HIGH-GRADE SEROUS OVARIAN CARCINOMA (HGSOC)

Abstract

The present invention relates to a method of determining the status of high-grade serous ovarian carcinoma (HGSOC) in a subject, the method comprising: providing a sample obtained from the subject; and detecting the presence of HGSOC biomarkers in the sample, wherein the method comprises detecting the presence of: a differentiated cell type; a KRT17 Cluster cell type; an epithelial-mesenchymal transition (EMT) cell type; a cell cycle cell type; and a ciliated cell type; wherein the level of the biomarkers is used to determine the fraction of EMT cells in the high-grade serous ovarian carcinoma in the subject. The invention further relates to associated kits, use and methods of treatment.

Claims

1. A method of determining the status of high-grade serous ovarian carcinoma (HGSOC) in a subject, the method comprising: providing a sample obtained from the subject; and detecting the presence of HGSOC biomarkers in the sample, wherein the method comprises detecting the presence of: a differentiated cell type by detecting one or more of differentiated cell biomarker proteins and/or nucleic acid encoding differentiated cell biomarker proteins selected from the group comprising LTBP4, PTGS1, SLC25A25, LAMC2, LRG1, DHCR24, PLK3 and LDLR; a KRT17 Cluster cell type by detecting one or more of KRT17 Cluster cell biomarker proteins and/or nucleic acid encoding KRT17 Cluster cell biomarker proteins selected from the group comprising SPP1, IL1B, IL1RN, KRT23, ALDH3B2, SUSD2, DEFB1, HLA-DQA2, CYP4B1, and PIGR; an epithelial-mesenchymal transition (EMT) cell type by detecting one or more of epithelial-mesenchymal transition (EMT) biomarker proteins and/or nucleic acid encoding EMT biomarker proteins selected from the group comprising SPARC, SERPINF1, DCN, SFRP4, CRISPLD2, TIMP3, CNN1, MYH11, MFAP4, ENG, EFEMP1, and RGS16; a cell cycle cell type by detecting one or more of cell cycle biomarker proteins and/or nucleic acid encoding cell cycle biomarker proteins selected from the group comprising FEN1, NUSAP1, UBE2C, ZWINT, PRC1, ASF1B, MCM4, GINS2, CENPM, MCM2, TK1, MCM6, SMC4, CENPU, and MAD2L1; and a ciliated cell type by detecting one or more of ciliated cell biomarker proteins and/or nucleic acid encoding ciliated cell biomarker proteins selected from the group comprising TEKT1, FAM92B, SNTN, LRRC46, EFCAB1, CDHR3, C6orf118, CCDC78, TUBA4B, C20orf85 and CAPSL; wherein the level of the biomarkers is used to determine the fraction of EMT cells in the high-grade serous ovarian carcinoma in the subject.

2. The method according to claim 1, wherein the fraction of EMT cells in the high-grade serous ovarian carcinoma in the subject is compared to a pre-determined threshold level to indicate if the high-grade serous ovarian carcinoma in the subject is an EMT subclass of high-grade serous ovarian carcinoma.

3. The method according to claim 1 or claim 2, wherein the level of the EMT biomarkers relative to the differentiated, KRT17 Cluster, cell cycle and ciliated biomarkers is indicative of the fraction of EMT cells, and the fraction of EMT cells above a pre-determined threshold level is indicative of an EMT subclass of high-grade serous ovarian carcinoma in the subject.

4. A method of detecting a panel of biomarkers in a sample of a subject, the method comprising: providing a sample obtained from a subject and detecting the presence of: one or more of differentiated cell biomarker proteins and/or nucleic acid encoding differentiated cell biomarker proteins selected from the group comprising LTBP4, PTGS1, SLC25A25, LAMC2, LRG1, DHCR24, PLK3 and LDLR; one or more of KRT17 Cluster cell biomarker proteins and/or nucleic acid encoding KRT17 Cluster cell biomarker proteins selected from the group comprising SPP1, IL1B, IL1RN, KRT23, ALDH3B2, SUSD2, DEFB1, HLA-DQA2, CYP4B1, and PIGR; one or more of EMT biomarker proteins and/or nucleic acid encoding EMT biomarker proteins selected from the group comprising SPARC, SERPINF1, DCN, SFRP4, CRISPLD2, TIMP3, CNN1, MYH11, MFAP4, ENG, EFEMP1, and RGS16; one or more of cell cycle biomarker proteins and/or nucleic acid encoding cell cycle biomarker proteins selected from the group comprising FEN1, NUSAP1, UBE2C, ZWINT, PRC1, ASF1B, MCM4, GINS2, CENPM, MCM2, TK1, MCM6, SMC4, CENPU, and MAD2L1; and one or more of ciliated cell biomarker proteins and/or nucleic acid encoding ciliated cell biomarker proteins selected from the group comprising TEKT1, FAM92B, SNTN, LRRC46, EFCAB1, CDHR3, C6orf118, CCDC78, TUBA4B, C20orf85 and CAPSL.

5. The method according to any preceding claim, wherein the nucleic acid encoding the biomarker comprises mRNA transcripts, or cDNA copies thereof, of the biomarkers.

6. The method according to any preceding claim, wherein detecting the level of a biomarker comprises the use of an oligonucleotide probe capable of binding to nucleic acid encoding the biomarker.

7. The method according to any preceding claim, wherein the method comprises determining the transcript level of the biomarkers.

8. The method according to any preceding claim, wherein the sample from the subject is ovarian cancer biopsy tissue.

9. The method according to any preceding claim, wherein all 52 of the biomarkers are detected.

10. A composition comprising a panel of probes, wherein the probes are for detecting: one or more of differentiated cell biomarker proteins and/or nucleic acid encoding differentiated cell biomarker proteins selected from the group comprising LTBP4, PTGS1, SLC25A25, LAMC2, LRG1, DHCR24, PLK3 and LDLR; one or more of KRT17 Cluster cell biomarker proteins and/or nucleic acid encoding KRT17 Cluster cell biomarker proteins selected from the group comprising SPP1, IL1B, IL1RN, KRT23, ALDH3B2, SUSD2, DEFB1, HLA-DQA2, CYP4B1, and PIGR; one or more of EMT biomarker proteins and/or nucleic acid encoding EMT biomarker proteins selected from the group comprising SPARC, SERPINF1, DCN, SFRP4, CRISPLD2, TIMP3, CNN1, MYH11, MFAP4, ENG, EFEMP1, and RGS16; one or more of cell cycle biomarker proteins and/or nucleic acid encoding cell cycle biomarker proteins selected from the group comprising FEN1, NUSAP1, UBE2C, ZWINT, PRC1, ASF1B, MCM4, GINS2, CENPM, MCM2, TK1, MCM6, SMC4, CENPU, and MAD2L1; and one or more of ciliated cell biomarker proteins and/or nucleic acid encoding ciliated cell biomarker proteins selected from the group comprising TEKT1, FAM92B, SNTN, LRRC46, EFCAB1, CDHR3, C6orf118, CCDC78, TUBA4B, C20orf85 and CAPSL.

11. A kit for determining the status of high-grade serous ovarian carcinoma (HGSOC) in a subject, the kit comprising a panel of probes, wherein the probes are for detecting: one or more of differentiated cell biomarker proteins and/or nucleic acid encoding differentiated cell biomarker proteins selected from the group comprising LTBP4, PTGS1, SLC25A25, LAMC2, LRG1, DHCR24, PLK3 and LDLR; one or more of KRT17 Cluster cell biomarker proteins and/or nucleic acid encoding KRT17 Cluster cell biomarker proteins selected from the group comprising SPP1, IL1B, IL1RN, KRT23, ALDH3B2, SUSD2, DEFB1, HLA-DQA2, CYP4B1, and PIGR; one or more of EMT biomarker proteins and/or nucleic acid encoding EMT biomarker proteins selected from the group comprising SPARC, SERPINF1, DCN, SFRP4, CRISPLD2, TIMP3, CNN1, MYH11, MFAP4, ENG, EFEMP1, and RGS16; one or more of cell cycle biomarker proteins and/or nucleic acid encoding cell cycle biomarker proteins selected from the group comprising FEN1, NUSAP1, UBE2C, ZWINT, PRC1, ASF1B, MCM4, GINS2, CENPM, MCM2, TK1, MCM6, SMC4, CENPU, and MAD2L1; and one or more of ciliated cell biomarker proteins and/or nucleic acid encoding ciliated cell biomarker proteins selected from the group comprising TEKT1, FAM92B, SNTN, LRRC46, EFCAB1, CDHR3, C6orf118, CCDC78, TUBA4B, C20orf85 and CAPSL.

