Worldwide, the management of cancer of the prostate remains a difficult problem. Each year close to 190,000 patients in the United States alone are diagnosed with prostate cancer.[1] The optimal initial management of these patients has been the subject of considerable debate. Typically, treatment options including radial prostatectomy, external- beam irradiation, brachytherapy, watchful waiting, or hormonal ablation have been recommended. Individual decisions are often guided by consideration of the published and well documented cancer control rates and complication risks. Very little has been published, however, to guide in the management of locally recurrent prostate cancer, despite the fact that 25% of patients will receive a secondary cancer treatment.[2] Epidemiologic Background In 1988, roughly half of all newly diagnosed prostate cancers were locally advanced or metastatic. Today, however, close to 80% of prostate cancer patients present with clinically localized disease[3] and are candidates for potentially curative therapy. Recent estimates from the National Cancer Data Base[4] and the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) program[ 5] suggest that 30% of presenting patients are managed initially with external-beam irradiation. Yet 10% of low-risk patients, and up to 50% of high-risk patients, may be expected to suffer a failure of primary treatment. Assuming an average case mix, there may therefore be up to 13,500 patients presenting annually in the United States with a recurrence following primary irradiation. For many men no further treatment is required, based on their advanced age or failing health. For others, the recurrent cancer might have a prolonged indolent course and not require further intervention. For those in whom intervention is recommended, 92% of patients treated secondarily after radiation failure are managed with androgen deprivation.[2] However, many of these patients have local recurrence as a component of their disease and might be considered for further treatment to the prostate. A significant- though not well documented- percentage of these patients can be expected to have only a local recurrence with no distant disease. While the number of patients who require salvage therapy is thus greatly reduced, it still represents a sizable and growing group of men. This review is intended to help address this potentially significant patient population and, in particular, to frame the following important questions:

• Can we identify patients who are most likely to require salvage treatment?
• Can we identify patients who are most likely to respond to salvage treatment?
• What is the expected morbidity of salvage brachytherapy?
• What other options should be considered?

Can We Identify Patients Who Are Most Likely to Require Salvage Treatment? An American Society for Therapeutic Radiology and Oncology (ASTRO) consensus panel acknowledged in 1997 that some component of local failure is common in radiotherapy- treated patients with a rising prostate-specific antigen (PSA) level.[6] As a result of this statement, they suggested a limited role for routine rebiopsy in patients experiencing a biochemical recurrence, unless further local treatments are being contemplated. Conservative therapy, with observation or hormonal intervention, is usually the treatment of choice for this population, and histologic confirmation offers little benefit. However, for the occasional patient being considered for a potentially aggressive local therapy, it seems prudent to require biopsy confirmation of prostatic recurrence as a minimum requirement for local intervention. As with men evaluated for primary treatment, the age and overall health of the patient should be the first and foremost factor in identifying appropriate candidates for salvage therapy. This requires an estimation of the individual's likely mortality from other causes, and the anticipated time to symptomatic progression of the cancer or death if left untreated. Unfortunately, neither will be known with any certainty. As with primary treatment, one is unlikely to have much beyond "good clinical judgment" to guide this choice. With that being said, there are some data to serve as a guide. A few authors have written about the clinical course of recurrent disease after radiation therapy. Kuban et al[7] reported that local recurrence diminishes the overall 5-year survival rate from 89% to 66%. Unfortunately, no PSA data were available during the years of this report, and their relevance in the PSA era is unclear. D'Amico et al[8] studied survival following PSA failure in radiotherapy patients. They reported 10-year prostate cancer-specific survival rates of 98%, 72%, and 44% in low-, intermediate-, and high-risk groups, respectively. For this patient