Skip to content

Study dataset

Uterine Carcinosarcoma (Johns Hopkins, Nat Commun 2014)

Tumor-normal exome sequencing of 22 carcinosarcomas: 17 uterine, 5 ovarian.

Listed May 5, 2026

  • Genomic
  • Clinical
  • Uterine sarcoma and carcinosarcoma

Overview

Study Design & Cohort

This was the first unbiased, systematic whole-exome sequencing study of gynecologic carcinosarcomas (also called Malignant Mixed Müllerian Tumors, MMMTs). The cohort consisted of 22 carcinosarcoma patients with matched tumor and normal tissue, collected with consent and approved by the OHSU IRB and Knight BioLibrary. The 22 tumors included:

  • 17 uterine carcinosarcomas
  • 5 ovarian carcinosarcomas

All samples were primary tumors with one exception (MM02, a recurrence at the primary site). Eleven patients had received chemotherapy or radiation after surgery. The median age of onset for this tumor type is ~62 years.


Biospecimens & DNA Extraction

Samples were macrodissected to remove contaminating normal tissue, achieving an average of 85% neoplastic cellularity. DNA was extracted from:

  • 10 fresh-frozen tumors (plus matched normals)
  • 12 FFPE (formalin-fixed paraffin-embedded) tumors (plus matched normals)

This mix of tissue types is notable — FFPE introduces sequencing artifacts, and the authors had to account for this in their filtering pipeline.


Sequencing Platform & Coverage

The study used whole-exome sequencing (WES).


Mutation Calls & Key Findings

The bioinformatics pipeline identified tumor-specific somatic single-nucleotide variants and small indels by comparing each tumor to its matched normal, removing known germline polymorphisms, and visually inspecting candidates.

The 22 tumors split into two groups based on mutation burden:

  • 18 microsatellite-stable (MSS) tumors: an average of 43 non-synonymous somatic mutations per tumor (range 4–88), dominated by C:G → T:A transitions consistent with CpG deamination — a signature common across gynecologic cancers.
  • 4 mismatch repair-deficient (dMMR) tumors: dramatically elevated mutation burden ranging from 904 to 5,913 somatic alterations, caused by inactivating mutations in MSH6 (MM20T) or MSH2 (MM04T, MM12T, MM18T).

Key recurrently mutated genes across the cohort:

  • TP53: 67% of cases
  • PIK3CA and PIK3R1: 9/22 cases each
  • ARID1A: 32% (7/22), predominantly in ovarian cases
  • ARID1B: co-mutated with ARID1A in 3 cases
  • MLL3 (KMT2C): 27% (6/22)
  • SPOP: 14% (3/22)
  • BAZ1A: 4/22 cases
  • FBXW7: 23% (5/22)
  • KRAS hotspot mutations: 2 MSS cases
  • Homologous recombination genes (BRCA1, BRCA2, PALB2): present

Overall, approximately two-thirds of cases harbored mutations in chromatin remodelling or modification genes, which was the central finding of the paper.


Germline Analysis

The study also interrogated germline DNA across 84 cancer predisposition genes. One notable finding: patient MM08 carried a germline nonsense mutation in BRCA2 (K3326X), a known pathogenic variant, and had a family history of cervical cancer (mother) and lung cancer (sister).

Inside the dataset

  • Library prep

    Illumina TruSeq

  • Capture

    Agilent SureSelect 51 Mb kit (version 4), targeting >20,000 genes

  • Sequencing

    Illumina HiSeq 2000, paired-end 100bp reads

  • Average coverage

    191-fold

  • Completeness

    >90% of bases covered by ≥10 high-quality reads

  • Alignment

    hg19 reference genome using the Eland algorithm (CASAVA 1.7)

Publications