Endometrial cancer is the most common malignancy of the female reproductive tract in the United States, with over 54,000 cases estimated to be diagnosed in 2016; its incidence exceeds that of cervical, ovarian, and vaginal cancers combined.1 Though many cases present with early-stage disease cured by surgery alone, total annual deaths in the United States from this disease exceed 10,000.1
Endometrial cancers have been classified into type I and type II cancers based on their risk factors and prognosis. Type I comprises 80% of all endometrial cancers and is associated with obesity, hyperestrogenism, nulliparity, diabetes mellitus, and hypertension.2 These tumors are usually low grade and have limited myometrial invasion. Patients with type I disease have a 5-year survival rate over 85%.2 Type II disease, which includes the more aggressive histologies of uterine papillary serous carcinoma (UPSC) and clear cell carcinoma (CC), occurs in older, non-obese women, and portends a poorer prognosis. In contrast to type I, type II disease is not typically estrogen-driven.2 Less is known about specific risk factors for type II because most studies have inadequate cases to examine these rare tumors in isolation. Although most endometrial cancers are type I, a disproportionate amount of the 10,000 annual endometrial cancer deaths are a result of type II cancers. For context, UPSC, CC, and grade 3 endometrioid adenocarcinoma (EAC) patients represent 10%, 3%, and 15% of endometrial cancers but are responsible for 39%, 8%, and 27% of endometrial cancer deaths, respectively.3
Even though EACs have increased overall survival (OS) compared with UPSC and CC, grade 3 (G3) EACs behave far worse than their low-grade counterparts. In several single-institution series, G3 EAC has been shown to have similar clinical outcomes to UPSC and CC in terms of OS, suggesting they should be classified as type II.4–9 However, other studies—including a retrospective review using the Surveillance, Epidemiology and End Results Program (SEER)—show that G3 EAC has a significantly better OS than the high-risk histologies.3,10,11 In studies on immunohistochemical analysis, G3 EAC patients have an “intermediate immunoprofile” between the 2 groups of low-grade EAC and UPSC.12
Given these conflicting results, determining the prognosis and optimal treatment for G3 EAC has been challenging. We sought to compare the OS of G3 EAC with that of grade 1 to 2 EAC, UPSC, and CC using the National Cancer Data Base (NCDB). Due to the relative rarity of G3 EAC, UPSC and CC, the NCDB is an ideal resource for such a comparison with the large number of patients included in this database. By gaining a better understanding of the prognosis of patients with G3 EAC, physicians may be able to recommend a more appropriate treatment. Furthermore, this insight can help design future clinical trials. To our knowledge, this is the largest study to date assessing survival outcomes of UPSC/CC and comparing them to G1-3 EAC by stage.
We used NCDB registry data from 1998 to 2012 to study OS in patients with International Federation of Gynecology and Obstetrics (FIGO) stage I to III endometrial cancer by performing a retrospective analysis. The NCDB was established in 1989 and is a nationwide, facility-based, comprehensive data set that currently captures 70% of all newly diagnosed malignancies in the US each year. The NCDB is a joint project of the American Cancer Society and the Commission on Cancer of the American College of Surgeons. The American College of Surgeons has a data use agreement with each of its Commission on Cancer accredited hospitals. This study was granted an exemption from our institution’s Program for the Protection of Human Subjects.
The NCDB was queried for patients diagnosed with FIGO stage I to III (2009 Classification) UPSC, CC, and endometrioid adenocarcinoma from 1998 to 2012 who underwent surgery as definitive treatment. A flow chart outlining patient selection is shown in Figure 1.
We had access to NCDB data for the uterine corpus from 1998 to 2012. Patients with the histologies of endometrioid adenocarcinoma, UPSC, and CC were identified by International Classification of Diseases for Oncology, 3rd edition (ICD-O-3) site codes (8011, 8210, 8261, 8262, 8263, 8380, 8381, 8382, 8383, 8480, 8481, 8482, 8560, 8570, 8934; 8010, 8050, 8260, 8441, 8460, 8461; and 8005, 8310, respectively).
Patients with unknown survival status, unknown grade/stage, non-surgical primary therapy, other histologies, and less than 30-day follow-up were excluded (Fig. 1). The primary outcome examined was OS, stratified by stage, for grade 1 EAC versus grade 2 EAC versus grade 3 EAC versus UPSC versus CC. The NCDB does not offer information on recurrence or cancer-specific survival; although the NCDB does collect data on recurrence, these data are not made available to researchers due to concerns of accuracy and completeness.
