Temporal and spatial discordance of programmed cell death-ligand 1 expression and lymphocyte tumor infiltration between paired primary lesions and brain metastases in lung cancer

The tumor microenvironments of paired primary lung cancers and brain metastases are significantly different, such that many of the metastases lose PD-L1 expression, lymphocyte infiltration or both with greater discrepancies over time. The spatial and temporal heterogeneity of PD-L1 expression may limit its use as a tissue-based predictive biomarker in lung cancer.The dynamics of PD-L1 expression may limit its use as a tissue-based predictive biomarker. We sought to expand our understanding of the dynamics of PD-L1 expression and tumor-infiltrating lymphocytes (TILs) in patients with lung cancer-related brain metastases. Paired primary lung cancers and brain metastases were identified and assessed for PD-L1 and CD3 expression by immunohistochemistry. Lesions with 5% or greater PD-L1 expression were considered positive. Agreement statistics and the chi(2) or Fisher's exact test were used for analysis. We analyzed 146 paired lesions from 73 cases. There was disagreement of tumor cell PD-L1 expression in 10 cases (14%, kappa = 0.71), and disagreement of TIL PD-L1 expression in 19 cases (26%, kappa = 0.38). Most paired lesions with discordant tumor cell expression of PD-L1 were obtained 6 or more months apart. When specimens were categorized using a proposed tumor microenvironment categorization scheme based on PD-L1 expression and TILs, there were significant changes in the classifications because many of the brain metastases lacked either PD-L1 expression, tumor lymphocyte infiltration or both even when they were present in the primary lung cancer specimens (P = 0.009). We identified that there are significant differences between the tumor microenvironment of paired primary lung cancers and brain metastases. When physicians decide to treat patients with lung cancer with a PD-1 or PD-L1 inhibitor, they must do so in the context of the spatial and temporal heterogeneity of the tumor microenvironment.