Comprehensive assessment of the genomic landscape of breast cancer brain metastases reveals targetable alterations and genomic signatures relevant to immune-checkpoint and PARP inhibitors

Abstract Background: Central nervous system (CNS) disease secondary to breast cancer (BC) is a growing clinical problem and a cause of significant morbidity and mortality. A better understanding of the genomic landscape is key to understanding its cause and to developing novel treatments. In this study, we examined the genomic landscape of a large cohort of breast cancer brain metastases (BMs) and compared them to a cohort of local breast cancers (BCs) and non-CNS metastases (N-CNS). Material and Methods: We analyzed 822 BMs and compared them to 11,988 local, breast-biopsied BCs and 15,516 N-CNS (unpaired samples) with comprehensive genomic profiling (CGP) for all classes of alterations in at least 324 genes (Foundation Medicine, Cambridge, MA, USA). Homologous recombination deficiency detected by genome-wide loss of heterozygosity (HRD-gLOH; cutoff 16%), tumour mutation burden (TMB; cutoff 10 mutations/Mb), microsatellite instability (MSI) and PD-L1 prevalence and expression by VENTANA SP142 immune cell immunohistochemistry (IHC; cutoff 1%) was also investigated within and across the cohorts. Results: As compared to local BCs, 31 genes were significantly enriched for genomic alterations in BMs with BM prevalence >3% and false-discovery rate (FDR) <0.1. The most enriched genes were: TP53 (71.8%), MYC (25.9%), ERBB2 (24.6%), PTEN (16.7%), CDKN2A and RB1 (10.3% each), NF1 (8.6%), GNAS (7.9%), BRCA1, CDKN2B and ZNF217 (7.8% each), ARID1A (7.3%), CCNE1 (6.7%), ESR1 (6.4%), MAP2K4 (5.8%), AURKA (5.5%) and PIK3R1 (5.1%). Of these 31 genes, 26 were also enriched between BM vs N-CNS. The BM-enriched. genes that lost significance in comparison to N-CNS were: ZNF217, ARID1A, MAP2K4, AURKA and ARFRP1. PIK3CA prevalence was BMs: 30.2%, BCs: 31.8% and N-CNS: 38.9% (no significant difference between cohorts). HRD-gLOH was higher in BMs: 52.0% vs BCs: 33.7% and N-CNS: 31.5%. The prevalence of immune checkpoint inhibition (ICPI) biomarkers in BM were as follows: TMB-High 15.5%, MSI-High 2.0%, PD-L1 (CD274) 3.5% and its paralogue PDCD1LG2 (PD-L2, CD273) 3.4%. All were significantly higher in the BM vs primary BC and N-CNS group (p<0.05, FDR<0.1 for all comparisons). PD-L1 copy number (CN) variation was also higher in BMs: 3.4% vs BCs: 1.6% and N-CNS: 1.1%. PD-L1 positive expression was lower in BMs: 38.7% vs BCs: 54.3% but higher than N-CNS: 26.2%. Higher PD-L1 positive expression was observed in ER-/HER2- BMs: 48.3% vs BCs: 50.0% vs N-CNS: 21.4%. Table 1 summarizes the genomic signatures and PD-L1 positivity in the BC, BM and N-CNS cohorts and within different receptor subtypes. Conclusion: A high prevalence of clinically-relevant genomic alterations in BM is identified in comparison to the local BCs and/or N-CNS disease. A high proportion of BMs have high HRD-gLOH, TMB, MSI and PD-L1 alterations rendering them potentially amenable to treatment with ICPIs and PARP inhibitors. These data reinforce the importance of acquiring BM material and/or cerebrospinal fluid for CGP to enable rationale treatment decisions as well for enrolment into genomically driven BM studies of targeted agents. Table 1.Prevalence (%) of the genomic signatures HRD-gLOH, TMB, MSI and PD-L1+ expression.Site% HRD-gLOH% TMB-H% MSI-H% PDL1+ALLlocal BC33.70%. (2573/7636)5.04%. (527/10464)0.37%. (39/10464)54.25%. (1092/2013)BM52.01%. (362/696)15.51%. (116/748)2.01%. (15/748)38.73%. (55/142)N-CNS31.55%. (3237/10260)9.82%. (1395/14210)0.23%. (32/14210)26.21%. (615/2346)ER+/HER2-local BC31.48%. (119/378)3.41%. (19/557)0.36%. (2/557)40.00%. (14/35)BM43.17%. (79/183)11.40%. (22/193)1.04%. (2/193)23.08%. (9/39)N-CNS29.65%. (126/425)8.92%. (51/572)0.17%. (1/572)12.99%. (10/77)ER+/HER2+local BC19.57%. (9/46)2.78%. (2/72)0.00%. (0/72)42.86%. (3/7)BM26.67%. (16/60)9.68%. (6/62)0.00%. (0/62)27.27%. (3/11)N-CNS47.37%. (18/38)19.57%. (9/46)0.00%. (0/46)14.29%. (1/7)ER-/HER2+local BC32.26%. (10/31)8.70%. (4/46)0.00%. (0/46)100.00%. (5/5)BM33.33%. (23/69)21.33%. (16/75)2.67%. (2/75)57.14%. (4/7)N-CNS34.38%. (11/32)4.26%. (2/47)0.00%. (0/47)42.86%. (3/7)ER-/HER2-local BC52.25%. (186/356)2.70%. (14/518)0.19%. (1/518)50.00%. (30/60)BM70.45%. (174/247)17.41%. (47/270)2.59%. (7/270)48.33%. (29/60)N-CNS43.52%. (84/193)10.04%. (27/269)0.74%. (2/269)21.43%. (6/28) Citation Format: Athina Giannoudis, Ethan Sokol, Shakti H Ramkissoon, Talvinder Bhogal, Kimberly McGregor, Allison Clark, Evangelia D Razis, Rupert Bartsch, Richard SP Huang, Carlo Palmieri. Comprehensive assessment of the genomic landscape of breast cancer brain metastases reveals targetable alterations and genomic signatures relevant to immune-checkpoint and PARP inhibitors [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr PD4-09.