Abstract 11: Artificial intelligence-based spatial analysis of the local tumor microenvironment in relation to c-MET expression in non-small cell lung cancer.

Le résumé fourni par la source

Abstract Introduction: MET is a well-known oncogenic driver that confers various genomic aberrations. Recent approval of antibody drug conjugates has expanded therapeutic options for c-MET expressing non-small cell lung cancer (NSCLC). However, the relationship between MET expression, tumor microenvironment (TME), and immunotherapy response remains unclear. This study explores the association between c-MET expression and spatial TME features in NSCLC to better understand its immunologic behavior. Methods: This study analyzed a total of 25,674 NSCLC samples from various cohorts, including AACR GENIE, TCGA, Ajou University Medical Center (AUMC), and Agilent Technologies. AI-powered analyzers (Lunit SCOPE IO and SCOPE uIHC) were previously developed using 19,112 H&E and 4,638 IHC whole slide images of 25 cancer types, stained with over 20 different antibodies. These platforms enabled the quantitative assessment of both the TME, and subcellular expression levels detected by IHC staining. Results: MET alterations occurred in 27 (2.9%) of TCGA and 909 (3.8%) of GENIE, including exon 14 skipping (n=388, 1.5%), amplification (n=380, 1.5%), and others (n=223, 0.9%). MET-altered tumors had higher MET RNA expression compared with wild-types (median: 0.4 vs. -0.2, p<0.001). TME analysis of samples with high (Z-score ≥2) and low RNA expression showed no significant difference in tumor-infiltrating lymphocyte (TIL) density (/mm2, median: 851 vs. 673, p=0.46). In 640 pairs of H&E and IHC slides from AUMC and Agilent, c-MET positive (3+, ≥50%) samples tended to have lower TIL density across the whole slide compared to c-MET negative samples (median, AUMC, 78.5 vs. 79.2, p=0.16; Agilent, 80.6 vs. 211.3, p=0.23, respectively). This trend became significant with cell and subcellular spatial analysis. The density of TIL was markedly reduced within 30um of tumor cells exhibiting strong (3+) c-MET expression (median:85.5 vs. 121.6, p<0.001; 132.2 vs. 162, p=0.013, respectively). Notably, a similar reduction in TIL density was also observed around tumor cells with membrane-specific c-MET expression (median: 106.6 vs. 122.3, p<0.001; 144.9 vs. 165.8, p=0.002, respectively). Conclusion: Spatial analysis of IHC demonstrated sparse immune cells near tumor cells with strong c-MET expression or membrane-specific localization. These findings suggest a mechanistic link between c-MET overexpression and immune evasion, indicating the potential benefit of combining MET-targeted and immunotherapy. Citation Format: Ji-Hyang Lee, Joshua Littrell, Chiyoon Oum, Taebum Lee, Sanghoon Song, Yoojoo Lim, Chang Ho Ahn, Jim Christian, Seokwhi Kim, Siraj M. Ali. Artificial intelligence-based spatial analysis of the local tumor microenvironment in relation to c-MET expression in non-small cell lung cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 11.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.