Purpose: To identify predictors of biochemical outcome following radiotherapy in patients with a rising prostate-specific antigen (PSA) after radical prostatectomy for prostate cancer.

Patients and Methods: One hundred fifteen patients with a rising PSA after radical prostatectomy received salvage three-dimensional conformal radiotherapy (3D-CRT) alone or with neoadjuvant androgen deprivation. Tumor-related and treatment-related factors were evaluated to identify predictors of subsequent PSA failure.

Results: The median follow-up time after 3D-CRT was 42 months. The 4-year actuarial PSA relapse-free survival, distant metastasis-free survival, and overall survival rates were 46%, 83%, and 95%, respectively. Multivariate analysis, which was limited to 70 patients receiving radiation without androgen deprivation therapy, showed that negative/close margins, absence of extracapsular extension (P < .01), and presence of seminal vesicle invasion (P < .01) were independent predictors of PSA relapse after radiotherapy. Neoadjuvant androgen deprivation did not improve the 4-year PSA relapse-free survival in patients with positive margins, extracapsular extension, and no seminal vesicle invasion. However, neoadjuvant androgen deprivation did improve PSA relapse-free survival when one or more of these variables were absent.

Conclusions: Salvage 3D-CRT can provide biochemical control in selected patients with a rising PSA after radical prostatectomy. Among patients with positive margins and no poor prognostic features, 77% achieved PSA control after salvage 3D-CRT. Salvage neoadjuvant androgen deprivation therapy may improve short-term biochemical control, but it requires further study.

OVER THE past several years, there has been increasing interest in more definitively determining the role of salvage radiotherapy for a rising prostate-specific antigen (PSA) in patients after radical prostatectomy. Risk factors that have been consistently associated with a postprostatectomy PSA relapse include elevated preoperative PSA level, Gleason score 7, positive surgical margins, extracapsular extension, seminal vesicle invasion, and lymph node involvement. Only a select group of patients with disease confined to the prostate bed will benefit from radiotherapy. How to clinically identify such patients remains a challenge.

The rationale for postoperative radiation therapy is to eradicate foci of residual prostate tumor cells; however, the optimal time to treat patients with an isolated PSA relapse remains unclear. The patterns of care of some urologists reflect skepticism about the clinical significance of an isolated PSA relapse. Consequently, urologists refer only a minority of patients for adjuvant or salvage therapy within 3 years following surgery. A recent study found that 54% of urologists recommend observation for an isolated case of a rising PSA, deferring radiotherapy until there has been a documented local recurrence.The absence of randomized clinical trials supporting a benefit for salvage radiotherapy may also affect the decision not to refer patients for treatment.

To identify those patients most likely to respond to salvage radiation therapy, guidelines are needed to predict whether a rising PSA represents tumor confined to the prostate bed, or whether it is an indicator of systemic disease. We have previously reported our early experience with postoperative three-dimensional conformal radiotherapy (3D-CRT). In this study, we update our results with the goal of identifying those clinicopathologic features at initial presentation and at the time of recurrence that predict the efficacy of salvage radiotherapy.

Treatment Toxicity
Twenty-seven (23%) patients had no acute urinary (GU) toxicity, whereas 61 (53%) and 24 (21%) patients had grade 1 and 2 toxicity, respectively. Three patients (3%) experienced grade 3 toxicity (one patient discontinued treatment after dilatation of a bladder neck contracture at 37.8 Gy, one patient required a catheter for urinary outlet obstruction at 30 Gy, and one patient required urethral dilatation of an anastomotic stricture 1 month after completion of radiotherapy). Forty-three (37%) patients had no acute rectal (GI) toxicity, whereas 54 (47%) and 18 (16%) patients had grade 1 and 2 toxicity, respectively. No patient experienced grade 3 or higher GI acute toxicity.

The 4-year actuarial rates of late grade 2 and 3 GU toxicity were 9% and 10%, respectively. Among the nine patients with late grade 2 GU toxicity, six (67%) had hematuria, with confirmed radiation cystitis on cystoscopy requiring no further treatment, and three (33%) required medications for increased urinary frequency. All nine patients with late grade 3 GU toxicity required dilatation of a urethral stricture or bladder neck contracture; however, three of these nine patients had evidence of locally recurrent tumor near the bladder neck after completion of salvage radiotherapy. In addition, three of the nine patients had prior dilatation or transurethral resection of the prostate or bladder neck. The 4-year actuarial rate of late grade 2 GI toxicity was 12%, consisting of moderate radiation proctitis. No relationship between dose of salvage radiotherapy and the appearance of late GU toxicity was observed. No patient experienced grade 3 or higher GI late toxicity.

