Sarcomas are malignant soft tissue and bone tumours affecting adults, adolescents and children. They represent a morphologically heterogeneous class of tumours and some entities lack defining histopathological features. Therefore, the diagnosis of sarcomas is burdened with a high inter-observer variability and misclassification rate. Here, we demonstrate classification of soft tissue and bone tumours using a machine learning classifier algorithm based on array-generated DNA methylation data. This sarcoma classifier is trained using a dataset of 1077 methylation profiles from comprehensively pre-characterized cases comprising 62 tumour methylation classes constituting a broad range of soft tissue and bone sarcoma subtypes across the entire age spectrum. The performance is validated in a cohort of 428 sarcomatous tumours, of which 322 cases were classified by the sarcoma classifier. Our results demonstrate the potential of the DNA methylation-based sarcoma classification for research and future diagnostic applications.

PURPOSE Meningiomas are the most frequent primary intracranial tumors. Patient outcome varies widely from benign to highly aggressive, ultimately fatal courses. Reliable identification of risk of progression for individual patients is of pivotal importance. However, only biomarkers for highly aggressive tumors are established ( CDKN2A/B and TERT), whereas no molecularly based stratification exists for the broad spectrum of patients with low- and intermediate-risk meningioma. METHODS DNA methylation data and copy-number information were generated for 3,031 meningiomas (2,868 patients), and mutation data for 858 samples. DNA methylation subgroups, copy-number variations (CNVs), mutations, and WHO grading were analyzed. Prediction power for outcome was assessed in a retrospective cohort of 514 patients, validated on a retrospective cohort of 184, and on a prospective cohort of 287 multicenter cases. RESULTS Both CNV- and methylation family–based subgrouping independently resulted in increased prediction accuracy of risk of recurrence compared with the WHO classification (c-indexes WHO 2016, CNV, and methylation family 0.699, 0.706, and 0.721, respectively). Merging all risk stratification approaches into an integrated molecular-morphologic score resulted in further substantial increase in accuracy (c-index 0.744). This integrated score consistently provided superior accuracy in all three cohorts, significantly outperforming WHO grading (c-index difference P = .005). Besides the overall stratification advantage, the integrated score separates more precisely for risk of progression at the diagnostically challenging interface of WHO grade 1 and grade 2 tumors (hazard ratio 4.34 [2.48-7.57] and 3.34 [1.28-8.72] retrospective and prospective validation cohorts, respectively). CONCLUSION Merging these layers of histologic and molecular data into an integrated, three-tiered score significantly improves the precision in meningioma stratification. Implementation into diagnostic routine informs clinical decision making for patients with meningioma on the basis of robust outcome prediction.