Fax: (509) 228-1180
Descriptive nomograms of adjuvant radiotherapy use and patterns of care analysis for stage I and II endometrial adenocarcinoma: A surveillance, epidemiology, and end results population study
Version of Record online: 11 SEP 2007
Copyright © 2007 American Cancer Society
Volume 110, Issue 9, pages 2092–2100, 1 November 2007
How to Cite
Lee, C. M., Szabo, A., Shrieve, D. C., Macdonald, O. K., Tward, J. D., Skidmore, T. B. and Gaffney, D. K. (2007), Descriptive nomograms of adjuvant radiotherapy use and patterns of care analysis for stage I and II endometrial adenocarcinoma: A surveillance, epidemiology, and end results population study. Cancer, 110: 2092–2100. doi: 10.1002/cncr.22997
- Issue online: 18 OCT 2007
- Version of Record online: 11 SEP 2007
- Manuscript Accepted: 24 JUN 2007
- Manuscript Revised: 20 JUN 2007
- Manuscript Received: 6 OCT 2006
- Cancer Center. Grant Number: 2P30CA042014-17
- endometrial cancer;
- adjuvant treatment;
- patterns of care analysis
Although endometrial cancer remains the most common gynecologic malignancy in the United States, differing approaches to adjuvant radiotherapy treatment for early disease exist within the medical community because of the lack of a national consensus.
The authors studied patterns of adjuvant care for stage I and II endometrial adenocarcinoma using a large United States population database. A retrospective analysis was conducted from the Surveillance, Epidemiology, and End Results (SEER) Program of the U.S. National Cancer Institute from 1988 to 2002, and 26,923 women with American Joint Committee on Cancer stage I and II endometrial adenocarcinoma were selected. The following prognostic factors were analyzed: age, race, stage, grade, year of diagnosis, SEER registry location, and use and type of postoperative radiotherapy (RT). Adjuvant RT was coded as none, external-beam RT (EBRT), brachytherapy (BR), or a combination of the 2 (EBRT + BR).
Higher tumor grade and stage led to greater use of RT. The odds ratio (OR) for adjuvant RT was 3.4 for stage IB versus stage IA and 51.8 for stage IC/II versus stage IA. The effect of grade depended on stage: for stages IA and IB, the OR was 2.9 for grade 2 versus grade 1 and 11.7 for grade 3/4 versus grade 1; whereas, for stage IC/II, the OR was 1.5 for grade 2 versus grade 1 and 2.0 for grade 3/4 versus grade 1. Within stage I, increasing substage and grade increased the odds of EBRT with or without BR compared with BR alone. Race did not effect the choice of therapy (all P > .1). Geographic location had a significant effect on overall RT use and therapy choice.
To the authors' knowledge, this was the largest patterns of care analysis to date of adjuvant RT in patients with stage I and II endometrial adenocarcinoma. The current study revealed that there is significant diversity in the use of adjuvant RT across the United States, and the results reflected the absence of a national consensus on adjuvant treatment for early-stage disease. Cancer 2007. © 2007 American Cancer Society.
Endometrial cancer is the most common gynecologic malignancy in the United States with more than 40,000 women newly diagnosed each year. Although approximately 85% of these patients are diagnosed with stage I or II disease, the role of adjuvant treatment in early-stage disease remains controversial despite published results from randomized trials.1–5 Results from the Phase III randomized Postoperative Radiation Therapy in Endometrial Carcinoma (PORTEC) trial and from the Gynecologic Oncology Group (GOG) 99 trial indicated that the addition of postoperative radiotherapy (RT) leads to a statistically significant reduction in locoregional recurrences in intermediate-risk disease. Our recent analysis of a large United States population database revealed that an overall survival advantage is associated with the adjuvant use of RT for specific high-risk cohorts of patients with stage I disease.6 In addition, data on 10,000 women from the American College of Surgeons National Data Base indicate a beneficial role of postoperative RT in the treatment of intermediate-risk endometrial cancer.7 Despite these encouraging results, uniformly defined treatment guidelines for patients with early-stage endometrial carcinoma remain difficult to find. Most practitioners advocate the use of external-beam RT (EBRT) or brachytherapy (BR) in the postoperative setting based on a patient's clinical and pathologic characteristics (a risk-stratified approach).1–3, 8–11 This notable diversity in treatment approaches is illustrated well by the clinical practice guidelines for early-stage endometrial cancer of the National Comprehensive Cancer Network.12 Despite this significant variability in practice patterns, a quantitative analysis of adjuvant RT practice patterns and of clinical factors that affect treatment decisions does not exist for this common gynecologic malignancy.
