Salvage radiotherapy for biochemical recurrence after radical prostatectomy


Akito Terai, Kurashiki Central Hospital, Urology, Kurashiki, Japan.



To evaluate the clinical outcome of salvage radiotherapy (RT) for biochemical recurrence after radical prostatectomy (RP) at our institution.


Between March 1999 and January 2004, 37 patients had salvage RT for prostate-specific antigen (PSA) failure after RP, including eight who had had neoadjuvant hormone therapy. After surgery, PSA was measured with ultrasensitive immunoassays. In all patients RT was delivered to the prostatic bed at a total dose of 60 Gy with a four-field box technique.


The median (range) PSA level before salvage RT was 0.146 (0.06–3.216) ng/mL and RT was started at a PSA level of <0.5 ng/mL in 34 of the 37 patients (92%). With a median follow-up of 31.9 (0–69.8), months, 11 patients (30%) had disease progression after RT and the 3- and 5-year progression-free probability was 74% and 54%, respectively. Univariate analysis showed that clinical and pathological tumour stages and PSA level before RT (>0.15 vs ≤ 0.15 ng/mL) were significant predictors of disease progression. There were no late adverse events related to RT.


Salvage RT for biochemical failure after RP at a low PSA level, using ultrasensitive immunoassays for monitoring, is a reasonably effective treatment. A relatively low radiation dose (60 Gy) seems to be effective.


radical prostatectomy




PSA doubling time


clinical target volume


American Society of Therapeutic Radiology and Oncology.


Radical prostatectomy (RP) is an effective and widely used treatment for clinically localized prostate cancer, but biochemical failure after RP occurs in 15–30% of patients within 10 years of surgery [1,2]. Although the site of the recurrence (local or distant) is critical for determining the best treatment, clinical evaluation can rarely differentiate between local and distant disease [3]. Salvage radiotherapy (RT) to the prostatic bed has been reported to have varying efficacy in treating presumed local recurrence of prostate cancer [4–18], but recent reports suggest that early salvage RT is better than delayed RT [4–19]. In the present study, we reviewed our institutional experience with salvage RT for biochemical failure after RP, as we began to use highly sensitive PSA assays for postoperative monitoring and an approach towards early salvage RT.


Between March 1999 and January 2004, 37 patients who had RP with pelvic lymphadenectomy for prostate cancer were treated with salvage RT because of PSA recurrence after RP. All patients underwent RP and RT at our institution. The patient and tumour characteristics are summarized in Table 1. Eight patients had neoadjuvant androgen-deprivation therapy before RP (two for 3 months and six for 6 months), but no patient received any adjuvant therapy after RP. Tumours were staged according to the 2002 TNM staging system [20]. The serum PSA level was routinely measured using the IMx (Dainabot, Tokyo, Japan; lower detection limit 0.1 ng/mL) until May 1999 and the ARCHITECT (Dainabot; lower detection limit 0.001 ng/mL) from June 1999. From August 1998 to April 1999, the Immulite third-generation PSA assay (Diagnostic Products Corp, Los Angeles, California; lower detection limit 0.003 ng/mL) was used to measure PSA levels of <0.1 ng/mL. In addition, selected frozen serum samples before August 1998 were examined later using either the ARCHITECT or Immulite third-generation PSA assays. The results of these assays were not interconverted. The PSA doubling time (PSADT) was calculated for each patient using all PSA values from the postoperative PSA nadir to the start of salvage RT. The PSADT was estimated as 0.693 (logn 2) divided by the slope from the linear regression of the natural log of PSA levels vs time of PSA measurement in months. Biochemical failure after RP was defined as a PSA level of ≥ 0.1 ng/mL at two consecutive visits, but the selection of time and referral for salvage RT were at the discretion of the treating urologist. For patients who had RT when the PSA level was ≤ 0.1 ng/mL, the disease-free interval was regarded as the time between RP and the start of RT.

