Small spatial-transcriptomics study links metabolism to T-cell exclusion in esophageal adenocarcinoma

A small clinical study suggests that lipid metabolism and mitochondrial respiration may help esophageal adenocarcinoma keep cytotoxic T cells at bay, potentially limiting responses to chemoradiotherapy and immune-checkpoint blockade. Researchers combined spatial transcriptomics, an independent cancer database and ex vivo metabolic measurements to compare tumors with abundant CD8-positive T cells against those with few of these cells. The strongest pattern was not a previously emphasized oncogenic pathway but a shared metabolic state involving cholesterol handling, ketone metabolism and oxidative phosphorylation.

The results identify associations rather than establish that metabolism directly causes T-cell exclusion, and the authors emphasize that the spatial analysis included only 12 patients, with particularly few usable immune-cell regions in the low-infiltration group. Previous work in this series of patients had linked denser pretreatment infiltration of CD8-positive T cells to a complete pathological response after neoadjuvant chemoradiotherapy. In the new analysis, four of six tumors with the highest CD8 scores showed a complete response, whereas none of the six with the lowest scores did.

That association prompted the team to search for cancer-cell-intrinsic and immune-cell features that differed between inflamed and excluded regions. Using the GeoMx Digital Spatial Profiler, the researchers profiled the whole transcriptome in treatment-naive, stage II or III esophageal or gastroesophageal-junction adenocarcinomas. A pathologist selected regions of interest, after which cancer cells identified by cytokeratin staining and immune cells identified by CD45 staining were analyzed separately.

After quality control, 34 cancer-cell segments and 22 immune-cell segments remained for further study, drawn from 41 regions across the 12 selected tumors. The selected patients represented the extremes of CD8 infiltration: the final set had a median of 20.6 CD8-positive cells per square millimeter in the low group and 224.2 per square millimeter in the high group. Cancer-cell profiles separated mainly according to the tumor’s CD8-infiltration status.

Regions from the same patient generally clustered together, suggesting that patient-specific programs were important, although one tumor contained both CD8-low and CD8-high transcriptional regions. These cancer cells expressed higher levels of APOA1 and APOA4, lipid transporters associated with cholesterol handling; PPARGC1A, which regulates mitochondrial biogenesis and oxidative metabolism; and OXCT1, a mitochondrial enzyme involved in ketone-body metabolism. Such metabolic differences were more prominent than differences in canonical oncogenic pathways.

Pathway analysis reinforced the metabolic pattern. Of 400 differentially expressed pathways, lipid-related programs—including high-density lipoprotein and chylomicron remodeling, lipoprotein remodeling and ketone metabolism—were elevated in cancer cells from CD8-low tumors. In contrast, CD8-high regions showed stronger antigen-presentation, interferon-alpha signaling, interleukin-7 signaling and Fas-mediated immune-apoptosis pathways.

The low-infiltration cancer cells also displayed several immune-suppressive features: reduced expression of antigen-presentation components in half of the affected patients, increased checkpoint-related transcripts, and higher expression of the tryptophan-depleting enzyme IDO1. At the same time, they expressed cancer-testis antigens that can elicit T-cell responses, underscoring the possibility that tumor cells may suppress presentation of antigens capable of attracting an immune response. The immune-cell compartments pointed in a similar direction, although the evidence was less balanced.

Only four CD45-positive segments from three CD8-low tumors passed quality control, compared with 18 segments from five CD8-high tumors, because immune-cell-depleted regions often yielded too few CD45-positive cells. In the available low-infiltration material, APOA1, APOA4, LGALS2, MMP1 and INHBA were enriched; these genes are associated with lipid metabolism or suppressive myeloid-cell populations. Single-sample pathway analysis also identified steroid-hormone metabolism, amino-acid and peptide transport, ubiquinol biosynthesis and the PGC-1alpha program in the low-infiltration immune compartments.

PGC-1alpha is a central regulator of mitochondrial biogenesis, making its enrichment compatible with a population of myeloid cells that depends more on oxidative phosphorylation than glycolysis. The researchers then tested whether the association appeared in a larger, independent dataset. They analyzed bulk messenger-RNA profiles from 65 treatment-naive, chromosomally unstable esophageal adenocarcinomas in The Cancer Genome Atlas and divided patients by the median cytotoxic T-lymphocyte signature.

