1457P - Tumor immune microenvironment subtypes of esophageal squamous cell carcinoma and their strong ability to predict the efficacy of neoadjuvant immunotherapy

Background

Heterogeneity of TIME organized by various immune and stromal cells, is a major contributing factor of drug resistance, metastasis and relapse, but how different TIME subtypes vary in ESCC and their connection and predicting role to response of neoadjuvant immunotherapy(nIOT) in ESCC remains unclear.

Methods

We analyzed sc-RNA sequencing data from 18 locally advanced ESCC patients who underwent nIOT in the SEEK-01 trial. Through clustering analysis, distinct categories of immune and stromal cells were identified. Correlation analysis explored co-occurrence patterns of TIME cell subpopulations. Hierarchical clustering and GSEA Analysis identified co-expression cell modules. Based on expression score, patients were classified into 4 TIME subtypes. To test the predictive ability of each module for irRVT score after nIOT, we used Wilcox test and multivariate logistic regression to construct a predictive model. We then conducted prospective validation using baseline sc-RNA sequencing results and clinical outcomes from 4 locally advanced ESCC patients.

Results

Analyzing 257,021 single cells, 9 immune cell types and 2 stromal cell types in TIME were categorized into 182 cell subgroups. Through correlation analysis and hierarchical clustering, we obtained 5 cell co-expression modules(CM1-CM5), and classified patients into 4 TIME subtypes based on expression score. These subtypes include 2 immune activation subtypes emphasizing effector CD4(CM1-dominated) or effector CD8 T cells(CM3-dominated), an inflammatory infiltration subtype(CM5-dominated), and an immune suppression subtype with myeloid cell infiltration(CM2-dominated). A predictive model for nIOT effectiveness was established, showing strong predictive capacity (r²=0.93, p=2*10^-5), incorporating TIME subtype, CM2 score, and age. Validation with 4 prospective samples demonstrated robust predictive capabilities.

Conclusions

This comprehensive study unraveled the complexity of TIME in locally advanced ESCC, delineated the features of 4 TIME subtypes and their varied responses to nIOT, and presented a reliable predictive model for accurately forecasting the efficacy of nIOTin baseline conditions.

Clinical trial identification

SEEK-01 trial: NCT05807542; First posted: April 11, 2023.

Editorial acknowledgement

Legal entity responsible for the study

The authors.

Funding

Scientific and Technical Department of Fujian Province.

Disclosure

All authors have declared no conflicts of interest.