CDK4/6i-treated HR+/HER2- breast cancer tumors show higher ESR1 mutation prevalence and more altered genomic landscape

Abstract

As CDK4/6 inhibitor (CDK4/6i) approval changed treatment strategies for patients with hormone receptor-positive HER2-negative (HR+/HER2-) breast cancer (BC), understanding how exposure to CDK4/6i affects the tumor genomic landscape is critical for precision oncology. Using real-world data (RWD) with tumor genomic profiling from 5910 patients with metastatic HR+/HER2- BC, we investigated the evolution of alteration prevalence in commonly mutated genes across patient journeys. We found that ESR1 is more often altered in tumors exposed to at least 1 year of adjuvant endocrine therapy, contrasting with TP53 alterations. We observed a similar trend after first-line treatments in the advanced setting, but strikingly exposure to aromatase inhibitors (AI) combined with CDK4/6i led to significantly higher ESR1 alteration prevalence compared to AI alone, independent of treatment duration. Further, CDK4/6i exposure was associated with higher occurrence of concomitant alterations in multiple oncogenic pathways. Differences based on CDK4/6i exposure were confirmed in samples collected after 2L and validated in samples from the acelERA BC clinical trial. In conclusion, our work uncovers opportunities for further treatment personalization and stresses the need for effective combination treatments to address the altered tumor genomic landscape following AI+CDK4/6i exposure. Further, we demonstrated the potential of RWD for refining patient treatment strategy and guiding clinical trial design.

Fig. 1. Cohort selection and study design.

a Selection of patients with HR+/HER2- BC in the CGDB. b Attrition of the eBC cohorts in our study. c Attrition of the 1L cohorts in our study and split into pre- and post-1L groups based on the line of treatment and timing of the sample collection for CGP testing. d Patients are selected based on their 1L regimen and are divided into subgroups based on the timing of their CGP. The effect of treatment on gene alterations is estimated by comparing alteration prevalence in tumors profiled prior to the 1L versus tumors profiled after 1L. Stratified analyses are based on a stratum of patients defined by clinical variables (e.g. de novo vs. recurrent disease) or sample characteristics. eBC/mBC stands for early/metastatic breast cancer, HR+/HER2- for hormone-receptor positive HER2-negative, CGP for comprehensive genomic profiling, 1L for first-line treatment, ETR for endocrine treatment resistance, CGDB for clinico-genomic database, AI for aromatase inhibitor, Fulv for fulvestrant.

Fig. 2. Prevalence of ESR1 and TP53 alterations in advanced HR+ breast cancer tumors is associated with treatment duration and CDK4/6i exposure.

a, b Prevalence of ESR1 (a) and TP53 (b) alterations in samples collected from patients prior to 1L split by duration of eBC ET (endocrine treatment) prior to relapse. Error bars represent the 95% confidence interval based on bootstrapping. c Prevalence of ESR1 alterations prior or after 1L treatment in the advanced setting in different cohorts. Arrow represents the difference in prevalence: its origin is the median prevalence pre-treatment and its end is the median prevalence post-treatment. Error bars represent the 95% confidence interval of prevalence; Color represents fold-change magnitude; * stands for p < 0.05, ** for p < 0.01 based on bootstrapping; Treatment cohorts are labeled on the y-axis. d Prevalence of ESR1 alterations after 1L treatment stratified by the 1L treatment duration prior to sample collection. Error bars represent the 95% confidence interval of prevalence. e Prevalence of TP53 alterations prior or after 1L treatment in the advanced setting in different cohorts. Same legend as c. f Prevalence of TP53 alterations after 1L treatment stratified by the 1L treatment duration prior to sample collection. Same legend as d.

Fig. 3. Prevalence of genomic alterations in tumors prior and after 1L treatment in the advanced setting.

a–d Median prevalence of alterations prior to (x-axis), or after (y-axis) treatment for a AI therapies (b) AI+CDK4/6i therapies (c) fulvestrant + CDK4/6i therapies, or d chemotherapies. Each point is an individual gene; genes of interest are labeled. Error bars represent the 95% confidence interval; Color reflects fold-change; Shape significance with an FDR cutoff of 0.2 based on bootstrapping and Benjamini-Hochberg procedure.

Fig. 4. Prevalence of pathway-level alterations is increased in tumors exposed to CDK4/6i.

a–d Distribution of the number of altered gene sets for pre- and post-treatment groups of the (a) AI, or b AI+CDK4/6i cohorts, as well as for the (c) pre-treatment and d post-treatment groups of AI-based cohorts. P-values based on a Kolmogorff-Smirnov test. e, f Distribution of the number of altered gene sets for the post-treatment samples of the (e) AI and f AI+CDK4/6i cohorts based on the 1L treatment duration prior to sample collection. g Distribution of the number of altered gene sets of samples collected after 2L or 3L based on treatment cohorts. P-value between samples exposed to CDK4/6i (left) or not (right) based on a Kolmogorff-Smirnov test.

Fig. 5. Prevalence of genomic alterations is higher for tumors from post-1L patients exposed to CDK4/6i in the acelERA trial.

a Median prevalence of alterations in samples from post-1L patients who received AI (x-axis) or AI+CDK4/6i (y-axis) as 1L treatment. Each point is an individual gene; genes of interest are labeled. Error bars represent the 95% confidence interval; Color reflects fold-change; Shape significance with an FDR cutoff of 0.2 based on bootstrapping and Benjamini-Hochberg procedure. b Distribution of the number of altered gene sets for the samples from post-1L patients who received AI or AI+CDK4/6i as 1L treatment. P-values based on a Kolmogorff-Smirnov test.