Abstract
Background: In recent years, great improvement has been made in immunotherapies for non-small cell lung cancer (NSCLC). Current data have suggested that Programmed cell death ligand 1 (PD-L1) expression might not be an ideal marker for patient selection in isolation. Evidence has been increasing that alternative markers, such as neutrophil-to-lymphocyte ratio (NLR), a biomarker of systemic inflammation response (SIR) previously associated with outcomes in a variety of cancers including NSCLC, might be a predictor for patient selection and the response to therapy. No reports have examined the prognostic value of combination of PD-L1 expression and inflammatory markers such as NLR in NSCLC. This retrospective study explores the relationship between NLR and PD-L1 expression in NSCLC as well as the prognostic value of combination of PD-L1 expression and NLR. Method: We evaluated tumor PD-L1 expression in 235 surgically resected NSCLC cases by immunohistochemical analysis. Carcinoma cells showing membranous staining for PD-L1 were considered PD-L1-positive cells (Figure 1). Cases with ≥1% tumor membrane staining were considered PD-L1-positive. The association of clinicopathological characteristics with PD-L1 expression was assessed by univariate and multivariate analyses. Moreover, univariate and multivariate analyses were performed to evaluate the predictive impact of PD-L1 expression and other factors on disease-free survival (DFS) and overall survival (OS). Result: PD-L1 protein expression was elevated in 34.0% of patients at cut-off value of 1%. Univariate analyses showed that PD-L1 expression was significantly higher in men (χ(2) =5.226, P=0.030), heavy smokers (χ(2) =18.650, P<0.001), and patients with squamous cell carcinoma (χ(2) =4.036, P=0.045). No correlations were noted between PD-L1 expression and age, EGFR mutation status or clinical stage. No significant correlations between PD-L1 protein expression and NLR were found. Multivariate logistic regression revealed that smoking index ≥400 was independent predictor of PD-L1 expression (odds ratio [OR], 3.375; P < 0.001). The results of univariate survival analyses showed that clinical stage (log-rank χ(2) =7.876, P=0.019) was associated with DFS. Smoking index (log-rank χ(2) =4.832, P=0.028), clinical stage (log-rank χ(2) =7.582, P=0.023) and adjuvant treatment (log-rank χ(2) =5.440, P=0.020) were significantly associated with OS. Neither PD-L1 expression nor NLR was found to be associated with DFS or OS. Of interest, when patients were divided in two groups according to combined PD-L1/NLR: patients with PD-L1+/ high NLR as Group 1, other patients as Group 2, Group 1 had significantly shorter DFS as well as OS than Group 2 (DFS: log-rank χ(2) =5.231, P=0.022, Figure 2A; OS: log-rank χ(2) =4.742, P=0.029, Figure 2B). In the multivariate analysis, Cox proportional hazards regression models showed that, PD-L1+/ high NLR was associated with a significantly shorter DFS and OS (hazard ratio [HR], 1.394, P=0.040; HR, 1.442, P=0.042, respectively). Stratified analysis showed that the prognostic value of combined PD-L1/NLR can only be observed in cases without epidermal growth factor receptor (EGFR) mutations (DFS: log-rank χ(2) =5.593, P=0.018, Figure 2C, OS: log-rank χ(2) =9.323, P=0.002, Figure 2D). In EGFR mutation subgroup, combination of PD-L1 expression and NLR has no relationship with DFS or OS. Conclusion: We found that combination of PD-L1 expression and NLR may be a promising prognostic indicator, and may also be a good marker for tumor recurrence, especially in the patients with wild-type EGFR.