Abstract
Background/Aim: This retrospective study aimed to investigate the outcomes of relapse-free survival (RFS) after salvage radiation therapy (SRT) to the prostate bed for postoperative biochemical recurrence of prostate cancer. Patients and Methods: A total of 87 patients were analyzed. There were 27, 32, and 24 patients with pathological grade groups of 1-2, 3, and 4-5, respectively. SRT doses of 64, 66 or 70 Gy were administered to 24, 3 and 60 patients, respectively. The Kaplan–Meier method was used to estimate time-to-event outcomes. The multiple imputations method was used to impute missing values, and Cox proportional-hazards models were applied for multivariate analyses. Results: The median follow-up period for patients overall was 58.6 months. The 5-year RFS rates of the whole cohort was 59.4% and those for pathological grade groups 1-2, 3 and 4-5 were 88.9%, 37.7% and 39.5%, respectively. In multivariate analyses, higher pathological grade group [4-5 vs. 3 vs. 1-2: hazard radio (HR)=8.65, p<0.01], negative surgical resection margin (positive vs. negative: HR=0.41, p=0.02) and higher pre-salvage treatment serum prostate-specific antigen (cutoff value 0.31 ng/ml: HR=3.50, p<0.01) were significantly associated with poorer RFS. The cumulative incidences of grade 2 or more late rectal bleeding and late hematuria were 4.9% and 8.7%, respectively, at 5 years and 4.9% and 15.7%, respectively, at 8 years. These toxicities occurred only in the 70 Gy-treated arm. Conclusion: Our study revealed that pathological grade group 3 prostate cancer patients experienced moderately unfavorable RFS after SRT. Higher radiation doses might increase late toxicities without improving RFS.
Biochemical recurrence (BCR) after radical prostatectomy occurs in approximately 20% of patients (1, 2). The median time to development of distant metastases was 8 years in the natural history after BCR, and prostate cancer-specific mortality was approximately 2-6% (2, 3). At the time of BCR, it is said that there is an equal risk of prostate cancer-specific mortality and death from competing causes (2). Therefore, less invasive treatment for selected patients is needed as secondary therapy for those with BCR. Salvage radiotherapy (SRT) is the only possible curative secondary therapy for BCR after radical prostatectomy. SRT administered within 2 years of BCR was reported to show improvement in prostate cancer-specific survival and overall survival (4). However, patients do not always obtain the benefit of SRT. To identify better candidates for SRT, many factors were investigated, and some prognostic factors were reported: serum prostate-specific antigen (PSA) level before SRT (ng/ml), PSA-doubling time (months), PSA velocity (ng/ml/year), surgical Gleason score, surgical margin status, seminal vesicle involvement, combined treatment with hormone therapy and extended SRT field (5-9). A high SRT dose (70 Gy) to the prostate bed also improved the relapse-free survival (RFS) rate compared with an SRT dose of 60 Gy (9). Because the same tendency was seen in our previous work (10), the SRT dose was escalated to 70 Gy at our Institute. Furthermore, relatively new treatments, such as robotic-assisted prostate cancer surgery and volumetric-modulated arc therapy (VMAT), have also been introduced. Use of the International Society of Urological Pathology grade group classification system has also become more common partly because of the prognostic difference between Gleason score 3+4 and 4+3 (11, 12). This study estimated the prognostic value of dose-escalated SRT, the pathological grade group, and these relatively new treatment modalities.
Patients and Methods
Ethics. This retrospective study was performed at a single institute and was approved by the Ethical Committee of Tohoku University Hospital (reference number: 2021-1-659). Informed consent was waived due to the retrospective study design. Nevertheless, the information about this study was released to the public on the Institute’s website to guarantee the chance to opt out of participation and opt-out consent was obtained. Furthermore, written-informed consent as a part of general consent for utilizing treatment data in future retrospective studies was obtained from all patients treated after April 2016.
