Comparison of Second Cancer Risks from Brachytherapy and External Beam Therapy After Uterine Corpus Cancer

Uterine corpus cancer (UCC) is the most common gynecologic malignancy in the United States and the fourth most common cancer among women (1). The disease is rare before the age of 45 years and the incidence peaks between 60 and 70 years of age (2). A high percentage of women are diagnosed with the disease confined to the uterus, resulting in overall 5-year relative survival rates of 80 to 90% (2). More than 90% of UCCs are endometrial adenocarcinomas or carcinomas of the uterine body, sarcomas, and other subtypes are uncommon (3). Approximately 90% of patients are treated surgically with removal of the uterus, cervix, ovaries, and fallopian tubes (3, 4). UCC patients often receive adjuvant radiotherapy to the pelvis, typically including external beam radiation alone, radioactive implants (brachytherapy) alone, or a combination of both types (3, 4).

High-dose radiotherapy to the pelvis increases the risk of second primary cancers, particularly for cancers of the colon, rectum, bladder, and leukemia (5-9). However, data on the second cancer risk for different types of radiotherapy are sparse, and to our knowledge, no studies directly compare the risks among UCC patients treated with external beam therapy versus brachytherapy. External beam therapy to the pelvis delivers substantially greater radiation doses to abdominal organs than brachytherapy, especially to organs not immediately adjacent to the uterus. We evaluated the risks of second malignancies in a large population-based cohort of women diagnosed with carcinoma of the uterine corpus and quantified the second cancer risk associated with radiation treatment compared with surgery alone. Further, we estimated the radiation dose to specific organs for typical treatments from each type of radiotherapy and quantified the risks of second primary cancers of organs receiving relatively high, moderate, and low doses.

Women with a first primary invasive adenocarcinoma or carcinoma of the uterine corpus (excluding sarcomas and other cell types) diagnosed between 1973 and 2003 at ages 15 to 79 y were identified (excluding women with prior malignancies) from nine population-based cancer registries reporting to the U.S. National Cancer Institute (NCI)'s Surveillance, Epidemiology, and End Results (SEER) Program (10). The SEER Program is a set of geographically defined, population-based tumor registries in the United States, operated by local nonprofit organizations under contract to the NCI. Each registry annually submits its cases to the NCI on a computer tape. These computer tapes are then edited by the NCI and made available for analysis. The SEER9 registries include the states of Connecticut, Iowa, New Mexico, Utah, and Hawaii, and the metropolitan areas of Atlanta, Detroit, San Francisco-Oakland, and Seattle-Puget Sound that together represent ∼10% of the U.S. population. The SEER Program records all treatment given for the first course of therapy in broad categories, that is surgery, radiotherapy, chemotherapy, and hormonal therapy; therapy given after the first course is not recorded.

Our study was restricted to women who survived ≥1 y following their initial diagnosis and who were initially treated with surgery only or with surgery and radiotherapy. The radiotherapy group was subdivided into external beam therapy, brachytherapy, and combination radiotherapy (external beam and brachytherapy).

The SEER files were searched for second primary cancers (excluding nonmelanoma skin cancers and female genital tract cancers) that were diagnosed at least 1 y after the initial UCC diagnosis. Higher-order cancers (third, fourth, etc.) occurring after the second cancer were not included.

Standardized incidence ratios (SIR) were calculated as the ratio of the observed number of second cancer cases occurring among women who survived ≥1 y divided by the expected number. For each patient, person-years at risk were accumulated from 1 y after UCC diagnosis until the date of death, date of last follow-up, date of diagnosis of second primary tumor, or end of the study (December 31, 2003), whichever came first. Site-specific cancer incidence rates for females were computed by registry, 5-y age and calendar year groups, and race (white, black, and other) and were multiplied by the person-years at risk to estimate the expected number of second cancers in the general population. Poisson-based 95% confidence intervals (CI) were calculated (11).

Because there are unmeasured risk factors related to UCC (obesity, exogenous hormones, etc.), we assessed the effect of radiotherapy by internally comparing the risk of second cancers among patients treated with surgery and radiotherapy to those treated with surgery alone using Poisson regression methods for grouped survival data (12). Incidence rate ratios (IRR) and 95% CIs were estimated with stratification by 5-y age and calendar year, cancer registry, and race.

Other studies have indicated that the minimal time between radiation exposure and the occurrence of a new cancer is approximately 5 to 15 y for solid cancer (7, 13) and about 1 to 5 y for leukemia (6, 14). Therefore, the analyses of solid cancers focused on the period ≥5 y from the initial diagnosis and the analyses of second lymphoma/leukemia cases were limited to the period ≥1 y after initial diagnosis. We also assessed the log-linear trends in latency-specific IRRs by incorporating a log-linear interaction between the treatment variable and four categories of time since first diagnosis (5-9, 10-14, 15-19, and ≥20 y).

