Fax: (434) 243-6878.
Cancer incidence after localized therapy for prostate cancer
Article first published online: 28 JUL 2006
Copyright © 2006 American Cancer Society
Volume 107, Issue 5, pages 991–998, 1 September 2006
How to Cite
Moon, K., Stukenborg, G. J., Keim, J. and Theodorescu, D. (2006), Cancer incidence after localized therapy for prostate cancer. Cancer, 107: 991–998. doi: 10.1002/cncr.22083
- Issue published online: 21 AUG 2006
- Article first published online: 28 JUL 2006
- Manuscript Accepted: 3 MAY 2006
- Manuscript Revised: 2 MAY 2006
- Manuscript Received: 9 MAR 2006
- National Institutes of Health. Grant Number: CA104106
- prostate neoplasms;
- radiation therapy;
- tumor markers
Second cancers may occur in patients who have undergone radiation therapy. The risk for these adverse events after therapy is uncertain. In this study, the authors examined the size and significance of the observed association between occurrences of secondary cancers 5 years after radiotherapy in a large population of men with incident prostate cancer.
Men with incident prostate cancer were identified from the Surveillance, Epidemiology, and End Results (SEER) registry and were distinguished by the type of treatment received, tumor stage, tumor grade, and age at diagnosis. SEER data also were used to identify occurrences of secondary cancer beginning 5 years after the date patients were diagnosed with prostate cancer. Multivariate logistic regression analysis was used to estimate the adjusted odds of the subsequent occurrence of other cancers associated with types of radiation therapy received and was adjusted for the type of surgery, tumor grade, stage, and patient age.
Compared with men who received no prostate cancer-directed radiation, men who received external beam radiation therapy (EBRT) as their only form of radiation therapy had statistically significant increased odds of developing secondary cancers at several sites potentially related to radiation therapy, including the bladder (odds ratio [OR], 1.63; 95% confidence interval [95% CI], 1.44–1.84) and rectum (OR, 1.60; 95% CI, 1.29–1.99). Men who received EBRT also had statistically significant higher odds of developing secondary cancers at sites in the upper body and other areas not potentially related to radiation therapy, including the cecum (OR, 1.63; 95% CI, 1.10–1.70), transverse colon (OR, 1.85; 95% CI, 1.30–2.63), brain (OR, 1.83; 95% CI, 1.22–2.75), stomach (OR, 1.38; 95% CI, 1.09–1.75), melanoma (OR, 1.29; 95% CI, 1.09–1.53), and lung and bronchus (OR, 1.25; 95% CI, 1.13–1.37) compared with the odds among men who received no radiation therapy. Men who received radiation therapy in the form of radioactive implants or isotopes, either in isolation or combined with beam radiation, did not have significantly different odds of secondary cancer occurring at any of the 20 most common sites.
Patients who received with EBRT had significantly higher odds of developing second cancers both overall and in the areas that were exposed to radiation. It is noteworthy that, to the authors' knowledge, this report shows for the first time that, despite the higher doses of radiation delivered, patients who received radioactive implants had the lowest odds of developing second cancers. Cancer 2006. © 2006 American Cancer Society.
External beam radiation therapy (EBRT), brachytherapy, or EBRT followed by brachytherapy boost1 are definitive therapies for patients with localized prostate cancer. However, radiation therapy has been linked to occurrences of secondary malignancies, including leukemia, sarcomas, thyroid carcinoma, lung carcinoma, and bladder carcinoma.2 Patients who receive radiation therapy may be at increased long-term risk for developing secondary cancers compared with patients who do not receive radiation therapy.3 Furthermore, the latency period between radiation exposure and radiation-induced malignancy is ≥5 years and ranges up to 15 years, and the radiation-associated malignancies would be expected to occur within the irradiated field.4–6 Thus, second cancers that arise shortly after radiotherapy cannot be regarded as radiation-related, and their inclusion has confounded prior studies.
Although the incidence of second cancers after definitive radiotherapy has been observed in various reports,1–4, 7–9 the contribution of radiation therapy for prostate cancer to the risk of second cancers remains unclear. Several earlier studies demonstrated that radiation therapy could not be expected to contribute to posttherapy malignancy induction.3, 7, 9 In contrast, other studies that used Surveillance, Epidemiology, and End Results (SEER) data bases demonstrated that radiation was associated with an increased risk of bladder and rectal carcinomas4 and other malignancies.2, 8 One study mentioned 5 second malignancies that developed after 374 transperineal brachytherapies, but the causal relation was unclear.1
To clarify these correlations, we carried out a retrospective study based on data extracted from tumor registries8 to determine whether radiotherapy for localized prostate cancer is associated with an increased risk of developing subsequent cancer in pelvic and distant organs. In addition, we compared this incidence as a function of the type of radiotherapy (external beam radiotherapy [EBRT], brachytherapy, or combined EBRT and brachytherapy) to determine whether it plays a role in determining the risk of secondary malignancy.
