OBJECTIVE: To compare prognostic variables and survival of radiation-associated endometrial cancers with sporadic second endometrial cancers.
METHODS: Patients with primary cancers of pelvic organs (urinary system, colorectal, cervix, vulva, and vagina) were identified from the Surveillance, Epidemiology, and End Results (SEER) database between 1973 and 2005. Among these patients, those who received pelvic radiation and subsequently developed endometrial cancer formed the radiation-associated endometrial cancers cohort (cases), whereas those who did not receive pelvic radiation but subsequently developed endometrial cancer formed the sporadic second endometrial cancers cohort (controls). Comparisons between radiation-associated endometrial cancers and sporadic second endometrial cancers used χ2, t tests, Kaplan-Meier, and cox proportional hazards analysis.
RESULTS: In 205 radiation-associated endometrial cancer patients and 1,001 sporadic second endometrial cancer patients, the mean age of diagnosis was 65 years and 68 years, respectively (P<.001). The mean latency between primary pelvic organ cancer and second endometrial cancer was 110 months for the radiation-associated endometrial cancers and 77 months for the sporadic second endometrial cancers (P=.03). The lesions in the radiation-associated endometrial cancers cohort (compared with sporadic second endometrial cancers) were far more likely to be nonendometrioid histology (76.2% compared with 51%, P<.001), poorly differentiated (58% compared with 28%, P<.001), and advanced stage (International Federation of Gynecology and Obstetrics III and intravenously combined) (43% compared with 16%, P<.001). The 5-year survival rate for radiation-associated endometrial cancers and sporadic second endometrial cancers was 27.1% and 57.1%, respectively (P<.001). In multivariable analysis, after controlling for age, race, histology, stage, grade, and treatment, the hazard ratio for death of sporadic second endometrial cancers was 1.4 (95% (CI) 1.2–3.6; P=.002).
CONCLUSION: The radiation-associated endometrial cancers carry a grave prognosis because they are more likely to be nonendometrioid, poorly differentiated advanced stage cancers. The longer latency and extensive spread at diagnosis among radiation-associated endometrial cancers may suggest a possible delay in clinical presentation and diagnosis.
LEVEL OF EVIDENCE: II
Radiation-associated endometrial cancers carry a grave prognosis, because they are more likely to be nonendometrioid, poorly differentiated advanced stage cancers.
From the Departments of 1Obstetrics and Gynecology, Division of Gynecologic Oncology, and 2Pathology Wayne State University School of Medicine; 3Wayne State University School of Medicine and Karmanos Cancer Institute; and 4The C. S. Mott Center for Human Growth and Development, Wayne State University School of Medicine, Detroit, Michigan.
From the Retrospective, Surveillance, Epidemiology, and End Results Program (SEER).
The authors thank Michael Kruger for his input in statistical analyses of the data.
Corresponding author: Dr. Sanjeev Kumar, MD, Department of Obstetrics and Gynecology, Wayne State University, 3990 John R, 7 Brush North, Hutzel Women's Hospital, Detroit, MI 48201; email: firstname.lastname@example.org.
Financial Disclosure The authors did not report any potential conflicts of interest.
Endometrial cancer developing after irradiation of the uterus directly or indirectly due to tumors of the adjacent body sites (such as cervix, rectum or the urinary system etc.), has been termed radiation-associated endometrial cancer.1
The radiation-associated endometrial cancer, although uncommon in occurrence, has been suggested to be a high-risk disease with worse prognosis.2 The experience described in the literature has several limitations, the foremost of which is the small number of patients reported in the various papers, mainly in form of case reports3,4 or case series.1,2,5 Second, although radiation-associated endometrial cancer can potentially develop after the uterus had been either in or adjacent to the irradiated field for the cancer of several organ sites, it is evident that majority of published reports have chosen to highlight radiation-associated endometrial cancer after irradiation of cervical cancer only. Third, the published reports on radiation-associated endometrial cancer are mainly from tertiary care centers and are subject to considerable referral bias due to the inherent tendency of these centers to attract the most severe or uncommon form of the disease. Fourth, the majority of authors did not have any control group to which to compare the outcomes of radiation-associated endometrial cancer, and in those that did, the control group was less than ideal.1 Obviously, almost all radiation-associated endometrial cancers are second cancers, and the proper control group should be the endometrial second cancers developing in absence of prior radiation, rather than the sporadic primary endometrial cancers as reported previously.1
The objective of the present study was to compare the prognostic variables and subsequent survival outcomes of radiation-associated endometrial cancer to that of second endometrial cancers developing in the absence of prior radiation.
