Digital Pathology-based Artificial Intelligence Biomarker Validation in Metastatic Prostate Cancer

Area of Science:

• Oncology
• Artificial Intelligence in Medicine
• Biomarker Discovery

Background:

• Personalized medicine for metastatic hormone-sensitive prostate cancer (mHSPC) requires prognostic and predictive biomarkers.
• Treatment options and clinical subgroups in mHSPC necessitate tailored therapeutic strategies.

Purpose of the Study:

• To evaluate a multimodal artificial intelligence (MMAI) biomarker for its prognostic capability in mHSPC.
• To assess the association of MMAI scores with overall survival (OS), clinical progression (CP), and castration-resistant prostate cancer (CRPC).

Main Methods:

• Utilized data from the phase 3 CHAARTED trial, including digital histopathology images and clinical variables from 456 patients with mHSPC.
• Generated MMAI scores and assessed their association with OS, CP, and CRPC using Cox proportional-hazards and Fine-Gray models.

Main Results:

• The MMAI biomarker demonstrated significant prognostic ability for OS, CP, and CRPC.
• Patients classified as MMAI-high risk had significantly worse 5-year OS (39%) compared to MMAI-intermediate (58%) and MMAI-low (83%) groups.
• The MMAI score remained prognostic across different clinical subgroups and after adjusting for treatment and clinical variables.

Conclusions:

• The MMAI biomarker is a valuable prognostic tool for patients with mHSPC, irrespective of clinical subgroup or treatment.
• Further research into MMAI biomarkers is warranted for advanced prostate cancer management.
• AI-driven tools show potential in enhancing clinical decision-making for prostate cancer patients.