Secondary cancers after intensity-modulated radiotherapy, brachytherapy and radical prostatectomy for the treatment of prostate cancer: incidence and cause-specific survival outcomes according to the initial treatment intervention


Michael J. Zelefsky, Department of Radiation Oncology, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, Room SM-05, New York, NY 10065, USA. e-mail:


Study Type – Therapy (case series)

Level of Evidence 4

What's known on the subject? and What does the study add?

Radiation Therapy for prostate cancer can increase the risk for the development of second cancers after treatment. This study highlights the fact that such second cancers within the pelvis do occur but are not as common as previously reported. In this report we also note that even among patients who develop second cancers, if detected earlier, the majority are alive 5 years after the diagnosis.


  • • To report on the incidence of secondary malignancy (SM) development after external beam radiotherapy (EBRT) and brachytherapy (BT) for prostate cancer and to compare this with a cohort contemporaneously treated with radical prostatectomy (RP).


  • • Between 1998 and 2001, 2658 patients with localized prostate cancer were treated with RP (n= 1348), EBRT (n= 897) or BT (n= 413).
  • • Using the RP cohort as a control we compared the incidence of SMs, such as rectal or bladder cancers noted within the pelvis, and the incidence of extrapelvic SMs.


  • • The 10-year SM-free survival for the RP, BT and EBRT cohorts were 89%, 87%, and 83%, respectively (RP vs EBRT, P= 0.002; RP vs BT, P= 0.37).
  • • The 10-year likelihoods for bladder or colorectal cancer SM development in the RP, BT and EBRT groups were 3%, 2% and 4%, respectively (P= 0.29).
  • • Multivariate analysis of predictors for development of all SMs showed that older age (P= 0.01) and history of smoking (P < 0.001) were significant predictors for the development of a SM, while treatment intervention was not found to be a significant variable.
  • • Among 243 patients who developed a SM, the 5-year likelihood of SM-related mortality among patients with SMs in the EBRT and BT groups was 43.7% and 15.6%, respectively, compared with 26.3% in the RP cohort; P= 0.052).


  • • The incidence of SM after radiotherapy was not significantly different from that after RP when adjusted for patient age and smoking history.
  • • The incidence of bladder and rectal cancers was low for both EBRT- and BT-treated patients.
  • • Among patients who developed a SM, the likelihood of mortality related to the SM was not significantly different among the treatment cohorts.

second malignancy




external beam radiotherapy


radical prostatectomy


intensity-modulated radiotherapy


androgen deprivation therapy




hazard ratio


Selecting a treatment option for patients with clinically localized prostate cancer is often challenging. For younger patients in particular, quality-of-life issues, long-term treatment–related morbidities and emotional considerations will play important roles in the decision-making process for a particular treatment. One issue for consideration is the potential risk for development of a secondary malignancy (SM) after radiotherapy-based approaches. Several reports have shown that patients with prostate cancer may be at a higher risk of developing SMs, such as bladder or rectal cancers [1–5]. Brenner et al.[6] noted increased relative risks of SM at 5 and 10 years after external-beam radiotherapy (EBRT) of 15% and 34%, respectively. However, there is paucity of information related to how relevant these SM incidence rates are for more sophisticated conformal-based EBRT, which achieves greater sparing of normal tissues such as the rectum and bladder, or for brachytherapy (BT), where much lower radiation dose levels are delivered to the bladder and rectal tissue. In addition, the prevalent notion that the virulence of a SM that develops after radiotherapy-based treatments is greater than that of a SM that develops after surgical intervention is not established.

In the present study we report the incidence of SM among patients with prostate cancer who received highly conformal-based treatment such as intensity-modulated radiotherapy (IMRT) and BT. We compared these incidence rates with those of a cohort of patients who underwent radical prostatectomy (RP) for localized disease and who served as a control group. As the control group represented a younger cohort of patients, and increasing age plays a significant role in the development of SM, a competing risk analysis was performed to account for the risk of dying from other causes, which could otherwise confound the SM incidence rates. In addition, we also studied the survival outcomes of patients who developed a SM to determine if a radiation-induced SM portends a worse prognosis compared with a SM that develops after RP. Our findings showed a low risk of SM development for both IMRT- and BT-treated patients not significantly different from that for patients treated with RP, and these data suggest that previously reported SM risk levels may not accurately reflect the actual incidence found using current treatment approaches.


Between 1998 and 2001, 2658 patients with clinically localized prostate cancer were treated with either RP (n= 1348), BT (n= 413) or EBRT (n= 897). The patient characteristics for all groups are shown in Table 1.

Table 1.  Patient characteristics
Patient demographicsBT, N= 413*EBRT, N= 897RP, N= 1348
n (%) n (%) n (%)
  1. *Median follow-up time, 92 months; †median follow-up time, 90 months; ‡median follow-up time, 113 months. §69 patients in the RP group did not have recorded baseline PSA information, 11 patients had T-stage 0 and two patients were without an assigned T-stage; these patients are not included in the table. NCCN, National Comprehensive Cancer Network.