With the lack of a broadly accepted standard for postoperative treatment of this common malignancy, population-based studies are needed to identify the impact of published trial results on the practices of physicians in the United States and also to document the variability in treatment approaches. By using population-based cancer data, the objective of this study was to quantitatively evaluate practice patterns through descriptive nomograms for the use of adjuvant RT in patients with stage I and II endometrial adenocarcinoma as they relate to clinical and pathologic risk factors.
MATERIALS AND METHODS
Data and Study Population
Data for this study were obtained from the National Cancer Institute (NCI) Surveillance, Epidemiology, and End Results (SEER) Program using the SEER 11 Registries plus Alaska 1988 to 2002 data set (November 2004 edition).13 SEER is comprised of a set of geographically defined, population-based, central cancer registries in the United States and is operated by local nonprofit organizations under contract with the NCI. The SEER database is a composite of cancer registry data from Connecticut, Iowa, Hawaii, New Mexico, Utah, metropolitan areas (Atlanta, Detroit, Los Angeles, Oakland, San Francisco, San Jose-Monterey, Seattle-Puget Sound), and Alaska Native populations. Serial registry data are submitted electronically without personal identifiers (deidentified) to the NCI on a biannual basis, and the NCI thereafter makes the data available to the public for research purposes.13 Because all SEER database information remains deidentified, approval by an ethics committee was not necessary to perform the current analyses, and informed consent by the study participants also was not necessary. The case-ascertainment rate from the SEER registries has been reported as 97.5%, and the SEER database is the authoritative source of population-based information on cancer incidence and survival in the United States. In general, the populations covered by SEER are known as representative of the United States as a whole.14
The analyzed study population included women who were diagnosed with American Joint Committee on Cancer (AJCC) stage I and II (uterine carcinoma stage I and II [T1-T2N0M0]: carcinoma is confined to the corpus, including the isthmus [T1]; carcinoma has involved the corpus and the cervix but has not extended outside the uterus [T2]; no regional lymph node metastasis [N0]; no distant metastasis [M0]) endometrial adenocarcinoma who underwent total abdominal hysterectomy and bilateral salpingo-oophorectomy (TAH-BSO). Histologic classification for endometrial adenocarcinoma was based on the International Classification of Disease for Oncology (ICD) codes (ICD code 8380).15 Patients who were diagnosed between 1988 and 2002 with complete data sets were included. The following pretreatment clinical and pathologic factors were included in the analysis: age, race, stage, grade, use and type of postoperative RT, year of diagnosis, and SEER registry location. The following were causes for patient exclusion from the analysis: clinical or pathologic lymph node involvement (N1 disease), distant spread (M1 disease), incomplete surgery (not TAH-BSO), and patients with missing tumor stage or grade information. For tabulation and plotting, we categorized the year of diagnosis into 3 approximately equal length intervals and categorized age at diagnosis into 3 approximately equal frequency intervals. In the statistical analyses, these were treated as continuous variables. This report included 26,923 patients.
We used logistic regression modeling to estimate jointly the effect of all the covariates. In a logistic regression, the log-odds of the probability of an event (in this instance, administering RT) is modeled as a linear combination of predictors. We present the fitted model in a graphic form, as a nomogram. On the nomogram for each predictor, we illustrate the effect of changing its value. The effect is expressed in a point value with each point representing a 2-fold increase in odds (corresponding point values are shown on the top of each plot). The point-contributions of all predictors are then totaled (because it is a linear model), and the resulting sum is converted to a probability according to a logistic scale (correspondence between the resulting sum and the probability of a specific adjuvant RT treatment are shown at the bottom of each nomogram).
Logistic regression models for adjuvant RT versus no RT, EBRT (either alone or in combination) versus BR, and EBRT + BR versus EBRT were constructed. The development of each of the models followed a similar route: we started with a base model that did not assume linearity of the effect for any of the predictors. Next, we examined interactions and included all interactions that had P values <.1. No main effects were removed even if they were nonsignificant (eg, race in the models of RT choice). We tested the importance of the nonlinear effects for age and year of diagnosis by using formal tests and examining the fitted curves and observed that the effect of the year of diagnosis was linear for the decision to administer RT, thus, we replaced it with a simple linear term in the model.
