• prostate cancer;
  • gastrointestinal toxicity;
  • radiation


  1. Top of page
  2. Abstract


The current study was designed to determine rates and predictors of late, lower gastrointestinal toxicity after radiation therapy in a population-based cohort of older men with prostate cancer.


The study population consisted of men with localized or regional stage prostate cancer who were age ≥66 years and were diagnosed between 1992 and 1999 who were identified in the Surveillance, Epidemiology, and End Results (SEER)-Medicare database. Gastrointestinal diagnoses were ascertained through claims from 6 to 60 months after diagnosis. The relative rates of diagnoses in the radiation group versus the nonradiation group were used as a means of estimating toxicity from radiotherapy. Cox modeling was used to determine factors associated with gastrointestinal diagnoses.


A total of 57,955 men were included, 24,130 of whom were treated with radiation therapy. Among patients with 5 years of follow-up, the rates of gastrointestinal diagnoses were 19.4% higher in irradiated patients than among patients who did not have local therapy. Hemorrhage was the most common diagnosis, and was increased by 18.9% among patients treated with radiation (39.6% of irradiated patients vs. comparison rates of 18.2% in patients treated with radical prostatectomy and 20.7% in patients with no local therapy). Diagnostic lower endoscopies were performed in an additional 20.9% of men (32.4% of men treated with radiation vs. 12.7% of men who underwent prostatectomy). In all, 4.4% of irradiated men were hospitalized with a gastrointestinal diagnosis versus comparison rates of 3.2% among men with no local therapy. In multivariate models, increasing patient age, hormonal therapy, comorbidity, diabetes, peripheral vascular disease, and hemorrhoids were all associated with gastrointestinal diagnoses consistent with toxicity, whereas tumor stage and grade were not predictors.


Lower gastrointestinal toxicity after radiation therapy for prostate cancer continues for at least 5 years and may be more common than previously reported. Cancer 2006. © 2006 American Cancer Society.

Prostate cancer is the most common noncutaneous malignancy among men in the U.S., and localized prostate cancer is often curable by radical prostatectomy or radiation therapy.1 Survival for patients with prostate cancer has been improving, with a 5-year relative survival rate of 99% for patients diagnosed between 1995 and 2000.1, 2 With increasing survival after diagnosis, the assessment of late side effects related to definitive therapies has become increasingly important.

External beam radiation therapy is effective for the treatment of prostate cancer. Gastrointestinal (GI) toxicity related to irradiation of the adjacent rectum, sigmoid colon, or bowel is a concern for patients and physicians. The most common toxicity is rectosigmoid radiation proctitis and often may manifest as rectal bleeding. Prior reports suggest that rates of any severe toxicity related to external beam radiation is less than 2%, but rates of less-severe rectal bleeding may be substantially higher.3, 4

Studies have also shown that higher doses of radiation may improve disease-free survival but at the cost of increased lower GI toxicity.5–9 Advances in radiation treatment planning and delivery over the past 10–15 years, such as 3-dimensional conformal radiation therapy (3D-CRT) and intensity-modulated radiation therapy (IMRT), have aimed at optimizing dose delivery to the target while minimizing the exposure to sensitive nearby structures like the rectum. These techniques have allowed higher doses of radiation to be delivered with lower GI side effects as long as the rectal dose is limited.7, 9, 10

Prior data on GI toxicity after radiation therapy have primarily been generated from clinical trials, which include highly selected patients treated at specialized centers.11–13 Another approach to determining toxicities has been surveys of treated patients.4, 14, 15 Whereas surveys may study a more representative patient population, they are limited by response bias and recall bias. The current study examines the incidence of GI-related diagnoses after local therapy in a large observational population-based cohort from the Surveillance, Epidemiology, and End Results (SEER)-Medicare database of men at least 66 years of age. Patients treated with radiation therapy were the primary focus of the analysis, but men who received surgery or no definitive local therapy were included to provide comparative rates of GI diagnoses.