Age, years<65166 (40.20)232 (25.90)1038 (77.00)
≥65247 (59.80)665 (74.10)310 (23.00)
PSA, ng/mL§<10375 (90.80)576 (64.20)1055 (78.30)
10–2035 (8.50)207 (23.10)173 (12.80)
>203 (0.70)114 (12.70)51 (3.70)
Gleasons score≤6314 (76.00)427 (47.60)612 (45.40)
785 (20.60)349 (38.90)645 (47.80)
≥814 (3.40)121 (13.50)91 (6.70)
T-stage§T1-T2a368 (89.10)673 (75.00)1038 (77.00)
T2b-T2c38 (9.20)129 (14.40)272 (20.20)
≥T3a6 (1.50)85 (9.50)36 (2.70)
Smoking historyYes248 (60.00)481 (53.60)431 (32.00)
No165 (40.70)416 (46.40)917 (68.00)
NCCN RiskNCCN1282 (68.30)272 (30.30)486 (36.10)
NCCN2110 (26.60)386 (43.00)713 (58.90)
NCCN321 (5.10)239 (26.60)149 (11.00)

Among patients who were treated with RP, there were known selection biases. Younger men were more often selected for surgery, and in general those ≥70 years old and those with serious comorbidities (with an American Society of Anesthesiologists score ≥3) were encouraged to consider radiation therapy. Surgical treatment included open bilateral pelvic lymphadenectomy. Patients with positive nodal disease were excluded. In addition, no patient in the RP cohort was treated with neo-adjuvant androgen-deprivation therapy (ADT) before surgery. Patients who were treated with adjuvant or salvage radiotherapy after RP for biochemical-relapsing disease were also excluded.

Among the 413 patients in the BT group, 322 were treated with a permanent interstitial iodine-125 (I-125) implant alone (prescription dose, 144 Gy) and 91 with more advanced disease were treated with either a permanent seed implantation (prescription dose, 110 Gy) or high-dose rate BT with iridium-192 (prescription dose, 21 Gy in three fractions) followed 2 months later by 50.4 Gy of supplemental IMRT to the prostate and seminal vesicles. Our permanent-seed prostate-implantation procedure technique using a real-time intraoperative-planned approach has been previously described [7]. An intraoperative planning system with a treatment-planning algorithm was used to achieve the optimum seed-loading pattern to deliver the prescription dose and maintain established dose constraints for the surrounding normal tissue. In general, BT-treated patients with gland sizes >50 cm3 (n= 167; 52%) were pretreated for 3 months with short-term neo-adjuvant ADT to reduce prostate volume before therapy.

In general, patients in the EBRT group were treated with IMRT using a five-field co-planar beam arrangement directed to the prostate and seminal vesicles via 15 MV photons in daily 1.8 Gy fractions. Treatment planning and delivery were as previously described [8]. The median EBRT prescribed dose was 8100 cGy. Before undergoing EBRT, 523 patients (58%) were treated with short-course (3-month) ADT. In general, short-course ADT was given to decrease the size of the prostate gland if enlarged before radiotherapy or for those with high-grade, high-risk disease. ADT was routinely discontinued after radiotherapy.

Follow-up evaluations were performed at 3–6-month intervals for 5 years, and yearly thereafter. Median follow-up times for the RP, BT and EBRT groups were 113, 92 and 90 months, respectively. For the purpose of this report, SM included skin cancers, yet the incidence rates of SM were calculated separately with and without the inclusion of skin cancers. The Kaplan–Meier method was used to determine the actuarial likelihood of developing a SM [9]. The risk of mortality that was related to SM and other causes was analysed by competing-risk analysis [10,11].

Institutional review board approval was obtained for this retrospective study.


The crude incidence of SM locations for each of the treatment groups as well as stage are shown in Table 2 and Table 3. The rate of bladder and rectal cancers that developed after treatment in the RP, BT and EBRT cohorts was 1.4% and 0.7%, 1.0% and 0.5% and 1.2% and 0.7%, respectively (P= 0.336) Among those patients who developed a SM, the median time from radiotherapy to the diagnosis of pelvic and extrapelvic SMs was 75 and 51 months for RP, 46 and 51 months for EBRT, and 39.5 and 37 months for BT cohorts, respectively.

