As suggested by an increasing amount of evidence, there is alternative splicing (AS) modification within malignancy, which is related to cancer occurrence and development. AS within acute myeloid leukemia (AML) has not yet been systematically analyzed yet. This study analyzed the transcriptomic profiling and corresponding clinical data from AML cases based on The Cancer Genome Atlas (TCGA). In addition, the prediction model, along with the splicing network, was used to analyze the prognosis for AML patients according to the seven different AS event types. Among the 34,984 AS events across the 8830 genes, 2896 AS events were detected among 1905 genes, showing marked correlation with the overall survival of patients. The risk scoring model based on all AS event types was the most efficient in identifying the prognosis for AML patients. Meanwhile, the area under the curve at 1-, 3-, 5-year were 0.852, 0.935, 0.955, respectively. At the same time, the splicing regulating network, which was constituted by 21 splicing factor genes as well as 32 AS events related to survival, was characterized. In conclusion, our predictive model constructed based on the AS events accurately predicts the survival for AML patients. In addition, the network between AS events and splicing factor is established, which may serve as a potential mechanism.
Keywords: TCGA; acute myeloid leukemia; alternative splicing; prognostic predictor; risk scoring model; splicing correlation network.