12. The composition according claim 10, or the kit according to claim 11, wherein the panel of probes comprises probes for all 52 of the biomarkers.

13. A method of selecting a patient for treatment with an agent, agent combination, or composition for treatment or prevention of HGSOC, the method comprising determining the status of high-grade serous ovarian carcinoma (HGSOC) in a subject according to the method of any of claims 1-3 and 5-9, wherein the determination of an EMT subclass of HGSOC indicates that the subject should or should not receive the agent, agent combination, or composition.

14. A PI3K pathway inhibitor and/or immunotherapeutic agent for use in the treatment of high-grade serous ovarian carcinoma (HGSOC) in a subject, wherein the treatment comprises selecting the patient for treatment based on the determination of an EMT subclass of high-grade serous ovarian carcinoma in the subject according to the method of any of claims 1-3 and 5-9.

15. A method of treatment of high-grade serous ovarian carcinoma, wherein the subject is determined to have an EMT subclass of high-grade serous ovarian carcinoma according to the method of any of claims 1-3 and 5-9; wherein the method of treatment comprises administrating a PI3K pathway inhibitor and/or immunotherapeutic agent to the subject.

16. A method of treating a high-grade serous ovarian carcinoma in a subject with, the method comprising the steps of: receiving results of a biomarker assay of a tissue sample from the subject to determine if the patient has an EMT subclass of high-grade serous ovarian carcinoma; and if the subject has an EMT subclass of high-grade serous ovarian carcinoma, then administrating a PI3K pathway inhibitor and/or immunotherapeutic agent to the subject, wherein the biomarker assay is in accordance with the method of any of claims 1-3 and 5-9.

17. Use of a panel of biomarkers for determining the fraction of EMT cells present in a tissue sample from a subject with HGSOC, or for determining the status of a HGSOC in a subject, wherein the biomarkers comprise: one or more of differentiated cell biomarker proteins and/or nucleic acid encoding differentiated cell biomarker proteins selected from the group comprising LTBP4, PTGS1, SLC25A25, LAMC2, LRG1, DHCR24, PLK3 and LDLR; one or more of KRT17 Cluster cell biomarker proteins and/or nucleic acid encoding KRT17 Cluster cell biomarker proteins selected from the group comprising SPP1, IL1B, IL1RN, KRT23, ALDH3B2, SUSD2, DEFB1, HLA-DQA2, CYP4B1, and PIGR; one or more of EMT biomarker proteins and/or nucleic acid encoding EMT biomarker proteins selected from the group comprising SPARC, SERPINF1, DCN, SFRP4, CRISPLD2, TIMP3, CNN1, MYH11, MFAP4, ENG, EFEMP1, and RGS16; one or more of cell cycle biomarker proteins and/or nucleic acid encoding cell cycle biomarker proteins selected from the group comprising FEN1, NUSAP1, UBE2C, ZWINT, PRC1, ASF1B, MCM4, GINS2, CENPM, MCM2, TK1, MCM6, SMC4, CENPU, and MAD2L1; and one or more of ciliated cell biomarker proteins and/or nucleic acid encoding ciliated cell biomarker proteins selected from the group comprising TEKT1, FAM92B, SNTN, LRRC46, EFCAB1, CDHR3, C6orf118, CCDC78, TUBA4B, C20orf85 and CAPSL.

18. The use according to claim 17, wherein the use comprise the use of the composition of claim 10 or 12, or the kit of claim 11 or 12.

Description

[0312] Embodiments of the invention will now be described in more detail, by way of example only, with reference to the accompanying figures.

[0313] FIG. 1. Landscape of single-cell transcriptome in fallopian tube epithelium. [0314] (A) Diagram shows the single-cell RNA sequencing and analysis workflow. [0315] (B) Diagram of Clincluster shows three steps that help it overcome the confounded batch effects and interpatient variability (see Methods). [0316] (C) Uniform manifold approximation and projection (UMAP) plot profiles 3,800 single-cell transcriptome from fallopian tubes. The cells are colored by their patient sources and annotated with cluster names. [0317] (D) Principal component (PC) plot shows the cells under different condition (fresh, overnight-cultured or long-term cultured) colored by the value of imputed pseudotime values. The violin plot in the smaller panel on the right top shows the pseudotime distribution for each condition. LT, long-term cultured. Pseudotime analysis of secretory cells suggested a more deviated transcriptome after overnight culturing compared to the long-time cultured group. We compared primary secretory cells across three conditions, namely fresh dissociated, overnight cultured and long-time cultured (2-day or 6-day). The cells are plotted based on principal components 1 and 2 (PCI & PC2) and colored by their pseudotime values. Overnight cultured cells have higher pseudotime values than long-time cultured cells

[0318] FIG. 2. The immunofluorescent staining validated the existence of the KRT7/CK7 positive and CAPS positive “intermediate” cell type in human fallopian tube epithelium. [0319] (A) The CAPS positive cell (arrow) is also positive for TUBB4, a ciliated marker, validating that CAPS is a ciliated marker. [0320] (B-C) The intermediate cells (arrows) are both KRT7 and CAPS positive, while another CAPS+ ciliated cell in (B) is KRT7 negative. CK7, cytokeratin 7 (KRT7). [0321] (D) The KRT7 positive secretory cells are CAPS negative, while the KRT7 negative ciliated cells (arrow) are CAPS positive. [0322] (E) IF staining in human FTE organoid shows one KRT7+CAPS+ intermediate cell (arrow).

[0323] FIG. 3. Subtyping fallopian tube secretory cells. [0324] (A) Heatmap profiling the scaled expression of top marker genes of each cluster in fresh secretory cells. [0325] (B) IHC staining shows the existence and low proportion of Cell Cycle Cluster by its marker STMN1, in human FTE. [0326] (C) IHC staining shows the validation of the marker of ECM Cluster, RGS16, in human FTE. [0327] (D-E) IF staining of KRT17, secretory marker KRT7 (CK7) and HLA-DR in human FTE shows that KRT17 population is secretory cells (D) and has high expression of HLA-DR, an MHC II protein (E). [0328] (F) IF staining of KRT17 and epithelial marker E-cadherin in organoid from human FTE shows that this is a stable population that exists in vivo model. [0329] (G-I) The immunofluorescent staining indicated that the EPCAM+ CD44+ peg cells were positive for lymphocyte markers CD45 and CD3 by double staining of CD44 and CD3 (C), CD45 and EpCAM (D) and CD45 and CD3 (E). The intra-epithelium CD44+CD3+ cells in were also CD45+ and EpCAM+ (yellow arrows). We also observed extra-epithelium CD44+CD3+CD45+EpCAM− cells (red arrows) in the stromal region. It suggests that the basal CD44+ cells are likely positive for lymphocytes markers CD3 and CD45. [0330] (J-K) Immunofluorescent staining suggests that the basal cells are positive for EPCAM and two memory T cell markers, CD103 (J) and CD69 (K).

