About 20–40% of all localized prostate cancer patients undergoing radical prostatectomy (RP) develop biochemical recurrence (1, 2). In the absence of overt distant metastasis, salvage radiotherapy (SRT) to the prostatic bed is commonly offered as a potential curative treatment for patients with biochemical or manifested local recurrence (LR).

Several studies have shown that SRT efficacy is highly dependent on the pre-salvage prostate-specific antigen (PSA) level (3–5). However, many urologists including at our institution are reluctant to offer early SRT in the absence of radiologically confirmed local disease. Overall, the two main arguments favoring this approach include lack of conclusive overall survival benefits from prospective clinical trials and a potential increased risk of worsened functional outcomes (6).

The purpose of the current study was to evaluate oncologic outcomes and toxicity rates in a cohort of patients treated with delayed dose-escalated image-guided SRT for macroscopic LR after RP.

Our work reports on a cohort of 69 patients homogeneously treated with dose escalated image-guided SRT due to macroscopic LR and a median pre-SRT PSA level of 2.7 ng/ml. At a median follow-up of 38 months, the 3- and 5-year bRFS were 58 and 44%, respectively. The 3- and 5-year cPFS were 91 and 76%, respectively. In this population, pathologic tumor stage and tumor size (GTV) were independent predictors of bRFS, but not of cPFS.

For high-risk prostate cancer patients, LR after RP is a predominant pattern of treatment failure (7). For patients with isolated LR, SRT represents a potentially curative therapeutic option. Indeed, many clinicians favor SRT over upfront adjuvant treatment to allow for postoperative recovery (8) and to avoid possible over-treatment as about 40–50% of the patients can achieve long-term stable disease without further treatment (9, 10). Against this rationale, the optimal timing for SRT and its impact on oncologic outcomes and health-related quality of life is often a matter of debate. Several large observational studies have reported better outcomes when SRT is administrated at a PSA ≤ 0.5 ng/mL (3, 11–14). Consistent with our findings, a large systemic review including 41 retrospective studies reporting on SRT following RP showed that pre-RT PSA is significantly associated with bRFS rates (15). The same work shows an average of 2.6% loss of bRFS for each incremental 0.1 ng/ml PSA increase at the time of SRT. When assessing studies using early salvage as defined by a pre-SRT PSA level of ≤ 0.5 ng/mL, the 5-year bRFS probability is 71% (range: 48–81.8%), which is much higher than the 44% probability reported in the current study (11).

Further, several large studies previously demonstrated the impact of SRT timing on metastasis-free survival. In a retrospective study including 1,106 patients, Stish et al. reported a significant improvement in metastasis-free survival and prostate cancer specific survival when post-RP patients with BCR received SRT at PSA levels ≤ 0.5 ng/mL (16). These findings again argue against prolonged monitoring of detectable post-RP PSA levels that delay initiation of SRT. Some groups have advocated SRT at very early stages, when pre-SRT PSA level ≤ 0.2 ng/mL (17). Although controversy persists, current international guidelines recommend the initiation of SRT at low serum PSA levels (5, 18).

A point of important debate is whether the superior outcomes related to earlier SRT are related to lead time bias (19). Important retrospective series cannot properly rule out this potential bias. Thus, caution should be taken when interpreting most of the results. Nevertheless, a recent analysis presented by Agrawal et al. demonstrated that prostate cancer–specific mortality and all-cause mortality calculated from the time of RP were indeed worse for patients treated with SRT at higher PSA levels (20) These findings corroborated those of Stish et al. (16).