Patient selection. Patients who received 60 Gy or more radiotherapy to the prostate bed field between January 2008 and December 2021 at Tohoku University Hospital were identified from our database. Eligible patients needed to meet the following criteria: cN0 and cM0 according to the eighth edition of the Union for International Cancer Control TNM Classification (13) at the time of prostatectomy; pN0 if lymph node dissection was performed; a nadir serum PSA level of 1.0 ng/ml or less after prostatectomy without any additional treatment confirmed BCR, which wars defined as a serum PSA level increased by more than 0.2 ng/ml or consecutively increased at two or more subsequent occasions; no detectable lesions on computed tomography (CT) and magnetic resonance imaging after BCR; and a pre-salvage treatment PSA level of 2.0 ng/ml or less (4, 13, 14). The criteria for administering hormone therapy during SRT varied among urologists, therefore, both SRT concurrent with hormone therapy regardless of agents or duration and SRT without hormone therapy were considered eligible. Patients whose follow-up period was 6 months or less were excluded (Figure 1).
Flow chart of patient selection from the database. PSA: Prostate-specific antigen.
Set-up, contouring, and SRT procedure. Patients were immobilized in the supine position with a full bladder and empty rectum for radiotherapy planning CT and radiotherapy. A CT scan (GE Light Speed Qxi - GE Healthcare, Waukesha, WI, USA; or SOMATOM Definition AS - Iselin, NJ, USA) was performed with a 2.0-2.5 mm slice thickness. The clinical target volume (CTV) was the prostate bed with or without a seminal vesicle bed, and this contouring was based on a prospective trial of the Japan Clinical Oncology Group (JCOG 0401) (15). Prostate bed CTV included the bladder-urethra anastomosis, posterior pubic symphysis, posterior rectal wall, and apex region with the level of the crus of the cavernous body of the penis. The seminal vesicle bed was included in the CTV only for patients with pathological invasion of prostate cancer to the seminal vesicle. The planning target volume was created from the CTV by an expansion of 1 cm to all directions except for 0.7 cm in a posterior direction. Three-dimensional conformal radiotherapy or VMAT with 10 MV or 15 MV photons using a linear accelerator (Clinac 23EX or TrueBeam STx; Varian Medical Systems, Palo Alto, CA, USA) was created using a radiotherapy planning system (Eclipse; Varian Medical Systems). Radiotherapy was applied at 60 Gy to 70 Gy in 2-Gy daily fractions, delivered using daily image guidance. The normal tissue dose constraints were as follows: 10%, 15%, 20%, 30%, and 45% of the rectum received <75 Gy, <70 Gy, <65 Gy, <60 Gy, and <50 Gy, respectively, and 25%, 35% and 50% of the bladder received <75 Gy, <70 Gy and <65 Gy, respectively.
Endpoints. The primary endpoint of this study was RFS after SRT for BCR after prostatectomy. The secondary endpoints of this study were RFS based on the Phoenix criteria (Phoenix-RFS), metastasis-free survival, toxicity, and building a nomogram for 5-year RFS. RFS was defined as the time from the start of SRT to the first day that BCR, initiation of hormone therapy or death was confirmed (6). If the PSA level never dropped below 0.2 ng/ml after SRT, the event date was defined as the last date of SRT. Phoenix-RFS was defined as the time from the start of SRT to the first day that serum PSA was the nadir level plus 2.0 ng/ml, or initiation of hormone therapy or death was confirmed (5). Metastasis-free survival was defined as the time from the start of SRT to the first day that distant metastasis, lymph node metastasis or death was confirmed. Toxicity was judged according to the National Cancer Institute Common Terminology Criteria for Adverse Events version 5.0, translated by the Japan Clinical Oncology Group (16).
Statistical analyses. Statistical analyses were performed using RStudio version 2022.12.0 and EZR version 1.54 (Saitama Medical Center, Jichi Medical University, Saitama, Japan), a modified version of R commander (R Foundation for Statistical Computing, Vienna, Austria) (17).
