17P - Spatial Characteristics Associated with the Chemo and Immuno-treatment Response of Gastric Cancer Revealed by Multi-omics Analysis

Background

Gastric cancer is a multifaceted condition exhibiting varied responses to treatment, emphasizing the need for tailored therapeutic strategies. This study endeavors to elucidate the cellular interactions and molecular mechanisms that underlie the response to Sintilimab plus SOX (fluorouracil plus oxaliplatin) therapy among gastric cancer patients.

Methods

Single-cell sequencing and multiplex immunohistochemistry (mIHC) were employed to elucidate the spatial characteristics linked to the response of gastric cancer to chemo- and immuno- treatment. By integration of mIHC, feature extraction, and machine learning algorithms, we elucidated the intricate interactions between cellular populations and developed a Support Vector Machine (SVM) model for the prediction of treatment response.

Results

We initially discovered a significant correlation between apical membrane cells and resistance to fluorouracil and oxaliplatin, both crucial components of the treatment regimen. This prompted us to delve into the involvement of apical membrane cells in treatment response. Through a thorough examination of cell interactions, we noted substantial connections between apical membrane cells and resident macrophages. Further analysis of ligand-receptor interactions unveiled specific molecular associations, with TGFB1-HSPB1 and LTF-S100A14 interactions standing out, indicating potential signaling pathways implicated in treatment response. To forecast treatment response, we developed an SVM model integrating six markers (DUOX2, HSPB1, S100A14, C1QA, TGFB1, and LTF), which demonstrated outstanding predictive capacity, achieving high area under the curve (AUC) values of 0.93 in the exploration cohort and 0.84 in the validation cohort.

Conclusions

Our research underscored the importance of integrating multi-omics data alongside spatial information in predictive model, holding the promise of steering personalized therapeutic decisions and enhancing treatment efficacy.

Legal entity responsible for the study

The First Affiliated Hospital, School of Medicine, Zhejiang University.

Funding

Has not received any funding.

Disclosure

All authors have declared no conflicts of interest.