Abstract
Soft tissue and bone sarcomas are diagnostically challenging due to their morphological overlap. Genome-wide DNA methylation profiling has emerged as an ancillary tool for classification, providing both a supervised diagnostic call and inferred copy-number alterations (CNAs) from a single assay. This systematic review synthesizes the evidence on its diagnostic performance, clinical utility, and integration into practice. Following PRISMA 2020 guidelines, a systematic literature search of PubMed and Google Scholar was conducted for studies published between 2010 and 2025. Eligible studies reported on the use of methylation-based classifiers for sarcoma diagnosis against an integrated histomolecular reference standard. Nine studies were included. In validation cohorts, methylation classifiers provided high-confidence predictions (calibrated score ≥0.9) in 55-75% of cases. When a high-confidence prediction was made, concordance with the reference diagnosis was high (83-88%). Discrepant high-confidence calls prompted diagnostic re-evaluation, leading to revised diagnoses in a significant proportion of cases. Methylation-derived CNAs provided critical diagnostic information for key differentials, such as MDM2 amplification in liposarcoma and 22q loss in SMARCB1-deficient tumors. Performance was robust in fusion-driven sarcomas but lower in selected benign entities. Technical success was influenced by tumor purity and specimen type (FFPE vs. frozen). Evidence for nanopore-based methods remains preliminary. DNA methylation-based classification is a clinically validated tool that augments the diagnosis of sarcomas, offering high diagnostic concordance and valuable copy-number insights. Its integration requires careful interpretation within a multidisciplinary framework. Further prospective validation and standardization of emerging sequencing technologies are warranted.