Diagnostic value of PRIMARY score and a combined model with SUVmax in 18F-PSMA-1007 PET/CT for prostate cancer

To investigate the diagnostic value of the PRIMARY score in 18F-PSMA-1007 PET/CT for prostate cancer (PCa), to compare its diagnostic efficacy with maximum standardized uptake value (SUVmax), and explore the value of their combined model.

A total of 180 patients with suspected prostate cancer were initially screened. After excluding 91 patients (70 with prior treatment, 16 without pathological confirmation, 3 with unclear pathological results, and 2 with poor image quality), 89 patients were finally enrolled. We retrospectively reviewed 89 patients with suspected PCa who underwent 18F-PSMA-1007 PET/CT from November 2023 to September 2025, including 61 with PCa and 28 with benign lesions. All patients underwent standard preoperative evaluation including PSA, digital rectal examination (DRE), ultrasound, or multiparametric MRI (mpMRI) before imaging. Prostate biopsy was performed within 2 weeks before or after PET/CT according to clinical indications. The SUVmax of the representative lesion was measured and the PRIMARY score was recorded. Intergroup differences were analyzed, optimal cutoff values were determined by ROC curve, and AUCs were compared by the DeLong test. Risk stratification was performed for clinically significant prostate cancer (csPCa), with focus on the combined diagnostic strategy for "gray-zone" lesions with PRIMARY score 4. Interobserver agreement was evaluated using the Kappa coefficient.

Age was comparable between the two groups (P > 0.05), while SUVmax and PRIMARY score were significantly different (both P < 0.05). The PRIMARY score showed excellent interobserver agreement (Kappa = 0.903). For PCa diagnosis, the AUCs of SUVmax, PRIMARY score, and the combined index PRE were 0.921, 0.960, and 0.963, and PRE was significantly superior to SUVmax alone, but not significantly different from PRIMARY score. For csPCa, the positive predictive value was 97.6% for PRIMARY score 5, 68.2% for score 4, and increased to 85.7% by applying the combined criterion (PRIMARY = 4 and SUVmax ≥ 7.5). Notably, 3 of 8 patients in the PRIMARY = 4 and SUVmax < 7.5 subgroup still had csPCa, indicating a trade-off between PPV and missed csPCa. PRIMARY score ≤ 3 showed a low csPCa prevalence of 7.7%.

In 18F-PSMA-1007 PET/CT, the PRIMARY score demonstrates comparable diagnostic efficacy to SUVmax with excellent reproducibility. The combined PRE model was superior to SUVmax alone, but not significantly different from PRIMARY score in this cohort, optimizing the stratification of PRIMARY score 4 "gray-zone" lesions. This approach effectively reduces false positives and provides a robust quantitative reference for biopsy decisions and individualized patient management.