Salvage radiotherapy for patients with PSA relapse after radical prostatectomy: a single institution experience

Background To assess the efficacy of salvage radiotherapy (RT) for persistent or rising PSA after radical prostatectomy and to determine prognostic factors identifying patients who may benefit from salvage RT. Methods Between 1990 and 2003, 59 patients underwent RT for PSA recurrence after radical prostatectomy. Patients received a median of 66 Gy to the prostate bed with 3D or 2D RT. The main end point was biochemical failure after salvage RT, defined as an increase of the serum PSA value >0.2 ng/ml confirmed by a second elevation. Results Median follow-up was 38 months. The 3-year and 5-year bDFS rates were 56.1% and 41.2% respectively. According to multivariate analysis, only preRT PSA ≥1 ng/ml was associated with biochemical relapse. Conclusion When delivered early, RT is an effective treatment after radical prostatectomy. Only preRT PSA ≥1 ng/ml predicted relapse.


Background
In Europe, the estimated incidence of prostate cancer is 238,000 new cases with 85,000 deaths per year [1]. Radical prostatectomy is the most widely used treatment for localized prostate cancer. Unfortunately, local recurrences occur in up to one-third of the patients by 5 years after surgery. It is generally accepted that 30% (27-32%) of all patients by 10 years after surgery suffer biochemical relapse, defined as increasing serum prostate-specific antigen (PSA) levels >0.2 ng/ml [2,3]. PSA relapse exposes to a 34% risk of metastatic disease at 5 years. After metastatic relapse, median survival is 5 years [4].
"Salvage" radiotherapy (RT) to the prostate bed for biochemical relapse achieved biochemical control in 10-66% of the patients at 5 years [5,6]. PSA failure after prostatectomy could reflect local relapse or metastatic disease. At present, modern imaging techniques lack the sensitivity to differentiate between these two types of relapse. Identification of the best candidates for RT should be based on factors predictive for local relapse. Numbers of positive margins, low Gleason score and/or long PSAdoubling time have been proposed to select patients for RT, but they are still discussed [7].
In this study, we evaluated RT efficacy and determine prognostic factors identifying patients who may benefit from salvage RT.

Methods
We reviewed the records of 59 patients who underwent RT between 1990 and 2003 for biochemical relapse of prostate cancer initially treated with radical prostatectomy. All patients had persistent or rising PSA >0.20 ng/ml at some time after surgery. None had imaging (bone scan and/or abdominal-pelvic computed tomography (CT) Scan) or clinical evidence of metastases at the time of the biochemical relapse.
RT delivered to the prostate bed a median of 66 Gy in 2.2 Gy daily fractions, four days per week, with 18 MV photon beams. Between 1990 and 1998, classical 2D RT was administered using a four-field box technique to 22 (37.3%) patients with fields of 10 cm × 10 cm shaped to protect small bowel, portions of the bladder and posterior rectal wall. The fields encompassed the prostatic/seminal vesicle bed and periprostatic tissues. Pelvic lymph nodes were not irradiated. After 1998, conformational 3D RT was adopted to define optimally the clinical target volume (CTV) and organs at risk (bladder and rectum). CTV included the prostatic/seminal bed, with a security margin to encompass subclinical disease in the periprostatic area. The planning target volume (PTV) was defined by extending the CTV 0.5 cm posteriorly and 1 cm in all other directions. No elective nodal irradiation was performed. Dose Volume Histograms were performed to decrease the dose at organs at risk. Treatment-related toxicity was graded according to the Radiation Therapy Oncology Group (RTOG) criteria [9] and the Expanded Prostate-cancer Index Composite (EPIC) score for urinary incontinence [10].
After radiation, patients were followed every 6 months by a radiation oncologist and a urologist with physical examination and PSA analysis. Imaging to exclude metastastic disease was performed at the physician's discretion, as was the prescription of hormone therapy for biochemical or clinical failure after RT. The interval between surgery and hormone therapy after RT failure was also recorded.
Biochemical failure after salvage RT was defined as an increase of the serum PSA value >0.2 ng/ml confimed by a second elevation.
Clinical failure was defined as evidence of clinical, sonographic, radiographic, or scintigraphic recurrence. The primary end point was biochemical relapse or introduction of hormone therapy before the criteria of PSA recurrence or clinical failure before biochemical relapse were met.
The other end points were overall and specific survival rates.

Statistical analysis
Survival curves were plotted using the Kaplan-Meier method. Survival rates were calculated from the last day of RT. The date of failure was defined at the time of the biochemical failure.
Patients who biochemically had no evidence of disease (bNED) were censured at the time of last follow-up.
Univariate and multivariate analysis using a Cox proportional hazards regression analysis were conducted to identify significant predictors of biochemical outcome for several clinical and pathological factors: pre surgery PSA; Gleason score; high-grade histological differentiation (4 or 5); extra prostatic extension (capsule or seminal vesicle invasion); positive surgical margin(s); ≥3 positive surgical margins; lymphovascular invasion; short PSA DT ≤12 months; persistently high PSA after surgery (>0.2 ng/ml); preRT PSA ≥0.5, ≥1 or ≥2 ng/ml; surgery-RT interval; duration of RT; classical 2D or 3D RT. For all analyses, the level of significance was set at 0.05.
Statistical analysis was performed using Statview software.