Survival times were calculated from the time of diagnosis to the date of death or last contact. Overall survival was calculated using the Kaplan-Meier product-limit method and compared using log-rank tests. χ2 Tests were used for comparing descriptive statistics. To determine survival differences between G3 EAC versus G1 EAC versus G2 EAC versus UPSC versus CC, χ2 tests and Kaplan-Meier analyses were used. P values less than 0.05 were considered statistically significant.
Cox proportional hazards regression was used to examine the impact of covariates on the outcome of death. Multivariate analyses were performed to determine likelihood of histology subgroups receiving adjuvant therapies of either chemotherapy or radiation therapy (RT). Covariates included in our model were race, age, stage, grade, histology, Charlson/Deyo comorbidity score, receipt of adjuvant chemotherapy, and receipt of adjuvant RT. Due to the small proportion of cases with a Charlson/Deyo comorbidity score over 2, the NCDB data were truncated to scores of 0 (no reported comorbidities), 1, or 2 (greater than 1 comorbidities reported). We categorized the covariate of race into the groups of white, black, or other. We bracketed the covariate of age into 2 groups: younger than 60 years and 60 years and older. The impact of covariates on survival was assessed among patients stratified by histology group (G1 EAC vs G2 EAC vs G3 EAC vs UPSC vs CC) within each stage. All statistical analyses were performed using SAS (version 9.4, SAS Institute, Cary, NC).
Baseline Patient Demographics
A total of 219,934 patients met our inclusion criteria, and their characteristics are summarized in Table 1. Of the patients included in this study, the median age at diagnosis was 62.3 years (interquartile range, 55–70), and the majority of patients were white (87.8%). The FIGO stages I, II, and III accounted for 174,361 (79.3%), 17,361 (7.9%), 28,212 (12.8%) of the study population, respectively. Within the study population, 100,532 (45.7%) were G1 EAC, 73,104 (33.2%) were G2 EAC, 28,510 (13.0%) were G3 EAC, 3391 (1.5%) were CC, and 14,397 (6.6%) were UPSC. Among these 5 subgroups, there was no significant difference in Charlson/Deyo comorbidity scores (p = 0.651); however, there was a significant variation in terms of age, race, stage, and receipt of adjuvant therapies among the subgroups (all Ps < 0.0001). Patients with UPSC and CC were older than EAC. There were more blacks among those with UPSC/CC than with EAC. The UPSC, CC and G3 EAC were more likely to have stage III disease than G1-2 EAC. Patients with G3 EAC, CC, and UPSC were more likely to receive chemotherapy and/or radiotherapy than low-grade tumors. Chemotherapy was given most frequently to patients with UPSC (43.4%) whereas RT was most commonly delivered to patients with G3 EAC (48.8%).
The radiation given was in the form of one of the following: EBRT, vaginal brachytherapy, a combination, or radiation “not otherwise specified.” With increasing stage, VB was given less often and EBRT was increasingly favored; in stage I, over half of all radiation given was VB but in stage III over half of radiation given was EBRT. The mean number of days from surgery to radiation was 68.0 days. The average dose delivered for EBRT was 50.4 and 20.8 Gy for VB.
The NCDB does not provide extensive details on chemotherapy regimen; patients were coded as either “chemotherapy administered, but the type and number of agents is not documented,” “single-agent chemotherapy,” or “multi-agent chemotherapy.” The breakdown by percentage among these choices is listed for the 3 high-risk histologies in Table S5 (http://links.lww.com/IGC/A426). The vast majority (>86%) received multiagent chemotherapy.
On MVA, adjuvant radiation and adjuvant chemotherapy were both significantly protective for improved OS for G3 EAC (RT hazard ratio [HR], 0.6; p < 0.0001; chemotherapy HR, 0.6; P < 0.0001), for UPSC (RT HR, 0.7; P < 0.0001; chemotherapy HR, 0.6; P < 0.0001) and for CC (RT HR, 0.8; P = 0.01; chemotherapy, 0.6; P < 0.0001).
Multivariate Analysis of Treatment Patterns
Multivariate analysis demonstrating likelihood of receiving adjuvant chemotherapy and radiotherapy is shown in Tables S1, http://links.lww.com/IGC/A423 and S2, http://links.lww.com/IGC/A424, respectively. Patients with UPSC (HR, 1.3; P <0.0001) histology were significantly more likely to receive chemotherapy. Patients with 60 years or older, black race, higher Charlson/Deyo comorbidity scores, higher stage, and higher grade were also associated with more chemotherapy use. Older age (≥60), higher stage, receiving adjuvant chemotherapy, UPSC histology, and higher grade were all associated with increased likelihood of receiving radiation. Breakdown by percentage of treatment (surgery alone, surgery plus adjuvant radiation, surgery plus chemotherapy, and all 3) for the 3 high-risk histologies, stratified by FIGO stage, is shown in Table S4, http://links.lww.com/IGC/A425. The trend for all 3 histologies is increased use of adjuvant therapies with increased stage; for stage III, all 3 histologies (G3 EAC, UPSC and CC) had about 32% of patients receive both adjuvant radiation and adjuvant chemotherapy after surgery.
