Impact of advanced radiotherapy techniques and dose intensification on toxicity of salvage radiotherapy after radical prostatectomy

The safety and efficacy of dose-escalated radiotherapy with intensity-modulated radiotherapy (IMRT) and image-guided radiotherapy (IGRT) remain unclear in salvage radiotherapy (SRT) after radical prostatectomy. We examined the impact of these advanced radiotherapy techniques and dose intensification on the toxicity of SRT. This multi-institutional retrospective study included 421 patients who underwent SRT at the median dose of 66 Gy in 2-Gy fractions. IMRT and IGRT were used for 225 (53%) and 321 (76%) patients, respectively. At the median follow-up of 50 months, the cumulative incidence of late grade 2 or higher gastrointestinal (GI) and genitourinary (GU) toxicities was 4.8% and 24%, respectively. Multivariate analysis revealed that the non-use of either IMRT or IGRT, or both (hazard ratio [HR] 3.1, 95% confidence interval [CI] 1.8–5.4, p < 0.001) and use of whole-pelvic radiotherapy (HR 7.6, CI 1.0–56, p = 0.048) were associated with late GI toxicity, whereas a higher dose ≥68 Gy was the only factor associated with GU toxicities (HR 3.1, CI 1.3–7.4, p = 0.012). This study suggested that the incidence of GI toxicities can be reduced by IMRT and IGRT in SRT, whereas dose intensification may increase GU toxicity even with these advanced techniques.

The dose-volume data of the rectum and bladder were analyzed, focusing on 50 patients treated at Nagoya City University Hospital, since the data were not available in most of the other institutions. The whole rectum and bladder were delineated. Among these 50 patients, 28 and 22 patients were treated by 3DCRT without IGRT and IG-IMRT, respectively. WPRT was not administered to these 50 patients.
The use of androgen deprivation therapy (ADT) combined with SRT and its duration was at the discretion of each urologist. A luteinizing hormone releasing hormone (LHRH) analog and/or anti-androgen therapy (i.e. bicalutamide) was administered to 64 patients (15%) as neoadjuvant, concurrent, and/or adjuvant ADT. Statistical analysis. The primary endpoint was the incidence of late gastrointestinal (GI) and genitourinary (GU) toxicity, as defined by National Cancer Institute Common Terminology Criteria for Adverse Events version 4.0 16 . Toxicities were always checked at each follow-up prospectively. A follow-up interval was 3 months in principle and was extended to every 4-6 months at 3-5 years after the start date of SRT when there was no complication with the maintenance of a low PSA level. The follow-up time was calculated from the start date of SRT. RT techniques in each era were compared by the Fisher's exact test. SRT doses and RT techniques in each era were compared by the Student's t-test. The distributions of toxicity were calculated by the cumulative incidence method. The significance of these outcomes was examined by the Gray's test. Fine-Gray proportional hazards models were used in the multivariate analysis. The following variables were included in the multivariate analysis as covariates: age at SRT, time between RP and SRT, SRT dose, fraction dose, use of WPRT, and use of ADT. As for advanced RT techniques, the factor of 3DCRT without IGRT vs IG-3DCRT vs IG-IMRT was used as the covariate in the multivariate analysis. All statistical analyses were carried out using EZR 17   Dosimetric analysis. Table 2 shows the dose-volume data of the rectum and bladder in 50 patients treated at Nagoya City University Hospital. Mean prescribed doses of 3DCRT and IG-IMRT were 63.6 and 68.3 Gy, respectively (p < 0.001). Figure 1a,b show the comparison of the average dose-volume histograms (DVHs) of the rectum and bladder between 3DCRT and IG-IMRT, respectively. IMRT reduced the rectum volume irradiated at 20-60 Gy (V20-60). However, the doses to 1% and 2% volume (D1 and D2) of the rectum were higher in the IG-IMRT group than the 3DCRT group because of the higher prescription in the IG-IMRT group. V20-60 of the bladder also tended to be lower in IG-IMRT than 3DCRT, but these differences were not significant. The high-dose areas (D1, D2, and V60) were higher in the IG-IMRT group than the 3DCRT group. www.nature.com/scientificreports www.nature.com/scientificreports/ GI and GU toxicity. The median (range) follow-up was 50 (4-162) months. Late grade 1, 2, and 3 GI toxicities were observed in 37 (8.8%), 6 (1.4%), and 10 patients (2.4%), respectively. Ten (2.4%) patients received argon plasma coagulation (APC) for hemorrhagic radiation proctitis. No grade 4 or higher GI toxicities were observed. The symptoms of GI toxicities are shown in Table 3. The cumulative incidence of late grade 2 or higher GI toxicities was 4.8% (95% confidence interval [CI], 2.3-7.1, Fig. 2a). Late grade 1, 2, 3, 4, and 5 GU toxicities were observed in 46 (11%), 26 (6.2%), 12 (2.9%), 1 (0.2%), and 2 (0.5%) patients, respectively. Eight patients (1.9%) received APC for hemorrhagic radiation cystis. The symptoms of GU toxicities are shown in Table 3. The cumulative incidence of late grade 2 or higher GU toxicities was 24% (95% CI, 5.6-39, Fig. 2a).
