|Home | About | Journals | Submit | Contact Us | Français|
Radiation therapy is commonly used to treat localized prostate cancer; however, representative data regarding treatment-related toxicities compared with conservative management are sparse.
To evaluate gastrointestinal (GI) toxicities in men treated with either primary radiation or conservative management for T1–T2 prostate cancer.
We performed a population-based cohort study, using Medicare claims data linked to the Surveillance Epidemiology and End Results data. Competing risk models were used to evaluate the risks.
GI toxicities requiring interventional procedures occurring at least 6 mo after cancer diagnosis.
Among 41 737 patients in this study, 28 088 patients received radiation therapy. The most common GI toxicity was GI bleeding or ulceration. GI toxicity rates were 9.3 per 1000 person-years after three-dimensional conformal radiotherapy, 8.9 per 1000 person-years after intensity-modulated radiotherapy, 5.3 per 1000 person-years after brachytherapy alone, 20.1 per 1000 person-years after proton therapy, and 2.1 per 1000 person-years for conservative management patients. Radiation therapy is the most significant factor associated with an increased risk of GI toxicities (hazard ratio [HR]: 4.74; 95% confidence interval [CI], 3.97–5.66). Even after 5 yr, the radiation group continued to experience significantly higher rates of new GI toxicities than the conservative management group (HR: 3.01; 95% CI, 2.06–4.39). Because our cohort of patients were between 66 and 85 yr of age, these results may not be applicable to younger patients.
Patients treated with radiation therapy are more likely to have procedural interventions for GI toxicities than patients with conservative management, and the elevated risk persists beyond 5 yr.
Patients who choose to undergo prostate cancer therapy have several options, including radical prostatectomy, radiation therapy, and androgen-deprivation therapy (ADT). Similar to the other treatment modalities, radiation therapy carries risks of significant side effects. For example, a randomized controlled trial of radiation dose escalation (78 Gy vs 70 Gy) revealed that patients receiving the higher dose had incidences of grade 2 or 3 rectal toxicity of 19% and 7%, respectively . Stone and Stock reported that brachytherapy conferred a risk of rectal bleeding of 24% in years 1–3 and a 2.8% risk of bleeding beyond 5 yr .
Further complicating the decision-making process is the fact that for most patients, prostate cancer is typically a slowly progressing disease that is diagnosed in older men, so conservative management is often appropriate. Understanding the potential side effects of various treatments is critical for decision making; however, data on long-term gastrointestinal (GI) toxicities following different radiation modalities are sparse. This population-based study investigates the long-term risk of GI toxicities that require intervention following radiation therapy for localized prostate cancer.
The main data source for this study is the Surveillance Epidemiology and End Results (SEER) –Medicare linked database, a unique resource that combines cancer, clinical, and sociodemographic data. The SEER cancer registry ascertainment rate exceeds 98% . Multiple studies have verified the validity of the Medicare database [4–6]. The agreement between SEER and Medicare data with regard to radiation treatments for prostate cancer is 93% . The study cohort consisted of patients 66–85 yr of age diagnosed with T1–T2 clinically localized prostate cancer between 1992 and 2005 (n = 181 171) who were residents of the SEER regions. They must have enrolled in both Medicare Parts A and B for the 12 mo before cancer diagnosis. Exclusion criteria included having another cancer before prostate cancer (n = 46 522), metastasis within 6 mo of cancer diagnosis (n = 2725), palliative radiation treatment within 12 mo of diagnosis (n = 3342), cryotherapy or radioisotope therapy (n = 838), repeated brachytherapy (n = 104), primary ADT not combined with radiotherapy (n = 7119) or radical prostatectomy (n = 13 060) in the first 12 mo after diagnosis, unless surgery was used after radiation treatment. Also excluded were patients who had existing GI toxicity in the year before diagnosis (n = 80). Enrollment in a health maintenance organization (HMO), having private health insurance, or coverage through the Veterans Administration during the study period (n = 65,644) were also exclusion criteria because these health agencies have historically not been required to submit claims data to Medicare, resulting in incomplete medical records in the SEER-Medicare database for these patients .
