A cancer-testis antigen signature for predicting prognosis and response to immunotherapy in acute myeloid leukemia.

Acute myeloid leukemia (AML) remains a challenging hematologic malignancy with poor survival rates, underscoring the need for precise prognostic biomarkers and therapeutic strategies. Cancer-testis antigens (CTAs), with tumor-restricted expression and immunogenicity, have not been systematically explored for prognosis prediction in AML. Here, we developed a novel CTA-based prognostic signature to predict survival and immunotherapy response in AML patients. Utilizing RNA-seq and clinical data from the TCGA-Acute Myeloid Leukemia cohort (n = 126, training set) and the GSE71014 dataset (n = 104, validation set), we identified 21 prognosis-associated CTAs via univariate Cox regression. Least absolute shrinkage and selection operator and multivariate Cox regression refined the model to 5 key genes: ACRBP, IGF2BP3, SPAG1, TEX101, and KDM5B. The CTA score, calculated from gene expression and regression coefficients, stratified patients into high- and low-risk groups with distinct overall survival (P < .001). The model exhibited robust predictive accuracy, with time-dependent area under the curve values of 0.863 (TCGA-Acute Myeloid Leukemia training set) and 0.723 (GSE71014 validation cohort) for 5-year overall survival. High CTA scores correlated with adverse prognosis, elevated monocytes and M0 macrophages infiltration, and enhanced immunotherapy responsiveness. Functional enrichment analysis revealed dysregulation in several key AML-related pathways, including leukocyte migration, PI3K-Akt signaling, and cytokine pathways. A nomogram integrating CTA score and age further improved prognostic precision (concordance index = 0.77). Drug sensitivity profiling highlighted differential therapeutic vulnerabilities between risk groups. This study established and validated a novel CTA-based prognostic tool for prognostic stratification and personalized treatment guidance in AML, bridging molecular insights with clinical applications.