Loss of heterozygosity (LOH) as a measure of whole-genome instability in ovarian cancer correlates with clinical outcomes.

Abstract

5027 Background: The aim of the study was to assay the amount of loss of heterozygosity (LOH) across the whole genome as a measure of genomic instability in ovarian tumors and to determine the correlation between the fraction of the ovarian cancer genome with LOH and the clinical outcomes of ovarian cancer patients. Methods: We assessed the prevalence of LOH in 100 unselected ovarian cancer tumors using whole genome 500K Single Nucleotide Polymorphism (SNP) arrays (Affymetrix, Inc.). Allele dosage, copy number, and degree of contamination with benign tissue were used to estimate the likelihood of various genotypes in tumor tissue. Regions with the copy number of more than 3 were eliminated from the analysis since LOH data are unreliable when copy number exceeds 3. Results: We found that copy number changes of 1, 2 and 3 composed 12.6%, 45.7% and 21.5% of the genome in ovarian cancer, respectively. These changes contributed to 42.4%, 43.8% and 13.8% of LOH in ovarian cancer, respectively. The average fraction of the whole genome with LOH in ovarian cancer was 35%. The fraction of the genome with LOH was significantly correlated with tumor grade. The fraction of the ovarian cancer genome with LOH was statistically highly correlated with both overall survival (OS;p=3.4x10-6) and progression-free survival (PFS; p=0.0016) after debulking surgery and platinum/taxane-based chemotherapy. In multivariate analyses with clinical variables, only the fraction of the genome with LOH and residual tumor size (RTS) after debulking surgery were independent predictive factors for OS and PFS. Conclusions: Increased fraction of the genome with LOH as a measure of genomic instability in ovarian cancer is a significant and independent negative prognostic factor for both OS and PFS. The validation of the fraction of the ovarian cancer genome with LOH as an independent predictor of RFS and OS in a second ovarian cancer sample set will also be presented at the meeting.