Tumor Microenvironment CD8 T and Treg Cells–related Genes Signature Distinguishes Distinct Prognosis and Targeted Therapies Response in Endometrial Cancer

Although most endometrial cancer (EC) patients have a favorable prognosis, the overall survival (OS) of metastatic and recurrent EC could hardly be improved by the current chemoradiotherapy. We aimed to reveal the tumor microenvironment immune infiltration characteristics to elucidate the underlying mechanism of EC progression and guide clinical decisions. In the Cancer Genome Atlas (TCGA) cohort, Kaplan-Meier survival curves confirmed Tregs and CD8 T cells were prognosis-protective factors in OS of EC ( P <0.05). Weighted gene coexpression network analysis identified 2 gene modules closely correlated with Tregs and CD8 T-cell infiltration. We randomly split the TCGA EC cohort into the training and testing cohorts at a ratio of 7:3. An immune-related prognosis risk index (IRPRI), including NR3C1, E2F1, OTOG, TTK, PPP1R16B, and FOXP3, was established by univariate, Least Absolute Shrinkage and Selection Operator, and multivariate Cox regression with area under the curve >0.67. Distinct clinical, immune, and mutation characteristics existed between IRPRI groups by multiomics analysis. Cell proliferation and DNA damage repair-related pathways were activated, and immune-related pathways were inactivated in the IRPRI-high group. Furthermore, patients in the IRPRI-high group had lower tumor mutation burden, programmed death-ligand 1 expression, and Tumor Immune Dysfunction and Exclusion scores, indicating a poor response to immune checkpoint inhibitors therapy ( P <0.05), which was also validated in the TCGA testing cohort and independent cohorts, GSE78200, GSE115821, and GSE168204. Also, the higher mutation frequencies of BRCA1, BRCA2, and genes enrolled in homologous recombination repair in the IRPRI-low group predicted a good response to PARP inhibitors. Finally, a nomogram integrating the IRPRI group and prognosis significant clinicopathological factors for EC OS prediction was developed and validated with good discrimination and calibration.