Skip to content
Open access

Multimodal deep learning model for predicting homologous recombination deficiency in prostate cancer: A multi-cohort study

2026 · The Innovation Oncology · 0 citations · 25 references

Abstract

Accurate prediction of homologous recombination deficiency (HRD) status is crucial for effective management of prostate cancer (PCa). However, genetic testing is expensive and inaccessible. We propose a multimodal deep learning approach that integrates clinical information, Hematoxylin & Eosin (H&E)-stained whole-slide images (WSIs), and multiparametric magnetic resonance imaging (MRI) to predict the HRD status of patients with PCa. Patients from The Cancer Genome Atlas (TCGA; n=486) and three Chinese hospitals (n=210) were used to develop and evaluate the prediction models. The Pathology Signature, Radiology Signature, and Patho-Radiology Signature showed predictive performance for HRD status in external validation or internal held-out evaluation (area under the receiver operating characteristic curve [AUC]=0.815, 0.833, and 0.900, respectively). Notably, four interpretable pathological features were identified in the Pathology Signature. These signatures may serve as a prescreening tool to select patients for confirmatory genetic testing. This multimodal framework may be adaptable to the prediction of molecular alterations in other malignancies, subject to independent validation. Because radiology and multimodal models were evaluated in small modality-complete subsets, these estimates should be interpreted as exploratory trends.

Read PDF

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.