Association of neutrophil-lymphocyte ratio (NLR) with clinical outcomes in patients (pts) with recurrent/metastatic head and neck squamous cell carcinoma (R/M HNSCC) treated with combination of cetuximab and nivolumab (C+N).

Abstract

6021 Background: Combination of anti-epidermal growth factor receptor (EGFR) and anti-programmed cell death-1 (PD-1) inhibitors such as C+N is highly efficacious in R/M HNSCC. While PD-1 ligand 1 (PD-L1) combined positive score (CPS) is predictive of anti-PD-1 inhibitor response, there is no predictive biomarker for the combination. NLR is a known prognostic biomarker in HNSCC pts treated with chemotherapy with/without radiation. We evaluated NLR as a potential predictive biomarker in R/M HNSCC pts treated with C+N. Methods: The clinical data were obtained from R/M HNSCC pts treated with C+N from a completed clinical trial (NCT03370276) and pts treated with nivolumab or pembrolizumab (N/P) as a standard of care in a retrospective chart review study. NLR were collected at baseline (NLR1) and on-treatment (1-month from the treatment initiation; NLR2). The median NLR was applied as a cut-point. Pts were stratified into NLR-high (H; >median) or -low (L; < median). Time-to-event endpoints were evaluated to assess survival and response at the initiation treatments using the Kaplan Meier and Cox Proportional Hazards Models. Selected variables included in the multivariate (MV) analyses were log2(NLR1 or 2), PD-L1 CPS, p16 and smoking status, and prior exposure to anti-PD-1 inhibitors. Results: Total 83 C+N and 40 N/P pts were included. In C+N, the median NLR cut-points were NLR1 6.27 and NLR2 5.98 while they were NLR1 5.87 and NLR2 8.21 in N/P. NLR1-H vs. -L in C+N showed no difference in OS (p = 0.26) and progression-free survival (PFS; p = 0.95), but NLR2-H had worse OS (P < 0.001) and PFS (P < 0.001) compared to NLR2-L. Similarly, in N/P, NLR1-H vs. -L did not show difference in OS (p = 0.23), but NLR2-H had worse OS (P < 0.046) compared to NLR2-L. In C+N, NLR2 (log2(NLR2)) consistently correlated with worse OS [univariate (UV), HR 1.69, CI 1.36-2.12; MV, HR 1.63, CI 1.26-2.11] and PFS (UV, HR 1.63, CI 1.22-2.19; MV, HR 1.70, CI 1.21-2.42). In N/P, PFS and MV analyses were not feasible due to lack of reliable data in the retrospective study. In C+N, NLR2 (p < 0.001), but not NLR1 (p = 0.93), was correlated with lower best overall response (BOR). Similarly, in N/P, NLR2 (p = 0.004), but not NLR1 (p = 0.12), was associated with lower BOR. In addition, a significantly larger reduction in NLR on-treatment from baseline (log2(NLR2) – log2(NLR1)) was observed in C+N compared to N/P (P = 0.005, Student t-test). Conclusions: On-treatment NLR2-H is associated with worse response and survival in pts with R/M HNSCC treated with C+N and N/P. C+N is associated with a greater reduction in NLR compared to N/P, suggesting that the combination enhances a favorable peripheral immunologic response. Further evaluation is warranted to determine the role of NLR as a biomarker of immunotherapy response.