Background: Recent studies suggest differences in outcomes among patients (pts) with metastatic breast cancer (MBC) treated with abemaciclib, ribociclib, or palbociclib, but whether these differences have a genomic basis is unknown. Here, we utilize a large real-world dataset to compare the mutational landscapes of HR+/HER2- MBC samples in which CDK4/6 inhibitor (CDK4/6i) treatment was initiated ≥6 months prior to biopsy to describe variations in tumor biology associated with exposure to each CDK4/6i. We also compare mutations detected by solid tumor sequencing and liquid biopsy to better understand each assay’s ability to identify relevant alterations in this population.

Methods: De-identified data from a cohort of pts with HR+/HER2- MBC (n=1172) sequenced with the Tempus xT (DNA-seq of 595-648 genes, whole exome-capture RNA-seq) solid tumor and xF (105-gene panel focused on detecting oncogenic and resistance mutations from cell-free DNA) liquid biopsy assays was retrospectively analyzed. For pts with multiple samples sequenced, the most recent sample was analyzed. Pts were selected based on receipt of CDK4/6i between metastatic diagnosis date and biopsy collection and excluded if <6 months elapsed between CDK4/6i initiation and biopsy collection. Demographics, clinical characteristics, and NGS findings were compared between groups by Chi-squared/Fisher’s Exact tests or Kruskal-Wallis tests, as applicable. The prevalence of individual gene alterations (consisting of pathogenic/likely pathogenic SNVs/indels and copy number alterations) were compared similarly with adjustment for false-discovery.

Results: We compared the immune biomarker and DNA mutational landscapes of 1172 samples collected after a period of treatment with abemaciclib, ribociclib, or palbociclib. Across all pts, the most commonly altered genes were TP53, PIK3CA, ESR1, CDH1, and GATA3. Abemaciclib-treated pts had the highest median TMB and MSI-high frequency (Table). Palbociclib-treated pts were less likely to have a high TMB (≥10 mutations/megabase) or RB1 mutations, a known biomarker of resistance to CDK4/6i (Table). We note that the total N for pts positive for TMB high or MSI-high was very low across all groups. We also compared DNA mutational landscapes between pts tested with solid tumor sequencing and liquid biopsy; the lower prevalence of RB1 mutations in palbociclib-treated pts trended towards significance in both groups (Table).

Conclusions: Results from our real-world dataset suggest that treatment with the different CDK4/6i drugs results in unique immune biomarkers and DNA mutational profiles. Detection of relevant alterations such as RB1 by both tissue testing and liquid biopsy supports a role for either assay in identifying mutations associated with CDK4/6i. Our findings raise the possibility that unique targeted treatment strategies and combination therapies may be warranted after progression on the different CDK4/6i drugs. Additional investigation into the differences in genomic alterations and outcomes among CDK4/6i drugs is necessary to further explore the hypotheses generated by this real-world study.

Table: Prevalence of immune biomarkers and DNA mutations in pts treated with each CDK4/6i drug

1Pearson’s Chi-squared test; Fisher’s exact test; Kruskal-Wallis rank sum test

2False discovery rate correction for multiple testing

3,4N=464; 708 pts with missing data

5N=1163; 9 pts with missing data

VIEW THE POSTER