磁振造影與經直腸超音波融合切片於攝護腺癌之診斷效力：病灶層級分析、攝護腺特異性抗原密度與機構學習曲線之探討

許庭瑞¹、黃銘德²、林偉銘²、楊浩誌¹、黃偉柏¹、陳鐸文¹、黃國財¹、黃雲慶¹、何東儒¹、林健煇¹、陳志碩¹、林威宇¹

嘉義長庚紀念醫院 ¹泌尿科，²放射診斷科

Diagnostic Performance of MRI-TRUS Fusion Biopsy for Prostate Cancer: Insights from Per-Lesion Analysis, PSA Density, and Institutional Learning Curve

Ting-Jui Hsu¹, Ming-De Huang², Wei-Ming Lin², Hao-Chih Yang¹, Wei-Po Huang¹, Duo-Wun Chen¹, Kuo-Tsai Huang¹, Yun-Ching Huang¹, Dong-Ru Ho¹, Jian-Hui Lin¹, Chih-Shou Chen¹, Wei-Yu Lin¹

Department of Urology and Diagnostic Radiology, Chang Gung Memorial Hospital, Chiayi, Taiwan

Purpose: To evaluate the diagnostic yield of MRI-TRUS fusion biopsy for clinically significant prostate cancer (csPCa), investigate the statistical discrepancy between per-patient and per-lesion analyses, and assess the predictive value of prostate-specific antigen density (PSAD) alongside our institutional learning curve.

Materials and Methods: We retrospectively reviewed 101 patients who underwent MRI-TRUS fusion biopsy at our institution between 2023 and 2026. Clinically significant prostate cancer (csPCa) was defined as a Gleason score >= 7 (Grade Group >= 2). Positive predictive values (PPVs) were calculated using a comparative approach: per-patient analysis (based on the highest PI-RADS score) and strict per-lesion analysis. Furthermore, the institutional learning curve was evaluated by comparing the diagnostic outcomes of the initial 30 cases with the most recent 30 cases.

Results: The overall cancer detection rate (CDR) was 55.4% (56/101), with csPCa accounting for 71.4% (40/56) of the positive cohort. The mean PSAD was significantly higher in positive cases (0.410 vs. 0.234 ng/mL²). The comparative analysis demonstrated a substantial discrepancy. The per-patient PPVs for PI-RADS 3, 4, and 5 were 31.0%, 59.3%, and 88.2%, respectively. In contrast, the per-lesion PPVs dropped notably to 19.8% (16/81) for PI-RADS 3, 47.2% (42/89) for PI-RADS 4, and 85.7% (18/21) for PI-RADS 5. Regarding the learning curve, comparing the initial 30 to the recent 30 cases, the overall CDR improved from 40.0% to 70.0%, and the csPCa detection rate significantly increased from 20.0% to 56.7%. This improvement was achieved despite a lower mean PSAD in the recent cohort (0.265 vs. 0.362 ng/mL²).

Conclusion: Conventional per-patient analysis overestimates the true PPV of PI-RADS scores. Our per-lesion data revealed that over 50% of PI-RADS 4 lesions are histologically benign, highlighting the crucial role of PSAD in guiding biopsy decisions for intermediate-risk lesions. Moreover, a steep learning curve was observed, indicating that accumulated surgical experience is essential for optimizing the detection of csPCa.