Our final cohort consisted of 28 088 patients treated with radiation therapy and 13 649 patients who were conservatively managed (no surgery, radiation therapy, or hormone therapy for at least a year after prostate cancer diagnosis). The radiation therapy cohort received radiation therapy (as defined by Current Procedural Terminology [CPT] codes in Addendum 1) within 1 yr of prostate cancer diagnosis. Radiation therapy was categorized as external-beam radiation therapy (EBRT), brachytherapy, or a combination (EBRT plus brachytherapy). Of the patients receiving EBRT, patients were subdivided into three-dimensional conformal radiotherapy (3D CRT), intensity-modulated radiotherapy (IMRT), or proton beam therapy. The 3D CRT was defined as only 3D CRT. IMRT was defined as IMRT only or the combination of 3D CRT and IMRT. Proton beam therapy was defined as protons only or protons combined with 3D CRT or IMRT.
In the United States, the Common Terminology Criteria for Adverse Events and Radiation Therapy Oncology Group/European Organization for Research and Treatment of Cancer (RTOG/EORTC) Late Radiation Morbidity Scoring Schema are common toxicity reporting systems [8,9]. To illustrate, the RTOG/EORTC scoring system for late small/large intestine toxicity defines grade 1 as mild diarrhea or cramping, bowel movement five times daily, slight rectal discharge or bleeding; grade 2 as moderate diarrhea and colic, bowel movement more than five times daily, excessive rectal mucus or intermittent bleeding; grade 3 as obstruction or bleeding requiring surgery; grade 4 as necrosis/perforation, fistula; and grade 5 as death directly related to radiation. To give an example of the relative frequency of these toxicities for a large series in the literature, Zelefsky et al reported on 743 patients treated with EBRT and found 14% with grade 1; 8%, grade 2; 0.8%, grade 3; and only one patient with grade 4 GI toxicity . In this study, we chose to focus on late grade 3/4 toxicities that required intervention and could be reliably identified by using International Classification of Diseases, ninth revision (ICD-9) procedure codes or CPT codes. The toxicities were broadly categorized into GI bleeding/ulceration, fistula, stricture, and colostomy (ICD-9 and CPT codes are defined in Addendum 2 and available on request). We found that using procedure codes gave more consistent results than using ICD-9 diagnosis codes. Because our goal was to document late (rather than acute) toxicity, only GI toxicities that developed ≥6 mo after diagnosis were counted. Patients in the radiation group who received a second course of radiation treatment or patients in the conservative management group who received any radiation treatment were censored at the time of treatment.
We estimated the probability of any GI toxicity by using cumulative incidence [11,12]. Because we had two competing outcomes (death or GI toxicity), a competing risk Cox proportional hazards model was used to analyze the data, with time to either event as the response variable [12,13]. The Cox proportional hazards model for cause-specific hazard was used to assess the effects of covariates on time to GI toxicity . Hazard ratios (HRs) were also obtained in pairwise comparisons between radiation modalities (3D CRT, IMRT, protons, brachytherapy) and conservative management. To examine whether the elevated risk of GI toxicities persisted >5 yr in the radiation group versus the conservative management group, we examined the 20 114 men who survived at least 5 yr after cancer diagnosis and were still free of GI toxicities at the 5-yr point and compared the GI toxicity rate using a competing risk model. We performed the analysis using Stata v.8 (StataCorp, College Station, TX, USA) and SAS v.9.1 (SAS Institute, Cary, NC, USA).
As shown in Table 1, the conservative management patients had higher comorbidity scores, were older, and were more likely to reside in a low-income area. Their cancers were more well differentiated, with lower T stage. With regard to radiation technique, of the 28 088 radiation patients, 19 063 patients were treated with EBRT, 5338 with brachytherapy monotherapy, and 3687 with a combination of EBRT and brachytherapy.