  1. Top of page
  2. Abstract

Data Source

We used the SEER-Medicare linked database for this study. The SEER program is a national population-based tumor registry that collects information on incident cancer cases. Medicare is the primary health insurer for 97% of the U.S. population age ≥65 years. Under an agreement between the National Cancer Institute and the Center for Medicaid and Medicare Services (CMS), SEER subjects who are eligible for Medicare have been linked to their Medicare records.16 Of persons who are reported by SEER as diagnosed with cancer at age ≥65 years, 93% were matched with their Medicare enrollment records. At present, cancer subjects diagnosed from 1973 to 1999 have been linked, and their complete Medicare claims are available from 1991 to 2002.

Study Population

The study population consisted of men in the SEER-Medicare database who were age 66 years and who had local or regional stage prostate cancer diagnosed between 1992 and 1999. A total of 97,241 subjects were identified using these criteria. To ensure complete information, patients not enrolled in both Medicare Part A and Part B from 12 months before to 6 months after their cancer diagnosis (9208 cases), patients who were members of a Health Maintenance Organization (21,008 cases), or patients who were diagnosed by autopsy or on a death certificate (229 cases) were excluded. Patients who received surgery or first radiation more than 6 months after diagnosis were excluded (4926 cases), as were patients who received both surgery and radiation (1110 cases), and those who received surgery or radiation treatment without specified dates (259 cases). Patients who were treated with chemotherapy were also excluded (2546 cases) due to potential toxicities of chemotherapy. Patients receiving brachytherapy were included and were grouped with the other patients receiving radiation therapy. Thus, a total of 57,955 patients were included in the analyses.

Patient demographics, dates of diagnosis, and extent of disease are available through the SEER registry data and are found in the SEER-Medicare PEDSF file. Patient characteristics were compared between those patients treated with radiation, surgery, and no treatment using the chi-square test for categorical variables and Student t test for continuous variables. Education was characterized by percentage of individuals who lived in a census tract area with less than 12 years education. Poverty was measured by percentage of individuals living in a given census tract living below the poverty level. The cut-off points for education and poverty were selected as the integers that were closest to the quartile of education and poverty. Comorbidity scores were calculated by the Klabunde modification of the Charlson method.17, 18

GI diagnoses were ascertained through claims up to 60 months after diagnosis and are presented separately for those patients treated with radiation, surgery, or those without local therapy.19 The specific International Classification of Diseases, 9th revision (ICD-09) and Current Procedural Terminology (CPT) codes used are shown in Table 1. We included diagnoses classified as follows: 1) “Any gastrointestinal diagnosis,” defines as relevant GI diagnoses identified on inpatient (Medical Provider Analysis and Review [MEDPAR] file), outpatient (Outpatient file), or physician (National Claims History [NCH] file) claims; 2) “hospitalizations,” defined as GI diagnoses serious enough to require hospitalization (MEDPAR file); and 3) “procedures,” defined as GI procedures or surgical interventions (MEDPAR, Outpatient, and NCH files). For “any diagnosis” and “procedures,” all relevant claims were included, regardless of ICD-09 diagnosis position. For “hospitalizations,” only claims in the first or second positions were counted.

Table 1. ICD-09 and CPT Diagnosis and Procedure Codes
 ICD-09 and CPT codes
  1. ICD-9 indicates the International Classification of Diseases, 9th revision; CPT, Current Procedural Terminology; GI, gastrointestinal.

