Related Experiment Video
Updated: Jul 12, 2026
13:19
Microarray-based Identification of Individual HERV Loci Expression: Application to Biomarker Discovery in Prostate Cancer
Published on: November 2, 2013
Combining pathology artificial intelligence and genomic biomarkers to refine long-term postprostatectomy outcome
Matthew R Cooperberg1,2, Kevin Shee1, Janet E Cowan1
1Department of Urology, UCSF Helen Diller Family Comprehensive Cancer Center, San Francisco, CA, USA.
Journal of the National Cancer Institute
|July 10, 2026
Summary
A multimodal artificial intelligence (AI) model showed prognostic value in prostatectomy specimens, offering complementary information when combined with genomic data for predicting cancer recurrence. Further validation is recommended for diverse clinical settings.
Area of Science:
- Urology
- Artificial Intelligence in Medicine
- Computational Pathology
Background:
- A multimodal AI (MMAI) model was previously validated for guiding treatment intensification in prostate biopsy specimens.
- The MMAI model's utility in prostatectomy patients and its relationship with genomic scores remained underexplored.
Purpose of the Study:
- To evaluate the prognostic performance of an MMAI biopsy model in prostatectomy specimens.
- To assess the MMAI model's association with established clinical (CAPRA) and genomic (CCP) scores.
Main Methods:
- The MMAI biopsy model was applied to a tissue microarray (TMA) cohort of 424 prostatectomy cases.
- MMAI scores were derived from digitized pathology images and clinical variables.
- Associations with biochemical recurrence and metastasis were analyzed using regression, adjusting for CAPRA and CCP scores.
Main Results:
- MMAI scores were successfully generated for 98% of patients.
- MMAI demonstrated significant associations with biochemical recurrence and metastasis in univariable analyses.
- While not independently prognostic after CAPRA adjustment, MMAI remained significant when adjusted for CCP.
- A combined MMAI and CCP model achieved the highest discrimination for metastasis.
Conclusions:
- The MMAI model, despite not being designed for TMAs, showed prognostic capability in prostatectomy specimens.
- The MMAI platform provided complementary information to genomic data, enhancing prognostic accuracy.
- Further refinement and validation of the MMAI platform in varied clinical settings are warranted.
