posted on 2025-11-05, 07:01authored byDorothy Hallberg, Alice C. Eastman, Shashikant Koul, Daniel C. Bruhm, Eniko Papp, Simon Davenport, Vilmos Adleff, Leonardo Ferreira, Noushin Niknafs, Jamie E. Medina, Stephen Cristiano, Carolyn Hruban, Jacob Fiksel, Kaui P. Lebarbenchon, Luis Aparicio, Nicholas A. Vulpescu, Kuan-Ting Kuo, Nita Ahuja, Ronny Drapkin, Euihye Jung, Sarah H. Kim, Mark A. Eckert, Ernst Lengyel, Kentaro Nakayama, Ayse Ayhan, Ie-Ming Shih, Tian-Li Wang, Ofer Lavie, Gad Rennert, Hariharan Easwaran, Stephen B. Baylin, Michael F. Press, Victor E. Velculescu, Robert B. Scharpf
<p>Ovarian endometrioid carcinomas are epigenetically similar to healthy uterine endometrium, whereas ovarian mucinous carcinomas are mixed with both mucinous- and endometrial-like signatures. <b>A,</b> We trained a LD model for the classification of samples as stomach mucinous, colorectal mucinous, or uterine endometrial from variably methylated CpG sites in TCGA samples. Ellipses denote the 95th percentile of the bivariate normal distribution for each of these histology classifications. <b>B,</b> Using the LD model from (<b>A</b>), the histologic classifications of samples were obtained and overlaid on the ellipses redrawn from <b>A</b>. The co-clustering of colorectal and stomach mucinous carcinomas with the corresponding tumor types from TCGA confirm that the methylation patterns that separate these tissues in TCGA broadly reflect histology differences (top). Whereas ovarian mucinous and ovarian endometrioid samples have no counterpart in TCGA (bottom), we evaluated whether these samples would be epigenetically similar to their morphologic counterparts in TCGA. <b>C,</b> Class probabilities from the LD model evaluated on the analyzed samples. As expected, patients with stomach and colorectal mucinous carcinomas were correctly classified as mucinous, and patients with ovarian endometrioid carcinomas were correctly classified as endometrial. Interestingly, classification of patients with ovarian mucinous carcinoma was more heterogenous, with nearly half of the samples exhibiting a mucinous signature and the remaining samples bearing more of a resemblance to endometrial methylation patterns. The two colorectal mucinous samples in <b>B</b> (top left) were sequencing replicates from a single patient’s tumor (patient CGCRC330).</p>
Funding
Dr. Miriam and Sheldon G. Adelson Medical Research Foundation (AMRF)
Gray Foundation
National Institutes of Health (NIH)
Commonwealth Foundation for Cancer Research Foundation
U.S. Department of Defense (DOD)
Honorable Tina Brozman Foundation (Tina’s Wish)
Stand Up To Cancer (SU2C)
National Institute of General Medical Sciences (NIGMS)
United States Department of Health and Human Services
Endometrioid and mucinous ovarian carcinomas represent nearly a fifth of ovarian cancers, but their molecular characteristics and pathologic origins are poorly understood. To identify the genomic and epigenomic alterations characteristic of these ovarian cancer subtypes and evaluate links to morphologically similar tumors from other sites, we performed a combination of sequence, copy number, mutation signature, and rearrangement analyses from tumor samples and matched normal tissues of 133 patients, as well as methylation analyses of these tumors and tissues of 150 patients from The Cancer Genome Atlas. Genomic analyses included samples from patients with ovarian endometrioid (n = 44), ovarian mucinous (n = 43), uterine endometrioid (n = 15), and gastrointestinal mucinous carcinomas (n = 31), including mucinous carcinomas of the stomach, colon, and pancreas. In addition to identifying genes previously known to be involved in these tumors, we identified alterations in RAD51C, NOTCH4, SMARCA1/4, and JAK1 in ovarian endometrioid, ESR1 in uterine endometrioid, and SMARCA4 in ovarian mucinous carcinomas. Whole-genome sequencing revealed rearrangements involving PTEN, NF1, and NF2 in ovarian endometrioid carcinomas and NF1 and MED1 in ovarian mucinous carcinomas. The number of alterations, affected genes, and genome-wide methylation profiles were not distinguishable between ovarian and uterine endometrioid carcinomas, supporting the hypothesis that these tumors share a tissue of origin. In contrast, mutation and methylation patterns in ovarian mucinous carcinomas were different from gastrointestinal mucinous carcinomas. These analyses provide insights into the genomic landscapes and origins of mucinous and endometrioid ovarian carcinomas, providing new avenues for early clinical intervention and management of patients with these cancers.
Integrative multi-omic analyses support a common tissue of origin between ovarian endometrioid and uterine endometrioid carcinomas but not between ovarian mucinous and gastric or pancreatic mucinous carcinomas.