population, with a median age of 73 years, slightly more than half of all deaths were from causes other than prostate cancer. Pound et al[9] addressed the natural history of recurrent prostate cancer following radical prostatectomy. After PSA failure, the median time to metastasis was 8 years, with death occurring an average of 5 years thereafter. The Gleason score, PSA doubling time, and interval between surgery and rising PSA were all significant factors predicting for a more aggressive disease course. Pound's study represents a highly selected patient population, and is perhaps not truly indicative of the average patient failing radiation therapy. The median age was not stated, but based on usual surgical criteria, might be expected to be at least a full decade younger than the average radiation therapy cohort. Until better data are available documenting the time to metastasis (or death) for radiation therapy populations, this might be considered a best-case scenario, expecting the average patient to present with clinical progression within 5 to 8 years after PSA failure. The risk of cancer progression must then be weighed against the individual patient's life expectancy. Death from unrelated causes, including heart disease, other cancers, and a host of other illnesses, remains the leading cause of death in patients with prostate cancer. Age-specific actuarial life expectancy tables can be found for populations, but never for individuals. For older and sicker patients, intervention targeting locally recurrent prostate cancer will often be unnecessary. In clinical practice, however, one is often faced with relatively young and healthy patients who must be considered at risk for death or morbidity from recurrent prostate cancer. These patients would be candidates for intervention if the cancer were expected to progress within their life expectancy. Historically, most radiation therapy practices have treated primary prostate cancer patients who averaged over 70 years of age.[8,10] Consequently, the typical patient showing a rising PSA several years after irradiation would generally not be a good candidate for aggressive second-line therapy. However, during the past decade, there has been a dramatic rise in the number of young prostate cancer patients (45-70 years old),[11] and almost a twofold increase in the number of younger men being treated with radiation. Even if only a minority of the treated patients fail, many men will still be young enough to reasonably consider definitive salvage treatment. Local salvage treatment of any sort only makes sense in patients without distant disease. In the PSA era, however, most failures now present with only a rising PSA level. Radionuclide bone scans, computed tomography (CT), or magnetic resonance imaging studies are still indicated but may miss subclinical metastases. Nevertheless, there is every reason to expect that the PSA may contain more information and give some indication of the probability of occult distant metastases at the time of biochemical failure in these patients. Both the level of PSA elevation and time course of the PSA findings since the primary treatment contain potentially useful information. In men given radiation therapy for local recurrence after radical prostatectomy, both the absolute value of the PSA at the time of second-line therapy and the interval from initial treatment to biochemical failure predict for freedom from a second failure, and thus are likely indicators of disease beyond the pelvis. Schild et al[12] determined that grade, dose, and PSA > 1.1 ng/mL at the time of salvage irradiation independently identify patients at high risk of failing salvage therapy. Several other series confirm that delaying treatment after surgical failure may more than halve the 5-year biochemical freedom from failure rates, with the critical PSA level ranging from 1.0 to 2.5 ng/mL.[13-15] Less information is available regarding the prognostic impact of the PSA level after radiation therapy. Given the nature of the treatments and differences in the definition of PSA failure between surgical and radiotherapy patients, one would reasonably expect that radiation patients would likely have a higher average PSA at the time of failure. Indeed, using the ASTRO definition of failure,[ 6] which requires three rising PSA intervals, few patients are likely to be identified as radiation failures while PSA is still less than 2 ng/mL. Tefilli et al[16] compared patients at a single institution having salvage radiotherapy following a radical prostatectomy with men having salvage prostatectomy following failed external- beam irradiation and noted a significant difference in serum PSA at the time second-line therapy was instituted. Patients having salvage radiation after surgery had