Dataset Information

DNA Methylation Profiling Enables Subclassification of Mucinous Ovarian Carcinoma and Distinguishes It from Extraovarian Mucinous Metastases

ABSTRACT: Mucinous ovarian carcinoma (MOC) is a rare and poorly characterized subtype of epithelial ovarian cancer. Pathogenetically, it is the only epithelial ovarian carcinoma for which the cell of origin remains unknown. Clinically, MOC presents a significant diagnostic challenge due to its close morphological resemblance to extraovarian metastases (EOM) involving the ovary. Accurate diagnosis is critical, as it directly informs treatment decisions. To improve our understanding of MOC and enhance its diagnostic accuracy, we performed genome-wide DNA methylation profiling on MOC (n = 58), EOM (n = 38), and mucinous borderline ovarian tumors (mBOT, n = 18) collected from three institutions. We developed a machine-learning pipeline based on DNA methylation data to support this challenging differential diagnosis. Our analysis revealed two distinct subtypes of mBOTâ€”one epigenetically similar to normal ovary and the other to MOCâ€”supporting the stepwise progression of mBOT to MOC. Furthermore, we identified two clinically relevant DNA methylation subtypes of MOC with prognostic implications, enabling molecular stratification of patients. Using these data and 389 additional DNA methylation profiles, we developed and validated a three-step molecular classifier. In the first step, MOC was identified with an accuracy of 98.04%. In the second step, EOM was detected and non-mucinous lesions excluded with 97.6% accuracy. In the third step, the tissue of origin for EOM was predicted with 86.27% accuracy. Overall classifier accuracy of all three steps was 89.07%. Additionally, we demonstrated the feasibility of implementing the classification process with nanopore sequencing, though high tissue quality remains essential. In conclusion, we identified two prognostically significant subtypes of MOC based on DNA methylation and developed a robust machine learning-based classifier that distinguishes MOC from EOM with high accuracy. These findings have the potential to improve diagnostic precision and guide therapeutic strategies.

ORGANISM(S): Homo sapiens

PROVIDER: GSE310580 | GEO | 2026/05/27

REPOSITORIES: GEO

ACCESS DATA

Dataset's files

Source:

			Action	DRS
		Other

Items per page:

1 - 1 of 1

Similar Datasets

Project description:Mucinous cystic neoplasms (MCN) remain a major riddle in hepato-pancreato-biliary pathology. These cystic tumors are defined by their mucinous epithelium and ovarian-like stroma, with an estimated 10% risk of progression to invasive carcinoma. The origin of the ovarian stroma remains a subject of debate. In this study, we conducted immunohistochemical profiling, targeted DNA sequencing, and genome-wide DNA methylation analysis on a cohort of 15 pancreatic MCNs (MCN-P) and 6 hepatic MCNs (MCN-L). Unsupervised analysis of DNA methylation profiles of pancreatic neoplasms (11 entities and normal pancreatic tissue from 224 unique samples) revealed that MCN-P predominantly forms a distinct group. In the DNA methylation landscape of liver tumors, encompassing 5 tumor types and normal bile duct tissue from 136 unique samples, MCN-L demonstrated a specific methylation profile when compared to all other entities. Furthermore, within the DNA methylation landscape of ovarian tumors—featuring 5 tumor types, normal fallopian tube, and normal ovarian tissue from 90 unique samples—we found that both MCN-P and MCN-L grouped with mucinous ovarian carcinoma and mucinous borderline ovarian tumors. Notably, low-grade MCNs exhibited greater DNA methylation similarities to mucinous borderline ovarian tumors, while high-grade or invasive MCNs were primarily associated with mucinous ovarian carcinomas. When analyzing all samples together (19 tumor types and 4 normal tissue types, n = 430), MCNs similarly grouped with mucinous ovarian tumors and normal ovarian tissue. Additionally, in anetwork analysis of differentially methylated probes indicated that MCN-P and MCN-L share significant methylomic traits, closely resembling mucinous ovarian tumors. In conclusion, our findings highlight that MCN-P and MCN-L are distinct entities in the landscape of pancreatic and hepatic tumors and show DNA methylation profile similarities with mucinous ovarian tumors, suggesting a potential common origin.

Project description:Epithelial ovarian cancer (EOC) are a heterogeneous collection of malignancies, each with their own developmental origins, clinical behavior and molecular profile. With less than 5% of EOC cases, mucinous ovarian carcinoma is a rare form with a poor prognosis with a 5-year survival of 11% for advanced stages (III/IV). Clinically, due to the large size and heterogeneity of mu-cinous tumors, Malignant forms are difficult to distinguish from Borderline (15%) and benign (80%) forms with a better prognosis. Improving their diagnosis is therefore a challenge with regard to the risk of under-treating a malignant form or of unnecessarily undertaking radical surgical excision. The involvement of microRNAs (miRNAs) in cancer and their potential as biomarkers of diagnosis are becoming increasingly recognized. In this study, miRNA expres-sion profiling generates using FFPE patient samples followed by qRT-PCR validation allows to identify 10 down-regulated and 5 up-regulated miRNAs between Malignant and Borderline tumors. To overcome issues linked to data normalization and to improve the accuracy of the results, a ratio analysis combining up-regulated and down-regulated miRNAs was used. Alt-hough 21/50 miRNA expression ratio were significantly different between Malignant and Bor-derline tumor samples, any ratio could perfectly discriminate the two groups. However, a combination of 14 miRNAs expression ratio (double ratio) showed high discriminatory poten-tial, with 100% of accuracy in distinguishing Malignant and Borderline ovarian tumors, and suggests that miRNAs may hold significant clinical potential as a diagnostic tool. Epithelial ovarian cancer (EOC) are a heterogeneous collection of malignancies, each with their own developmental origins, clinical behavior and molecular profile. With less than 5% of EOC cases, mucinous ovarian carcinoma is a rare form with a poor prognosis with a 5-year survival of 11% for advanced stages (III/IV). Clinically, due to the large size and heterogeneity of mu-cinous tumors, Malignant forms are difficult to distinguish from Borderline (15%) and benign (80%) forms with a better prognosis. Improving their diagnosis is therefore a challenge with regard to the risk of under-treating a malignant form or of unnecessarily undertaking radical surgical excision. The involvement of microRNAs (miRNAs) in cancer and their potential as biomarkers of diagnosis are becoming increasingly recognized. In this study, miRNA expres-sion profiling generates using FFPE patient samples followed by qRT-PCR validation allows to identify 10 down-regulated and 5 up-regulated miRNAs between Malignant and Borderline tumors. To overcome issues linked to data normalization and to improve the accuracy of the results, a ratio analysis combining up-regulated and down-regulated miRNAs was used. Alt-hough 21/50 miRNA expression ratio were significantly different between Malignant and Bor-derline tumor samples, any ratio could perfectly discriminate the two groups. However, a combination of 14 miRNAs expression ratio (double ratio) showed high discriminatory poten-tial, with 100% of accuracy in distinguishing Malignant and Borderline ovarian tumors, and suggests that miRNAs may hold significant clinical potential as a diagnostic tool.

Dataset Information

DNA Methylation Profiling Enables Subclassification of Mucinous Ovarian Carcinoma and Distinguishes It from Extraovarian Mucinous Metastases

Dataset's files

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets