Abstract 4343: Revealing ovarian cancer copy number variation in single cells

Abstract

The mortality rate associated with ovarian cancer (OvCa) is disproportionately high in comparison to its incidence rate. This is partly due to the heterogeneous nature of the disease, which reduces treatment efficacy and contributes to high rates of relapse and chemotherapy resistance. Most OvCa are epithelial in origin and can be classified into four main subtypes: serous, mucinous, endometrioid, and clear cell. Of these, high grade serous ovarian cancer (HGSOC) is the deadliest. Epithelial ovarian carcinomas (EOC) typically exhibit widespread chromosomal and arm-level copy number abnormalities across most of the genome; in HGSOC, focal amplifications and microdeletions are especially prevalent and indicative of high genomic instability. To understand the heterogeneity of aneuploidy in EOC and HGSOC, we performe single-cell whole genome sequencing on four EOC samples: two HGSOC, one clear cell, and one mixed clear cell and endometrioid. All samples were late stage and treatment naïve, and one sample had a known BRCA2 mutation. Sequencing data was processed by two complementary methods to call copy number alterations. First, we used the Cell Ranger DNA pipeline (10x Genomics) to align cell-identified sequencing reads to human reference genome GRCh38 for coverage-based copy number estimation. Resulting copy number calls were cleaned up for mappability, quality, and noisiness. Each sample was then subject to clustering and subclustering analysis using maximum likelihood genetic clustering algorithms. All samples exhibited a high level of aneuploidy, including characteristic alterations known to be associated with EOC. Two tumors contained readily distinguishable clonal populations, and all samples contained main tumor clones that could be further divided by unique subclonal characteristics. Evidence of polyploidy was also seen in all four specimens, with some tumor clusters exhibiting triploid and tetraploid baselines. In parallel, sequencing data was analyzed by the Copy-number Haplotype Inference in Single-cell by Evolutionary Links (CHISEL) algorithm. CHISEL utilizes both binned read depth ratio and B-allele frequency data to determine allele- and haplotype-specific copy numbers in single cells. Results from CHISEL confirmed the copy number calls from Cell Ranger DNA, and revealed widespread loss of heterozygosity in all samples. These findings were corroborated with allele-specific copy number data derived from matched tumor-normal whole exome sequencing. Furthermore, CHISEL detected polyploidy in one-third of the tumor cells with no preference for the A or B alleles. Overall, our findings highlight that the known heterogeneity of ovarian cancer extends to the level of aneuploidy and CNAs, shedding light on factors which pose significant barriers to effective personalized medicine implementation.