Abstract
Retroperitoneal leiomyosarcoma (RLMS) remains a major therapeutic challenge because of frequent postoperative recurrence and the limited benefit of current adjuvant therapies. The marked molecular heterogeneity of RLMS and its incompletely characterized oncogenic drivers have hindered the development of effective targeted therapies. This review proposes an integrative framework that combines transcriptomic subtyping with surgical risk stratification to support artificial intelligence (AI)-guided drug repurposing. The delineation of RLMS subtypes and the identification of potential therapeutic targets through transcriptomic analysis are described, including PDGFRα and VEGFA. The AI-guided screening of approved and investigational drug libraries to identify compounds predicted to reverse subtype-specific molecular programs; preclinical studies highlight candidates such as pazopanib and histone deacetylase (HDAC) inhibitors is discussed. Finally, the outline of a personalized strategy is proposed, in which surgical decision-making integrates anatomic risk with molecular signatures to inform the selection of neoadjuvant or adjuvant therapies. Integrating surgical management, multi-omics, and computational pharmacology helps bridge the gap from bench to bedside and, ultimately, improve outcomes for patients with RLMS. In contrast to prior work that addresses molecular subtyping or surgical management in isolation, this review presents an integrative framework that links surgical risk stratification with transcriptomic profiling to enable AI-guided drug repurposing and provides a roadmap for personalized RLMS therapy.