Any lower GI diagnosis
 Diarrhea, noninfectious enteritis, colitis787.91, 555, 556, 557, 558
 Hemorrhage (GI hemorrhage and hemorrhage of rectum and anus)578, 569.3
 Rectal/anal fissure, abscess, stenosis565, 566, 569.2
 Other unspecified anal, rectal, and intestinal disorders569.4, 569.8
 Hospitalization787.91, 555, 556, 557, 558, 578, 569.3, 565, 566, 569.2, 569.4, 569.8
 Endoscopies (proctosigmoidoscopy, colonoscopy, anoscopy)45.23, 48.23, 49.21, 45300-45345, 45355-45387, 46600-46615
 Biopsies (rectal biopsy, cecal biopsy, and anal biopsy)48.24, 45.25, 49.23
 Resections/major surgical interventions (colectomy, laparoscopic colectomy, proctectomy, colostomy)46.10, 48.62, 46.14, 46.03, 46.13, 46.11, 44140-44160, 44204-44212, 45110-45123
 Procedures for strictures (dilation of rectal stricture or anal sphincter, anoplasty for stricture)96.22, 96.23, 45905, 45910, 46700
 Argon laser destruction and electrocautery48.33, 48.31, 48.32
 Other surgical interventions (e.g., Proctoplasty for stenosis, closure of fistula, fistula repair, anal sphincter repair)48.93, 58.43, 57.83, 48.73, 49.1, 49.6, 49.7, 48.4 -48.7, 48.9, 49.73, 45000, 45005, 45020, 45500, 45800-45805, 45820-45825

Statistical Analyses

We plotted the percentage of patients with any GI diagnosis for each month starting 12 months before diagnosis through 60 months after diagnosis. A clear difference was apparent between diagnoses occurring around the time of radiation and surgery and those that emerged later. Therefore, the diagnoses were divided into categories of “early” and “late,” consistent with known patterns of treatment toxicity. Early diagnoses were defined as occurring within 6 months after first radiation, after radical prostatectomy, or after diagnosis for those patients who did not have definitive therapy. Late diagnoses were defined as occurring from 6 to 60 months after first radiation, after radical prostatectomy, or after diagnosis, respectively.

In a subset of 24,820 patients who were diagnosed 1992–1996 and had complete follow-up for 60 months after diagnosis, we calculated the percentages of patients with each diagnosis. Differences by local treatment were compared using the chi-square test. Using the entire patient population, Cox proportional hazards models were used to determine factors associated with any late diagnosis and any late diagnosis requiring hospitalization. The overall model included local therapy, androgen deprivation, age at diagnosis, race, stage, grade, year of diagnosis, SEER region, education, poverty, comorbidity index, hemorrhoids, prediagnosis toxicity, and early toxicity. To determine which factors predicted late GI diagnoses among patients treated with radiation, Cox modeling was performed on the 24,130 radiation-treated patients. An additional model, which included the specific comorbidities of diabetes, myocardial infarction, peripheral vascular disease, and hemorrhoids instead of a comorbidity score was also performed, because previous work has suggested that these conditions may be associated with radiation toxicity.20–23 Adjusted time-to-event curves were estimated from stratified Cox proportional hazards models. All P-values are 2-sided. Statistical analyses were performed using SAS software (SAS Institute, Inc., Cary, NC).


  1. Top of page
  2. Abstract

A total of 57,955 men with localized and regional stage prostate cancer were included in this study, 24,130 of whom (41.6%) were treated with radiation therapy, 11,918 (20.6%) underwent radical prostatectomy, and 21,907 (37.8%) received neither surgery nor radiation. Significant differences in patient characteristics are apparent between those patients who received irradiation, those who had surgery, and those without local therapy (Table 2). Generally, the patients treated with radiation were older and had a higher burden of comorbidities than those patients treated with radical prostatectomy, but were younger and with fewer comorbidities than those patients who did not receive any local therapy. Similarly, the use of androgen deprivation among patients treated with radiation (28.8%) was intermediate between patients treated surgically (12.4%) and those who had neither treatment (34.0%).

Table 2. Patient Characteristics
  Radiation (n = 24,130), %Radical prostatectomy (n = 11,918), %Neither (n = 21,907), %P
  1. SEER indicates the Surveillance, Epidemiology, and End Results program.

