Use of baseline plasma circulating tumor DNA (ctDNA) to predict duration of endocrine therapy (ET) and CDK4/6 inhibitor (CDK4/6i) therapy (tx) and to analyze intrinsic vs acquired endocrine resistance.

Abstract

1075 Background: ET + CDK4/6i is standard-of-care for patients (pts) with hormone receptor-positive/HER2-negative (HR+/HER2-) metastatic breast cancer (MBC). We aimed to identify predictors ET + CDK4/6i tx duration and to compare genomic profiles in pts with intrinsic vs acquired resistance. Methods: Plasma samples were collected from pts with HR+/HER2- MBC enrolled in the EMBRACE cohort study who had plasma collection within 3 months (mo) prior to CDK4/6i initiation to 14 days after initiation. The primary outcome was duration of ET+CDK4/6i tx, defined as time from tx initiation to end of tx. Intrinsic resistance was defined as pts with tx duration < 180 days. Plasma samples were analyzed using the Guardant360 assay, which includes genotyping of > 700 genes and tumor fraction (TF) score. TF was estimated by normalizing cancer-specific differentially methylated regions with matched control regions in each sample. The predictive value of baseline TF (0 vs > 0) was tested using a Cox regression model including age, line of tx, and liver metastases. For comparison of pts with intrinsic vs acquired resistance, analysis was limited to samples with TF >1% to minimize the impact of variation in tumor shed. Gene frequency between intrinsic and acquired resistance samples were compared using q-tests(q < 0.25). Results: A total of 188 pts were included. Median age at MBC diagnosis was 57.5 yrs. ET+CDK4/6i was given in the first-line (1L) in 115 pts, second-line (2L) in 37 pts, and > 2L in 36 pts. Of 167 pts, TF was undetectable (TF = 0) in 19 (11%) and detectable (TF > 0) in 148 (89%). In Cox regression, baseline TF (p = 0.001), line of tx (n = 0.002), and presence of liver metastasis (p = 0.014), but not age, were predictors of duration of tx. Median duration of tx was 44.6 mo in pts with baseline TF = 0 vs. 5.8 mo in pts with baseline TF > 0 (HR 0.28, 95% CI 0.14-0.56). Similar results were found when restricting the analysis to those receiving tx in the 1L or 2L. There were notable differences in the frequency of ESR1 (63% vs 48%), CDH1 (38% vs 18%), PTEN (21% vs 9%), RB1 (32% vs 20%), and CDKN2A (20% vs 5%) alterations in pts with intrinsic vs acquired resistance, though these did not reach statistical significance in the setting of small sample size. ESR1 fusions were seen in 14% (8/56) pts with intrinsic resistance vs 7% (3/44) pts with acquired resistance. Among pts with intrinsic resistance, ERBB2 copy number loss was present in 7(13%) (6 het loss, 1 homozygous deletion), RB1 copy number loss in 10 (18%) (all het loss), and CDKN2A copy number loss in 9(16%) (7 het loss, 2 homozygous deletions). Conclusions: Baseline TF in ctDNA is highly predictive of time on ET+CDK4/6i tx in pts with HR+/HER2- MBC. Baseline genomic profiles differ qualitatively between pts with intrinsic vs acquired resistance. If validated, baseline plasma may provide a valuable tool in tx selection.