Retrospective review of a cohort of 501 patients at 5 US academic tertiary referral centers who received salvage radiotherapy between June 1987 and November 2002 for detectable and increasing prostate-specific antigen (PSA) levels after radical prostatectomy. Disease progression after salvage radiotherapy, defined as a serum PSA value 0.1 ng/mL above the postradiotherapy PSA nadir confirmed by a second PSA measurement that was higher than the first by any amount, by a continued increase in PSA level after treatment, or by the initiation of androgen deprivation therapy after treatment.

Radiation was delivered to the prostatic fossa by a variety of techniques, including a 4-field technique, 6-field conformal technique, and 5-field intensity-modulated radiotherapy approach, in daily fractions of 180 to 200 rad (1.8-2.0 Gy) using 10- to 23-MV photons. Twenty-four patients (5%) received radiation to pelvic lymph nodes. The median radiation dose was 6480 rad (64.8 Gy) (range, 3780-7560 rad [37.8-75.6 Gy]); 6 patients (1.1%) received doses less than 5940 rad (59.4 Gy) and 65 patients (13%) received doses greater than 7000 rad (70 Gy).

Results  Over a median follow-up of 45 months, 250 patients (50%) experienced disease progression after treatment, 49 (10%) developed distant metastases, 20 (4%) died from prostate cancer, and 21 (4%) died from other or unknown causes. The 4-year progression-free probability (PFP) was 45%. By multivariable analysis, predictors of progression were Gleason score of 8 to 10 (hazard ratio, preradiotherapy PSA level greater than 2.0 ng/mL, negative surgical margins, PSA doubling time (PSADT) of 10 months or less  and seminal vesicle invasion. Patients with no adverse features had a 4-year PFP of 77%. When treatment was given for early recurrence (PSA level 2.0 ng/mL), patients with Gleason scores of 4 to 7 and a rapid PSADT had a 4-year PFP of 64% and of 22% when the surgical margins were positive and negative, respectively. Patients with Gleason scores of 8 to 10, positive margins, and receiving early salvage radiotherapy had a 4-year PFP of 81% when the PSADT was longer than 10 months and of 37% when the PSADT was 10 months or less.

Conclusions  Gleason score, preradiotherapy PSA level, surgical margins, PSADT, and seminal vesicle invasion are prognostic variables for a durable response to salvage radiotherapy. Selected patients with high-grade disease and/or a rapid PSADT who were previously thought to be destined to develop progressive metastatic disease may achieve a durable response to salvage radiotherapy.

For patients with recurrent prostate cancer after radical prostatectomy, salvage radiotherapy remains the only potentially curative therapy. However, results from the Cancer of the Prostate Strategic Urologic Research Endeavor (CaPSURE) longitudinal registry indicate that less than 50% of patients who receive secondary treatment after radical prostatectomy will undergo salvage radiotherapy.The majority will receive ADT, a treatment that offers no hope for cure. Patients with high-grade disease, a rapidly increasing PSA level, and a short disease-free interval after radical prostatectomy are at the highest risk for progression to distant metastases.^. Urologists may be reluctant to treat patients with salvage radiotherapy because previous studies have reported that it is ineffective for cancers with these aggressive features. There is also a widespread perception that the majority of patients with PSA recurrence have occult metastatic disease, precluding successful local salvage therapy.

We conducted a multicenter analysis of 501 patients undergoing salvage radiotherapy for an increasing PSA level after radical prostatectomy in order to reliably identify prognostic variables associated with a durable response. In this cohort, a Gleason score of 8 to 10, preradiotherapy PSA level greater than 2.0 ng/mL, negative surgical margins, PSADT of 10 months or less, and seminal vesicle invasion were significant predictors of disease progression despite salvage radiotherapy. Yet we demonstrated that subsets of patients with high-grade disease and/or a rapid PSADT who were thought to be incurable could still achieve a durable response to salvage radiotherapy when the treatment was administered early in the course of recurrent disease. These results suggest that salvage radiotherapy may prevent metastatic disease progression for those patients at the highest risk.

