Ten-year outcomes of high-dose, intensity-modulated radiotherapy for localized prostate cancer




The authors investigated long-term tumor control and toxicity outcomes after high-dose, intensity-modulated radiation therapy (IMRT) in patients who had clinically localized prostate cancer.


Between April 1996 and January 1998, 170 patients received 81 gray (Gy) using a 5-field IMRT technique. Patients were classified according to the National Comprehensive Cancer Network-defined risk groups. Toxicity data were scored according to the Common Terminology Criteria for Adverse Events Version 3.0. Freedom from biochemical relapse, distant metastases, and cause-specific survival outcomes were calculated. The median follow-up was 99 months.


The 10-year actuarial prostate-specific antigen relapse-free survival rates were 81% for the low-risk group, 78% for the intermediate-risk group, and 62% for the high-risk group; the 10-year distant metastases–free rates were 100%, 94%, and 90%, respectively; and the 10-year cause-specific mortality rates were 0%, 3%, and 14%, respectively. The 10-year likelihood of developing grade 2 and 3 late genitourinary toxicity was 11% and 5%, respectively; and the 10-year likelihood of developing grade 2 and 3 late gastrointestinal toxicity was 2% and 1%, respectively. No grade 4 toxicities were observed.


To the authors' knowledge, this report represents the longest followed cohort of patients who received high-dose radiation levels of 81 Gy using IMRT for localized prostate cancer. The findings indicated that high-dose IMRT is well tolerated and is associated with excellent long-term tumor-control outcomes in patients with localized prostate cancer Cancer 2011. © 2010 American Cancer Society.

Recent randomized controlled trials and several single-institution studies have confirmed the advantage of delivering high doses of external beam radiotherapy to achieve optimal tumor-control outcomes in patients with localized prostate cancer.1-9 Several studies have demonstrated an association between high-dose radiotherapy that is delivered using conventional treatment planning and elevated risks of late urinary and rectal toxicities.5, 10, 11 An advanced form of 3-dimensional, conformal radiation therapy is intensity-modulated radiotherapy (IMRT), which delivers nonuniform beam intensities to an irregular target volume to create a highly sculpted dose distribution. These techniques have facilitated the safe delivery of increased doses of radiation to the prostate and seminal vesicles with concurrent dose reductions to adjacent normal tissues.

At Memorial Sloan-Kettering Cancer Center, IMRT was introduced as a dose-escalation tool in 1996 to allow for the delivery of higher dose radiation levels of 81 grays (Gy) to 86.4 Gy and for the delivery of as much as 91 Gy to significant portions of the target. These ultrahigh dose levels have been possible and are well tolerated.6 We previously reported the long-term results from 81 Gy delivered with IMRT, which has been associated with a low risk for treatment-related complications. In this report, we summarize the 10-year results of IMRT. To our knowledge, the current study represents the longest followed cohort of patients to date who have received high-dose IMRT for localized prostate cancer. Our findings indicate that delivery of high-dose IMRT continues to be associated with excellent long-term tolerance outcomes and disease-control rates.


Between April 1996 and January 1998, 170 patients with histologically proven prostate cancer received IMRT up to a prescribed dose of 81 Gy at Memorial Sloan-Kettering Cancer Center. The median patient age was 69 years (range, 51-82 years). The characteristics of these patients are listed in Table 1. Ninety-one patients (54%) received 3 months of neoadjuvant androgen deprivation therapy (N-ADT) before radiotherapy. In general, short-course N-ADT was given to decrease the size of the enlarged prostate before radiotherapy or to patients who had high-grade, unfavorable-risk disease. ADT routinely was continued during the course of IMRT and then was discontinued at the completion of therapy. On-treatment evaluations were performed at least weekly during the treatment course, and follow-up evaluations were performed at intervals of 3 to 6 months for 5 years and yearly thereafter. The median follow-up was 99 months and was calculated from the completion of radiation therapy.

Table 1. Patient Characteristics
CharacteristicNo. of Patients%
  1. AJCC indicates American Joint Committee on Cancer; PSA, prostate-specific antigen; NCCN, National Comprehensive Cancer Network.