Age, y66–6921.650.511.7<.0001
Clinical stageT1/T292.588.293.9<.0001
GradeWell-differentiated Gleason 2-411.38.421.1<.0001
Moderately differentiated Gleason 5-764.472.052.2
Poorly differentiated Gleason 8-1019.618.519.1
EthnicityNon-Hispanic white82.484.677.3<.0001
SEER regionSan Francisco7.18.07.9<.0001
New Mexico4.15.85.9
San Jose3.14.34.6
Los Angeles10.618.813.8
Year of diagnosis199216.421.314.4<.0001
Census tract education<10;24.227.919.6<.0001
(percentage of adults with <12 y education)10 to <2035.037.433.3 
20 to <3024.421.525.4
Census tract poverty<322.819.416.3<.0001
(percentage living below poverty line)3 to <732.533.029.4 
7 to <1425.629.328.5
Comorbidity index075.585.068.7<.0001
Myocardial infarctionNo99.399.599.0<.0001
Peripheral vascular diseaseNo98.099.297.3<.0001
Androgen deprivationNo71.287.666.0<.0001

Figure 1 shows the percentage of radiation-treated patients presenting with lower GI diagnoses for each month, staring 12 months before diagnosis and extending to 60 months after diagnosis. As a comparison, separate curves are shown for those patients treated with radical prostatectomy and for those with no local therapy. For each population of patients, there is an increase in diagnoses potentially relating to GI toxicity around the time of prostate cancer diagnosis and the initial treatment (early GI diagnoses). Approximately 6 months after diagnosis, a more marked increase in GI diagnoses is seen among those patients who were treated with radiation (late GI diagnoses). The late GI diagnoses appear to peak around 18 to 24 months after diagnosis and then decline. However, the percentage of patients with lower GI diagnoses remains persistently higher for irradiated patients out to 5 years.

thumbnail image

Figure 1. Percentage of patients each month with a Medicare containing gastrointestinal (GI) diagnosis, stratified by local therapy (radiation, radical prostatectomy, or none).

Download figure to PowerPoint

We next determined the percentage of patients with specific lower GI diagnoses within 5 years of diagnosis. For this analysis, we limited the patient population to the 24,820 patients who had complete follow-up for 5 years (Table 3). Within the first 6 months of diagnosis or treatment, 6.9% of men treated with radiation, 6.1% of those treated with surgery, and 6.3% of those without local therapy had a lower GI diagnosis (P = 0.07). Overall, men treated with radiation had an absolute increase in GI diagnoses of 19.4% in the period between 6 and 60 months after diagnosis (51.3% irradiated men vs. comparative rates of 28.6% of men treated with radical prostatectomy and 31.9% of patients who had neither surgery nor radiation). The most common diagnosis was “hemorrhage”; rates were 18.9% higher among irradiated men (39.6% of irradiated patients vs. 20.7% of patients who did not receive local therapy). Only 4.4% of patients who received radiation experienced lower GI diagnoses serious enough to require hospitalization over this 5-year period, as compared with 3.2% of men who did not receive local therapy. Diagnostic endoscopies (colonoscopy or sigmoidoscopy) were commonly performed in patients treated with radiation (32.4% of patients, compared with 12.7% of surgically treated patients and 11.5% of patients with no local therapy). The percentage of patients who required major GI surgical interventions or resections did not differ between the treatment groups. As a further control, we also examined a GI diagnosis that would not be expected to result from radiation for prostate cancer, upper GI bleed. There was no increase in diagnoses of upper GI bleeds relative to patients not receiving treatment (11.2% after radiation and 11.9% after no treatment).

Table 3. Percentages of the 24,820 Patients with 60 months of Complete Follow-up Who Had Gastrointestinal (GI) Diagnoses, Hospitalizations, and Procedures
Primary therapy radiation (n = 10304) Radical prostatectomy (n = 6700) None (n = 7816)12 months before diagnosis6-60 months after diagnoses (late)Net increase in late GI diagnoses
  1. GI indicates gastrointestinal.