The results of the multivariate analysis predicting GI and GU toxicity after SRT are shown in Table 4. Non-use of advanced RT techniques (hazard ratio [HR] 3.1, 95% CI 1.8-5.4, p < 0.001) and use of WPRT (HR 7.6, CI 1.0-56, p = 0.048) were significantly associated with late grade 2 or higher GI toxicities. Adjusted HRs of non-use of IMRT and non-use of IGRT for GI toxicity were 4.5 (CI 1.7-12, p = 0.003) and 9.3 (CI 3.9-22, p < 0.001), respectively. The comparison of the cumulative incidence of late grade 2 or higher GI toxicities between the 3DCRT and IMRT groups (p = 0.16) and between the non-IGRT and IGRT groups (p < 0.001) is shown in Fig. 2b,c, respectively. Based on the multivariate analysis predicting GU toxicity, a higher SRT dose ≥68 Gy was the only factor associated with GU toxicities (HR 3.1, CI 1.3-7.4, p = 0.012). The comparison of the cumulative incidence of late grade 2 or higher GU toxicities between the dose < 68 Gy and the dose ≥68 Gy groups (p < 0.001) is shown in Fig. 2d.

Discussion
This large multi-institutional retrospective study showed that the prescribed dose increased significantly in SRT after RP, from 61.   www.nature.com/scientificreports www.nature.com/scientificreports/ The cumulative incidence of late ≥grade 2 GI complications was only 4.8% in our current study, whereas that after definitive RT ranged from 7% to 15% [19][20][21] . This may be because a lower dose of 64-70 Gy is usually used in SRT as compared with a dose of 76-80 Gy in definitive RT. Despite the low incidence of GI toxicity after SRT, additionally, IMRT could decrease GI toxicity compared with 3DCRT, as shown in Fig. 2b. This finding was in line with the results of a retrospective study in which the 5-year rate of ≥grade 2 GI toxicities was 1.9% in IMRT vs 10.2% in 3DCRT in SRT 22 . Reduction of GI complications with IMRT is expected because dosimetric studies showed that IMRT could improve the DVHs of the rectum even in SRT 6,23 .
Our current study showed that IGRT also contributed to the reduction of GI toxicity in SRT. Considering the results shown in Fig. 2b,c, IGRT may have a higher impact on the reduction of GI toxicity than IMRT. This may be partly because physicians might reduce the margin of the target volume when IGRT was used. In definitive RT, the combined use of IGRT may affect the DVHs difference as compared with the non-IGRT treatment 24 . The most significant reason may be simply the correction of the rectum position against intrafractional changes in rectal and bladder filling and digestion gas. The prostate bed is situated precisely between the bladder and rectum. In the case of SRT, the bladder and rectum lie next to each other because of the absence of the prostate. In SRT, intrafraction variations of the rectum position can affect the irradiated volume of the rectum directly more so than in definitive RT. IGRT and IMRT are often used together and have been reported to prevent GI and GU complications in post-prostatectomy RT 25 . On the other hand, in the national population-based study of RT after RP, the use of IMRT could not improve rates of severe GI and GU toxicity 26 . The authors considered one of the reasons for this finding to be the missing of information on the use of IGRT.
The cumulative incidence of late ≥grade 2 GU toxicities was 24% and a few patients developed severe (i.e. ≥grade 3) GU toxicities. Our current study showed that a SRT dose ≥68 Gy was an only significant factor www.nature.com/scientificreports www.nature.com/scientificreports/ associated with GU toxicities. Thus, this large multi-institutional retrospective study suggested that dose intensification might increase GU toxicity in SRT even with the advanced RT techniques. The incidences of ≥grade 2 GU toxicities varied among the reports of post-prostatectomy RT, and were between 10% and 34% 12,25,27,28 . In definitive RT, the volume of the bladder receiving high doses was a risk factor of GU toxicity 29 . The reason for the increase of GU toxicity with dose intensification may be the same as that for definitive RT. De Meerleer et al. 12 reported that when 75 Gy in 37 fractions was prescribed in IMRT, ≥grade 2 GU toxicities were observed in 34% (23/68) patients and urethral stricture was observed in 7.4% (5/68). Anastomotic stricture is a common complication after RP, with a reported incidence of approximately 5-10% 30 . A higher SRT dose may affect the development of anastomotic stricture. The target volume of SRT necessarily encompasses a certain range of the bladder and an anastomosis site, which is a favored site of recurrence 5 . Therefore, the irradiated volume of an anastomosis site cannot be reduced by IMRT or IGRT. Although the GU toxicity of urinary tract obstruction was observed in only 7 patients (1.7%) in our current study, mild toxicities, such as grade 1 toxicities, were unable to be recorded. Additionally, the GU toxicities, such as urinary frequency and urinary tract infection, may have been due to anastomotic stricture. In fact, Table 2 and Fig. 1b suggested that even IMRT might not control the high-dose area of the bladder (i.e., doses around 65 Gy) in dose-escalated RT. Thus, dose intensification may increase GU toxicities even with these advanced techniques. The guideline of the Australian and New Zealand Radiation Oncology Group recommends SRT doses of 66-70 Gy to the prostate bed 31 . We also advocate that physicians must be careful in prescribing SRT doses ≥70 Gy in SRT after RP.
