Conclusion: We have illustrated a simple technique which corrects for the daily prostate motion, allowing for extremely preciseprostate cancer treatment. This technique has significant implications in dose escalation and for decreasing rectal complicationsin the treatment of prostate cancer.

2177 Adjuvant And Salvage Radiotherapy after Radical Prostatectomy for Prostate Cancer

N.A. Taylor, J.F. Kelly, D.A. Kuban, R.J. Babaian, L.L. Pisters, A. Pollack

Radiation Oncology, M.D. Anderson Cancer Center, Houston, TX

Purpose: The optimal role of radiotherapy (RT) to the prostate bed after radical prostatectomy (PR) is the subject of muchdebate. In this study, the results of adjuvant and salvage RT were examined in terms of prognostic factors and relative efficacy.

Materials and Methods: From 1987 to 1998, 146 lymph node negative patients received RT to the prostate bed following PR.Seventy-five patients had an undetectable PSA and were treated adjuvantly (ART) after PR for adverse pathologic features only,to a median dose of 60 Gy (range 51 to 70 Gy). A positive margin was identified in 73 of 75 (96%); 2 of the 3 margin negativepatients had seminal vesicle involvement (SVI). Altogether, there were 13 ART patients with SVI. Salvage radiotherapy (SRT)was administered for either persistently detectable PSA post-PR (PD-PSA; n�25) or for delayed rise in PSA post-PR (DR-PSA;n�46) to a median dose of 70 Gy (range 60 to 78 Gy). A 4-field box was used in all but 1 patient, who received 78 Gy usinga conformal boost for a partial-PR. Adjuvant androgen ablation (AA) was given to 37 patients; 2 patients (2.7%) who receivedART and 35 patients (49%) who received SRT. The median duration of AA was 24 months. Biopsy proven local recurrencewas documented in 14 SRT patients treated for DR-PSA. After RT, an undetectable PSA was necessary to be consideredbiochemically free of disease (bNED). Pre-PR and Pre-RT PSA values, clinical stage, use of adjuvant AA, PR Gleason score,PR SVI, PR extracapsular extension (ECE), margin positivity (Marg�), time to RT post-PR (RT-time), RT dose, and indicationfor radiotherapy were analyzed for impact on bNED, as from the date of radiotherapy by the Kaplan-Meier method. Medianfollow-up was 53 months for all patients, 68 months for the ART patients and 39 months for the SRT patients.

Results: For the ART group, 8 patients subsequently developed a rising PSA. The 5 yr bNED rate was 88%. SVI stronglyinfluenced outcome, with a 5 yr bNED of 94% for those without SVI and 65% for those with SVI (p�0.0002). Other significantadverse factors in univariate analysis were a Pre-PR PSA�10 ng/ml (5 yr bNED 76% vs. 100%, p�0.01) and a PR-Gleasonscore�7 (5 yr bNED 73% vs. 96%, p�0.006); the presence of ECE was borderline (p�0.09). No association was observedbetween the clinical stage, AA, RT-time, or RT dose and the bNED rate. The only factor significant in multivariate analysiswas SVI for the ART cohort.

For the SRT group, 20 patients developed failure to this salvage treatment. The 5 yr bNED for the SRT group (66%) wassignificantly worse than that of ART group (p�0.0008). Of patients treated with SRT, the 5 yr bNED for patients with aDR-PSA was significantly better than for those with a PD-PSA; 78% vs 43% (p�0.01). SVI had no effect on bNED in SRTpatients overall (�60% at 5 yr), but, AA was associated with an improvement in bNED from 54% to 81% (p�0.03). However,since 16 of the 37 AA treated patients were still on AA at last follow-up, these results are difficult to interpret. None of the otherfactors tested were significant in univariate analysis in SRT patients. In multivariate analysis, both AA and indication ofradiotherapy (DR-PSA vs. PD-PSA) were significant.

Conclusion: In the absence of SVI, ART resulted in an actuarial failure rate of only 6%. Moreover, there were no failuresbeyond 5 years in these ART patients and there were 13 patients at risk at 96 months. Although SRT for DR-PSA was alsoassociated with a relatively low failure rate at 5 years (22%), half were treated with extended AA and many were still beingtreated with AA at last contact. ART is highly effective at reducing biochemical failure, as this pathologically high risk groupof PR patients would be expected to have greater than twice the failure rate observed.

2178 Online Image Guided External Beam Radiation Therapy for the Treatment Prostate Cancer UsingElectronic Portal Imaging: Prostate Motion and Setup Evaluation

E. Vigneault1, L. Beaulieu1, S. Aubin1, L.M. Girouard1, J. Pouliot2, S. Pouliot1, J. Laverdiere1

1Radiation Oncology, Centre Hospitalier Universitaire de Quebec, Quebec, PQ, Canada, 2Radiation Oncology, UC SanFrancisco, San Francisco, CA

Purpose: To demonstrate the feasibility of online target setup verification, automatic deviation measurements and positionningcorrections for patients treated with 3D conformal radiotherapy (3D-CRT) and IMRT of the prostate gland.

Materials and Methods: Participating prostate cancer patients had three gold markers implanted before the treatment. Thepatients were CT planned in dorsal position with empty rectum and full bladder and treated with a 3D-CRT 7 field technique.Electronic Portal Images (EPI) was used in conjunction with an automatic maker detection algorithm to track the prostatemotion. EPIs were taken for each patients for all fractions and for 6 of the 7 treatment fields. The ability of the automatic markerdetection algorithm (and EPID) to adequately and consistently detect the markers has been tested. The results are used todetermine the random and systematic prostate motion and as a feedback loop to strenghten the algorithm. CT-sim exams arealso performed to provide snapshots in time of the markers relative to the CTV and anatomical structures.

Results: This abstract report the results of the first 6 patients enrolled in this study giving 308 inter-fraction prostate motionmeasurements. For each patients the RPO and RL EPIs were recorded and reviewed. On each EPI the markers were detectedby an automated algorithm and displacements were compared to the reference EPI. The population average seed displacementsobserved were 0.2 mm, 0.9 mm, 0.3 mm in the lateral, cephalo-caudal and antero-posteriror direction respectively. Thedistributions of these deviations in the same three directions are fairly large with standard deviations of 1.5 mm in the lateraldirection, 3.2 mm in the cephalo-caudal direction and 2.4 mm in the antero-posteriror direction. Maximum displacements of 4.4mm, 15.4 mm and 9.7 mm were observed in the lateral, cephalo-caudal and antero-posteriror direction respectively. Thedisplacement were mainly in the cephalo-caudal and antero-posteriror direction. Following a statistical analysis, the random andsystematic errors were calculated. The standard deviation of the random errors are 1.5 mm, 2.2 mm and 2.2 mm in the

317Proceedings of the 43rd Annual ASTRO Meeting