SEER*Stat version 6.1.4 was used to extract case-level data from the SEER cancer public-use database (1973–1999; November 2003 submission). R version 2.1.0 was used for statistical and data analyses.16 All P values <.05 were considered statistically significant.
Table 1 illustrates descriptive statistics for the selected groups. The frequency of adjuvant RT use increased with increasing tumor grade and increased tumor stage, as expected. The decade of treatment did not have a significant effect on adjuvant RT use, and those patients who had a surgical lymph node examination had an increased rate of adjuvant RT use. The mean age of diagnosis is 63.1 years (range, 14–99 years). The median follow-up was 50 months (mean, 59.9 months; range, 0–179 months). The relative frequency of adjuvant RT use based on stage and grade of disease is detailed in Table 2, which also illustrates that the relative frequency of adjuvant RT increased for patients with increased stage and grade of disease. For example, California (the San Francisco, San Jose, and Los Angeles registries) seems to be characterized by rare use of BR alone; whereas New Mexico, Detroit, Hawaii, and Rural Georgia have much higher use of combination RT than the other areas. It should be emphasized that these findings are marginal effects alone and do not take into account the interactions of various factors. For example, individual registries within the SEER database have been started in various years, and each is comprised of unique patient subgroups with respect to disease status and racial composition.
|Variable||No. of patients||Percentage of patients (No.)|
|No RT||EBRT||BR alone||EBRT + BR|
|Tumor grade at diagnosis|
|1||12,190||88 (10,783)||6 (714)||2 (281)||3 (412)|
|2||9715||73 (7124)||14 (1336)||4 (393)||9 (862)|
|3–4||5018||53 (2664)||27 (1354)||6 (297)||14 (703)|
|Tumor stage at diagnosis|
|IA||7608||95 (7248)||2 (185)||1 (92)||1 (83)|
|IB||12,537||83 (10,459)||9 (1097)||4 (488)||4 (493)|
|IC||3485||44 (1548)||36 (1259)||5 (163)||15 (515)|
|II||3293||40 (1316)||26 (863)||7 (228)||27 (886)|
|Age at diagnosis, y|
|<58||8970||81 (7260)||9 (833)||3 (306)||6 (671)|
|58–70||9032||75 (6753)||13 (1185)||4 (362)||8 (732)|
|≥70||8921||74 (6558)||16 (1386)||3 (303)||8 (674)|
|Period of diagnosis|
|Before 1994||6366||72 (4610)||15 (975)||4 (228)||9 (553)|
|1994–1998||9961||76 (7543)||14 (1396)||3 (317)||7 (705)|
|After 1998||10,596||79 (8418)||10 (1033)||4 (426)||7 (719)|
|Black||1287||71 (916)||16 (211)||3 (36)||10 (124)|
|White||23,519||76 (17,935)||10 (202)||3 (54)||7 (138)|
|Other||2024||81 (1630)||13 (2989)||4 (880)||7 (1715)|
|Missing||93||97 (90)||2 (2)||1 (1)||0 (0)|
|AlsN||24||88 (21)||8 (2)||4 (1)||0 (0)|
|AltM||1295||74 (958)||13 (173)||6 (74)||7 (90)|
|Cnnc||2775||66 (1821)||13 (363)||10 (272)||11 (319)|
|DtrM||3904||75 (2785)||11 (409)||2 (82)||12 (428)|
|Hawa||896||84 (757)||6 (50)||1 (13)||8 (76)|
|Iowa||3117||72 (2232)||15 (462)||3 (108)||10 (315)|
|LsAn||5561||80 (4459)||13 (741)||2 (116)||4 (245)|
|NwMx||1001||74 (737)||10 (105)||1 (11)||15 (148)|
|RrlG||40||80 (32)||8 (3)||2 (1)||10 (4)|
|SF-S||3263||80 (2622)||14 (447)||1 (38)||5 (156)|
|SJ-M||1243||89 (1112)||8 (97)||1 (13)||2 (21)|
|StPS||2828||75 (2123)||13 (379)||7 (206)||4 (120)|
|Utah||1176||78 (912)||15 (173)||3 (36)||5 (55)|
|Surgical LN examination|
|Performed||12,313||70 (8563)||16 (1975)||5 (647)||9 (1128)|
|Not performed||14,610||82 (12,008)||10 (1429)||2 (324)||6 (849)|
|Overall||26,923||76 (20,571)||13 (3404)||4 (971)||7 (1977)|
|Variable||No. of patients||Percentage of patients (No.)|