Before radiotherapy, 53 patients (46%) were fully continent, whereas 37 (32%) patients had stress incontinence not requiring pads and 25 (22%) patients had stress incontinence requiring pads. After radiotherapy, 56 (49%) patients were fully continent, whereas the remaining patients had chronic stress incontinence requiring pads or without requiring pads (26% and 25%, respectively). Among the 53 patients who were fully continent before radiotherapy, nine (17%) developed long-term stress incontinence, but only one (2%) required pads. The remaining 43 (84%) patients maintained complete continence. Among the 62 patients with stress incontinence before radiotherapy, 12 (19%) patients regained complete control. For the 25 patients requiring pads before radiotherapy, three (12%) patients improved, whereas the remaining 22 (88%) patients still required pads.

We used the absence of extracapsular extension, negative/close margins, and seminal vesicle invasion to identify a cohort of patients with a favorable outcome after salvage radiotherapy alone or after salvage radiotherapy and androgen deprivation therapy (see figure above). Equal weighting was given to all three risk factors. Among the 70 patients receiving radiotherapy alone, the 4-year PSA relapse-free survival was 77% for those with zero risk factors and 20%  for those who had one or more risk factors. Among the 45 patients who received radiotherapy and neoadjuvant androgen deprivation therapy, the 4-year PSA relapse-free survival was 91% for those with zero risk factors and 43% for those who had one or more risk factors. In both subgroups, there was no significant difference in PSA outcome regardless of whether the patient had one, two, or three risk factors. For the 34 patients without any risk factors for PSA failure, the addition of neoadjuvant androgen deprivation did not improve PSA relapse-free survival. However, there was a significant improvement in PSA control with neoadjuvant androgen deprivation among the 81 patients with risk factors.

In this article, we observed significantly different PSA outcomes after salvage radiotherapy based on the presence of prognostic risk variables, which underscores the need for careful patient selection for therapy. In our study, only patients with positive margins in the absence of other poor prognostic features fared well, achieving PSA control in 77% in this subgroup. Reported PSA relapse-free survival rates after radiotherapy for postprostatectomy PSA failures have ranged from 10% to 64%, with variable definitions of PSA failure and length of follow-up. Some series have reported poor results with salvage radiotherapy with PSA control rates of only 10% to 31%. However, the median preradiation PSA levels were higher (1.4 to 2.8 ng/mL) in those previous series than in our series (0.87 ng/mL). Furthermore, the radiation dose was relatively low in some series, ranging from 58.7 to 61.2 Gy, and heterogeneous radiation techniques were sometimes used.Finally, some reports contained more patients with Gleason scores greater than 7, SVI, or negative margins compared with other studies. Our study demonstrates that margin status, extracapsular extension, and SVI, all well-recognized pathologic predictors of biochemical relapse after radical prostatectomy, are also relevant predictors of outcome with salvage radiation therapy after surgery. By offering salvage 3D-CRT to select patients with a higher likelihood of residual localized disease, radiation can be avoided in those patients that are more likely to have undetected micrometastatic disease.

We have demonstrated that salvage 3D-CRT is reasonably well tolerated. The majority of our patients experienced grade 2 toxicity or less and had a low risk of moderate to severe late complications; however, these toxicities were documented by physician-report toxicity grading, which is now recognized to be less sensitive to treatment-related morbidity than validated patient-report health-related quality-of-life methodology. Conformal techniques minimize the volume of the bladder and rectum included in the treatment field and allow the delivery of high doses of radiation with acceptable toxicity. Although there was a 10% risk of requiring a urethral dilatation after radiation in our study, as many as 15% to 20% of patients develop bladder neck contracture or anastomotic stricture requiring intervention after radical prostatectomy alone. We used a higher dose than in most other series, with 97 patients (84%) receiving a minimum of 66.6 Gy prescribed to the planning target volume, with doses ranging from 70 to 75 Gy at the isocenter. No increase in toxicity was observed at the higher isocenter dose levels. A recent American Society of Therapeutic Radiology and Oncology (ASTRO) consensus panel recommended doses of 64 Gy or higher, and two large series confirmed that high-dose levels of radiation improve biochemical and clinical relapse-free survival. Longer follow-up is necessary, but dose escalation may be important in patients with clinically documented local failures given their worse prognosis compared with patients with only a biochemical relapse.

