The investigation began as a large-scale computational sweep. The team integrated bulk transcriptomic datasets, single-cell RNA sequencing and spatial transcriptomics from lung adenocarcinoma patients, cross-referencing these molecular maps with clinical cohorts that had received immune checkpoint blockade. Using weighted gene co-expression network analysis, survival meta-analysis and tissue enrichment scoring, they hunted for stromal cell subpopulations whose presence tracked with aggressive disease and poor immunotherapy outcomes. That search converged on a single culprit: a myofibroblastic cancer-associated fibroblast subset defined by high expression of MMP11, alongside the canonical contractile genes ACTA2 and TAGLN. These MMP11-positive myCAFs were enriched not only in primary lung tumors but also in metastatic lesions, and their abundance correlated with advanced tumor stage, metastatic spread, shorter overall survival and markedly reduced benefit from checkpoint inhibitors.

What makes this finding technically significant is the way the team moved beyond correlation. Spatial transcriptomic deconvolution, a computational method that infers which cell types occupy which physical niches within a tissue section, showed that MMP11-positive myCAFs consistently co-localized with SPP1-positive macrophages, a pro-tumor myeloid population that was itself enriched in malignant and metastatic lung adenocarcinoma tissue. The researchers then confirmed this spatial partnership in human clinical specimens using multiplex immunohistochemistry, a staining technique that visualizes several protein markers simultaneously on a single tissue slide. The two cell types were not merely neighbors in the same tumor; they occupied the same microanatomical neighborhoods, suggesting a functional relationship rather than coincidental distribution.

To identify the molecular language spoken between these cells, the team performed ligand-receptor interaction analysis on their single-cell and spatial data, complemented by virtual gene knockout experiments that computationally predicted the downstream consequences of removing specific signaling molecules. WNT5A emerged as the dominant macrophage-derived mediator linking SPP1-positive macrophages to myCAF activation. WNT5A is a member of the Wingless-related integration site family, a group of secreted glycoproteins that orchestrate embryonic development, tissue patterning and, in cancer, a bewildering array of pro-tumor behaviors. In this context, the analysis indicated that WNT5A released by macrophages acts on fibroblasts, driving them toward a matrix-remodeling, contractile state.

The mechanistic story has two halves. On the macrophage side, the researchers found that SPP1-positive macrophages exhibited activation of transcriptional programs governed by the non-canonical NF-kappaB factors NFKB2 and RELB. When they generated macrophages that overexpressed SPP1 in vitro, levels of Nfkb2, Relb and WNT5A protein all rose, placing SPP1 upstream of a signaling cascade that culminates in WNT5A production. This is a notable detail because it ties a well-known pro-tumor macrophage marker to a specific transcriptional pathway and a specific secreted output, giving the field a concrete molecular handle on a cell type that has often been described only by its marker genes.

On the fibroblast side, the team built a macrophage-CAF co-culture system in which SPP1-overexpressing macrophages and MMP11-overexpressing fibroblasts could interact directly. When fibroblasts were stimulated with WNT5A, they showed activation of YAP-associated signaling and increased expression of INHBA, the gene encoding the activin A subunit. YAP, the transcriptional co-activator downstream of the Hippo pathway, is a well-established sensor of mechanical forces and tissue stiffness, and its activation in fibroblasts is a hallmark of the myofibroblastic state. Functionally, WNT5A-stimulated fibroblasts migrated more readily, contracted more forcefully and deposited more extracellular matrix than their untreated counterparts. Each of these behaviors contributes to the desmoplastic reaction, the dense fibrous scar tissue that many solid tumors build around themselves.

The in vivo experiments brought the story into living systems. The researchers established murine tumor models and evaluated them with in vivo imaging, flow cytometry, atomic force microscopy, hematoxylin and eosin staining, Masson staining and multiplexed immunohistochemistry. Atomic force microscopy deserves particular mention: it measures tissue stiffness at the nanoscale by probing the physical resistance of the extracellular matrix, providing a direct readout of the mechanical remodeling that the fibroblasts were predicted to drive. The results were consistent with the in vitro findings. When WNT5A signaling was activated in the tumor microenvironment, collagen deposition increased, tissue stiffness rose, and MMP11-positive CAFs accumulated.

