Heterogeneity-based stratification identifies CKMT2 as a prognostic marker in osteosarcoma

Background Osteosarcoma (OS) is the most common primary malignant bone tumor. Intratumoral heterogeneity (ITH) contributes to therapy resistance and disease spread. Understanding the molecular structure and cellular groups contributing to ITH, as well as their clinical impacts, remains limited. Methods To build our investigation, we started by integrating single-cell transcriptomic data with multi-cohort transcriptomic datasets sourced from public repositories. Our bioinformatics pipeline was then designed to systematically unpack this information: first, we identified malignant cells by inferring their copy number variation (CNV); next, we assessed the degree of heterogeneity and defined distinct molecular subtypes; and finally, we conducted an in-depth analysis of the functional pathways mediating these heterogeneous characteristics. To sharpen the prognostic model linked to ITH, we employed machine learning to develop and validate a prognostic risk model for ten critical genes and performed both clinical and functional assays. Meanwhile, we validated the pro-tumorigenic role and therapeutic viability of CKMT2 through immunohistochemistry (IHC) and a series of in vitro functional assays. Results Analysis of transcriptomic data from single-cells highlighted ten primary cell types in osteosarcoma. The dominant malignant population as per CNV analysis has been identified as osteoblastic cells (OBs). The subsequent subclassification of OB led to the discovery of five malignant subtypes, resulting in stratification of high and low ITH. Consensus clustering, driven by the differential expression genes discovered in ITH, successfully divided the population into two distinct molecular subtypes. These subtypes exhibited a robust correlation with clinical prognosis and immune infiltration in multiple independent cohorts. Through machine learning modeling, we identified a highly promising therapeutic target, Creatine Kinase Mitochondrial 2 (CKMT2), and performed both clinical and functional assays. IHC validated high CKMT2 expression in tumor tissues corresponding to clinical stage, and in vitro CKMT2 knockdown significantly decreased OS cell proliferation and colony formation. Conclusion This study provides a systematic map of osteosarcoma’s ITH landscape, identifying distinct molecular subtypes with varying clinical outcomes and immune profiles. CKMT2 is established as a key driver of osteosarcoma progression, serving both as a prognostic biomarker and a potential therapeutic target.