The optimal SRT dose in the setting of macroscopic LR remains unknown. In the absence of macroscopic disease, most guidelines recommend a dose of 64–66 Gy for adjuvant or SRT (5). Retrospective data clearly support dose-escalation to the prostate bed, however the only prospective study assessing dose escalation in the postoperative setting (64 Gy vs. 70 Gy), in the absence of detectable LR, is the SAKK 09/10 trial. This study recently showed low rates of acute grade 2 and 3 GU and GI toxicity with a minor impact in urinary quality of life (21). The ongoing MAPS trial (NCT01411345) randomizes patients with macroscopic recurrence to 68 Gy in 34 fractions vs. 68 Gy in 34 fractions plus boost to MRI-guided macroscopic disease to 76.5 Gy (2.25 Gy per fraction), equivalent to 80 Gy in 2.0 Gy fractions. In our series, 72–74 Gy was used in an unfavorable group of patients with macroscopic LR (mostly biopsy proven) at a median pre-SRT PSA level of 2.7 ng/ml. We found a low toxicity profile in this series with the use of 72-74Gy; however, longer follow-up is required. Although some experiences have shown acceptable toxicity rates with doses up to 80 Gy in the postoperative setting (22), severe GI and GU toxicity may occur with increased RT doses at a longer follow up (23).

Although direct comparisons between studies are challenging, tumor control rates of the present cohort are inferior to prior published series that used SRT at an earlier time-point (Table 4). This is corroborated by results of the phase 3 GETUG-AFU 16 study (29), in which patients with rising post-RP PSA of 0.2–2.0 ng/ml were randomized to SRT vs. SRT plus short term ADT. Patients who received SRT plus short term ADT had a 5-year bRFS of 80% (vs. 44% in our cohort).

Within limitations of retrospective series along with selection bias, our data suggest that offering early SRT to patients with biochemical relapse at low PSA levels, i.e., before the LR becomes manifest, is essentially preferable. Nevertheless, at low PSA levels, conventional imaging modalities are unable to differentiate loco-regional vs. systemic recurrence. Therefore, prostate bed irradiation may expose patients to unnecessary treatment. One could argue that image-guided delayed SRT could spare many patients from potential SRT-related side effects and quality of life changes. Pre-SRT PSA level was not an independent predictor of cPFS in this cohort. In addition, the vast majority of patients who received delayed SRT survived without clinical progression; albeit at a relatively short follow-up in a group of patients with a long-life expectation (median age at RP was 63 years).

The use of new imaging modalities has been rapidly growing and changing clinical management despite lack of prospective data. Emmet et al. recently showed that PSMA PET is independently predictive of treatment response to radiation. A negative PSMA-PET was a strong predictor of a high response to radiation to the prostate bed (30), also arguing against delayed treatment. In our opinion, delayed SRT should be an option only for unfit patients with limited life expectancy.

The present study has obvious limitations. Although this cohort included a group of patients who were homogeneously treated, it has a relatively small sample. Also, staging procedures were not homogeneously performed (e.g., not all patients had PET-CT scans); thus, some patients may have had regional or systemic disease at the time of SRT. Lastly, 40% of the patients did not have a confirmatory biopsy, hence we relied on the accuracy of multiparametric MRI for the detection and treatment of the LR. Furthermore, a longer follow up is certainly required for more conclusive statements on the impact of delayed SRT in cPFS, distant metastasis and prostate cancer-specific survival.

Taken together, a better stratification of patients who develop biochemical recurrence with or without macroscopic LR after RP is essential for more tailored approaches. Several interesting tools have been studied including patient-specific molecular signatures that could be used to identify the optimal radiotherapy dose (31), radiotherapy timing (32), the efficacy of post-operative radiotherapy (33), and also a promising combination of quantitative multiparametric MRI data with tissue-based gene expression (34).

Delayed image-guided SRT to macroscopic LR after RP led to 44% bRFS and 76% cPFS at 5 years, suggesting poorer tumor control compared to early SRT. Implementation of novel stratification tools is of utmost importance to advance clinical decision making in the postoperative setting.

MS contributed to study conception, clinical analysis, statistical analysis, and manuscript preparation. SO contributed to clinical annotation, clinical analysis, statistical analysis, and manuscript preparation. BB contributed to clinical analysis. DN contributed to clinical analysis and manuscript preparation. GT contributed to manuscript preparation. DA contributed to study conception and manuscript preparation. AD contributed to study conception, clinical analysis and manuscript preparation.