Computed Tomography-Based Radiomics Prediction of Biochemical Failure and Distant Metastasis in Patients With High-...

Area of Science:

• Oncology
• Radiology
• Medical Imaging

Background:

• Prostate cancer (PCa) management relies on predicting treatment outcomes.
• Radiomics analysis of planning computed tomography (pCT) images offers potential for improved prognostication.

Purpose of the Study:

• To evaluate the effectiveness of radiomics features from pCT scans in predicting biochemical failure (BF) and distant metastasis (DM) in localized prostate cancer (PCa).

Main Methods:

• A retrospective study of 608 high- and very high-risk localized PCa patients treated with intensity modulated radiation therapy.
• Clinical variables and 1316 radiomics features were extracted from pCT images.
• Random survival forest models (clinical, radiomics, hybrid) were developed and validated to predict BF and DM, considering competing risks.

Main Results:

• Over 8.2 years median follow-up, 29.3% experienced BF and 8.6% DM.
• The hybrid model (clinical + radiomics) showed improved BF prediction (C-index=0.687) compared to clinical (0.637) or radiomics (0.640) alone (P < .05).
• No significant improvement in DM prediction was observed across models.

Conclusions:

• Integrating clinical and radiomics data from pCT enhances prediction of biochemical failure in localized prostate cancer.
• Radiomics analysis did not significantly improve distant metastasis prediction in this cohort.