Numerous studies have demonstrated better outcomes with salvage radiotherapy when it is administered at the earliest evidence of disease progression, ie, when PSA has just begun to increase above detectable levels. A high preradiotherapy PSA level has consistently been shown to negatively influence the outcome of patients undergoing salvage radiotherapy. Studies have proposed preradiotherapy PSA cut points from 0.6 to 4.0 ng/mL for predicting the most favorable response to salvage radiotherapy. The American Society for Therapeutic Radiology and Oncology (ASTRO) consensus panel concluded that the appropriate PSA cut point "seemed to be 1.5 ng/mL" based on the available evidence. In our multivariable analysis, only preradiotherapy PSA level greater than 2.0 ng/mL had a significant association with progression after salvage radiotherapy. However, patients receiving treatment at very low PSA levels (0.6 ng/mL) had an improved outcome compared with patients with a preradiotherapy PSA level between 0.61 and 2.0 ng/mL when other variables were considered and when corrected for multiple hypothesis testing. The use of very low PSA thresholds risks overtreating patients whose PSA level is detectable due to residual benign prostatic tissue. For example, one recent report demonstrated that a single PSA elevation of less than 0.4 ng/mL after radical prostatectomy is associated with subsequent stable, nonprogressing disease in up to 50% of patients. Thus, while outcomes are better when salvage radiotherapy is administered earlier in the course of recurrent disease, physicians should avoid the risk of overtreatment by confirming a trend of increasing serum PSA levels rather than simply relying on a single PSA cut point to establish clinically significant disease recurrence.

The identification of high-grade disease and a rapid PSADT as risk factors for disease progression after salvage radiotherapy is consistent with the observed association of these variables with the development of metastatic disease in cohorts of patients treated with radical prostatectomy and with primary radiotherapy. Tumors with these features are more likely to metastasize early in the course of the disease (either before radical prostatectomy or from residual pelvic disease).

We confirmed our previous observation that positive surgical margins are a powerful predictor of a durable response to salvage radiotherapy. A positive surgical margin suggests a greater likelihood that recurrence is due to residual pelvic disease. Therefore, a patient with positive margins who relapses is more likely to benefit from salvage radiotherapy than a patient with negative margins, whose PSA level is more likely to represent distant disease. Rates of positive surgical margins have been reported in 5% to 53% of patients undergoing radical prostatectomy. While the risk of positive surgical margins generated at radical prostatectomy is associated with adverse clinical and pathological features of prostate cancer, the surgeon's technique appears to be the most significant parameter controlling the status of surgical margins.  Furthermore, positive surgical margins have been demonstrated to be an important predictor of disease recurrence after radical prostatectomy. Our results further establish that disease recurrence in the presence of positive surgical margins often represents local recurrence, even for patients with aggressive features such as a Gleason score of 8 to 10 or a rapid PSADT.

We could not demonstrate the significance of positive lymph nodes for disease progression after salvage radiotherapy, perhaps because the number of patients with positive nodes was low. However, only 1 patient with positive lymph nodes has been followed up for more than 3 years without relapse. We believe that positive lymph nodes are an indicator of systemic disease and that patients with nodal disease are unlikely to benefit from additional local therapy.

Unlike previous studies, this study did not identify an association between the use of neoadjuvant ADT or radiation dose and the success of salvage radiotherapy.  However, the high durable response rates observed in our study were achieved using radiation doses comparatively lower than doses currently administered safely by intensity-modulated technology. A higher radiation dose may be beneficial for these patients, as has been demonstrated in patients treated with primary radiotherapy, but clinical trials would be required to test this hypothesis.

To identify subgroups of patients who are most likely to benefit from salvage radiotherapy, we stratified the outcome using Gleason score, preradiotherapy PSA level, surgical margin status, and PSADT. Overall, patients with a Gleason score of 4 to 7 and a slow PSADT (<10 months) had a 4-year PFP greater than 70% when radiation was delivered early, when the PSA level was lower. A significant proportion of selected patients with adverse features known to be associated with metastatic disease may nevertheless achieve a durable response to salvage radiotherapy. For patients with Gleason scores of 4 to 7 and a rapid PSADT, 67% will have a durable response to early salvage radiotherapy if they have positive surgical margins, compared with 22% for those having negative margins. Likewise, more than half of patients with Gleason scores of 8 to 10 and positive margins will achieve a durable response to salvage radiotherapy, and over one third of these patients with a rapid PSADT will be free of disease at 4 years. These results have important implications for patients with recurrent prostate cancer, as it has been generally believed that high-grade disease and a rapidly increasing PSA level signify the presence of distant metastases, precluding successful local salvage therapy.

