Tumor Subtype Classification Tool for HPV-associated Head and Neck Cancers

Area of Science:

• Oncology
• Genomics
• Bioinformatics

Background:

• Human Papillomavirus-associated Head and Neck Squamous Cell Carcinoma (HPV+ HNSCC) exhibits distinct molecular subtypes: IMU (immune strong) and KRT (highly keratinized).
• These subtypes possess different carcinogenic pathways, clinical outcomes, and may require tailored treatment approaches.
• A standardized method for classifying HPV+ HNSCC tumors into these subtypes is currently lacking.

Purpose of the Study:

• To develop and validate a robust machine learning classifier for standardizing HPV+ HNSCC subtyping.
• To investigate the clinical, demographic, and molecular features associated with each subtype through a meta-analysis of patient cohorts.

Main Methods:

• RNA sequencing (RNA-seq) data from 229 HPV+ HNSCC samples across four cohorts were utilized.
• An ensemble machine learning approach, incorporating five algorithms and multiple gene sets, was employed for classification.
• Classifier performance was rigorously tested and validated on independent patient cohorts.

Main Results:

• The developed classifier achieved 100% accuracy in the test set and demonstrated successful separation in validation cohorts.
• Significant associations were identified between subtypes and 24 of 39 tested clinicodemographic and molecular variables.
• The IMU subtype correlated with epithelial-mesenchymal transition and immune cell infiltration, while the KRT subtype showed higher keratinization and a higher proportion of female patients.

Conclusions:

• This study presents a reliable RNA-seq-based classifier for subtyping HPV+ HNSCC into IMU and KRT categories.
• The findings enhance the understanding of HPV+ HNSCC molecular subtypes and their associated features.
• The developed tool and insights are crucial for advancing personalized treatment strategies in HPV+ HNSCC.