An artificial neural network model based on DNA damage response genes to predict outcomes of lower-grade glioma patients. Issue 6 (20th May 2021)

• Record Type: Journal Article
• Authors: Chen, Jian; Qian, Xiaojun; He, Yifu; Han, Xinghua; Pan, Yueyin
• Is Part Of: Briefings in bioinformatics. Volume 22:Issue 6(2021)
• Publication Date: 2021-05-20
• Subjects: artificial neural network -- DNA damage response -- lower-grade glioma -- prognosis; Genetics -- Data processing -- Periodicals; Molecular biology -- Data processing -- Periodicals; Genomes -- Data processing -- Periodicals; 572.80285
• DOI: 10.1093/bib/bbab190
• Languages: English
• ISSNs: 1467-5463

Abstract

Abstract: Although the prognosis of lower-grade glioma (LGG) patients is better than others, outcomes are highly heterogeneous. Isocitrate dehydrogenase ( IDH ) mutation and 1p/19q codeletion status can identify patient subsets with different prognosis. However, in the era of precision medicine, there is still a lack of biomarkers that can accurately predict the individual prognosis of each patient. In this study, we found that most DNA damage response (DDR) genes were aberrantly expressed in LGG patients and were associated with their prognosis. Consequently, we developed an artificial neural network (ANN) model based on DDR genes to predict outcomes of LGG glioma patients. Then, we validated the predictive ability in an independent external dataset and found that the concordance indexes and area under time-dependent receiver operating characteristic curves of the predict index (PI) calculated based on the model were superior to those of the mutation markers. Subgroup analyses demonstrated that the model could accurately identify patients with the same mutation status but different prognosis. Moreover, the model can also identify patients with favorable prognostic mutation status but poor prognosis or vice versa. Finally, we also found that the PI was associated with the mutation status and with the altered immune microenvironment. These results demonstrated that the ANN model can accurately predict outcomes of LGG patients and will contribute to individualized therapies. In addition, a web-based application program for the model was developed.