The extra dose to the bladder due to an inappropriate IMRT planning is also of concern. Radiation oncologists usually pay more attention to the DVH of the rectum rather than of the bladder because the definite dose constraint of the bladder is unclear, whereas the recommended dose constraint of the rectum has been reported 32,33 . Several studies have reported conflicting results for the significance of IMRT in managing adverse effects of SRT 23,34,35 . A population-based cohort study related to postoperative RT 34 showed that the IMRT utilization increased from 33% in 2000-2004 to 78% in 2005-2007, and suggested that IMRT is associated with a lower rate of GI toxicities and a higher rate of GU complications than 3DCRT in RT after RP 34 . The authors considered one of the reasons for this finding to be that the increased use of IMRT facilitated the high-dose prescription during the study period. Thus, our finding was in line with the results of this population-based study in which higher doses were associated with GU toxicity. The long-term efficacy and toxicity of dose-escalated SRT should be evaluated in a randomized control trial (RCT) 36 .
Our study presented some relevant issues about SRT after RP. Several retrospective studies reported that elective pelvic lymph node irradiation (i.e. WPRT) improved biochemical control in selected patients 37,38 . With limited available evidence, WPRT is currently an option for patients with adverse features such as pathological nodal metastasis and Gleason score ≥8 30 . Our study suggested that WPRT increases the incidence of GI toxicity (HR 7.6, CI 1.0-56, p = 0.048). This was because WPRT encompasses a larger irradiated volume of the rectum than SRT to the prostate bed alone. Therefore, even when WPRT is performed in SRT, the advanced RT techniques with IGRT and IMRT may enable to escalate dose while constraining GI toxicity. The results of the RTOG 0534 phase 3 trial 39 , the first RCT investigating the efficacy of WPRT and concurrent ADT, are awaited.
No randomized data has shown that the combination of ADT potentiates the adverse effects in SRT 40 , and the same is true in our current study. Although we recognize the efficacy of adding ADT to SRT in consideration of the promising results of the recently published RCT 40 , only 15% patients received ADT combined with SRT in our current study. This was partly because of the lack of evidence for the combination of ADT in the era of our current study. Moreover, the low rate of the ADT combination may be partly due to race differences 41 because most Asian patients with prostate cancer have a better prognosis than whites 42 . Japanese physicians might think that the combination of ADT is not necessarily needed in the setting of SRT.
Of patients who experienced grade 3 or higher GI or GU toxicities, 10 (2.4%) and 8 (1.9%) patients received the endoscopic mucosal cauterization for hemorrhagic radiation proctitis and cystis, respectively. APC can reduce bleeding and blood infusion by cauterizing mucosal telangiectasias 43 , and remedied bleeding due to hemorrhagic RT proctitis in most cases 44,45 . Therefore, this effective treatment is recommended for moderate and severe RT proctitis.  Table 4. Multivariate analyses evaluating late grade 2 or higher toxicities in 421 patients after salvage radiotherapy. HR hazard ratio, 95% CI 95% confidence interval, SRT salvage radiotherapy, RP radical prostatectomy, EQD2 equivalent dose in 2-Gy fractions, IMRT intensity-modulated radiation therapy, IGRT image-guided radiation therapy, WPRT whole-pelvic radiotherapy, ADT androgen deprivation therapy. *Advanced RT techniques means IMRT or IGRT or both. (2020) 10:114 | https://doi.org/10.1038/s41598-019-57056-9 www.nature.com/scientificreports www.nature.com/scientificreports/ This large multi-institutional retrospective study had some limitations. First, the follow-up time was relatively limited. On the other hand, late GI and GU toxicities occasionally occurred after a median follow-up of 50 months, as shown in Fig. 2a. Additionally, the follow-up time was different between the advanced RT and conventional RT groups, and the cumulative incidence of toxicity may have become higher because of the longer follow-up time in the conventional RT group. Second, the lack of absolute information on the DVHs was previously mentioned. Despite these limitations, the findings of our study suggested the impact of advanced RT techniques and dose intensification on the toxicity of SRT after RP.
In conclusion, this study demonstrated that the SRT dose increased significantly with increasing use of IMRT and IGRT from 2004 to 2017. Our results suggested that the incidence of GI toxicity can be reduced by IMRT and IGRT for SRT after RP, whereas high-dose SRT may increase the incidence of GU toxicity even with these advanced techniques. These findings will facilitate the design of a clinical trial for SRT after RP.

Data availability
Data sharing is not applicable to this article due to ethical reasons.