Any lower GI diagnosis  <.0001 <.000119.4
Radical prostatectomy8.228.6
Diarrhea, noninfectious enteritis, and colitis  .131 <.00019.4
Radical prostatectomy2.611.1
Hemorrhage  <.0001 <.000118.9
Radical prostatectomy5.018.2
Rectal, anal fissure, abscess, and stenosis  .098 <.00011.7
Radical prostatectomy0.62.0
Other unspecified anal, rectal, intestinal  .435 <.000115.5
Radical prostatectomy0.94.5
Any lower GI hospitalization  .0002 <.00011.2
Radical prostatectomy0.11.6
Any lower GI procedures  .042 <.000120.8
Radical prostatectomy3.013.0
Endoscopies  .046 <.000120.9
Radical prostatectomy3.012.7
Biopsies  .362 <.00014.8
Radical prostatectomy0.21.4
Resections/major surgical interventions  .342 .3450.1
Radical prostatectomy0.00.3
Strictures  .625 .0010.2
Radical prostatectomy0.00.1
Interventions (argon laser destruction or electrocautery)  .270 <.00010.6
Radical prostatectomy0.00.0
Other surgical interventions  .307 .5660
Radical prostatectomy0.00.2

We then performed multivariate analyses to determine factors associated with late lower GI diagnoses. Primary therapy with radiation was associated with higher risk of any lower GI diagnosis (hazards ratio [HR], 1.84; 95% confidence interval [95% CI], 1.78–1.90) and diagnoses requiring hospitalization (HR, 1.36; 95% CI, 1.23–1.51). Adjusted time-to event curves are illustrated in Figure 2. A separate multivariate model was calculated for the 24,130 men treated with radiation to determine which factors were associated GI diagnoses and hospitalizations among irradiated men (Table 4). In this model, the use of androgen deprivation therapy was associated with a slightly higher hazard of any GI diagnosis (HR, 1.07; 95% CI, 1.02–1.12) and diagnoses requiring hospitalization (HR, 1.22; 95% CI, 1.05–1.41). Older patients were at higher risk of GI diagnoses, with a 51% higher hazard of hospitalization for each 10-year increase in age. Black men had a higher hazard of diagnoses requiring hospitalization when compared with white men (HR, 1.49; 95% CI, 1.19–1.86). Patients who had a diagnosis of GI disease in the 12 months preceding the prostate cancer diagnosis were more likely to develop subsequent GI diagnoses (HR, 1.59; 95% CI, 1.50–1.68 for any diagnosis; and HR, 2.60; 95% CI, 1.39–4.87 for hospitalization), as were patients who experienced early treatment toxicity (HR, 2.10; 95% CI, 1.97–2.23 for any diagnosis; and HR, 3.50; 95% CI, 2.10–5.85 for hospitalization).

thumbnail image

Figure 2. Adjusted time-to-event curves for (A) time until any gastrointestinal (GI) diagnosis and (B) time until any GI hospitalization.

Download figure to PowerPoint

Table 4. Multivariate Analysis of Factors Associated with Any Late GI Diagnosis or Hospitalization with a Late GI Diagnosis among 24,130 Patients with Prostate Cancer Treated with Primary Radiation Therapy
  Any late GI diagnosis adjusted HR (95% CI)Any late GI hospitalization adjusted HR (95% CI)
  1. GI indicates gastrointestinal; HR, hazards ratio; 95% CI, 95% confidence interval.

