Skip to main content
. 2015 May 11;6(19):17261–17275. doi: 10.18632/oncotarget.3746

Figure 1.

Figure 1

A. Scheme, showing all types of tissues and methods used in this study. Ovarian tumor mass (P, primary tumor), peritoneal, tumor mass (M, metastasis) of both tumor spread types, miliary and non-miliary, ascites single cells (A) and ascites spheroids (S) were collected. Solid tumors were disaggregated. Ascites cells were separated by filtration into ascites single cells and spheroids, i.e. tumor cell aggregates. For RNA-sequencing, P, M, and A samples were enriched for EpCAM+ cells, while S samples were not further enriched. For flow cytometry, cells from disaggregated P and M tissues were depleted of CD45+ cells, S samples were disaggregated, and A samples were analyzed without further treatment. B. Principal Component Analysis (PCA) of all flow cytometric determined subpopulation frequencies (CD133+, CD44+, EpCAM+, and L1CAM+; n = 125; cf. Supplementary Tables S2) from ascites single cells, ascites spheroids, ovarian tumor masses, and peritoneal tumor masses. Each dot represents one patient. Light green indicates patients without peritoneal metastases, green, patients of the non-miliary tumor spread type, and blue, patients of the miliary tumor spread type. For statistical feasibility, only patients with available ascites samples were included. C. Boxplots showing the frequencies (in percent with reference to live cell counts) of CD44+/CD45− cells in ascites single cell and spheroid samples (upper graph) and of EpCAM+/CD45− cells in ovarian and peritoneal tumor masses (lower graph). In green, samples of non-miliary tumor spread (nM) and in blue samples of miliary tumor spread (M) are shown. P-values were calculated with Wilcoxon rank-sum tests. CD45+ cells, dead cells, and cell doublets and triplets were excluded from further analysis.