Temporal patterns of selected late toxicities in patients treated with brachytherapy or brachytherapy plus external beam radiation for prostate adenocarcinoma


Richard G. Stock, Department of Radiation Oncology, Mount Sinai School of Medicine, Box 1236, 1 Gustave L. Levy Place, New York, NY 10029, USA. e-mail: richard.stock@mountsinai.org


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

While the frequencies and severity of late toxicities following prostate brachytherapy are well known, less has been published with regard to time to first onset. Several series with limited median follow-up have published time to onset. An extensive analysis of timing to late toxicity following brachytherapy for cervical cancer has also been published.

This study is the largest of its kind with the longest median follow-up to capture very late events. It can provide a basis for physician and patient education about when late toxicities can reasonably be expected to occur. The study also shows that a significant amount of erectile dysfunction might be more age related than radiation induced.


  • • To assess the timing of first onset of late rectal bleeding, late haematuria and erectile dysfunction (ED) following brachytherapy with or without external beam radiation therapy (EBRT) for prostate adenocarcinoma.
  • • To identify treatment factors and patient characteristics that affect the time to first onset.


  • • In all, 2046 patients were definitively treated for prostate adenocarcinoma with a full 125I or 103Pd implant or a partial 103Pd implant followed by EBRT with 6 years median follow-up (range 2–17 years).
  • • Patients were selected for an event of Radiation Therapy Oncology Group (RTOG) grade 2 or greater rectal bleeding, ≥RTOG grade 2 haematuria, or a drop in the Mount Sinai Erectile Dysfunction Score from potent to impotent (excluding patients who received androgen deprivation therapy).
  • • Life tables were generated to calculate actuarial incidence rates of toxicity.
  • • Wilcoxon rank sum and Cox regression were utilized to identify treatment factors affecting time to onset.


  • • The incidence rate per 1000 patients for 0–2 years, 2–5 years and 5–10 years following radiation for rectal bleeding is 14.3, 15.9 and 6.5, respectively; for haematuria, 14.0, 8.2 and 1.3, respectively; and for ED, 82.4, 48.2 and 42.2, respectively.
  • • Just 5% of rectal bleeding occurs after 5 years from radiation vs 18% of haematuria cases and 22% of ED.
  • • On multivariate analysis, time to first onset of rectal bleeding was affected by the addition of EBRT only whereas the time to onset of haematuria was affected by the biological effective dose of the radiation and the addition of EBRT.
  • • The only factor on multivariate analysis to affect time to onset of ED was the age of the patient at treatment, independent of radiation dose or technique.


  • • Unique temporality to first onset of selected toxicities was observed in patients after radioactive implant for prostate adenocarcinoma with or without EBRT.
  • • Clinicians and patients should be counselled when to expect late toxicities.
  • • The only factor to affect time to onset of ED is the age of the patient, suggesting possible over-reporting of radiation-induced ED in the light of normal age-related events.

external beam radiation therapy


erectile dysfunction


Radiation Therapy Oncology Group


androgen deprivation therapy


Mount Sinai Erectile Function Score


National Institute of Standards and Technology


biological effective dose


hazard ratio.


The use of brachytherapy alone or in combination with external beam radiation therapy (EBRT) and/or hormonal therapy has become a standard of care among the many treatment options available for the definitive treatment of prostate adenocarcinoma [1]. Disease-free survival with this technique is extremely high for early and intermediate risk disease and might be superior to EBRT alone [2]. In many respects, balancing the toxicity profile of this treatment against the therapeutic gains is the most important decision for a patient choosing this modality. Detailed prospective quality of life studies have been performed comparing patient responses to different treatment modalities for prostate cancer including brachytherapy, EBRT and surgery showing an acceptable short-term toxicity profile compared with other treatments [3]. Prostate brachytherapy also has predictable well-documented low rates of long-term morbidity with regard to urinary, rectal and sexual functioning as well [4,5].

While extensive literature exists with regard to the frequencies and predictors of the complications associated with brachytherapy, less attention has been paid to the temporal sequence of late toxicities and when patients can reasonably expect to experience them for the first time. A short-term prospective analysis was performed to track the incidence and timing of different treatment modalities for early stage prostate cancer and found unique temporal sequences for different treatment modalities [6]. Another study explored the relationship between time as a continuous or categorical variable and gastrointestinal, genitourinary or erectile toxicity and found no association between these toxicities and time [7]. Most recently, a paper compared brachytherapy with brachytherapy and EBRT and published median times to onset for selected gastrointestinal and genitourinary toxicities [8]. However, there was insufficient follow-up in these studies to report clearly when a patient can expect to experience late symptoms for the first time nor were factors affecting time to onset clearly identified.

