This study identified environmental pollutants associated with activation of the AHR–NF-κB–C/EBPβ signaling axis and examined their effects in colorectal cancer (CRC). ToxCast indicators and TCGA-COAD/READ RNA-seq data were integrated to construct a pollutant-response matrix; an SVM model prioritized phenmedipham for mechanistic testing. In vitro, phenmedipham increased AHR, p-p65, and C/EBPβ expression and pro-inflammatory cytokine secretion, whereas pathway-component silencing attenuated these effects. In vivo, phenmedipham increased colonic tumor burden and histopathological damage; pathway inhibition mitigated these phenotypes. Immunological analyses indicated impaired dendritic-cell activation and T-cell effector responses. These results provide preclinical evidence that phenmedipham can modulate inflammatory signaling and tumor-associated phenotypes in the tested models. They do not establish human carcinogenicity, CRC risk, or population-level exposure–disease causality; epidemiological validation is required.
Accurate preoperative prediction of lymph node metastasis (LNM) could facilitate more individualized axillary management in breast cancer by improving identification of patients at lower risk of nodal involvement while maintaining appropriate staging for those at higher risk. Yet, reliable molecular predictors of LNM remain lacking. VEGF-C, a key driver of lymphangiogenesis, is expressed in both cancer cell and stromal compartments, but its clinical utility is unclear. In a prospectively enrolled cohort, we assessed stromal and cancer cell VEGF-C expression by immunohistochemistry using both manual scoring and a novel automated image analysis script enabling objective, reproducible classification of stromal versus epithelial cells at scale. Stromal VEGF-C was independently associated with LNM after adjustment for established predictors (P = 0.004), with prediction models reaching AUCs of up to 0.784. Cancer cell VEGF-C added no predictive value. In exploratory analyses, stromal VEGF-C provided predictive information comparable to that of vascular invasion, a strong predictor of LNM that is generally unavailable preoperatively. These findings identify stromal VEGF-C as a potential preoperative biomarker for nodal risk stratification. However, given the observational and predictive nature of the present study, stromal VEGF-C should be considered complementary to, rather than a replacement for, established pathological predictors. Its potential contribution to future axillary surgery de-escalation strategies warrants prospective validation.
Tumor-associated macrophages (TAMs) and cancer-associated fibroblasts (CAFs) are major components of the tumor microenvironment (TME), influencing tumor immunity, cancer progression, and treatment responses. Interactions between CAF and TAM subpopulations have only recently been investigated, revealing complex mechanisms that may shape tumor immunity and therapeutic outcomes. In this study, we performed single-cell RNA sequencing and spatial transcriptomics on primary patient-derived HPV-negative head and neck squamous cell carcinoma (HNSCC) tumors to characterize cellular heterogeneity and stromal-immune crosstalk. We identified five TAM and four CAF subsets, with IL1B+ TAMs and IL-iCAFs forming a distinct, spatially co-localized inflammatory niche. In vitro co-culture experiments revealed that IL-iCAF-derived paracrine signals induced transcriptional features associated with the IL1B⁺ TAM state. Clinically, high infiltration of IL1B⁺ TAMs correlated with poorer survival and higher recurrence risk, likely driven by increased inflammatory signaling from IL1B+ TAMs that promotes regulatory T cell recruitment and exhaustion. Together, our findings identify a spatially organized inflammatory niche involving IL-iCAFs and IL1B⁺ TAMs and suggest that fibroblast-derived signals may contribute to the acquisition of inflammatory TAM features. These findings provide a basis for further investigation of IL-iCAF–TAM interactions and the potential clinical relevance of IL1B⁺ TAMs in HPV-negative HNSCC.
Oncology computed tomography (CT) foundation models require large, high-quality training datasets. However, the collection and annotation of such data are costly and time-consuming because tumors occur across multiple anatomical regions and have complex three-dimensional (3D) structures. Strict privacy constraints and institutional variation further limit the availability of large-scale oncology CT datasets. Here, we propose MedGenAgent, a closed-loop agent that supports oncology CT foundation model pretraining in data-constrained settings by iteratively constructing a lesion-rich, anatomically balanced 3D pretraining dataset. MedGenAgent incorporates a Specification resolver tool for structured task parsing and a central Orchestrator tool to coordinate the entire closed-loop agent workflow by leveraging both an Execution layer (including the MAISI synthesis model node for 3D oncology CT sample synthesis, the MedGemma vision-language model node for automated quality control, and the foundation model training node) and a Memory store tool for iterative, feedback-driven updates. Through the synergy of these tools, MedGenAgent synthesized 11,159 CT samples and retained 11,000 quality-controlled samples. By first leveraging this synthetic dataset for pretraining and then aligning with real-world data using only 4400 DeepLesion samples, representing 38.37% of the data used for baseline pretraining, the MedGenAgent-pretrained foundation model achieved performance comparable to the full-data baseline across three retrospective downstream tasks: lesion anatomical site classification, nodule malignancy classification, and cross-institutional NSCLC overall survival classification. Furthermore, automated quality control, independent physician assessment, and workflow analysis demonstrated the construction of a quality-controlled synthetic CT dataset while revealing regional variation in sample quality. Together, these results show that MedGenAgent can autonomously construct quality-controlled synthetic CT data that support the competitive performance of oncology CT foundation models when annotated real-world data are limited.
Stromal cells are key components of the breast cancer tumor microenvironment, yet their heterogeneity, spatial organization, and interactions with tumor and immune cells remain incompletely defined. We integrated single-cell RNA sequencing data from 233,518 non-immune stromal cells across 349 patients with spatial transcriptomic data from 51 patients. This analysis identified 36 fibroblast, endothelial, and mural cell subsets with distinct transcriptional programs, spatial distributions, and functional associations. ABCA8+ lipid-associated fibroblasts were observed at tumor margins, showed predicted interactions with malignant cells, and were associated with exhaustion-associated T-cell phenotypes. CCL19+ inflammatory fibroblasts showed spatial patterns associated with tertiary lymphoid structure formation, whereas tumor-associated pericytes showed predicted interactions with PGF+ tip endothelial cells that may be involved in extracellular matrix deposition and vascular remodeling. Together, these findings provide a high-resolution stromal atlas of breast cancer, resolving stromal subtypes, spatial context, and potential intercellular communication. This resource provides a framework for investigating stromal contributions to tumor progression and immune regulation, although further experimental validation is required.