The number of excess second cancers associated with the specific radiation modalities was calculated as [(IRR − 1)/IRR] × (number of observed cases in the exposure group of interest). The total number of excess cases was obtained by summing over the three radiotherapy groups. The proportion of cancer cases attributed to a specific type of radiotherapy was calculated as (IRR − 1)/IRR].

We estimated typical organ-specific radiation doses in gray (Gy) and weighted doses to the active bone marrow (Table 1). Doses were estimated separately for each type of radiotherapy, based on techniques typically used during the study period. Dose calculations were done with a three-dimensional mathematical phantom and measurements in water, using methods previously described (15). We did not have detailed information to individualize the brachytherapy dose based on placement of the radioactive source. Typical doses for brachytherapy were based on the assumption that sources were placed in the uterus and vagina (16), which will overestimate the site-specific doses compared with using sources placed in vaginal only. For brachytherapy, a rapid fall-off in dose occurs with distance from the implant; organs adjacent to the sources such as urinary bladder, colon, and rectum receive high doses. External beam therapy results in high doses of radiation to the same organs in the pelvic region (about 40-50 Gy), whereas the stomach and pancreas receive moderate radiation doses in the range of 0.4 to 5.0 Gy. The doses to colon were heterogeneous with lower doses to the transverse colon compared with the ascending colon. Using the estimated doses, we evaluated the risk of second cancers for three separate groups: heavily (≥5 Gy), moderately (0.4-5 Gy), and lightly irradiated sites (<0.4 Gy).

We observed 7,428 second cancers in the cohort of 60,949 women who survived at least 1 year following UCC and who received surgery with or without radiotherapy as their first course of treatment. In total, 62% received surgery alone and these patients were similar to those treated with radiotherapy in terms of age, survival, race, and initial treatment with chemotherapy (Table 2). Women treated with surgery alone were more likely to have tumors diagnosed at a localized stage of disease and were more frequently diagnosed in later calendar years compared with irradiated women. Patients treated with brachytherapy had longer follow-up (data not shown) and more localized uterine tumors compared with those in other radiotherapy groups.

The risk of developing a second cancer was lower than that expected in the general population among women treated with surgery only for their first course of the therapy (SIR = 0.89 for all second cancers combined; SIR = 0.90 for second solid cancers; Table 3). Overall, women receiving radiotherapy did not differ from the general population in their second cancer risk, either for all cancers combined (SIR = 1.02) or second solid cancers (SIR = 1.02). However, significant elevations in risk were observed for several sites, including cancers of the colon (SIR = 1.25), rectum (SIR = 1.27), urinary bladder and ureter (SIR = 1.78), and soft tissue sarcomas (SIR = 1.76). Although the risk of nonchronic lymphocytic leukemia (CLL) was significantly elevated following radiotherapy (non-CLL, SIR = 1.55), a significant deficit in risk was observed for CLL (SIR = 0.65). Lower than expected risks were also found for smoking-related second cancer sites (buccal cavity and pharynx, esophagus, and the respiratory system) among irradiated and nonirradiated women.

Table 4 shows the risks of second cancers among women receiving radiotherapy compared with women treated with surgery only. Among women who survived ≥1 years, the IRRs of developing a second cancer were increased among irradiated patients compared with patients having surgery only (combination external beam and brachytherapy, IRR = 1.26; external beam therapy alone, IRR = 1.15; brachytherapy alone, IRR = 1.07). External beam therapy alone was associated with a 2-fold risk of leukemia excluding CLL (IRR = 2.03) and combination radiotherapy with a similarly elevated risk of non–Hodgkin lymphoma (IRR = 1.79). Among 5-year survivors, elevated risks for second solid cancers were concentrated in heavily irradiated organs, specifically the colon, rectum, and pelvic area of soft tissues, following combination radiotherapy and external beam therapy alone. Excess cancers of the bladder/ureter were found after all three radiation modalities (IRR = 2.66, combination radiotherapy; IRR = 2.30, external beam therapy; IRR = 1.73, brachytherapy). Elevated risks of stomach cancer and cancers of the liver and thyroid (based on small numbers) were limited to the brachytherapy group. Results for non-CLL and non–Hodgkin lymphoma were similar for 1- and 5-year survivors but with lower precision for 5-year survivors due to the smaller number of survivors.