In order to educate clinicians and patients as well as possibly obtain a better understanding of the individual tissue responses to brachytherapy with or without EBRT, we explored the temporal sequence for patients to experience their first symptoms of rectal bleeding, haematuria or erectile dysfunction (ED) following treatment. Treatment factors affecting time to first onset were also analysed.


Patients definitively treated for biopsy-proven adenocarcinoma of the prostate from 1990 to 2007 who received a full 125I implant, a full 103Pd implant or a partial 103Pd implant followed by EBRT were selected from our database. 125I implants were prescribed to 160 Gy and full 103Pd implants were prescribed to 124 Gy. Partial 103Pd implants were prescribed to 100 Gy followed by EBRT to the prostate and seminal vesicles to 45 Gy. All implants were performed by a single physician (R.G.S.). Post-implant dosimetry was performed for all patients. Patients with less than 2 years of follow-up or who received a salvage implant following local recurrence were excluded from our analysis. Patients were followed prospectively by a physician with a standardized institutional questionnaire to address toxicities.

Patients were screened for the first onset of selected long-term toxicities as follows: for rectal bleeding, any patient who experienced grade 2 or higher rectal bleeding using the Radiation Therapy Oncology Group (RTOG) long-term toxicity scale following implant was recorded (Table 1). Patients with a known history of bleeding haemorrhoids with no formal diagnosis of radiation proctitis were excluded from analysis (n= 1). For haematuria, any reported grade 2 haematuria or above on the RTOG long-term toxicity scale, excluding the first 3 months immediately following the implant, was recorded as an event (Table 1). Patients with a known history of bladder cancer were excluded. Finally, for ED, any patient who received androgen deprivation therapy (ADT) as part of treatment or had previous ED was excluded from analysis (n= 1333). We utilized the physician-assigned Mount Sinai Erectile Function Score (MSEFS) to assess potency which has been validated using correlation studies with a patient-assessed potency tool (Table 1) [9]. First onset of ED was scored as a drop in the MSEFS from above to below 2 [10]. A patient was considered potent as long as he could maintain an erection sufficient for intercourse with or without use of a phosphodiesterase type 5 inhibitor.

Table 1. Toxicity scales
 Grade 1Grade 2Grade 3Grade 4
Radiation Therapy Oncology Group Late Radiation Morbidity Scoring Schema
 HaematuriaMicroscopic haematuriaIntermittent macroscopic haematuriaFrequent haematuriaSevere haemorrhagic cystitis
 Rectal bleedingSlight rectal discharge or bleedingExcessive rectal mucus or intermittent bleedingBleeding requiring surgeryNecrosis/perforation Fistula
Mount Sinai Erectile Dysfunction Score
 Erectile dysfunctionUnable to maintain an erectionAble to achieve an erection but insufficient for intercourseErection sufficient for intercourse but suboptimalNormal erection

Life tables were utilized to calculate actuarial incidence rates. Censoring was assumed to be continuous across all intervals. Graphical displays of time to first incidence were generated with the Kaplan–Meier method. Univariate analysis was performed using the Wilcoxon rank sum test. Cox regression was used for multivariate analysis to compare significant treatment factors from univariate analysis. National Institute of Standards and Technology (NIST) corrected implant D90 was used, and biological effective dose (BED) was calculated as shown elsewhere [11]. anova was used to compare attained age in the ED group vs the age of censored controls.


A total of 2046 patients met inclusion criteria (summarized in Table 2). Of these, 1353 patients received a definitive 125I implant, 12 received a full 103Pd implant and 681 were treated with a partial 103Pd implant followed by EBRT. The median age of patients at treatment was 67 (range 39–88), and the median length of follow-up was 6 years (range 2–17.7 years). Ninety-two patients (4.4%) experienced a grade 2 or greater rectal bleeding event and 168 (8.2%) experienced grade 2 or greater haematuria. Eight of the haematuria patients had a transurethral resection of the prostate prior to implant and 14 had one following implant. Of the 713 patients that did not receive hormone therapy, 219 (31%) experienced a drop in erectile function on the MSEFS.