Table 2 summarizes the results for GI toxicity for the respective radiation modalities and conservative management. EBRT (8.8 per 1000 person-years) had a higher rate of GI toxicity than brachytherapy (5.3 per 1000 person-years). Conservative management, predictably, had the lowest rate (2.1 per 1000 person-years). Of the external-beam modalities, the rates of GI toxicities in descending order were protons (20.1 per 1000 person-years), 3D CRT (9.3 per 1000 person-years), and IMRT (8.9 per 1000 person-years). Of the late GI toxicities that were screened for, GI bleeding/ulceration was by far the most common.
Figure 1 compares the cumulative incidence of GI toxicities among patients treated with radiation therapy versus patients who were conservatively managed. At 10 yr the GI toxicity rate was 1.5% for conservative management versus 5.8% for radiation therapy. Figure 2 compares the cumulative incidence of GI toxicities for the various radiation therapy modalities and conservative management. Patients undergoing proton therapy and 3D CRT had the highest rates of GI toxicities; brachytherapy and IMRT offered the lowest rates among the radiation modalities. At 4 yr, the GI toxicity rates were protons (8.5%), 3D CRT (4.8%), combination therapy (3.5%), IMRT (3.3%), and brachytherapy (2.5%). Figure 3a and and3b3b demonstrate that the GI toxicity associated with proton therapy and IMRT decreased markedly as the year of diagnosis progressed.
As shown in Table 3, radiation therapy is the most significant factor associated with increased risk of GI toxicities (hazard ratio [HR]: 4.74; 95% confidence interval [CI], 3.97–5.66). To assess the long-term impact of radiation therapy, we examined those patients who had not developed any GI toxicity within the first 5 yr of diagnosis and compared the risk of GI toxicities after 5 yr. Patients receiving radiation therapy continued experiencing significantly higher rates of new GI toxicities than the conservative management group (HR: 3.01; 95% CI, 2.06–4.39; p < 0.001). However, the absolute difference was small: about 1% at 10 yr.
When considering pairwise comparisons (Table 4), all EBRT modalities had more GI toxicity than did conservative management. When compared with each other, IMRT had less GI toxicity than 3D CRT (HR: 0.67; 95% CI, 0.55–0.82) or protons (HR: 0.30; 95% CI, 0.19–0.47), and protons had higher GI toxicity than either 3D CRT (HR: 2.13; 95% CI, 1.45–3.13) or IMRT (HR: 3.32; 95% CI, 2.12–5.20).
In this population-based study, we found that of the radiation modalities, brachytherapy (monotherapy) had the lowest rates of procedural intervention for late GI toxicity (5.3 per 1000 person-years). Somewhat surprising was that among the EBRT modalities, proton therapy had the highest rate of grade 3/4 toxicity, followed by 3D CRT and then IMRT (20.1, 9.3, and 8.9 per 1000 person-years, respectively). New technologies such as proton therapy and IMRT show substantial improvement over time, suggesting that experience with the technology may result in reduced risk of late toxicities.
That IMRT in recent years and brachytherapy offer relatively low rates of grade 3/4 toxicity is well borne out by the literature. Memorial Sloan-Kettering Cancer Center reported only a 1% risk of grade 3 GI toxicity in a cohort of prostate IMRT patients who were dose escalated to 81 Gy . Gelblum and Potters reported a large series of 825 patients treated mostly with brachytherapy monotherapy and some with combined EBRT and brachytherapy. They reported a moderate amount of grade 2 rectal toxicity (6.6%) but very rare grade 3 toxicity (0.5%) . Clinical decision making for prostate cancer is often extremely difficult. Unlike many other cancers, the decision to treat is often equivocal due to the relatively slow natural progression of prostate cancer and the typically advanced age of the patients. In this situation, where there is a clear option to manage conservatively rather than intervene, there is added emphasis to “do no harm.” This large population-based study demonstrates that IMRT and brachytherapy monotherapy are associated with very low levels of grade 3/4 GI toxicity and may offer an excellent risk–benefit ratio to patients and physicians struggling with these difficult choices. Because the theoretical advantage of proton therapy is reduced toxicity due to its particular dose deposition characteristics (the Bragg peak), it was somewhat surprising that protons were associated with the highest GI toxicity of the various radiation modalities. However, it must be pointed out that the sample size for the proton cohort was quite small because our study included patients diagnosed from 1992 to 2005, a period when proton therapy was in its relative infancy. In fact, our data showed a major decline in GI toxicity associated with protons over time (Fig. 3a). There have since been reports of favorable toxicity outcomes with proton therapy. Long-term follow-up of 1255 patients treated with proton plus photon therapy at Loma Linda University revealed late grade 3 and 4 GI toxicity of 1% and 0.2%, respectively . A recent Japanese proton study reported 18%, 3%, and 0% grade 1, 2, and 3 late rectal toxicity, respectively .