AJCC tumor classification
Gleason score
 Median (range)6 (5-9) 
PSA, ng/mL
 Median (range)8 (47-95) 
NCCN risk group
 Low risk4929
 Intermediate risk8952
 High risk3219

Detailed IMRT techniques for treatment planning and delivery as well as quality assurance have been reported elsewhere.12-15 Briefly, all patients were treated in the prone position within a thermoplastic mold and were instructed to empty their bladders immediately before treatment to ensure reproducibility of the daily setup and to minimize organ motion, as described previously.16 Patients were scanned in the treatment position from the L5-S1 levels to 10 cm caudal to the ischial tuberosities on a computerized tomography (CT) simulator. CT slices were reconstructed at 0.3-cm increments to produce high-resolution, 3-dimensional images and digitally reconstructed radiographs. The CT images were downloaded into the Memorial Sloan-Kettering treatment-planning system. Permanent localization marks were placed at a standardized treatment isocenter: a midline point near the center of the prostate located 1 cm inferior to and 6 cm posterior to the symphysis pubis. The clinical target volume included the prostate and seminal vesicles. A margin of 1.0 cm was added to the clinical target volume except at the prostatorectal interface, where a 0.6-cm margin was used. An additional 0.5-cm margin was added around the planning target volume (PTV) (except superiorly and inferiorly, where 1.0 cm was used) to account for penumbra. The rectum, bladder, bowel, and femora were contoured as critical normal tissue structures. An isocentric, 5-field technique comprising a posterior field (0°), a right posterior oblique field (75°), a right anterior oblique field (135°), a left-anterior oblique field (225°), and a left posterior oblique field (285°) was used. Treatment plans then were optimized with an inverse optimization algorithm,17, 18 which is used to minimize the sum of quadratic differences between the desired and computed dose distributions. Optimization was performed using dose or dose-volume constraints and penalties to control the PTV dose, homogeneity within the PTV, and doses to critical structures. The parameters that were used to produce the desired dose distribution included dose uniformity (100%) to the PTV, and limits that no greater than 30% of the rectal wall volume should receive >75.6 Gy (V75.6 <30) and that no greater than 53% of the rectal and bladder walls should receive >47 Gy (V47 <53). In the overlap region between the PTV and these critical organs, the constraint was set at 88% of the prescription dose for the rectum and at 98% for the bladder. Once the intensity profiles of the 5 IMRT beams were determined, leaf-motion files were created, and dose distributions were generated. Beam's-eye-view, digitally reconstructed radiographs were generated, onto which fluence apertures were projected. The fluence apertures of each intensity-modulated field were specified to encompass the area with an intensity >1% of the maximum. The desired beam-intensity profiles were delivered by dynamic multileaf collimation using the sliding window technique.15

All patients were treated to a prescribed dose of 81 Gy in daily fractions of 1.8 Gy in 45 fractions using 15-megavolt photons. The prescribed dose represented the minimum dose to the PTV, but portions of the target volume, including the isocenter (International Commission on Radiation Units and Measurements prescription point), received up to 10% higher doses. During the treatment period for patients in this study, the patient position was verified with weekly port films. More recently, we instituted a policy of routine fiducial marker placement and daily, 2-dimensional-kV imaging for prostate localization for all patients who receive prostate IMRT.

Late gastrointestinal (GI) and genitourinary (GU) toxicities were graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events, version 3.0, on the following 4-point scale: grade 1, minimal side effects not requiring medications; grade 2, side effects requiring medications for symptom management (or an increase in the dose of pre-existing medication); grade 3, side effects requiring minor procedures (ie, cauterization, catheterization, transfusions) to control or affect activities of daily living; or grade 4, life-threatening toxicity requiring major surgery and hospitalization. Erectile function was evaluated according to patient reports at the time of each follow-up and was defined as the inability to achieve an erection sufficient for penetration.

Patients were classified into groups according to their risk of recurrence as defined by National Comprehensive Cancer Network (NCCN) guidelines (www.nccn.org; accessed May 14, 2009). Patients were classified with low-risk disease if they had clinical T1 to T2a tumors, a Gleason score ≤6, and a pretreatment prostate-specific antigen (PSA) level <10 ng/mL. Patients with clinical T2b or T2c tumors, a Gleason score of 7, or a pretreatment PSA level between 10 ng/mL and 20 ng/mL were classified with intermediate-risk disease. Patients who had T3a or higher tumors, a Gleason score ≥8, or a pretreatment PSA level >20 ng/mL were classified with having high-risk disease.

