Melanoma, a malignant tumor of the skin, presents challenges in its treatment process involving modalities such as surgery, chemotherapy, and targeted therapy. However, there is a need for an ideal model to assess prognosis and drug sensitivity. Programmed cell death (PCD) modes play a crucial role in tumor progression and has the potential to serve as prognostic and drug sensitivity indicators for melanoma.
We analyzed 13 PCD modes including apoptosis, necroptosis, ferroptosis, pyroptosis, netotic cell death, entotic cell death, lysosome-dependent cell death, parthanatos, autophagy-dependent cell death, oxeiptosis, disulfidptosis, and alkaliptosis. These modes were used to construct a model that incorporated genes related to these 13 PCD modes to establish a cell death index (CDI) to conduct prognosis analysis. Transcriptomic, genomic, and clinical data were collected from cohorts including TCGA-SKCM, GSE19234, and GSE65904 to validate this model.
A CDI consisting of ten gene signatures was established using machine learning algorithms and divided into two groups based on CDI values. The high CDI group exhibited relatively lower numbers of immune-infiltrating cells and showed resistance to commonly used drugs such as docetaxel and axitinib. Our validation results demonstrated good discrimination in PCA analysis between CDI groups, and melanoma patients with higher CDI values had worse postoperative prognoses (all p < 0.01).
The CDI model, incorporating multiple PCD modes, accurately predicts the clinical prognosis and drug sensitivity of melanoma patients.