Most consequentially for patients, WNT5A activation impaired the therapeutic efficacy of PD-1 blockade in these animal models. Flow cytometry revealed reduced infiltration of CD8-positive T cells into the tumors, the very effector cells that anti-PD-1 antibodies are designed to unleash. The picture that emerges is a coherent immunosuppressive circuit: SPP1-positive macrophages, driven by NFKB2/RELB programs, secrete WNT5A; WNT5A activates YAP/INHBA signaling in fibroblasts; the fibroblasts remodel and stiffen the matrix; and the resulting dense, rigid stroma excludes cytotoxic T lymphocytes, leaving checkpoint inhibitors with little to work with even when the drug successfully engages its target on T cells.

The clinical implications are substantial. Immune checkpoint blockade has reshaped outcomes in lung adenocarcinoma, yet primary and acquired resistance remain stubborn problems, and stromal exclusion of T cells is one of the best-documented mechanisms of failure. By identifying SPP1-positive macrophages and MMP11-positive myCAFs as a spatially coupled, functionally linked pair, and by pinpointing WNT5A as the bridge between them, the study offers a set of potential biomarkers and drug targets. Patients whose tumors are rich in MMP11-positive myCAFs might be identified in advance as unlikely to benefit from PD-1 monotherapy and steered toward combination strategies. Meanwhile, the WNT5A-YAP axis suggests that disrupting macrophage-fibroblast communication could soften the tumor matrix, restore T-cell infiltration and resensitize tumors to immunotherapy. The authors note that their findings highlight a potential therapeutic strategy to overcome immune checkpoint blockade resistance, though translating that insight into clinics will require inhibitors capable of safely modulating WNT signaling in humans.

Methodologically, the study is also a demonstration of how modern cancer biology is increasingly conducted. No single technique carried the argument; instead, bulk transcriptomics established population-level associations, single-cell RNA sequencing resolved the relevant cell states, spatial transcriptomics and multiplex immunohistochemistry anchored those states in physical tissue architecture, virtual knockout and co-culture experiments tested causal mechanisms in vitro, and mouse models with mechanical measurements confirmed the circuit in vivo. The human specimens were obtained under ethics approval from Zhejiang Cancer Hospital, and the work was supported by the National Natural Science Foundation of China and the Traditional Chinese Medicine Scientific Research Foundation of Zhejiang Province. As with any study that combines computational inference with experimental validation, the next step will be independent confirmation in larger immunotherapy cohorts and the development of clinically tractable ways to interrupt the WNT5A signal. If those hurdles can be cleared, the stromal conversation between macrophages and fibroblasts may become one more vulnerable link in the armor that lung adenocarcinomas use to shield themselves from the immune system.

Subject of Research: The role of SPP1-positive macrophage-derived WNT5A in MMP11-positive myofibroblastic cancer-associated fibroblast activation, matrix remodeling, and immunotherapy resistance in lung adenocarcinoma.

Article Title: SPP1⁺ macrophage-derived WNT5A contributes to MMP11⁺ myCAF-mediated matrix remodeling and immunotherapy resistance in lung adenocarcinoma

Article References: Wang, M., Zhang, P., Li, X., Shao, L., Dai, S., & Chen, Z. (2026). SPP1⁺ macrophage-derived WNT5A contributes to MMP11⁺ myCAF-mediated matrix remodeling and immunotherapy resistance in lung adenocarcinoma. Journal of Translational Medicine. https://doi.org/10.1186/s12967-026-09069-y

Image Credits: AI Generated

DOI: 10.1186/s12967-026-09069-y

Keywords: lung adenocarcinoma, WNT5A, SPP1 macrophages, cancer-associated fibroblasts, MMP11, matrix remodeling, immunotherapy resistance, PD-1 blockade, tumor microenvironment, YAP signaling, single-cell RNA sequencing, spatial transcriptomics