Among patients who did not receive neoadjuvant androgen deprivation, we identified three independent predictors of biochemical failure: the absence of extracapsular extension, negative/close margins, and SVI. Positive margins increase the likelihood of biochemical failure after radical prostatectomy, but close margins do not.Previous smaller retrospective series evaluating salvage radiation therapy have not found a statistically significant difference in PSA failure rates based on margin status, and two other large series did not report on its influence.However, we believe that positive margins in the setting of postprostatectomy PSA relapse favor a local failure that may respond to radiotherapy. Similarly, extracapsular extension also favors a localized source of detectable PSA after surgery. Although SVI has also been associated with higher local failure rates,it is also associated with lymph node involvement and is a strong predictor of subsequent distant metastases.On the basis of the poor results among patients with any one of the risk factors we have identified, many such patients are likely to have a systemic component to biochemical relapse after prostatectomy that cannot be treated effectively by radiotherapy alone.

The role of androgen deprivation and its benefit in combination with radiation therapy after radical prostatectomy is poorly defined. Only a few retrospective series have included patients that received androgen deprivation, ranging from 4% to 29% of the patients evaluated. Although one series identified the short-term use of androgen deprivation therapy as the only independent predictor of biochemical control with salvage radiotherapy, 46% of those patients also received treatment to the entire pelvis, rather than the prostate bed alone.Using PSA failure as an end point can be problematic because of the variable time necessary for testosterone normalization after androgen deprivation therapy. Although we did not identify the use of androgen deprivation therapy as an independent predictor of biochemical outcome, there may be a role for systemic therapy after prostatectomy for some patients. The RTOG is presently conducting phase III randomized trials to assess the benefit of androgen deprivation and radiation in the adjuvant and salvage setting after prostatectomy (ie, RTOG P-0011 and RTOG 96–01, respectively). In our study, the majority of patients (70%) had at least one unfavorable pathologic feature (absence of extracapsular extension, positive seminal vesicles, or negative/close margins). However, the role and optimal timing of salvage androgen deprivation after radical prostatectomy remains unclear. Whether salvage androgen deprivation therapy will alter the natural history of PSA failure and delay the subsequent onset of distant metastases is uncertain. Thus, this therapy requires further study.

In this study, we included all patients with a measurable postoperative PSA level. One report indicated that postoperative PSA levels of as low as 0.1 ng/mL signified a higher risk of clinical relapse. In contrast, a recent report indicates that a PSA cutoff of 0.4 ng/mL more consistently correlates with clinical progression than lower values, and observation of the patient may be a reasonable option until the PSA rises to this level. However, preradiation PSA is the most consistently identified PSA variable for biochemical outcome after salvage radiotherapy. We found that both presalvage and preradiation PSA were significant variables from cutoffs of 0.4 to 1.0 ng/mL, but only when all 115 patients were analyzed. When we limited the analysis to the 70 patients who did not receive neoadjuvant androgen deprivation therapy, PSA was not an independent predictor of biochemical outcome after salvage radiotherapy. Nevertheless, higher PSA levels may represent a larger volume of hypoxic prostate cancer, and longer times to radiotherapy may allow proliferating localized clonogens to transform into metastatogenic phenotypes. Although we did not identify presalvage PSA as an independent predictor of biochemical outcome among the patients that received radiotherapy alone in this study, we believe it may still be an important variable to consider when offering salvage therapy. PSA doubling time also may help predict the site of clinical recurrence after radical prostatectomy, and it has been examined in the setting of salvage radiotherapy, with mixed results. We are currently evaluating the effect of PSA kinetics as an outcome variable, and this effect will be the subject of future analysis and publication.

Given the recent interest in salvage radiotherapy, all reports have short follow-ups relative to the long natural history of prostate cancer, including the follow-up of our own study. The retrospective nature of this study also introduces several biases. First, our patients had variable surgical techniques, and uniform criteria were not used for referral for radiotherapy. Second, interobserver variability in reporting pathologic features may overstate the importance of our findings. Eight patients could not be staged because some pathologic features were not described, and the use of neoadjuvant androgen deprivation in 12 patients before prostatectomy could influence the pathologic findings. Finally, the use of different assays to measure PSA levels complicates evaluation of serologic variables. However, variable surgical technique, referral patterns, and pathology reporting may reflect more accurately the efficacy of radiotherapy in the community at large than do reports based on patients of a single academic surgeon.

In conclusion, high-dose salvage 3D-CRT can provide durable biochemical control in select patients with a rising PSA after radical prostatectomy. We have identified positive margins, extracapsular extension, and the absence of SVI as independent predictors of 4-year biochemical control. The absence of any of these predictors of biochemical control indicates that a systemic component may be present that radiation to the prostate bed cannot eradicate. Prospective trials are currently underway to help identify which patients can be cured with postoperative radiation therapy. In the future, molecular staging and emerging genetic biomarkers may lead to better patient selection. However, at this time, we believe our findings may assist in the decision to offer salvage radiotherapy to patients after radical prostatectomy.