Leveraging multiple cell-death patterns based on machine learning to decipher the prognosis, immune, and immune therapeutic response of soft tissue sarcoma.

Soft tissue sarcomas (STS) imposes a substantial healthcare burden on society. The progression of these tumors is significantly influenced by diverse modes of programmed cell death (PCD), which can serve as valuable indicators for assessing prognosis and immune therapeutic response in STS. Nonetheless, the precise role of multiple cell death patterns in STS is yet to be clarified. We employed 96 machine-learning algorithm combination frameworks to identify novel cell death-related signatures (CDSigs) with the highest mean c-index, indicating their excellence. The independence test and comparison with previously published models further confirmed the stability and quality of these signatures for survival prediction in STS. The nomogram, comprising the cell death score (CDS) and clinical features, exhibited excellent predictive performance. Additionally, the CDSigs revealed associations with immune checkpoint genes and the immune microenvironment in STS. Furthermore, the results demonstrated that patients with lower CDS had the potential for greater benefit from immune therapeutic responses compared to those with higher CDS. Moreover, STS patients with low-risk scores exhibited heightened sensitivity to doxorubicin, axitinib, cisplatin, and camptothecin. Finally, the RT-qPCR results underscored significant differences in expression levels of several CDSigs genes between STS and normal cells. Overall, we comprehensively analyzed the multiple PCD in STS and established a novel CDSig for STS patients. This novel CDSig holds great promise in deciphering the prognosis, immune, and immune therapeutic response of STS.