A 13-gene prognostic cluster associated with poor survival in EGFR-mutant LUAD is identified and the EGFR-regulated USP15/KRT17 axis is position as a novel framework for patient risk stratification, with potential implications for guiding targeted therapeutic strategies.
Abstract
Background
EGFR-mutant lung adenocarcinoma (LUAD) exhibits significant prognostic heterogeneity despite the efficacy of tyrosine kinase inhibitors. To uncover the transcriptomic markers of rapid disease progression, we integrated clinical and transcriptomic data to develop robust prognostic biomarkers for risk stratification.
Methods
We utilized ensemble machine learning algorithms, including Random Survival Forest and XGBoost, to analyze transcriptomic data from 293 patients with EGFR-mutant LUAD. Single-cell transcriptomic analysis combined with copy number variation (CNV) profiling was employed to assess the cellular distribution and tumor epithelial contribution of the identified prognostic signature. The functional effects of the candidate biomarkers, USP15 and KRT17, and their dynamic regulation by oncogenic EGFR signaling were experimentally validated in EGFR-mutant cell lines using siRNA knockdown, pharmacological blockade (osimertinib), and gain-of-function assays.
Results
The optimized XGBoost model identified a 13-gene cluster with strong predictive performance (C-index = 0.863), which was independently validated in an external cohort (log-rank P = 0.0117; 1-year AUC = 0.698). Single-cell transcriptomic analysis demonstrated that the detectable 11-gene subset exhibited a mixed tumor-microenvironment distribution, including expression within CNV-defined malignant epithelial cells. Functional experiments further demonstrated that USP15 and KRT17 promote tumor cell proliferation, clonogenic growth, and migration. Moreover, the expression of both genes was responsive to downstream EGFR signaling. Collectively, these findings establish a robust 13-gene prognostic gene cluster with a mixed cellular origin and identify USP15- and KRT17-associated biology as a promising direction for future mechanistic investigation in EGFR-mutant LUAD.
Conclusions
We identify a 13-gene prognostic cluster associated with poor survival in EGFR-mutant LUAD. These findings position the EGFR-regulated USP15/KRT17 axis as a novel framework for patient risk stratification, with potential implications for guiding targeted therapeutic strategies.
Keratins (KRTs) are intermediate filament proteins expressed in epithelial cells and serve as diagnostic cancer biomarkers. They serve pivotal roles in tumor progression and metastasis, but their prognostic value in lung adenocarcinoma (LUAD) and relationship with the tumor immune microenvironment remain unclear. Gene expression of KRTs in The Cancer Genome Atlas (TCGA)-LUAD was analyzed; candidate genes were selected using LASSO and random forest. An XGBoost classifier with SHAP interpretation distinguished tumor from normal tissues. Prognostic models were developed using 101 algorithms with 10-fold cross-validation and validated in two independent GEO cohorts. Immune cell composition and pathway activity were assessed by CIBERSORT, MCP-counter, ssGSEA and GSEA; the TIDE score estimated potential immunotherapy response. Functional validation of KRT81 was performed via shRNA knockdown in LUAD cell lines, followed by proliferation, apoptosis and migration assays. A KRT-based diagnostic and prognostic signature was established. The XGBoost model achieved high accuracy (TCGA AUC=0.996; GSE31210 AUC=0.860); SHAP identified KRT81 as a key contributor. A four-gene prognostic model (KRT27, KRT80, KRT16, KRT81) stratified patients into high- and low-risk groups. The risk score was associated with advanced tumor stage and independently predicted overall survival (multivariate HR=2.16, 95% CI 1.43–3.28, P=2.0×10-4). High-risk tumors enriched proliferation/stroma pathways (cell cycle, DNA repair, ECM-receptor interaction, focal adhesion, p53 signaling); low-risk tumors enriched immune pathways. Immune profiling revealed reduced T/B cell infiltration, increased endothelial cells and higher T-cell exclusion scores in high-risk patients, indicating an immunosuppressive microenvironment. Higher risk scores also associated with chemotherapy resistance. Functional validation confirmed that shRNA-mediated KRT81 knockdown reduced LUAD cell proliferation, migration and invasion, while promoting apoptosis. These findings link KRT expression to clinical outcomes, the tumor microenvironment and therapeutic response in LUAD, suggesting roles for KRTs in cancer progression, chemotherapy resistance and predictive potential for immunotherapy response. This study provides new insights for prognostic evaluation and therapeutic decision-making in LUAD.
Aohui Chen, Ting Gao, Fengqi Liu et al.· Oncology Letters· 0 citations
A six-gene-fibrosis-based prognostic model based on six genes stratifies survival risk and correlates with immune features and drug sensitivity, but provides a preliminary framework requiring prospective clinical validation.
