4607 Background: The external validity of randomized trials (generalizability) has been called into question following the finding that patients treated on clinical trials vary substantially from patients treated in the usual care setting. The purpose of this secondary analysis is to examine whether results of prostate randomized trials within the RTOG differ according to enrollment setting. Methods: 4,154 patients accrued into one of 4 RTOG trials between 1987-1999 were included in this analysis. There were 224 North American institutions enrolling in one or more trials. Patient attributes were compared by enrollment settings, controlling for trial. Survival regression models stratified by trial and including patient attributes were used to evaluate survival differences by enrollment setting. Results: Academic institutions enrolled 36% of patients; community centers enrolled 49%; CCOPs enrolled 14%, and VHAs enrolled 2%. With respect to patient attributes, men from Academic and VHA centers tended to have better performance status, black men were less frequently enrolled from non-CCOP community institutions, and Gleason scores tended to be higher in men from CCOPs. Age at diagnosis did not differ materially by enrollment setting. Adjusted for patient attributes, survival by enrollment setting did not differ significantly (Table). Conclusions: Attributes of prostate cancer patients enrolled in these RTOG trials Phase III differed modestly according to enrollment site type. Accounting for these differences, overall survival did not depend on the type of institution from which patients were enrolled. These findings support the external validity of RTOG prostate trials.This project was supported by RTOG grant U10 CA21661, and CCOP grant U10 CA37422 from the National Cancer Institute (NCI). Contrast* Hazard ratio 95% confidence interval CCOP vs. academic 1.00 0.87–1.14 Other community vs. academic 0.91 0.83–1.00 VHA vs. academic 1.08 0.72–1.61 * Adjusted for age at diagnosis, race, performance status, Gleason score, node status.

The identification of surrogate endpoints for survival would accelerate the conduct of prostate cancer clinical trials. We evaluated the incidence of distant metastasis (DM) and general clinical failure (GCF) as potential surrogate markers for cause-specific survival (CSS) using data from RTOG 9202. Men with locally-advanced prostate cancer were treated with 4 months of neoadjuvant and concurrent androgen deprivation (AD) with external beam radiation therapy (RT) followed by no additional therapy (standard arm) or 24 additional months of AD (experimental arm). General clinical failure (GCF) was defined as local progression, regional/distant metastasis, initiation of hormonal therapy, or PSA ≥ 25 ng/mL after completion of RT (Taylor et al, IJROBP 50:1212, 2001). DM and GCF with 5 years follow up were formally tested for surrogacy for CSS at 10 years. Surrogacy testing utilized Prentice's criteria: 1) that the treatment is prognostic for the true endpoint; 2) that the treatment is prognostic for the surrogate endpoint; 3) that the surrogate endpoint is prognostic for the true endpoint; and 4) that the full effect of treatment on the true endpoint is explained by the surrogate. Landmark analyses at 3 and 5 years were performed to address potential follow up length bias. Among 1521 eligible, randomized patients, there were 218 cause-specific deaths, 291 DMs and 726 GCFs. The 10-year CSS was 85% in the standard arm, compared with 89% in the experimental arm (p = 0.0089). The 5-year DM and GCF rates were 14% and 43% in the standard arm, compared with 9% and 33% in the experimental arm (HR = 1.7 [1.2–2.4] and 1.6 [1.3–1.9], respectively). Both 5-year DM and 5-year GCF were significantly associated with 10-year CSS (p < 0.0001 and p = 0.0005, respectively). 10-year CSS was independent of treatment among patients who experienced DM (p = 0.63) or GCF (p = 0.23), and 10-year CSS was independent of treatment among patients who did not experience DM (p = 0.60) or GCF (p = 0.69), indicating that the 5-year surrogates both capture the full effect of treatment on 10-year CSS. Therefore, without adjusting for length bias, 5-year DM and GCF both satisfied all of Prentice's criteria for 10-year CSS. The analysis using only the 1364 patients alive at landmark time 3 years, showed that both 3-year DM and GCF satisfied all of Prentice's criteria for 10-year CSS. In the 1178 patients alive at a landmark time of 5 years, there was no statistical difference in CSS between the treatment arms (p = 0.08). Nonetheless, 5-year DM and GCF met the most important of the remaining Prentice's criteria for 10-year CSS. This analysis of patient data from RTOG 9202 demonstrates that DM and GCF at 5 years satisfy Prentice's surrogacy criteria for CSS at 10 years. These endpoints should be validated in future randomized clinical trials for localized prostate cancer.

Purpose/Objective: In our prior report, nadir PSA post-radiotherapy was found to be a predictor of freedom from biochemical failure (bNED), distant metastasis (FDM), and cause specific survival. The relationship between post-treatment PSA nadir, clinical failure as manifested only by a rising PSA, and true clinical relapse is unclear. For nadir to be of any clinical value, it must be a correlate of outcome in patients who are free of evidence of relapse when accurate measurements of nadir are made, at a minimum of 1.5 years after radiotherapy (RT). In this study, nadir PSA was examined as an independent predictor of bNED and FDM in patients selected to be at low risk of micrometastatic disease and who were treated with contemporary RT doses. Materials/Methods: A population of 615 men treated at Fos Chase Cancer Center between April 1989 and December 1995 with 3DCRT alone have previously demonstrated the importance of post-treatment PSA nadir in predicting distant failure and death due to prostate cancer. Four-hundred and thirteen of the 615 men are the basis of this report, with the exclusions detailed as: 184 men achieved a post-treatment PSA nadir within eighteen months of treatment and/or experienced a nadir PSA level above 2.0 ng/mL; 18 men were treated with ICRU dose levels <70 Gy. The median dose in the study cohort was 74 Gy (70–81) and median follow-up was 80 months (18–154). The outcome measures of interest were bNED as defined by the ASTRO consensus panel and FDM. Cox multivariate regression was used to identify the independent predictive ability of nadir while adjusting for other previously determined prognostic factors, including dose, stage, grade, and pre-treatment PSA. Outcome rates were estimated univariately using Kaplan-Meier methodology and comparisons were accomplished using the log-rank test. Results: The median post-treatment PSA nadir was 0.5 ng/mL (0.1–2.0). The median time to nadir was 44 months (18–127). One-hundred and eight men experienced biochemical failure, with an overall eight-year bNED rate of 68%. Cox multivariate analysis demonstrated that lower nadir continues to predict for improved bNED (p=.002), after adjusting for pre-treatment PSA (p<.0001), Gleason Score (p=.006), and dose (p=.03). Univariate eight-year bNED rates were 80% (n=128), 65% (n=116), 66% (n=125), and 36% (n=44) for patients achieving nadir levels ≤0.2, 0.3–0.5, 0.6–1.0, and 1.1–2.0 ng/mL, respectively (p<.0001). Ten men developed distant failure, with an overall eight-year FDM rate of 97%. Multivariate analysis demonstrated nadir ≤1.0 ng/mL to be the strongest predictor of improved FDM (p=.0003). Eight-year FDM rates were 98% and 90% for patients achieving nadir levels ≤1.0 and 1.1–2.0 ng/mL, respectively (p=.0003). Conclusions: This study suggests that post-treatment PSA nadir remains highly predictive of biochemical and clinical outcome within a patient population treated with contemporary dose levels and carefully selected to be at low risk of harboring micrometastatic disease at initial presentation. A nadir PSA over 1.0 ng/mL denotes a particularly poor prognosis, with a biochemical failure rate of 64% and a five-fold increase in distant failure over that of those achieving a lower nadir, and may be used as an indicator for early salvage intervention.