Table 2.  SMs by therapy treatment
n (%)Median time to SM, months n (%)Median time to SM (months) n (%)Median time to SM, months
In-field SM2246639.54075
 Rectal/Colorectal/Sigmoid5 (23)242 (33)529 (23)69
 Bladder12 (55)494 (67)39.516 (40)98
 Other5 (23)70  15 (38)69
Out-field SM7051303711551
 Melanoma12 (17)53.52 (7)4610 (9)25
 Colon6 (9)606 (20)35.5  
 Lymphoma11 (16)465 (17)3715 (13)56
 Lung12 (17)703 (10)5822 (19)49.5
 Liver4 (6)32.52 (7)738 (7)29.5
 Stomach4 (6)202 (7)51.51 (1)105
 Head and neck5 (7)436 (20)2710 (9)42
 Other16 (23)364 (13)4449 (43)65
Table 3.  SMs by stage
 SM stagingEBRT, n (%)BT, n (%)RP, n (%)
OverallStage I–II45 (49)21 (58)56 (36)
Stage III–IV31 (34)9 (25)39 (25)
Unknown16 (17)6 (17)60 (39)
In-fieldStage I–II16 (73)6 (100)19 (48)
Stage III–IV3 (14) 10 (25)
Unknown3 (14) 11 (28)
Out-fieldStage I–II29 (41)15 (50)37 (32)
Stage III–IV28 (40)9 (30)29 (25)
Unknown13 (19)6 (20)49 (43)

As shown in Fig. 1, the 10-year actuarial likelihood for all SM development for the RP, BT, and EBRT groups was 11%, 13% and 17%, respectively (P= 0.002) and the 10-year likelihood for developing a pelvic SM was 3%, 2%, and 4%, respectively (P= 0.29). The 10-year incidence for developing an extrapelvic SM was 8%, 11%, and 14%, for RP, BT and EBRT, respectively (P= 0.02). In a Cox regression analyses (Table 4), only increasing age and a history of smoking were predictors for SM development, and the treatment intervention was not a significant variable.

Figure 1.

Overall SM-free survival curve for patients who received RP vs EBRT (P=0.002) and BT vs RP (P=0.37). Probability of 5-year SM-free was 94%, 90% and 92% for RP, EBRT and BT, respectively) and 10-year SM–free probability was 89%, 87% and 83%.

Table 4.  Univariate and multivariate analyses for predictors of overall SM risk (pelvic and extrapelvic)
VariableUnivariate analysisMultivariate analysis
  • *

    11 patients had T-stage 0 disease and two patients were not assigned a T-stage.

  • 3D-CRT, three-dimensional conformal radiotherapy; NCCN, National Comprehensive Cancer Network.

 EBRT vs RP1.560.002  
 BT vs RP1.200.37  
Treatment mode    
 IMRT vs RP1.560.002  
 3D-CRT vs RP1.370.75  
 BT only vs RP1.170.47  
 BT combo vs RP1.280.51  
Age (continuous)1.03<0.0011.03<0.001
Age (>69 vs ≤69 years)1.270.11  
 T2b-T2c vs T1-T2a1.010.96  
 ≥T3a vs T1-T2a1.360.27  
Gleason score1.060.52  
Pretreatment PSA1.000.34  
NCCN risk    
 Intermediate vs low1.
 High vs low1.600.0091.460.038
Smoking (yes vs no)1.81<0.0011.71<0.001
Neo-hormone therapy (yes vs no)1.370.021  


Table 5 shows the incidence of SM for each of the treatment groups as well as according to the assigned stage of the SM. As the staging systems vary from one cancer sub-site to another, for the purpose of this analysis, stages I and II were characterized as early stage and stages III and IV were characterized as advanced stage. Using a competing-risk analysis to account for other causes of death, the 10-year incidence of mortality related to an in-field SM (such as colorectal and bladder cancers) for the RP, EBRT and BT groups was 0.35% (95% CI 0.01–0.68%), 0.12% (95% CI 0–0.36%), and 0%, respectively. The 10-year incidence of mortality related to an extrapelvic SM for the RP, EBRT and BT groups was 0.92% (95% CI 0.38–1.46%), 1.97% (95% CI 1.01–2.92%) and 0.78% (95% CI 0. 01–1.67%), respectively.

Table 5.  Competing risks regression analysis limited to patients with SMs
VariableUnivariate analysisMultivariate analysis
  1. In the univariate analysis for Stage of SM only 176 patients were included. For all other variables, all 243 SM patients were included in the univariate analysis. As the multivariate analysis included only observations with complete data on all the variables, 176 patients were used in the multivariate analysis.

EBRT vs RP1.770.031.880.054
BT vs RP0.820.670.880.83
Age (continuous)
Stage of SM (high vs low)6.6<0.0016.92<0.001
Smoking (yes vs no)1.000.990.690.21

Secondary malignancies after radiotherapy (EBRT or BT) were not associated with a worse survival outcome compared with that of patients who developed SMs after RP. Among 243 patients who developed a SM, the likelihood of mortality related to that SM was calculated. Using a competing-risk analysis to account for other causes of death, the 5-year cause-specific (attributable to SM) mortality rates among EBRT and BT patients with SMs (calculated from the date of SM to last follow-up or death from SM) were 43.7% and 15.6%, respectively, compared with 26.3% in the RP cohort; P= 0.052) The 5-year cause-specific mortality rates for pelvic SMs in the EBRT and BT groups were 5.7% and 0%, respectively, compared with 5.7% in the RP group (P= 0.44).


