ABSTRACT
PURPOSE: To identify a novel system for scoring intratumoral immune response that can improve prognosis and therapy decisions in early stage non-small cell lung cancer (NSCLC). METHODS/PATIENTS: Eighty-four completely resected stage I/II NSCLC without adjuvant therapy were classified by expression profiling using whole genome microarrays. An external cohort of 162 tumors was used to validate the results. Immune cells present in tumor microenvironment were evaluated semiquantitatively by CD20, CD79, CD3, CD8, CD4 and CD57 immunostaining. Univariate and multivariate analyses of variables associated with recurrence-free survival were performed. RESULTS: Initial molecular classification identified three clusters, one with significantly better RFS. A reduced two-subgroup classification and a 50-gene predictor were built and validated in an external dataset: high and low risk of recurrence patients (HR = 3.44; p = 0.001). Analysis of the predictor´s genes showed that the vast majority were related to a B/plasma cell immune response overexpressed in the low-risk subgroup. The predictor includes genes coding for unique B lineage-specific genes, functional elements or other genes that, although non-restricted to this lineage, have strong influence on B-cell homeostasis. Immunostains confirmed increased B-cells in the low-risk subgroup. Gene signature (p < 0.0001) and CD20 (p < 0.05) were predictors for RFS, while CD79 and K-RAS mutations showed a tendency. CONCLUSIONS: Favorable prognosis in completely resected NSCLC is determined by a B-cell-mediated immune response. It can be differently scored by a 50-gene expression profile or by CD20 immunostaining. That prognosis information not reflected by traditional classifications may become a new tool for determining individualized adjuvant therapies.
