Whole-body MRI-based multivariate prediction model in the assessment of bone metastasis in prostate cancer.

ABSTRACT

PURPOSE: A whole-body MRI (WB-MRI) including T1, short time inversion recovery (STIR), diffusion-weighted imaging (high b value) was applied in our center for the detection of bone metastasis in prostate cancer (PCa) patients. We intended to assess the diagnostic performance of this examination. METHODS: 547 cases of PCa patients with higher risk of metastasis were referred to bone scintigraphy with SPECT/CT (BS + SPECT/CT) and whole-body MRI in Shanghai Changhai Hospital. Best valuable comparator (BVC) was applied for the final diagnosis of metastasis. A panel of radiologists interpreted the results. Decision curve analysis (DCA) and receiver operating characteristic curve (ROC) analysis were applied. RESULTS: Bone metastasis was diagnosed in 110 cases, and others were non-metastatic by BVC. The area under the receiver operating characteristic curve (AUC) was higher in WB-MRI (0.778) than BS + SPECT/CT (0.634, p < 0.001). A WB-MRI-based prediction model was established with AUC of 0.877. Internal validation showed that the predictive model was well-calibrated. The DCA demonstrated that the model had higher net benefit than the BS + SPECT/CT-based model. CONCLUSION: WB-MRI is more effective in identifying metastasis in PCa patients than BS + SPECT/CT. The prediction model combined WB-MRI with clinical parameters may be a promising approach to the assessment of metastasis.