The study, published in the British Journal of Cancer, combined two of the most powerful tools in modern genomics: spatial transcriptomics, which records gene activity while preserving the physical architecture of tissue, and single-cell RNA sequencing, which profiles gene expression in thousands of individual cells. The researchers applied spatial transcriptomics to colorectal tumor tissue and matched normal tissue from a patient with early-stage colon cancer, then validated their discoveries across 41 publicly available spatial transcriptomics slides, single-cell datasets representing 472 colorectal cancer patients, and bulk transcriptomic data from 617 more. This layered approach allowed them to see not just which cells were present in the tumor microenvironment, but exactly where they sat, whom they were talking to, and what molecular messages they were exchanging.

The first major insight concerned the geography of invasion. Using computational deconvolution methods such as SpaCET and RCTD, the team estimated the proportions of different cell types within each microscopic spot of tissue. They found that tumor clusters, defined by markers like CD44, EPCAM, and CDH1, extended beyond the boundaries visible under the microscope, infiltrating the surrounding stroma. A complementary reference-free analysis with STdeconvolve revealed even greater spatial complexity, showing tumor-like gene signatures pushing further into normal tissue than standard spot-based resolution could detect. Along this invasive trajectory, the researchers observed a coordinated upregulation of genes driving epithelial-to-mesenchymal transition, matrix metalloproteinases such as MMP2, MMP9, and MMP11, and inflammatory chemokines, while genes maintaining intestinal integrity were switched off.

Perhaps the most striking structural discovery was a ring of SPP1-positive macrophages encircling the tumor, a barrier-like arrangement invisible in routine hematoxylin and eosin staining. These macrophages displayed a classic M2-like, immunosuppressive phenotype, expressing high levels of CCL18, CCL22, CD276, and metalloproteinases that remodel the extracellular matrix in ways that support cancer invasion. Crucially, the spatial data showed that regulatory T cells, or Tregs, the immune cells whose job is normally to restrain immune responses, were concentrated within the tumor core, nestled inside this macrophage ring. Ligand-receptor interaction analysis revealed the communication channels at work: CCL18 signaling through CCR8, and CCL22 through CCR4, both acting as chemical beacons drawing Tregs into the tumor. In publicly available spatial datasets, these SPP1 macrophage and Treg neighborhoods were predominantly localized at the invasive margins of tumors, suggesting a conserved architectural principle of immune evasion.

The tumor cells themselves were far from passive. All tumor clusters showed elevated expression of TGFB1, a potent immunosuppressive cytokine, along with the inhibitory checkpoint molecule CD276. The researchers also documented a metabolic form of immune suppression: tumor clusters exhibited heightened KYNU-AHR signaling, a pathway tied to tryptophan deprivation, which is known to starve anti-tumor T cells of an essential amino acid. This same axis appeared to induce CXCL5, a macrophage attractant, reinforcing the recruitment of M2-like macrophages, particularly in late-stage disease. Tumor clusters also activated the CD24-SIGLEC10 interaction, a so-called don’t-eat-me signal that prevents macrophages from engulfing cancer cells, and upregulated CD47 in the tumor core, another molecular shield against phagocytosis that has previously been linked to poor responses to PD1 and PD-L1 checkpoint inhibitors.

Not all immune structures in the tumor microenvironment are hostile. The team identified immune cell aggregates resembling tertiary lymphoid structures, organized clusters of immune cells that form in non-lymphoid tissue and are generally associated with favorable prognosis. Interaction analysis confirmed active T cell homing through CXCL13-CXCR5, CCL19-CCR7, and CCL21-CCR7 signaling, robust antigen presentation via MHC molecules, and T cell receptor signaling. Yet even within these seemingly protective structures, the researchers detected immunosuppressive undertones: enhanced PVR-TIGIT and PVR-CD96 inhibitory signaling, chemokine interactions capable of recruiting CCR4-positive Tregs, and elevated CD276. The picture that emerged was one of a battleground where anti-tumor immunity and immune suppression coexist in uneasy proximity.