In a direct comparison among radiation modalities (using brachytherapy alone as a reference group), women treated with any external beam therapy (combination or beam therapy alone) had significantly higher risks of second cancers. The IRRs were 1.10 (95% CI, 1.02-1.20) for cancers in 1-year survivors and 1.44 (95% CI, 1.19-1.75) for heavily irradiated sites in 5-year survivors. Combination radiotherapy had the overall highest estimated organ-specific doses and the strongest association with new solid malignancies, with significantly elevated risks observed for heavily irradiated segments of the colon (IRR = 1.62; 95% CI, 1.18-2.22) and bladder/ureter (IRR = 2.15; 95% CI, 1.34-3.46), compared with women receiving brachytherapy.

Of the 3,203 second cancers observed in women treated with radiotherapy, 405 excess cancer cases overall (including 327 solid cancers) occurred among 1-year survivors and 275 cases overall (including 213 solid cancers) among 5-year survivors could be explained by radiotherapy (Table 4). Of the 151 cancers of the bladder/ureter occurring among irradiated patients surviving ≥5 years, 81 cases or 54% were estimated as attributed to radiation therapy. External beam therapy (either combination or external beam alone) accounted for 100 excess colon cases (36%) and 22 excess rectal cancer (39%), compared with only 5 excess colorectal cancer cases following brachytherapy (4%).

Figure 1 shows the IRRs of second solid cancers among 5-year survivors by type of radiation and according to the three categories of estimated radiation dose to specific organ sites. Statistically significant increased risks following combination radiotherapy were found for second solid cancers occurring at heavily (IRR = 1.93; 95% CI, 1.60-2.32) and moderately irradiated sites (IRR = 1.55; 95% CI, 1.17-2.06). Increased risks for heavily irradiated sites were also seen following external beam therapy alone (IRR = 1.71; 95% CI, 1.47-1.98); however, elevated risks were not evident for moderately irradiated sites. Risks of second solid cancers following brachytherapy alone were increased for both heavily (not significant) and moderately irradiated sites, but were overall lower than the risks observed for any external beam therapy.

Table 5 shows IRRs by latency (time since initial diagnosis) for heavily, moderately, and lightly irradiated solid cancer sites. For both combination and beam therapy, significantly increased IRRs of second cancers were detected for most latency intervals among the heavily irradiated sites (P < 0.01 over 5-9, 10-14, 15-19, ≥20 years) with women followed ≥10 years having >2-fold increases in risk. For brachytherapy, radiation-related risks were substantially lower and a statistically significant elevated risk (IRR = 1.36) was seen only for the 10- to 19-year time interval (P = 0.02). The IRRs of second solid cancers were lower among the moderately irradiated sites compared with the heavily irradiated sites, but statistically significant risks were detected for the 10- to 19-year latency interval for women receiving either combination radiotherapy (IRR = 1.87) or external beam therapy (IRR = 1.39). P trends over the latency periods of ≥5 years were significant for each radiation modality. Among patients treated with brachytherapy, there was only a modest nonsignificant elevation in second solid cancers risks at moderately irradiated sites. For cancer sites that received low radiation exposures, we observed a significantly elevated risk only for the time period of 1 to 4 years after brachytherapy.

We evaluated the effect of age at UCC diagnosis among 10-year survivors, comparing those treated with radiotherapy with those receiving surgery alone. There was no evidence of substantial differences in risk for heavily irradiated solid cancer sites by age at diagnosis (<50, 50-59, 60-69, ≥70 years; data not shown).

In a sensitivity analyses, we found no statistically significant heterogeneity by cancer registry and results were similar to those reported above when analyses were restricted to women with localized stage. Excluding patients known to be treated with initial chemotherapy did not change the results.

Our study quantified the risk of radiation-related second cancers in a large cohort of UCC survivors treated with surgery, with or without radiotherapy, who have been followed for new malignancy over a 30-year period and, for the first time, directly compared the risk of solid cancers among women treated with different radiotherapy modalities. We found that women initially treated with surgery and radiotherapy had significantly higher risks compared with those treated with surgery alone, with substantially lower risks following brachytherapy compared with any beam therapy. When we evaluated cancer risk for organ sites that were ranked according to estimated radiation dose received, the greatest risks were found for non-CLL leukemia and for heavily irradiated solid organs sites (e.g., colon, rectum, and bladder). Among 5-year survivors receiving radiotherapy, we estimate that 11% (213 of 2,012 cases) of the solid second cancer cases could be explained by radiotherapy with only 14 total excess cases related to brachytherapy compared with 199 for any external beam therapy. Risk of radiation-related cancer did not depend on age at exposure in this population of mostly older women (>60 years).