[0331] FIG. 4. Deconvolution of bulk expression matrix of HGSOC by FTE cell subtypes revealed a prognostic signature. [0332] (A)Diagram shows the 5-signature panel calculated based on FTE scRNA-seq data by BSEQ-sc. The columns in the heatmap (bottom panel) correspond to five cell subtypes in FTE (top panel). The heatmap stands for the strength of 53 marker genes (rows) in five signatures (columns). A heatmap showing deconvolution results of 308 tumors from the TCGA Ovarian Carcinoma study by Cibersort. The color of cell denotes the proportion (0-1) of five signatures (columns) across tumor samples (rows). [0333] (B) The violin plots showing that expression levels of three key EMT drivers, Twist (TWIST1 and TWIST2) and Snail (SNAI2), are significantly increased in EMT-high tumors (log-FC>1.8, FDR<2e-14). [0334] (C) Volcano plot showing that miRNAs are differentially expressed between EMT-high and -low tumors, including the miRNA-200 family (miR-200a, miR- 200b, miR-200c, miR-141, miR-429), which are the suppressors of EMT. The green and blue dots are significantly differentially expressed miRNAs (log-FC >0.5, FDR<0.05). The blue dots are the ones with text labels next to them. [0335] (D) Shows a diagram of the same 5-signature panel as (A) in a different format. [0336] (E) Shows the same 5-signature panel and heat map as (A) with genes listed according to a first panel of genes, [0337] (F) Shows the same 5-signature panel and heat map with a second panel of genes listed,

[0338] FIG. 5. Landscape of single-cell transcriptome in fallopian tube epithelium and quality control. [0339] (A)Heatmap shows the differentially expressed genes between fresh, overnight-cultured and LT-cultured groups that are enriched in the annotated gene ontology and pathway. [0340] (B) Violin plots show that represented genes in three pathways that are differentially expressed between fresh and overnight-cultured groups (FDR <0.05). [0341] (C) Violin plots show the expression of represented genes (LGR5, RSPO1 and WNT7A) in the Wnt signaling that is disturbed by the culturing condition (FDR<0.05). [0342] (D) Dot plots show that culture condition changes the proportion of cells that express genes related to Wnt pathway, including LGR5, RSPO1, WNT7A and HES6. [0343] (E) Violin plots show the disturbed expression of three genes (CD44, ESR1 and OVGP1) after overnight culturing. [0344] (F) Violin plots show that genes that are enriched in fatty acid processing are transiently switched off after overnight culturing. [0345] (G)Violin plots show that genes (STMN1, CCNA1 and TK1) that are enriched in cell cycle are significantly upregulated and expressed in most of cells after LT culturing. [0346] (H) Violin plots show the expression of marker genes of secretory cells in fresh cells (KRT7 and PAX8). [0347] (I) Violin plots show the expression of marker genes of ciliated cells in fresh cells (CCDC17, CCDC78 and CAPS). [0348] (J) IHC staining of ciliated cell marker CCDC17 [0349] (K) IHC staining of ciliated cell marker CAPS

[0350] FIG. 6. Intermediate cell subtype. [0351] (A)PCA plots show that the intermediate cell population (grey circle) has the expression of secretory markers (KRT7 and PAX8) and ciliated markers (CCDC17 and CAPS). [0352] (B) IF staining of both TUBB4 and CAPs shows that TUBB4 positive ciliated cells are also CAPS positive. It demonstrated that CAPS is a marker of ciliated cells. [0353] (C) IF staining of CK7 (KRT7) and CAPS shows that the majority of secretory cells are CK7 positive and CAPS negative, while the intermediate cells are double positive. [0354] (D)PCA plots show that the intermediate cell population (grey circle) has the expression of secretory markers (KRT7 and PAX8) and ciliated markers (CCDC17 and CAPS). [0355] (E) IF staining of both TUBB4 and CAPs shows that TUBB4 positive ciliated cells are also CAPS positive. It demonstrated that CAPS is a marker of ciliated cells. [0356] (F) IF staining of CK7 (KRT7) and CAPS shows that the majority of secretory cells are CK7 positive and CAPS negative, while the intermediate cells are double positive.

[0357] FIG. 7. Novel secretory subtypes and their molecular features. [0358] (A) Quality control of the single FTE cells by the copy number variant referred from expression data. [0359] (B) Violin plots show the upregulation of representative marker genes for Cell cycle cluster (C10), which are enriched in Cell cycle, DNA repair and Chromatin remodeling pathways. [0360] (C) Violin plots show the upregulation of representative marker genes for KRT17 cluster (C4), which involve MHC II, cytokeratin, aldehyde dehydrogenases and p21 (CDKN1A).

[0361] FIG. 8. Deconvolution of bulk expression data of HGSOC. [0362] (A) Violin plot shows that the ciliated signature was enriched in the Grade 1 tumors compared to Grade 2-3 tumors in the AOCS (Australian Ovarian Cancer Study) dataset. [0363] (B) Violin plots show that 6 out of 12 EMT markers are negatively expressed or downregulated in LCM stroma samples compared to LCM tumour samples.

[0364] FIG. 9. Validation of the secretory subtypes in the FTE of benign donors by using scRNA-seq. [0365] (A) UMAP shows the populations in the FT samples from 5 benign patients. Each dot is a cell colored by its donor. [0366] (B) UMAP plots show the populations in the FT samples from cancer patients (n =5) and benign patients (n=5). The left, middle and right subpanels contain the cells from all 10 patients, 5 cancer patients and 5 benign patients respectively. [0367] (C) UMAP plot shows the populations in the FT samples from 5 benign patients. Each dot represents a secretory cell from a benign patient. The dots are colored by their donors as shown in the legend. [0368] (D) Scatter plots show the transcriptomic characteristics of each subtype in benign and cancer patients. Cells (dots) are colored by the score of each transcriptomic signature (subtitle). The score of a transcriptomic signature was computed by the scaled and centered sum of expression levels of the marker genes in each transcriptomic signature. The scores correspond to the expression of marker genes of each cluster. The transcriptomic signatures are listed in Table S7.

[0369] FIG. 10, related to FIG. 3. Validation of the secretory subtypes in the FTE of benign donors by using scRNA-seq. [0370] (A) Flowchart shows the processing of the validation set, in which we profiled 2185 single-cell transcriptomes from five benign patients. After the initial filtering, 1875 cells were left as shown in FIG. 3A. By using the data integration, we removed the batch effects between the discovery set and the validation set and then merged the two datasets. We next clustered the merged datasets to identify the four secretory subtypes in the FTE secretory cells from benign patients. [0371] (B) Scatter plots show the expression of marker genes in the FT cells from five benign patients. The x- and y-axes represent the first two components of the UMAP analysis. Each dot (cell) is colored by the expression level of the marker gene (subtitle). The result shows the CD45+ leukocytes, COL1A1+ stromal cells, KRT7+ PAX8+ EPCAM+ secretory cells and CAPS+ EPCAM+ ciliated cells. [0372] (C) IHC shows the STMN1 positive cell cycle subtype in the FTE of a benign patient. [0373] (D) IF staining shows a KRT17 and EPCAM double positive cell (KRT17 subtype) in the FTE of a benign patient. [0374] (E) IHC images show the SPARC and PAX8 double positive cells in the FTE in the FTE of multiple benign patients (arrows and dashed circles).

[0375] FIG. 11. A—The EMT-high tumours have significantly higher proportions of macrophages M2 compared to EMT-low tumours. p-value=4.093e-05 by one-sided Welch t-test. The y-axis shows the proportion of Macrophages M2. Each dot is a tumour sample from TCGA. B—The EMT-high tumours have significantly higher proportions of monocytes compared to EMT-low tumours. p-value=8.021e-12 by one-sided Welch t-test. The y-axis shows the proportion of Monocytes. Each dot is a tumour sample from TCGA. C—The EMT scores and the expression levels of most macrophage markers are positively correlated. The x-axis is the markers of macrophages M2. The y-axis is the Person correlation coefficient between the EMT scores and the expression levels of marker genes.

EXAMPLE 1

[0376] Introduction

[0377] Limited understanding regarding of FTE has hindered further investigation into HGSOC; therefore, cellular subtypes in FTE need to be thoroughly studied at the transcriptomic level. Herein, we profiled the fallopian tube epithelium from patients with HGSOC or endometrium cancer to delineate subtypes in FTE secretory cells and their marker genes. These markers from FTE single cells were then used to stratify HSGOCs and identified a tumor subtype with poor overall survival.

[0378] Results

[0379] A Cell census of human fallopian tubes in cancer patients We analyzed 3,877 single cells from the fallopian tubes of six ovarian cancer patients and four endometrial cancer patients using Smart-Seq2 technique (Picelli et al., 2014) (FIG. 1A and Table 2). Flowcytometry was used to identify and sort single epithelial cells (EPCAM+, CD45−), leucocytes (EPCAM-, CD45+) and stromal cells (EPCAM−, CD45−) prior to sequencing. To overcome the confounding of batch effects and patient-specific variability in clinical samples, we used differential-expression-based clustering. This clustering approach is based on a functional similarity assumption, in which the differential expression (DE) patterns across batches are similar but distinguishable across cell populations with miscellaneous functions (FIG. 1B). Using this approach, we were able to differentiate between epithelial and non-epithelial cells (FIGS. 1C).