To trace how these immune aggregates change as normal tissue becomes cancerous, the researchers applied pseudotime trajectory analysis, a computational method that orders cells or tissue spots along a developmental path based on gene expression. Two lineages emerged: a normal lineage and a tumor lineage. The tumor lineage showed heightened antibacterial and cytotoxic gene expression, active antigen presentation, and markers of ongoing angiogenesis, including MMP2-PECAM1 and TIMP3-CD44 interactions. But it also displayed a troubling signature: decreased co-stimulatory CD28 and increased expression of SIT1 and SIRT2, both negative regulators of T cell receptor signaling. SIT1, a transmembrane adaptor protein previously shown in mouse models to inhibit TCR-mediated signaling, had never been implicated in colorectal cancer immune suppression. Its elevated expression in tumor-associated immune aggregates, confirmed in public spatial datasets, suggests a previously unrecognized mechanism by which tertiary lymphoid structures might be functionally disarmed even as they form.

The study’s most novel discovery, however, was an entirely new subpopulation of regulatory T cells. Within FOXP3-positive spots, the hallmark of Tregs, in a specific tumor cluster, the researchers found high expression of IFI6, an interferon-induced gene, alongside ISG15 and a suite of other interferon-related genes including IFIT1, IFIT3, IFI44, and the IFITM family. Validation in single-cell data from 472 patients confirmed that IFI6-positive Tregs were significantly enriched in tumors and metastases compared with peripheral blood, healthy tissue, adjacent normal tissue, and lymph nodes, and were present across all stages of disease progression. Bulk transcriptomic analysis of 617 patients independently confirmed the elevation of this signature in primary tumors. These cells also expressed high levels of LGALS9, or galectin-9, an immune checkpoint molecule that has shown promise as a therapeutic target in early clinical trials.

The clinical implications of the IFI6-positive Treg population are considerable. The proportion of these cells was significantly increased in patients who did not respond to therapy, and a signature score combining Treg fraction with IFI6 and LGALS9 expression was associated with significantly reduced survival in The Cancer Genome Atlas colorectal cohort. Interaction analysis suggested that these Tregs are recruited to the tumor through multiple channels, including TFF3-CXCR4 signaling from cancer cells, CXCL16-CXCR6 from dendritic cells and SPP1 macrophages, and CXCL13 and CXCL14 from other immune and stromal cells. Once inside, they appear to suppress immunity through MHCII-mediated interactions with antigen-presenting cells, CD86-CTLA4 engagement with dendritic cells, and inhibition of NK cells via the HLA-E-KLRD1 checkpoint axis. IFI6 itself, an apoptosis suppressor, has been linked to colorectal cancer proliferation and, in esophageal carcinoma, to a mesenchymal and immunosuppressive microenvironment, reinforcing the plausibility of its role here.

The authors are careful to note the limitations of their work. The spatial transcriptomics data derive from a limited number of patient samples, and the computational discoveries await functional validation in laboratory and animal models. Protein-level confirmation in larger independent cohorts will be needed before these findings reach the clinic. Nevertheless, the study establishes a compelling framework: colorectal tumors construct layered immunosuppressive niches, with SPP1-positive macrophages forming a physical and chemical barrier that recruits Tregs via CCL18 and CCL22, tumor cells deploying tryptophan metabolism and don’t-eat-me signals, tertiary lymphoid structures being quietly undermined by SIT1, and a newly identified interferon-primed Treg population associated with treatment failure and poor survival. Each of these molecular players, from CCR8 and CCR4 to LGALS9 and HLA-E, represents a potential point of therapeutic intervention, offering a roadmap for the next generation of personalized immunotherapies against one of the world’s most formidable cancers.