Purpose/Objective: The proportion of tumor cells stained immunohistochemically for the proliferation marker Ki-67, the Ki-67 staining index (Ki67-SI), has been associated with prostate cancer patient outcome. Few studies have studied radiotherapy (RT)-treated patients and no prior study has investigated patients treated with short term (STAD; 4 mo) or long term (LTAD; 28 mo) androgen deprivation plus RT. In this report, the association of Ki67-SI to local failure (LF), biochemical failure (BF), distant metastasis (DM), cause-specific death (CSD) and overall death (OD) was determined in men randomized between STAD+RT and LTAD+RT. Materials/Methods: There were 537 cases in RTOG 92–02 (35.5%) that had sufficient tissue for Ki67-SI analysis. Median follow-up was 96.3 months. Ki67-SI cut points of 3.5% and 7.1% were previously found to be related to patient outcome and were examined here in Cox proportional hazards multivariate analysis (MVA). Ki67-SI was also tested as a continuous variable. Time events were measured from randomization. Unless mentioned, covariates were dichotomized in accordance with stratification and randomization criteria as follows: median Age (≤70, >70 years), initial pretreatment PSA (iPSA; ≤30, >30 ng/ml), T-category (T2c, T3-T4), Gleason score (2–6, 7–10), and assigned treatment (STAD+RT, LTAD+RT). Results: Median Ki67-SI was 6.5% (range 0–58.2). There was no difference in the distribution of patients in the Ki-67 analysis cohort (n=537) and other patients in RTOG 92–02 (n=977) by any of the covariates or end points tested. In MVAs, Ki67-SI (continuous) was associated with LF (p=0.08), BF (p=0.0445), DM (p<0.0001), CSD (p<0.0001) and OD (p=0.0094). For these analyses iPSA was also included as a continuous variable. When categorical variables were used in MVAs, the 3.5% Ki67-SI cut point was not significant. In contrast, the 7.1% cut point was related to BF (p=0.09), DM (p=0.0008), and CSD (p=0.017). Ki67-SI was the most significant correlate of DM and CSD. In univariate analysis, the 5 year DM rate (95%CI) was 8.6% (5.2, 12.1%) for Ki67-SI ≤7.1%, compared to 17.3% (12.2, 22.3%) for Ki67-SI >7.1% (p=0.0004). The 5 year CSD rates were 3.3% (1.2, 5.5%) and 12.0% (7.4, 15.9%) for Ki67-SIs ≤7.1% and >7.1% (p=0.0041). In analyzing patients by assigned treatment, Ki67-SI was the strongest correlate of DM for each group. Because one of the principal effects of LTAD+RT over STAD+RT was the diminution of DM, a detailed analysis of the hazard rates for DM in all possible covariate STAD+RT subgroup combinations (n=16), as referenced to the most favorable LTAD+RT subgroup, was performed. The goal was to identify subgroups treated with STAD+RT that had the same DM rate as those who received LTAD+RT and would therefore not require LTAD. The 4 favorable subgroups identified had T2c disease, a Ki67-SI ≤7.1%, a PSA ≤30 or >30 ng/ml, and a Gleason score of 2–6 or 7–10. The 5 year DM rates were 5.2% and 3.6% for the favorable STAD+RT (n=63) and LTAD+RT (n=59) populations (p=0.69). Conclusions: Ki67-SI was the most significant determinant of DM and CSD, and was also associated with OD. The relationship of Ki67-SI to patient outcome is a continuous function, wherein the higher the Ki67-SI the greater the risk of an adverse result. The 7.1% cut point has been confirmed as correlating with DM and CSD, and appears to be useful in the selection of patients who will not benefit from LTAD+RT over STAD+RT. The KI67-SI should be incorporated into the stratification of patients in future trials.

Purpose/Objective: Patient decision-making is enhanced with knowledge of what others going through similar treatments prefer and what the impact on quality of life (QL) is with different treatment alternatives. This study assessed the preferences, utilities and QL in patients treated with 3DCRT for prostate cancer, as well as predictors of these variables. Materials/Methods: Patients participating in a prostate cancer risk registry recruited through the Department of Radiation Therapy were analyzed. Preferences for health states of impotence (IP) and incontinence (IC) associated with prostate cancer therapies were elicited using a modified Time Trade-off (TTO) interview. Utility (a number between 0, death and 1, best possible health) for health states were calculated from stated preferences. QL (using the Functional Assessment of Cancer Therapy-Prostate, FACT-P; made up of a general QL scale, FACT-G and a prostate specific subscale), sexual adjustment (via a modified Sexual Adjustment Questionnaire, SAQ), and urinary function (using the AUA-Symptom Problem Index, SPI) were evaluated in patients with prostate cancer receiving 3DCRT. Stepwise multiple regressions were used to assess predictors (sociodemographics and pretx and treatment characteristics) of utilities and QL. Results: Fifty-six men (mean age 63 yrs.) treated between 12/90 and 2/99 with 3DCRT (mean dose 73 Gy) and a mean follow-up of 55 mos (range 17-131) completed the TTO. 83% were stage T1-T2 and 30% were on hormones. 33/56 (59%) pts completed the FACT and 38-39/56 (68%) pts completed the SAQ & SPI. Utilities followed a linear trend with declining scores for increasing risk of poorer health states. Men showed an increased preference and higher utility for health states associated with radiotherapy as compared to surgery or hormone therapy (Table 1). Global QL scores were higher than reports of a population of 466 mixed cancer patients (Cella, '94). SAQ scores were low but urinary function scores were excellent. Predictors of utilities for IP included marital status with being married associated with higher utilities, and for IC younger age and higher income were associated with higher utilities. The only treatment variable that predicted utilities for either IP or IC was RT dose, with higher dose associated with higher utilities. QL scores did not predict utilities. Predictors of sexual function included hormone therapy, while higher RT dose predicted worse urinary function, higher income predicted better global QL and higher tumor grade predicted worse FACT-P subscale scores.Table 1Utility Values for Prostate Cancer Treatment Related Health States and QL Scores of Men Treated with 3DCRTUtilities Health StateIncontinence(IC)Impotence(IP)(IP)% Risk of Health State10% Risk30% Risk40% Risk80% Risk99% RiskUtility Value, Mean (SD).9321 (.21).8961 (.22).9415 (.14).9043 (.15).8750 (.16)Qualtiy of LifeInstrumentFACT- GGlobal QL#higher scores are better;FACT-P Subscale#higher scores are better;FACT-P Total#higher scores are better;SAQ#higher scores are better;SPI∗lower scores are betterPossible Score Range0–1120–480–1608–10028-0QL Score Mean (SD)97 (13.2)37.8 (8.1)134.8 (19.5)39.5 (15.8)4.28 (5.66)# higher scores are better;∗ lower scores are better Open table in a new tab Conclusions: Men in this study treated with 3DCRT had relatively high utility for health states associated with treatment and higher utilities for health states associated with RT than hypothetical health states associated with surgery or hormonal therapy. All aspects of QL were quite good with the exception of sexual function. Predictors of both utilities and QL can assist in identifying patients at risk for poorer outcomes.

