P4-16-02: Problems with Identifying Bone Metastasis-Specific Genes without Considering Biological Differences between ER-Positive and ER-Negative Breast Cancers.

Abstract

Abstract Background: Bone metastasis-specific genes in breast cancer have been reported without considering the significant differences in ER status between bone and non-bone metastases. The aims of this study were to validate genes that had been reported as bone metastasis-specific genes using our data set and to identify bone metastasis-specific genes on the basis of biological differences between ERpositive and ER-negative breast cancers. Patients and Methods: We used Affymetrix GeneChip arrays to analyze tumor samples obtained from 365 primary invasive breast cancer patients who underwent surgery from 1999 to 2008. We excluded patients with HER2−positive breast cancer (normalized HER2 mRNA expression [probe set 216836_s_at] > 12.54). We classified the samples into 3 cohorts according to first metastatic site: bone, non-bone, or no metastasis. Differential expression of genes between bone and non-bone cohorts that were differentially expressed was identified using the Cox proportional hazards model, and gene sets was assessed using gene-set analysis. Results: Of the 365 patients, 34 (9.3%) were included in the bone cohort and 32 (8.8%) in the non-bone cohort. Two hundred fourteen (58.6%) had ER-positive and 151 (41.4%) had ER-negative breast cancer. First, we performed gene-set analysis using 5 gene sets that had been reported to be associated with bone metastasis. One gene set, which had been detected using an ER-negative breast cancer cell line, was validated as predicting bone metastasis in ER-positive breast cancer. None of the 5 gene sets predicted bone metastasis in ER-negative breast cancer. We then determined the levels of individual genes associated with bone metastasis by ER status using all 16,712 probe sets filtered by average gene expression level. When we analyzed all patients without any stratification by ER status, as in previous studies, 592 probe sets were significantly overexpressed in the bone cohort compared with the non-bone cohort, with a false discovery rate of ≤0.05. However, when we analyzed ER-positive and ER-negative breast cancers separately, no genes were found with significant differences between bone and non-bone cohorts. Finally, we used 2,246 functionally annotated gene sets assembled from Gene Ontology to examine possible biological differences between bone and non-bone cohorts. In the bone cohort, 151 and 125 gene sets were significantly overexpressed in ER-positive and ER-negative breast cancers, respectively (P ≤ 0.05). Ppathways related to Cellular growth and proliferation, and intracellular and second-messenger signaling were overexpressed in ER-positive breast cancer, whereas pathways related to nuclear receptor and cytokine signaling were overexpressed in ER-negative breast cancer. Most bone-metastasis-related pathways were different in ER-positive and ER-negative breast cancers (91.4% and 89.6% of the gene sets, respectively). Discussion: No genes were found that can predict bone metastasis. ER-positive and ER-negative breast cancers have different biological potentials for bone metastasis. Therefore, we need to assess the prediction model of bone metastasis based on the biological features for each ER status separately. Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P4-16-02.