Age at diagnosisEach 10-year increase in age1.15 (1.10–1.20)1.51 (1.31–1.73)
EthnicityNon-Hispanic White1.001.00
Black0.93 (0.86–1.01)1.49 (1.19–1.86)
Hispanic1.00 (0.88–1.14)1.15 (0.79–1.68)
Other1.04 (0.93–1.16)1.32 (0.94–1.84)
Year of diagnosis19921.001.00
19930.94 (0.88–1.01)0.91 (0.73–1.13)
19941.03 (0.96–1.11)0.84 (0.67–1.06)
19950.99 (0.92–1.07)0.92 (0.73–1.17)
19961.01 (0.93–1.09)0.98 (0.77–1.23)
19971.06 (0.99–1.14)1.05 (0.83–1.32)
19981.03 (0.95–1.12)0.97 (0.76–1.25)
19991.12 (1.04–1.22)0.87 (0.65–1.16)
T3/T41.05 (0.98–1.14)1.13 (0.90–1.42)
GradeWell-differentiated Gleason 2-41.001.00
Moderately differentiated Gleason 5-71.02 (0.95–1.08)1.02 (0.83–1.25)
Poorly differentiated Gleason 8-101.02 (0.95–1.10)1.08 (0.85–1.36)
Androgen deprivationNo1.001.00
Yes1.07 (1.02–1.12)1.22 (1.05–1.41)
Toxicity in the 12 months before diagnosisNo1.001.00
Yes1.59 (1.50–1.68)2.60 (1.39–4.87)
Early toxicityNo1.001.00
Yes2.10 (1.97–2.23)3.50 (2.10–5.85)
Comorbidity index01.001.00
11.22 (1.16–1.29)1.56 (1.33–1.83)
21.16 (1.05–1.29)2.02 (1.55–2.65)
≥31.35 (1.24–1.47)2.72 (2.22–3.34)

Patient comorbidity score was strongly associated with hazard of lower GI diagnoses. Even those patients with comorbidity scores of 1 had higher risk of GI diagnoses than those patients without any significant comorbidities (HR, 1.22; 95% CI, 1.16–1.29). The risk of GI disease increased with increasing comorbidity. In a separate model, we substituted the specific comorbidities of 1) diabetes, 2) myocardial infarction, 3) peripheral vascular disease, and 4) hemorrhoids for the Charlson comorbidity score to determine the impact of these specific comorbidities. Patients with a diagnosis of diabetes had an increased risk of any GI diagnosis (HR, 1.12; 95% CI, 1.05–1.19) and diagnoses requiring hospitalization (HR, 1.60; 95% CI, 1.35–1.90). No significant association was seen for patients with a history of myocardial infarction. Patients with a diagnosis of peripheral vascular disease were at increased risk for developing GI diagnoses after radiation (HR, 1.29; 95% CI, 1.14–1.47 for any toxicity; and HR, 1.59; 95% CI, 1.12–2.25 for hospitalization). Patients with hemorrhoids had an increase in risk of any GI diagnosis (HR, 1.26; 95% CI, 1.16–1.37) but not for hospitalization (HR, 0.99; 95% CI, 0.74–1.31).


  1. Top of page
  2. Abstract

In this study we evaluated late lower GI toxicity after radiation therapy. The study had the advantage of a population-based cohort of nearly 25,000 men with complete follow-up for 5 years. Among the over 10,000 patients treated with radiation, greater than half had outpatient visits for diagnoses suggestive of lower GI toxicity. Lower GI hemorrhage, which was the most common diagnosis, was reported in 18.9% more men who received radiation as compared with men treated with no local therapy. GI diagnosis peaked at 18 to 24 months, with a subsequent leveling off consistent with the time-course of radiation toxicity. Diagnostic procedures, particularly endoscopy, were also elevated among men treated with radiation, with approximately one-third of irradiated men having a diagnostic lower endoscopy over the subsequent 5 years. Despite these high rates of office visits for GI diagnoses and procedures, the toxicity appears to have been mild because only 4% of irradiated patients required hospital admission for a lower GI diagnosis.

These rates of lower GI toxicity are higher than reported in previous studies.4, 14, 24, 25 For instance, in a survey of 621 Medicare beneficiaries treated with radiation for prostate cancer, only 10% reported problems with bowel dysfunction.4 Similarly, in a prospective cohort study of 410 men, only 9.4% of irradiated patients reported bowel problems at 24 months of follow-up.25 Few previous studies have evaluated toxicity at 5 years of follow-up.14, 24 In the Prostate Cancer Outcomes Study of 1187 men with prostate cancer, of whom 286 were treated with radiation, 29% of irradiated men reported diarrhea and 33% reported bowel urgency at 5 years after treatment.14

It is difficult to directly compare results of patient surveys and toxicities assessed via medical claims, because they measure different aspects of GI toxicities and compare patient-reported symptoms with physician-reported diagnoses. A claims-based approach may ascertain more mild toxicities, ones that are noted by a physician but are not particularly bothersome to the patients, particularly if physicians have incentives to document all diagnoses, however minor. For instance, the diagnosis of “hemorrhage” could reflect hemorrhoidal bleeding rather than more serious GI bleeding. However, claims data also may more accurately capture toxicities over time, especially those symptoms that are transient. An additional benefit of claims data is that baseline rates of GI diagnoses can be addressed through the comparison groups of untreated and surgically treated patients, which allow estimation of absolute increases in risk.