Purpose/Objective: To determine the importance of biochemical failure (BF) as a correlate of distant metastasis (DM) and death and to identify factors associated with early DM and death in those who failed biochemically. Materials/Methods: There were 1236 patients who were treated from 1987–1998 with external beam radiotherapy (EBRT) alone or EBRT plus short course androgen deprivation (AD; ≤6 mo). Of these, 942 had sufficient PSA determinations in follow-up for the analyses described and represent the study cohort. All had a pretreatment PSA (iPSA) available. Median radiation dose was 72 Gy (range 62–81 Gy), with 91% treated with conformal EBRT. Median iPSA was 9.9 ng/ml, median age was 69 years and median follow-up was 73 mo. Pretreatment PSA levels were 10–<20 in 29%, and ≥20 ng/ml in 20%. AD was given to 13% for a median of 3 mo (range 1–6 mo); long course hormone therapy patients were not included. Stage T3/T4 disease was seen in 12% and Gleason 7–10 in 27%. The ASTRO consensus definition was used to define BF. Kaplan-Meier calculations were from the start of EBRT or from BF, as indicated. Cox proportional hazards regression multivariate analysis (MVA) of factors associated with DM and death were investigated. Results: Biochemical failure was observed in 316 (34%), while 66 (7%) experienced DM and 230 (24%) died during the study period. Kaplan-Meier 5 and 10 yr estimates from the start of EBRT for BF, DM, and death were 38% and 43%, 6% and 13%, and 13% and 39%, respectively. For those with an event, median times to BF, DM, and death were 24, 34, and 58 mo. Factors related to increased DM in MVA were T3/T4 disease (p=0.007), BF (p=0.004), conventional EBRT (p=0.0015), shorter time to BF (TTBF; p<0.0001), earlier year of treatment (p<0.0001) and high nadir PSA (p<0.0001). Factors related to death in MVA were Gleason 7–10 (p=0.02), conventional EBRT (p=0.002), high nadir PSA (p=0.0009), high iPSA (p=0.0001), age>70 years (p<0.0001), year of treatment (p<0.0001), absence of BF (p<0.0001), and shorter TTBF (p<0.0001). The 5 year OS rates for the cases with and without BF were 92% and 84% (p=0.02), respectively. The relationship of no BF to increased death may be due to competing risks; longer survival allowing for a greater chance of manifesting BF. Of the cases with BF, there were 281 in whom sufficient information for PSA doubling time (PSADT) calculation was present. Median values for the parameters examined were 58 mo follow-up, 14.9 ng/ml iPSA, 69 years of age, 72 Gy EBRT dose, 23 months TTBF and 13 mo PSADT. Salvage AD (SAD) was administered to 28%. Gleason 7–10 was seen in 34%, Stage T3/T4 in 18%, and pretreatment PSAs of 10–19.9 and ≥20 ng/ml in 30% and 36%. Median times to DM and death from BF were 28 mo and 81 mo, respectively. All calculations described below are post-BF. Factors related to increased DM risk in MVA were no SAD administration (p=0.006), high nadir PSA (p<0.0001), and shorter PSADT (p<0.0001). Factors related to increased risk of death were T3/T4 disease (p=0.038), no short course AD (p=0.004), age >70 years (p=0.001), and shorter TTBF (p<0.0001). One possible explanation for why AD was associated with reduced survival is tumor growth in relapsing patients was slower after AD+RT (28 mo PSADT mean), as compared to RT alone (17 mo PSADT mean; p=0.004 in multivariate linear regression). However, AD was not related to reduced DM, perhaps because SAD overshadowed any effect. Conclusions: While BF was a significant correlate of DM, increased BF was seen in those who survived longer. This was probably due to using overall survival and competing risks of death. There is every indication that BF will eventually reduce survival over time. The results from the time of BF indicated that BF progression to DM and subsequently death was associated with the timing (TTBF <12 mo) and rate of the PSA rise (PSADT <12 mo) as well as other factors; such patients should be entered into novel trials. Early SAD reduced DM, but did not impact on survival. The data also suggest that short course neoadjuvant/adjuvant AD may ultimately impact on survival by reducing the rate of tumor progression after BF. Confirmation by others of this observation is needed.