Disease status was determined according to an analysis in October 2008. The date patients completed radiotherapy was used as the starting time for all survival endpoints in this analysis. PSA relapse was defined according to the Phoenix consensus definition as an absolute nadir PSA level plus 2 ng/mL dated at the call.19 None of the patients received postirradiation ADT or other anticancer therapy before documentation of a PSA relapse. For cause-specific survival (CSS) analysis, patients with documentation of biochemical or metastatic, relapsing disease who subsequently died were scored as deaths from prostate cancer. Death from other causes was considered a competing risk. Distributions of PSA relapse-free survival and distant metastases-free survival (DMFS) were calculated using the Kaplan-Meier method. Cause-specific mortality rates were calculated using competing-risks analysis tools. The log-rank test was used to assess whether subgroups had nonparametrically and univariately identical distributions of the survival endpoints. When multivariate analysis was applicable or continuous covariates were involved, the Cox proportional hazards regression model was used to determine the effect of covariates, and a stepwise model selection tool was used to construct the final multivariate model. Statistical significance was achieved when P ≤ .05.


A typical dose distribution depicting improved tumor coverage with 81 Gy and reduced volume of critical normal structures using IMRT is shown in Figure 1.

Figure 1.

The planned dose distribution for a patient with prostate cancer is illustrated. cGy indicate centigrays.

Biochemical Tumor Control

PSA relapse-free survival rates according to NCCN prognostic risk groups are shown in Figure 2. The 10-year actuarial PSA relapse-free survival was 81% (95% confidence interval [CI], 64%-97%), 78% (95% CI, 66%-91%), and 62% (95% CI, 44%-80%) for the low-risk, intermediate-risk, and high-risk groups, respectively (P = .013). Clinical tumor classification (P = .0002) and a pretreatment PSA level >10 ng/mL (P = .008) were identified as significant predictors of biochemical relapse on multivariate analysis (Table 2). Gleason score, receipt of short-term N-ADT, and patient age had no demonstrable impact on the biochemical relapse rate in these patients.

Figure 2.

These Kaplan-Meier curves illustrate the actuarial probability of achieving prostate-specific antigen relapse-free survival based on the prostate-specific antigen (PSA) nadir plus 2 ng/mL definition according to National Comprehensive Cancer Network risk group. Int indicates intermediate.

Table 2. Cox Regression Model Results for Prostate-Specific Antigen Relapse-Free Survival Analysis
VariableUnivariate AnalysisMultivariate Analysis
HR (95% CI)PHR (95% CI)P
  1. HR indicates hazard ratio; CI, confidence interval; NCCN, National Comprehensive Cancer Network; NS, nonsignificant; HT, hormone therapy; PSA, prostate-specific antigen.

NCCN risk group1.99 (1.18-3.34).009NSNS
Age, continuous1.02 (0.96-1.08).51  
Age, >69 y vs ≤69 y1.09 (0.54-2.22).80  
HT, 1 vs 00.89 (0.44-1.81).75  
Pre-PSA, >10 ng/mL vs ≤10 ng/mL2.41 (1.19-4.88).0152.64 (1.29-5.43).008
Tumor classification1.42 (1.17-1.71).00031.47 (1.20-1.80).0002
Gleason score1.44 (0.92-2.24).11NSNS

Distant Disease Control and Survival Outcomes

Distant metastasis developed in 8 patients (5%) at a median of 20 months (range, 7-70 months) after the completion of therapy. The 10-year DMFS rate was 95% (95% CI, 91%-98%) (Fig. 3). The 10-year DMFS outcomes were 100%, 94%, and 90% for low-risk, intermediate-risk, and high-risk patients, respectively (P = .12). Univariate analysis demonstrated that clinical tumor classification (P = .0002) and a pretreatment PSA level >10 ng/mL (P = .014) were predictors of long-term DMFS. Because of the limited number of events, a multivariate analysis could not be performed (Table 3).

Figure 3.

This Kaplan-Meier curve illustrates the actuarial probability of achieving distant metastasis-free survival.

Table 3. Competing Risks Analysis Results for Cause-Specific Mortality Rates
VariableUnivariate Analysisa
HR (95% CI)P
  • HR indicates hazard ratio; CI, confidence interval; NCCN, National Comprehensive Cancer Network; HT, hormone therapy; PSA, prostate-specific antigen.

  • a

    Multivariate analysis was not applicable because there were only 5 events.