MATERIALS AND METHODS
For this study, we used data from the SEER Program for men with who had an incident diagnosis of prostate cancer from 1973 through 1999 and who also were enrolled in the Medicare Program at any time either before or after their diagnosis.10 The SEER data used for this study included cases reported by registries in San Francisco, Connecticut, Detroit, Hawaii, Iowa, New Mexico, Seattle, Utah, Atlanta, San Jose, Los Angeles, and Rural Georgia, which collectively draw cases from approximately 14% of the United States population (available at URL: http://seer.cancer.gov/publicdata/). SEER Program data were used to define primary and secondary tumor characteristics, treatment history, and other related information for each member of the study population. Dates of death for men who died during the study period were obtained from the linked Medicare Program enrollment data. The data used in this study were reviewed and determined to be exempt human subjects research by the Human Investigations Committee of the University of Virginia (HIC no. 11621.)
Incidence Analysis of Second Cancers
Prior research indicates that the latency period between radiation exposure and radiation-induced secondary cancers is from 5 years to 15 years.6, 11 To account for this, we limited our primary analysis to men with histologically confirmed, incident prostate cancer diagnosed from 1973 and 1994 who survived for ≥5 years after the date of their diagnosis. We also limited our analysis to occurrences of secondary cancers that were diagnosed ≥5 years after the date of prostate cancer diagnosis.
Although the 5-year threshold provides a method of accounting for the latency period, this approach may underestimate overall incidence. To assess the amount of underestimation, we calculated the product-limit function2 of the probability of survival without any secondary cancer across the available follow-up period for all men with incident prostate cancer diagnosed from 1973 through 1999. The available follow-up period extended from the date of diagnosis through December 31, 2002 or to the date of death for men who died before that date. The product-limit function provides an estimate of the cumulative incidence of any secondary cancer over the available follow-up period among men with prostate cancer diagnosed from 1973 through 1999.
The frequency of occurrence was calculated for 90 separate secondary cancer sites by using SEER classification groups. The frequency calculated for each site excluded all men who had an occurrence of secondary cancer at that site prior to the 5-year anniversary of their prostate cancer diagnosis. Thus, the frequency of occurrence for each secondary cancer site was calculated as the proportion of occurrences at that site among all men who survived for ≥5 years who did not have a prior diagnosis of a secondary cancer at that site. All secondary cancers that occurred in less than 0.10% of the study population were excluded from further analysis. The remaining secondary cancer sites were divided into 2 groups: pelvic and other secondary cancer sites potentially related to radiation exposure (i.e., lymphomas or leukemias) or upper-body and unrelated sites of secondary cancers.
Multivariate logistic regression analysis was used to estimate the adjusted odds of occurrence for each secondary cancer site associated with the type of radiation therapy received during the first 4 months after diagnosis. Radiation therapy was classified by using the SEER-defined categories of EBRT, radioactive implants or isotopes (brachytherapy), combined EBRT with radioactive implants or isotopes, radiation not otherwise specified, or no radiation therapy received. Adjusted odds ratios (ORs) for men who received each type of radiation therapy were calculated with reference to men who received no radiation therapy.
Each OR was calculated with adjustments for differences in patient age at diagnosis, the type of cancer-directed surgery received, tumor stage, and grade. Type of cancer-directed surgery received was defined by the following categories: radical prostatectomy, transurethral radical prostatectomy, cancer-directed surgery not otherwise specified, and no cancer-directed surgery. Prostate tumor grade was defined for each patient as well differentiated, moderately differentiated, poorly differentiated, undifferentiated, or grade unknown. Prostate tumor stage was characterized for each patient as American Joint Committee on Cancer Stage I, Stage II, Stage III, Stage IV, or unknown stage.
Incidence of Second Cancers in Men with Prostate Cancer
The SEER-Medicare data included records for 297,069 men with histologically confirmed, incident prostate cancer diagnosed from 1973 through 1999. There were 29,529 men (9.9%) who had a secondary cancer diagnosed after the date of their prostate cancer diagnosis. The mean number of days available for follow-up was 2569 (7.0 years), and the median number of days was 2342 (6.4 years).