Purpose/Objective: Adoption of the ASTRO Consensus Statement on PSA After Radiation Therapy has been critical in evaluating and comparing outcome following treatment with radiation. However, since its near universal adoption, several points have remained controversial, notably backdating the date of failure (DOF) to the point midway between the post-treatment PSA nadir and the first rise. The purposes of this study is to evaluate the impact of backdating on bNED (biochemical no evidence of disease) control and suggest changes in the definition, which can modify the influence of backdating on bNED control. Materials/Methods: Between 4/1/1989 and 11/30/1998, 1017 patients with non-metastatic prostate cancer were treated with 3D Conformal Radiation Therapy (3DCRT) alone. No patient received hormonal therapy prior to, during, or after radiotherapy unless local, distant, or biochemical failure was documented. bNED control was defined using the ASTRO Consensus definition. bNED failure was the time midway between the post-treatment PSA nadir and the first of the three consecutive rises in PSA. To evaluate the impact of backdating on bNED control, two alternate failure time-points were chosen, backdating to the first rise in PSA after nadir versus using the third post-treatment PSA rise. To further evaluate the impact of follow-up time on outcome, a subset analysis of 480 patients treated between 4/1/1989 and 11/30/1994 was evaluated. bNED control was estimated univariately using Kaplan-Meier methodology and comparisons in failure times were based on paired t-tests. Patients were also stratified by prognosis: favorable (PSA < 10, Gleason score [GS] 2–6, T1c/T2a), intermediate (not favorable or unfavorable), and unfavorable (T3, PSA > 20 or GS 8–10). The median follow-up was 53 months (range: 2 to 144 months) for the entire group and 74 months (range: 2 to 144 months) for the subset. Results: The 10-year actuarial bNED control rate was 66%, 59%, and 31% using the ASTRO definition, 1st rise and 3rd rise, respectively. Differences between the 3rd rise and other 2 definitions were statistically significant (p < 0.001). These differences persisted when patients were stratified by prognostic group (favorable: 81%, 77%, 55%; intermediate: 66%, 59%, 36%; unfavorable: 50%, 41%, 10%). These same differences in bNED control were observed for the long-term follow-up subset- 10-year actuarial bNED control rate was 57%, 52%, and 32% using the ASTRO definition, 1st rise and 3rd rise, respectively. These differences were statistically significant (p < 0.001). Hazard functions were estimated to better characterize risk of failure over time for the 3 DOFs. Hazard analysis was also accomplished using two additional points: 1) between the 1st and 2nd rise, and 2) between the 2nd and 3rd rise. The hazard functions demonstrated that earlier failure dates (e.g., ASTRO definition) are associated with the highest risk of failure in the first four years following radiotherapy. Similarly, later failure dates result in the highest risk of biochemical failure at up to 8 years post-treatment. These trends were seen for all of the DOFs compared to the ASTRO definition. These differences persisted with the subset analysis. Conclusions: Adoption of the ASTRO consensus definition has been crucial in evaluating outcome in the radiation oncology community. However, modifications are necessary to improve the sensitivity and specificity of the definition by changing the DOF to the point of the first consecutive rise. This point is about midway between the post-treatment nadir and the point of the third consecutive rise although validation with large patient datasets will be necessary to confirm this modified consensus definition.

Purpose/Objective: The goal of this study was to investigate the relationship between PSA doubling time (PSADT) and short-term initial androgen deprivation (ADi; <6 mo), and the impact of these factors on prostate cancer progression after radiation therapy.Materials/Methods: Between 5/89 and 10/98, 248 patients treated with 3DCRT experienced biochemical relapse as defined under the ASTRO consensus statement. All patients had sufficient follow-up data for PSA doubling time (PSADT) calculations. The median follow-up and dose was 85 months (15-144) and 73 Gy (62-80). Median ADi was 3 mo. Stepwise linear regression models were used to assess predictors of PSADT (log transformation) among ADi, time to biochemical failure (<12 mos vs 12+ mos), Gleason Score (2-6 vs 7-10), 1997 AJCC stage (T1/T2a vs T2b/T3), dose (continuous), post-treatment PSA nadir (continuous), pretreatment PSA (continuous), and age (continuous). Stepwise Cox regression models were used to assess predictors of distant metastasis (DM) and overall survival (OS) among the same set of covariates along with PSADT.Results: There were 33 (13%) patients who developed distant metastasis and 36 (15%) who died. The mean and median PSADT was 19 and 12 months (1-375). Predictors of a longer PSADT in linear regression were ADi (p=.08), longer interval to biochemical failure (p=.0002), Gleason Score 2-6 (p=.01), and T1/T2a tumors (p=.06). Predictors of improved DM rates in Cox modeling were ADi (p=.0065), Gleason Score 2-6 (p=.0014), lower nadir (p=.0055), and longer PSADT (p

Purpose/Objective: The American Joint Commission on Cancer staging classification for prostate cancer permits the use of “all information available prior to first definitive treatment” for clinical staging. Such information includes not only the results of digital rectal exam (palpable stage), but also findings from radiographic imaging and prostate biopsy. Ultrasound imaging has been shown to add little to palpable stage, while endorectal coil MRI holds more promise. Inclusion of positive prostate biopsy laterality status (unilateral vs bilateral) into clinical staging is controversial. The addition of biopsy laterality status would, for example, up-stage many patients from category T2a to T2b with the potential for stage migration. The purpose of this project was to determine the effect of the addition of positive prostate biopsy information on staging and the influence of this information on freedom from biochemical failure (bNED) outcome. Materials/Methods: Between 4/1/89 and 8/31/99, 1,038 clinically localized prostate cancer patients were treated with 3-dimensional conformal radiation therapy (3D-CRT) alone. The influence of unilateral versus bilateral positive biopsy on bNED outcome was determined using Kaplan-Meier survival combined with the log-rank test for univariate analysis and Cox proportional hazards regression for multivariate analysis. Results: Median patient age was 68 years and median pretreatment PSA was 8.5ng/ml (range: 0.4ng/ml to 191.0ng/ml). Median radiotherapy (RT) dose prescribed to the prostate was 75 Gy and median follow-up was 46 months. The proportion of patients with bilateral positive biopsies for category T1c (n=334) was 45%, T2a (n=335) was 46%, T2b (n=41) was 6%, and T3 (n=26) was 4% (See Table 1). There was no significant difference in 7-yr bNED based on biopsy laterality status (unilateral vs bilateral) for palpable category T1c (76% vs 78%, p=0.55) or T2a, when subdivided by those having unilateral versus bilateral positive biopsies. There was a marginally statistically significant difference in 7-yr bNED in the T2b category based on laterality (46% vs 74%, p=0.04). This is probably the consequence of random statistical error, since those with T2b and positive biopsies bilaterally did as well as patients with no palpable disease (74% & 77%, respectively), and those with T2b and positive biopsies unilaterally did as poorly as those with T3 disease (46% & 43%, respectively). Last, 7-yr bNED was not significantly different based on biopsy laterality status for palpable category T3 (47% vs 40%, p=0.52).Table 17-yr bNED for T-Category and Positive Biopsy LateralityAJCC T-CategoryPositive Biopsy Laterality7-yr bNED (%)T1c, T2, T3Unilateral & Bilateral (n=1,038)-Unilateral (n=736)69Bilateral (n=302)71 (p=0.95)†Univariate log-rank test for unilateral vs bilateralT1cUnilateral & Bilateral (n=440)77Unilateral (n=334)76Bilateral (n=106)78 (p=0.55)T2aUnilateral & Bilateral (n=427)66Unilateral (n=335)66Bilateral (n=92)66 (p=0.84)T2bUnilateral & Bilateral (n=116)64Unilateral (n=41)46Bilateral (n=75)74 (p=0.04)T3Unilateral & Bilateral (n=55)43 (p<0.0001)‡Univariate overall log-rank test for T1c vs T2 vs T3Unilateral (n=26)47Bilateral (n=29)40 (p=0.52)† Univariate log-rank test for unilateral vs bilateral‡ Univariate overall log-rank test for T1c vs T2 vs T3 Open table in a new tab Conclusions: Inclusion of positive biopsy laterality status into clinical staging did not change bNED outcome for patients with clinically localized prostate cancer treated with 3D-CRT alone. Those with unilateral positive biopsies had about the same bNED outcome as those with bilateral positive prostate biopsies. There is no indication that biopsy laterality results should be included in the clinical staging of prostate cancer. However, such staging information is now required by tumor registries. Positive biopsy laterality status does not impact on outcome and should not be used to alter palpable clinical stage.

