TCGA Gynecologic Cohorts

Cancers of the ovary, uterus and cervix are often grouped together, but their genomes go wrong in different ways, and that shapes which treatments can work. The Cancer Genome Atlas (TCGA) profiled thousands of tumors with the same methods, which makes these differences directly comparable. Here we compare 1,468 tumor samples from its four gynecologic cohorts: ovarian (OV), endometrial (UCEC), cervical (CESC) and uterine carcinosarcoma (UCS).

Explore the data: Open the dataset in the Discovery Lab TRE to run reproducible analyses where the data resides.

Dataset page · Open in the Discovery Lab TRE

Key Findings

• Ovarian cancer is driven by copy-number change. Its tumors gain and lose large stretches of chromosomes, more than half of the genome in a typical tumor, while carrying relatively few mutations.
• Endometrial cancer often breaks the other way. Many endometrial tumors carry heavy mutation loads, and a quarter are microsatellite unstable, a form of DNA repair failure that is almost absent from ovarian cancer.
• Ovarian cancer survival remains poor. Half of the TCGA ovarian cancer patients died within 45 months of diagnosis.

The Cohorts

• OV: 585 samples (high-grade serous ovarian cancer).
• UCEC: 529 samples (uterine corpus endometrial carcinoma).
• CESC: 297 samples (cervical squamous cell carcinoma and endocervical adenocarcinoma).
• UCS: 57 samples (uterine carcinosarcoma).
• Source: cBioPortal's PanCancer Atlas clinical and genomic summaries. The aneuploidy score comes from Taylor et al. (2018) and survival endpoints from the TCGA Clinical Data Resource (Liu et al., 2018).

Papers: Hoadley KA et al. Cell 173, 291–304 (2018), doi:10.1016/j.cell.2018.03.022 · Liu J et al. Cell 173, 400–416 (2018), doi:10.1016/j.cell.2018.02.052 · Taylor AM et al. Cancer Cell 33, 676–689 (2018), doi:10.1016/j.ccell.2018.03.007

Ovarian Cancer Breaks Its Chromosomes, Not Its Genes

Tumors can damage their genomes in two broad ways: by gaining and losing whole chromosome arms (aneuploidy) or by piling up small mutations. Each point here is a tumor. Ovarian cancer (median aneuploidy score 13) and carcinosarcoma (19) sit high on aneuploidy and low on mutations, while endometrial cancer (3) runs the other way, with a tail of hypermutated tumors.

More Than Half of a Typical Ovarian Tumor's Genome Is Altered

Measured as the fraction of the genome with a copy-number change, the contrast is stark: 57% in the median ovarian tumor against 7% in the median endometrial tumor. Mutation burden shows the mirror image.

Microsatellite Instability Is Common in Endometrial Cancer and Rare in Ovarian Cancer

Microsatellite instability is the signature of a failed DNA mismatch repair system, and it matters clinically because these tumors often respond to immunotherapy. It is common in endometrial cancer, at 132 of 528 tumors, and nearly absent from ovarian cancer, at 4 of 512.

Half of Ovarian Cancer Patients Died Within About Four Years

For the 571 ovarian cancer patients with follow-up, 341 died during the study, and the estimated median survival is 45.1 months. Each step down the curve is a death, and the curve shows how many patients were still alive at each point after diagnosis.

About These Charts

Survival is shown for TCGA-OV only: the endometrial, cervical and carcinosarcoma exports carry no survival fields. Stage is not shown because every stage column is empty in these exports.

Fill rates differ by measurement and cohort. For example, aneuploidy is recorded for 552 of 585 ovarian samples and mutation burden for 523.

These charts were built in the Discovery Lab TRE from cBioPortal's PanCancer Atlas studies (OV 6ab033ba, UCEC da148455, CESC ba4a2a04, UCS 53108189; clinical tables 61690970).