[0380] However, we observed striking effects of the culture conditions on the single cell transcriptomes. Most notably, overnight culturing, induced profound differential expression changes in pathways related to cell cycle (e.g. RGCC, p21 and MCM4), RNA processing (e.g. POLR2B, PRPF3 and METTL3) and stress response (e.g. NR4A1, FOS and EGR1) (FIGS. 5A and B). In addition, the Wnt signaling pathway was also significantly affected, where LGR5 and RSPO1 were downregulated and WNT7A was upregulated after culturing (FIGS. 5C and D). Importantly, overnight culture induced the expression of genes that are known to be rarely expressed in FTE cells such as CD44 (log.sub.e fold-change [log-FC]=3.8) (Paik et al., 2012) and reduced the expression of key markers of secretory cells such as Estrogen Receptor alpha (ESR1) and Oviductal Glycoprotein 1 (OVGP1) (FIG. 5E) (Cerny et al., 2016; Wu et al., 2016). Cilium organization was also downregulated in the overnight-cultured ciliated cells. Moreover, Pseudotime analysis (Campbell and Yau, 2018) across three conditions from the same patients revealed that the transcriptomes of fresh cells were more similar to the long-term (LT) cultured cells compared to the overnight group of cells (FIG. 1D). For instance, fatty acid metabolic process (e.g. BDH2, ALKBH7 and PTGR1) was provisionally downregulated after overnight culturing and then upregulated in the LT group, while RNA processing pathway was upregulated (FIGS. 5A and 5F). This suggested that including the overnight-cultured cells in subsequent analysis may introduce significant biases that would preclude meaningful conclusions. Similarly, although the LT group resembled the fresh group of cells, they also showed a unique split into two subgroups and disturbed expression of Stathmin and cell cycle genes that probably represent an artefact of long-term culture (FIG. 5G). To avoid the substantial effects from preservation methods, we focused our analysis on fresh cells only.

[0381] Within the epithelial cells, we identified the two previously established subtypes, secretory and ciliated cells (FIG. 1C). Secretory cells were characterized by the expression of PAX8 and KRT7 (FIG. 5H) as well as a large number of newly identified markers of secretory cells. The ciliated population was represented by high expression of the FOXJ1 and members of the coiled-coil domain containing protein family, such as CCDC17 and CCDC78 (FIGS. 51 and J). This protein family is essential for cilia functioning (Klos Dehring et al., 2013). We also identified a list of previously unrecognized markers of fallopian tube ciliated cells such as the calcium binding protein Calcyphosin (CAPS) that were enriched in the cilium-related pathways (FIGS. 5I, K and 2A) (Wang et al., 2002).

[0382] In addition to the two established cell types, we discovered a rare intermediate type that was characterized by the expression of the secretory cell marker KRT7 and high expression of ciliated marker CAPS (FIGS. 2B, C and 6A), whilst other KRT7+ secretory cells were CAPS negative and validated this population in human FTE (FIGS. 2D and 6B). PAX8 was expressed in a subset of this population, possibly because that its moderate expression level caused a higher dropout rate. Additionally, this subtype was enriched in overnight cultured cells and recapitulated in organoids cultured from human FTE (FIG. 2E). However, due to the low proportion of this intermediate population, it is challenging to conduct DE analysis to identify specific markers for this population. This intermediate population might represent an intermediate cell state between secretory and ciliated cells, which accords with the previously assumed transition between these two cell types (Ghosh et al., 2017; Hellner et al., 2016).

[0383] Four Novel Secretory Subtypes in FTE

[0384] We next attempted to classify secretory cells based on their transcriptomes. To ensure the purity of secretory cells, the cell was only kept for further analysis if it had strong expression of KRT7 and EPCAM and no expression of CCDC17 or PTPRC. In addition, to avoid including contaminating cancer cells, we excluded cells that had detectable copy numbers variants or loss-of-heterogeneity (FIGS. 7A) (Fan et al., 2018). By applying the DE-based clustering on fresh secretory cells, we found that they were clustered into nine clusters (FIGS. 3A). Except for a patient-specific cluster (C8) that was enriched in inflammatory markers, all other clusters contained cells from multiple patients. Three out of nine clusters (C1, C2 and C5) had no particular distinguishing features suggesting that they probably represented a quiescent population of cells. Cluster C6 had evidence of cell stress as shown by high expression of early response genes such as FOS and JUN (Honkaniemi et al., 1992).

[0385] Surprisingly, C7 showed high expression of a Regulator of G protein signaling (RGS16) and genes that were enriched in the extracellular matrix (ECM) pathway (false discovery rate [FDR]=1.80E-17), such as TIMP3, SPARC and COL1A (FIGS. 3B and C). We validated the presence of this subtype using IHC (FIGS. 3B). This cell type may be generated by the epithelial-mesenchymal transition (EMT), which can be induced by the chronic exposure to the oxidative stress (Mahalingaiah et al., 2015) and might be related to cancer development (Hanahan and Weinberg, 2011). This cell type is not a contamination from FT mesenchymal cells because it strongly expressed KRT7 and EPCAM (FIGS. 7D).

[0386] Cluster C3 had upregulation of genes that are involved in RNA synthesis and transport (e.g. PTBP1, ZNF259 and PRPF38A). It probably represented a transient differentiating cell population. Cluster 4 is characterized by the upregulation of major histocompatibility complex (MHC) Class II genes (e.g. HLA-DQA1, HLA-DPA1 and HLA-DPB1), cytokeratins (KRT17 and KRT23), aldehyde dehydrogenases (e.g. ALDH1A1 and ALDH3B2) and CDKN1A (also called p21) (FIGS. 3A and 7D). KRT17 was reported to be expressed in around 5% of FTE cells (Comer et al., 1998), but the fact that it was enriched in MHC class II expression was unknown. Importantly, this KRT17 positive cluster was validated in human FTE using IF and recapitulated in organoid cultures grown from human fallopian tube epithelial cells suggesting that they represent a robust group of cells with potentially important biological functions (FIGS. 3G-I).

[0387] C9 cluster (−1.6% of fresh FTESCs) most probably represented cycling cells because the marker genes of this cluster were enriched in three pathways, namely cell cycle (e.g. MCM2-7, MKI67, TK1 and STMN1), DNA repair (e.g. FANCD2, FANCI and MSH2) and chromatin remodeling (e.g. HMGB2 and SMC1A) (FIGS. 3A, B and 7B). MKI67 (also known as Ki-67) is a well-known marker for proliferation in FTE and other cells (Kuhn et al., 2012). The two Fanconi Anemia genes, FANCD2 and FANCI, can form a heterodimer that is essential for DNA repair and can interact with MCM2-7 (Nalepa and Clapp, 2018). The relatively low percentage of cycling cells is consistent with the age of the patients from whom the cells were obtained.

[0388] We also confirmed the CD45+EPCAM+ population that was located as basal cells in FTE by IF staining (FIG. 3H). This population was also positive for CD3, CD44, CD69 and CD103 (FIGS. 3G and I), suggesting that they are likely tissue resident T lymphocytes.

