Validation of genomic markers that predict distant recurrence risk via FISH assays in women with early stage breast carcinoma

Abstract

10506 Background: Biomarkers offer the potential to better predict recurrence risk in early stage invasive breast cancer, especially if adaptable to common assays such as fluorescent in situ hybridization (FISH). We studied the ability of tumor copy numbers (CN) of 3 genes, BIRC5, CYP24, and PDCD6IP, to predict distant recurrence in women with breast cancer. Methods: We conducted a blinded, multi-site study of 350 women with lymph node negative, estrogen/progesterone receptor positive (ER/PR+), invasive ductal carcinomas with archived paraffin-embedded biopsy specimens available. Eligible women received local surgical and radiation treatment only (93) (LTO) or local treatment and tamoxifen therapy only (257), and had distant metastases or no evidence of distant recurrence with at least 5 years follow-up (FU). Tumors were subjected to FISH with DNA probes for the 3 genes to assess CN. A prognostic index (PI) was calculated for each patient based on the 3 genes’ CN. A predetermined threshold (3.0) was used to categorize PI's as low or high. Data were assessed by Fisher's exact analysis and by categorical and continuous Cox modeling. Results: The study included 350 women with Stage I (288) and II (62) disease. FU averaged 7.3 years. Distant recurrence occurred in 32 women (9.1%) overall and in 5.2% of women with Stage I disease. In a multivariate Cox analysis including continuous PI, age, tamoxifen treatment, grade, and tumor size, PI remained a significant predictor of recurrence for Stage I patients: OR=1.56, p=0.019, and Stage I/LTO patients: OR=6.02, p=0.00007. Using the PI as a categorical classifier, the high/low recurrence rates were significantly different: OR=2.24, p= 0.045. Similar results were seen in Stage I patients: OR=3.53, p=0.021 with recurrence rates of 3.5% and 11.5% in the low and high risk groups, respectively, and for the Stage I/LTO patients: OR=5.33, p=0.041. Conclusions: FISH analysis of 3 genes allows metastasis risk prediction in women with early stage ER/PR+ cancers. This may be particularly useful for Stage I cancers traditionally considered low risk. The markers are prognostic in that they predict metastasis in patients receiving only local therapy. No significant financial relationships to disclose.