a mean PSA of 1.1 ng/mL, in contrast to 9.1 ng/mL for salvage after radiotherapy. Salvage treatment after radiation therapy was more likely to be delayed, averaging 15.6 months from the time of recurrence, compared to 4.9 months for patients receiving radiation therapy after failing prostatectomy. There are even fewer data addressing the level of PSA at the time of salvage brachytherapy following external- beam irradiation. One report suggests a PSA > 10 ng/mL as indicative of a high risk of failure,[17] but other levels were not addressed. It is not clear if this dramatic difference in the PSA cut-point following surgery and radiation therapy is a reflection of a biologic difference between these two initial treatment modalities, a delay in recognition of recurrence, or simply a reflection of the paucity of studies evaluating salvage brachytherapy. Most likely all three of these possibilities are to some extent true; it seems reasonable, however, to conclude that the risk of failing second-line therapy rises with PSA elevation. The disease-free interval since initial treatment has also been suggested as an indicator of disease site. Following radical prostatectomy, patients who show signs of PSA failure within the first 24 months have a significantly increased risk of developing distant metastases, whereas those with a later PSA failure are more likely to have only local failure.[9] In addition, a post-initial therapy PSA doubling time of less than 6 months has been suggested as a strong predictor of metastatic disease.[18] Similar results have been reported following radiation therapy, with no clinical signs of progression seen at 28 months in patients with a PSA doubling time of more than 9 months.[19] Guidelines
Thus, it appears that the patients most likely to require additional local therapy are those with

• Pathologically documented local failure
• Clinically and radiographically absent distant metastases
• Life expectancy > 5-10 years based on age and health
• Disease-free interval > 2 years
• PSA < 10 ng/mL
• Long PSA doubling time (ie, > 6-9 months).

Can We Identify Patients Who Are Most Likely to Respond to Salvage Treatment? The concept of using brachytherapy as a salvage treatment is not new. Shortly after the introduction of iodine(Drug information on iodine)-125 as a clinically useful isotope for primary prostate cancer, Goffinet et al[20] reported a small preliminary series of patients treated with retropubic implantation for recurrent disease. Implanting 14 to 23 mCi they delivered 9,000 to 22,500 cGy. Early results were encouraging, with 11 of 14 patients locally controlled and 8 completely disease-free. This report is unfortunately limited by use of the now outdated retropubic technique, lack of PSA results, and short followup. Complications were modest, except in the group implanted with high activity (> 0.5 mCi) sources. Wallner et al[21] reported a unique series of 13 patients with locally recurrent disease following retropubic iodine-125 implantation who underwent a second implant with a median matched peripheral dose of 170 Gy. A variety of techniques including retropubic implantation, open perineal placement, and fluoroscopic and CTguided brachytherapy were used. A 5-year freedom from local progression of 51% was achieved, although 10 of 13 patients developed distant metastases, with 59% survival at the 5-year mark. By today's standards, these were fairly advanced cancers. The PSA assay was not yet available, most had moderately or poorly differentiated cancers, and all patients had palpable tumors larger than 1.5 cm or extraprostatic extension. Loening and Turner reported on 31 patients treated with transperineal ultrasound- guided gold-198 brachytherapy.[ 22] Follow-up was short, and no PSA information was presented. They primarily reported biopsy results, showing that 40% of biopsies became negative while 33% showed cancer with radiation effect. Three patients developed metastases, and the 5-year survival rate was 67% with only two cancer-related deaths. Taken together, it appears from these early series that several conclusions can be reasonably justified. First, prior to the introduction of PSA, it was difficult to select patients with truly localized disease, and overall survival of just over 50% at 5 years could be expected. Local control was generally in excess of 50% with salvage brachytherapy, although it remains difficult to accurately assess. Complications were relatively modest but frequent enough to discourage widespread adoption of these procedures. During the past decade, there has been a virtual revolution in prostate brachytherapy with the introduction of transperineal ultrasound-guided techniques, improved dosimetry, and general acceptance of the procedure for early cancer. During