Osteosarcoma (OS) is one of the most prevalent bone malignancies with a poor prognosis. Various types of programmed cell death patterns can influence cancer progression and response to treatment. We aimed to integrate different molecular characteristics of cell death for risk stratification and personalized therapy. First, we obtained transcriptomic, single-cell transcriptomic, and clinical information from the TARGET-OS and GEO databases as well as analyzed genes in fourteen cell death patterns to establish the cell death index (CDI) signature. A nomogram constructed from the CDI calculated from seven genes in combination with metastasis could effectively predict the prognosis of OS patients. Subsequently, the prognostic value and immune characteristics in CDI-defined subgroups were analyzed. A construct nomogram model was also constructed with clinical information. Notably, immunohistochemistry confirmed that the expression of GALNT14, a core gene in CDI model, correlated with poor survival. Deficiency of the highly expressed prognostic gene GALNT14 significantly repressed OS progression and OS cell proliferation by promoting apoptosis. We subsequently demonstrated that Bortezomib, a targeted inhibitor of GALNT14, can be used to enhance chemosensitivity. Finally, it was further elucidated that Bortezomib reduces MT2A glycosylation and improves its stability to promote apoptosis in OS cells by inhibiting GALNT14 expression. In summary, integration of multiple cell death genes may improve the ability to stratify risk in patients with OS, and targeting GALNT14 with Bortezomib improves chemotherapy sensitivity and induces apoptosis.