Time-to-event outcomes were calculated from the first day of SRT to the day that an event was confirmed. The Kaplan–Meier method was used to estimate time-to-event outcomes, and a log-rank test was used to compare Kaplan–Meier curves in the univariate analyses. When the cumulative incidence of toxicity was calculated, death without toxicity was regarded as a competing risk. Continuous covariates were dichotomized using survival receiver operating characteristic (ROC) curves at 5 years of RFS. Fisher’s exact tests were used to compare categorical variables. The possible prognostic factors from the known pre-SRT nomogram and elements of interest, such as the type of surgery or radiotherapy, were tested in a univariate model (6). The factors with a p-value of 0.1 or less and those with a variance inflation factor (VIF) for each factor of 2.0 or less were used for the multivariate analyses. When there were 30 or more events in RFS, 5-9 events per variable in the multivariate analysis model were considered acceptable based on previous findings (18). Multiple imputations using the areg impute function was used to impute missing values, and Cox proportional-hazards models were applied for multivariate analyses. Then, a nomogram for RFS probability at 5 years was created based on the results of multivariate analyses. Internal validation of the nomogram was performed using bootstrapping with 10,000 resamples, and the concordance index (c-index) of optimism-corrected ROC was calculated. Furthermore, a calibration of the nomogram was calculated using 10,000 bootstrap resamples. A p-value less than 0.05 was defined as significant.
Results
A total of 87 patients were identified, and their characteristics at the time of prostatectomy and at the time of SRT are shown in Table I and Table II, respectively. Eight patients were administered concurrent hormone therapy, which consisted of leuprorelin acetate, bicalutamide or both, and the median duration of hormone therapy was 6.1 months [interquartile range (IQR)=3.7-8.1 months]. Regarding SRT doses, no patient received less than 64 Gy, and more than half received 70 Gy. Fifty-nine patients received 3-D conformal radiotherapy, and 28 patients received VMAT. One patient whose planned radiation dose was 70 Gy using the VMAT technique ceased at 66 Gy because of radiation-induced proctitis. The median PSA-doubling time was 6.1 months (IQR=3.6-10.9 months), and the median pre-salvage treatment PSA level was 0.33 ng/ml (IQR=0.22-0.49 ng/ml).
Prostatectomy characteristics.
Characteristics of eligible patients and treatments at the time of radiotherapy.
The median follow-up period after SRT was 58.6 months (IQR=35.6-89.8 months). During follow-up, one patient died of their comorbidity without any evidence of recurrence, and two patients died of prostate cancer. Among patients without concurrent hormone therapy, in 58 out of 79 patients (73%) a PSA value of 0.1 ng/ml or less was achieved after SRT. There were 37 RFS events, the 5-year RFS was 59.4% [95% confidence interval (CI)=47.1-69.8%], and the median RFS was 79.5 months (95% CI=54.8 months-not estimable; Figure 2A). The median interval to relapse was 9.8 months (IQR=1.7-32.7 months). RFS curves for the pathological grade group are also shown (Figure 2B and C). The 5-year RFS rates of grade groups 1-2, 3 and 4-5 were 88.9%, 37.7% and 39.5%, respectively (p<0.01 for groups 1-2 vs. 3 and p=0.03 for groups 3 vs. 4-5), and the 5-year RFS rates for grade groups 2 and 3 were 87.0% and 37.7%, respectively (p=0.01). The prescribed radiation doses for each group did not significantly differ (p=0.94; Table III). The 5-year Phoenix-RFS was 66.2% (95% CI=56.4-77.5%; Figure 2A), but 18 out of 31 patients who showed increasing PSA had started hormone therapy before their serum PSA value reached the Phoenix criteria. Distant or lymph node metastasis occurred in 12 patients, at a median interval of 84.9 months (IQR=29.9-142.1 months), and the 5-year metastasis-free survival was 94.5% (95% CI=85.6-97.9%; Figure 2A). The most frequent sites of first metastasis or recurrence were the bone (n=6), followed by the lymph nodes (n=5) and both the bone and lymph nodes (n-1).
A: Kaplan–Meier curves for relapse-free survival (RFS), RFS based on the Phoenix criteria (Phoenix-RFS), and metastasis-free survival. The 5-year RFS, Phoenix-RFS and metastasis-free survival rates were 59.4%, 66.2% and 94.5%, respectively. B: Kaplan–Meier RFS curves according to pathological grade group: 1-2, 3 and 4-5. The 5-year RFS rates of grade groups 1-2, 3 and 4-5 were 88.9%, 37.7%, and 39.5%, respectively (p<0.01 for groups 1-2 vs. 3, and p=0.03 for groups 3 vs. 4-5). C: Kaplan–Meier RFS curves comparing pathological grade group 2 with group 3. The 5-year RFS rates of grade groups 2 and 3 were 87.0% and 37.7%, respectively (p=0.01).
