The article describes the development and validation of a new framework, utilizing genome-wide DNA methylation profiling and a neural network called MARLIN, for the rapid and precise classification of acute leukemia. Researchers first created a comprehensive reference cohort of over 2,500 samples to define 38 distinct methylation classes across various acute leukemia lineages, demonstrating that this epigenetic classification complements and refines conventional diagnostic methods. MARLIN was trained to classify acute leukemia subtypes from sparse nanopore sequencing data, achieving high concordance with clinical diagnoses in validation cohorts and proving capable of providing accurate predictions, often within two hours of sample receipt, in real-time prospective cases. This framework offers a significant advancement for acute leukemia diagnostics, potentially accelerating treatment planning by providing timely molecular information.

References:

• Steinicke T L, Benfatto S, Guerra M R C, et al. Rapid Epigenomic Classification of Acute Leukemia[J]. Blood, 2024, 144: 273.