Table 2. Patient characteristics
Median age (years)67
PSA (ng/mL) 
 ≤101532 (75)
 10–20367 (18)
 >20147 (7)
 1a–1c991 (48)
 2a406 (20)
 2b419 (21)
 2c172 (8)
 3a47 (2)
 3b or 3c11 (1)
Gleason score 
 ≤61373 (67)
 7450 (22)
 8–10224 (11)
 Implant + EBRT702 (34)
 Iodine1353 (66)
 Full palladium12 (1)
 Partial palladium681 (33)
 ADT1123 (55)
Median dose (Gy) 
 NIST D90 125I180.3
 NIST D90 103Pd115.5
 NIST D90 partial 103Pd105.0
 BED (Gy2)193.2

Significantly different temporal patterns of first onset were seen for these late toxicities. Kaplan–Meier survival curves provide a graphical illustration of the differences in time to first onset (Fig. 1). While rectal bleeding shows a relatively rapid time to onset with a plateau thereafter, haematuria and ED show a significant number of late events, well beyond 5 years. The rate of decline for haematuria and ED remains relatively constant over time after an initial rapid decline.

Figure 1.

Kaplan–Meier curves for time to first onset of (A) rectal bleeding, (B) haematuria and (C) erectile dysfunction.

Life tables were generated and used to calculate the incidence rates of the selected toxicities at different time points (Table 3). All toxicities showed relatively high incidence rates per 1000 persons during the first 2 years following radiation. However, while the incidence rates for rectal bleeding and haematuria significantly decreased by 5 years, the incidence rate for ED remained approximately as high from 5 to 10 years after radiation as it did from 2 to 5 years following radiation. The incidence of rectal bleeding after 5 years is exceedingly low.

Table 3. Incidence rates over time
Years post radiationIncidence of toxicity (cases per 1000)
Rectal bleedingHaematuriaErectile dysfunction

In order to present these data in a pragmatic format for clinicians and patients, we made the assumption that all toxicity events were captured within the scope of this study and calculated percentiles of events to occur within specified time frames (Table 4). The greatest difference between the groups exists for the latest events: while almost all rectal bleeding events occurred within 5 years, approximately 20% of cases of haematuria and ED happened more than 5 years after implant.

Table 4. Percentiles and time frame for patients to experience symptoms
 Time for percentiles to experience first symptom (years)Percentage of patients with symptoms who experience first symptom within given time (years)
Rectal bleeding0.81.73.024587995
Erectile dysfunction0.62.05.428496178

Using the Wilcoxon rank sum test, we performed univariate analysis to identify factors affecting the time of first onset of toxicity (Table 5). Significant univariate variables were then utilized to create Cox regression models for each of the toxicities. For rectal bleeding, the only significant variable affecting the time to first onset was the addition of EBRT to brachytherapy (hazard ratio [HR] 2.3, 95% CI 1.55–3.51). Time to onset of haematuria was significantly influenced by the addition of EBRT (HR 1.6, 95% CI 1.16–2.18) and a BED greater than 200 Gy2 (HR 1.6, 95% CI 1.16–2.20). The only factor to affect the time to onset of ED was the patient's age at treatment (HR 3.0, 95% CI 2.39–3.75).

Table 5. Univariate analysis
 Wilcoxon rank sum P 
HaematuriaRectal bleedingErectile dysfunction
Age > 65 years0.9510.546<0.0001
Brachytherapy ± EBRT<0.0001<0.00010.042
Prostate cancer risk group0.5580.224<0.0001
NIST adjusted D90 > 150 Gy0.2980.103<0.0001
BED > 200 Gy20.0030.1720.075

In order to clarify the role of age in the time to onset of ED, we compared the attained age of patients with ED vs that of the patients censored at last follow-up without an ED event using anova. The median age for patients with ED was 69.4 years vs 66.9 for those patients censored without ED, suggesting a significant difference between groups (< 0.001).


The analyses in this paper demonstrate that late toxicities following radioactive prostate implantation with or without EBRT follow unique temporal patterns in terms of time to first onset. While rectal bleeding tends to occur relatively early within the first several years following treatment, haematuria and ED are more indolent in onset. These data are best reflected in the variable incidence rates for each of these toxicities. Rectal bleeding shows a fall in incidence after 2 years and a very low incidence after 5 years. In contrast, haematuria does not decrease in incidence until after 5 years. Most striking, ED has fairly constant incidence from 2 to 10 years.

