The aim of this investigation was to see if a panel of biomarkers (maspin, CD105, and nm23-H1) could be used to stratify patients with laryngeal squamous cell carcinoma (LSCC) in homogeneous disease recurrence risk clusters.

Cluster analysis was used to classify 89 patients based on their immunohistochemical expression of nm23-H1, CD105, and maspin.

Our analysis identified seven homogeneous clusters: the LSCC recurrence rate was lowest in cluster 6 (non-nuclear maspin pattern, nuclear nm23-H1 expression ≥10%, endothelial CD105 expression <6%; P = .009), and highest in cluster 3 (non-nuclear maspin pattern, nuclear nm23-H1 expression <10%, endothelial CD105 expression ≥6%; P <.001).

Similar panels of biological variables identified by cluster analysis should be tested in prospective clinical trials to establish whether treating patients identified as being at higher risk of LSCC recurrence more aggressively could significantly improve their recurrence rate and/or disease-specific survival.