Distribution of dose administered in salvage radiotherapy (SRT) to patients according to pathological grade group.
In addition to well-known prognostic factors, some exploratory factors were tested using univariate analyses to construct the multivariate prognostic model, but none were finally selected (Table IV). VIF was calculated among the candidates for the multivariate model; then velocity was removed from the model because the VIF between velocity and pre-salvage treatment PSA was more than 2.0. Therefore, the multivariate model was constructed using five factors (7.4 events per variable): serum PSA pre-prostatectomy, pathological grade group, extra-prostatic extension, surgical resection margin status and pre-salvage treatment PSA level. As a result of multivariate analyses, pathological grade groups (three groups: 1-2, 3 and 4-5), surgical resection margins and pre-salvage treatment PSA level (cutoff value: 0.31 ng/ml) showed significance [hazard ratio (HR)=10.93, p<0.01; HR 0.44, p=0.03; and HR 3.50, p<0.01, respectively; Table V]. Finally, a nomogram predicting the 5-year RFS probability was calculated and drawn (Figure 3A). In the internal validation, the c-index of bootstrap optimism-corrected ROC was 0.74; the calibration plot is shown in Figure 3B.
Univariate log-rank tests for recurrence-free survival (RFS).
Multivariate analyses for recurrence-free survival.
A: Nomogram for 5-year relapse-free survival (RFS) after salvage radiotherapy (SRT). B: Calibration plot of the nomogram. The nomogram was validated with 10,000 bootstrap resamples. The difference between the predicted probability from the nomogram and the actual probability was plotted. The black line represents the values for the observed data, the gray line represents the ideal values and dashed line represents the optimally corrected values. EPE: Extra-prostatic extension; PSA: prostate-specific antigen; RM: surgical resection margins.
Grade 2 and 3 acute gastrointestinal toxicities occurred in 47 and one patient, respectively. Grade 2 acute genitourinary toxicities occurred in three patients, and grade 3 or more genitourinary toxicities did not occur. Regarding late toxicities by Common Terminology Criteria for Adverse Events v5.0, grade 2 and 3 rectal bleeding occurred in three and one patient, respectively. Grade 2 and 3 hematuria occurred in five and two patients, respectively. Four patients suffered grade 2 or more urinary retention: one required temporary urinary catheterization, two required bougienage, and one required urethrotomy (grade 3). The cumulative incidence of grade 2 or more late rectal bleeding was 4.9% (95% CI=1.6-11.2%) at 5 and 8 years, and the cumulative incidence of grade 2 or more late hematuria was 8.7% (95% CI=3.1-18.0%) at 5 years and 15.7% (95% CI=6.1-29.4%) at 8 years. Grade 2 or more late rectal bleeding and hematuria occurred only in patients who received 70 Gy (p=0.15 and p=0.01, respectively; Figure 4A and B).
A: Cumulative incidence of grade 2 or more late rectal bleeding (A) and late hematuria (B) after salvage radiotherapy for biochemical recurrence.
Discussion
This study showed the clinical effectiveness of SRT for BCR after radical prostatectomy. Some changes have occurred since our previous report: the introduction of robotic-assisted surgery to prostatectomy, VMAT to SRT, and the concept of pathological grade group classification (10, 12). Among them, the pathological grade group showed significance in multivariate analyses. The RFS curve for pathological grade group 3, which represents a Gleason score of 7 but a Gleason grade pattern of 4+3, was in the middle between grade groups 1-2 and 4-5, and a significant difference was seen between pathological grade group 2 (Gleason grade pattern 3+4) and group 3 (Figure 2B and C). Although prognostic significance after SRT between pathological grade groups 2 and 3 has hardly been discussed, some reports have discussed it outside of SRT. Comparing Gleason grade patterns 3+4 and 4+3, there were differences in progression-free probability in surgical series and overall survival or cancer-specific survival differences in database analyses of prostate cancer (11, 19, 20). The difference in cancer-specific mortality between patients with Gleason scores of 7 or less and 8 or more has also been reported (4). A Gleason score of 8 or more is also one of the well-known unfavorable factors after SRT (21, 22). This study added the significance of moderately unfavorable factors of pathological grade group 3 after SRT. When treating pathological grade group 3 (or higher), other factors, such as surgical resection margins and pre-salvage treatment PSA level, should be considered when performing SRT. If the predicted RFS rate is low, an extended field of SRT (i.e., pelvic lymph node radiotherapy) and concurrent hormone therapy should be considered (5, 8, 23).