Figure 1 presents a graph of the estimated product-limit function for survival. The function plotted in the graph is the probability of survival without occurrence of secondary cancer over time among 297,069 men with incident prostate cancer during the available follow-up period, with patients censored according to occurrences of secondary cancer or the end of the available follow-up period. The function demonstrates that most secondary cancers diagnosed in the study population occur at ≥5 years after prostate cancer diagnosis.
The final study population included 140,767 men with prostate cancer diagnosed from 1973 through 1994 who remained alive 5 years after the date of their diagnosis. The mean number of days available for follow-up was 3867 (10.6 years), and the median number of days was 3651 (10.0 years). The probability of survival without occurrence of secondary cancer over time in the final study population essentially was identical to that observed for the entire cohort of men with histologically confirmed, incident prostate cancer in the SEER-Medicare data plotted in Figure 1 (data not shown).
Table 1 lists the distribution of types of radiation therapy and/or surgery, tumor stage, and tumor grade characteristics of the patients in this final study population. The mean age at diagnosis was 70 years (standard deviation, 7.4 years). Most men (67%) did not receive any form of radiation therapy, whereas 28% of men received EBRT only, and 1.6% of men received EBRT with brachytherapy boost. Most men underwent some form of cancer-directed surgery (64%).
|Variable||Percent of patients||No. of patients|
|EBRT with radioactive implants or isotopes||1.58||2219|
|Radioactive implants or radioisotopes||0.91||1285|
|No prostate cancer-directed radiation||67.16||94,541|
|Transurethral radical prostatectomy||21.51||30,286|
|No prostate cancer-directed surgery||35.25||49,621|
Table 2 lists the 20 most common sites of secondary cancers diagnosed ≥5 years after the date of prostate cancer diagnosis, in total and for mutually exclusive categories of prostate cancer-directed radiation therapies received within 4 months of the incident diagnosis. These 20 secondary cancer sites were divided according to their potential etiologic relation to radiation therapy. The bladder (1.06%), rectum (0.34%), and sigmoid colon (0.36%) were the most frequently occurring sites of secondary cancers in the pelvic region and in other sites potentially related to prostate cancer-directed radiation. The lung and bronchus (1.79%), melanomas of the skin (0.64%), and the cecum (0.34%) were the most common sites of secondary cancers of the upper body and of regions not considered potentially related to prostate cancer-directed radiation.
|Secondary cancer||EBRT||EBRT and implants||Implants only||Radiation, NOS||No radiation||Total|
|Pelvic and other potentially related secondary cancers†|
|Non-Hodgkin lymphomas (nodal)||0.37||148||0.27||6||0.16||2||0.31||9||0.31||292||0.33||457|
|Non-Hodgkin lymphomas (extranodal)||0.15||58||0.05||1||0.00||0||0.10||3||0.15||139||0.14||201|
|Chronic lymphocytic leukemia||0.16||63||0.23||5||0.08||1||0.10||3||0.16||148||0.16||220|
|Acute myeloid leukemia||0.14||57||0.09||2||0.00||0||0.03||1||0.10||97||0.11||157|
|Upper body and other unrelated secondary cancers†|
|Kidney and renal pelvis||0.27||105||0.32||7||0.24||3||0.35||10||0.26||241||0.26||366|
|Lung and bronchus||2.05||809||1.68||37||1.18||15||2.46||71||1.64||1543||1.79||2,475|
Men who received EBRT had higher incidence rates of several pelvic and other potentially related secondary cancers, including secondary cancers of the bladder (1.46%), rectum (0.44%), and sigmoid colon (0.40%). Men who received radioactive implants or isotopes only had the lowest incidence rates. It is noteworthy that men who received EBRT also had higher incidence rates of several upper body and other unrelated secondary cancers, including secondary cancers of the lung and bronchus (2.05%) and the cecum (0.40%).
Risk of Second Cancers as a Function of the Type of Radiation Treatment
Multivariate logistic regression analysis was used to assess the statistical significance of differences in the risk of secondary cancer occurrence by type of radiation treatment. Figure 2 presents the ORs calculated for each of the 20 most common sites of secondary cancers that occurred among men who received EBRT only (Fig. 2A), men who received EBRT together with radioactive implants or isotopes (Fig. 2B), men who received radioactive implants or radioisotopes only (Fig. 2C), and men who received radiation not otherwise specified (Fig. 2D). In each of the graphs, the ORs represent the odds of occurrence of the secondary cancer for men who received that type of therapy compared with men who received no radiation therapy. Each OR is adjusted for differences in surgical therapy, tumor stage, grade, and age at diagnosis.
