Purpose: Salvage radical prostatectomy (RP) may potentially cure patients who have isolated local prostate cancer recurrence after radiotherapy (RT). We report the long-term cancer control associated with salvage RP in a consecutive cohort of patients and identify the variables associated with disease progression and cancer survival.

Results: Overall, the 5-year progression-free probability was 55%  and the median progression-free interval was 6.4 years. The preoperative PSA level was the only significant pretreatment predictor of disease progression in the multivariate analysis (p = 0.01). The 5-year progression-free probability for patients with a preoperative PSA level of <4, 4–10, and >10 ng/mL was 86%, 55%, and 37%, respectively. The 10-year and 15-year cancer-specific mortality after salvage RP was 27% and 40%, respectively. The median time from disease progression to cancer-specific death was 10.3 years (95% confidence interval, 7.6–12.9). After multivariate analysis, the preoperative serum PSA level and seminal vesicle or lymph node status correlated independently with disease progression.

Conclusions: Greater preoperative PSA levels are associated with disease progression and cancer-specific death. Long-term control of locally recurrent prostate cancer after definitive RT is possible when salvage RP is performed early in the course of recurrent disease.

Discussion

Despite these advances, initial attempts at curative therapy fail in many men. Specifically, in one-third of men primary RT will fail, with 25% still having localized disease. These patients are candidates for salvage RP. This procedure, however, has not gained wide acceptance owing to concerns regarding morbidity and lack of evidence demonstrating long-term cancer control in the absence of systemic therapy. Only 25% of surveyed urologists and radiation oncologists would recommend salvage RP to a patient aged 45–65 years with definitive local recurrence after RT, 54% would either observe or treat with ADT, and 20% would consider salvage brachytherapy

In a recent extensive analysis aimed at morbidity and long-term functional outcomes, an improving trend was noted likely owing to the improved delivery of RT. In our early experience, the incontinence rate was 58% and the major complication rate was 33% (including a 15% risk of rectal injury). Others have reported incontinence and complications in 65% and 50% of patients, respectively . However, the high complication rates in these early series were observed mainly in patients who had undergone prior staging pelvic lymph node dissection and/or open retropubic brachytherapy, both associated with extensive pelvic fibrosis, which increases the difficulty of dissection and the likelihood of complications. In our recent analysis among patients who had not undergone prior pelvic lymph node dissection and/or open seed implantation, the median operative time, blood loss, and transfusion rates associated with salvage RP were similar to our results with standard RP. For patients treated after 1993, the major complication rate was 13%, significantly less than the 33% rate reported in our early experience. In addition, only 1 of 58 patients treated since 1993 had a rectal injury, which was closed primarily without sequelae. In terms of urinary incontinence, an estimated 68% of our patients required one urinary pad daily or less after salvage RP. Although 23 patients required placement of an artificial urinary sphincter for moderate to severe incontinence, most patients had a good outcome after sphincter insertion and only 1 required a revision procedure. With less pelvic fibrosis encountered after EBRT and transperineal brachytherapy, a significant proportion of salvage RP patients are candidates for unilateral or bilateral nerve-sparing salvage RP. We estimated that 28% of these recovered erections sufficient for intercourse after nerve-sparing salvage RP, including 5 of 7 previously potent patients who had preservation of both neurovascular bundles.

Because of the increasing use of salvage RP, understanding its long-term cancer control is of utmost importance. Cheng et al.  studied 86 patients who underwent salvage RP before 1996 and reported a 10-year cancer-specific survival rate of 64%. Prostate cancer mortality was associated with a greater Gleason score and DNA ploidy. Our results demonstrated a 10- and 15-year cancer-specific survival rate of 73% and 60%, respectively. Salvage RP alone totally controlled the disease in 30% of the patients 10–15 years after the procedure. These outcomes likely reflected better patient selection, primarily suggested by the declining median preoperative PSA levels during the study period.

In our assessment of risk factors, the only preoperative variable associated with cancer survival was the serum PSA level. The postoperative finding of seminal vesicle invasion and/or lymph node metastasis (present in 42% of this cohort) carried a poor prognosis, as did greater Gleason scores. These same factors were associated independently with PSA progression after multivariate analysis. The differences in prognosis to this endpoint for patients with a PSA level <4 ng/mL and those between 4 and 10 ng/mL were statistically significant (p = 0.03). These findings suggest the need for improved methods of detecting radio-recurrent cancers early in their course when the best outcomes could be achieved. However, too often, salvage RP is contemplated too late.

