Spatial architecture contributes to failure of bulk biomarker-guided neoadjuvant immunotherapy selection in bladder cancer: The DUTRENEO study.
Summary
The prospective phase 2 DUTRENEO study demonstrated that a retrospectively validated 18-gene tumor inflammation signature (TIS) failed to predict response to neoadjuvant immune checkpoint inhibitors (ICIs) in muscle-invasive bladder cancer. This outcome indicates that bulk gene-expression stratification is insufficient to enrich for responders. Single-cell spatial transcriptomic analysis revealed that response is instead governed by spatial architectures, such as CD8+ T cell proximity to cancer cells and localized checkpoint co-expression, which are invisible to bulk assays. A quantitative framework was developed, suggesting that at least 77 genes and tissue regions of at least 3-mm diameter are required to preserve predictive spatial signals.
Analysis
Clinique: The phase 2 DUTRENEO trial represents a critical example of the failure of retrospectively validated bulk biomarkers to translate into prospective clinical utility for ICI patient selection. The failure to meet its primary endpoint highlights that bulk gene-expression stratification is insufficient to identify responders. This underscores an urgent need for more sophisticated predictive biomarkers, potentially based on spatial architecture. The immediate clinical impact of this study is to discourage the use of similar bulk gene signatures for neoadjuvant selection in bladder cancer. In the longer term (5+ years), it justifies clinical trials evaluating spatial biomarkers to guide therapies. Biomol: This study discovered that ICI response is intrinsically linked to the spatial architecture of the tumor microenvironment, an aspect invisible to bulk assays. Key elements include CD8+ T cell proximity to cancer cells, localized checkpoint co-expression within epithelial cancer-rich neighborhoods, and fibroblast-rich immune-excluded communities in non-responders. The technology used is single-cell spatial transcriptomics, analyzing 377 genes across large tissue areas. This approach allowed for the definition of a quantitative framework to preserve predictive spatial signals, requiring at least 77 genes and tissue regions of at least 3 mm. These findings have a major translational impact, shifting future biomarker development towards spatial methods, potentially involving liquid biopsies or advanced imaging analyses, for improved patient selection and understanding of resistance mechanisms.