|No RT||EBRT||BR||EBRT + BR|
|Grade 1||4692||98 (4597)||1 (42)||1 (35)||0 (18)|
|Grade 2||2108||95 (1995)||3 (66)||1 (24)||1 (23)|
|Grades 3–4||808||81 (656)||10 (77)||4 (33)||5 (42)|
|Grades 1–4||7608||95 (7248)||2 (185)||1 (92)||1 (83)|
|Grade 1||5633||93 (5245)||3 (176)||2 (120)||2 (92)|
|Grade 2||4744||83 (3956)||8 (369)||5 (225)||4 (194)|
|Grades 3–4||2160||58 (1258)||26 (552)||7 (143)||10 (207)|
|Grades 1–4||12,537||83 (10,459)||9 (1097)||4 (488)||4 (493)|
|Grade 1||1007||53 (529)||30 (305)||5 (50)||12 (123)|
|Grade 2||1483||43 (631)||38 (563)||4 (63)||15 (226)|
|Grades 3–4||995||39 (388)||39 (391)||5 (50)||17 (166)|
|Grades 1–4||3485||44 (1548)||36 (1259)||5 (163)||15 (515)|
|Grade 1||858||48 (412)||22 (191)||9 (76)||21 (179)|
|Grade 2||1380||39 (542)||24 (338)||6 (81)||30 (419)|
|Grades 3–4||1055||34 (362)||32 (334)||7 (71)||27 (288)|
|Grades 1–4||3293||40 (1316)||26 (863)||7 (228)||27 (886)|
Logistic regression models for adjuvant RT versus no RT, EBRT (alone or in combination) versus BR, and EBRT + BR vs EBRT were constructed. The effect of age at diagnosis on adjuvant RT use was similar for all 3 models: No change in the effect was observed up to approximately age 70 years; and, thereafter, the effect started to decrease linearly. Descriptive nomograms were used in this study to further allow clinicians to understand and observe how specific clinical factors (when taken both alone and in combination) affect the probability of a specific adjuvant RT treatment being delivered. Figure 1 shows the contribution of each effect to the probability of receiving adjuvant RT in any form. For example, the probability of an individual patient being treated with adjuvant RT can be calculated based on their specific factors. The predictor point values (listed across the top of each nomogram), which correspond to each clinical factor, can be added together to create a composite “total points.” Then, this total point value can be correlated with the estimated probability of adjuvant RT treatment (ie, a patient aged 78 years [2.8 predictor point value] who was a woman diagnosed in 1998 [.7 predictor point value] with stage IC/grade 2 endometrial adenocarcinoma [6.3 predictor point value] without a surgical lymph node examination at the time of TAH-BSO [0 predictor point value] in Utah [1.7 predictor point value] would have an estimated probability of being treated with adjuvant RT of 49% [based on total points of 2.8 + .7 + 6.3 + 0 + 1.7 = 11.5]). Thus, differences in the estimated probabilities of adjuvant RT use can be calculated and compared based on differing specific clinical scenarios and geographic locations. For additional clarity regarding the contribution of each factor to the model, we have rescaled the predictor points so that 1 point corresponds to an odds ratio (OR) of 2. For example, going from 2 to 4 points means an OR of 2(4-2) = 4. Higher tumor grade and stage are correlated with greater use of RT: The ORs for RT are 3.4 (stage IB vs stage IA) and 51.8 (stage IC/II vs stage IA). The effect of grade depends on stage: The ORs for stage IA and IB disease are 2.9 (grade 2 vs grade 1) and 11.7 (grade 3–4 vs grade 1); whereas, for stage IC and II, the ORs are 1.5 (grade 2 vs grade 1) and 2.0 (grade 3–4 vs grade 1). Within stage I, increasing substage and grade increase the odds of EBRT (with or without BR) compared with BR alone, with similar EBRT-to-combination therapy ratios. There was a correlated increase in the use of combination therapy as stage increased from stage I to stage II (OR, 2.6).
