High-grade glioma (HGG) is an aggressive brain tumor. Sex is an important factor that differentially impacts survival outcomes in HGG. We employed an end-to-end deep-learning approach on Hematoxylin and eosin (H&E) scans to (1) identify explainable, sex-specific histopathological attributes of the tumor microenvironment (TME) that may be associated with patient-outcomes, and (2) create sex-specific risk profiles to prognosticate overall survival. Surgically resected H&E-stained tissue slides were analyzed in a two-stage approach using ResNet18 deep-learning models, first, to segment the viable tumor regions, and second, to build sex-specific prognostic models for prediction of overall survival. Our mResNet-Cox model yielded C-index (0.696, 0.736, 0.731, 0.729) for the female cohort and C-index (0.729, 0.738, 0.724, 0.696) for the male cohort across training and three independent validation cohorts, respectively. End-to-end deep-learning approaches using routine H&E-stained slides, trained separately on male and female HGG patients, may allow for identifying sex-specific histopathological attributes of the TME associated with survival and, ultimately, build patient-centric prognostic risk-assessment models.

Data The patches and the associated tumor segmentation labels (expert-vetted) from our analysis are available in Patches.pytable file.CodesThe codes for training tumor segmentation models and conducting survival analysis are available in the following files ResNet-train: Code to train Resnet18 model for Tumor Segmentation Tumor_Segmentation: Code to segment tumor regions from WSI using ResNet18 model ResNet_Cox_train: Code to train ResNet-Cox model in 5 folds cross-validation setting Evaluate_ResNetCox: Code to evaluate ResNet-Cox model