Introduction:

Pelvic lymph node dissection (PLND) is the gold standard for nodal staging in prostate cancer, despite ongoing debate about its therapeutic benefit. Multiparametric MRI (mpMRI), risk calculators, and increasingly adopted PSMA PET/CT are used to estimate lymph node involvement (LNI) risk and guide surgical planning. Nonetheless, their diagnostic performance in high-risk prostate cancer is not well established. However, their diagnostic accuracy in high-risk prostate cancer is not well established. We evaluated the diagnostic performance of mpMRI, [¹⁸F]-DCFPyL PSMA PET/CT, the Briganti 2023 nomogram, and a combined Briganti+PSMA model for preoperative LNI detection in high-risk prostate cancer patients.

Methods:

We retrospectively analyzed patients with high-risk prostate cancer (clinical stage ≥T3, PSA >20 ng/mL, or biopsy Grade Group ≥4) undergoing radical prostatectomy and PLND. Sensitivity, specificity, positive and negative predictive values, and balanced accuracy  were assessed via confusion matrices and bootstrap confidence intervals

Results:

In 180 high-risk prostate cancer patients, 20% had pathologic lymph node involvement. MRI showed no diagnostic discrimination (AUC 0.50) and 100% NPV but zero sensitivity. [¹⁸F]-DCFPyL PSMA PET/CT demonstrated low sensitivity (26.9%) but high specificity (91.6%) and NPV (88.1%). The Briganti 2023 nomogram had even lower sensitivity (19.6%) with similar specificity (92.1%). A combined Briganti+PSMA model modestly improved AUC to 0.71 but had 0% sensitivity at the 0.5 threshold despite an NPV of 99.4%. Balanced accuracy was poor across all models, reflecting significant false negatives.

Conclusion:

Current imaging modalities and risk models lack sufficient sensitivity to replace PLND for nodal staging in high-risk prostate cancer. Until more reliable non-invasive methods emerge, PLND remains essential to ensure accurate lymph node assessment.

Funding: N/A