NCCN risk group5.95 (1.48-24.03).012
Continuous1.12 (1.00-1.24).05
Ages >69 y vs ≤69 y1.66 (0.28-9.98).58
HT, 1 vs 01.43 (0.24-8.57).70
Pre-PSA, >10 ng/mL vs ≤10 ng/mL8.58 (1.00-73.74).05
Tumor classification2.16 (1.40-3.34).0005
Gleason score1.19 (0.54-2.59).67

The overall 10-year cause-specific mortality rate was 3.8%. The 10-year cause-specific mortality rates were 0%, 3%, and 14% for low-risk, intermediate-risk, and high-risk patients, respectively (P = .0012).


The incidence of late GU and GI toxicities is shown in Table 4. The median time to the development of late grade 2 GU and grade 3 GU toxicities were 31 months and 42 months, respectively. No late grade 4 GU toxicities were observed. The 10-year actuarial risk of developing late grade ≥2 GU toxicity was 17% (Fig. 4). Late grade 2 GI toxicity occurred in 4 patients (2%), and late grade 3 GI toxicity occurred in 2 patients (1%). Among patients who developed late grade 2 GI toxicity, 2 patients developed grade 2 rectal bleeding at a median of 24.5 months after the completion of therapy. Among the patients who developed late grade 3 GI toxicity, 2 patients developed grade 3 rectal bleeding that required 1 or more transfusions or a laser cauterization procedure at a median 17 months after the completion of IMRT. No grade 4 rectal toxicities were noted. The 10-year incidence of late grade ≥2 GI toxicity was 3.7% (Fig. 5). On multivariate analyses, the presence of acute grade ≥2 GU toxicity was predictive for the development of late grade ≥2 GU toxicity (P = .001). The receipt of ADT and patient age were not associated with developing late grade ≥2 GU toxicity. The presence of acute grade ≥2 GI toxicity was the only predictor on multivariate analyses of late grade ≥2 GI toxicity (P = .0025).

Figure 4.

This Kaplan-Meier curve illustrates the actuarial probability of grade ≥2 late genitourinary (GU) toxicity and indicates that 130 patients will be at risk at 5 years and that 8 patients will be at risk at 10 years.

Figure 5.

This Kaplan-Meier curve illustrates the actuarial probability of grade ≥2 late gastrointestinal (GI) toxicity and indicates that 141 patients will be at risk at 5 years and that 8 patients will be at risk at 10 years.

Table 4. Late Toxicity From Intensity-Modulated Radiotherapy (n=170)
Toxicity GradeNo. of Patients (%)
None108 (63)132 (78)
139 (2332 (19)
215 (9)4 (2)
38 (5)2 (1)
40 (0)0 (0)

Erectile Dysfunction

Forty-one patients were impotent before they received radiotherapy. Of the 105 patients who were potent (ie, had erections sufficient for penetration) before IMRT, 44 patients (42%) became impotent. The 10-year actuarial incidence of developing postradiation erectile dysfunction was 44%. The 10-year actuarial incidence of developing postradiation erectile dysfunction in patients who did and did not receive N-ADT was 54% and 35%, respectively. The receipt of N-ADT was predictive for the development of erectile dysfunction on multivariate analyses (P = .02).


The current results demonstrate that radiation dose levels of 81 Gy with IMRT are tolerated well in these patients, who have been followed for many years after treatment. The limitations of current reports on IMRT included smaller numbers of patients who were followed for ≤5 years, which did not allow adequate time to observe clinical failures or treatment-related toxicities. We observed relatively low failure rates 10 years after IMRT. The 10-year biochemical control rates were 81%, 78%, and 62% in patients with low-risk, intermediate-risk, and high-risk features, respectively.

It is now well established that conventional doses for external beam radiotherapy in the range of 70 Gy are not sufficient for the eradication of local prostatic disease. Three-dimensional, conformal radiotherapy and IMRT represent improvements in radiation planning and delivery that make it possible to deliver higher doses of radiation to the prostate, which can lead to improved outcomes for biochemical tumor control in patients with clinically localized prostate cancer.2, 3, 5, 7, 8, 20 The improvements in biochemical tumor-control rates with the use of higher radiation doses have been correlated not only with superior local control but also with improved distant metastases-free and CSS outcomes.21-25 The significant dose effect was demonstrated for improved DMFS in patients who received ≥81 Gy compared with patients who received 75.6 Gy.24 In the current report, we observed that a pretreatment PSA level ≥10 ng/mL and advanced clinical tumor classification were significant predictors of biochemical control, distant disease control, and CSS rates. The lack of a benefit from ADT in high-risk patients probably is related to the relatively short course of only 5 to 6 months of ADT that was received by patients at that time. Currently, we advocate the use of longer courses of ADT, and particularly for high-risk patients.