Combination treatment with hormone therapy and SRT is also a well-known favorable factor after SRT. Although adding hormone therapy to SRT was not a significant factor in this study, the effectiveness of hormone therapy has already been revealed in some phase III trials (5, 8, 23). One of these trials showed that the efficacy of adding antiandrogen therapy (bicalutamide) was more effective as the PSA level at trial entry (i.e., pre-salvage treatment) was higher (8). Interestingly, the subgroup with the lowest PSA level showed no difference between treatment with and without bicalutamide or showed somewhat worse survival in SRT with bicalutamide. The 12-year overall survival rates of patients treated with SRT with versus without bicalutamide in patients with PSA concentrations greater than 1.5 ng/ml and less than 0.7 ng/ml were 73.5% vs. 48.9% and 76.8% vs. 80.7%, respectively, in that report. In our study, the PSA level of 0.7 ng/ml was exceeded in less than a quarter of patients; therefore, the benefit of combination treatment with hormone therapy was minimized. Furthermore, only eight patients were administered hormone therapy, and the agents and duration were inconsistent. As a result, hormone therapy showed no significance (Table II and Table IV). In addition to these results, some urologists in Japan may opt for SRT monotherapy based on a phase III trial (JCOG 0401) conducted in the country (15). The study revealed that SRT without concurrent hormone therapy resulted in a longer time to bicalutamide failure than salvage bicalutamide alone. It should be noted that other clinical trials have confirmed the effectiveness of adding hormone therapy to SRT, which is the mainstream of SRT (5, 23).
In multivariate analyses for RFS, surgical resection margin status and pre-salvage PSA level also showed significance. The significance of the pre-salvage PSA level has been reported in many studies (6, 24, 25), and was also confirmed in this study. On the other hand, the effect of surgical resection margin status is controversial. The multivariate analyses and nomograms analyzing more than 1,000 patients revealed that a positive surgical margin was a significant favorable factor in RFS, and our results also supported this (6, 24). In contrast to these findings, other reports showed that a negative surgical margin was a significant favorable factor in RFS (23). A recent report that used prostate-specific membrane antigen positron-emission tomography (PSMA-PET) in SRT also showed that a negative surgical margin was a significant favorable factor of freedom from BCR (26). In this manner, there is inconsistency in the literature regarding prognostic factors; therefore, the accuracy of nomograms might be limited (27). In addition, the resultant c-index of our nomogram for RFS was 0.74 in the internal validation, which fell short of perfection (c-index=1.0). In other nomograms for RFS after SRT, the c-index was reported to range from 0.68 to 0.74 in internal validation (6, 24, 25). It is true that nomograms can be a useful clinical decision-making tool, but they are not a perfect tool (25).
The approach in SRT dosing, especially dose escalation, differed from initial prostate cancer radiation therapy. In initial definitive radiotherapy, higher radiation doses contributed to a lower PSA recurrence rate with up to 200 Gy biologically effective dose (28). For SRT, it has been said that biochemical progression-free survival increased by 2.5% per Gy of SRT doses whose median value ranged from 60 to 72 Gy; on the other hand, grade 3 or more late gastrointestinal and GU toxicities also increased by 1.2% per Gy and 0.7% per Gy of SRT doses, respectively, in a review article with the inclusion of retrospective studies (29). A multi-institutional observational study reported that SRT doses of 66 Gy or more, whose IQR ranged from 64.8 to 69.0 Gy, were associated with reduced risk of secondary BCR (30). However, a more recent phase III trial comparing SRT doses of 64 Gy with 70 Gy to the prostate bed without concurrent hormone therapy revealed no significant differences in secondary BCR, clinical progression-free survival, hormone treatment-free rates, or overall survival. In addition to these findings, there was a significant increase in late gastrointestinal toxicity (31). The result was confirmed by our findings that an SRT dose of 70 Gy led to no difference in RFS (Table IV). Furthermore, our study also revealed a significant increase in late hematuria (Figure 4). Because the timing of postoperative radiotherapy (adjuvant or salvage setting) is useless in obtaining an improved therapeutic ratio, one solution will be the introduction of a moderately hypofractionated SRT regimen with the consideration of biologically effective doses to prostate cancer and surrounding organs (32, 33). Another solution may be the introduction of PSMA-PET because PSMA-PET-guided SRT led to a significant reduction in secondary BCR at higher (more than 70 Gy) SRT doses (26).