Purpose/Objective: Standard radiotherapy doses of 70 Gy have proven effective for the treatment of favorable risk prostate cancer. However, there may be subgroups which would benefit from dose escalation. In this study, recursive partitioning was used to identify favorable risk subgroups that require higher than standard doses for optimal control.Materials/Methods: This analysis included 453 patients treated with radiotherapy for prostate cancer with favorable features (T1c/T2a using 1992 AJCC staging system, PSA<10 ng/ml, and Gleason score 2–6). A two-phase approach was used. First, recursive partitioning was performed using the covariates of T1c vs T2a, PSA (continuous), and Gleason score (ordinal 2, 3, 4, 5, and 6). Second, the subgroups identified were then subjected to further analysis of optimal dose cut-points using recursive partitioning coupled with Kaplan-Meier estimation. All patients were treated with 3DCRT alone between 5/89 and 7/00. The median follow-up was 38 months. The median dose was 74 Gy (range 63–82 Gy). The primary outcome measured was freedom from biochemical failure (bNED) defined using the ASTRO consensus guidelines.Results: Four subgroups were identified using the recursive partitioning method: (1) T1c and PSA≦6; (2) T1c and PSA>6; (3) T2a and PSA≦7.5; (4) T2a and PSA>7.5. The 5-year Kaplan-Meier bNED rates for the four subgroups were 93%, 84%, 80%, and 49% respectively, p<.0001 (Table). Using a combination of recursive partitioning and Kaplan-Meier estimation, the dose cut-point found for each subgroup was in the 71–72 Gy range. For patients receiving >71 Gy, the 5-year bNED rates for the four subgroups were 92%, 87%, 78%, and 54% respectively, p<.0001 (Table).The number of patients receiving <71 Gy in each subgroup was too small to provide meaningful results. Pairwise comparisons were performed showing that bNED was significantly worse for subgroup "4" (T2a, PSA>7.5) as compared to each of the other subgroups "1", "2", and "3" (p<.0001, p<.0001, and p<.0006, respectively). Restricting the analysis to patients receiving >71 Gy, bNED was still significantly worse for subgroup "4" as compared to each of the other subgroups "1", "2", and "3"(p<.0001, p=.0002, and p=.0047, respectively). Subgroups "1" and "3" were significantly different (p=.02), even when the comparison was restricted to patients receiving >71 Gy (p=.02). Subgroups "1" and "2" were significantly different when all patients were included (p=.03), but not when the comparison was restricted to patients receiving >71 Gy. Subgroups "2" and "3" were not significantly different. The mean and median doses for the subgroups treated to >71 Gy were 74–76 Gy; there were no significant differences in dose amongst the subgroups.Table5-year bNED (n=453)T1c & PSA≦6T1c & PSA>6T2a & PSA≦7.5T2a & PSA>7.5p valueAll patients93% (n=139)84% (n=178)80% (n=110)49% (n=26)p<.0001>71 Gy only92% (n=134)87% (n=171)78% (n=101)54% (n=23)p<.0001 Open table in a new tab Conclusions: Recursive partitioning defines four risk subgroups among "favorable" risk patients with prostate cancer. Subgroups "2" and "3" do not appear to be significantly different and may be grouped together. The subgroup of patients with T2a and PSA>7.5 is associated with significantly worse outcome even when treated to >71 Gy. Our results suggest that this subgroup is more intermediate risk and should be treated more aggressively to doses above the median value used here (>74 Gy). Purpose/Objective: Standard radiotherapy doses of 70 Gy have proven effective for the treatment of favorable risk prostate cancer. However, there may be subgroups which would benefit from dose escalation. In this study, recursive partitioning was used to identify favorable risk subgroups that require higher than standard doses for optimal control. Materials/Methods: This analysis included 453 patients treated with radiotherapy for prostate cancer with favorable features (T1c/T2a using 1992 AJCC staging system, PSA<10 ng/ml, and Gleason score 2–6). A two-phase approach was used. First, recursive partitioning was performed using the covariates of T1c vs T2a, PSA (continuous), and Gleason score (ordinal 2, 3, 4, 5, and 6). Second, the subgroups identified were then subjected to further analysis of optimal dose cut-points using recursive partitioning coupled with Kaplan-Meier estimation. All patients were treated with 3DCRT alone between 5/89 and 7/00. The median follow-up was 38 months. The median dose was 74 Gy (range 63–82 Gy). The primary outcome measured was freedom from biochemical failure (bNED) defined using the ASTRO consensus guidelines. Results: Four subgroups were identified using the recursive partitioning method: (1) T1c and PSA≦6; (2) T1c and PSA>6; (3) T2a and PSA≦7.5; (4) T2a and PSA>7.5. The 5-year Kaplan-Meier bNED rates for the four subgroups were 93%, 84%, 80%, and 49% respectively, p<.0001 (Table). Using a combination of recursive partitioning and Kaplan-Meier estimation, the dose cut-point found for each subgroup was in the 71–72 Gy range. For patients receiving >71 Gy, the 5-year bNED rates for the four subgroups were 92%, 87%, 78%, and 54% respectively, p<.0001 (Table).The number of patients receiving <71 Gy in each subgroup was too small to provide meaningful results. Pairwise comparisons were performed showing that bNED was significantly worse for subgroup "4" (T2a, PSA>7.5) as compared to each of the other subgroups "1", "2", and "3" (p<.0001, p<.0001, and p<.0006, respectively). Restricting the analysis to patients receiving >71 Gy, bNED was still significantly worse for subgroup "4" as compared to each of the other subgroups "1", "2", and "3"(p<.0001, p=.0002, and p=.0047, respectively). Subgroups "1" and "3" were significantly different (p=.02), even when the comparison was restricted to patients receiving >71 Gy (p=.02). Subgroups "1" and "2" were significantly different when all patients were included (p=.03), but not when the comparison was restricted to patients receiving >71 Gy. Subgroups "2" and "3" were not significantly different. The mean and median doses for the subgroups treated to >71 Gy were 74–76 Gy; there were no significant differences in dose amongst the subgroups. Conclusions: Recursive partitioning defines four risk subgroups among "favorable" risk patients with prostate cancer. Subgroups "2" and "3" do not appear to be significantly different and may be grouped together. The subgroup of patients with T2a and PSA>7.5 is associated with significantly worse outcome even when treated to >71 Gy. Our results suggest that this subgroup is more intermediate risk and should be treated more aggressively to doses above the median value used here (>74 Gy).

