Can We Predict Skeletal Lesion on Bone Scan Based on Quantitative PSMA PET/CT Features?

Area of Science:

• Oncology
• Radiology
• Nuclear Medicine

Background:

• Prostate cancer (PCa) restaging with PSMA-PET/CT is increasingly used.
• PSMA-PET/CT detects more metastatic lesions than conventional imaging (CI).
• Therapeutic pathways are based on CI, creating a discrepancy with PSMA-PET/CT findings.

Purpose of the Study:

• To determine if PSMA-PET/CT and clinical data can predict lesion visibility on bone scans (BS).
• To bridge the gap between conventional and advanced imaging in prostate cancer management.

Main Methods:

• Retrospective analysis of 43 PCa patients with concurrent BS and PSMA-PET/CT.
• Collected PSMA-PET/CT parameters (SUVmax, SUVmean, PSMAvol, etc.) and CT features (density, diameter, erosion).
• Utilized a random forest model with 10-fold cross-validation to predict BS visibility.

Main Results:

• 58.1% of 222 bone lesions identified on PSMA-PET/CT were visible on BS.
• All PSMA-PET/CT parameters significantly correlated with BS visibility (p < 0.001).
• The random forest model achieved 77.6% accuracy in predicting BS visibility.

Conclusions:

• PSMA-PET/CT parameters can predict bone scan visibility of prostate cancer lesions.
• This predictive capability may enhance comparability between imaging modalities.
• Findings could support improved therapy selection in prostate cancer patients.