Multi-omics analysis identifies a liquid-liquid phase separation-related subtypes in head and neck squamous cell carcinoma

TL;DR

This study identifies three subtypes of head and neck cancer based on liquid-liquid phase separation patterns and develops a tool to predict prognosis and treatment response.

Contribution

A novel classification system and prognostic signature for HNSCC based on LLPS-related genes and immune microenvironment patterns.

Findings

Three distinct LLPS subtypes (LS1, LS2, LS3) were identified with differences in prognosis and immune response.

The LPRS showed strong predictive power for HNSCC prognosis and immunotherapy response across multiple cohorts.

Potential small-molecule compounds targeting LLPS were identified for HNSCC treatment.

Abstract

Growing evidence indicates that abnormal liquid–liquid phase separation (LLPS) can disrupt biomolecular condensates, contributing to cancer development and progression. However, the influence of LLPS on the prognosis of head and neck squamous cell carcinoma (HNSCC) patients and its effects on the tumor immune microenvironment (TIME) are not yet fully understood. Therefore, we aimed to categorize patients with HNSCC based on LLPS-related genes and explored their multidimensional heterogeneity.

We integrated the transcriptomic data of 3,541 LLPS-related genes to assess the LLPS patterns in 501 patients with HNSCC within The Cancer Genome Atlas cohort. Subsequently, we explored the differences among the three LLPS subtypes using multi-omics analysis. We also developed an LLPS-related prognostic risk signature (LPRS) to facilitate personalized and integrative assessments and then screened and validated potential therapeutic small molecule compounds targeting HNSCC via experimental analyses.

By analyzing the expression profiles of 85 scaffolds, 355 regulators, and 3,101 clients of LLPS in HNSCC, we identified three distinct LLPS subtypes: LS1, LS2, and LS3. We confirmed notable differences among these subtypes in terms of prognosis, functional enrichment, genomic alterations, TIME patterns, and responses to immunotherapy. Additionally, we developed the LPRS, a prognostic signature for personalized integrative assessments, which demonstrated strong predictive capability for HNSCC prognosis across multiple cohorts. The LPRS also showed significant correlations with the clinicopathological features and TIME patterns in HNSCC patients. Furthermore, the LPRS effectively predicted responses to immune checkpoint inhibitor therapy and facilitated the screening of potential small-molecule compounds for treating HNSCC patients.

This study presents a new classification system for HNSCC patients grounded in LLPS. The LPRS developed in this research offers improved personalized prognosis and could optimize immunotherapy strategies for HNSCC.

Figures