Potential effects of TGF-β inhibition on immune resistance by targeting immunosuppressive microenvironment in advanced gastric cancer (AGC): Multi-omics post-hoc analysis of the K-Umbrella-06 trial.
Abstract
2612 Background: TGF-β signaling promotes immune exclusion by driving fibrosis and hindering T-cell infiltration. We performed multi-omics post-hoc analyses of K-Umbrella-06 to identify mechanisms and predictive biomarkers for TGF-β/PD-L1 dual blockade in HER2-negative AGC. Methods: K-Umbrella-06 (NCT04835896) was an investigator-initiated, single-arm, phase Ib/II study enrolling HER2-negative AGC pts who had progressed on 1st-line treatment. Pts received bintrafusp alfa (TGF-β trap/anti–PD-L1) and weekly paclitaxel. Pretreatment H&E whole-slide images (WSIs) were analyzed with an AI-powered WSI analyzer to quantify fibroblast and endothelial cell (EC) densities and immune phenotypes. A separate cohort treated with nivolumab + paclitaxel (n=26) served as a control. Integration of baseline tissue transcriptomics and serial ctDNA sequencing (baseline, 1 st RECIST assessment, and progression) was performed to identify predictive biomarkers and resistance mechanisms. Results: At a median follow-up of 46 months in 30 enrolled patients, the mPFS and mOS were 3.8 and 8.9 months, respectively; notably, 5 patients achieved durable responses with PFS exceeding 3 years. AI-WSI analysis revealed that high fibroblast density (≥median) was a significant predictor of superior outcomes (mPFS: 4.3 vs. 2.0 months, HR 0.29, P=0.007; mOS: 20.4 vs. 5.8 months, HR 0.28, P=0.012). Conversely, patients with higher intratumoral EC density had shorter mPFS (2.0 vs. 5.9 months, HR 2.29, P=0.058) and mOS (4.0 vs. 8.9 months, HR 1.61, P=0.285). Notably, patients with the immune-excluded phenotype achieved a higher ORR (60.0% vs. 18.2%) and prolonged survival compared to inflamed/desert types. These biomarkers were not predictive in the nivolumab + paclitaxel cohort, suggesting a TGF-β specific effect. Transcriptomic profiling of responders showed enrichment in TGF-β signaling (normalized enrichment score [NES] 2.87, P<0.001) and SMAD2/3 activity (NES 2.99, P<0.001), alongside increased naive CD8+ T cells and inflamed EC cells. Conversely, non-responders exhibited higher M2 macrophage and abundant angiogenic tip cells. Serial ctDNA analysis identified emergent TGF-β pathway alterations as a potential mechanism of acquired resistance (OR 7.30, P=0.08). Conclusions: Integrated AI-WSI, transcriptomic, and longitudinal ctDNA analyses suggest that TGF-β pathway activation and an immune-excluded phenotype are associated with clinical benefit from bintrafusp alfa plus paclitaxel, and may identify a responder subset distinct from conventional PD-1/PD-L1–based therapy. These findings support further evaluation of anti–TGF-β strategies for AGC with an immunosuppressive tumor microenvironment. Clinical trial information: NCT04835896 .
Article Details
Journal Info
Journal of Clinical Oncology
Lippincott Williams & Wilkins
Authors (13)
Choong-kun Lee
Kwang Seob Lee
Department of Laboratory Medicine, Yonsei University College of Medicine, Seoul, South Korea
Dong Hyun Seo
School of Electrical Engineering and Computer Science Gwangju Institute of Science and Technology Gwangju 61005 Republic of Korea
Sangwon Shin
Chang Ho Ahn
Lunit Inc., Seoul, South Korea
Hyunki Kim
Department of Pathology, Severance Hospital, Yonsei University College of Medicine, Seoul, South Korea
Minkyu Jung
Department of Internal Medicine, Yonsei Cancer Center, Yonsei University College of Medicine, Seoul, South Korea
Hyo Song Kim
Division of Medical Oncology, Department of Internal Medicine, Yonsei Cancer Center, Yonsei University College of Medicine, Songdang Institute for Cancer Research, Seoul, South Korea
Sejung Park
Songdang Institute for Cancer Research, Yonsei University College of Medicine, Seoul, South Korea
Chan-Young Ock
Woo Sun Kwon
Hyun Cheol Cheol Chung
Yonsei University College of Medicine, Seodaemun-Gu, South Korea
Sun Young Rha