MATERIALS AND METHODS
Women having pelvic radiation for tumors of various organs (cervix, vulva, vagina, colon and rectum, anus, and urinary system) with subsequent development of endometrial cancer were identified from the limited use Surveillance, Epidemiology, and End Results (SEER) Program 1973–2005.6 This was termed the radiation-associated endometrial cancer group. The uterus may lie either within the radiation field or adjacent to the radiation field when radiation is used for the cancers of these organ systems. In a second search, women with tumors of the same organ sites (Table 1), who did not have radiation, but developed endometrial cancers subsequently, were identified and termed as the second sporadic endometrial cancer group. The rational for the structure of the two cohorts hence created was based on the assumption that the endometrial cancers that develop in association with radiation for a primary cancer are, in fact, second cancers, and the optimal comparison group should be a cohort of endometrial second cancers with no prior radiation exposure, rather then the primary sporadic endometrial cancer itself.
Exclusion criteria (along with number of excluded cases in parenthesis) were as follows: unknown age (4), unknown histology (22), benign and borderline behavior of the tumor (2), diagnosis by autopsy or death certificate (7), unknown radiation status (4), and unknown latency period of development of second endometrial cancers (1). Cases that were not actively followed (7), or where endometrial carcinoma was higher than a secondary tumor (for example third or fourth tumor) were also excluded (13).
The SEER program collects information about all incident cancer cases, including age at diagnosis, subsequent tumors, marital status, race, ethnicity, tumor site, stage at diagnosis, first course of treatment, and annual follow up for vital status. The SEER program currently collects cancer incidence and survival data from population-based cancer registries covering approximately 26% of the U.S. population (SEER Web site: http://seer.cancer.gov/about/).
A second cancer was defined as any tumor, classified by the treating clinical team as a distinct tumor (independent of the primary tumor) and occurring at least 12 months after the diagnosis of the primary tumor until the date of death, date of last follow-up or the end of the study period (2005), whichever came first. Latency period is the time difference between development of first and second cancers calculated for each person in months.
The International Federation of Gynecology and Obstetrics (FIGO) stage is only applicable to the cases diagnosed after 1988. Therefore, a simplified form of staging (localized, regional, and distant) was also used in our analysis, which is available for 1973 to 2005 and provides a fair idea of the extent of the disease at diagnosis. The histology of the tumors was classified according to the International Classification of Diseases, 3rd Revision codes. The major histology types are shown in the Table 2. Note that a large number of cases are recorded in the SEER database as adenocarcinoma NOS (not otherwise specified) or simply as adenocarcinoma. The main reason for such form of reporting was the lack of wide realization of the importance of various histology types in 1970s and 1980s and was supported by our analysis that as we progress from 1970s to the 1980s and 1990s, the relative proportion of cases reported as adenocarcinoma NOS decrease drastically. Although these cases are included in the survival analysis, they were excluded during the statistical comparison of the relative abundance of various histology types in the radiation-associated endometrial cancer and second sporadic endometrial cancer cohorts, because the exact histology subtype to which these tumors belong was unknown.
Clinical and pathologic data were extracted in a tabular format and were analyzed using SEER*Stat 6.4 and SPSS 15.0 (SPSS Inc., Chicago, IL) software. Comparisons between Cohort A and B used χ2 test and t tests. Survival comparisons were obtained using the log rank test in an unadjusted Kaplan-Meier model. Cox proportional hazards model in a forward stepwise method (conditional logistic regression) was used to determine independent effect of different variables on survival. These variables included age of diagnosis, race, histology, grade, stage, and a history of radiation use before development of endometrial cancer. Survival analysis used deaths attributable to endometrial cancer rather than other causes. Variables that did not have a statistically significant independent effect on survival were excluded from the multivariable analysis (exclusion with P>.05). All P values reported are two-tailed, and a P value of less than or equal to 0.05 was considered to be statistically significant.
A retrospective power analysis was performed using n-query Advisor 7.0 (Statcon, Witzenhausen, Germany) to determine the number of patients needed to find a statistical difference (if one existed) at a power of 0.80 and an alpha of 0.05. The authors arbitrarily assumed a survival difference of 20% or 40% to be of clinical value. To detect a 20% survival difference at 5 years, the required sample size would be 360, whereas to detect a 40% difference, it would be 80. Review and approval for this study was obtained from the Institutional Review Board of Wayne State University.
