Background: CDK4/6 inhibitors increase tumor immunogenicity in preclinical models of breast cancer and several trials combining CDK4/6 inhibitors and anti-PD1/PDL1 therapies are underway. However, immune response data in tumor samples from patients (pts) treated with CDK4/6 inhibitors are scarce. Here, we present exploratory results of the CORALLEEN trial, which evaluated the efficacy of ribociclib and endocrine therapy in patients with high-risk Luminal B disease (Gavilá et al. submitted to SABCS 2019).
Methods: CORALLEEN is a randomized exploratory study in postmenopausal women with operable stage I-III breast cancer, hormone receptor positive (HR+)/HER2-negative and Luminal B by Prosigna®. Pts were randomized 1:1 to receive either 6 cycles of ribociclib (600mg; 3-weeks-on/1-week-off) plus daily letrozole (R+L) or chemotherapy (CHT): 4 cycles of AC followed by 12 doses of weekly paclitaxel. The primary endpoint is the rate of PAM50 Risk of Relapse (ROR)-low disease at surgery in each arm. Baseline and surgical specimens were also collected for stromal tumor infiltrating lymphocyte (TIL) determination and gene expression analysis. Expression of 770 genes and 31 biological signatures were determined using the Breast360TM panel (nCounter). In order to identify genes whose expression correlated with TIL, a significance of microarrays (SAM) quantitative analysis was used with a false discovery rate (FDR) <5%. Finally, interaction tests between each variable and tumor ROR response (i.e. relative decrease in ROR score) according to type of therapy were explored in logistic regression models.
Results: From July-2017 to Nov-2018, 198 pts were screened and 106 (54%) pts with Luminal B disease were recruited. baseline characteristics were: mean age 64, mean tumor size 3.8 cm, N+ (39%), mean Ki67 33.2%; 86.8% of pts were ROR-high. A total of 95 pts (90%) were included in this analysis (46 pts in the CHT arm and 49 in the R+L arm). At baseline, 21.7% and 32.7% of pts in the CHT and R+L arms had ≥10% of TILs, respectively. Compared to baseline, no significant change patterns in TILs expression were observed. In the CHT arm, 32.6%, 28.3 % and 39.1% of tumors increased, decreased or did not show any change in TILs. In the R+L arm, 30.6%, 44.9% and 24.5% of tumors increased, decreased or did not show any change in TILs. At surgery, 15.2% of pts in the CHT arm and 26.5% of pts in the R+L arm had ≥10% of TILs. Moreover, 5 of the 13 pts (38.5%) with ≥10% of TILs at surgery following R+L had TILs < 10% at baseline. In both arms, high expression of immune-related genes and signatures tracking CD8 T-cells (i.e. CD8A, PD1, LAG3 and CD8T-cell signature) were found associated with high TILs (FDR<5%). when immune response was evaluated based on tumor ROR response (as a continuous variable), high TILs at surgery were associated with better response to CHT but not to R+L (interaction P=0.03). In the CHT arm, mean % of TILs at surgery in low and high ROR responders (defined as <50% or ≥50% relative decrease in ROR score) were 4.2% and 11.2%, respectively. In the R+L arm, mean % of TILs at surgery in low and high ROR responders were 16.2% and 6.1%, respectively. Similar results were found with genes/signatures such as CD8A mRNA (interaction P=0.03), CD8 T-cell signature (interaction P=0.04), Tumor Inflammation Signature (interaction P=0.04) and GZMA mRNA (Granzyme A, interaction P=0.02).
Conclusions: An increase in TILs following 24-weeks of R+L occurs in ~30% of pts with high-risk Luminal B tumors, regardless of tumor ROR response. These findings suggest that immune checkpoint blockade might be an interesting strategy to explore in patients with a low ROR response after R+L.