Mapping the Tumor Microenvironment: Immune Pathway Signatures Predict Immunotherapy Outcomes in Melanoma
核心洞察
A new study reveals that tumor-infiltrating lymphocyte (TIL) patterns in melanoma (搜索) are linked to distinct immune pathway activation, with brisk TILs showing enrichment in antigen presentation and T-cell receptor signaling.
Gene expression scores developed separately for brisk and non-brisk TIL groups demonstrated strong predictive performance for immunotherapy response, with AUCs of 1.000 and 0.852 respectively.
The gene SPP1 (搜索) showed context-dependent behavior, associated with response in brisk TIL tumors but with non-response in non-brisk tumors, highlighting the complexity of the tumor immune microenvironment.
A study published online in Genes & Immunity on 15 July 2026 provides a refined map of the immune landscape within melanoma (搜索) tumors following immunotherapy, offering new insights into why patients with similar clinical histories can experience divergent treatment outcomes. The research, led by de Souza and colleagues at a tertiary cancer center in Brazil, integrates high-resolution immune profiling with pathway-level inference to characterize the tumor microenvironment (TME) beyond conventional blood-based biomarkers.
The retrospective observational study examined 32 patients diagnosed with melanoma (搜索) who received anti–PD-1 (搜索) therapy (nivolumab or pembrolizumab) as monotherapy for advanced disease. Using the NanoString nCounter Human Immunology v2 Panel on primary tumor samples, the team profiled 594 immune-related genes and correlated expression patterns with histopathological TIL classification—brisk versus non-brisk—and clinical outcomes including immunotherapy response, overall survival (OS), progression-free survival (PFS), and post-immunotherapy survival (PIS).
Brisk TILs Align with Broader Immune Transcriptional Programs
Among the 594 genes evaluated, 68 were differentially expressed between brisk and non-brisk TIL patterns. Of these, 64 genes were overexpressed in the brisk group, while only 4 were overexpressed in the non-brisk group. After Benjamini–Hochberg correction, four genes—HLA-C, SELL, TLR1, and ZAP70 (搜索)—remained significantly overexpressed in brisk tumors. External validation using The Cancer Genome Atlas (TCGA-SKCM) cohort of 96 evaluable primary melanoma (搜索) cases confirmed that ZAP70 remained significantly overexpressed in brisk tumors across both cohorts.
"ZAP70 (搜索) is central to T-cell receptor signaling, which could be directly associated with brisk TILs pattern," the authors note. Conversely, CXCL1 showed an inconsistent pattern—upregulated in brisk TILs in the discovery cohort but overexpressed in non-brisk TILs in TCGA—which the researchers suggest may reflect differences in cellular admixture and the non-T-cell inflamed microenvironment.
Pathway analysis using the Gene Set Analysis module revealed that brisk tumors were enriched for multiple immune programs. Cell adhesion (directed global significance score [DGSS] = 2.15) and MHC Class II antigen presentation (DGSS = 2.128) showed the highest scores. At the sample level, T-cell receptor signaling, MHC class I antigen presentation, chemokine signaling, adaptive immune system, type II interferon signaling, and lymphocyte activation remained significant (p < 0.05), though none survived adjustment for multiple testing.
Cell Composition Reveals Functional Differences Beyond Quantity
Cell-type profiling demonstrated that brisk TIL patterns were associated with significantly higher levels of B cells, CD45⁺ cells, cytotoxic cells, neutrophils, T cells, and Treg cells (adjusted p < 0.05). Critically, immune cell ratios revealed a more nuanced picture: non-brisk tumors exhibited higher exhausted CD8⁺ T cells/TILs, macrophages/TILs, NK CD56dim/TILs, and NK cells/TILs ratios (adjusted p < 0.05), consistent with a microenvironment that is "not simply less inflamed, but functionally less effective."
Predictive Gene Scores Differ by TIL Architecture
Genes associated with immunotherapy response differed markedly between brisk and non-brisk groups. Using elastic net penalized logistic regression with bootstrap resampling, the researchers developed a 4-gene score for the brisk group (C5, MAPKAPK2, IL12A, and SPP1 (搜索)) and a 3-gene score for the non-brisk group (IL13RA1, TRAF1, and CFI). Both scores demonstrated strong predictive performance: AUC of 1.000 for the brisk group and 0.852 for the non-brisk group.
After multivariable adjustment, the score remained a significant independent predictor of immunotherapy response in the non-brisk group (p = 0.029), in a model incorporating regression, stage at diagnosis, and line of immunotherapy. In the brisk group, the 4-gene score was significantly associated with OS (p = 0.024) and PIS (p = 0.033), while in the non-brisk group, the 3-gene score was significantly associated with PFS (p = 0.037) and PIS (p = 0.034).
SPP1 (搜索): A Context-Dependent Mediator
A particularly striking finding involved SPP1 (搜索) (osteopontin), which was overexpressed in responder patients within the brisk group but in non-responder patients within the non-brisk group. In brisk tumors, SPP1 was negatively correlated with lymphocyte trafficking, phagocytosis and degradation, chemokine signaling, TNF family signaling, and cytotoxic cells. In non-brisk tumors, SPP1 was positively correlated with lymphocyte trafficking, phagocytosis and degradation, and macrophage proportions, while negatively correlated with cytotoxic cells, Treg cells, and T cells.
The authors interpret this duality as reflecting distinct immune topologies: in non-brisk tumors, high SPP1 (搜索) aligns with an immune-excluded, myeloid-skewed microenvironment where SPP1⁺ macrophages may impair CD8⁺ T-cell infiltration. In brisk tumors, higher SPP1 expression in responders may reflect specific SPP1⁺ macrophage subsets capable of supporting response to checkpoint blockade.
Limitations and Future Directions
The researchers emphasize that their findings are "exploratory and hypothesis-generating," citing key limitations including small cohort size, bulk gene expression analysis that cannot resolve cell-type–specific versus tumor-intrinsic contributions, and the use of primary tumor samples to infer biology relevant to the metastatic setting. Although TCGA validation showed directional consistency, the associations did not reach statistical significance—a finding the authors attribute partly to the fact that TCGA-SKCM was established largely before routine immune checkpoint inhibitor use.
"Validation in larger, independent cohorts and with higher-resolution approaches (e.g., spatial or single-cell profiling) could contribute to additional information regarding prognostic significance of TILs in melanoma (搜索) patients treated with anti-PD-1 (搜索)," the authors conclude. The study nonetheless supports a future where immune monitoring emphasizes the TME's pathway architecture, potentially enabling clinicians to move from measuring immune presence to interpreting immune functionality when making treatment decisions.