A total of 205 radiation-associated endometrial cancers and 1,001 second sporadic endometrial cancers met the inclusion criteria for this study. The mean age of diagnosis was 65 years (range 20–96 years) for the radiation-associated endometrial cancer group and 68 years (range 25–99 years) for the second sporadic endometrial cancer (P=.007). In the entire study population, whites were the most prevalent race (89%), followed by African Americans (6.5%), and others (4.6%). Racial distribution of the two groups was markedly different, with the radiation-associated endometrial cancer group being three times more likely to contain African American patients (odds ratio) than the second sporadic endometrial cancer group (P<.001). On the contrary, the second sporadic endometrial cancer population was more likely to be white than the radiation-associated endometrial cancer group, whereas “other” racial fractions were equally distributed in the two groups (Table 2). The mean latency period of diagnosis between the first cancer and the second endometrial cancer was 110 months for the radiation-associated endometrial cancer and 77 months for the second sporadic endometrial cancer groups (P=.03).
Histology was coded as “adenocarcinoma not otherwise specified” in 592. Excluding these, endometrioid carcinoma was most frequent histology in the second sporadic endometrial cancer group (23%) as compared with the radiation-associated endometrial cancer group, in which the uterine sarcoma was the most frequent histology (26.3%). Whereas the second sporadic endometrial cancer subgroup was divided equally between the endometrioid and the nonendometrioid histology, more than 75% of lesions in the radiation-associated endometrial cancer group were nonendometrioid. The clear cell and papillary serous cancers were twice as likely, and sarcoma was four times more likely to be in the radiation-associated endometrial cancer group when compared with the second sporadic endometrial cancer subgroup (P<.001).
Relatively well-differentiated tumors (FIGO grade I and II) with respect to the histological grade were more common in the second sporadic endometrial cancer group as compared with radiation-associated endometrial cancer group, 72% compared with 42%, respectively. On the contrary, in the radiation-associated endometrial cancer group, 58% of tumors were poorly differentiated (FIGO grade III and intravenously) compared with 27% in second sporadic endometrial cancer (P>.05). Tumors in the radiation-associated endometrial cancer group were more likely to be in the advanced stage at diagnosis. As shown in Figure 1, the second sporadic endometrial cancer tumors were likely to be localized to the uterus, whereas radiation-associated endometrial cancer were more likely to be regional or at distant spread at diagnosis. Specifically, 43% tumors in the radiation-associated endometrial cancer group were advanced stage (FIGO III and intravenously) as compared with only 16% in second sporadic endometrial cancer group (Table 2).
An approximately equal fraction of patients in both radiation-associated endometrial cancer and second sporadic endometrial cancer groups received surgery as the primary treatment (82% compared with 84.7%, respectively; P=.19). Twenty-seven patients (13%) in the radiation-associated endometrial cancer group received radiation as compared with 321 (32.1%) in the second sporadic endometrial cancer group (P<.001). Status of lymph node dissection was known in 168 patients in radiation-associated endometrial cancer and 731 in second sporadic endometrial cancer. Lymphadenectomy was performed in 50 (29.8%) and 242 (33.1%) of radiation-associated endometrial cancer and second sporadic endometrial cancer groups, respectively (P=.4). Of these patients who underwent lymphadenectomy, lymph node metastasis was present in 16% in radiation-associated endometrial cancer and 10% in second sporadic endometrial cancer (P=.2).
There were 148 events (deaths) in the radiation-associated endometrial cancer group compared with 612 in second sporadic endometrial cancer group. The overall mean survival for the entire study cohort was 68 months (standard error [SE] 4.5). The overall median survival for radiation-associated endometrial cancer was 19 months (SE 3.6) as compared with 82 months (SE 5.2) for second sporadic endometrial cancer. The 5-year survival rates for these two groups was 27.1% and 57.1%, respectively (Fig. 2, log rank P<.001). After controlling for age, race, histology, stage, grade, and treatment provided, on a multivariable analysis, the hazard ratio for death of the radiation-associated endometrial cancer patients was 1.4 (95% CI 1.2–3.6; P=.002; Table 3).
Multivariable analysis using Cox proportional hazards model in a forward stepwise (conditional logistic regression) method revealed that even after controlling for age, race, stage, grade, histology, and treatment provided, the radiation-associated endometrial cancer carried a worse prognosis. The reference groups for the respective variables have a hazard ratio of 1. All other variables not shown in the table were eliminated from the model due to P>.05.