Because serum PSA levels typically drop continuously for up to 18 months after RT and, uncommonly, become undetectable, assessing cancer control may be difficult. The question of what is post-RT PSA failure, a subject of controversy, was standardized with the American Society Therapeutic Radiology Oncology (ASTRO) definition (3). However, this definition carries two fundamental caveats. The first is a lack of an absolute nadir value. This shortcoming is particularly important and troublesome, because most long-term studies have suggested that the nadir value has definitive predictive power. Shipley et al. (2) examined 1,765 men with clinically localized prostate cancer (clinical Stage T1-T2) treated with RT between 1988 and 1995. Men who had a PSA nadir of ?0.5, 0.6–0.9, 1.0–1.9, and ?2.0 ng/mL had a 5-year PSA recurrence-free survival rate (using the ASTRO definition for PSA recurrence) of 83%, 68%, 56%, and 28%, respectively. A recent study of 4,829 patients treated at nine institutions validated these results, demonstrating that patients who reached a PSA nadir of ?0.5 ng/mL had the best 5-year and 8-year PSA recurrence-free survival rates (75% and 70%, respectively). For those whose PSA nadir was between 0.5 and 1 ng/mL, the corresponding rates were 60% and 53%. Importantly, only 65% and 34% of the reported cohort had a PSA nadir of <1 and <0.5 ng/mL, respectively. In addition, 4% of the patients developed clinical evidence of disease before their PSA met the ASTRO definition of failure (1, 2). The second caveat is the characterization of local recurrence. A persistent rise in the PSA level after RT most likely represents disease recurrence. However, men with local radio-recurrent prostate cancer may not have a steady PSA rise, delaying fulfillment of the criteria set by the ASTRO definition for recurrence. This has been highlighted by the results from the M. D. Anderson Cancer Center dose-escalation trial that required a biopsy 24 months after RT. Men with a positive biopsy were more likely to experience cancer recurrence (hazard ratio range, 1.1–6.0), even though they were considered disease free by the ASTRO definition (17). Other investigators have reported similar results (4, 5). Overall, the inadequacy of the current definition of PSA failure was clearly illustrated in our study, in that the contemporary patients underwent a median of 10 PSA tests, yet 5 years elapsed between RT and the diagnosis of local recurrence. In synthesis, we are missing our window of opportunity to eradicate the disease and cure these patients.

Owing to the potential limitations of serial PSA levels after RT, we suggest a more aggressive strategy to identify local recurrence. For men whose PSA level never drops to <1 ng/mL, we recommend prostate biopsy 24–36 months after RT. In addition, a PSA rise, even if it does not meet the ASTRO definition, should be investigated with a prostate biopsy. If the results are positive, endorectal MRI should be recommended to assess the disease extent and determine whether salvage local therapy is indicated. However, a negative result should lead to careful PSA monitoring. We would not hesitate to repeat a biopsy in 6–12 months if clinical suspicion remains high owing to suspicious digital rectal examination findings or an elevation or inconsistent pattern in the PSA dynamics. Only by decreasing our threshold to evaluate local recurrence will we be able to diagnose and alter the disease course.

We recognize the limitations of our findings. The patients in our cohort were carefully selected and had prolonged life expectancies, and two experienced surgeons performed the procedures. The variations among surgeons that have been shown to affect the prostatectomy results are also very likely to occur in the salvage surgery setting. We caution surgeons against performing this operation without formal uro-oncologic training with exposure to such patients and the hazards within. Importantly, our patients were treated during a 20-year span, and the disease presentation has changed significantly. Few cancer death events occurred, limiting our ability to detect potential interactions and effects from risk factors. In addition, the initiation of ADT was not controlled or standardized. Nevertheless, the cancer control rates and overall survival were remarkably good, and we expect this to continue or improve.

Conclusion

Salvage RP, although technically challenging, provides excellent local control of radio-recurrent cancer and can eradicate the disease in a high proportion of patients who are treated while the cancer is confined to the prostate or immediate periprostatic tissue. Suspicion of radio-recurrent disease by an inconsistent PSA pattern in patients with good health and a life expectancy >10 years should prompt an early biopsy to identify the patients who may benefit from additional local therapy with acceptable morbidity.