Uncovering potential molecular biomarkers for cancer-associated secondary lymphedema through integrated analyses of RNA-sequencing, machine learning, and clinical data.

Background: Cancer-associated secondary lymphedema (CASL) commonly occurs after tumor-related lymph node dissection and radiotherapy. Nevertheless, the mechanisms of CASL remain unclear, and there are no specific molecular markers for its diagnosis and treatment. Methods: In this study, RNA sequencing was performed on adipose tissues from 10 normal controls and 40 patients with CASL. Differentially expressed genes were screened using two machine learning algorithms to identify potential molecular markers for CASL. Subsequently, seven machine learning algorithms were employed to develop predictive models based on the identified markers. The contribution of each feature to the predictive outcomes was interpreted using Shapley additive explanation (SHAP). Immune cell infiltration was profiled through CIBERSORT and MCP-counter algorithms, and single-cell RNA sequencing (scRNA-seq) data were integrated to explore interactions between characteristic genes and immune cell subpopulations. Furthermore, associations between characteristic genes and clinical parameters were also assessed. Results: IL2RG, HOXD10, and TSPAN1 were identified as potential biomarkers of CASL. Diagnostic models built on these three genes showed excellent performance. Functional enrichment analysis suggested that the dysregulation of cytokine-cytokine receptor interactions and immune pathways underlies the pathological progression of CASL. In addition, immune infiltration analysis indicated that T cell and macrophage infiltration were intricately involved in CASL progression. Intriguingly, single-cell transcriptomic analysis further revealed elevated expression of IL2RG, HOXD10, and TSPAN1 in T cell subsets. Finally, RT-qPCR validated that these genes were expressed at higher levels in CASL tissues than in normal tissues. Moreover, IL2RG expression was strongly associated with clinical parameters. Conclusions: This study identified IL2RG, HOXD10, and TSPAN1 as novel potential molecular markers for CASL, providing valuable biological evidence for the diagnosis and intervention of CASL.