Intensity-modulated radiation therapy from 70Gy to 80Gy in prostate cancer: six- year outcomes and predictors of late toxicity

Objective To report grade ≥2 overall late rectal and urinary toxicities in patients (pts) with prostate cancer treated by intensity-modulated radiotherapy (IMRT) at 3 dose-levels. Identify predictors of radiation toxicity and report biochemical progression free survival (bPFS). Methods A total of 277 pts were treated with 70Gy (10.8%), 74Gy (63.9%) and 80 Gy (25.3%) using IMRT without pelvic irradiation were analyzed. Short or long-course androgen deprivation therapy (ADT) was allowed in 46.1% of pts. The toxicity was described using the Common Terminology Criteria for Adverse Events (CTCAE) v4.0 scale. Cox regression models addressed demographics, disease and dosimetry characteristics as potential predictors of late grade ≥2 toxicity after adjusting for other modifying factors. Results The median follow-up was 77 months (range 15; 150). There was no grade ≥4 toxicity. The 5-year cumulative rate of grade ≥2 late rectal and urinary toxicities was 6.3% (95% CI = 3.8%; 10.3%) and 25.3% (95% CI = 19.8%; 31.8%) respectively. In multivariate analysis, only the dose (80Gy vs 74 and 70Gy) was found to increase the risk of rectal toxicity (HR = 2.96 [1.07; 8.20]). For pts receiving 74 Gy, International Prostate Symptom Score (IPSS) at baseline ≥8 (HR = 2.40 [1.08; 5.35]) and dose ≥73Gy delivered in more than 2% of bladder (D2%) were found to be predictors of bladder toxicity (HR = 3.29 [1.36; 7.98]). The 5–year biochemical relapse free survival was 81.0% [74.5%; 86.0%] in the entire population, 97.5% [83.5%; 99.6%] in the low risk group, 84.9% [76.7%; 90.3%] in the intermediate risk group and 66.4% [51.8%; 77.4%] in the high-risk group. D’Amico low (HR = 0.09 [0.01; 0.69]) and intermediate risk groups (HR = 0.50 [0.28; 0.88]) as well as PSA nadir ≥0.2 ng/ml (HR = 1.79 [1.01; 3.21]) were predictive of biochemical relapse. Conclusions The rate of late rectal toxicity increased with higher doses, while Dmax ≥74Gy, D2% ≥ 73Gy for bladder wall and baseline IPSS ≥8 increased late urinary toxicity.


Introduction
External beam radiation therapy is commonly used to treat with curative intent prostate cancer. During the past two decades, 7 randomized trials and one metaanalysis have shown improvements of 10 to 20% in biological progression-free survival rates (bPFS) and freedom from metastatic disease (in D' Amico high risk patients) by using dose-escalated radiotherapy [1][2][3][4][5][6][7][8]. However, increasing the dose carries a potential risk of severe rectal and urinary side effects [9][10][11][12], especially when former conventional radiation techniques like three-dimensional conformal radiotherapy (3D-CRT) are used.
Nowadays, the use of modern intensity-modulated radiotherapy (IMRT) has shown to significantly reduce acute toxicity rates compared with what has been observed with 3D-CRT, and biochemical control rates are promising [13]. However, mature data on late toxicity and biochemical outcome is still lacking even if earlier studies of prostate IMRT are beginning to report longer follow-up.
The primary objective of our study was to report outcomes, including grade ≥2 overall late rectal and urinary toxicity and biochemical control rates in pts treated with IMRT to 70Gy, 74Gy and 80Gy. We also aimed to identify potential clinical as well as dosimetric predictors of late toxicity in pts receiving moderate dose IMRT (74Gy-group).