Yanyan Qiu, Cui Lv, Shubo Ding· Clinical and Translational O...· 0 citations
It is suggested that glycolytic activity is closely associated with both drug resistance and clinical prognosis in LUAD, and the proposed integrative model holds promise for enhancing precision treatment decision-making through optimized risk assessment and rational selection of chemotherapeutic regimens.
Qian Zheng, Yunxiao Liu, Tianli Li et al.· Frontiers in Oncology· 0 citations
Background The tumor immune microenvironment influences non-small cell lung cancer (NSCLC) progression and therapy response. Programmed death-ligand 1 (PD-L1) is a key biomarker for immune checkpoint blockade, yet the genomic correlates of its expression status require elucidation. Methods We analyzed 97 advanced lung adenocarcinoma patients, (13 cases—all from the PD-L1-unknown subgroup—excluded due to insufficient sequencing quality, leaving 84 cases for downstream analyses) categorized by PD-L1 expression (positive: n=32; negative: n=45; unknown: n=20). Genomic profiling was performed using a 95-gene targeted NGS panel. Analyses included mutational profiling, pathway enrichment (GO/KEGG), gene interaction patterns, and genomic complexity. Results The cohort was predominantly metastatic (99.0%). Dominant mutations were in EGFR(43.80%), TP53(38.02%), and KRAS(11.57%). PD-L1-positive tumors more frequently harbored EGFRE746_A750del, while TP53 mutations were enriched in PD-L1-negative cases. Somatic interaction analysis revealed significant EGFR/TP53 co-occurrence in PD-L1-positive tumors and TP53/RB1 co-occurrence in negative tumors. GO analysis showed PD-L1-positive tumors were enriched for kinase-related functions, reflecting a distinct EGFR-driven biology characterized by a striking 59.26% prevalence of EGFR mutations. KEGG analysis indicated immune pathway enrichment in this subgroup, supporting an oncogene-induced immune phenotype frequently synergized by TP53 co-alterations. In contrast, PD-L1-negative tumors correlated with co-mutation-prone, metabolism-oriented profiles and immune suppression. Mutual exclusivity patterns differed: in PD-L1-positive tumors, EGFRmutations were mutually exclusive with KRAS/NRAS/HRASmutations, suggesting a single dominant driver. PD-L1-negative tumors displayed more co-occurring alterations (e.g., TP53 with STK11 or PTEN). Furthermore, 24% of PD-L1-positive tumors had multiple concurrent alterations, a proportion significantly higher than in PD-L1-negative tumors (8%; P<0.05). Conclusion This study delineates distinct genomic landscapes associated with PD-L1 expression. PD-L1-positive status is linked to an immune-active, kinase-signaling-driven phenotype with mutually exclusive drivers. PD-L1-negative status correlates with a more immunosuppressive, co-mutation-prone, and metabolism-oriented profile. These findings provide molecular insights into immune heterogeneity and may inform tailored combination therapies.
Mukun Huang, Zhongqiang Yao, Hua Zhang et al.· Frontiers in Oncology· 0 citations
Hepatocellular carcinoma (HCC) is one of the malignant tumors with high incidence and mortality rates worldwide. Given the poor prognosis of patients with HCC, it is crucial to explore the molecular mechanisms underlying HCC development and to evaluate prognostic markers. Differential expression analysis followed by univariate Cox, LASSO, and multivariate Cox regression identified four genes (EPO, SOCS2, IL18RAP, and KPNA2), and a Cox-based risk score was evaluated in the TCGA-LIHC cohort and externally in GSE14520 using Kaplan–Meier and time-dependent ROC analyses. Bulk, single-cell, and protein resources provided convergent expression context. Survival machine-learning analysis using observed overall-survival time and censoring status identified Cox–Ridge as the best-performing model in TCGA-LIHC, with more modest performance in GSE14520, and immune profiling revealed risk-group-associated differences in estimated immune and stromal components, immune-cell composition, and immune-checkpoint expression. The oncoPredict/GDSC2 screen highlighted five potential drug candidates for experimental prioritization. Because the drug screen is based on computationally predicted sensitivities, these findings should be regarded as hypothesis-generating and require validation in prospective cohorts and experimental systems before clinical translation.
Yuxian Liu, Xing-Jie Chen, Junyuan Zhang et al.· International Journal of Mol...· 0 citations
A robust prognostic model based on four RhoGTPase-related prognostic genes was established, effectively stratifying LUAD patients and provides valuable insights into the heterogeneity of LUAD and has the potential to inform personalized therapeutic strategies.
Tianchuan Li, Danhong Wu, A. Yang et al.· Scientific Reports· 0 citations