Previous studies have reported an increased risk of second cancers among patients typically treated with pelvic external beam radiotherapy for cancers of the cervix, ovary, testis, and prostate (5, 9, 17-21). However, only sparse data exist on the carcinogenic effect of brachytherapy when given without external beam therapy, and no earlier investigation has evaluated solid cancer risk following this therapy for UCC. Brachytherapy alone has been used with increasing frequency to treat prostate cancer since the mid-1990s; however, numbers in the long-term follow-up intervals remain small and evidence for differences in second cancer risk by radiation modality have been inconsistent (22-24).

Our current investigation expands upon earlier SEER investigations (19, 25) by directly comparing the risks of developing a new malignancy following three different radiation modalities for UCC compared with surgery alone, and by evaluating the risk in groups of cancer sites defined by their average radiation doses. The two previous SEER investigations (19, 25) had somewhat different inclusion criteria and design (including age, calendar years, follow-up time, and definitions of radiation and second cancer) and were therefore not directly comparable with our study, but the overall second cancer results were similar. Our overall risk of second solid cancers among UCC 5-year survivors was only slightly increased following brachytherapy, whereas risk was significantly elevated by 13% following external beam therapy and 28% after combination radiotherapy. We found a 44% higher risk for heavily irradiated solid organ sites following any beam therapy compared with brachytherapy alone. Differences in the carcinogenic effect of the two radiation modalities is not surprising because we estimated that radiation dose to organs in close proximity to the uterus is 5- to 10-fold higher for beam therapy than brachytherapy. In addition, since today's patients largely receive only vaginal brachytherapy (26), risks would probably be even lower for current brachytherapy regimens.

Our results are in general agreement with studies of women treated with radiotherapy for cervical cancer, most of whom received only external beam therapy or combination radiotherapy (5, 9, 14). UCC survivors treated with any external beam therapy had significant excesses of second cancers of the colon, rectum, bladder, pancreas (combination therapy only), and soft tissues (beam therapy only). Risks for heavily irradiated sites were elevated for both combination and external beam therapy alone and were highest among 10-year survivors. Similar to previous reports (6, 19), risks of acute nonlymphocytic leukemia after UCC were elevated by 70% to 80% after external beam therapy. In a new finding, women treated with combination therapy in our study developed an increased risk of non–Hodgkin lymphoma, a cancer that has been infrequently associated with radiation in the literature (7).

Although brachytherapy was associated with prominent excesses of second cancers of the bladder/ureter and stomach among 5-year UCC survivors, little excess was seen for cancers of the rectum, pelvic soft tissues, parts of the colon, or for leukemia. Overall, we found no evidence of an increased risk following brachytherapy to the group of heavily irradiated sites for the first 10 years of follow-up, although risk seemed to increase thereafter. From previous studies, increased risks of cancers related to radiation have been observed among primarily younger women treated with brachytherapy given at lower doses for benign gynecologic bleeding disorders (27, 28). Excess deaths were due to cancers of the uterus, bladder, colon, other genital sites, and leukemia, but not rectal cancer (28).

Advantages of the current study include the large number of UCC women treated with radiation and with surgery alone in a population-based setting, and the availability of detailed information on type of radiotherapy received. However, we lacked the information on other risk factors for endometrial cancer, treatment after the first therapy, and individual organ doses. Because treatments have changed over time, some of the risks we observed are associated with old treatments. Moreover, increased medical surveillance of heavily irradiated organs, such as the colon, rectum, and bladder, may have resulted in increased risks for irradiated compared with nonirradiated women.

In summary, radiotherapy for UCC increases the risk of second cancers, emphasizing the need for continued long-term medical surveillance of irradiated cancer patients. Our study shows that organs that receive high doses from any external beam radiation due to their close proximity to the uterus are particularly susceptible to iatrogenic cancers. The current results lend support to clinical studies emphasizing the more favorable toxicity profile of vaginal brachytherapy in early-stage UCC (3, 29), although we were unable to address the late effects of high-dose rate brachytherapy, the latter being seen in wider use in the last decade (26). It is important that our findings be interpreted in light of the substantial benefits of adjuvant radiotherapy in reducing the risk of recurrence and possibly prolonging survival (29, 30). Thus, physicians need to continually weigh the risk of developing a second cancer versus the benefits of radiotherapy.

No potential conflicts of interest were disclosed.

We thank Nathan Appel, Information Management Services, Rockville, MD, USA, for the expert computer support and data management.

Grant Support: Intramural Research Program, NCI, NIH, USA.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.