Tumors in the lower half ranked highly for mitochondrial translation, respiratory electron transport, the tricarboxylic acid cycle and oxidative phosphorylation. Single-sample analysis also associated cholesterol metabolism with greater macrophage infiltration and ketone-body metabolism with lower CD8-positive T-cell infiltration. These results support a broad association between mitochondrial metabolic programs and lymphocyte-poor tumors, but the retrospective bulk data cannot show that the pathways caused the difference in immune composition.

Fresh tissue provided a functional test of the transcriptomic observations. Using SCENITH, a flow-cytometry assay that estimates dependence on glycolysis or mitochondrial oxidative phosphorylation from changes in protein synthesis, the team profiled five resected tumors, including esophageal, gastroesophageal-junction and gastric adenocarcinomas. Three specimens were dominated by macrophages and classified as T-cell-low, whereas two lymphocyte-rich tumors included a microsatellite-instable cancer used as a positive control.

Macrophage-like cells from the T-cell-low cancers showed greater mitochondrial dependence, while those from lymphocyte-rich tumors were more glycolysis-dominant. Because the five specimens differed in anatomical origin, prior treatment and immune composition, the experiment supports the proposed metabolic association but does not isolate tumor biology from specimen-level differences. Separate co-culture experiments offered a possible mechanism.

Cancer cell lines with greater oxidative-phosphorylation reliance tended to produce a lower ratio of the macrophage marker CD80 to the alternative-activation marker CD163 than lines relying more on glycolysis, suggesting a shift toward a less inflammatory macrophage phenotype. The result varied among monocyte donors, and the most glycolytic line did not increase the CD80-to-CD163 ratio, indicating that cytokines, direct cell contact and other factors also shape macrophage state. The experiment therefore supports an association between tumor-cell metabolism and macrophage polarization rather than proving that OXPHOS alone causes suppression.

The authors noted that lipid-rich environments can favor mitochondrial respiration in regulatory and myeloid cells, while providing less favorable metabolic conditions for antitumor T cells. Together, the findings nominate lipid metabolism and mitochondrial respiration as possible contributors to the T-cell-excluded state in esophageal adenocarcinoma, while also highlighting reduced antigen presentation as a complementary mechanism. They provide a rationale for testing metabolic interventions alongside immune activation, but they do not demonstrate that such treatment will improve outcomes in patients.

The study was limited by its small selected cohort, unequal immune-segment numbers, retrospective validation dataset and heterogeneous ex vivo samples. The authors have initiated a clinical trial using metformin as a metabolic modulator to test more directly whether changing metabolism can modify the antitumor immune response. Until controlled clinical results are available, the central conclusion is that oxygen-consuming, lipid-linked metabolic states mark esophageal adenocarcinomas that exclude CD8-positive T cells and may help explain resistance to existing immune-directed therapy.

Subject of Research: Spatial analysis of metabolic programs associated with T-cell exclusion in esophageal adenocarcinoma

Article Title: Spatial transcriptomic sequencing identifies lipid metabolism and mitochondrial respiration as central features of T cell excluded esophageal adenocarcinomas

Article References: Sanders, J., Hahn, N., Bos, E. N., van Schie, D. M., Nijman, I. J., van Grieken, N. C. T., van der Peet, D. L., van Laarhoven, H. W. M., Grootjans, J., Bijlsma, M. F., Van den Bossche, J., de Gruijl, T. D., & Derks, S. (2026). Spatial transcriptomic sequencing identifies lipid metabolism and mitochondrial respiration as central features of T cell excluded esophageal adenocarcinomas. British Journal of Cancer. https://doi.org/10.1038/s41416-026-03635-3

Image Credits: AI Generated

DOI: 10.1038/s41416-026-03635-3

Keywords: esophageal adenocarcinoma, spatial transcriptomics, T-cell exclusion, lipid metabolism, mitochondrial respiration, oxidative phosphorylation, tumor microenvironment, macrophages, antigen presentation, immune suppression, Spatial, transcriptomic