Subject of Research: Spatial and single-cell transcriptomic characterization of immunosuppressive niches in colorectal cancer

Article Title: Integrated spatial and single-cell transcriptomics uncover IFI6⁺ regulatory T-cells and SPP1⁺ macrophages immunosuppressive niches in colorectal cancer

Article References: Miroszewska, D., Song, S., Urbiola-Salvador, V., Cázares Olivera, M., Jabłońska, A., Bolcewicz, M., Kalinowski, L., Duzowska, K., Drężek-Chyła, K., Zdrenka, M., Śrutek, E., Szylberg, Ł., Jankowski, M., Bała, D., Zegarski, W., Nowikiewicz, T., Filipowicz, N., Piotrowski, A., Dumanski, J. P., … Chen, Z. (2026). Integrated spatial and single-cell transcriptomics uncover IFI6⁺ regulatory T-cells and SPP1⁺ macrophages immunosuppressive niches in colorectal cancer. British Journal of Cancer. https://doi.org/10.1038/s41416-026-03637-1

Image Credits: AI Generated

DOI: 10.1038/s41416-026-03637-1

Keywords: colorectal cancer, spatial transcriptomics, single-cell RNA sequencing, regulatory T cells, SPP1 macrophages, IFI6, tumor microenvironment, immune evasion, LGALS9, tertiary lymphoid structures, immunotherapy, CCL18-CCR8 signaling

The research began with a deceptively simple question: which cells inside a head and neck tumor are actually burning LDL cholesterol? Using single-cell RNA sequencing, the team scored every cell in the tumor microenvironment for the activity of LDL-related metabolic genes. The answer was striking. Rather than the cancer cells themselves, it was the myeloid compartment, the family of innate immune cells that includes monocytes, dendritic cells, and macrophages, that showed the highest LDL metabolic activity. Within that compartment, macrophages stood out as the most metabolically active population, suggesting that these scavenger cells, which are known to engulf lipids and cellular debris, are the primary consumers of cholesterol traffic inside the tumor.

This finding matters because macrophages are not passive bystanders. Depending on the signals they receive, they can either rally anti-tumor immune responses or actively suppress them, and their metabolic state is increasingly understood to dictate which behavior dominates. The researchers went a step further and dissected the macrophage population into subtypes, uncovering a subset defined by expression of the secreted protein SPP1, also known as osteopontin. These macro_SPP1 macrophages displayed a double dose of pathological activity: heightened LDL metabolism and elevated extracellular matrix remodeling, two processes closely linked to tumor invasion and fibrosis. Critically, this subset also showed upregulation of ITGA5, the gene encoding integrin alpha 5, a cell-surface receptor subunit that anchors cells to fibronectin in the extracellular matrix.

To translate these single-cell observations into something clinically useful, the team built a machine learning-based prognostic model centered on LDL metabolism-associated genes. When applied to independent validation datasets, the model reliably stratified head and neck squamous cell carcinoma patients into high-risk and low-risk groups with significantly different overall survival. The stratification held up in multivariate Cox regression analyses that accounted for standard clinical variables such as tumor stage, nodal status, and grade, indicating that the LDL metabolic signature carries prognostic information beyond what clinicians already measure. Receiver operating characteristic analysis and concordance index calculations confirmed that the model’s predictive performance was robust across cohorts.

Perhaps the most provocative results emerged when the researchers overlaid the risk score onto immunotherapy data. In an independent cohort of patients treated with immune checkpoint blockade, high-risk patients responded more poorly to therapy. The high-risk group also exhibited a distinctive immune landscape: elevated tumor mutational burden, which in many cancer types predicts better immunotherapy outcomes, but simultaneously reduced immune cell infiltration into the tumor. This paradoxical combination, a genetically noisy tumor that somehow keeps immune cells at arm’s length, is precisely the scenario in which checkpoint inhibitors often fail, and it suggests that LDL metabolic reprogramming may be one of the mechanisms by which tumors construct an immunologically cold microenvironment despite carrying abundant mutations for the immune system to recognize.

The team then turned to spatial transcriptomics to verify that these molecular patterns correspond to actual geography within tissue. The technique, which preserves the physical layout of cells while reading their gene expression, confirmed that ITGA5 was enriched in tumor regions with high LDL metabolic activity. This spatial concordance is important because it rules out the possibility that the single-cell findings were artifacts of cell dissociation, and it places the ITGA5-LDL axis squarely within the tumor proper rather than the surrounding stroma. Correlation analyses in both bulk transcriptomic datasets and single-cell data reinforced the link, showing that ITGA5 expression tracks with canonical cholesterol metabolism genes including LDLR, SQLE, and SOAT1 across multiple independent cohorts.