One might question the validity of second sporadic endometrial cancer as a better control group than the primary endometrial cancer, based on the assumption that these two groups may be similar. To address this issue, data on primary endometrial cancers was also analyzed. In 56,497 patients with primary endometrial cancer in absence of radiation, the mean age was 62 years. Whites formed an 86% (47,990) fraction of this group. The vast majority of these lesions were FIGO stage I (83%), endometrioid histology (67%), and relatively well-differentiated (82%) tumors. The 5-year survival rate in this group was 88%. All the comparisons of these data to the corresponding variables of second sporadic endometrial cancer and radiation-associated endometrial cancer (Table 2) yielded statistically significant results (P<.01). These comparisons imply that significant differences exist between second sporadic endometrial cancer and primary endometrial cancers; which in turn carry a much better prognosis than either radiation-associated endometrial cancer or second sporadic endometrial cancer.
The results of the present study show that radiation-associated endometrial cancers carry a grave prognosis because they are more likely to be nonendometrioid, poorly differentiated advanced-stage cancers. The longer latency and extensive spread at diagnosis among radiation-associated endometrial cancer may suggest a possible delay in clinical presentation and diagnosis.
The mean latency period of diagnosis of radiation-associated endometrial cancer in our study (10 years) was shorter than that reported by one previous study (14 years)1 but still remained in range of that generally reported by others (5–20 years).7 This difference may be explained by the heterogeneous methods employed by other investigators to include cases in their respective analysis,8 exclusion of certain histology types1,2 and exclusion of radiation-associated endometrial cancer developing after radiation of other organs except the cervix. Further, our study reports a longer follow-up (1973–2005) by 8 years when compared with the previous largest study on radiation-associated endometrial cancer,1 which included patients from 1976 to 2000. The latter study was based at Memorial Sloan-Kettering and MD Anderson Cancer Center. It compared 23 radiation-associated endometrial cancer developing after cervical cancer to 527 sporadic endometrial cancers and found that the median survival in the radiation-associated endometrial cancer group was significantly worse (24 months) than the sporadic group, where it was not reached (P<.001).
Similar to our data, Pothuri et al1 reported that endometrioid carcinoma was the most frequent lesion in the control group as compared with nonendometrioid histology, which was far more frequent in the radiation-associated endometrial cancer group. Many other studies also support the development of aggressive nonendometrioid histology tumors after radiotherapy of the uterus.3 The radiation-associated endometrial cancers are far more likely to be poorly differentiated advanced stage tumors, as displayed in Table 2, a finding corroborated by other studies as well.3 A median survival of 19 months for radiation-associated endometrial cancer and a 5-year survival of 27% are similar to that of 24 months and 21%, respectively, as reported by Pothuri et al.1 The survival analysis from both these reports agree on radiation-associated endometrial cancer being a disease that carries dismal prognosis.
A longer latency period and a much higher prevalence of advanced stage lesions in the radiation-associated endometrial cancer group as compared with second sporadic endometrial cancer support the hypothesis that the radiation-associated endometrial cancers may have protracted in situ growth without making themselves evident for early diagnosis; a feature well-displayed by sporadic endometrial cancer. The mode of presentation of the patients with the radiation-associated endometrial cancer is also different. Whereas the majority of sporadic endometrial cancers disclose themselves with vaginal bleeding, the radiation-associated endometrial cancer patients on the other hand may present only with nonspecific symptoms such as abdominal pain or cramping.2 Cervical stenosis is a well-recognized complication of irradiation of the uterus,4 and may explain why women with radiation-associated endometrial cancer may not have any vaginal bleeding at all. In fact, the finding of a fluid-filled endometrial cavity on ultrasonography, CT, or magnetic resonance imaging in the setting of prior irradiation of the pelvis can be due to the stenotic cervix blocking the passage of blood or pus and must raise suspicion of radiation-associated endometrial cancer.10
The health care professionals should be aware of the fact that radiation-associated endometrial cancer, although rare, has a grave prognosis. Its presentation may be atypical. Although the incidence of radiation-associated endometrial cancer is sufficiently rare that no specific screening program can be recommended for these women, one must have a very low threshold for initiating the workup to diagnose radiation-associated endometrial cancer if women with a prior history of pelvic irradiation present with nonspecific symptoms such as abdominal pain or cramping after long latency periods of exposure. An attempt must be made to retrieve endometrial tissue, either by a formal curettage or with a traditional endometrial biopsy pipette. Difficulty may be anticipated due to a stenotic/fibrotic cervix.11 Ultrasonography-guided aspiration of endometrial contents with a wide-bore needle has been proven to be useful in this situation.12 Further, patients must be educated about radiation-associated endometrial cancer and prompted to report any pertinent symptoms to their clinicians to facilitate early diagnosis.
