Plasma exosomes from individuals with type 2 diabetes drive breast cancer aggression in patient-derived organoids

Abstract

Women with obesity-driven type 2 diabetes (T2D) face worse breast cancer outcomes, yet metabolic status does not fully inform current standards of care. We previously identified plasma exosomes as key drivers of tumor progression; however, their effect on immune cells within the tumor microenvironment (TME) remains unclear. Using a novel patient-derived organoid (PDO) system that preserves native tumor-infiltrating lymphocytes (TILs), we show that T2D plasma exosomes induce a 13.6-fold expansion of immunosuppressive TILs relative to nondiabetic controls. This immune dysfunction may promote micrometastatic survival and resistance to checkpoint blockade, a known issue in T2D cancer patients. Tumor-intrinsic analysis revealed a 1.5-fold increase in intratumoral heterogeneity and 2.3-fold upregulation of aggressive signaling networks. These findings reveal how T2D-associated metabolic dysregulation alters tumor-immune crosstalk through previously underappreciated exosomal signaling, impairing antitumor immunity and accelerating progression. Understanding these dynamics could inform tailored therapies for this high-risk, underserved patient population.

Fig. 1. Cellular characterization of PDOs.

A Illustration of scRNAseq workflow in PDO model of breast cancer TME. B scVI-integrated UMAP visualization of PDOs colored by broad cell type annotation. N = 24,599 from three patients. BT (n = 4226) denotes basal-like tumor cells; CAF (n = 1149) denotes cancer-associated fibroblasts; CYC (n = 1223) denotes cycling epithelial cells; LEC (n = 1254) denotes normal luminal epithelial cells; LP (n = 3574) denote luminal progenitor-like cells; LT (n = 7514) denotes luminal-like tumor cells; MEC (n = 2382) denotes normal myoepithelial cells; PVL (n = 1487) denotes perivascular-like cells. C Relative proportion of cell types in each patient highlighting conservation of major lineages. Colors denote the same cell types in (A). D, E scVI-integrated UMAP colored by treatment (D) and malignancy score as calculated from genomic instability (E). F Heatmap showing expression of canonical marker genes per lineage.

Fig. 2. Increased aggression in T2Dexo-PDOs compared to NDexo-PDOs.

A Bubble plot summarizing statistically significant gene sets enriched in differentially expressed gene sets between T2Dexo-PDOs (blue) and NDexo-PDOs (orange) identified with GSEA. Size of dot is scaled to adjusted p-value. B GSEA enrichment plots for representative gene sets upregulated (top) and downregulated (bottom) in T2Dexo-PDOs compared to NDexo-PDOs. C Violin plots of COSMIC CGC (left) and canonical CSC gene expression (right). Significance calculated via linear mixed effects model, ***<0.001, **<0.01, *<0.05. D Violin plots of top differentially expressed genes. E Representative confocal images of UT (left), NDexo (middle), and T2Dexo (right) per patient. Scale bar = 100 μm. F Violin plots of calculated circularity of PDOs. Singlets were removed during thresholding. Significance calculated via pairwise Wilcoxon Rank Sums test with Bonferroni correction, ***<0.001, **<0.01, *<0.05. G Projection of composite gene signatures capturing diabetes-associated exosomal effects onto TCGA (left) and METABRIC (right) cohorts. Cox proportional hazards model adjusted for age, proliferation score, inflammation score, and molecular subtype.

Fig. 3. Tumor population dynamics in T2Dexo-PDOs.

A UMAP visualization of reclustered scVI-integrated epithelial cells. BT1-2 (n = 4084; n = 142) denote basal-like tumor subclones; CYC (n = 1223) denotes cycling epithelial cells; LEC (n = 1254) denotes normal luminal epithelial cells; LP (n = 3574) denotes luminal progenitor-like cells; LT1-5 (n = 3397; n = 2669; n = 925; n = 398; n = 125) denote luminal-like tumor subclones; MEC (n = 2382) denotes normal myoepithelial cells. B Heatmap showing expression of canonical and top differentially expressed marker genes per subclone. C Relative proportion of luminal-like tumor subclones per treatment group. Red (blue) stars indicate significantly expanding (contracting) clusters. Significance calculated via binomial linear regression model, ***<0.001, **<0.01, *<0.05. D GSEA enrichment plots for representative gene sets upregulated in LT1. E Bubble plot summarizing enrichment of predicted and validated targets for individual T2Dexo-associated miRNAs among downregulated genes in LTs treated with T2Dexo vs. NDexo. Odds ratios from Fisher’s exact test are shown for each miRNA; bubble size corresponds to the number of targets downregulated. F Violin plots showing expression of negative regulators of MMPs.