Patient characteristics
Patient and tumor characteristics are reported in Table 1. Fifty-nine patients were treated within a 13-year interval. It should be noted that 11 (18.6%) patients had no lymph-node dissection at the time of radical prostatectomy and did not receive any lymph-node irradiation at the time of RT. One (2%) patient had a nodal involvement at the time of RT. Seven patients (12%) received short-term (≤6 months) hormonetherapy, with Luteinizing Hormone Releasing Hormone agonist, after surgery, at the urologist's discretion.
Thirty-five patients had positive margins but no mention of margin status was made in one patient's report.
RT was delivered to 80% of the patients because of a rising PSA levels after postoperative negativity and for persistently high PSA after surgery for the remaining 20%. 3D and 2D salvage RT was performed in 63% and 37% of the patients, respectively. The median RT dose delivered was 66 Gy (59.4-70.4) in a mean of 60 days.

Survival and bDFS
Median follow-up was 38 months after completing RT. At 3 years, the overall and specific survival rates were 93% and 100% respectively. The estimated 5-year overall survival rate was 87% and specific survival was 96%. bDFS at 3 years was 56.1%. The estimated 5-year bDFS was 41.2% ( Figure 1).
Comparisons of bDFS after the end of RT according to dif-ferent preRT PSA thresholds  bDFS after the end of RT according to PSA >0.2 ng/ml defini-tion Figure 1 bDFS after the end of RT according to PSA >0.2 ng/ml definition.
High preRT PSA was associated with poor biochemical control after RT regardless of the biochemical definition used. This observation is consistent with the previously reported finding that preRT PSA was the most frequently selected factor predictive of bDFS . Those authors described poorer prognoses associated with higher PSA values before RT using thresholds ranging from 0.4 to 2 ng/ml. An ASTRO consensus panel recommended that RT be delivered before the PSA level reaches 1.5 ng/ml [32]. For our patients with a preRT PSA <1 ng/ml, the 3-year bDFS was significantly higher than for those with a PSA ≥1 ng/ml (70 vs 30% respectively). We analyzed 3-year bDFS as a function of different preRT PSA thresholds: rates declined as PSA concentrations increased from <0.5 (66.4%) to ≥2 ng/ml (only 33%). These significantly different rates (p = 0.008) are strong arguments supporting early treatment after biochemical relapse. We think that, in the setting of RT, the earlier the better. When biochemical failure is confirmed, and a sufficient number of factors suggestive of local relapse are present, patients should be irradiated without waiting for PSA to reach 1 or 1.5 ng/ml.
Recently, the randomized EORTC 22911 study demonstrated significant improvement for adjuvant RT vs salvage RT in terms of bDFS and clinical local control at 5 years but not for overall survival [33]. Similarly, adjuvant RT significantly increased bDFS vs observation in the ARO 96-02 and the SWOG 8794 studies [34,35]. But adjuvant RT based only on unfavorable histological prognostic features (positive surgical margins, seminal vesicle involvement or extra capsular effraction) would expose some of these patients to over treatment with the risk of incontinence and urethral stricture resulting from the accumulation of the two treatments. Indeed 40-50% of the patients with positive surgical margins would develop biochemical relapses at 5-years [36,37]. To conclude definitively, we should compare adjuvant radiotherapy vs early salvage radiotherapy: GETUG and RADICALS ongoing trials will try to answer to this question.
In this study, we did not observe any difference in terms of incontinence (grade ≥2) according to EPIC score or rectal toxicity (grade ≥2) between salvage 3D and 2D RT. Urinary tract toxicities (stricture, hematuria) were also similar for the two techniques, with very low frequencies in both groups which is consistent with the MSKCC experience, in which adjuvant/salvage 3D RT was associated with 5% grade ≥2 toxicity [38].

Conclusion
Salvage RT is an effective treatment after radical prostatectomy. bDFS 3-and 5-years after salvage RT were 56% and 41%, respectively. RT was well tolerated in terms of urinary toxicity, especially with 3D RT. PreRT PSA was the most powerful prognostic of bDFS before RT delivery and surpassed all other factors evaluated. To increase its efficacy, RT should be given earlier after biochemical relapse, ideally when preRT PSA <1 ng/ml, to obtain the best biochemical control.

Competing interests
The author(s) declare that they have no competing interests.

Authors' contributions
LQ wrote the manuscript, collected and interpreted the data, performed the analysis and reviewed the literature. PMA performed the surgical treatment and revised the manuscript.
VR performed the surgical treatment and revised the manuscript. CM performed the radiotherapy treatment and revised the manuscript. FD performed the surgical treatment and revised the manuscript. CH was involved in the writing of the manuscript, performed the statistical analysis and revised the manuscript. All authors read and approved the final manuscript.