Age at diagnosis did not affect the overall use of RT or the use of EBRT until age 70 years, after which, both decrease rapidly. No racial effect on choice of therapy was evident (all P > .1). However, since 1988, the overall use of RT has decreased much more for blacks than for other populations (OR for 2002 vs 1988: 0.6 for whites and 0.3 for blacks). We also observed time-related trends for treatment choice: The use of EBRT(with or without BR) has been decreasing since 1994 for patients who had a surgical lymph node examination at the time of TAH-BSO (but not the use of other adjuvant RT types), whereas the use of combination therapy (EBRT + BR), compared with EBRT alone, has been increasing for all patients during the same time. We observed that the largest variability in RT use was between different geographic regions within the SEER Program. The proportion of patients receiving no RT remained stable, whereas the frequency of specific RT modality use varied substantially. Geographic location had a large effect on both overall RT use (the largest OR was 8.7 for Cincinnati vs San Jose) and therapy choice (for EBRT with or without BR vs BR, the largest OR was 11.8 for New Mexico vs Seattle; for combination therapy vs EBRT alone, the largest OR was 7.0 for New Mexico vs San Jose).
Next, we assessed patterns in the selection of specific radiation therapies. For patients who receive postoperative RT, Figure 2 is a descriptive nomogram and illustrates the logistic regression model of adjuvant EBRT (with or without BR) versus BR alone. For patients who receive postoperative EBRT with or without BR, an additional descriptive nomogram is illustrated in Figure 3, which reveals the logistic regression model of adjuvant EBRT + BR versus EBRT alone. For these analyses, we excluded the Native Alaska and Rural Georgia registries, because as both had <10 irradiated cases during the analyzed time frames.
The objective of this study was to evaluate practice patterns quantitatively regarding the use of adjuvant RT in patients with stage I and II endometrial adenocarcinoma in a large United States population database. This study, which, to our knowledge, is the largest reported patterns of care analysis to date of adjuvant RT in stage I and II endometrial adenocarcinoma, reveals a significant diversity in adjuvant RT use across the United States and reflects the lack of a national consensus on adjuvant treatment for early-stage disease. In addition to differences in frequency of adjuvant RT use and differences in the specific RT modality used, geographic and age-based practice patterns also were identified. On the basis of these population data, descriptive nomograms for adjuvant RT were constructed to illustrate further the estimated probability of postoperative RT and RT modality based on clinical and pathologic factors.
A central strategy and goal of physicians is to optimize patient outcomes and customize therapy based on risk stratification. In the past, multiple studies have demonstrated that patients with stage I endometrial carcinoma who receive treatment according to risk-stratified models experience 5-year overall survival rates of 80% to 90% and 5-year cancer-specific survival rates of 90% to 95%.1–3, 8–11 Randomized trial results reported by Aalders et al. revealed that the addition of EBRT after TAH-BSO and vaginal BR led to reduced vaginal and pelvic recurrence rates, although a statistically significant survival benefit was not observed (5-year survival rate: 89% vs 91%).1 It is noteworthy that their report also suggested a survival benefit for the specific subgroup of patients with grade 3 tumors and deep (stage IC) myometrial invasion. Because of these previous reports, it was decided to exclude this higher risk subgroup from random assignment within the multicenter PORTEC trial of stage I disease.2 Creutzberg et al. recently reported that those high-risk patients (stage IC/grade 3) who were excluded from randomization but who were registered experienced a 5-year locoregional recurrence rate of 14%, a distant metastasis rate of 31%, and an overall survival rate of 58% (P < .0001) compared with a 5-year locoregional recurrence rate of 1%, a distant metastasis rate of 3%, and an overall survival rate from 83% to 85% for stage IA through IC/grade 1 and 2 tumors within the randomized portion of the PORTEC trial. Those investigators reported that grade 3 disease was the most important adverse prognostic factor on multivariate analysis associated with recurrence and death as a result of endometrial cancer.4 Their study illustrates that, despite published results, practice patterns across the United States have remained nonuniform and vary greatly by geography and patient clinical factors.
The GOG staging study revealed that microscopic pelvic lymph node metastases were present in 18% of patients who had clinical stage I with deep myometrial invasion (defined as the outer one-third of the myometrial wall) compared with <10% for patients who had more superficial disease invasion into the myometrial wall.17 Meerwaldt et al. also reported an increase in the locoregional recurrence rate for the specific subgroup of patients with stage IC/grade 3 disease.18
Although multiple, elegantly performed patterns of care analyses studying RT use have been performed for other less common gynecologic malignancies, such as cervical cancer,19–24 few reports exist in the literature that provide details on postoperative practice patterns for the more common of gynecologic malignancies, endometrial cancer. Recently, Small et al. reported important results from an American Brachytherapy Society survey regarding the specifics of vaginal BR practice patterns.25 That survey provided clinicians a glimpse of the variability in practice patterns that exist for this procedure in dose fractionation, frequency of use, applicator type, and frequency of low-dose-rate or high-dose-rate approaches. The most common factors used in daily practice to determine whether patients are good candidates for adjuvant therapy are patient age, stage, grade, presence of lymphovascular involvement, surgical margin status, number of lymph nodes sampled at the time of surgery, low uterine segment involvement, and tumor size.
