Evaluating Ki67 and Oncotype DX Breast Recurrence Score during neoadjuvant treatment with letrozole/abemaciclib or chemotherapy in patients with highly proliferative HR+/HER2- breast cancer participating in the GEICAM CARABELA trial.

Abstract

576 Background: CARABELA (NCT04293393) is a phase 2 randomized trial comparing 12 months of neoadjuvant letrozole/abemaciclib (let/abema) vs. 6 months of neoadjuvant chemotherapy (CT) for stage II-III, Ki67≥20%, HR+/HER2- breast cancer patients. Let/abema failed to show similar residual cancer burden (RCB) 0-I rates compared to CT (13% vs. 18%), highlighting the need to identify response biomarkers to CDK4/6 inhibitors and to determine which patients cannot avoid CT. Methods: We explored the predictive role of baseline (BL) Ki67 and Recurrence Score (RS) in tumor samples, and their changes during treatment. Ki67 was centrally determined using a standardized assay (Ki67 MIB-1 pharmDx, Dako Omnis, Agilent Technologies) at BL, 2 weeks (2w), and surgery. RS result was assessed at BL and at surgery (Exact Sciences), in the case of RCB=0, RS and Ki67 was assigned a value of 0. We examined the association of BL Ki67 and RS with RCB 0-I rates as independent variables, by logistic regression (adjusting by disease stage and cN), checking for treatment interaction. Results: BL Ki67 and RS showed similar distributions between arms; median Ki67: let/abema 38% vs CT 36% ( P=0.528); median RS: let/abema 27 vs CT 30 ( P=0.188). Although we could not detect a differential treatment effect based on BL Ki67 or RS, rates of RCB0-I were higher in the CT arm both for high Ki67 (≥40%) and high RS (>25) (table 1). In addition, BL Ki67 as a continuous variable was associated to RCB 0-I only in the CT arm (OR 0.954; 95%CI 0.922-0.988, P=0.006), with an interaction P=0.05. Ki67 suppression was higher in let/abema arm, with a lower median Ki67 at 2w: let/abema 0% vs. CT 21.7% ( P<0.0001), but not at surgery: let/abema 0% vs. CT 2.25% ( P=0.157). In addition, rates of Ki67 <2.7% were higher with let/abema compared to CT at 2w (68% vs. 6%, P<0.0001), but not at surgery (53% vs. 45%, P=0.157). Moreover, 17% of patients in the let/abema arm shifted from 2w Ki67<2.7% to surgery Ki67≥2.7%, while only 2% did so with CT ( P<0.0001). Median RS at surgery was higher in let/abema arm (22 vs. 19 in CT, P=0.028), and CT induced higher downstaging from BL high (>25) to low (≤25) RS at surgery (36% vs. let/abema 20%, P=0.024). Conclusions: Highly proliferative tumors Ki67 (≥40%) or those with high Recurrence Score (>25) exhibited higher rates of RCB 0-I when treated with chemotherapy, than when treated with letrozole/abemaciclib. This suggests that relying solely on letrozole/abemaciclib as systemic treatment for these tumors may be insufficient. Clinical trial information: NCT04293393 . [Table: see text]