Multi-modal integration of histopathology and transcriptomics reveals STAB1+ macrophage-associated efferocytosis as a suppressive immune mechanism in colon adenocarcinoma

Zhanhao Chang; Chongli Zhong; Shuo Xu; Yuyang Zhang; Xingqi Guo; Jielin Yu; Zitong Xu; Shukun Han; Bing Han; Chao Lv; Yu Tian

doi:10.1186/s12967-025-07348-8

Multi-modal integration of histopathology and transcriptomics reveals STAB1⁺ macrophage-associated efferocytosis as a suppressive immune mechanism in colon adenocarcinoma

J Transl Med. 2025 Dec 8;23(1):1386. doi: 10.1186/s12967-025-07348-8.

Authors

Zhanhao Chang¹, Chongli Zhong¹, Shuo Xu¹, Yuyang Zhang¹, Xingqi Guo², Jielin Yu³, Zitong Xu³, Shukun Han¹, Bing Han⁴, Chao Lv¹, Yu Tian⁵

Affiliations

¹ Department of General Surgery, Shengjing Hospital of China Medical University, Shenyang, Liaoning, 110004, China.
² Department of General Surgery, Cancer Hospital of China Medical University, Liaoning Cancer Hospital & Institute, Cancer Hospital of Dalian University of Technology, Shenyang, Liaoning, 110042, China.
³ Department of Pathology, Shengjing Hospital of China Medical University, Shenyang, Liaoning, 110004, China.
⁴ Department of Gastroenterology, First Affiliated Hospital of China Medical University, Shenyang, Liaoning, 110001, China.
⁵ Department of General Surgery, Shengjing Hospital of China Medical University, Shenyang, Liaoning, 110004, China. yu.tian@cmu.edu.cn.

Abstract

Background: Colon adenocarcinoma (COAD) has a limited response to immunotherapy due to its immunologically “cold” tumor microenvironment (TME). Efferocytosis is an important process that regulates the TME, but its mechanism and clinical significance in COAD are unclear.

Methods: We integrated histopathological images, transcriptomic profiles, and clinical data from 387 COAD patients. Image features were extracted using ResNet50 and CellProfiler, followed by construction of a multimodal machine learning model to evaluate prognostic risk. We further combined bulk RNA-seq, single-cell RNA-seq, and spatial transcriptomics to comprehensively characterize efferocytosis-associated immune cell subsets and signaling pathways.

Results: The efferocytosis-based risk model demonstrated strong prognostic performance across multiple time points and remained independent of conventional clinical indicators. Mechanistically, we identified a subset of STAB1⁺ tumor-associated macrophages (TAMs) enriched in COAD tumors, exhibiting enhanced efferocytosis activity, M2-like polarization, and mTORC1 signaling activation. In vitro, STAB1 expression was essential for IL-4-induced M2 polarization, and its inhibition attenuated the formation of immunosuppressive TAMs. Single-cell and spatial transcriptomic analyses revealed that this macrophage population was transcriptionally distinct and increased in abundance following neoadjuvant therapy.

Conclusion: This study establishes a multimodal prognostic system that integrates histopathological imaging with molecular profiling, and for the first time reveals the pivotal role of STAB1⁺ TAMs in orchestrating the immunosuppressive TME via efferocytosis and mTORC1 activation. Our findings provide both a clinically applicable risk assessment tool and a potential therapeutic target. Targeting STAB1 may broaden the benefit of immunotherapy for COAD patients with limited responses to immune checkpoint blockade.

Graphical Abstract:

Supplementary information: The online version contains supplementary material available at 10.1186/s12967-025-07348-8.

Keywords: Colon adenocarcinoma; Deep learning; Efferocytosis; Histopathology; STAB1; Tumor-associated macrophages.