the proposed metabolic signatures showed a lower prognostic power than the CTC status, an FDA-approved prognostic marker, the combination of the Cox selected metabolites with the CTC status displayed a lower integrated prediction error than CTC status alone. Therefore, the identified metabolic markers might add prognostic value in combination with other biomarkers such as CTC status determination. The majority of the here identified metabolites have previously shown functional roles in cancer and metastasis development, thus laying a supposed mechanistic basis for their differential levels observed in plasma. Lastly, comparative profiling of plasma metabolites in ovarian cancer patients and healthy controls were applied to identify metabolites associated with ovarian cancer. Remarkably, 71 significantly differentially expressed metabolites were identified (FDR < 0.05). Most of them were down-regulated in ovarian cancer patients. A combination of seven metabolites, including His, Trp, C18:1, lysoPC a C18:2, PC aa C32:2, PC aa C34:4, PC ae C34:3, were identified to differentiate ovarian cancer cases from healthy controls with an AUC of 0.95. Furthermore, this panel could distinguish ovarian cancer from primary breast cancer patients with an AUC of 0.93. In conclusion, we identified specific signatures of plasma metabolites associated with primary breast cancer, metastatic breast cancer, and ovarian cancer. Further, we identified sets of metabolites correlated with the prognosis of metastatic breast cancer patients. If these identified metabolic marker signatures can be verified in large, multi-centric, prospective studies, they might add value to the development of blood-based diagnostic tests.