1094 Background: The genomics of patients with metastatic breast cancer (MBC) who develop brain metastases (BM) is not well understood given the difficulty in obtaining brain tumor for genotyping. We compared tumor genotyping results via cell-free DNA (cfDNA) collected at MBC diagnosis in patients who developed BM after MBC diagnosis with those who did not develop BM (non-BM). Methods: Patients at an academic institution who had cfDNA testing (Guardant 360/Next generation sequencing, 73 gene assay) at MBC diagnosis between 1/2016-12/2017, with ≥ 6 months of follow-up post testing, were identified. A chart review was done to identify tumor subtype, demographics, cfDNA results, and development of BM at or after MBC diagnosis. Pearson’s chi-squared and Wilcoxon rank sum tests were used to determine differences in clinical and cfDNA characteristics in BM vs. non-BM (p<0.05 for statistical significance). Results: CfDNA results were available for 49 patients, of whom 13 (27%) developed BM (4 with BM at MBC diagnosis). The median time to BM development was 11 months. While patients with BM were younger at MBC diagnosis than non-BM (median age BM 53 vs. non-BM 61, p=0.05), they had similar subtype (BM vs. non-BM: HR+/HER2- 62% vs. 69%, HER2+ 8% vs. 14%, TNBC 23% vs. 17%, unknown 8% vs. 0%, p=0.3), de-novo vs. recurrent disease (BM vs. non-BM: de-novo 8% vs. 14%, recurrent 92% vs. 86%, p=0.6), and visceral disease (BM vs. non-BM: 77% vs. 56%, p=0.2) distributions. All patients with BM had ≥1 detectable cfDNA mutation vs. 88% of non-BM. While the median mutant allele frequency of the most common mutation was similar in BM vs. non-BM (2.4% vs. 3.7%, p=0.5), the mutation pattern varied. Patients with BM more often had mutations in BRCA1 (15% vs. 3%, p=0.1), APC (15% vs. 0%, p=0.02), and CDKN2A (15% vs. 0%, p=0.02), compared to non-BM. In 4 patients with BM at MBC diagnosis, mutations in APC (50%), CDKN2A (50%), and BRCA 1/2 (25%) were noted; 1 had coexisting APC and BRCA1/2 mutations and another had coexisting APC and CDKN2A mutations. Conclusions: Patients with MBC who develop BM may have different cfDNA genomics, particularly BRCA1, APC, and CDKN2A mutations. Further research is needed to determine the predictive value of cfDNA at MBC diagnosis in the identification of patients at higher risk of developing BM. [Table: see text]
Read full abstract