[0389] Deconvolution Revealed a Poor-Prognostic Tumor Subtype

[0390] We hypothesized that FTE cell subtypes might be correlated with HSGOC tumor types. Based on the four novel secretory subclasses and the ciliated cell type, we firstly computed a reference matrix with cell-type derived transcriptomic signature from five major FTE cellular subtypes (Cell cycle, EMT, Differentiated, KRT17 cluster and ciliated) as previously described (FIG. 4A) (Baron et al., 2016). The resulting signature matrix was then used for the deconvolution analysis (Newman et al., 2015) on bulk ovarian cancer RNA-seq data from The Cancer Genome Atlas (TCGA) and the microarray date from AOCS study (Bell et al., 2011; Tothill et al., 2008) to generate the fractions of five subtypes within each tumor. Whereby, we found a dispersed proportion of composition across tumors for these five cell types (FIG. 4A). Over 75% (233/308) tumors from TCGA were dominated by one cellular subtype (fraction >0.5), while the rest tumors have main components from multiple subtypes. For example, the ciliated tumor subtype was enriched in the Grade 1 tumors compared to Grade 2-3 in the AOCS dataset (p<1e-06, one-sided Wilcox test, FIG. 8A), suggesting that the grade of serous ovarian carcinoma may be related to their ability to differentiate. Most notably, we were able to identify a class of EMT-enriched tumors in multiple data sets. These tumors were enriched in genes that were previously linked to the “mesenchymal” HGSOC subtype. We found that the marker genes (FDR<0.05, log-FC>1) of these tumors were enriched in focal adhesion and PI3K-Akt signaling pathway (FDR<0.0002, by DAVID), which are critical for tumor cell survival (Fresno Vara et al., 2004; McLean et al., 2005). Furthermore, three key EMT genes, TWIST1, TWIST2 and SNAI2 (Ansieau et al., 2008; Kang and Massagué, 2004; J. Yang et al., 2004), were upregulated in the EMT-high tumors (FIG. 4B), suggesting that EMT may be the underlying mechanism of this tumor subtype. To test whether the EMT-high tumor subtype was merely caused by stromal cell impurities in tumor samples, we performed RNA sequencing on 36 laser capture microdissected (LCM) tumors from samples collected from 15 patients and classified tumors based on the deconvolution analysis. We compared the expression of genes in EMT signature between laser capture microdissected tumor and stroma samples. As expected, expression levels of PAX8 and EPCAM were significantly higher in tumor samples compared to stromal samples in which these markers showed almost no expression (FIG. 8B). In contrast, EMT-high markers (SPARC, TIMP3 and MFAP4) were highly expressed in both tumor and stroma confirming that EMT-high tumors truly express these genes. Furthermore, we also verified that the EMT-high tumors were not the by-product of ploidy or copy number aberrations of the involved highly expressed genes.

[0391] We next tested whether any of the five tumor subtype scores from the deconvolution analysis correlated with survival. The EMT score was significantly associated with poor overall survival and was independent of the effect of ages, stages and residual diseases (p<0.05, by Cox proportional hazard model). The robustness of the association was confirmed by the permutation test (n=500) leaving out 10% samples each time (empirical p-values=0.012 [TCGA] and 0 [AOCS], permutation test).

[0392] SPARC, one of the 12 genes that comprise the EMT signature, was previously described in the mesenchymal subtype of HGSOC (Tothill et al., 2008), while some other markers were reported to be related to EMT in ovarian cancer or other cancers, such as SFRP4 (Ford et al., 2013), TIMP3 (Anastassiou et al., 2011), MYH11 (Y.-R. Li and W.-X. Yang, 2016) and EFEMP1 (Yin et al., 2016). Nevertheless, the link between this tumor types and a particular FTE cellular subtype was previously unrevealed. The mesenchymal subtype was previously thought to have an association with poor prognosis, but the reproducibility of the observation was inconsistent probably because of the difficulty in defining this group of tumors. Using the EMT scores from deconvolution, we reached a robust classification with consistently significant correlation with poor survival (p<0.03) in another seven independent datasets, including the AOCS dataset (Tothill et al., 2008) and six additional microarray datasets (N>100) from the CuratedOvarianData database (Ganzfried et al., 2013) (Table 3).

[0393] A DE analysis of TCGA miRNA data revealed that the miRNA-200 family (miR-200a, miR-200b, miR-200c, miR-141 and miR-429) was downregulated in EMT-high tumors (FDR<0.01, log-FC<−0.5), which agrees with the previous finding that this miRNA family suppresses EMT process and that its loss can activate EMT in invasive breast cancer cell lines with a mesenchymal phenotype (Gregory et al., 2008). We also found that miRNA-483 and miRNA-214 were significantly upregulated in EMT-high tumors, while miRNA-513c, miRNA-509 and miRNA-514 were downregulated (FIG. 4C). Although previous studies suggested that miRNA-483 and miRNA-214 play an important role in cancer progression (Chandrasekaran et al., 2016; Liu et al., 2013), their connection with EMT or ECM has not been fully studied.

TABLE-US-00003 TABLE 2 Patient information CELLS FOR SAMPLE.ID PATIENT.ID SOURCE AGE DIAGNOSIS TUBE STATUS ANALYSIS 11511L&R 11511 cryopreservation & 69 Endometrial cancer 63 overnight cultured 11519L&R 11519 cryopreservation & 50 HGSOC 610 overnight cultured 11528L 11528 cryopreservation & 56 HGSOC 258 overnight cultured 11529L 11529 cryopreservation & 78 HGSOC 146 overnight cultured 15062L&R 15062 cryopreservation & 73 Endometrial cancer 100 overnight cultured 11543L&R 11543 fresh 73 Advanced L: Normal; 225 ovarian cancer R: STIC 11545L&R 11545 fresh 66 Primary Normal 518 peritoneal cancer 15066L&R 15066 fresh 52 High-grade Normal 606 endometrial cancer 11553L&R 11553 fresh 77 HGSOC L: Normal; 464 R: mucosal carcinoma 15072L&R 15072 fresh 62 squamous cell 319 carcinoma of endometrium 11553-LT 11553 long-term cultured \ \ \ 26 15072-LT 15072 long-term cultured \ \ \ 135 11553-ON 11553 overnight cultured \ \ \ 229 15072-ON 15072 overnight cultured \ \ \ 178 Note: L—left tube R—right tube LT—long-term ON—overnight

TABLE-US-00004 TABLE 3 AOCS dataset and seven microarray datasets (N > 100) from the CuratedOvarianData database GEO HAZARD LOWER UPPER DATABASE SAMPLES.sup.ζ EVENTS RATIO P-VALUE CI95 CI95 CITATION E.MTAB.386 129 73 3.31 0.0120 1.30 8.41 (Bentink et al., 2012) GSE49997 194 57 3.08 0.0038 1.44 6.60 (Pils et al., 2012) GSE13876 157 113 3.03 0.0082 1.33 6.87 (Crijns et al., 2009) GSE26712 185 129 2.60 0.0051 1.33 5.09 (Bonome et al., 2008) GSE26193 107 76 2.42 0.0286 1.10 5.35 (Mateescu et al., 2011) GSE51088 152 112 1.95 0.0121 1.16 3.30 (Karlan et al., 2014) GSE32062.GPL6480 260 121 1.88 0.0556* 0.98 3.58 (Yoshihara et al., 2012) AOCS 253 106 2.69 0.0004 1.56 4.66 (Tothill (GRADE (H) et al., 2008) .sup.ζValidation survival analysis was restricted to eight microarray datasets with over 100 samples. *Except for GSE32062.GPL6480, the EMT scores and overall survival are negatively correlated in other seven datasets (P < 0.05). The hazard ratios of EMT scores in all eight datasets are larger than 1 (range: 1.88-3.31).

EXAMPLE 2

[0394] To exclude the potential paracrine effect of cancer cells on non-cancer FTE cells, we validated the existence of the four secretory subtypes in the FTE cells obtained from benign (non-cancer) donors. We first analyzed 1857 single-cell transcriptomes of fallopian tubes from five patients with benign conditions (FIGS. 9A, 10A-B). Next we integrated the fresh secretory cells from the benign patients with the annotated ones from cancer patients by computing batch-correcting anchors (Stuart et al., 2019. Cell 177, 1888-1902.e21). Clustering of the integrated data illustrated that the four secretory subtypes also existed in the FTE of non-cancer donors (FIGS. 9B-D). Further validation using immunofluorescence (IF) and immunohistochemistry (IHC) in FT samples from benign donors confirmed the above results (FIGS. 10C-E). Overall, these results demonstrate that the new secretory subtypes were not artefacts caused by the influence of nearby cancer cells or by systemic effects of cancer burden.

