Tumor cell budding as a prognostic and potentially therapeutically targetable biomarker in head and neck cancer.

Abstract

6053 Background: Several studies have demonstrated a negative prognostic impact of tumor cell budding (TCB) in HPV- head and neck squamous cell carcinoma (HNSCC), but analyses of its prognostic impact in HPV+ HNSCC and of the underlying molecular alterations are lacking. Methods: The study cohort included 331 HPV- and HPV+ HNSCC from TCGA with digitalized H&E-stained slides available. Corresponding mutation, methylation, and gene expression data were obtained from the pan-cancer atlas web page. Tumor buds were defined as clusters of up to four tumor cells separated from the tumor mass. The numbers of tumor buds were evaluated in ten digital high-power fields by a senior pathologist. A two-tier cellular dissociation grading system was introduced by separating tumors with six or more tumor buds (TCB-high) from tumors with fewer tumor buds (TCB-low). The impact of TCB on overall survival (OS) was analyzed using Cox proportional hazard models. Results with p < 0.05 were considered significant. Association of TCB with mutations was analyzed using the Wilcoxon test. Association of TCB with gene expression and methylation was analyzed using Spearman correlations. Lists of altered genes were compiled correcting the p-values with the Benjamin-Hochberg method and controlling the FDR at 5%. Results: In a univariate analysis, OS was significantly shorter in TCB-high tumors compared to TCB-low tumors in HPV- HNSCC (HR = 1.4, 95% CI 1.0-2.3) and HPV+ HNSCC (HR = 5.0, 95% CI 1.9-13.7). Shorter OS in TCB-high HNSCC was confirmed in a multivariate analysis including age, sex, HPV status, tumor site, tumors stage, and tumor margin status (HR = 1.7, 95% CI 1.2-2.4). Significant association of TCB with mutations was detected for two genes: NSD1 mutations correlated negatively with TCB in HPV- HNSCC, while TP53 mutations correlated positively with TCB in HPV+ HNSCC. Methylation of 126 (1%) genes was associated with TCB in HPV- HNSCC, while methylation of 511 (3%) genes was associated with TCB in HPV+ HNSCC. Expression of 422 (2%) genes was associated with TCB in HPV- HNSCC, while expression of 786 (4%) genes was associated with TCB in HPV+ HNSCC. Among these genes, those annotated to the epithelial mesenchymal transition were highly significantly enriched in both HPV- HNSCC (5.5-fold enrichment) and HPV+ HNSCC (2.7-fold enrichment). Conclusions: Evaluation of TCB based on digital HE slides was conducted in a large clinically and molecularly characterized HNSCC cohort. The prognostic impact of TCB could be validated in HPV- HNSCC, while a prognostic impact of TCB could be demonstrated for the first time in HPV+ HNSCC. TCB correlated with mutations of only two genes and theses correlations were imperfect. A plethora of genes correlated with TCB on the level of methylation and gene expression level. These genes should be further analyzed and prioritized for the evaluation of therapeutic targeting.