The variability and controversy of postoperative therapy in early-stage endometrial cancer is not limited to the role of adjuvant therapy but also exists regarding primary therapy. A recent analysis from Florida documented significant differences in primary therapy among women who were treated by a general gynecologist or a gynecologic oncologist.26 When they were treated by a gynecologic oncologist, women in the study were more likely to undergo comprehensive surgical staging and were less likely to be referred for adjuvant therapy. Likewise, Macdonald et al. from our institution have reported that significantly more women will undergo adequate staging (79% vs. 23%) with fewer referrals for adjuvant RT (27% vs. 39%) when they are treated primarily by a gynecologic oncologist rather than a general gynecologist.27 Unfortunately, the data available within the SEER database are not adequate to investigate the role of specialty care further within this patient population. This analysis did reveal that, as expected, the relative frequency of adjuvant RT increased for patient subsets with higher grade and disease stage. It is also noteworthy that this analysis revealed an increase in adjuvant RT use for patients who have undergone a surgical lymph node examination, but it did not demonstrate a large difference in adjuvant RT use based on treatment decade.
Several limitations of the current United States population-based study must be considered. Although there is a reported 97.5% ascertainment at the participating SEER sites, these sites only comprise approximately 10% of the United States population.14 In addition, specific RT treatment details with known prognostic significance are not readily available, including information regarding RT dose (dose per fraction and total dose), timing of RT treatment (EBRT and BR) after surgery, BR applicator type, and/or whether high-dose-rate or low-dose-rate BR was used.13 Therefore, we were unable to adjust for these factors in our analyses.
Endometrial cancer remains the most common gynecologic malignancy in the United States. This study, which, to our knowledge, is the largest reported patterns of care analysis to date of adjuvant RT in early-stage endometrial adenocarcinoma, reveals that higher tumor grade and stage lead to more common use of adjuvant RT. In addition, as the largest reported patterns of care analysis to date of adjuvant RT in stage I and II endometrial adenocarcinoma, this study reveals a significant diversity in adjuvant RT use across the United States and reflects the lack of a national consensus on adjuvant treatment for early-stage disease. Along with differences in the frequency of adjuvant RT use and differences in the specific RT modality used, geographic and age-based practice patterns also were identified. On the basis of these population data, descriptive nomograms for adjuvant RT were constructed that further illustrated the estimated probability of postoperative RT and RT modality based on clinical and pathologic factors. It should be emphasized that statistical analysis cannot replace clinical judgment when considering the individual patient, the tumor characteristics, and the potential risks and benefits of adjuvant therapy. With further treatment advances and information learned through research, our hope is that treatment algorithms will continue to evolve such that a uniform consensus can be reached regarding postoperative treatment decisions for specific patient cohorts with early-stage endometrial adenocarcinoma.
- 10Benefit of external irradiation in pathologic stage I endometrial carcinoma: a prospective clinical trial of 605 patients who received postoperative vaginal irradiation and additional pelvic irradiation in the presence of unfavorable prognostic factors. Gynecol Oncol. 1990; 38: 99–104., , .
- 11A prospective trial of postoperative vaginal radium/cesium for grade 1–2 less than 50% myometrial invasion and pelvic radiation therapy for grade 3 or deep myometrial invasion in surgical stage I endometrial adenocarcinoma. Cancer. 1990; 66: 1133–1138., .
- 12National Comprehensive Cancer Network. NCCN guidelines website. Available at URL: http://www.nccn.org Accessed March 9.
- 13National Cancer Institute. Surveillance, Epidemiology, and End Results (SEER) Program public-use data (1973–1999). National Cancer Institute, DCCPS, Surveillance Research Program, Cancer Statistics Branch, released April 2002, based on the November 2001 submission. Available from URL: http://seer.cancer.gov/publicdata/ Accessed May 6, 2005.
- 14National Cancer Institute. Surveillance, Epidemiology, and End Results (SEER) Program web site, data quality. Available from URL: http://seer.cancer.gov/about/quality.html Accessed May 6, 2005.
- 15International Classification of Diseases for Oncology, 2nd ed. Geneva, Switzerland: World Health Organization; 1990., , .