Materials and methods
We retrospectively reviewed medical and dosimetry records of pts who received radical IMRT in our center for a prostate cancer. Eligibility criteria included: untreated histologically confirmed adenocarcinoma of the prostate, stage T1c-T4 N0 M0 according to the 1992 American Joint Committee on Cancer staging system. Pts were stratified by prognostic risk groups based on D' Amico criteria [14]. Pts with positive pelvic lymphadenectomy or radiological positives nodes were excluded. Androgen deprivation therapy (ADT) was allowed in D' Amico intermediate and high-risk pts. The following data were extracted from medical records: age, surgical history (prior abdominal surgery, transurethral resection of prostate), anticoagulant or antiplatelet treatment, diabetes, hypertension and coronary insufficiency as well as tumor characteristics (T stage, Gleason score, pretreatment PSA). Dosimetric parameters were extracted for pts receiving 74Gy and are detailed in statistical methods.
The CTCAE v.4 criteria was used to evaluate late toxicity (≥6 months from the start of treatment) for all patients. Toxicity scores were recalculated for patients treated before publication of CTCAE v4.0 scale. The radiation oncologist and urologist recorded peak toxicity grades every 6 months for at least 5 consecutive years.
The RTOG-ASTRO Phoenix Consensus definition of biochemical relapse (nadir + 2ng/ml) was applied for pts treated after 2006 [15]. For all others, 3 consecutive PSA rises after a nadir with the date of failure as the point halfway between the nadir date and the first rise or any rise great enough to provoke initiation of therapy was also considered as a biochemical failure [16].

Treatment planning
All pts underwent treatment planning computed tomography (CT) (Philipps Brilliance 4.0) with a slice thickness of 3 mm. Images were then transferred to the treatment planning system (Eclipse Varian Medical Systems v7-10).
Pts were asked to empty their rectum and bladder 1 h prior to the treatment planning CT scan and daily treatment and invited to drink water in order to have 100-400 ml in the bladder. Bladder filling was verified with a portable bladder-scan. Target and organ at risk delineations followed the recommendations of the French Study Group on Urogenital Tumors (GETUG) [3]. Briefly, the planning target volume (PTV) was calculated by adding a 10-mm margin to the prostate and seminal vesicles in all directions except posteriorly (5mm). Pts were treated in the supine position with COMBIFIX™ device (CIVCO Medical Solutions) for pelvic setup. The radiation dose was escalated over time starting from 70Gy to 74Gy and finally to 80Gy. The dose was prescribed and normalized so that the PTV was included within 95% isodose line. The dose per fraction was specified at the International Commission on Radiation Units and Measurements point. All pts were treated with 10 and 23 MV photons using 5 static IMRT sliding-windows beams at angles of (IEC-scale) 0°, 80°, 130°, 230°, 280°, respectively. Dose constraints used for validation of treatment planning are detailed in Table 1. Two hundred and eighteen pts (78.7%) were treated with a simultaneous integrated boost (SIB), which delivered 2Gy/fraction to the prostate and 1.4Gy to 1.6Gy/fraction to the seminal vesicles. All others were treated in 2 phases: phase I, 46 Gy in 2 Gy/fraction, 5 times/week to the PTV1 (prostate and seminal vesicles), while in phase 2, pts received 24 or 34 Gy to the prostate alone (PTV2). IMRT was delivered with online imaging guidance at day 1, 2, 3 and then weekly. This study was approved by our Institutional Review Board and Ethics Committee and conducted in accordance with the Declaration of Helsinki, Good Clinical Practice and French regulatory requirements.

