Diagnostic accuracy of MRI radiomics in predicting lymph node metastasis in prostate cancer: A systematic review

Purpose

Prostate cancer (PCa) is frequently associated with pelvic lymph node metastasis (PLNM), which may be missed by conventional imaging, particularly in micrometastatic disease. MRI-based radiomics offers potential to improve detection. This review evaluates recent advancements and diagnostic accuracy of MRI radiomics for predicting PLNM in PCa patients.

Methods

PubMed, Embase, and Web of Science were systematically searched through January 1, 2025, using terms like “prostate cancer,” “radiomics,” and “pelvic lymph node metastasis.” Eligible studies were assessed using the Radiomics Quality Score (RQS). Study characteristics and performance metrics were narratively synthesized. Pooled area under the receiver operating characteristic curve (AUC) was calculated for PLNM prediction in studies using prostate as regions of interest (ROI), reported with 95 % confidence intervals (CI); p-value < 0.05 was considered significant.

Results

Nine studies (2021–2024) involving 2344 PCa patients were included. Radiomics models using prostate as ROI achieved a pooled AUC of 0.78 (95 %CI: 0.72–0.84) with mild heterogeneity (I² = 19.81 %, p < 0.38). Models with lymph nodes as ROI showed AUCs of 0.93–0.95. Integrating imaging reports and clinical data often improved diagnostic accuracy. Radiomics outperformed clinical nomograms in five studies, although the difference was insignificant in one study (p > 0.05). Median RQS was 16/36; studies lacked prospective design and cost-effectiveness analysis.

Conclusion

MRI radiomics predicts PLNM with moderate accuracy, particularly when using pelvic lymph nodes as ROI. Standardized protocols, feature extraction, and clinical data integration are crucial for consistency. Prospective studies with larger cohorts are needed to validate these findings.