To test whether ITGA5 is merely a bystander marker or an active driver of malignancy, the researchers performed functional experiments in head and neck cancer cell lines. When ITGA5 expression was knocked down using short hairpin RNA, the cells lost key malignant capabilities: proliferation measured by CCK-8 and EdU assays declined, and both migration and invasion were impaired. Conversely, the survival analysis showed that patients with low ITGA5 expression had more favorable clinical outcomes. Intriguingly, treatment of HN-5 and UMSCC-47 cells with LDL or oxidized LDL modulated the expression of ITGA5 and several macrophage-associated lipid genes including SPP1, TREM2, APOE, and LPL, hinting at a biochemical conversation between circulating lipids and the tumor’s transcriptional state.

The study also explored what distinguishes ITGA5-high from ITGA5-low macrophages within the macro_SPP1 subset. Gene set enrichment analysis revealed that the ITGA5-high macrophages were enriched for pathways including epithelial-mesenchymal transition, TNF/NF-kappaB signaling, hypoxia, glycolysis, mTORC1 signaling, and interferon-gamma response, while showing alterations in oxidative phosphorylation and antigen processing and presentation. This transcriptional profile paints a picture of macrophages that are metabolically rewired, pro-inflammatory in a potentially tumor-promoting way, and less capable of properly presenting antigen to T cells, a combination that would be expected to blunt anti-tumor immunity while supporting tissue invasion and matrix remodeling.

Taken together, the findings establish LDL metabolic reprogramming as a feature of head and neck squamous cell carcinoma that operates on two fronts simultaneously. In the immune compartment, it defines a macrophage state associated with matrix destruction and immunosuppression; in the malignant compartment, ITGA5 acts as a tumor cell-intrinsic effector that directly promotes proliferation, migration, and invasion. The dual role is unusual and clinically meaningful, because it suggests that a single metabolic axis could be targeted to influence both the cancer cells and the immune cells that surround them. The prognostic model developed from this axis could, in principle, help clinicians identify patients unlikely to benefit from immune checkpoint blockade and direct them toward alternative or combination strategies.

As with any multi-omics study, important caveats remain. The prognostic model was developed and validated on retrospective datasets, and prospective clinical validation would be needed before the risk score could guide treatment decisions. The functional experiments, while convincing at the cellular level, do not yet establish whether manipulating LDL metabolism in patients would alter tumor behavior, and the relationship between dietary or pharmacological cholesterol management and cancer outcomes remains an open question. Nevertheless, by demonstrating that a metabolic pathway classically associated with heart disease is woven into the architecture of the head and neck tumor microenvironment, the study opens a genuinely new line of investigation. If future work confirms that targeting the ITGA5-LDL axis can reprogram macrophages and sensitize tumors to immunotherapy, the humble cholesterol particle may find an unexpected second act in oncology.

Subject of Research: LDL metabolism-associated myeloid signatures and tumor cell-intrinsic ITGA5 in head and neck squamous cell carcinoma

Article Title: Single-cell analysis identifies an LDL metabolism–associated myeloid signature and tumor cell–intrinsic ITGA5 in HNSCC

Article References: Ma, H., Wang, S., Xiong, L., Hahan, Z., Zhao, P., Wu, Y., Wei, M., & Wang, X. (2026). Single-cell analysis identifies an LDL metabolism–associated myeloid signature and tumor cell–intrinsic ITGA5 in HNSCC. BMC Cancer. https://doi.org/10.1186/s12885-026-17006-x

Image Credits: AI Generated

DOI: 10.1186/s12885-026-17006-x

Keywords: head and neck squamous cell carcinoma, LDL metabolism, macrophages, single-cell RNA sequencing, ITGA5, tumor microenvironment, immunotherapy, tumor mutational burden, spatial transcriptomics, cholesterol metabolism, prognostic model, SPP1 macrophages