OS of Each Histology Subgroup by Stage
The median follow-up of the entire study population was 56.7 months (range, 1–191 months). For each stage of disease, OS for patients, stratified by histologic grade/type, is shown in Figures 2 to 4. For patients with stage I disease (n = 174,361), 5-year OS was for 92.4% for G1 EAC, 87.8% for G2 EAC, 77.5% for G3 EAC, 74.9% for CC, and 74.6% for UPSC (Fig. 2). For stage II patients (n = 17,361), 5-year OS was 86.7% for G1 EAC, 80.2% for G2 EAC, 62.7% for G3 EAC, 64.3% for CC, and 56.7% for UPSC (Fig. 3). For stage III patients (n = 28,212), 5-year OS was 79.7% for G1 EAC, 68.9% for G2 EAC, 49.6% for G3 EAC, 40.2% for CC, and 35.7% for UPSC (P values <0.0001). The P values for log-rank tests of equality over strata was less than 0.0001, indicating a statistically significant difference in OS among the 5 histologies at each stage.
Other Factors That Affected OS
On multivariate analysis, aged 60 or older (HR, 3.1; 95% confidence interval [CI], 2.7–2.9), black race (HR, 1.13; 95% CI, 1.2–1.3), higher Charlson/Deyo comorbidity scores, higher stage, receiving chemotherapy, UPSC histology (HR, 1.6; 95% CI, 1.4–1.8), CC histology (HR, 1.4; 95% CI, 1.2–1.6) and higher grade were all significantly (P < 0.0001) associated with poorer OS (Table 2). Of note both UPSC and CC histology were associated with worsened survival. The model demonstrated that the strongest quantitative overall predictors of death were age ≥60 (HR, 3.1) and stage III (HR, 3.0). In addition, receiving RT was protective and associated with improved OS (HR, 0.7).
This analysis of over 200,000 patients shows that G3 EAC clearly and consistently has lower OS than G1-2 EAC. However, UPSC and CC have lower OS when compared to G3 EAC. Even when age and all other variables were controlled for, UPSC persisted as an independent statistically significant predictor of worse mortality. There were statistically significant differences in age, race, and adjuvant therapy among the 5 cohorts, though there was no difference in Charlson/Deyo comorbidity scores. Although previous studies have shown that patients with UPSC and CC tend to be older in age compared with patients with EAC, this study joins a SEER analysis by Hamilton et al in being among the first that are adequately powered to demonstrate a statistically significant increase in age at diagnosis for high-risk histologies.6,13–15 Furthermore, patients with UPSC and CC have worse OS despite being treated more aggressively with adjuvant chemotherapy and/or RT. The lower OS associated in patients receiving chemotherapy in our study is most likely due to it being given in those patients with the poorest risk factors.
The optimal role of chemotherapy and RT in endometrial cancer has yet to be clearly elucidated. A recent survey of SGO members by Pereira et al. highlighted the discrepancy in adjuvant therapy for stage I endometrial cancers amongst clinicians.16 Due to the paucity of data and lack of randomized trials for rare histologies such as UPSC and CC, there is debate over the ideal adjuvant therapy for these high-risk subgroups. One hypothesis is that combined therapy would confer a survival benefit in endometrial cancer patients by improving both local and distant control; GOG 249, GOG 258, and PORTEC 3 are 3 large randomized collaborative group trials currently investigating this. The design and objectives of these trials demonstrate the complexity and uncertainty in defining endometrial cancer patients who are at risk for recurrence after surgery and choosing the appropriate adjuvant therapy.
Although UPSC and CC have been lumped together with G3 EAC over the past 2 decades as “high-risk” tumors, there has been controversy over whether there is a significant survival difference between patients with these histologies and whether to treat them as distinct disease entities. The above mentioned cooperative group trials may or may not answer this question since UPSC/CC will likely be under represented due to their rarity. Our data, which includes a relatively large number of patients with rare histologies, suggests G3 EAC has better survival compared to UPSC/CC.