The definition of relapse after SRT (i.e., secondary BCR) varied in trials: PSA levels of 0.2-0.5 ng/ml with or without consecutive rise or nadir value plus 2.0 ng/ml (5, 6, 8, 15, 23). The Phoenix criteria were used in the Radiation Therapy Oncology Group trial (RTOG0534) because the Phoenix definition was associated with clinical failure, distant metastasis, and power overall survival (34). The specificity, sensitivity, and positive predictive value for clinical failure by various PSA cutoff levels are shown in the protocol: 56%, 95%, and 23%, respectively, for a PSA level of 0.2 ng/ml, and 83%, 91% and 43%, respectively, for the Phoenix criteria (5). Phoenix-RFS was also investigated in this study, but unfortunately, more than half of the patients (58%) started hormone therapy before the PSA level reached their nadir plus 2.0 ng/ml. Although no patients underwent PSMA-PET in this study, introducing this modality to follow-up after SRT or work-up after elevation of PSA might provide a new PSA cutoff level (35).
Study limitations. This study was a retrospective single-institute study; therefore, there was a limitation due to its retrospective nature. The number of patients was relatively small, the multivariate analysis model had relatively few events per variable, and some factors were lacking, partly because some patients underwent prostatectomy outside our Institute. In this study, only eight patients were administered hormone therapy alongside SRT, and the agents and duration of hormone therapy were inconsistent. As a result, despite its well-known prognostic importance, hormone therapy administration did not show any significance in this study. Furthermore, patients with a short follow-up were excluded from the analyses, and the follow-up periods varied.
Conclusion
Five-year RFS of patients treated with SRT for BCR was 59.4% in this study, and pathological grade group 3 demonstrated significantly less favorable RFS compared with grade groups 1-2 and significantly more favorable RFS compared with grade groups 4-5. The RSF difference was significant between pathological grade groups 2 and 3, despite both reflecting a Gleason score of 7. Additionally, positive surgical resection margins were found to be significantly associated with a higher RFS rate in this study, which is a controversial prognostic factor. When aiming to improve RFS, it is important to note that higher radiation doses (70 Gy) significantly increased the risk of late hematuria without improving RFS compared to doses of 64-66 Gy.
Acknowledgements
The Authors are grateful to the radiation oncologists, urologists, medical physicians, and radiation technologists at Tohoku University Hospital who contributed to SRT and patient follow-up and supported data acquisition.
Footnotes
Authors’ Contributions
Conception and study design: Takaya Yamamoto and Rei Umezawa. Data acquisition, data analysis and interpretation: Takaya Yamamoto, Rei Umezawa, Shuichi Shimada, Noriyoshi Takahashi, Kazuya Takeda, Yu Suzuki, Keita Kishida, So Omata, Yuta Sato, Hinako Harada, Akihiro Ito, and Keiichi Jingu. Article draft: Takaya Yamamoto. Article editing and revision: Akihiro Ito and Keiichi Jingu. Final approval of article: All Authors.
Conflicts of Interest
TY has received honoraria for lectures from AstraZeneca KK, Amgen KK and AiRato Inc. KT has received honoraria for lectures from Bayer Yakuhin, Ltd. KJ has received consulting fees from Varian Medical Systems and honoraria from AstraZeneca KK, Varian Medical Systems and Elekta KK. RU, SS, NT, YS, KK, YS, HH and AI have no conflict of interest.
- Received August 22, 2023.
- Revision received September 24, 2023.
- Accepted September 26, 2023.
- Copyright © 2023 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.