Purpose: In patients treated with definitive 3DCRT for localized prostatic adenocarcinoma, we sought to evaluate the relationship between pretreatment prostate gland volume and post-treatment PSA nadir, as well as the relationship of volume and nadir with biochemical control (bNED). Two subgroups are studied: favorable (PSA <10 ng/ml, Gleason Score 2-6, and T1/T2A) and unfavorable (one or more: PSA >=10, Gleason score 7-10, T2B/T3). Materials and Methods: 655 men (n=271 favorable and 384 unfavorable) were treated with 3DCRT alone between 5/89 and 11/97. All patients had information on prostate gland volume and a minimum of 24 months follow-up (median 56 months, range 24-126). 481 men (n=230 favorable and 251 unfavorable) remained bNED at time of analysis, with biochemical failure defined in accordance with the ASTRO consensus definition. Univariate estimates and comparisons of bNED were accomplished using Kaplan-Meier methodology, the log-rank test, and Cox regression. Multivariate analysis of bNED was accomplished using proportional hazards modeling. Factors analyzed included pretreatment prostate gland volume, post-treatment PSA nadir, pretreatment PSA, palpation stage, Gleason score (grade), center of prostate dose, and perineural invasion (PNI). For bNED patients, the relationship between PSA nadir and prostate volume was evaluated using Pearson's correlation and predictors of PSA nadir were established using linear regression. Results: For all patients, a positive correlation between pre-treatment volume and post-treatment PSA nadir was demonstrated (p<0.0001). Subgroup analysis of the favorable and unfavorable patients also demonstrated a positive correlation between volume and nadir (p=0.0017 favorable and p=0.0002 unfavorable). Using multiple regression, the following were found to be predictive of post treatment nadir in all bNED patients: prostate gland volume p<0.0001), pre-treatment PSA (p<0.0001), stage (p=0.0003), and dose (p=0.004). Grade and PNI were not statistically significant. For the favorable group, stage (p=0.001), pretreatment PSA (p=0.0082), prostate volume (p=0.0128), and grade (p=0.0596) were found to be predictive of nadir. Dose and PNI were not statistically significant. In the unfavorable group, prostate volume (p=0.0024), dose (p=0.0039), pretreatment PSA (p=0.0182), and stage (p=0.0296) were found to be predictive of nadir. Grade and PNI were not statistically significant. On multivariate analysis, nadir is not associated with bNED in the favorable subgroup (p=.13) but is strongly associated in the unfavorable prognosis group (p=.0001). On multivariate analysis,volume is associated with bNED in the favorable subgroup (p=.02) independent of nadir effects [bNED control at five years for favorable patients is 85% for volume <=80 cc and 70% for volume >80 cc (p=.04)] and is not associated in the unfavorable groupings (p=.75). bNED control at five years for unfavorable patients is 71% for nadir <=1 ng/ml and 31% for nadir >=1 ng/ml (p=<.0001). Conclusion: 1) This is the first demonstration that prostate gland volume is predictive of PSA nadir for cured patients both favorable and unfavorable subgroups. It is logical that the larger volume of normal tissue present pretreatment in larger glands would result in a higher PSA nadir after irradiation. 2) Prostate gland volume is also predictive of bNED failure in favorable but not unfavorable patients. This is independent of nadir and may be explained by the inaccurate assessment of tumor volume in large prostate glands. 3) PSA nadir is not associated with bNED status in favorable patients (p=.21) and nadir should not be a component of the definition of cure in the favorable group. 4) PSA nadir is strongly predictive of bNED in the unfavorable patients (p=<.0001).

Purpose: The impact of inter-observer variability on evaluation of post-implant dosimetry of permanent Iodine-125 implants has been studied extensively using CT based evaluations. However, CT provides poor soft tissue contrast and hence delineation of organs such as the prostate may be inaccurate. Consequently, the post-implant evaluations may also be inaccurate, and provide little information on the spatial and temporal dosimetric impact of seed placements inside the Evaluation Target Volume (ETV). The purpose of this study was to quantify dosimetric evaluation parameters (D90, V100, V90 and ETV) as a function of patient, observer and imaging modality (CT and MRI) contributions.Materials and Methods: Ten consecutive patients were selected for this study. All patients in this study had T1c/T2a disease with pre-treatment PSA levels ≤10 ng/ml and Gleason scores 2-6. Pre-planning dosimetry was performed using RTOG 98-05 protocol guidelines for defining a Planning Target Volume (PTV) derived by expanding a Clinical Target Volume (CTV), which was the pre-implant TRUS definition of the prostate. The CTV was expanded by 2-3 mm in the lateral dimensions and 2-3 mm in the anterior dimension for each TRUS axial image. CT and axial gradient echo MRI scans (0.23T, TE=3.9 ms, TR=7.9 ms) were obtained 2 hours and 30 days post-implant. Both sets of scans were completed within one hour and the ETV was outlined on the CT and MRI images by four radiation oncologists. Commercial treatment planning software was modified to enable post-implant dosimetry evaluation from MRI images. I-125 seeds were identified on the CT images using a seed sort algorithm with redundancy correction; spatial seed locations were registered on the fused MRI images and post-implant dosimetry was performed following the AAPM TG-43 protocol. Multiple regression was used to establish the significance of patient, observer and imaging modality (MRI or CT) contributions to the various dosimetric evaluation parameters.Results: Linear modeling demonstrates that patient and imaging modality significantly contributes to dosimetric evaluation parameters D90(p=.012 and p=.009, respectively), V100 (p=.025 and p=.004) and V90 (p=.090 and p=.004). Observer contribution was insignificant for these parameters. Table 1 represents p-values for equality of variance tests according to imaging modality for the three parameters. Standard deviations for these evaluation indices were smaller for the MRI scans, indicating that volumes drawn on MRI are more consistent than those on CT. ETV only showed an observer dependence, which was attributed to a single relatively inexperienced observer. Exclusion of this observer rendered the ETV independent of imaging, observer and patient.Conclusion: We have shown that MRI based post implant dosimetry provides results that are more consistent between observers compared to CT based post implant evaluations. MRI based post-implant dosimetry is a superior modality of evaluating the spatial and temporal effects of seed placement in the prostate and consequently provides an excellent feedback mechanism for perfecting peripheral implant methodologies. Tabled 1 Purpose: The impact of inter-observer variability on evaluation of post-implant dosimetry of permanent Iodine-125 implants has been studied extensively using CT based evaluations. However, CT provides poor soft tissue contrast and hence delineation of organs such as the prostate may be inaccurate. Consequently, the post-implant evaluations may also be inaccurate, and provide little information on the spatial and temporal dosimetric impact of seed placements inside the Evaluation Target Volume (ETV). The purpose of this study was to quantify dosimetric evaluation parameters (D90, V100, V90 and ETV) as a function of patient, observer and imaging modality (CT and MRI) contributions. Materials and Methods: Ten consecutive patients were selected for this study. All patients in this study had T1c/T2a disease with pre-treatment PSA levels ≤10 ng/ml and Gleason scores 2-6. Pre-planning dosimetry was performed using RTOG 98-05 protocol guidelines for defining a Planning Target Volume (PTV) derived by expanding a Clinical Target Volume (CTV), which was the pre-implant TRUS definition of the prostate. The CTV was expanded by 2-3 mm in the lateral dimensions and 2-3 mm in the anterior dimension for each TRUS axial image. CT and axial gradient echo MRI scans (0.23T, TE=3.9 ms, TR=7.9 ms) were obtained 2 hours and 30 days post-implant. Both sets of scans were completed within one hour and the ETV was outlined on the CT and MRI images by four radiation oncologists. Commercial treatment planning software was modified to enable post-implant dosimetry evaluation from MRI images. I-125 seeds were identified on the CT images using a seed sort algorithm with redundancy correction; spatial seed locations were registered on the fused MRI images and post-implant dosimetry was performed following the AAPM TG-43 protocol. Multiple regression was used to establish the significance of patient, observer and imaging modality (MRI or CT) contributions to the various dosimetric evaluation parameters. Results: Linear modeling demonstrates that patient and imaging modality significantly contributes to dosimetric evaluation parameters D90(p=.012 and p=.009, respectively), V100 (p=.025 and p=.004) and V90 (p=.090 and p=.004). Observer contribution was insignificant for these parameters. Table 1 represents p-values for equality of variance tests according to imaging modality for the three parameters. Standard deviations for these evaluation indices were smaller for the MRI scans, indicating that volumes drawn on MRI are more consistent than those on CT. ETV only showed an observer dependence, which was attributed to a single relatively inexperienced observer. Exclusion of this observer rendered the ETV independent of imaging, observer and patient. Conclusion: We have shown that MRI based post implant dosimetry provides results that are more consistent between observers compared to CT based post implant evaluations. MRI based post-implant dosimetry is a superior modality of evaluating the spatial and temporal effects of seed placement in the prostate and consequently provides an excellent feedback mechanism for perfecting peripheral implant methodologies. Tabled 1

