Multimodal data integration using deep learning predicts overall survival of patients with glioma

Gliomas are highly heterogenous diseases with poor prognosis. Precise survival prediction could benefit further clinical decision-making, clinical trial incursion and health economics. Recent research has emphasized the prognostic value of magnetic resonance imaging, pathological specimens and circulating biomarkers. However, the integrative potential and efficacy of these modalities require to be further validated. After incorporating 218 patients of TCGA glioma datasets of and 54 patients of Huashan cohort with complementary prognostic information, we established Squeeze-and-excitation deep learning feature extractor (SE-DLFE) for T1-contrast enhanced images and histological slides, and explored to screen significant circulating 5-hydroxymethylcytosines (5hmC) profiles for glioma survival by LASSO-Cox regression. Therefore, a prognostication predictive model with high efficiency was obtained through survival support vector machine (SVM) multimodal integration of radiologic imaging, histopathologic imaging features, genome-wide 5hmC in circulating cell-free DNA (cf-DNA) and clinical variables, suggesting a valid strategy (C-index = 0.897; Brier score = 0.118) for improved survival risk stratification of glioma patients.