The prognostic landscape of tumor-infiltrating immune cells in cervical cancer

Abstract

Tumor-infiltrating immune cells (TICs) are highly relevant to tumor development and are promising prognostic biomarkers. However, the precise assessment of TICs is limited by the deficiencies of traditional measurements, such as the lack of phenotypic markers. Here, we analyzed the composition of TICs in cervical cancer based on RNA expression data with a metagene approach called CIBERSORT and evaluated the prognostic value of TICs. The immune infiltration profiles functioned as intrinsic features to distinguish cervical cancer from normal tissue. According to the Cox regression analysis, higher levels of activated memory CD4+ T cells were independently associated with favorable overall survival (OS) (hazard ratio [HR] = 0.71, 95% confidence interval [CI]: 0.57–0.89; p = 0.003), whereas a higher fraction of activated mast cells was independently associated with adverse outcomes (HR = 1.53, 95% CI: 1.23–1.91; p < 0.001). Furthermore, a novel prognostic model named aTMNs (activated memory CD4+ T cells, activated mast cells and activated natural killer [NK] cells) was constructed to predict OS in cervical cancer with high accuracy (area under the curve [AUC] = 0.723, concordance index [C-index] = 0.738): risk score = −0.34508 × (proportion of activated memory CD4+ T cells) + 0.426841 × (proportion of activated mast cells) + 0.272202 × (proportion of activated NK cells). The aTMNs model outperformed the immunomodulator model (AUC = 0.673, C-index = 0.693). Overall, TICs are important prognostic determinants in cervical cancer and may be a useful resource for the development of effective immunotherapy.