Men who received EBRT as their only form of radiation therapy had statistically significant increased odds of secondary cancers at several sites potentially related to radiation therapy, including the bladder (OR, 1.63; 95% confidence interval [95% CI], 1.44–1.84) and rectum (OR, 1.60; 95% CI, 1.29–1.99). The odds of nodal non-Hodgkin lymphomas (OR, 1.26; 95% CI, 1.00–1.57) and acute myeloid leukemia also were greater, although these differences were of borderline statistical significance.
Men who received EBRT also had statistically significant higher odds of secondary cancers occurring at sites in the upper body and other areas not potentially related to radiation therapy. Higher odds were observed for occurrences of secondary cancers of the cecum (OR, 1.63; 95% CI, 1.10–1.70), transverse colon (OR, 1.85; 95% CI, 1.30–2.63), brain (OR, 1.83; 95% CI, 1.22–2.75), stomach (OR, 1.38; 95% CI, 1.09–1.75), melanoma (OR, 1.29; 95% CI, 1.09–1.53), and lung and bronchus (OR, 1.25; 95% CI, 1.13–1.37) compared with men who received no radiation therapy.
Men in the study population who received radiation therapy of unspecified type (not otherwise specified in the SEER records) had higher odds of secondary cancers of the rectum (OR, 2.34; 95% CI, 1.50–3.65), bladder (OR, 1.87; 95% CI, 1.42–2.48), and melanomas (OR, 1.88; 95% CI, 1.29–2.74). Men who received radiation therapy in the form of radioactive implants or isotopes, either in isolation or in combination with beam radiation, did not have significantly different odds of secondary cancer occurring at any of the 20 most common sites.
Table 3 lists adjusted ORs that were estimated for each covariate in the multivariate logistic regression models for the 4 most commonly occurring pelvic and other potentially related secondary cancers. Table 4 lists adjusted ORs that were estimated for each covariate in models that predicted the occurrence of the 4 most common upper body and other unrelated sites of secondary cancers outside the pelvic field. Complete results for the other 12 models are available in supplementary data.
|Predictor†||Adjusted odds ratio|
|Age at diagnosis of prostate cancer||1.00||0.99||1.00||0.99|
|No prostate cancer-directed radiation (reference)|
|EBRT with implants or isotopes||0.93||1.59||1.08||0.78|
|Radioactive implants or radioisotopes||0.25||0.30||1.40||0.50|
|No prostate cancer-directed surgery (reference)|
|Grade 1 (reference)|
|Stage 1 (reference)|
|Predictor†||Adjusted odds ratio|
|Cecum||Pancreas||Lung and bronchus||Melanoma|
|Age at diagnosis of prostate cancer||1.03‡||1.00||0.97‡||0.99|
|No prostate cancer-directed radiation (reference)|
|EBRT with implants or isotopes||1.67||1.55||0.89||1.47|
|Radioactive implants or radioisotopes||0.82||0.58||0.72||0.95|
|No prostate cancer-directed surgery (reference)|
|Grade 1 (reference)|
|Stage 1 (reference)|
Radiotherapy includes unavoidable irradiation of organs and tissues outside the target volume, even though 3-dimensional conformal radiotherapy or intensity-modulated radiation therapy (IMRT) may reduce these side effects. Accurate delivery of the prescribed target dose while minimizing the dose to surrounding critical organs is a key objective of treatment.
The majority of second tumors postradiotherapy were observed within the margin region of the treatment volume, which is defined as the volume from the margin of the planning target volume to 5 cm outside the edge of that volume.13 In addition, the same authors reported the induction by radiation, dose dependence, and spatial relation to irradiated volume.13
Two different approaches can be taken to quantify the risk of radiation-related second malignancies. In the direct approach, a group of patients treated a single institution may be followed, and the observed frequencies may be compared with the appropriate comparison group.8 The problem with this approach lies in the small numbers of patients and the lack of appropriate comparison groups (because of the single-institution referral bias and strict treatment protocols).8 Examination of the statistical power in these single-institution studies indicates that even the largest study (with 534 patients) had very limited power to detect and quantify realistic increases in the rates of secondary malignancies.8 An alternative to this approach involves retrospective studies based on data extracted from tumor registries.8 We used to latter approach in the current study.
In the past, some authors used the standardized incidence ratio and relative risk from the SEER data base.2, 8 Even though the SEER data do not provide complete risk factor information, they provided sufficient data for the current analysis. Neugut et al. reported the incidence of bladder carcinoma and other second malignancies after radiotherapy for prostate carcinoma with 8 years of follow-up and concluded that the risk of bladder carcinoma was elevated 1.5 8 years after patients received radiotherapy for prostate carcinoma.2 Furthermore, no increased risk of rectal carcinoma or leukemia was noted after this type of radiation exposure.2 In contrast to our study, their work did not evaluate the type of radiation used.
