Background
The tumor microenvironment is a site of intense metabolic competition between cancer cells and immune cells. In melanoma, tumor-intrinsic glutamine metabolism and the anti-tumor interferon-gamma (IFNγ) immune response are critical opposing factors. We hypothesized that the interplay between these two pathways, termed a ‘metabolic conflict’, could be captured by a transcriptional signature to stratify patients and predict clinical outcomes.
Methods
We developed transcriptional signatures for glutamine metabolism and IFNγ response pathways and calculated per-sample enrichment scores using Gene Set Variation Analysis (GSVA). We analyzed a discovery cohort of 469 melanoma patients from The Cancer Genome Atlas (TCGA-SKCM) to assess the signature’s prognostic value. We then tested its predictive value for anti-PD-1 therapy in an independent validation cohort of 49 melanoma patients from GEO (GSE91061).
Results
In the TCGA cohort, the glutamine metabolism and IFNγ response scores were negatively correlated (Pearson r = -0.412), supporting the concept of a metabolic-immune conflict. Stratifying patients into four metabolic-immune groups revealed a trend towards improved overall survival for the ‘Favorable’ group (Low Glutamine/High IFNγ), but this did not reach statistical significance (log-rank p = 0.081). Cox proportional hazards models confirmed the signature lacked independent prognostic value when adjusted for clinical covariates. In the immunotherapy-treated validation cohort, the signature failed to predict clinical response to anti-PD-1 therapy (Chi-squared p = 0.706).
Conclusion
A transcriptional signature based on the interplay between glutamine metabolism and IFNγ signaling, while biologically plausible, does not function as an independent prognostic or predictive biomarker in melanoma. Our findings underscore the value of rigorously testing well-founded hypotheses and highlight the complexity of translating metabolic-immune concepts into robust clinical tools.