Monitoring PD-L1 expression in cancer-associated macrophage-like cells as predictor of clinical outcomes in metastatic cancer patients treated with PD-L1 immunotherapies.

Abstract

2555 Background: Studies have described the efficacy of immunotherapies (IMT) utilizing programmed death 1 receptor and its ligand (PD-L1) for treating solid tumors. However, many patients (pts) fail to respond to IMT, necessitating better predictive biomarkers for improved stratification. Poor IMT responses are often attributed to the dynamic nature of PD-L1 likely changing after chemotherapy or radiation, but typically quantified by static immunostaining. Recent studies have described PD-L1 upregulation in giant phagocytic stromal cells, i.e. Cancer associated macrophage-like cells (CAML), circulating macrophages that engulf tumor before entering circulation and may predict IMT responses. We conducted a pilot study to monitor the peripheral blood of n = 111 metastatic cancer pts undergoing systemic treatment with IMT in combination with other therapies, to evaluate CAML PD-L1 prior to & post IMT induction with clinical correlation at 2 years. Methods: In a prospective pilot study of n = 111 metastatic cancer pts, breast (n = 42), lung (n = 46), renal cell (n = 10), prostate (n = 5), esophageal (n = 5) & colon (n = 3), all starting new lines of systemic chemotherapy in combination with IMT (pembrolizumab [n = 69], nivolumab [n = 23], Durva [n = 4], or atezolizumab [n = 13]) for new recurrent metastasis (n = 45) or with previously treated progressive metastatic disease (n = 66). We isolated CAMLs from 7.5 ml baseline (T0) blood using the LifeTracDx PD-L1 test and scored PD-L1 as high or low. If possible, a follow-up sample (T1) was taken (~56 days) after IMT induction. Pts’ progressive free survival (PFS) and overall survival (OS) hazard ratios (HRs) were analyzed by censored univariate analysis based on RECIST v1.1 over 2 years. Results: T0 PD-L1 CAML data was available for 78% (n = 86/111) of pts, with 34% (n = 29/88) having high CAML PD-L1 which was not correlated with improved PFS (HR = 0.99, p = 0.9416, CI = 0.6-1.6) or OS (HR = 1.1, p = 0.9472, CI = 0.6-1.8). T1 PD-L1 CAML data was available for 74% (n = 82/111) of pts, with 44% (n = 36/82) having high CAML PD-L1, which significantly correlated with improved PFS (HR = 3.1, p = 0.0002, CI = 1.8-5.5) & OS (HR = 6.6, p < 0.0001, CI = 3.4-12.7). In comparing CAML PD-L1 change post IMT, it was found that consistently low PD-L1 at T0 & T1 had the poorest responses, median PFS (mPFS) = 4.9 months & median OS (mOS) = 7.2 months. In contrast, consistently high PD-L1 at T0 & T1 had better responses, mPFS = 8.4 months & mOS = 18.3 months. Further, pts who increased in CAML PD-L1 after IMT had the best OS response rates, mPFS = 3.8 months & mOS = 20.9 months. Conclusions: We utilized a cancer agnostic blood-based biopsy to monitor PD-L1 changes in circulating tumor immune cells and predict clinical benefit to PD-L1 IMTs in several cancer types. While this initial pilot study appears to stratify pts with better IMT responses, larger validation studies are needed.