Purpose: The benefit of long-term adjuvant hormones (LTH) combined with conventional doses of radiation has been demonstrated in the treatment of locally advanced prostate cancer in several randomized prospective studies. However, the benefit of LTH with high dose 3D conformal radiation therapy (3DCRT) has not been investigated. The purpose of this study is to compare biochemical endpoints between patients with locally advanced prostate cancer treated with high dose 3DCRT with and without LTH. Materials and Methods: Patients included in this study demonstrated T1/T2a disease with Gleason score (GS) 8-10 or T2b/T3 disease with any GS. Between January 1991 and December 1999, 236 patients meeting these criteria with locally advanced non-metastatic prostate cancer were treated with high dose 3DCRT at Fox Chase Cancer Center. 65 patients received LTH (≥ 6 months) with 3DCRT and 171 patients were treated with 3DCRT alone. The median dose to the center of the prostate overall and for both groups was 78 Gy (60-84 Gy). The median follow-up for all patients was 35 months (3-102 months), 40 months (11-93 months) for the LTH patients and 34 months (3-102 months) for the 3DCRT alone patients (p= 0.02). bNED control was defined using the ASTRO Consensus definition. bNED control was estimated univariately using Kaplan-Meier methodology and comparisons were based on the log-rank statistic. Multivariate analysis was performed using Cox proportional hazards modeling with stage, grade, pretreatment PSA, and adjuvant hormones as covariates. Results: The 5-year actuarial bNED control rate was 80% for the 3DCRT and LTH patients compared with 57% for the 3DCRT patients alone (p= 0.036). There were 9/65 failures versus 42/171 for the 2 groups, respectively. Univariate analysis demonstrated that pre-treatment PSA, T stage, and the use of LTH to be predictive of bNED control. LTH use remained predictive of bNED control (p= 0.007) even after adjusting for pre-treatment PSA (p= 0.0002), GS (p= 0.023), and T stage (p= 0.007). Conclusion: For patients with locally advanced non-metastatic prostate cancer, combining LTH with high dose 3DCRT resulted in improved bNED control compared with high dose 3DCRT alone.

Purpose: It is well established that pre-treatment PSA, post-treatment nadir, stage, grade, and dose are all predictors of biochemical relapse in men treated with radiation therapy for prostate cancer. As series have matured, the power to demonstrate predictors of clinical failure has increased, with reports indicating that clinical characteristics are the strongest factors associated with distant failure and cause-specific survival. This study attempts to link post-treatment biochemical profiles to distant failure and subsequent cause-specific death by assessing the relationship between post-treatment PSA nadir and doubling time (PSADT) with these outcome measures. Materials and Methods: 615 men were treated at our institution between 4/89 and 12/95 with 3DCRT alone and median ICRU dose of 73 Gy (64-82). Kaplan-Meier methodology was used to estimate rates for freedom from distant metastasis (FDM) and cause-specific survival (CSS); the log-rank test was used for comparisons. Stepwise Cox models were used to assess multivariate predictors of outcome among pretreatment PSA, stage, grade, dose, post-treatment PSA nadir, PSADT, and androgen deprivation use upon PSA failure (AD). Results: With a median follow-up of 64 months (2-135), 186/615 patients experienced biochemical relapse, 40 developed distant metastasis, and 18 died of prostate cancer. 48 of 186 biochemical failures were treated with AD for disease relapse. Multivariate analysis demonstrated that post-treatment PSA nadir (p<.0001), stage (p<.0001), grade (p=.0009), dose (p=.004), and AD (p=.02) were all highly predictive of FDM, while pretreatment PSA was not (p=.18). Ten-year FDM rates were 96%, 89%, and 49% for nadir values <1.0, 1.1-2.0, and >2.0 ng/mL, respectively (p<.0001, see Figure). Patients achieving nadir in the 0-1.0 ng/mL range experienced similar FDM rates. Multivariate analysis of CSS demonstrated that nadir (p<.0001) and palpation stage (p=.001) were the only predictors of death due to prostate cancer. Ten-year CSS rates were 96%, 96%, and 78% for nadir values <1.0, 1.1-2.0, and >2.0 ng/mL, respectively (p<.0001). 136 of 186 biochemical failures had sufficient PSA information for PSADT calculations; 17/136 failed distantly and 2 died of disease. Multivariate analysis of FDM from time of biochemical failure demonstrated that PSADT (p=.0085), nadir (p=.0091), and AD (p=.005) were all independent predictors of FDM. There was insufficient data for an analysis of PSA doubling time and CSS. Conclusion: To our knowledge, this is the first study to demonstrate the overwhelming predictive power of post-treatment PSA nadir and PSADT for distant metastasis and death due to prostate cancer following external beam irradiation (3DCRT). These results may be used for the early identification of patients at high risk for distant failure and who may be directed to applicable Phase III clinical trials.

