Monitoring PD-L1 expression in circulating cancer associated cells for prediction of clinical outcomes in metastatic breast cancer patients treated with immune checkpoint inhibitors.

Abstract

e14535 Background: Anti-programmed cell death ligand PD-L1/PD-1 immune checkpoint inhibitors (ICIs) are approved for metastatic breast cancer (mBC) patients (pts) with a tumor combined positive score (CPS) ≥10% in their tissue. However, many pts with < 10% CPS may also respond to ICIs. Recent studies in lung cancer have shown that circulating tumor cells (CTCs) and tumor-macrophage fusion cells (TMFCs) both express, and may upregulate, PD-L1 which is predictive for ICI responses. We initiated a prospective study to monitor PD-L1 in CTCs & TMFCs in mBC pts prior to receiving systemic ICIs to correlate their PD-L1 expression to progression free survival (PFS) and overall survival (OS) at 24 months. Methods: We conducted a multi-institutional prospective pilot study of n = 102 pts with pathologically confirmed mBC, having failed ≥1 prior lines of therapy, prior to receiving ICIs (n = 52 Pembrolizumab, n = 39 Retifanlimab, n = 5 Atezolizumab, and n = 1 Nivolumab). Anonymized blinded peripheral blood samples (7.5ml) were procured prior to ICI induction (T0) and (~40 days after ICI induction (T1). TMFCs & CTCs were isolated using Cellsieve microfilters with average PD-L1 expressions measured then categorized as negative/low or high. Tumor CPS was categorized as negative/low (< 10%) or high (> 10%). We compared TMFC PD-L1 to tissue CPS by linear regression. TMFC/CTC PD-L1 expression and tissue PD-L1 CPS were compared to pts’ PFS and OS by Cox proportional univariate and multivariate analysis over 24 months. Results: Of n = 102 pts, median age = 59 years (range 32-77); 42% (43/102) were ER+, 33% (33/102) were PR+, 5% (5/102) were HER2+, and 56% (57/102) were triple-negative. 32% (32/102) had an ECOG score of 0, 25% (26/102) had ECOG 1, and 43% of scores were unavailable. N = 42 pts had available tumor CPS data, which did not correlate to TMFC/CTC PD-L1 at T0 or T1. Pts with high CPS did not significantly predict for PFS (HR = 0.78; 95% CI, 0.3-1.9; P = 0.7627) nor OS (HR = 0.54; 95% CI, 0.2-1.7; P = 0.4471). Pts with high TMFC/CTC PD-L1 at T0 also did not significantly predict for PFS (HR = 0.74; 95% CI, 0.5-1.2; P = 0.2799) nor OS (HR = 0.84; 95% CI, 0.4-1.6; P = 0.7220). However, of the n = 82 pts who provided T1 samples, high CTC/TMFC PD-L1 at T1 was significant for better PFS (HR = 2.55; 95% CI, 1.6-4.2; P = 0.0004) and better OS (HR = 2.06; 95%CI, 1.1-3.8; P = 0.0309). Conclusions: In this study of mBC pts treated with PD-L1 ICIs, tissue CPS did not correlate with PD-L1 expression in CTCs or TMFCs. Further, neither CPS nor baseline PD-L1 in CTC/TMFCs predicted for clinical outcomes in ICI therapies. However, PD-L1 in CTC/TMFCs ~40 days post-induction of ICI did predict for better response rates. While this study suggests predictive value of monitoring PD-L1 in blood during ICI therapies, further studies are required to refine and validate these findings.