Fig. 4. Tumor cell evolution.

A Violin plots of pseudotime calculated for luminal-like epithelial cells per treatment group. Significance calculated via linear mixed effects model, ***<0.001, **<0.01, *<0.05. B Scatter plots of gene expression as a function of pseudotime. Dot color denotes epithelial cell subtype assigned to cells. Black line showing linear regression of best fit. Genes determined via graph autocorrelation. C Heatmap of gene module expression in epithelial cell subtypes. D Violin plots of expression of Module 5 in luminal-like epithelial cells. Significance calculated via linear mixed effects model, ***<0.001, **<0.01, *<0.05.

Fig. 5. Perturbed immune signaling in T2Dexo-PDOs.

A Unintegrated UMAP visualization of clustered immune cells. ChopT (n = 135) denote DDIT3 + T cells; MAIT (n = 284) denote mucosal-associated invariant T cells; Tcm (n = 343) denotes central memory T cells; Tem (n = 274) denotes effector memory T cells; Teff (n = 374) denotes effector T cells; Th1 (n = 90) denote T helper 1 cells. B Relative proportion of immune compartment per treatment. Red (blue) stars indicate significantly expanding (contracting) clusters. Significance calculated via binomial linear regression model, ***<0.001, **<0.01, *<0.05. C Heatmap showing expression of canonical marker genes per immune cell state. D Bubble plot summarizing statistically significant gene sets enriched in differentially expressed genes between the immune compartments of T2Dexo-PDOs (blue) and NDexo-PDOs (orange) identified during gene set enrichment analysis (GSEA). Size of dot is scaled to adjusted p-value.

Fig. 6. Pseudotime analysis reveals divergent fates in the T cell compartment.

A Trajectory inferred from T cells. B Phylogenetic tree showing branch point leading either toward typical development (blue) or toward ChopT phenotype (red). C, D Bubble plot summarizing statistically significant gene sets enriched along the trajectory toward proper development (C, blue) or toward the ChopT phenotype (D, red). Dot size reflects the number of genes detected per gene set. E Volcano plot of top differentially expressed genes at the branch point.

Fig. 7. T2Dexo modulates intercellular communication networks.

A Circle plot showing the number of interactions among different cell types. The blue (orange) colored edges represent increased (decreased) signaling in T2Dexo compared to NDexo-PDOs. B Circle plot showing the differential interaction strength among different cell types. The blue (orange) colored edges represent increased (decreased) interaction strength in T2Dexo compared to NDexo-PDOs. C, D Circle plot displaying the number of interactions and the strength of interactions between cell types in NDexo-PDOs (C) and T2Dexo-PDOs (D). The number of lines represents the number of interactions, and the thickness of lines is proportional to the strength of the interactions. E Stacked bar plot showing the overall information flow of each significant signaling pathway (p < 0.01). The vertical dashed line indicates the position where the sample accounts for 50% of the overall information flow. Label color is determined by a significantly larger contribution from T2Dexo (blue) or NDexo-PDOs (orange). Pathways equally important in both datasets are labeled in black. P value determined by paired Wilcoxon test according to CellChat. Grey dots indicate pathways highlighted within the main text. F Scatter plot of dominant senders and receivers for NDexo (left) and T2Dexo-PDOs (right). G Scatter plot of the signaling changes associated with T cells in T2Dexo (blue) or NDexo-PDOs (orange). Shared pathways are labeled in gray. Collagen and laminin pathways have been removed from plot for ease of visualization.