Background: Molecular surveillance adoption in hormone receptor-positive, HER2-negative metastatic breast cancer (HR+/HER2- mBC) has potential to inform critical treatment decisions in patients (pts) treated with aromatase inhibitor (AI) plus CDK4/6 inhibitor (CDK4/6i) therapy with detected ESR1 mutations (ESR1m). ESR1m patterns remain poorly characterized in real-world practice. We characterized ESR1m emergence and clinical impact in a large RWD cohort of HR+/HER2- mBC pts monitored with longitudinal testing.
Methods: Study cohort included pts in Tempus AI’s RWD, multimodal database with HR+/HER2- mBC diagnosis prior to January 2023 who received AI+CDK4/6i therapy, had reported results of comprehensive genomic profiling from tissue (Tempus xT) or liquid (Tempus xF) biopsy within 24 months (mo) of receiving AI+CDK4/6i therapy, and ≥1 on-treatment xF tests. Pts were stratified by ESR1m and ESR1-wildtype (ESR1wt) status during AI+CDK4/6i. Real-world overall survival (rwOS) was estimated using Kaplan-Meier survival curves. Survival differences between groups were assessed with a log-rank test. Followup was measured from AI+CDK4/6i therapy initiation. Delayed entry, due to first biopsy sequencing occurring after therapy initiation, was accounted for using risk set adjustment in rwOS estimates. Pts were censored at 60 mos after AI+CDK4/6i initiation or at last known followup. Pts who switched to elacestrant, ESR1m targeted therapy, were censored at therapy initiation. Incidence and timing of ESR1m emergence and treatment patterns post ESR1m detection were also assessed.
Results: Among 301 HR+/HER2-mBC pts treated with AI+CDK4/6i and undergoing longitudinal molecular testing, median time to on-treatment xF was 23.1 mo. Median time from AI+CDK4/6i initiation to ESR1m detection was 19.1 months overall, 21.0 months for patients with ESR1m detected during 1L therapy, and 12.3 months for those detected in 2L+ settings. Baseline characteristics were similar between ESR1m and ESR1wt groups (median age 64/63 years; 55%/ 57% White, 16%/12% Hispanic, 7%/7% Black, and 4%/4% Asian). ESR1m status was associated with reduced survival (median rwOS 44.0 mo vs. not reached for ESR1wt; log-rank test p=0.02). Median follow-up time overall was 33.8 mo. From on-treatment xF, median time to therapy switch was 41 days overall in ESR1m (41 days in 1L, 36 days in 2L+) and 107.5 days overall in ESR1wt (137.5 days in 1L, 81 days in 2L+). 18 pts went on elacestrant and 46 pts went on fulvestrant after ESR1m detection. ESR1m incidence was 33.9% overall (102/301 pts), 34.0% (69/203) in 1L and 33.7% (33/98) in 2L+ settings. From on-treatment xF, most common ESR1 point mutations were D538G (64.4%), Y537S (24.4%), Y537N (21.1%), Y537C (10.0%), and E380Q (5.6%). Multiple(M) ESR1 missense mutations were observed within individual tumor samples, with higher death rates in pts with multiple(M) than single(S) mutations(D538G(S): 38.5%; Y537S(S): 40.0%; Triple(M): 68.8%). Rare mutations (e.g., S463P, V422del) were also observed. PIK3CA co-mutations occurred in 35.9% (108 pts), more frequently in ESR1m pts (42.2%) than ESR1wt pts (32.7%) (chi-square test, p=0.13).
Conclusion: This large multimodal RWD outcome analysis from longitudinal molecular surveillance testing in HR+/HER2- mBC pts treated with AI+CDK4/6i sheds light on the continuum of ESR1m emergence and pt outcomes from 1L and beyond outside of clinical trial data. Our analyses show higher ESR1m incidence is associated with reduced survival regardless of line of therapy. RWD analyses also indicate the complexity of resistance within ESR1 gene including multiple mutations or co-mutations with PIK3CA and occurrence of rare ESR1m variants. The impact of these variants on therapeutic decision-making will be investigated in future research.
VIEW THE PUBLICATION
VIEW THE POSTER