Statistical methods
Cumulative incidence of bladder toxicity was described with the Kaplan Meier method. The event of interest was overall late urinary toxicity with a grade ≥2. The follow-up time was censored at 84 months for pts with no events. For pts with loco-regional or metastatic progression, the observation of late toxicity was censored at the moment of relapse. Potential prognostic factors of urinary toxicity were assessed with bivariate Cox proportional hazard model (CPHM). Age was dichotomized according to its median value and IPSS according to a threshold of ≥8 [17]. Parameters with a p-value less than 0.2 and age were included into a multivariate CPHM with backward selection. Our dosimetry constraints on bladder wall (Table 1) were tested by a bivariate analysis in the whole population. Patients "were stratified" according to deviations from DVH constraints (Table 1). For example, group 0 = full compliance with all DHV constraints; group 1 = minor or major deviation from 1st DHV constraint (V50 ≤ 35% for 70Gy-arm, V50 ≤ 40% for 74-arm, V70 ≤ 50% for 80-arm); group 2 = minor or major deviation from 2nd DHV constraint, or a combination of non-compliance to all constraints. Our bivariate model aimed to test if noncompliance to one or all of the constraints was predictive of grade ≥2 toxicity. A second analysis of dosimetry data was only performed in the 74-Gy subgroup. The 3 constraints detailed in Table 1 as well as close values to these given constraints were tested as prognostic factors for bladder toxicity by a bivariate CPHM. In case of significant result, the threshold that maximized the loglikelihood function was chosen in order to dichotomize the parameter.
Cumulative incidence of rectal toxicity was described with the Kaplan Meier method and bivariate and multivariate CPHM as described for the urinary toxicity. For dosimetry analysis, our constraints on rectal wall (Table 1) were tested in the whole population and the in the 74Gy-arm subgroup by the same bivariate CHPM model as described above for urinary toxicity.
The bPFS survival was also described by Kaplan Meier method. The event of interest was a biochemical relapse or death (all causes) whichever occurred first. Other events were ignored. Patients alive and free of biochemical relapse were censored at 84 months. Potential prognostic factors were assessed with bivariate CPHM. Threshold for PSA nadir was assessed according to the maximization of the log-likelihood function in CPHM. Parameters with a p-value less than 0.2 and the age were included into a multivariate CPHM with backward selection. Note that stage, Gleason and PSA at baseline were not selected in this procedure as these are strongly linked to D'amico Risk group. The stability of all multivariate CPHM was investigated using a bootstrap resampling method [18].
The median PSA nadir (PSAn) and time to PSA nadir (tPSAn) were compared between radiation doses with a Kruskal Wallis test after testing the normality of the distribution by a Shapiro-Wilk test. For patients without ADT receiving 74Gy, tPSAn was tested as a prognostic factor of biochemical relapse free survival in a bivariate CPHM and dichotomized according to the maximization of the log-likelihood function in CPHM. Multivariate CPHM was performed in order to adjust the result on the prognostic factors found in the entire population.
Cumulative rate of locoregional and metastatic failure was estimated by the Kaplan Meier method. All these results were expressed with hazard ratio and 95% confidence interval. A p-value less than 0.05 was considered statistically significant. All analyses were performed with SAS software, version 9.2 (SAS, Cary, NC, USA).

Results
Between June 2000 and December 2010, two hundred seventy-seven pts were treated by IMRT for a prostate cancer in our center. The median age was 69 years (range 51; 79). The median follow-up was 77 months (15; 150) for pts alive at last follow-up. All pts completed the planned treatment, except one patient who stopped treatment at 78 Gy and was analyzed in the 80Gy-group. Two patients were excluded from the morbidity analysis because follow-up was shorter than 6 months.
One hundred and twenty-eight pts (46.1%) had undergone ADT (≤ 6 months for 70 pts and ≥6 months for 57 pts). The median treatment duration was 57 days (46; 76). The dose planning data from 237 pts was available for analysis. Pretreatment patients' characteristics are listed in Table 2.

I. Late toxicity a) Rectal
There was no grade ≥4 toxicity. The cumulative rate of overall grade ≥2 late toxicity at 5 years was 6.3% (95% CI = 3.8-10.3). Grade 2 was experienced by 9 pts and consisted of minor rectal bleeding, diarrhea and rectal pain. Six pts experienced grade 3 toxicity namely rectal bleeding that needed invasive intervention. All grade 2-3 rectal side effects were observed in pts with no ADT or ≤6 months ADT. Bivariate analyses of prognostic factors are presented in Table 3. Doses of 74Gy versus 70Gy had no statistically significant impact on rectal toxicity. In multivariate analysis only the dose of 80 Gy, compared to 70 and 74Gy, was significantly associated with increased toxicity (HR = 2.96 [1.07; 8.20], p = 0.04) ( Table 3).
