Follicular Helper T-Cell-Based Classification of Endometrial Cancer Promotes Precise Checkpoint Immunotherapy and Provides Prognostic Stratification.

ABSTRACT

Despite the fact that management of EC is moving towards four TCGA-based molecular classifications, a pronounced variation in immune response among these molecular subtypes limits its clinical use. We aimed to investigate the determinant biomarker of ICI response in endometrial cancer (EC). We characterized transcriptome signatures associated with tumor immune microenvironment in EC. Two immune infiltration signatures were identified from the TCGA database (n = 520). The high- and low-infiltration clusters were compared for differences in patient clinical characteristics, genomic features, and immune cell transcription signatures for ICI prediction. A Lasso Cox regression model was applied to construct a prognostic gene signature. Time-dependent receiver operating characteristic curve, Kaplan-Meier curve, nomogram, and decision curve analyses were used to assess the prediction capacity. The efficacy of potential biomarker was validated by the Karolinska endometrial cancer cohort (n = 260). Immune signature profiling suggested that T follicular helper-like cells (Tfh) may be an important and favorable factor for EC; high Tfh infiltration showed potential for clinical use in the anti-PD-1 treatment. A Tfh Infiltration Risk Model (TIRM) established using eight genes was validated, and it outperformed the Immune Infiltration Risk Model. The TIRM had a stable prognostic value in combination with clinical risk factors and could be considered as a valuable tool in a clinical prediction model. We identified CRABP1 as an individual poor prognostic factor in EC. The Tfh-based classification distinguishes immune characteristics and predicts ICI efficacy. A nomogram based on Tfh-related risk score accurately predicted the prognosis of patients with EC, demonstrating superior performance to TCGA-based classification.