The research effort was led by Ilkyu Park, Jae-Ghi Lee, and Jisup Kim, who contributed equally to the work, under the corresponding authorship of Jung Ho Kim at Gachon University Gil Medical Center in Incheon. Rather than relying on bulk sequencing, which averages molecular signals across large chunks of tissue and destroys the crucial architectural context in which cells operate, the team embraced a dual-platform strategy. GeoMx Digital Spatial Profiling allowed them to measure gene expression across precisely annotated regions of interest, while CosMx Spatial Molecular Imaging delivered single-cell and subcellular resolution across hundreds of fields of view. Together, these technologies enabled the researchers to observe not merely which genes were active, but exactly where in the tissue those genes were being switched on and which neighboring cells were talking to one another.

The scale of the study is one of its distinguishing strengths. The investigators analyzed tissue from 89 colorectal cancer patients across 288 carefully annotated regions, encompassing tumor compartments, adjacent normal mucosa, stromal zones, and adipose tissue. They then validated and extended their observations in an independent cohort of 24 patients, evenly split between 12 individuals with obesity and 12 without, spanning 969 individual fields of view. This depth of sampling across both pre-neoplastic and tumor states allowed the team to trace molecular trajectories through time, comparing obese-normal mucosa, non-obese-normal mucosa, tumors from patients with obesity, and tumors from patients without obesity, using formalin-fixed paraffin-embedded clinical specimens that reflect real-world pathology practice.

The first major discovery emerged from an unexpected place: tissue that looked perfectly healthy under the microscope. In the normal-appearing mucosa of patients with obesity, the researchers identified a transcriptionally shifted epithelial state, meaning the genes expressed by the intestinal lining cells had already drifted into a different program even though the tissue retained its normal morphology. Within this obese-normal tissue, the team detected two distinct expression trajectories among a set of 30 genes linked to colorectal cancer. Crucially, a subset of these genes displayed early transcriptional alterations in the obese-normal tissue that were maintained all the way through tumor development. In other words, obesity appears to imprint a molecular signature on the gut epithelium before any visible lesion arises, and that signature persists as cancer takes hold, functioning as a kind of molecular breadcrumb trail from healthy tissue to malignancy.

This pre-neoplastic reprogramming carries significant implications for how scientists think about cancer initiation. The conventional view of carcinogenesis often centers on mutations accumulating over decades, but this work suggests that the systemic metabolic environment created by obesity can push epithelial cells into altered transcriptional states long before genetic damage is fully manifest in tissue architecture. Pathways related to epithelial identity and metabolic reprogramming, including signatures consistent with shifts in processes such as the pentose phosphate pathway and epithelial-mesenchymal transition programs, appear to be engaged in this early phase. If the epithelial lining of a person with obesity is already operating under a different gene-expression regime, interventions aimed at preventing colorectal cancer in this population may need to begin earlier and target these early transcriptional changes rather than waiting for polyps or tumors to appear.

The second phase of the obesity-driven reprogramming unfolds within the tumor itself, and it is immunological in nature. As tumors progressed in patients with obesity, the spatial organization of immune cells inside the tumor microenvironment changed in a distinctly unfavorable direction. The researchers documented a localized enrichment of macrophages expressing SPP1, a gene encoding the secreted protein osteopontin that has been implicated in tumor progression, metastasis, and immune evasion across multiple cancer types. These SPP1-positive macrophages were not scattered uniformly through the tissue; instead, they accumulated in specific spatial niches, clustering near cancer cells where their positioning could most effectively shape tumor behavior.

Even more telling was the ratio shift the team observed. In obese tumors, the balance between SPP1-positive macrophages and pro-inflammatory macrophages changed in favor of the SPP1-positive population, particularly in the regions immediately adjacent to cancer cells. At the same time, the spatial analysis revealed enhanced growth-factor and cytokine-mediated signaling emanating from these localized immune niches. This pattern describes a form of spatially restricted immunosuppression: rather than a wholesale depletion of immune cells, obesity appears to redistribute the immune landscape so that pro-tumor, tissue-remodeling macrophage states dominate the critical interfaces where tumors and immune defenses meet. Such remodeling could help explain clinical observations that obesity-associated colorectal cancers often behave more aggressively and respond differently to immunotherapy than tumors arising in patients of normal weight.

