Integrated metabolic and immune profiling analysis reveals distinct prognostic signatures for neoadjuvant immunotherapy in locally advanced esophageal cancer

Abstract Background There is an unmet demand for non-invasive biomarker assays to identify patients benefiting from neoadjuvant immune checkpoint inhibitors (ICIs). Here, we aimed to characterize the composition and alterations in plasma metabolites and peripheral blood immune cell subtypes associated with neoadjuvant ICI efficacy in locally advanced esophageal cancer (LAEC) and to investigate potential novel therapeutic targets and non-invasive biomarkers for predicting ICI efficacy. Methods 89 patients with LAEC treated with neoadjuvant programmed cell death 1 blockade combined with chemotherapy were included in this study. We performed an untargeted metabolomic analysis of 606 metabolites on 72 plasma samples using high-performance liquid chromatography-mass spectrometry and an immune profiling analysis of 9 immune cell subtypes on 33 peripheral blood mononuclear-cell samples using flow cytometry. Furthermore, we conducted correlation network and pathway enrichment analyses for potentially beneficial and pathogenic metabolites to explore the metabolite-mediated ICI responses. Finally, a metabolite-based prediction model was established using the least absolute shrinkage and selection operator regression analysis. Results Comparative metabolomics revealed that pyrimidine and purine metabolic pathways were disturbed in ICI non-responders, with significant enrichment of dihydrothymidine, ureidoisobutyric acid, and deoxyadenosine, which were significantly associated with poor survival. Conversely, jasmonic acid increased dramatically in responders and was significantly associated with better survival. Strikingly, tryptophan metabolism intermediate-indole-3-acetic acid and arachidonic acid metabolism intermediate-16(R)-HETE levels were positively correlated with cytotoxic T lymphocyte levels but inversely correlated with polymorphonuclear-myeloid-derived suppressor cells levels, which were markedly associated with a favorable prognosis. Notably, the area under the receiver operating curve for the metabolite-based model predicting 12-month overall survival was 87.7% and 82.6% in the discovery and validation cohorts, respectively, demonstrating promising performance. Conclusions Our work identified potential non-invasive biomarkers based on plasma metabolic signatures for predicting neoadjuvant ICI responses and prognosis in patients with LAEC, which provides novel insight into ICI precision medicine in the management of LAEC.