Mitochondrial metabolism emerges as a signature of T cell exclusion in esophageal adenocarcinoma
Spatial transcriptomic profiling is sharpening a long-standing question in esophageal adenocarcinoma: why do some tumors remain cold, with few intratumoral CD8+ T cells, while others are infiltrated enough to respond more readily to treatment? In treatment-naive tumors collected before neoadjuvant chemoradiotherapy, cancer cell programs separated cleanly by immune infiltration status. The dominant pattern was not just immune presence or absence. It was metabolic.
Tumors with low CD8+ T cell infiltration showed reduced antigen presentation machinery and increased lipid and mitochondrial metabolism in cancer cells. Tumors with higher CD8+ T cell content moved in the opposite direction, with stronger cytokine signaling and more active apoptotic pathways. The contrast points to a tumor ecosystem in which immune exclusion is not simply a passive failure of T cells to arrive, but a biologically active state maintained by the cancer cells themselves.
The practical consequence is direct. If the tumor cell compartment is rewired toward mitochondrial fitness and lipid use, the immune microenvironment may be pushed away from effective antitumor surveillance.
Spatial mapping separated cancer and immune programs inside the same tumor regions
The study used spatial transcriptomics on treatment-naive CD8-low and CD8-high esophageal adenocarcinomas, with 12 tumors included in the spatial analysis. Regions of interest were chosen from pathologist-annotated tissue, and cancer cells and immune cells were examined within the same spatial context. That design matters because bulk sequencing can blur together cell types that are physically and functionally distinct inside a single lesion.
Within the cancer cell compartment, whole-transcriptome analysis clustered tumors according to T cell infiltration status. Low-CD8 tumors were marked by diminished antigen presentation, a finding that fits with reduced visibility to the adaptive immune system. At the same time, the same tumors showed stronger lipid and mitochondrial metabolic activity. Those changes are consistent with a cancer cell state that may be better equipped to survive hostile conditions while resisting immune pressure.
In the higher-CD8 tumors, transcriptomes were enriched for cytokine signaling and apoptotic pathways. In plain terms, the tissue looked more immunologically engaged and more vulnerable to immune-mediated damage. The split between the two groups was not subtle; it reflected a coordinated shift in both tumor biology and the surrounding immune response.
PGC-1α-linked immune suppression appeared inside the CD45+ compartment
The immune cell compartment added another layer of mechanistic signal. In CD45+ regions from CD8-low tumors, enrichment was seen for the PGC-1α pathway, a program tied to mitochondrial biogenesis. That pathway has a particular relevance in cancer immunology because it can support metabolic states associated with immune suppression rather than immune activation.
The same spatial regions also aligned with immunosuppressive myeloid populations. This is an important detail because T cell exclusion is rarely caused by one pathway alone. In these tumors, the myeloid compartment and the cancer cell compartment appear to reinforce one another. A metabolically active tumor may help recruit or sustain suppressive myeloid cells, and those cells may then help maintain the exclusion of cytotoxic T cells.
The metabolic orientation was not limited to the spatial data. Bulk transcriptomic analyses from TCGA supported the same overall pattern, strengthening the argument that the finding is not an artifact of a small tissue set. Fresh tumor profiling using SCENITH, performed in five specimens, confirmed mitochondrial dependence in macrophages from CD8-low tumors. In other words, the immune compartment in these tumors was not metabolically neutral. It was using mitochondrial respiration in a way that tracked with a suppressed immune state.
Why antigen presentation and metabolism may belong to the same immune evasion program
The biology fits together in a way that is clinically plausible. If cancer cells lower antigen presentation, fewer tumor antigens reach the immune system in a form that T cells can detect efficiently. If the same tumors also heighten lipid handling and mitochondrial respiration, they may become better suited to persist under oxidative and inflammatory stress. Those two features can cooperate: one reduces immune recognition, and the other helps the tumor survive despite immune pressure.
This is especially relevant in esophageal adenocarcinoma, where response to neoadjuvant chemoradiotherapy remains inconsistent. Some patients achieve complete histopathological response, but many do not, and recurrence remains common even after surgery. Existing immunotherapy data in gastroesophageal cancers have shown benefit in selected settings, yet the gains have been more modest than those seen in more immunotherapy-sensitive tumors such as melanoma. The present findings offer one biological explanation for that gap: the tumor may be metabolically configured to keep T cells out.
The study does not claim that mitochondrial metabolism is the only driver of immune exclusion. It does, however, place metabolism close to the center of the problem rather than at its margins. Cancer cell-intrinsic metabolic programs, immune cell mitochondrial dependence, and impaired antigen presentation appear to converge on the same phenotype: a T cell-poor tumor microenvironment.
Therapeutic implications point toward metabolism as an immunologic target
The most immediate translational implication is not a ready-made treatment but a testable strategy. If mitochondrial metabolism helps sustain the cold immune state in these tumors, then metabolic intervention could in theory make the tumor more permissive to T cell entry and improve response to chemoradiotherapy. That hypothesis is especially relevant for patients whose tumors show low baseline CD8+ infiltration, because they are the ones least likely to benefit from immune activation on their own.
The study also reinforces the value of spatial methods in solid tumor immunology. Without spatial transcriptomics, the link between cancer cell metabolism and immune exclusion could have remained hidden inside averaged bulk profiles. Here, the tissue architecture itself was part of the signal. Cancer cells, immune cells, and suppressive myeloid populations were analyzed in the same local environment, revealing a metabolic pattern that would have been harder to reconstruct from dissociated tissue alone.
What remains unresolved is how best to translate this biology into a therapeutic sequence. The findings support further work on mitochondrial targeting, but they do not identify which agent, dose, or combination would be most effective, or whether such an approach would meaningfully improve pathologic response in patients receiving neoadjuvant chemoradiotherapy. Those answers will require functional studies and prospective clinical testing, because the current data define a mechanism more clearly than they define a treatment.