Synthesizing these observations, the authors propose a two-phase model of obesity-driven colorectal carcinogenesis. In the first phase, obesity induces early epithelial transcriptional reprogramming in morphologically intact mucosa, priming the tissue for transformation. In the second phase, once tumors have formed, obesity fosters spatially constrained immunosuppressive remodeling, concentrating SPP1-positive macrophages and pro-tumor signaling at the tumor-immune frontier. The two phases are complementary: an epithelium already nudged toward a cancer-prone transcriptional state faces a tumor microenvironment that has been simultaneously engineered to suppress the immune responses that might otherwise eliminate emerging malignant cells. This dual mechanism reframes obesity not simply as a statistical risk factor but as an active biological agent that rewrites both the target tissue and its defensive ecosystem.

The technical achievement underlying these conclusions deserves emphasis, because it illustrates why spatial transcriptomics has become one of the most transformative tools in modern cancer biology. Traditional single-cell RNA sequencing requires dissociating tissue, losing the spatial coordinates that define how cells actually interact in the body. The GeoMx and CosMx platforms used in this study preserved those coordinates, allowing the researchers to interrogate tumor cores, invasive margins, stromal compartments, adipose depots, and normal mucosa within the same specimens. The identification of cell-cell interaction networks, immune composition shifts, and expression trajectories would have been impossible with dissociative methods, and the concordance of findings across two independent platforms and two patient cohorts strengthens the reliability of the results considerably.

Clinically, the study opens several concrete avenues. The early epithelial gene-expression signature identified in obese-normal mucosa could serve as the basis for biomarker panels that stratify colorectal cancer risk in patients with obesity, potentially guiding the timing and intensity of colonoscopy surveillance. The spatially defined enrichment of SPP1-positive macrophages suggests a specific, targetable node in the immunosuppressive architecture of obesity-associated tumors; therapies designed to deplete, repolarize, or block the recruitment of this macrophage subset could restore anti-tumor immunity where it matters most. Moreover, because the macrophage remodeling is spatially restricted, combinatorial approaches that pair such targeted interventions with existing checkpoint inhibitor immunotherapies may prove particularly effective for this patient population. The work was supported by the National Research Foundation of Korea and approved by the institutional review board at Gachon University Gil Medical Center, with written informed consent obtained from all participants.

As obesity rates continue to climb worldwide and colorectal cancer incidence rises in younger populations, understanding the mechanistic links between the two has become a global health priority. This study demonstrates that the damage inflicted by obesity on colorectal tissue is not a late consequence of established disease but an early, detectable, and spatially organized process that unfolds across years of tumor evolution. By mapping that process at unprecedented resolution, the researchers have transformed obesity from a crude demographic risk marker into a set of actionable molecular and cellular targets, offering new hope that the elevated cancer burden carried by patients with obesity can one day be anticipated, intercepted, and treated with far greater precision than is possible today.

Subject of Research: Obesity-driven spatial reprogramming of the epithelial and immune ecosystem in colorectal cancer

Article Title: Obesity reprograms the spatial ecosystem of colorectal cancer through early epithelial reprogramming and immunosuppressive remodeling

Article References: Park, I., Lee, J.-G., Kim, J., Kim, S.-H., Nam, S., Lee, W.-S., Lee, H., Jang, Y., Chung, J.-W., Kim, K. O., Kwon, K. A., & Kim, J. H. (2026). Obesity reprograms the spatial ecosystem of colorectal cancer through early epithelial reprogramming and immunosuppressive remodeling. Genome Medicine. https://doi.org/10.1186/s13073-026-01758-z

Image Credits: AI Generated

DOI: 10.1186/s13073-026-01758-z

Keywords: colorectal cancer, obesity, spatial transcriptomics, tumor microenvironment, epithelial reprogramming, SPP1 macrophages, immunosuppression, immune remodeling, cancer risk, Genome Medicine, immunotherapy, risk stratification