Table 1. 
Clinical and pathological characteristics of patients treated with salvage RT
Median (range):
 Age at RP, years: 65 (53–77)
 Pretreatment PSA, ng/mL 9.37 (4.58–97.21)
Neoadjuvant androgen deprivation   therapy, n (%) 8 (22)
Biopsy Gleason score, n (%)
 ≤ 626 (70)
 ≥ 7 11 (30)
Clinical tumour stage, n (%)
 T1c28 (76)
 T2a–T2c 4 (11)
 T3a–T3b 5 (14)
RP Gleason score, n (%)
 ≤ 620 (54)
 ≥ 717 (46)
Pathological tmour stage, n (%)
 pT2a–pT2c26 (70)
 pT3a–pT3b 11 (30)
Positive surgical margin, n (%)13 (35)
Median (range):
Nadir PSA after RP, ng/mL  0.020 (0.001–1.14)
PSA level before RT, ng/mL 0.146 (0.060–3.22)
PSADT after PSA nadir, months 4. 8 (1.1–18.7)
Disease-free interval before RT, months 11.8 (0–58)
Time to RT after RP, months 11.9 (3–58.6)
Persistent PSA ≥ 0.1 ng/mL after RP, n (%) 3 (8.1)

Salvage RT was delivered with a four-field box technique using a 10-MV X-ray with the patient supine. A total dose of 60 Gy was prescribed to the isocentre, at a daily fraction of 2 Gy for all patients. Treatment was planned with a CT-based simulator. The clinical target volume (CTV) included the bladder neck, urethrovesical anastomosis and urethra (mean 9.2 mL). The treatment fields were shaped by a 2-cm multileaf collimator, with a 1.5-cm margin around the CTV until 40 Gy, and thereafter reduced to the CTV + 0.9 cm.

After RT, patients were monitored for disease recurrence at regular intervals, usually every 3 months, with measurements of serum PSA level. Progression of PSA after RT was defined as a PSA value of ≥ 0.1 ng/mL confirmed by a second PSA measurement that was higher than the first, or as a continuous increase in PSA level after RT (defined as disease progression at 0 months).

The progression-free probability was estimated using the Kaplan-Meier method and the log-rank test was used to analyse differences between patient subgroups categorized by potentially prognostic variables, with P < 0.05 considered to indicate statistical significance. All variables were dichotomized using predetermined threshold values identified from the literature or median values in our patients (nadir PSA, PSA level before RT, and PSADT).


At the start of salvage RT, the median (range) PSA level was 0.146 (0.060–3.216) ng/mL. Nine patients had RT after three or more consecutive increases in serum PSA level at <0.1 ng/mL PSA, 17 at 0.1–0.2 ng/mL, and eight at 0.2–0.5 ng/mL. Thus, in 34 of 37 patients (92%), RT was started at ≤ 0.5 ng/mL PSA. The median delay between biochemical failure and starting RT was 2.4 (0–23.8) months. The median PSADT was 4.8 (1.1–18.7) months, and 32 patients (86%) had a PSADT of <10 months. Although three patients discontinued RT for several days because of proctitis, cholecystitis and pneumonia, all patients finally received the total radiation dose. Cholecystitis and pneumonia were adverse events unrelated to RT.

Over a median (mean, range) follow-up of 31.9 (34.3, 0–69.8) months after salvage RT, there was disease progression in 11 patients (30%). Three patients had a continuous increase in PSA level after treatment (progression at time 0) but 34 had a median PSA nadir after RT of 0.012 (0.001–0.108) ng/mL, after a median of 21.8 (2.5–68.8) months. Although the timing of additional therapy after disease progression was at the discretion of the treating urologist, two of nine patients started hormone therapy before clinical progression, and one died from prostate cancer 54.3 months after RT. None of the patients had late adverse events related to RT, such as rectal bleeding secondary to radiation proctitis

The 3- and 5-year progression-free probability after salvage RT was 74% and 54%, respectively (Fig. 1A). Of the patients who had disease progression, the median time to progression was 26.6 (0–48) months. Certain variables before RT were analysed for prognostic significance (Table 2). On univariate analysis, clinical and pathological tumour stages and PSA level before RT (Fig. 1B) were significant predictors of disease progression after RT, while surgical margin status showed marginal significance (P = 0.052).

