Integrating multimodal data to predict the progression of hormone-sensitive prostate cancer.

Identifying the population at risk of rapid progression from hormone-sensitive prostate cancer (HSPC) to lethal castration-resistant prostate cancer (CRPC) is a challenge. This work has highlighted important prognostic insights based on proteomics data, magnetic resonance imaging (MRI) and histopathological specimens. We retrospectively developed a multi-omics-based model based on 77 patients with HSPC. In order to identify the features related to survival time under each mode, we used the Boruta algorithm for feature screening. In order to demonstrate the effectiveness of our selected features, we used six machine learning methods to validate the classification of the selected features for each mode. A total of 63 proteome signatures, 60 HE signatures, 56 T2WI signatures, and 54 ADC signatures were identified as features related to the speed of HSPC progression. Ultimately, 30 multi-omics-based features were determined by the least absolute shrinkage and selection operator (LASSO) method and multivariate cox regression. In order to stratify patients with significant disparities in progress, a nomogram model was developed, of which the C-index was 0.906. Accordingly, the developed model could help identify patients who are at a high risk of rapid CRPC progression, and aid clinicians in guiding personalized clinical management and decision-making.