The non-compliance of IMRT treatment plans to our constraints on rectal wall (Table 1) was not a significant predictor of rectal toxicity (data not shown).

Discussion
Our results suggest that radiation doses of 80 Gy are associated with a greater likelihood of long-term overall grade ≥2 rectal toxicity but failed to identify other predictors. However, rates of late rectal toxicity, grade 3 CTCAE in particular, were low with IMRT (6.2%). Concerning urinary toxicity, moderate to severe baseline IPSS (≥ 8) worsened radiation-induced toxicity. Dosimetric parameters, especially D2% ≥73Gy, may be used as a surrogate predictor of late urinary toxicity in moderate escalation IMRT (74Gy). To the best of our knowledge, our report is the first to compare 70Gy, 74Gy and 80Gy with IMRT in prostate cancer. We believe that our findings are an important addition to the previously published data on prostate IMRT, and this for several reasons.
Firstly, our study reports overall late toxicity outcomes and biochemical control rates from a comparison between 3 radiation doses levels, the lower and upper bounds being set at 70Gy and 80 respectively. Indeed, it is now well established that doses of 70Gy or over are required for eradication of local disease while the highest radiation dose ever reported in a randomized dose escalation trial using conventionally fractionated radiation was 80Gy. The intermediate bound was set at 74Gy considering that dose escalation with 3D-CRT up to 74Gy was shown to achieved high control rates without an increase in treatment morbidity as compared to 70Gy [19]. We have observed no difference on bPFS in our population regardless of radiation dose. This may be explain by the fact that the majority of our patients (59.9%) had a baseline PSA < 10 ng/ml whereas the advantage of high dose radiotherapy seems to be more important in patients with PSA > 10-15ng/ml [3,6]. We acknowledge the fact that for patients in oldest cohort it is likely possible to have counted PSA recurrences earlier than for those who were analyzed using Phoenix definition of PSA relapse. However, less than 20 pts (roughly 7%) completed treatment 2 years short from publication of Phoenix criteria. The unbalance in ADT use was accounted in both uni-and multivariate analysis as ADT may overcast the impact of radiation dose.
Secondly, the analytical part of our study was focused on intermediate doses (74Gy group), which are often Abbreviations: HR hazard ratio, 95% CI 95% confidence interval, PSA prostate specific antigen, RTH radiation therapy a Parameters included in backward multivariate analysis used in clinical practice but nowadays often overlooked in the literature. Elderly pts may experience more severe radiation induced toxicity in contrast with younger men thus more modest dose radiation schedules such as those used in our series (74Gy) may be an alternative option. In a series of 132 pts treated with prostate-only radiotherapy, 35% of men over 75 years developed acute grade ≥2 rectal toxicity compared to 15% of pts under 75 years. Indeed, Sundar and colleagues [20] found that age above 75 years increases by 3.8 times the risk of rectal toxicity (OR 95% CI = 1.06-13.5). Furthermore, it seems that elderly pts with prostate cancer prefer lower radiation doses over efficacy [21]. Thirdly, we noted excellent rates of overall late grade ≥2 rectal toxicity which sustains other observations showing that IMRT is associated with decreased rectal toxicity compared to 3D-CRT [22]. We have applied the same PTV margins for 70 and 80Gy -groups as those used with 3D-CRT in the GETUG 06 protocol. We obtained a low rate of 5-year Grade ≥2 toxicity (6.2%) which can be thus considered as indirect evidence of the effectiveness of GETUG 06 constraints on rectal wall [3]. The actuarial grade ≥2 toxicity rates issued from investigations are in line with ours: 7.4% at 2 years in the series by Jereszek-Fossa et al. using 3D conformal two-dynamic arc therapy (3D-ART) to deliver a median dose of 76 Gy [23], 11% at 2 years according to De Meeleer et al. when using IMRT from 74 to 76Gy [24] and peak grade ≥2 toxicity of 10.9% for Vora et al. when delivering a median radiation dose of 75.6Gy [25]. The largest study of the Memorial Sloan Kettering Cancer Center (MSKCC) with high-dose IMRT (86.4Gy) showed 7-year actuarial rates of 4.4% Grade ≥2 toxicity. The PTV margins were basically identical to ours, 10mm in all directions except 6 mm posteriorly [26]. Due to a small number of adverse events (15 grade ≥2 toxicities) and the lack of some information on demographics and comorbidities, we failed to show that age, past medical history of pelvic or abdominal surgery, hemorrhoids, anticoagulation treatment are predictive of late rectal toxicity as some studies have suggested it [27][28][29][30]. No patient with ADT >6 months in our cohort had a rectal toxicity ≥ grade 2. The impact of ADT on late gastrointestinal toxicity is still controversial, previous reports showing no significant differences in late toxicity rates with or without the addition of ADT to radiation [31].
