Digital Pathology-Based Multimodal Artificial Intelligence Scores and Outcomes in a Randomized Phase III Trial in Men...

Area of Science:

• Oncology
• Digital Pathology
• Artificial Intelligence in Medicine

Background:

• The SPARTAN trial established apalutamide's efficacy in nonmetastatic castration-resistant prostate cancer (nmCRPC).
• Digital histopathology offers novel avenues for predicting clinical outcomes.
• Multimodal artificial intelligence (MMAI) algorithms can integrate diverse data for prognostic assessments.

Purpose of the Study:

• To apply a digital histopathology-based MMAI algorithm to estimate clinical outcomes in the SPARTAN trial cohort.
• To evaluate MMAI as a prognostic and predictive biomarker in nmCRPC patients treated with apalutamide.

Main Methods:

• Digitized hematoxylin and eosin-stained slides from 420 nmCRPC patients were analyzed.
• MMAI scores were generated using digital histopathology and baseline clinical data.
• Patients were stratified into MMAI non-high-risk and high-risk groups; survival analyses (MFS, PFS2, OS) were performed.

Main Results:

• MMAI risk score independently predicted shorter metastasis-free survival (MFS), second progression-free survival (PFS2), and overall survival (OS).
• MMAI high-risk patients receiving apalutamide showed significant improvements in MFS, PFS2, and OS compared to non-high-risk patients.
• A significant interaction between MMAI risk and treatment arm indicated greater apalutamide benefit in MMAI high-risk patients.

Conclusions:

• MMAI serves as a prognostic marker in nmCRPC.
• MMAI may function as a predictive biomarker, identifying high-risk patients who derive maximal benefit from apalutamide.
• This study represents the first extension of an MMAI classifier to nmCRPC, highlighting the need for further validation.