Figure 1.

Disease progression-free probability after salvage RT, A, for all 37 patients and B, stratified by PSA level before RT; ≤ 0.15 ng/mL (21 patients, green) vs > 0.15 ng/mL (16 patients, red).

Table 2. 
Differences in progression-free probability between patient subgroups categorized by potentially prognostic variables
VariableCategorization (n)Log-rank test P
Pretreatment PSA, ng/mL>10 (16) vs ≤ 10 (21)0.781
Neoadjuvant therapyyes (8) vs no (29)0.598
Biopsy Gleason score≥ 7 (11) vs ≤ 6 (26)0.665
Clinical T stagecT3 (5) vs cT1–2 (32)0.037
Prostatectomy Gleason score≥ 7 (17) vs ≤ 6 (20)0.551
Pathological T stagepT3 (11) vs pT2 (26)0.044
Surgical marginpositive (13) vs negative (24)0.052
Nadir PSA after RP, ng/mL>0.020 (18) vs ≤ 0.020 (19)0.908
Pre-RT PSA, ng/mL>0.150 (16) vs ≤ 0.150 (21)0.042
PSA doubling time after PSA nadir, months<5 (19) vs ≥ 5 (18)0.108
Disease-free interval before RT, months<12 (20) vs ≥ 12 (17)0.214


Many investigators have reported the efficacy of salvage RT to the prostatic bed for PSA recurrence after RP [4–19]. Despite conflicting results, the PSA level before RT appears to be the most important factor for predicting the outcome of RT. The American Society of Therapeutic Radiology and Oncology (ASTRO) recently published a consensus panel report advising that early treatment (PSA < 1.5 ng/mL and ≥ 64 Gy) is more likely to be successful than later treatment [21]. The largest retrospective study to date showed that patients with PSA levels of ≤ 0.6 ng/mL had a significantly better prognosis than those with higher PSA levels, after adjusting for other significant prognostic variables [19].

Table 3 shows the most recent studies reporting the results of salvage RT started at a median PSA level of <1.5 ng/mL [4,7–9,12–17]. In these studies, the 5-year recurrence-free rates achieved by RT alone are 16–46%. The reasons for this wide variation include heterogeneous patient populations and different definitions of PSA failure after RT.

Table 3. 
Published results of salvage RT for biochemical failure, for reports with a median PSA level before RT of ≤ 1.5 ng/mL
ReferenceNMedian PSA level before RT, ng/mLMedian radiation dose, GyMedian follow-up, monthsProgression-free rate, %Endpoint, years
  • *

    Mean value

Garg et al. [4]781.2662568–783
Pisansky et al.[7]1660.96452465
Anscher et al.[8]891.46648504
Vanuytsel et al.[9]530.86633*463
Song et al.[12]610.866.636394
Choo et al.[13]620.8–1.26040–5126–394
Chawla et al.[14]541.364.845355
MacDonald et al.[15]600.6964.851455
Katz et al.[16]700.7566.642394
Liauw et al.[17]510.42565.746563
Present study370.146603274 543 5