Example 3

[0395] According to the invention herein, a first panel of cell-signature markers was identified, as provided in Table 3 below. After further analysis, where the threshold for selecting the marker genes was adjusted, a second panel of cell-signature markers was identified, as provided in Table 4 below. Whilst both panels prove useful for identifying the cell-signatures, the second panel generated more significant (p<0.05) and reproducible results across multiple datasets.

TABLE-US-00005 TABLE 3 First Panel. Entrez_ KRT17 ID HGNC_symbol Signature gene_id Ensembl_gene_id Differentiated cluster EMT Cell cycle Ciliated  1 LTBP4 Differentiated 8425 ENSG00000090006 249.443114 57.3172043 112.25 103.173913 21.7034483  2 PTGS1 Differentiated 5742 ENSG00000095303 433.125749 238.209677 307.325 203.043478 20.7448276  3 SLC25A25 Differentiated 114789 ENSG00000148339 320.88024 236.344086 187.625 192.956522 114.358621  4 LAMC2 Differentiated 3918 ENSG00000058085 694.94012 451.069892 329.35 339.782609 439.137931  5 LRG1 Differentiated 116844 ENSG00000171236 358.568862 191.586022 203.225 140.173913 322.524138  6 DHCR24 Differentiated 1718 ENSG00000116133 677.347305 524.88172 319.975 337.130435 92.9241379  7 LDLR Differentiated 3949 ENSG00000130164 383.580838 245.55914 226.65 390.956522 35.0758621  8 SPP1 KRT17 Cluster 6696 ENSG00000118785 0.04191617 203.94086 0.05 10.9130435 5.56551724  9 IL1B KRT17 Cluster 3553 ENSG00000125538 70.3173653 254.069892 33 208.347826 7.52413793 10 IL1RN KRT17 Cluster 3557 ENSG00000136689 42.2335329 128.655914 21.325 59.826087 1.42068966 11 KRT23 KRT17 Cluster 25984 ENSG00000108244 21.9341317 120.349462 19.75 31.3913043 112.786207 12 ALDH3B2 KRT17 Cluster 222 ENSG00000132746 15.0718563 98.6935484 14 15.2608696 19.8827586 13 SUSD2 KRT17 Cluster 56241 ENSG00000099994 22.239521 168.704301 12.8 3.69565217 3.44827586 14 DEFB1 KRT17 Cluster 1672 ENSG00000164825 10.1916168 67.0752688 5.65 34.0434783 1.82068966 15 HLA-DQA2 KRT17 Cluster 3118 ENSG00000237541 7.46107784 61.4677419 20.5 10.7826087 2.06206897 16 CYP4B1 KRT17 Cluster 1580 ENSG00000142973 75.5868263 180.758065 57.325 39.6521739 472.606897 17 PIGR KRT17 Cluster 5284 ENSG00000162896 8.92814371 197.983871 23.2 27.6521739 40.0965517 18 SPARC EMT 6678 ENSG00000113140 12.1497006 12.2634409 234.9 20.7826087 2.11034483 19 SERPINF1 EMT 5176 ENSG00000132386 0.91017964 9.05376344 128.825 0 16.537931 20 DCN EMT 1634 ENSG00000011465 1.19760479 7.29569892 509.575 1.91304348 1.36551724 21 SFRP4 EMT 6424 ENSG00000106483 2.41916168 6.68817204 688.275 0 1.84827586 22 CRISPLD2 EMT 83716 ENSG00000103196 9.04191617 18.9731183 259.925 7.86956522 0.67586207 23 TIMP3 EMT 7078 ENSG00000100234 3.49700599 4.02688172 444.3 7.34782609 0.53793103 24 CNN1 EMT 1264 ENSG00000130176 10.0479042 0.49462366 116.625 6.13043478 10.5517241 25 MYH11 EMT 4629 ENSG00000133392 1.05389222 2.10215054 266.65 3.13043478 7.32413793 26 MFAP4 EMT 4239 ENSG00000166482 1.44311377 0.07526882 344.85 0 0.00689655 27 ENG EMT 2022 ENSG00000106991 3.8502994 3.73655914 103 0.39130435 3.07586207 28 EFEMP1 EMT 2202 ENSG00000115380 39.6706587 138.83871 273.275 13.0434783 119.427586 29 RGS16 EMT 6004 ENSG00000143333 2.31137725 2.70967742 259.075 25.173913 95.9931034 30 FEN1 Cell cycle 2237 ENSG00000168496 13.0239521 12.3387097 24.9 87.8695652 13.2344828 31 NUSAP1 Cell cycle 51203 ENSG00000137804 7.43712575 2.2688172 8.575 135.391304 1.82758621 32 UBE2C Cell cycle 11065 ENSG00000175063 0.02994012 0.24731183 2.725 203.043478 1.4137931 33 ZWINT Cell cycle 11130 ENSG00000122952 9.18562874 10.0483871 4.85 190 3.42758621 34 PRC1 Cell cycle 9055 ENSG00000198901 10.742515 5.19892473 15.675 122.913043 5.66206897 35 ASF1B Cell cycle 55723 ENSG00000105011 0.58083832 0 0.2 146.173913 3.84137931 36 MCM4 Cell cycle 4173 ENSG00000104738 42.9161677 28.0806452 65.225 209.608696 24.6275862 37 GINS2 Cell cycle 51659 ENSG00000131153 6.88622754 1.12365591 8.15 74.1304348 1.03448276 38 CENPM Cell cycle 79019 ENSG00000100162 1.19760479 1.44623656 24.575 77.5652174 80.6551724 39 MCM2 Cell cycle 4171 ENSG00000073111 34.9101796 11.9677419 39.325 92.3478261 33.6275862 40 TK1 Cell cycle 7083 ENSG00000167900 3.45508982 7.1344086 3.85 351.913043 32.7655172 41 MCM6 Cell cycle 4175 ENSG00000076003 14.7065868 8.90322581 53.025 131.26087 2.66896552 42 SMC4 Cell cycle 10051 ENSG00000113810 14.1616766 4.79569892 14.25 71.9565217 15.9931034 43 CENPU (MLF1IP) Cell cycle 79682 ENSG00000151725 1.75449102 1.30645161 8.725 71.173913 4.19310345 44 MAD2L1 Cell cycle 4085 ENSG00000164109 8.0239521 4.31182796 3 87.173913 37.0344828 45 TEKT1 Ciliated 83659 ENSG00000167858 5.5988024 0.8655914 0.3 0.82608696 800 46 FAM92B Ciliated 339145 ENSG00000153789 0.02994012 0.05913978 0 0 589.331034 47 SNTN Ciliated 132203 ENSG00000188817 1.03592814 4.53225806 0 2.47826087 800 48 LRRC46 Ciliated 90506 ENSG00000141294 2.08982036 3.40860215 5.075 1.17391304 501.317241 49 EFCAB1 Ciliated 79645 ENSG00000034239 0.98203593 0.79032258 0 0.04347826 611.6 50 CDHR3 Ciliated 222256 ENSG00000128536 1.51497006 2.12365591 0 0 659.655172 51 C6orf118 Ciliated 168090 ENSG00000112539 17.7125749 2.11827957 0 14.3913043 493.765517 52 CCDC78 Ciliated 124093 ENSG00000162004 0.04790419 0.19354839 0 0 702.027586 The table lists 52 marker genes. The HGNC gene symbol is listed in the second column, the Entrez gene ID in the fourth column and the Ensembl gene ID in the fifth column. The third column describes which signature the gene belongs to. The sixth to tenth columns show the scaled expression levels of each gene in a certain cell state signature. The numbers are used in the deconvolution step to calculate the cell state proportions.