Purpose: Conformal external beam radiation and IMRT for prostate cancer are planned primarily based upon CT. MRI has been shown to permit more accurate delineation of the prostate than CT. The purpose of this study was to determine the impact of MRI delineation on target coverage, treatment time, and dose to the bladder and rectum for IMRT in prostate cancer. Materials and Methods: From 2/01-3/01, all patients with prostate cancer treated by three of the authors with conformal external beam radiation underwent routine MRI scanning. The study population consists of 8 of 13 patients chosen for IMRT because of one or more of the following characteristics: PSA > 10 ng/ml, Gleason 7-10, palpable T2B or T3 tumor, or presence of perineural invasion. All underwent CT simulation followed by an MRI using a dedicated open 0.23 T unit in the supine casted treatment position. The prostate, bladder and rectum were delineated first on CT. The studies were fused and the MRI was used to redefine the internal anatomy. Both sets of contours were input into an inverse planning system. For each patient a separate IMRT treatment plan was performed based upon the CT and MRI defined prostate, bladder and rectum. The 16 treatment plans for IMRT were accepted after 45 iterations. This included two sets of plans for one patient who had required a repeat scan after hormonal therapy to meet acceptance criteria. All patients were initially treated to 56 Gy using a four field conformal technique. The planning target volume (PTV) of prostate plus margin of 5 mm was then treated for 23 Gy using IMRT. Results: The MRI-based prostate volumes were larger than on CT for 4 of 8 patients (mean 39.4 cc vs. 34.7 cc, 13.5% increase) with a range of −12.4% to 64.6% (mean 18.3%, median 3.4%). There was no significant difference in treatment planning time required between CT and MRI data sets. Most patients required the same number of beam orientations (5) and iterations (median 1.5) to derive a solution for each data set. In general the MRI-based plans required slight alterations of beam directions to increase separation between the MRI-PTV, bladder and rectum. There was a modest increase in the mean number of segments (35 vs. 38) for the MRI-PTV plans, so that the mean total treatment time was increased by 48 seconds. There was no difference in the dose inhomogeneity between CT and MRI plans (mean 115.7% vs. 116%, median 115.1% vs. 115.1%). Since MRI produces more accurate delineation of the prostate, the coverage of the MRI-PTV with the plan based upon the CT contours represents the potential miss in target coverage if CT alone were used. Dose-volume histogram analysis revealed that the plans generated from the CT data set resulted in coverage of 95% of the MRI-PTV by a mean of 80.0% of the dose (range 56.8%-93.4%, median 80.4%). This compares to a mean of 96.8% of the dose (range 91.8%-100.9%, median 96.6%) covering 95% of the MRI-PTV by the plans generated from the MRI data set. The mean dose received by 20% of the rectum was 59.6% for the CT plans and 60.2% (range 51%-67%) for the MRI plans. The mean dose received by 20% of the bladder was 44.7% and 44.6% (range 23.6%-52.6%) between CT and MRI-based plans respectively. Conclusion: The use of MRI simulation added significantly to the coverage of the PTV during IMRT compared to CT simulation alone. Despite a larger delineation of the prostate by MRI compared with CT, the MRI-based treatment plans did not result in increased dose to the bladder or rectum. There was also no significant increase in planning or overall treatment time with MRI-based plans. The routine use of MRI simulation is a feasible means of improving the therapeutic ratio for IMRT and high dose conformal external beam radiation for prostate cancer.

Purpose: We have previously reported that hypoxic regions exist in human prostate cancer (CaP). In this study, we investigated whether low pO2 in CaP predicts for biochemical control (bNED) after brachytherapy. Materials and Methods: Custom-made Eppendorf pO2 microelectrodes were used to obtain □&100 pO2 readings from both the pathologically involved region of the prostate (as determined by sextant biopsies) and normal muscle (as an internal control) per patient (pt). Eligibility included localized disease, KPS >70, pre-tx PSA, central pathologic review, and an IRB-approved signed consent. 57 pts were prospectively studied; all received brachytherapy implants (48 LDR and 9 HDR) under spinal anesthesia. 9 pts had received prior hormonal therapy. Biochemical failure was defined as 2 consecutive rises in PSA, without a return to baseline. Cox proportional hazards regression was used to evaluate the influence of hypoxia on bNED control univariately and multivariately, while adjusting for: PSA, Gleason score, stage, avg (mean) of the median prostate pO2 levels (P), avg median muscle pO2(M), prostate pO2/muscle pO2 ratio (P/M), type of implant (LDR vs HDR), perineural invasion and hormonal therapy; the P/M ratio was analyzed to control for possible inter-patient and technical variations. Results: With a median f/u of 19 months (4-31), 9 pts developed biochemical failure. A stepwise MVA found the P/M ratio (on a continuum) to be the only predictor of bNED control (p=0.002). A univariate threshold analysis of P/M demonstrated that bNED control at 2 yrs differed significantly at a ratio of <0.05 vs ≥0.05 (31% vs 92%, p=0.001, see Figure). Yet, the classic prognosticators were similar in these two groups of pts with otherwise good-prognosis CaP (table). For P/M separations of 0.1, 0.2, or 0.3, 2 yr bNED differences were less dramatic (e.g., 71% vs 88%, p=0.16 for P/M 0.1). Conclusion: To our knowledge, this is the first study to correlate the degree of hypoxia in CaP with tx outcome after RT. The P/M pO2 ratio is the strongest predictor for bNED on stepwise MVA. Longer f/u in more patients is planned to confirm this result. The P/M threshold analysis supports the radiobiological concept that cellular radioresistance becomes significant only at very low pO2 levels. This study has important implications with the advent of novel hypoxic strategies, including anti-angiogenesis therapies. Tabled 1P/M RatioNMedian F/U (mo)Median Age (yrs)Median PSA (ng/ml)Gleason 6 (%)T1/T2A (%)Muscle pO2 (M) (mmHg)Prostate Po2 (P) (mmHg)<0.051218 (4-25)64 (55-74)7.1 (0-22)83%83%32 (23-72)1.0 (0-2)≥0.054519 (4-31)66 (45-75)6.5 (0.5-22)89%84%27 (1-55)5.3 (0.2-65) Open table in a new tab