Although the toxicity relating to radiation treatment parameters, such as dose and volume, has been extensively studied, to our knowledge less attention has been paid to individual patient characteristics that may predispose to treatment toxicity. In this study the patient factors of age and comorbidities predicted lower GI toxicity after radiation. Older patients were at higher risk of developing GI diagnoses after radiation, even after controlling for baseline comorbidity and baseline GI diagnoses. For each 10-year increase in age, men had a 15% increase in any GI diagnoses and a 51% increase in risk of hospitalization for a GI diagnosis. Previous studies on predictors of toxicity have had conflicting results, with some studies showing an increase in risk of GI toxicity with increasing age20, 22 and others showing no association.5, 26 However, age should not be considered a contraindication to radiation therapy, as older patients also have increased surgical risks.

Patients with a higher burden of comorbidity and those with the specific comorbidities of diabetes, peripheral vascular disease, and hemorrhoids were also at significantly higher risk of developing subsequent GI diagnoses, although patients with a previous history of myocardial infarction were not. Among patients with diabetes and vascular disease, the likely mechanism of an increase in toxicity is related to impaired repair of radiation-damaged tissue. These findings are consistent with previous studies that suggested that patients with diabetes and hemorrhoids may be at increased risk of toxicity.20–23, 26

Treatment with hormone therapy was a significant predictor of GI diagnoses. This association has been previously seen in smaller studies and to our knowledge is of unclear etiology.5, 27, 28 It is possible that patients receiving hormone therapy had worse features and therefore were treated with larger fields or a higher dose, both of which would be expected to produce more side effects. Our analyses adjust for T classification and for Gleason score, but residual selection biases may still be present. However, neither tumor grade nor stage were significant predictors of toxicity, which suggests that selection biases are not operating here. Another possibility is that androgen deprivation therapy has an impact on GI toxicity by unknown mechanisms, such as prostate shrinkage, exposing more of the rectum to radiotherapy.5, 27

Our study has some limitations. We ascertained diagnoses consistent with toxicity through Medicare claims. Our rates of diagnoses are likely an underascertainment of all toxicity, as some patients may have toxicities that are not captured through billing codes. We do not have any information on pretreatment prostate-specific antigen levels. In addition, no details regarding radiation treatment were available. We do not know the radiation fields or doses, which are both important to a complete understanding of toxicity. There have been changes in radiation therapy for prostate cancer over the 1990s, both in technique and dose. This study probably covered a period when many patients were treated using “conventional” treatment planning in the early part of the decade and “3-dimensional-conformal” treatment planning (computer tomography-based), which has lower toxicity, in the latter part of the decade.10, 12, 29 However, no clear decrease in toxicity over time was apparent in either the unadjusted or multivariate analyses.

In conclusion, we found that a high proportion of older men developed lower GI diagnoses over the 5 years after radiation for prostate cancer, although the severity of the GI disease appears to have been low. Older patients, patients with diabetes, hemorrhoids, or peripheral vascular disease, and those patients treated with hormonal therapy appear to be at higher risk for treatment toxicity after radiation therapy. Given that most men with localized prostate cancer will have the choice of radical prostatectomy or definitive radiation therapy as their primary treatment modality, it is important to have accurate information on the long-term toxicities of therapy. This study illustrates the persistent late lower GI toxicities associated with radiation therapy. These toxicities, while rarely serious enough to result in hospitalization, could negatively impact quality of life in prostate cancer survivors.


  1. Top of page
  2. Abstract