Transcript isoform diversity defines molecular subtypes and prognosis in acute myeloid leukemia through long-read sequencing.

Acute myeloid leukemia (AML) is a genetically complex and clinically heterogeneous hematopoietic malignancy. This study employs long-read transcriptome analysis using oxford nanopore technologies sequencing on 60 primary AML bone marrow samples. This approach delivers comprehensive isoform-level resolution of splicing abnormalities and overcomes limitations of short-read sequencing. We detect extensive AML-specific splicing anomalies and identify 119,278 previously unannotated transcript isoforms. Of these, 80,294 (67.31%) contain complete open reading frames, with 9,812 (12.22%) validated using liquid chromatography-tandem mass spectrometry. Quantitative analysis in 175 RNA sequencing samples enables non-negative matrix factorization clustering, defining distinct molecular subtypes. These isoform-defined subtypes exhibit strong correlations with patient prognosis, indicating their potential as biomarkers for clinical classification. The findings highlight alternative splicing as a major contributor to AML molecular heterogeneity and provide a valuable foundation for advancing precision medicine and developing innovative therapeutic strategies targeting splicing abnormalities in AML.