Although the present report has shortcomings, including relatively few patients (37) and a short follow-up (median 31.9 months), the salient feature is that, by using highly sensitive PSA assays after RP, most patients underwent early salvage RT (<0.5 ng/mL PSA before RT). It has been reported that ultrasensitive PSA assays that detect 0.001–0.1 ng/mL PSA can reveal biochemical relapse after RP even 18 months earlier than the conventional assay [22,23]. The present 3- and 5-year progression-free probabilities of 74% and 54%, respectively, are similar to those in other reports (Table 3), but the present study is characterized by the lowest PSA level before RT, and by the lowest radiation dose. We used a total radiation dose of 60 Gy mainly because of safety limitations imposed on the RT device (four-field box nonconformal technique). Although this dose is lower than recommended by the ASTRO and recent reports, our strategy might be consistent with an oncological principle that a lower dose is required to treat the fewer tumour cells left behind. Since February 2005 our institution has had a new linear accelerator, and we started three-dimensional conformal RT with a radiation dose of ≥ 64.8 Gy. However, we plan to maintain our strategy of using salvage RT at the earliest time of evident PSA recurrence (e.g. 0.1 ng/mL PSA).

The definition of recurrence after RP has generally relied on a single elevated PSA level, and various PSA thresholds have been used, including >0.1, >0.2, >0.4 and >0.5 ng/mL. Freedland et al.[24] recently showed that a PSA threshold of >0.2 ng/mL was appropriate to define PSA recurrence, as such patients had a 100% 3-year risk of PSA progression. However, by ultrasensitive PSA analysis it was possible to identify rapidly relapsing patients whose PSA levels were >0.1 ng/mL within 4 years of surgery [22]. In the present study, the PSA level changes after RP (Fig. 2) in most patients were classified as being a fast relapse, with a median PSADT of 4.8 months. In nine patients, salvage RT was started at <0.1 ng/mL PSA. The PSADT was not significantly different between these nine patients and 25 with 0.1–0.5 ng/mL PSA before RT (median PSADT 2.5 vs 5.1 months, P = 0.545, Mann–Whitney U-test). Among these nine patients, the median nadir PSA values after RP was 0.013 (0.01–0.05) ng/mL and after salvage RT was 0.006 (0.001–0.03) ng/mL, respectively, suggesting that the rise in PSA level represented tumour persistence and not re-growth of incompletely resected benign prostate tissue. We therefore consider that a threshold of >0.1 ng/mL PSA in the context of a rising PSA level by ultrasensitive assays is appropriate to recommend referral for salvage RT.

Figure 2.

Changes of serum PSA levels over time, excluding three patients with persistent PSA 3 0.1 ng/mL after RP. PSA values between post-RP PSA nadir and starting salvage RT were plotted. The solid and broken lines show nine patients who had salvage RT at <0.1 ng/mL PSA, and 25 who had RT at 0.1-0.5 ng/mL PSA.

Various prognostic factors for salvage RT have been suggested, including PSA level before RT, Gleason score or grade, stage, surgical margin status, seminal vesicle invasion, PSADT, and total radiation dose [4–19]. The present study also showed clinical and pathological stages, PSA level before RT, and surgical margin status to be significant prognostic factors. However, compared to other reports the present patients were relatively free from unfavourable pathological features such as a RP Gleason score of ≥ 8 (two men) or seminal vesicle invasion (one). Notably, the PSA level before RT was a significant prognostic variable, supporting the ‘earlier the better’ approach to salvage RT [25] even in low PSA ranges (<0.5 ng/mL in 92% of patients). Inevitably there is some bias in our patient selection for salvage RT. We began to use the highly sensitive ARCHITECT PSA immunoassay in June 1999; between then and December 2003, 124 patients underwent RP in our institution without adjuvant hormonal therapy and had ≥ 12 months of follow-up. To date, PSA failure has developed in 30 men (24%); salvage RT was used in 20, four had hormonal therapy and six were treated by watchful waiting. Furthermore, four more patients had salvage RT when their PSA level was still <0.1 ng/mL. Recently, adjuvant RT after RP has been reported to improve the biochemical outcome for patients with poor pathological features, e.g. extracapsular extension, seminal vesicle invasion, and positive surgical margins [26]. However, we have no experience with adjuvant RT in our institution. Although a definitive conclusion cannot be drawn because of the few patients and short follow-up, our preliminary findings are promising and are similar to those in other reports. Further data with a longer follow-up are needed to confirm the validity of our treatment strategy.


None declared.