Precision Selection for Targeted Radionuclide Therapy in Prostate Cancer: Biparametric MRI Radiomics Combined with Serum Bone Turnover Biomarkers for Predicting Postoperative Bone Metastasis.
Abstract
Objective
Particularly for patients with bone-dominant illness, targeted radionuclide treatments are becoming a viable treatment option for metastatic prostate cancer (PCa). Precision treatment selection and improved therapeutic outcomes may result from early detection of patients who are at high risk of developing bone metastases. To identify early-stage PCa patients at increased risk of postoperative bone metastasis who might benefit from targeted radionuclide therapy, this study sought to develop and validate a multimodal predictive model that integrated biparametric magnetic resonance imaging (bpMRI) radiomics features and serum bone turnover biomarkers.
Methods
A total of 143 patients with clinically localized PCa (cT1-2N0M0) receiving laparoscopic radical prostatectomy were included in this prospective single-center cohort analysis. Bone metabolism indicators, such as osteocalcin N-terminal mid-fragment and alkaline phosphatase, were measured in serum prior to surgery. bpMRI was used to extract radiomics variables that reflected the tumor microenvironment and perfusion characteristics (apparent diffusion coefficient mean and Ktrans mean). Serum and imaging biomarkers were integrated using a multivariate logistic regression model. Receiver operating characteristic (ROC) analysis, calibration curves, decision curve analysis, and Kaplan-Meier survival analysis for bone metastasis-free survival were used to assess the model's performance.
Results
With an area under the ROC curve of 0.984, the integrated multimodal model outperformed individual biomarkers and imaging features by a significant margin (all p < 0.001). In decision curve analysis, the model demonstrated significant clinical net benefit and strong calibration (Hosmer-Lemeshow test p = 0.830). Patients were categorized into high-risk (n = 53) and low-risk (n = 90) groups according to risk stratification based on the model, with significantly different bone metastasis-free survival results (log-rank p < 0.001).
Conclusions
A reliable method for anticipating postoperative bone metastases in early-stage PCa is a multimodal framework that combines bpMRI radiomics with blood bone turnover indicators. This strategy may facilitate early treatment intervention in individuals with increased metastatic risk and support precision patient selection for targeted radionuclide therapy.