Given the insensitivity of current imaging modalities, a durable response to salvage radiotherapy may be the most accurate means of identifying isolated local recurrence. The incidence and clinical characteristics of locally recurrent prostate cancer can thus be inferred from patients who achieve a durable response to salvage radiotherapy. The findings of our study run counter to existing concepts regarding the incidence and clinical significance of locally recurrent prostate cancer. In modern radical prostatectomy series, the reported incidence of isolated local recurrence ranges from 6% to 19%. Because these studies defined local recurrence as abnormal digital rectal examination findings or a positive prostatic fossa biopsy result, the incidence of locally recurrent cancer was likely underappreciated and biased toward patients with higher-volume disease. Based on the 4-year PFP for patients receiving early salvage radiotherapy, 52% of patients with an increasing PSA level will have disease initially confined to the pelvis. The 67% complete response rate to salvage radiotherapy that we observed also suggests that residual pelvic disease initially accounts for the bulk of recurrent prostate cancer, even in patients with combined local and distant recurrence. As this estimate of local recurrence assumes that all patients with isolated local recurrence were cured by salvage radiotherapy, the true rate of isolated local recurrence may be considerably higher than 50%. This suggests that the incidence of isolated local recurrence after radical prostatectomy appears to be closer to the estimate of Lightner et al, who reported a biopsy-proven local recurrence rate of 42% in patients with an increasing PSA level after radical prostatectomy.

Until now, it has been widely believed that locally recurrent prostate cancer is biologically different from metastatic disease and results from inadequate surgery rather than from inherent aggressive tumor biology. Partin et al proposed criteria to distinguish local recurrence from distant metastases for patients with an increasing PSA level after radical prostatectomy. A Gleason score of 8 to 10, seminal vesicle invasion, positive lymph nodes, and a rapid PSA velocity were associated with distant metastases. Local recurrence was more frequently observed in patients with low-grade and organ-confined disease, a slow PSA velocity, and a disease-free interval greater than 3 years, suggesting that isolated local recurrence has a low metastatic potential.

In contrast to these beliefs, our study demonstrates that locally recurrent prostate cancer frequently has features that are often associated with the development of distant metastases if the disease is left untreated. Of the patients who were free of PSA progression for a minimum of 4 years after salvage radiotherapy, 15% had Gleason scores of 8 to 10, 38% had a PSADT of 10 months or less, and 70% had a disease-free interval of 12 months or less after radical prostatectomy. In our study, the majority of patients with positive surgical margins and either a Gleason score of 8 to 10 or a rapid PSADT achieved a durable response to salvage radiotherapy. This evidence suggests the existence of a substantial number of patients with aggressive, recurrent prostate cancer initially confined to the pelvis that has not yet metastasized and that may be effectively treated with radiation therapy if it is delivered early in the course of recurrent disease.

This estimate of the incidence and disease characteristics of local recurrence assumes that the patients in our study are representative of all patients with postprostatectomy PSA recurrence. Our cohort is composed of selected patients whom clinicians believed would likely benefit from salvage radiotherapy and therefore may reflect a biased selection of this patient population. However, the clinical features of our cohort are similar to those of 2 studies examining predictors of clinical disease progression (in the absence of salvage ADT) in consecutive patients with recurrent disease after radical prostatectomy with respect to preoperative PSA level, Gleason score, pathological stage, disease-free interval after radical prostatectomy, and PSADT.  The proportion of patients in our study with positive lymph nodes was similar to that in the study by Roberts et al  but significantly lower than that in the study by Pound

The role of salvage radiotherapy in the treatment of patients with an increasing PSA level after radical prostatectomy remains controversial. No study has shown that salvage radiotherapy improves survival or prevents the development of distant metastases. However, our study demonstrates that salvage radiotherapy can interrupt the natural history of patients with PSA recurrence after radical prostatectomy, even for those patients at the highest risk of progression to distant metastases and death from prostate cancer. In general, salvage radiotherapy appears to be an underused treatment option for patients who experience relapse after radical prostatectomy. Based on our results, we believe that patients with positive surgical margins who experience relapse after radical prostatectomy should be strongly considered for salvage radiotherapy, even those with high-grade disease and/or a rapid PSADT. We have developed a predictive model to estimate the likelihood of treatment success for a given individual that will help guide physicians in the selection of patients for this therapy.