Related Experiment Video
Updated: Jan 6, 2026
Predictive Immune Modeling of Solid Tumors
Published on: February 25, 2020
Combining molecular patterns and clinical data for better immune checkpoint inhibitor prediction in metastatic
Background:
The therapeutic landscape of advanced urothelial carcinoma (UC) is evolving, making the prediction of immune checkpoint inhibitor (ICI) therapy efficacy crucial. Standalone biomarkers offer limited predictive value, necessitating integrative approaches combining clinicopathological, laboratory, and molecular factors to enhance accuracy. This study aimed to evaluate clinical and molecular factors, including the real-life performance of PD-L1 IHC, to improve treatment outcome prediction in ICI-treated UC patients, ultimately developing a more precise therapy selection model.
Methods:
We conducted a retrospective, multicenter study on advanced or metastatic UC patients with available formalin-fixed, paraffin-embedded tumor samples who underwent ICI therapy (n = 100). NanoString technology was used to analyze 770 immune-related genes and a 60-gene panel for molecular subtype classification. Identified genes were validated in the IMvigor210 dataset. Whole tissue PD-L1 expression was assessed using the Dako 22C3 antibody.
Results:
Our findings show that PD-L1 IHC has limited predictive value for ICI response. However, among multigene molecular factors, the neuronal signature, MDA p53-like, and TCGA-luminal-infiltrated subtypes were linked to improved OS. Additionally, we identified and validated five novel ICI-predictive genes (PSMB10, HLA-E, IRF7, CXCL12, and C3), and by combining molecular and clinicopathological parameters, we developed a model with enhanced predictive value.
Conclusions:
Our real-life cohort analysis confirms the limitations of standalone biomarkers like PD-L1. We identified gene expression-based markers with strong prognostic and predictive value for ICI treatment outcomes.