The present study is the first report in the literature to compare the incidence at a single institution of SMs in patients treated with EBRT or BT with a control cohort of patients contemporaneously treated with RP. It dispels a number of misconceptions regarding the frequency and presumed behaviour of SMs in patients treated with RT. It has been reported by some [3,5,6,11] that the frequency of rectal and bladder cancers is significantly higher after prostate RT, but in the present study at 10-year follow-up the incidence of rectal and bladder cancers after EBRT or BT was <2% and not significantly different from the observed incidence of such tumours among patients treated with RP. Another widely cited misconception is that SMs that develop after radiotherapy are highly lethal and not as responsive to standard oncological regimens [11]. Based on the data presented in this report we have not found this to be the case. In fact, among the patients who developed a SM after radiotherapy in the present study, the overwhelming majority experienced 5-year survival outcomes of >70% and these outcomes were not significantly different from patients who developed SMs after RP. As we have previously noted, patients who developed bladder and rectal cancers in the present study population were in general diagnosed at early stages and this could possibly be the reason for the improved survival outcomes in these patients [12].

In a recent report, Hinnen et al.[13], from the Netherlands Cancer Institute, compared the risk of SMs after I-125 prostate BT and surgery. With a median follow-up of 7 years for both treatment groups, the crude incidence of rectal and bladder cancers for patients in the BT and RP groups was low and similar to each other (BT: 0.8% and 1.4%; RP: 1.3% and 1.4%, respectively). Secondary bladder cancer risk was noted to be increased after BT in patients <60 years of age. Their results were similar to the low incidence of second pelvic tumours that we observed across all three forms of treatment in the present study, but we could not confirm that younger patients treated with BT experienced a higher incidence of bladder cancers after treatment. In the present study, multivariate analysis showed that age and smoking history were predictors for the overall risk of a SM after treatment, but the particular treatment intervention, i.e. BT, EBRT or RP, did not influence this endpoint.

The lower incidence rates of secondary bladder and rectal cancers after EBRT found in the present study could possibly be related to the conformal treatment techniques used, e.g. IMRT. Our results differ from those reported by Bhohani et al.[14] from the Montreal Health Center, who noted a higher incidence of rectal and lung cancers in EBRT-treated patients than in surgery-treated patients at 10 years after treatment (2% vs 1%, hazard ratio [HR] 2.0; and 7% vs 4%, HR 2.1, respectively). Their results, however, were derived from patients treated with EBRT before the availability of conformal-based techniques, with larger volumes of normal tissue being exposed to the radiation doses. The notion that conformal radiotherapy may be associated with a lower incidence of SM is also suggested in a recent report by Huang et al.[15]. These authors compared 2120 patients treated with various forms of EBRT or BT with a surgical cohort extracted from a population-based cancer registry from the Detroit region of the USA using a matched-pair analysis. They noted that among patients who were treated with non-conformal treatment techniques (i.e. two-dimensional planning), which are associated with larger volumes of normal tissue exposed to the radiation doses, higher risks of bladder cancers and lymphomas were observed compared with the incidence among patients treated with more conformal techniques such as three-dimensional conformal radiotherapy or IMRT. One of the limitations of their findings was that the follow-up for the two-dimensional radiotherapy was twice as long as that for the patients who underwent conformal treatment (9.2 years vs 4.9 years, respectively). The present data, with longer follow-up, show that the incidence of bladder and rectal cancers after highly conformal radiotherapy delivery was low and similar to the incidence observed in the RP cohort.

The limitations of the present study include its retrospective nature, the relatively small number of patients included (SM is a relatively rare event), and the limited follow-up beyond 12 years after treatment. Nevertheless, the present data suggest that, using sophisticated conformal treatment approaches for radiation dose delivery, the incidence of post-treatment malignancies is low. Longer follow-up observations will be necessary to corroborate these findings. In addition, the present data highlight the importance of careful attention in the follow-up period for the possibility of SM development. While we do not advocate routine follow-up cystoscopic or sigmoidoscopic procedures for patients after EBRT or BT, we nevertheless recommend baseline endoscopies and follow-up procedures at least every 3–5 years. For patients who develop gross haematuria after treatment, we recommend follow-up evaluation including urine cytology, CT urogram and cystoscopy to rule out a secondary bladder cancer. The present data suggest that the early diagnosis of such SMs may have contributed to the favourable survival outcomes among patients who developed a SM after radiotherapy interventions.


None declared.