Cluster Analysis According to Immunohistochemistry is a Robust Tool for Non-Small Cell Lung Cancer and Reveals a...

Clustering in medicine is the subgrouping of a cohort according to specific phenotypical or genotypical traits. For breast cancer and lymphomas, clustering by gene expression profiles has already resulted in important prognostic and predictive subgroups. For non-small cell lung cancer (NSCLC), however, little is known. We performed a cluster analysis on a cohort of 365 surgically resected, well-documented NSCLC patients, which was followed-up for a median of 62 months, incorporating 70 expressed proteins and several genes. Our data reveal that tumor grading by architecture is significant, that large cell carcinoma is likely not a separate entity, and that an immune signature cluster exists. For squamous cell carcinomas, a prognostically relevant cluster with poorer outcome was found, defined by a high CD4/CD8 ratio and lower presence of granzyme B+ tumor-infiltrating lymphocytes (TIL). This study shows that clustering analysis is a useful tool for verifying established characteristics and generating new insights for NSCLC. Importantly, for one "immune signature" cluster, the signature of the TIL (especially the amount of CD8+ TIL) was more crucial than the histologic or any other phenotypical aspect. This may be an important finding toward explaining why only a fraction of eligible patients respond to immunomodulating anticancer therapies.