Purpose: To compare dosimetric outcomes between two real-time prostate seed implantation (PSI) techniques to evaluate the impact of three-dimensional (3D) intraoperative computer planning on target coverage, conformality, and preset urethral and rectal dose constraints.
Methods and materials: One hundred and fourteen patients with clinically localized prostate cancer underwent ultrasound-guided transperineal PSI of the prostate with (125)I sources as monotherapy. From 1999 to 2001, 69 patients were implanted in real-time using a standard look-up nomogram (Group 1: NG-PSI). All patients were implanted with a modified peripheral loading technique in which 75-80% of the calculated total activity was delivered to the gland periphery, with the remaining 20-25% activity placed in the gland interior, to achieve a prescribed dose (PD) of 144 Gy to cover the gland with acceptable homogeneity. No preoperative or intraoperative planning was performed to set dose constraints to the urethra or anterior rectal wall. Dosimetric outcome from this group was compared with 45 patients subsequently implanted after 2001 using an intraoperative 3D computer planning system (Group 2: 3D-PSI). A similar modified peripheral loading technique was used as an option in the planning system. Preoperative dose constraints were placed on the urethra (V150 < 35%), prostate (V100 > 95% of PD; D90: 140-180 Gy), and rectal wall (V110 < 1.5 cc) with real-time dosimetric feedback performed after peripheral loading. Manual dose optimization was performed to determine interior needle position and remaining number and placement of (125)I sources to adhere to urethral and rectal constraints and target coverage goals. Both groups underwent postimplant CT analysis to determine dosimetric outcome with regard toV100(prostate), D90(prostate), V150(urethra), and V110(rectum). Univariate and multivariate analysis was performed to determine variables impacting on dosimetric outcome.
Results: Analysis of preimplant and postimplant variables demonstrated no difference in the median preimplant gland volume (33 cc vs. 35 cc; p = 0.31), median mCi/seed strengths (0.4 vs. 0.45 mCi; p = 0.23), median V100 (94% vs. 94%), or median D90 at postimplant Day 30 (165 Gy vs. 160 Gy; p = 0.26) between Groups 1 and 2. However, for Group 2 (3D-PSI) the median total mCi implanted (26 vs. 33 mCi; p < 0.0001) and the median number of seeds implanted (67 vs. 83; p < 0.0001) were reduced substantially. The percent of patients exceeding a D90 > 180 Gy was reduced from 29% in Group 1 to 16% in Group 2 (p = 0.08). A reduction was observed in the percent of patients receiving a D90 < 140 Gy (14% Group 1 vs. 9% Group 2, p = 0.56). The median V150(urethra) for Group 2 was reduced dramatically with 3D-PSI compared with NG-PSI (63% vs. 17%; p < 0.0001). A V150(urethra) > 30% was observed in 88% in Group 1 compared with 29% in Group 2, p < 0.0001. Similarly, the median V110(rectum) for Group 1 was significantly higher than that in Group 2 (1.93 vs. 0.26 cc; p < 0.0001). The percent of patients with V110(rectum) > 1.5 cc in Group 1 and Group 2 was 57% and 13%, respectively (p < 0.0001).
Conclusions: The adoption of 3D computer intraoperative dose planning and optimization for prostate seed implantation resulted in dramatic reductions in urethral and rectal wall doses, while consistently producing excellent target coverage with reduced dose variability above 180 Gy and below 140 Gy, compared with the use of a standard look-up nomogram. Additionally, the reduction in total mCi and number of seeds needed to achieve improved conformality was substantial and may have implications for cost savings.