Fourthly, we have tried to identify patient and radiation related risk factors correlated with late urinary toxicity. So far, this has been rarely examined and very little is known on the topic. When designing this study authors had in mind to investigate dose constraints for urinary bladder with moderate dose IMRT (74Gy). The premise was again the study conducted on the behalf of the French GETUG who showed that dose escalation from 70 Gy to 80 Gy with 3D-CRT is more frequently associated with grade ≥2 late urinary toxicity when maximum radiation dose was above 75Gy (p = 0.0064) and 50% of the bladder received not more than 44.7 Gy (p = 0.04) [32]. The GETUG 06 cutoff values were used as dose constraints for bladder wall, having in mind that planning margins were identical. Volume receiving 65Gy (V65) ≤25% was equally defined as an additional constraint for the bladder wall (Table 1, column 2). Similar to the GETUG 06 results, we were able to show that bladder wall high dose spots, quantified as Dmax ≥ 74 Gy and D2% ≥ 73Gy are related to late urinary toxicity indicating that the maximal dose with IMRT seems to be important in determining toxicity.
The 5-year incidence of grade ≥2 late urinary toxicity was found to be of 24.9%, slightly higher to those reported in the literature ranging from 12 to 22% at 3 years [25,33,34]. Excellent outcomes were reported by Jereszek-Fossa [23] who found actuarial rates of 8.5% of grade 2 bladder toxicity at 2-years, which is probably due to a shorter follow-up period (23.5 months). Mild genitourinary toxicity is known to develop even into the second decade after radiotherapy although severe toxicity seems to be rare [35].
We also observed significant differences in grade ≥2 urinary toxicity between pts with a baseline IPSS ≥8 and those with an IPSS ≤8. To our knowledge, few studies have explored this potential predictor for radiation toxicity. In the series by Malik R et al., the 4-year freedom from grade ≥2 toxicity was significantly different in men with baseline IPSS ≥15 vs. ≤14 (38% vs. 64%, p < 0.0001). Grade 3 side effects were equality more common in pts with an IPSS ≥15 but this difference was not statistically significant [36]. Recent data from an analysis of 1002 pts treated with high-dose IMRT (86.4Gy) demonstrated that baseline IPSS >15 vs. ≤15 was an independent predictor of grade ≥2 late toxicity [26]. Hoffman et al. demonstrated that men with larger prostate pre-treatment volume have an increased risk of late toxicity [37], knowing that larger prostate volumes are correlated with higher baseline IPSS [38]. Nevertheless, no clear threshold value could be identified from the published literature.
Because this study is retrospective, it is subject to bias in patient selection. Further limitation of this study was the fact that our toxicity grading was limited to case report forms performed during consultations and follow-up visits without patient based selfassessment questionnaires. Combing both seems to be more accurate in quantifying the true incidence of radiation induce toxicity [39]. Furthermore, the use of ADT has been reported to increase urinary toxicity and thus induce biases in toxicity analysis. Like mention before, the unbalanced in ADT use between our groups was accounted in the multivariate analysis.