TABLE-US-00006 TABLE 4 Second Panel. Entrez_ KRT17 ID HGNC_symbol Signature gene_id Ensembl_gene_id Differentiated cluster EMT Cell cycle Ciliated  1 LTBP4 Differentiated 8425 ENSG00000090006 249.443114 57.3172043 112.25 103.173913 21.7034483  2 SLC25A25 Differentiated 114789 ENSG00000148339 320.88024 236.344086 187.625 192.956522 114.358621  3 LAMC2 Differentiated 3918 ENSG00000058085 694.94012 451.069893 329.35 339.782609 439.137931  4 DHCR24 Differentiated 1718 ENSG00000116133 677.347305 524.88172 319.975 337.130435 92.9241379  5 PLK3 Differentiated 1263 ENSG00000173846 151.700599 111.655914 78.725 106.608696 65.6275862  6 LRG1 Differentiated 116844 ENSG00000171236 358.568862 191.586022 203.225 140.173913 322.524138  7 LDLR Differentiated 3949 ENSG00000130164 383.580838 245.55914 226.65 390.956522 35.0758621  8 SPP1 KRT17 Cluster 6696 ENSG00000118785 0.04191617 203.94086 0.05 10.9130435 5.56551724  9 IL1B KRT17 Cluster 3553 ENSG00000125538 70.3173653 254.069893 33 208.347826 7.52413793 10 IL1RN KRT17 Cluster 3557 ENSG00000136689 42.2335329 128.655914 21.325 59.826087 1.42068966 11 KRT23 KRT17 Cluster 25984 ENSG00000108244 21.9341317 120.349462 19.75 31.3913044 112.786207 12 ALDH3B2 KRT17 Cluster 222 ENSG00000132746 15.0718563 98.6935484 14 15.2608696 19.8827586 13 SUSD2 KRT17 Cluster 56241 ENSG00000099994 22.239521 168.704301 12.8 3.69565217 3.44827586 14 DEFB1 KRT17 Cluster 1672 ENSG00000164825 10.1916168 67.0752688 5.65 34.0434783 1.82068966 15 HLA-DQA2 KRT17 Cluster 3118 ENSG00000237541 7.46107784 61.4677419 20.5 10.7826087 2.06206897 16 CYP4B1 KRT17 Cluster 1580 ENSG00000142973 75.5868264 180.758065 57.325 39.6521739 472.606897 17 PIGR KRT17 Cluster 5284 ENSG00000162896 8.92814371 197.983871 23.2 27.6521739 40.0965517 18 SPARC EMT 6678 ENSG00000113140 12.1497006 12.2634409 234.9 20.7826087 2.11034483 19 SERPINF1 EMT 5176 ENSG00000132386 0.91017964 9.05376344 128.825 0 16.537931 20 DCN EMT 1634 ENSG00000011465 1.19760479 7.29569893 509.575 1.91304348 1.36551724 21 SFRP4 EMT 6424 ENSG00000106483 2.41916168 6.68817204 688.275 0 1.84827586 22 CRISPLD2 EMT 83716 ENSG00000103196 9.04191617 18.9731183 259.925 7.86956522 0.67586207 23 TIMP3 EMT 7078 ENSG00000100234 3.49700599 4.02688172 444.3 7.34782609 0.53793103 24 CNN1 EMT 1264 ENSG00000130176 10.0479042 0.49462366 116.625 6.13043478 10.5517241 25 MYH11 EMT 4629 ENSG00000133392 1.05389222 2.10215054 266.65 3.13043478 7.32413793 26 MFAP4 EMT 4239 ENSG00000166482 1.44311377 0.07526882 344.85 0 0.00689655 27 ENG EMT 2022 ENSG00000106991 3.8502994 3.73655914 103 0.39130435 3.07586207 28 EFEMP1 EMT 2202 ENSG00000115380 39.6706587 138.83871 273.275 13.0434783 119.427586 29 RGS16 EMT 6004 ENSG00000143333 2.31137725 2.70967742 259.075 25.173913 95.9931035 30 FEN1 Cell cycle 2237 ENSG00000168496 13.0239521 12.3387097 24.9 87.8695652 13.2344828 31 NUSAP1 Cell cycle 51203 ENSG00000137804 7.43712575 2.2688172 8.575 135.391304 1.82758621 32 UBE2C Cell cycle 11065 ENSG00000175063 0.02994012 0.24731183 2.725 203.043478 1.4137931 33 ZWINT Cell cycle 11130 ENSG00000122952 9.18562874 10.0483871 4.85 190 3.42758621 34 PRC1 Cell cycle 9055 ENSG00000198901 10.742515 5.19892473 15.675 122.913044 5.66206897 35 ASF1B Cell cycle 55723 ENSG00000105011 0.58083832 0 0.2 146.173913 3.84137931 36 MCM4 Cell cycle 4173 ENSG00000104738 42.9161677 28.0806452 65.225 209.608696 24.6275862 37 GINS2 Cell cycle 51659 ENSG00000131153 6.88622755 1.12365591 8.15 74.1304348 1.03448276 38 CENPM Cell cycle 79019 ENSG00000100162 1.19760479 1.44623656 24.575 77.5652174 80.6551724 39 MCM2 Cell cycle 4171 ENSG00000073111 34.9101796 11.9677419 39.325 92.3478261 33.6275862 40 TK1 Cell cycle 7083 ENSG00000167900 3.45508982 7.1344086 3.85 351.913044 32.7655172 41 MCM6 Cell cycle 4175 ENSG00000076003 14.7065868 8.90322581 53.025 131.26087 2.66896552 42 SMC4 Cell cycle 10051 ENSG00000113810 14.1616767 4.79569893 14.25 71.9565217 15.9931035 43 CENPU (MLF1IP) Cell cycle 79682 ENSG00000151725 1.75449102 1.30645161 8.725 71.173913 4.19310345 44 MAD2LI Cell cycle 4085 ENSG00000164109 8.0239521 4.31182796 3 87.173913 37.0344828 45 TEKT1 Ciliated 83659 ENSG00000167858 5.5988024 0.8655914 0.3 0.82608696 760.510552 46 TUBA4B Ciliated 80086 ENSG00000243910 0.13772455 1.17741936 0.05 0.26086957 674.096552 47 C20orf85 Ciliated 128602 ENSG00000124237 0.83233533 2.70967742 0 0 760.510552 48 CAPSL Ciliated 133690 ENSG00000152611 4.94610778 5.50537634 6.775 0.04347826 760.510552 49 LRRC46 Ciliated 90506 ENSG00000141294 2.08982036 3.40860215 5.075 1.17391304 501.317241 50 EFCAB1 Ciliated 79645 ENSG00000034239 0.98203593 0.79032258 0 0.04347826 611.6 51 C6orf118 Ciliated 168090 ENSG00000112539 17.7125749 2.11827957 0 14.3913044 493.765517 52 CCDC78 Ciliated 124093 ENSG00000162004 0.04790419 0.19354839 0 0 702.027586 The table lists 52 marker genes. The HGNC gene symbol is listed in the second column, the Entrez gene 10 in the fourth column and the Ensembl gene ID in the fifth column. The third column describes which signature the gene belongs to. The sixth to tenth columns show the scaled expression levels of each gene in a certain cell state signature. The numbers are used in the deconvolution step to calculate the cell state proportions.

[0396] Comparison between the First Panel and the Second Panel

[0397] By comparing the survival analysis results, the second gene panel generated more significant (p<0.05) and reproducible results across multiple datasets.

TABLE-US-00007 TABLE 5 Multivariate survival analysis of patients' overall survival against EMT scores, grades and stages by using the first panel in the deconvolution analysis N N Hazard Lower Upper Dataset samples events ratio p CI95 CI95 E.MTAB.386 128 73 2.4 0.056 1.0 5.7 GSE13876 144 105 1.8 0.119 0.9 3.9 GSE26193 79 60 2.3 0.057 1.0 5.5 GSE26712 185 129 2.0 0.040 1.0 3.7 GSE32062.GPL6480 260 121 1.8 0.067 1.0 3.5 GSE49997 170 47 2.5 0.049 1.0 6.1 GSE51088 113 93 2.4 0.010 1.2 4.8 TCGA 184 307 2.2 0.011 1.2 4.0 AOCS 109 253 2.2 0.005 1.3 3.9

TABLE-US-00008 TABLE 6 Multivariate survival analysis of patients' overall survival against EMT scores, grades and stages by using the second panel in the deconvolution analysis N N Hazard Lower Upper Dataset samples events ratio p CI95 CI95 E.MTAB.386 128 73 3.1 0.019 1.2 7.9 GSE13876 144 105 2.9 0.013 1.3 6.7 GSE26193 79 60 2.6 0.041 1.0 6.5 GSE26712 185 129 2.0 0.031 1.1 3.9 GSE32062.GPL6480 260 121 1.9 0.054 1.0 3.6 GSE49997 170 47 2.8 0.022 1.2 6.9 GSE51088 113 93 2.1 0.022 1.1 3.9 TCGA 184 307 2.2 0.009 1.2 3.9 AOCS 109 253 2.1 0.008 1.2 3.7

EXAMPLE 4

The Immunophenotype

[0398] We investigated if the EMT scores correlate with the immunophenotype of SOC. We computed the proportion of multiple types of leukocytes in the TCGA data by using CIBERSORT. We used both the LM22 and LM6 signatures, which generates two sets of deconvolution results. In the results generated by using LM22, the EMT-high tumors have significantly higher proportion of macrophage M2 (FIG. 11A). In the results generated by using LM6, the EMT-high tumors have significantly higher proportion of monocytes (FIG. 11B). We next conducted association analysis between the EMT scores and the expression levels of macrophage marker genes, which shows that the positive correlation also existed between the EMT scores and the expression of macrophage markers (FIG. 11C). Overall, the results suggest that there is a positive association between macrophages M2 and EMT components in serous ovarian tumours. Therefore, the method of determining the status of high-grade serous ovarian carcinoma (HGSOC) in a subject could indicate that immunotherapy can be used to target the tumour.

