Journal: Nature Medicine

Article Title: Integrated epidemiological and molecular data inform the relationship between precancer and cancer states of esophageal adenocarcinoma

doi: 10.1038/s41591-026-04331-8

Figure Lengend Snippet: a , Distribution of T stage and BE status in the surveillance versus full cohort and surveillance durations for patients with prior BE diagnosis. The proportions of early-stage cancers (T1) and BE-positive EAC were higher in the surveillance cohort (214 patients) than in the full cohort (3,100 patients), as expected. When excluding patients with unknown BE phenotypes, the proportions of patients with BE-positive EAC are 87.3% (145/166) in the surveillance cohort and 58.4% (1,235/2,115) in the full cohort. ‘*’ indicates a statistically significant difference in T stage distribution between the surveillance cohort and the full cohort within the same phenotype. ‘^’ indicates a statistically significant difference in T stage distribution compared to BE-positive EAC within the same cohort. χ 2 analysis with FDR and Bonferroni corrections for multiple comparisons was applied. Exact P values for each pair of comparisons are presented in Supplementary Table . Among 214 patients with a self-reported prior BE diagnosis, recorded surveillance duration was available for 116 patients, and the median surveillance time was longer in patients with BE-negative EAC (114.0 months (IQR: 54.0–127.0)) compared to patients with BE-positive EAC (49.2 months (IQR: 21.6−97.9)). b , Spatial transcriptomic analysis using the 10x Genomics Visium HD platform. EAC regions express BE marker genes, whereas areas of intestinal metaplasia do not express EAC-associated genes, supporting a unidirectional progression from BE to EAC. Column 1 displays expression overlays for four gene sets. Tumor genes: MK167, SPINK1, ERBB2 and CLDN4 (dark blue); BE genes: MUC2, TFF3, REG4 and CDX2 (light blue); squamous genes: DSG3, KRT5, KRT14 and TP63 (dark green); and stroma genes: DSG3, KRT5, KRT14 and TP63 (light green). Column 2 highlights spatial expression of the key established protein markers for intestinal metaplasia TFF3 and REG4. c , IHC staining of BE-associated biomarkers (TFF3 and REG4) from BE-positive and BE-negative phenotype EAC. In BE-negative EAC cases, positive staining was observed in cytoplasmic vacuoles of glandular epithelium infiltrated by tumor cells in areas where differentiation was maintained. By contrast, BE-positive EACs showed strong expression of both markers within regions of intestinal metaplasia adjacent to the tumor. Note that the cases shown in b and c are different; only the group labels are consistent across both. d , Summary graphic to show how the clinical and pathological measurable evidence of BE varies according to disease stage once an adenocarcinoma is developing. The molecular ‘signatures’ or characteristics of the BE metaplasia from which the EAC arises are persistent over time. These signatures include REG4 and TFF3 lineage markers ascertainable at the transcriptomic or protein level, and SBS17 and loss of CDKN2A hallmarks retained in the tissue, but this list in not exclusive.

Article Snippet: Spatial transcriptomics profiling was performed using the 10x Genomics Visium HD WT platform measuring approximately 18,000 genes, which achieves 2-μm resolution, enabling near-subcellular spatial granularity.

Techniques: Biomarker Discovery, Marker, Expressing, Immunohistochemistry, Staining

Journal: Nature Medicine

Article Title: Integrated epidemiological and molecular data inform the relationship between precancer and cancer states of esophageal adenocarcinoma

doi: 10.1038/s41591-026-04331-8

Figure Lengend Snippet: A . Patient selection workflow for spatial transcriptomics analysis. B . Spatial localization of transcriptional clusters in an EAC sample. Individual panels show the spatial distribution of the four primary Leiden clusters overlaid on the H&E-stained tissue section. Cluster 1 (blue) represents the main proliferative tumor compartment, marked by secretory and cell cycle–associated genes including REG1A and TOP2A, notably BE lineage genes were also markers of this cluster in EAC samples with and without adjacent BE. Cluster 2 (orange) corresponds to cytokine-producing tumor regions enriched for CCL20, CXCL8, and DUOX2. Cluster 3 (yellow) localizes to fibroblast-rich stromal regions, expressing ECM-associated genes such as COL1A1 and SPARC. Cluster 4 (red) highlights spatially confined tumor-immune interface regions, characterized by CD74, CXCL5, and MMP3. Composite overlay of all clusters demonstrates the mosaic-like spatial organization of transcriptional programs within the tissue, cluster 3 is represented in yellow for better visualization. Distinct zones of proliferative, inflammatory, stromal, and immune-interacting activity emerge. This case is shown as a representative example illustrating how spatial clustering resolves modular transcriptional architecture within an EAC sample. C . Comparison between unsupervised transcriptomic clustering and morphology-informed tissue classification in BE. Unsupervised transcriptomic clustering of the same tissue section identifies four distinct domains. In the left panel, Cluster 1 (blue) corresponds to canonical BE epithelium. Cluster 3 (green) defines a spatially confined epithelial patch with elevated CEACAM5, CEACAM6, and CLDN4, suggestive of early neoplastic transformation. Cluster 2 (orange) localizes to the columnar–squamous interface and expresses squamous markers (KRT5, KRT13) along with inflammatory genes (S100A8, S100A9), consistent with a reactive squamous phenotype. The region of high-grade dysplasia (HGD) was labeled. In the right panel, Spatial transcriptomics map of tissue compartments derived from morphology-aligned gene expression using Visium HD at 2 µm resolution. BE epithelium (light blue) expresses MUC2, TFF3, REG4, CDX2 consistent with intestinal metaplasia, while EAC regions (dark blue) are defined by expression of MKI67, SPINK1, EiRBB2, and CLDN4. Squamous epithelium (dark green) is defined by DSG3, KRT5, KRT14, and TP63. Stromal regions (light green) express ACTA2, PDGFRA, COL1A1, COL3A1, FAP, MMP2, and MMP9, indicating fibroblast activation and extracellular matrix remodeling.This case serves as a representative example demonstrating the spatial distribution of transcriptomic clusters within a BE sample containing.

Article Snippet: Spatial transcriptomics profiling was performed using the 10x Genomics Visium HD WT platform measuring approximately 18,000 genes, which achieves 2-μm resolution, enabling near-subcellular spatial granularity.

Techniques: Selection, Spatial Transcriptomics, Staining, Expressing, Activity Assay, Comparison, Transformation Assay, Labeling, Derivative Assay, Gene Expression, Activation Assay