Head-to-head comparison of MRI-based PI-RR scoring system and 68 Ga-PSMA-PET/CT: enhancing prostate cancer recurrence detection after radical prostatectomy

Abstract Purpose To compare the diagnostic performance of MRI using the PI-RR scoring system and PSMA-PET/CT in detecting prostate cancer (PCa) local recurrence after radical prostatectomy (RP), and to identify clinical and imaging variables independently associated with recurrence. Materials and Methods This retrospective, multicenter study evaluated 119 patients with suspected PCa recurrence who underwent both MRI and PSMA-PET/CT; MRI was performed within a time window ranging from up to 3 months before to a maximum of 1 month after the PET/CT examination; the time range was defined within MDTs. Both radiologists and nuclear medicine physicians were blinded to clinical and opposite imaging data during image interpretation. Exclusion criteria included insufficient follow-up, recent pelvic radiation (within 6 months), and history of hormone therapy. Reference standard was defined using histopathologic findings, follow-up imaging, or clinical response to treatment. Sensitivity, specificity, PPV, NPV, and accuracy were calculated. The Cohen’s kappa coefficient was used to determine inter-reader agreement. Univariable and multivariable logistic regression analyses were performed. Results MRI, considering a PI-RR ≥ 3, detected recurrence in 69 patients versus 48 with PSMA-PET/CT ( p = 0.006). MRI showed significantly higher sensitivity (86%, 95% CI 75–92% vs. 71%, 95% CI 59–80%; p = 0.03) and a trend toward higher specificity (78%, 95% CI 65–87% vs. 52%, 95% CI 39–65%; p = 0.06), PPV (74% vs. 51%), NPV (89% vs. 73%). Overall accuracy was also superior for MRI (81%, 95% CI 74–87% vs. 63%, 95% CI 54–71%). ROC analysis confirmed a higher AUC for MRI (0.87, 95% CI 0.80–0.93) compared with PSMA-PET/CT (0.62, 95% CI 0.52–0.72; ΔAUC = 0.25, p < 0.001). Multivariable analysis showed that only age ( p = 0.002) and PI-RR score ≥ 3 ( p = 0.001) were independently correlated with recurrence. Postoperative risk grouping was correlated only at univariable analysis. Among the 119 included patients, histopathologic confirmation was available in 3% ( n = 4), treatment response (either biochemical or imaging based) in 31% (n = 37), follow-up with imaging in 33% ( n = 39), and clinical-laboratory follow-up in 33% ( n = 39).The median follow-up was 12 months (IQR, 9–18). The effective recurrence rate was 58%. Conclusions MRI demonstrated strong diagnostic performance in detecting prostate cancer local recurrence. These findings suggest that MRI may be particularly useful for rule out local recurrence in low-risk patients, while PET/CT could be more appropriate for rule in, both local and distant in higher-risk patients. However, this proposed risk-adapted imaging strategy should be considered hypothesis-generating and warrants further validation through prespecified risk-stratified analyses. Limitations include the retrospective design, the composite reference standard, and the relatively high recurrence prevalence, which may have influenced predictive values.