the same years, the introduction of routine PSA testing in screening, pretreatment selection, and posttreatment evaluation has brought about parallel changes in our understanding of the disease and in the stage of newly diagnosed cancers. A few studies have now been published addressing the use of salvage brachytherapy in this more modern era, taking advantage of these advances in implantation and PSA evaluation. Grado et al[23] performed salvage brachytherapy in 46 patients who had developed a recurrence 2 to 5 years following conventional external-beam irradiation (median: 66 Gy) and in 3 patients with recurrence following an iodine-125 implant. They implanted a median of 31.76 mCi iodine-125 or 126 mCi palladium-103 to deliver a median matched peripheral dose of 160 Gy and 120 Gy, respectively. All implants were performed prior to the introduction of TG-43 and NIST-1999 standards. (These conventions were established within the physics community in the late 1990s to correct some discrepancies in implant radiation dose calculations. As such, the equivalent doses using today's calculation methodologies would be approximately 144 Gy and 125 Gy for iodine-125 and palladium-103, respectively.) Local control of 98% was reported, although no biopsy confirmation was provided. Biochemical disease-free survival rates of 48% and 34% were achieved at 3 and 5 years, respectively, using a definition of failure that required two rising PSA values. Patients who reached a PSA nadir < 0.5 ng/mL (47% in this report) fared dramatically better than those who did not. At 5 years, the biochemical disease-free survival rate was 56% for those who achieved this nadir, and 15% for those who failed to do so. This series presents a mixed cohort of adversely selected patients. Four men had also previously undergone radical prostatectomy and had grossly palpable recurrence. A total of 11 cancers were hormone-refractory, and 16 patients had previously undergone a transurethral resection of the prostate. It is highly unlikely that any patient who has hormone-refractory disease or has failed a prior prostatectomy and radiation therapy has only local failure and can be controlled. It is encouraging, however, to look at the subgroup who achieved a PSA nadir < 0.5 ng/mL. One might presume that these patients were less likely to have metastatic disease and did enjoy a 5-year PSA-based tumor control rate of 56%. Our own preliminary results for salvage brachytherapy (at Arizona Oncology Services) were published for the first 17 consecutively treated patients.[17] Doses of 120 Gy prior to TG-43 (or 110 Gy post-TG-43) were used for iodine-125 and 90 Gy for palladium-103, prior to NIST-99. Slightly better overall results were achieved with 53% biochemical disease- free survival and 93% prostate cancer-specific survival reported at 5 years. Gleason sum (< 7) and PSA < 10 ng/mL at the time of salvage predicted for low-risk patients and improved outcomes, although not at a statistically significant level. It is encouraging to note that 83% of these low-risk patients are still free of a second recurrence 5 years following salvage therapy. With additional treated patients and longer follow-up, a retrospective review of our results confirms many of these initial impressions.[24] Currently, 30 patients have been followed up to 125 months (median: 46 months). The biochemical diseasefree survival is shown in Figure 1 for the same risk groups; consisting of PSA ≤ 10 ng/mL vs PSA > 10 ng/mL (Figure 1A) and Gleason score 2-6 vs Gleason score 7-10 (Figure 1B). As in the initial analysis, a lower break point for PSA failed to discriminate any better than PSA greater than or less than 10 ng/mL, although the size of the cohort remains too small to make any statistically significant claims. As might be expected with more than 5 years' follow-up, late prostate cancer deaths have now been seen. At 10 years, the prostate cancer-specific survival rate is 60%. Gleason score and PSA both predict for survival, as can be seen in Figure 2. No cancer deaths have been seen in patients with a Gleason score of 6 or less at the time of salvage brachytherapy. Guidelines
In conclusion, it appears that iodine- 125 doses in the 110- to 145-Gy range (accounting for TG-43) are effective as a second-line treatment for patients failing initial radiation therapy. Incorporating the recommendations of NIST-99, palladium-103 doses of 100 to 115 Gy would be indicated. Overall, approximately 50% of patients should be locally controlled clinically and demonstrate PSA control. As with primary treatment, however, the PSA and Gleason score at the time of salvage can help select patients with better prognoses. Those most likely to respond are men with

• PSA < 10 ng/mL
• Gleason score < 7
• PSA nadir < 0.5 ng/mL.