REFERENCES

[0399] All references cited herein are incorporated by reference.

[0400] Ansieau, S. et al. Cancer Cell 14, 79-89. doi:10.1016/j.ccr.2008.06.005

[0401] Baron, M., et al., 2016. Cell Syst 3, 346-. doi:10.1016/j.cels.2016.08.011

[0402] Bell, D. et al. 2011. Integrated genomic analyses of ovarian carcinoma. Nature 474, 609-615. doi:10.1038/nature10166

[0403] Campbell, K. R., Yau, C., 2018. Nat Commun 9. doi:10.1038/s41467-018-04696-6

[0404] Cerny, K. L., et al. PLoS ONE 11, e0147685. doi:10.1371/journal.pone.0147685

[0405] Chandrasekaran, K. S. et al. 2016. Br. J. Cancer 115, 741-751. doi: 10.1038/bjc.2016.234

[0406] Chen, G. M., et al. 2018. Clin. Cancer Res. 24, 5037-5047. doi:10.1158/1078-0432.CCR-18-0784

[0407] Comer, M. T., et al. 1998. Human Reproduction 13, 3114-3120. doi:10.1093/humrep/13.11.3114

[0408] Fan, J., et al. 2018. Linking transcriptional and genetic tumor heterogeneity through allele analysis of single-cell RNA-seq data. Genome Res. gr.228080.117. doi:10.1101/gr.228080.117

[0409] Ford, C. E., et al. 2013. PLoS ONE 8, e54362. doi:10.1371/journal.pone.0054362

[0410] Fresno Vara, J. A., et al. 2004. Cancer Treat. Rev. 30, 193-204. doi: 10.1016/j .ctrv.2003 .07.007

[0411] Ganzfried, B. F., et al. 2013. CuratedOvarianData: clinically annotated data for the ovarian cancer transcriptome. Database (Oxford) 2013, bat013. doi:10.1093/database/bat013

[0412] Ghosh, A., et al. 2017. Development 144, 3031-3041. doi:10.1242/dev.149989

[0413] Gregory, P. A., et al. 2008. Nat. Cell Biol. 10, 593-601. doi:10.1038/ncb1722

[0414] Hanahan, D., Weinberg, R. A., 2011. Hallmarks of Cancer: The Next Generation. Cell 144, 646-674. doi:10.1016/j.ce11.2011.02.013

[0415] Hellner, K., et al. 2016EBIOM 10, 137-149. doi:10.1016/j.ebiom.2016.06.048

[0416] Honkaniemi, J. et al. 1992. Neuroreport 3, 849-852.

[0417] Kang, Y., Massagué, J., 2004. Cell 118, 277-279. doi:10.1016/j.ce11.2004.07.011

[0418] Klos Dehring, D. A., et al. 2013. Dev. Cell 27, 103-112. doi: 10.1016/j .devce1.2013.08.021

[0419] Kuhn, E., et al. 2012. International Journal of Gynecological Pathology 31, 416-422. doi:10.1097/PGP.0b013e31824cbeb4

[0420] Li, Y.-R., Yang, W.-X., 2016. Oncotarget 7, 46785-46812. doi:10.18632/oncotarget.8800

[0421] Liu, M., et al. 2013. Genes Dev. 27, 2543-2548. doi:10.1101/gad.224170.113

[0422] Mahalingaiah, P. K. S., Ponnusamy, L., Singh, K. P., 2015. Cells. J. Cell. Physiol. 230, 1916-1928. doi:10.1002/jcp.24922

[0423] McLean, G. W., et al. 2005. Nature Reviews Cancer 5, 505-515. doi:10.1038/nrc1647

[0424] Nalepa, G., Clapp, D. W., 2018. Nature Reviews Cancer 18, 168-185. doi:10.1038/nrc.2017.116

[0425] Newman, A. M., et al. 2015. Nat Meth 12, 453-. doi:10.1038/NMETH.3337

[0426] Paik, D. Y., et al. 2012. Stem Cells 30, 2487-2497. doi:10.1002/stem.1207

[0427] Picelli, S., et al. 2014. Nature Protocols 9, 171-181. doi:10.1038/nprot.2014.006

[0428] Wang, S., et al. 2002. Biochem. Biophys. Res. Commun. 291, 414-420. doi:10.1006/bbrc.2002.6461

[0429] Wu, R., et al. 2016. J. Pathol. 240, 341-351. doi:10.1002/path.4783

[0430] Yang, J., et al. 2004. Cell 117, 927-939. doi:10.1016/j.ce11.2004.06.006

[0431] Yin, X., et al. 2016. Oncotarget 7, 47938-47953. doi:10.18632/oncotarget.10296

[0432] Bentink, S., et al. 2012. PLoS ONE 7, e30269. doi:10.1371/journal.pone.0030269

[0433] Bonome, T., et al. 2008. Cancer Res. 68, 5478-5486. doi:10.1158/0008-5472.CAN-07-6595

[0434] Crijns, A. P. G., et al. 2009. PLoS Med. 6, e24. doi:10.1371/journal.pmed.1000024

[0435] Karlan, B. Y., et al. 2014. Gynecol. Oncol. 132, 334-342. doi:10.1016/j.ygyno.2013.12.021

[0436] Mateescu, B., et al. 2011. Nat. Med. 17, 1627-1635. doi:10.1038/nm.2512

[0437] Pils, D., et al. 2012. Cancer Sci. 103, 1334-1341. doi:10.1111/j.1349-7006.2012.02306.x

[0438] Tothill, R. W., Tinker, A. V., George, J., Brown, R., Fox, S. B., Lade, S., Johnson, D. S.,

[0439] Trivett, M. K., Etemadmoghadam, D., Locandro, B., Traficante, N., Fereday, S., Hung, J. A., Chiew, Y.-E., Haviv, I., Australian Ovarian Cancer Study Group, Gertig, D., DeFazio, A., Bowtell, D. D. L., 2008. Novel molecular subtypes of serous and endometrioid ovarian cancer linked to clinical outcome. Clin. Cancer Res. 14, 5198-5208. doi:10.1158/1078-0432.CCR-08-0196

[0440] Yoshihara, K., Tsunoda, T., Shigemizu, D., Fujiwara, H., Hatae, M., Fujiwara, H., Masuzaki, H., Katabuchi, H., Kawakami, Y., Okamoto, A., Nogawa, T., Matsumura, N., Udagawa, Y., Saito, T., Itamochi, H., Takano, M., Miyagi, E., Sudo, T., Ushijima, K., Iwase, H., Seki, H., Terao, Y., Enomoto, T., Mikami, M., Akazawa, K., Tsuda, H., Moriya, T., Tajima, A., Inoue, I., Tanaka, K., Japanese Serous Ovarian Cancer Study Group, 2012. High-risk ovarian cancer based on 126-gene expression signature is uniquely characterized by downregulation of antigen presentation pathway. Clin. Cancer Res. 18, 1374-1385. doi:10.1158/1078-0432.CCR-11-2725