This study has certain limitations. The results of retrospective SEER-Medicare studies depend largely on the accuracy of the ICD-9 and CPT codes used. In this case, we chose to focus on patients requiring procedural invervention because the accuracy of procedures is much higher than that of diagnosis codes alone. Another limitation is that our cohort of patients were between 66 and 85 yr of age, and so our results may not be completely applicable to younger patients. There are subtle differences between SEER-Medicare and the general US population: The SEER-Medicare population is less frequently white, less poverty prone, more urban dwelling, and may have lower rates of cancer mortality. Because HMO enrollees tend to be younger and healthier than their non-HMO counterparts, excluding them may bias this cohort toward an older population . Finally, we have little data as to the specifics of treatment. For example, why did GI toxicity for protons decrease for patients diagnosed in 2004–2005? Was it due to improved equipment, a change in radiation dosage, or was there simply a learning curve to treating with protons?
There is a significantly elevated risk of GI toxicities following radiation therapy compared with conservative management, and this risk persists beyond 5 yr. Of the radiation modalities, IMRT in more recent years and brachytherapy monotherapy are associated with relatively low rates of late grade 3/4 GI toxicities and may be excellent choices in terms of optimizing the risk–benefit ratio. The choice of treatment modality is made all the more critical when one considers the advanced age of most prostate patients, the typically slow natural progression of prostate cancer, and the dramatic worsening of quality of life associated with serious GI toxicities. Somewhat surprisingly, proton therapy had the highest GI toxicity of the radiation modalities, although it was very time dependent, and by 2004–2005, GI toxicity associated with protons had decreased significantly. These data show the latest technology may not necessarily be superior to existing ones and that a substantial learning curve often exists for new interventions.
The authors acknowledge the efforts of the Applied Research Branch, Division of Cancer Prevention and Population Science, National Cancer Institute (NCI), the Office of Information Services, and the Office of Strategic Planning, Center for Medicare and Medicaid Services (CMS); Information Management Services (IMS), Inc., and the Surveillance Epidemiology and End Results (SEER) Program tumor registries in the creation of the SEER-Medicare database.
Funding/Support and role of the sponsor: This study was supported by NCI Challenge grant RC1CA145722, Robert Wood Johnson Foundation 60624, and CINJ Biometrics shared resource (NCI CA-72720-10).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Author contributions: Grace L. Lu-Yao had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.Study concept and design: Lu-Yao, Kim, Shao.
Acquisition of data: Lu-Yao.
Analysis and interpretation of data: Lu-Yao, Kim, Moore, Shih, Lin, Shen, Li, Dolan, Shao.
Drafting of the manuscript: Lu-Yao, Kim, Moore, Shih, Lin, Shen, Li, Dolan, Shao.
Critical revision of the manuscript for important intellectual content: Lu-Yao, Kim, Moore, Shih, Lin, Shen, Li, Dolan, Shao.
Statistical analysis: Lu-Yao, Moore, Shih, Lin, Shen.
Obtaining funding: Lu-Yao.
Administrative, technical, or material support: None.
Other (specify): None.
Financial disclosures: I certify that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/ affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.
Disclaimers: This study uses the Linked SEER-Medicare Database. The project described was supported by award number RC1CA145722 from the National Cancer Institute (NCI). NCI is not involved in the design or conduct of this study. The content is solely the responsibility of the authors and does not necessarily represent the official view of NCI or the National Institutes of Health.