Multi-dimensional immune profiling in GMAP project informs PD-1/PD-L1 blockade efficacy

Keren Jia,Yang Chen,Yi Xie,Xiaoyi Chong,Yilin Li,Yingcheng Wu,Jiajia Yuan,Yanyan Li,Xujiao Feng,Yajie Hu,Yu Sun,Jifang Gong,Xiaotian Zhang,Jian Li,Lin Shen

DOI: https://doi.org/10.1016/j.ejca.2023.05.019

IF: 10.002

2023-06-02

European Journal of Cancer

Abstract:Background Immunotherapy has resulted in impressive objective response rates and durable tumor remission, but only in a subset of gastric cancer (GC) patients. The PD-L1 combined positive score is the most widely used tissue-based biomarker for anti–PD-1/PD-L1 therapy; however, this unidimensional method has limitations. Next-generation exploration of tissue-based biomarkers for GC requires characterization of various cellular marker and key immunoregulatory molecule expression in situ . Thus, a complete, stepwise solution covering the entire process from staining samples to cross-site utilization of pathomics data is urgently needed. Methods With the advanced multispectral imaging analysis method, web-based data repository, and interactive sharing technology, we conducted a project entitled Gastric Cancer Multiplex Immunohistochemistry Atlas from Peking University Cancer Hospital (GMAP). We propose a standard pipeline covering sample collection, staining, scanning multispectral images, constructing a spectral library, identifying and phenotyping cells, positioning each element, and quantitatively extracting immune features. We designed an open-access relational database to explore tissue-based biomarkers to determine PD-1/PD-L1 blockade efficacy. Results The GMAP project detected the functional status and spatial location of more than 50 million cells using 15 markers in 80 GC patients, based on which billions of cell pairs were recognized, highlighting the rich spatial arrangement information and the fine tumor microenvironment structure. We generated a tumor-immune atlas using the count and spatial features of 65 immune cell types. We eventually selected the indicators and built a comprehensive risk-scoring system. Patients with higher risk score showed superior irPFS (Hazard ratio: 3.19; P < 0.001; median irPFS: 4.87 vs. 19.87 months, respectively) and irOS (Hazard ratio: 3.10; P = 0.001; median irPFS: 10.03 vs. 24.87 months, respectively), compared with lower risk patients, demonstrating their potential for guiding anti-PD-1/PD-L1-based immunotherapy. Importantly, an easy-to-use and versatile web server was built to promote tissue-based biomarker exploration in GC. Conclusions The GMAP project highlighted the clinical value of tissue-based immune features as biomarkers for immunotherapeutic decision-making. We present a well-designed, detailed workflow for the orderly generation and use of a high-quality, spatially resolved pathological database.

oncology