The effectiveness of high-dose radiation therapy is limited by possible increased risks to normal tissues for long-term morbidity. The incidence of late grade ≥2 GU and GI toxicity reportedly ranges from 13% to 40% and from 26% to 30%, respectively, in high-dose (78 Gy) radiotherapy arms of recent studies10, 26, 27 More recently, Cahlon et al reported our results using ultrahigh dose (86.4 Gy) radiotherapy.6 The 5-year incidence of late grade ≥2 GU and GI toxicity was 16% and 4%, respectively, in the study. In the current report, using an 81-Gy dose level with IMRT, the risks of developing late grade ≥2 GU and GI toxicities were 17% and 3.7%, respectively, at 10 years, and the 10-year incidence of developing grade ≥2 rectal bleeding was 2%. Taken together, these data demonstrate that IMRT is the safest way to deliver high doses of external beam radiotherapy and that our routinely used normal tissue margins are sufficient to cover the PTV.

Grade 2 and greater acute GU symptoms and GI symptoms developed in 5 patients (3%) and 2 patients (1%), respectively, during the course of radiotherapy. Fifty-four percent of our patients received short-course N-ADT, and we did not observe more toxicity in those patients. Similarly, Zelefsky et al reported that the receipt of ADT did not have a demonstrable influence on the incidence of acute symptoms.28 In contrast to our report, other studies have suggested higher toxicities when longer courses of hormone therapy were used.8, 29 We also could not demonstrate any predictive effect of ADT for developing higher GU or GI toxicity in our multivariate analyses. Similarly, the Radiation Therapy Oncology Group 94-06 trial has demonstrated no effect of neoadjuvant hormone therapy on late GI or GU toxicity.30 That group noted that only patients with poor pretreatment urinary function who were receiving hormone therapy had increased acute toxicity. Conversely, a relation between acute and late GI toxicities was reported. In a study by Heemsbergen et al, the authors observed that late effects were a direct consequence of the initial tissue injury.31 In that study, the presence of diarrhea during treatment predicted an increased risk of late grade ≥2 rectal toxicity. Zelefsky et al recently reported that the presence of acute GI and GU symptoms during treatment conferred a 5-fold and 3-fold increased risk of late GI and GU toxicities, respectively, in 1571 patients with prostate cancer who had a long follow-up after receiving 3-dimensional, conformal radiotherapy or IMRT.28 Similarly, in our current multivariate analyses, we observed that the presence of acute grade ≥2 GI and GU toxicity was a significant predictor of late grade ≥2 GI and GU toxicity. The other important finding in our report is the relatively high incidence at 10 years of erectile dysfunction in men with prostate cancer. The actuarial incidence of developing postradiation erectile dysfunction was 44% at 10 years. Certainly the multitude of factors that affect the likelihood of developing erectile dysfunction are well known and include pretreatment status; the presence of comorbidities, such as diabetes and heart disease; medications; and patient age. With a median age of 69 years at the time of treatment, it is expected that a significant number of patients would have developed a decline in erectile function with normal aging over the 10 years of follow-up. Although IMRT would allow for the delivery of lower doses to erectile tissues and potentially lead to improved potency rates,32 we could not demonstrate that the use of this modality was associated with a lower incidence of erectile dysfunction in these patients compared with other treatment techniques, such as 3-dimensional, conformal radiotherapy. In our report, the use of short-course N-ADT played a significant role in the development of erectile dysfunction, as noted in our multivariate analyses. Similar findings have been reported by others.14, 33, 34

In conclusion, the current data indicate that IMRT is associated with excellent clinical outcomes in patients with localized prostate cancer who were followed for 10 years. IMRT requires more time and effort from physicians and physicists, but much of the planning and delivery have become automated. Computer algorithms are available to support treatment-planning optimization and treatment delivery with dynamic multileaf collimation. Using dynamic multileaf collimation and inverse-planning treatment software, IMRT can deliver a highly conformal and targeted radiation dose to the prostate while also lowering the concomitant dose to the rectum and bladder. Our findings indicate that high-dose IMRT is tolerated well and is associated with excellent long-term tumor-control outcomes in patients with localized prostate cancer.


Supported by Program 2219 of the Scientific and Technical Research Council of Turkey (TUBITAK).