Brenner et al. reported second malignancies, such as bladder cancer, rectal cancer, lung cancer, and sarcoma, within the treatment after prostatic EBRT compared with surgery. Their result suggested that radiotherapy for prostate carcinoma was associated with a statistically significant, although fairly small, enhancement in the risk of second solid tumors (6%), particularly for long-term survivors (15% for ≥5 years; 34% for ≥10 years).8 In addition, Baxter et al. reported a significant increase in the development of rectal cancer after radiation for prostate cancer, indicating that the effect was specific to directly irradiated tissue. Those authors noted that, as part of the standard EBRT radiation fields to the prostate, a portion of the rectum will receive a higher radiation dose.4 The observed hazard ratio for radiation therapy and subsequent rectal cancer was 1.7 (95% CI, 1.4–2.2).4 That study was directed to patients with rectal and colon cancer only.4
Recently, Chrouser et al. reported on the risk of bladder cancer after primary and adjuvant EBRT for prostate cancer. Their retrospective review suggested that there is no evidence of an increased risk of bladder cancer after primary radiation therapy; however, their adjuvant radiation subset analysis showed a significant increase in subsequent bladder cancer.9 The reason for this difference was unclear.
To our knowledge, there is no analysis in the literature that shows the relation between brachytherapy and subsequent malignancy. Brachytherapy was introduced only in 1998; therefore, the mean follow-up for this cohort of individuals likely would be very short. Pickles and Phillips reported that 5 of 374 patients who received transperineal brachytherapy developed second malignancies.1 These included 2 pancreatic cancers, 1 gastric cancer, 1 tonsillar cancer, and 1 small cell carcinoma of the lung. The follow-up for this group was short, and the tumors occurred after a median interval of only 8 months and were located outside the radiated area. Hence, it is highly unlikely that these cancers were associated with radiation treatment.1 Furthermore, that study did not use a control group (i.e., a nonirradiated comparison group).
Our current results suggest that the incidence of second malignancy is low. Examining the adjusted ORs of developing a second cancer in each radiation subgroup, the EBRT-only subgroup had significantly higher odds than others, whereas the radioactive implants subgroup had the lowest odds. This is the first demonstration to our knowledge of differences in second cancer incidence among radiotherapy modalities. It needs to be emphasized that, given the time frame of the study, it is unlikely that most patients received more focused EBRT treatments, like those delivered by 3-dimensional conformal or IMRT techniques.
It is noteworthy that the adjusted odds of developing second malignancies in unrelated sites also were higher in the EBRT group compared with other treatment modalities. The reason for this is unclear but may be because of such confounding factors as differences in underlying cancer rates in the populations studied, genetic predisposition to spontaneous cancer, fractionation, and the combination of radiotherapy with adjuvant chemotherapy (although evidence for risk modification because of this factor was weak).14
Differences in the occurrence of secondary cancers observed in this study may reflect in part differences among patients in their indications for treatment, smoking history, or in other confounders that were not included in the multivariate adjustments. Other limitations of the study included biases caused by the quality and character of the data available for patients from the SEER data base. Tumor registries may underestimate the incidence of secondary neoplasms, because they may not have been recorded or documented properly in the source documentation, especially when the information is based on a death certificate with limited information.3 Information regarding underlying risk factors is not present typically in this context, limiting the analysis of possible concomitant exposures. Based on the atomic bomb studies, we know that secondary malignancies may appear from 5 years to 15 years after exposure.11 Because our mean follow-up was 10.6 years, there is a chance that our own analysis may have missed some of the later developing cancers. A shortcoming of our study, and of other studies that deal with prostate cancer, is the relatively short follow-up for many men. This is an inherent difficulty in the study of diseases like prostate cancer, because it tends to affect men in the last years of their lives.1
In conclusion, the current results suggest that the incidence of second malignancy after prostate therapy is low. Compared with men who received no prostate cancer-directed radiation, men who received EBRT had significantly greater odds of developing second cancers both overall and in the areas that were exposed to radiation. It is noteworthy that, in the current report, we have shown for the first time to our knowledge that, despite the delivery higher doses of radiation, patients who received radioactive implants had the lowest odds of developing a second cancer.
- 10Overview of the SEER-Medicare data: content, research applications, and generalizability to the United States elderly population. Med Care. 2002; 40( 8 Suppl): IV3–IV18., , , , .
- 12Analysis of Survival Data. London: Chapman and Hall; 1984..