Several limitations of our study should be mentioned. Foremost is the fact that this is a retrospective study, and we examine the association of radiation to the development of subsequent endometrial cancers, rather than its causative role. We also lack data on the other comorbidities, chemotherapy, dose of radiation, frequency and site of recurrence, and central pathology review of these tumors. Nevertheless, this is a large cohort of radiation-associated endometrial cancers reported from a wide and diverse geographic area of the United States. This study overcomes the referral bias that is inherent in single institutional studies, because patients reported to SEER are treated both in the community and at tertiary care centers. Virnig et al13 analyzed the concordance of radiotherapy as reported in SEER to that of Medicare from 1991 to 1996 for a variety of common cancers. The level of agreement reported by these authors was 88% for lung, 93% for prostate, 94% for rectal and breast, and 95% for endometrial cancers respectively. Further, the level of agreement persisted for the entire study period and across a wide geographical area as well. These data indicate that the information on radiotherapy in the SEER database is fairly accurate. Moreover, the results of the present study may be more generalizable to the external population because the SEER data presently covers approximately 26% of the U.S. population.
1. Pothuri B, Ramondetta L, Eifel P, Deavers MT, Wilton A, Alektiar K. Radiation-associated endometrial cancers are prognostically unfavorable tumors: a clinicopathologic comparison with 527 sporadic endometrial cancers. Gynecol Oncol 2006;103:948–51.
2. Pothuri B, Ramondetta L, Martino M, Alektiar K, Eifel PJ, Deavers MT. Development of endometrial cancer after radiation treatment for cervical carcinoma. Obstet Gynecol 2003;101:941–5.
3. Behtash N, Tehranian A, Ardalan FA, Hanjani P. Uterine papillary serous carcinoma after pelvic radiation therapy for cancer of the cervix. J Obstet Gynaecol 2002;22:96–7.
4. Park MH, Cho SH, Kang HJ, Kim SR, Hwang YY. Uterine papillary serous carcinoma following radiation therapy for carcinoma of cervix: a case report. Int J Gynecol Cancer 2000;10:253–6.
5. Parkash V, Carcangiu ML. Uterine papillary serous carcinoma after radiation therapy for carcinoma of the cervix. Cancer 1992;69:496–501.
6. Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov
) SEER*Stat Database: Incidence-SEER 9 Regs Limited-Use, Nov 2007 Sub (1973–2005) < Katrina/Rita Population Adjustment > Linked To County Attributes- Total U.S., 1969-2005 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Cancer Statistics Branch, released April 2008, based on the November 2007 submission.
7. Hoffman M, Roberts WS, Cavanagh D. Second pelvic malignancies following radiation therapy for cervical cancer. Obstet Gynecol Surv 1985;40:611–7.
8. Rodriguez J, Hart WR. Endometrial cancers occurring 10 or more years after pelvic irradiation for carcinoma. Int J Gynecol Pathol 1982;1:135–44.
9. Gallion HH, van Nagell JR Jr, Donaldson ES, Powell DE. Endometrial cancer following radiation therapy for cervical cancer. Gynecol Oncol 1987;27:76–83.
10. Aartsen EJ. Fluid detection in the uterus during and after irradiation for carcinoma of the cervix—clinical implications. Eur J Surg Oncol 1990;16:42–6.
11. Newman C, Finan MA. Hysterectomy in women with cervical stenosis. Surgical indications and pathology. J Reprod Med 2003;48:672–6.
12. Hammoud AO, Deppe G, Elkhechen SS, Johnson S. Ultrasonography-guided transvaginal endometrial biopsy: a useful technique in patients with cervical stenosis. Obstet Gynecol 2006;107:518–20.
© 2009 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.
13. Virnig BA, Warren JL, Cooper GS, Klabunde CN, Schussler N, Freeman J. Studying radiation therapy using SEER-Medicare-linked data. Med Care 2002 Aug; 40 suppl:IV:49–54.