The phenomenon of differential rate of onset for late radiation toxicities was described with regard to intracavitary brachytherapy for cervical cancer where it was found that different tissues had unique patterns of onset of late toxicity [12]. It is difficult to make direct comparisons between the studies in the light of the differences in technique, radiation dose, sex and average ages of the patients involved. However, the data are consistent in showing a rapid fall off in incidence for late rectal toxicity and a slower decline in incidence rate for urinary morbidity. Of note, the median time of onset for toxicity in our study was longer than that of a recent publication [8]. However, the median overall follow-up in our analysis is longer, probably pushing the median time to onset to a later point.

The only treatment variable affecting the time to first onset of rectal bleeding was the addition of EBRT to the seed implant. This effect might result from the additional posterior dose distribution when supplementing with EBRT or the longer number of days exposed to radiation which might prolong the healing process and make the rectum more vulnerable to late injury. From a radiobiological perspective, late rectal bleeding probably represents mucosal injury. Significant fibrosis has been observed in studies of rat intestinal mucosa following radiation [13]. Combination therapy might aggravate this process. Of note, a higher absolute late gastrointestinal toxicity rate with the addition of EBRT to brachytherapy has been documented elsewhere, possibly supporting these hypotheses [8]. We have demonstrated that using intensity-modulated radiation therapy or image-guided radiation therapy can decrease the risk of rectal bleeding as opposed to three-dimensional conformal radiation [14]. Likewise, there are reports of using space creating injections to decrease the risk of rectal bleeding [15]. However, it is difficult to assess how these advances will affect the time to first onset which this analysis attempted to characterize.

In contrast, the time to first onset of late haematuria was affected by the BED of radiation treatment and the addition of EBRT. Post-implant haematuria is a complex process that we hypothesize might represent necrosis in the prostatic urethra. On post-implant cystoscopy, we often find areas of oozing from necrotic sites in the prostatic urethra (unpublished results). The timing for this necrosis appears to be a much slower process than that for rectal bleeding reflecting a different mechanism of action.

Perhaps the most interesting finding from this analysis is the relatively slow time to onset of ED. ED is hypothesized to represent damage to the neurovascular bundles articulating around the prostate. The mechanism for neuronal cell death is complicated but probably involves cell cycle machinery after DNA damage [16]. Given the slow division time of neurons, a slow time to first onset of ED is plausible.

However, multivariate analysis found the only significant factor affecting time to onset was the age of the patient at implant. This was further corroborated by finding the age at onset of first ED to be higher than the median age at last follow-up in the control group. The increase in incidence of ED as men age from 40 to 70 is a well documented phenomenon from the Massachusetts Male Aging Study [17]. The statistically significant dependence on age for the time of onset of ED in this study might be a reflection of this fact. If this hypothesis is correct, the reported values for radiation-related ED might overestimate true treatment-related toxicity. In the absence of pathognomonic findings for radiation injury, identifying true radiation-related ED vs normal aging might prove very difficult. In contrast, another recent publication did not find an increase in ED with time since treatment [18]. However, that study looked at a single time point with a questionnaire sent to patients and perhaps missed the spectrum of reported ED events observed with longitudinal analyses.

Limitations of this study include the fact that our analysis began before the use of accepted toxicity scales and reports values based on less commonly utilized scales. In addition, as with any retrospective analysis, the potential for selection bias is present in this sample pool. Finally, to record percentages of events accrued at certain time points, we had to assume that we had captured all events at the furthest follow-up points which might underestimate the true frequencies.

Regardless of aetiology, this paper provides a useful tool for physician and patient education. Consistent with prior published results, the majority of symptoms happen early within the first year to 2 years following implant [6,7]. Still, a significant number of patients will not experience first onset of symptoms until many years following implant and should be educated accordingly. This might reflect the biological differences of late toxicity or be a byproduct of the extended and rigorous screening of this study population. A further assessment of the temporality of long-term toxicity might be better able to determine what long-term effects can truly be linked to treatment.


Nelson Stone is a Consultant for Nihon Mediphysics, Ferring, Amgen, and Janseen.