Background: CDK4/6i have established roles in treating patients (pts) with high-risk earlystage and metastatic HR+/HER2- breast cancer (HR+ MBC). Understanding the genomic and molecular factors that dictate CDK4/6i response is a critical area of research. Low-level copy number gains in the mitotic spindle regulator AURKA have been implicated in promoting CDK4/6i resistance via whole-exome sequencing of tumor samples, and constitutive overexpression provokes resistance in vitro [Wander et al. Cancer Discov 2020]. Developing a clinical biomarker of AURKA copy number gain poses unique challenges given the generally low-level AURKA gain observed in resistant samples. Validation of a reliable assay to detect AURKA-dependent CDK4/6i resistance in the clinic may enable personalized use of emerging AURKA inhibitors such as alisertib, which is currently being evaluated for this indication (ClinicalTrials.gov identifier NCT06369285).

Methods: De-identified clinical and genomic records of pts with HR+ MBC in the Tempus database whose tumors were sequenced using xT (DNA) and xR (RNA) were investigated. Tempus xT is a targeted, tumornormal matched DNA panel that detects single-nucleotide variants, insertions and/or deletions, and copy number variants in 648 genes. Tempus xR is a whole-exome capture, next-generation sequencing assay that identifies transcriptional evidence of chromosomal rearrangements resulting in expression of fusion RNA species. Copy number analyses were performed with a proprietary algorithm assessing tumor purity, ploidy, and B-allele frequencies to estimate copy numbers and minor allele counts. RNA-based analyses classified low and high expressors based on median gene expression values. Real-world progression-free survival (rwPFS) was the time from start of CDK46/i to first recorded disease progression (determined by the primary treating oncologist), death from any cause, or initiation of a subsequent treatment line. Survival analyses were rightcensored at 5 years.

Results: 848 pts with HR+ MBC with DNA sequencing prior to CDK4/6i were identified. Common pathogenic genomic alterations included PIK3CA (SNV: 43.7%; CNV: 2.4%) and ESR1 (SNV: 12.2%; CNV: 6%; fusion: 1.8%). 4.1% of biopsy samples harbored RB1 alterations. AURKA copy number gains were identified in 15.0% of pts, with most demonstrating low-level amplification (copy number 47: 13.8%; copy number >7: 1.2%). Of note, these low-level amplifications would not routinely be reported by available sequencing platforms in clinical settings. Linear regression modeling showed that AURKA copy number correlated with RNA expression (p-value <0.001). Overall, 713 pts with DNA sequencing prior to CDK4/6i exposure also had rwPFS information. Those whose tumors harbored high- or low-level AURKA amplifications (n=116) had inferior rwPFS on CDK4/6i (9.9 vs 17 months for no AURKA amplification (n=597; p=0.00032).

Conclusions: In this large, real-world translational research cohort of pts with HR+ MBC, low-level AURKA copy number gain was common. These data report the first evidence suggesting that low-level amplifications in AURKA, which conventional sequencing platforms miss, can provoke meaningful changes in gene expression as assessed via RNA transcriptome analysis. Further, low-level AURKA amplifications predict inferior outcomes on CDK4/6i for pts with HR+ MBC. Real-world clinical data for pts whose tumors harbored RB1 mutations, which have been associated with alisertib sensitivity in both the preclinical and clinical settings, trended with shorter rwPFS on CDK4/6i in this dataset and will be presented at the meeting. Efforts to refine these and other potential biomarkers that may guide clinical deployment of alisertib are ongoing.

VIEW THE PUBLICATION