Clinical parameters and nomograms for predicting lymph node metastasis detected with 68 Ga‐PSMA‐PET/CT in prostate cancer patients candidate to definitive radiotherapy

Background Defining the extent of disease spread with imaging modalities is crucial for therapeutic decision-making and definition of treatment. This study aimed to investigate whether clinical parameters and nomograms predict prostate-specific membrane antigen (PSMA)-positive lymph nodes in treatment-naive nonmetastatic prostate cancer (PC) patients. Materials and methods The clinical data of 443 PC patients (83.3% high-risk and 16.7% intermediate-risk) were retrospectively analyzed. Receiver operating characteristic (ROC) curves with areas under the curve (AUC) were generated to evaluate the accuracy of clinical parameters (prostate-specific antigen [PSA], T stage, Gleason score [GS], International Society of Urological Pathology [ISUP] grade) and nomograms (Roach formula [RF], Yale formula [YF], and a new formula [NF]) in predicting lymph node metastasis. The AUCs of the various parameters and clinical nomograms were compared using ROC and precision-recall (PR) curves. Results A total of 288 lymph node metastases were identified in 121 patients (27.3%) using 68 Ga-PSMA-11-positron emission tomography (PET)/computed tomography (CT). Most PSMA-avid lymph node metastases occurred in external or internal iliac lymph nodes (142; 49.3%). Clinical T stage, PSA, GS, and ISUP grade were significantly associated with PSMA-positive lymph nodes according to univariate logistic regression analysis. The PSMA-positive lymph nodes were more frequently detected in patients with PSA >20 ng/ml, GS ≥7 or high risk disease compared to their counterparts. The clinical T stage, serum PSA level, GS, and ISUP grade showed similar accuracy in predicting PSMA-positive metastasis, with AUC values ranging from 0.675 to 0.704. The median risks for PSMA-positive lymph nodes according to the RF, YF, and NF were 31.3% (range: 12.3%-100%), 22.3% (range: 4.7%-100%), and 40.5% (range: 12.3%-100%), respectively. The AUC values generated from ROC and PR curve analyses were similar for all clinical nomograms, although the RF and YF had higher accuracy compared to NF. Conclusion The clinical T stage, PSA, GS, and ISUP grade are independent predictors of PSMA-positive lymph nodes. The RF and YF can be used to identify patients who can benefit from 68 Ga-PSMA-11 PET/CT for the detection of lymph node metastasis. Together with nomograms, 68 Ga-PSMA-11 PET/CT images help to localize PSMA-positive lymph node metastases and can thus assist in surgery and radiotherapy planning.