Interestingly, some endometrial cancer trials in the past decade have included UPSC and CC while others have chosen to exclude them. Most prior studies of this nature were performed at single-institutions with limited patient numbers and were often inadequately powered to detect significant differences in survival (see Supplementary Table 3, http://links.lww.com/IGC/A422).
The SEER analysis by Hamilton et al3 analyzed 4180 women from the SEER program (SEER database) from 1988 to 2001. Of these patients, 1473 had UPSC and 391 had CC while 2,316 had G3 EAC. As in our study, the UPSC and CC cohorts tended to be older and had a higher proportion of black patients. One advantage of SEER over the NCDB is the inclusion of disease-specific survival. 5-year disease-specific survival (DSS) for UPSC, CC and G3 EAC (Stages I-IV) were 55%, 68%, and 77%, respectively, and DSS differences persisted after controlling for stage.3 The SEER study concluded that UPSC and CC predict for poorer survival compared to G3 EAC, which is consistent with our findings.3
In contrast, other smaller studies have found no statistically significant survival difference between G3 EAC and UPSC/CC. A single-institution study by Cirisano et al7 of 574 patients from 1967 to 1990 (53 UPSC and 18 CC patients) demonstrated a 5-year OS for UPSC + CC of 56% compared to 71% for G3 EAC; however, results were not statistically significant (P value = 0.11), perhaps due to the limited number of patients with UPSC and CC. Creasman et al5 performed a similar study of 523 patients from the FIGO database and found no survival difference between Stage IB-C UPSC compared to G3 EAC. Alektiar et al6 studied 83 patients with Stage I-II endometrial cancer and did not find an OS difference when directly comparing G3 EAC (n = 41) to UPSC/CC (n = 42). Voss et al4 included 131 patients with either G3EAC or UPSC/CC and found disease-specific and recurrence-free survival to be similar between these 2 subgroups. This study concluded that G3 EAC would be better categorized as a type II endometrial cancer and treated with the same adjuvant therapies as UPSC/CC.4 However, our data suggest that UPSC and CC may require even more aggressive treatment than G3 EAC since UPSC/CC have a lower OS than G3 EAC.
It is unclear why prior retrospective studies have shown that G3 EAC has similar OS as UPSC and CC whereas our data along with the SEER study demonstrate worse OS with UPSC and CC. Since UPSC and CC are relatively rare it is difficult to analyze large numbers of these patients and to demonstrate statistical significance. It is unlikely that there will be a randomized study to evaluate these histologies separately so physicians will be limited to retrospective studies to determine prognosis and treatment. Our study is the largest study on this topic thus far (219,934 patients) and may help guide future directions.
In addition to its retrospective nature, our study is limited by the lack of certain data points in the NCDB. As noted, the lack of data on recurrence, patterns of failure, and disease-specific survival are drawbacks. Furthermore, the Charlson/Deyo comorbidity score, while an important benchmark, does not specify the particular comorbidities of each patient. Not knowing the nature of comorbidities does limit conclusions that can be made from NCDB studies. Also, many patients in the initial data set were excluded from our study due to unknown stage. Our results may have changed if this information had been available. Since this was a large hospital-based study the lack of standardization in pathology review and in adjuvant therapy are shortcomings.
Despite these weaknesses, our results provide convincing evidence for the OS of the various grades/histologies in endometrial cancer. The relatively large numbers of UPSC and CC patients in the NCDB allowed for such an analysis. This data may help clinicians in treatment recommendations and guide researchers on future studies.
With one of the largest study populations to date, our results demonstrate that G3 EAC, regardless of more intensive therapy, portends worse survival than G1-2 tumors, but that their OS is not as poor as UPSC or CC. The survival difference between G1-2 EAC versus G3 EAC versus UPSC/CC should be considered in the design of future trials examining adjuvant therapy in endometrial cancer. Further research is needed into refining optimal treatments for these patients.
1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2015. CA Cancer J Clin
2. Zhang Y, Wang J. Controversies in the management of endometrial carcinoma. Obstet Gynecol Int
3. Hamilton CA, Cheung MK, Osann K, et al. Uterine papillary serous
and clear cell
carcinomas predict for poorer survival compared to grade 3 endometrioid corpus cancers. Br J Cancer
4. Voss MA, Ganesan R, Ludeman L, et al. Should grade 3 endometrioid endometrial carcinoma be considered a type 2 cancer-a clinical and pathological evaluation. Gynecol Oncol
5. Creasman WT, Kohler MF, Odicino F, et al. Prognosis of papillary serous, clear cell
, and grade 3 stage I carcinoma of the endometrium. Gynecol Oncol
6. Alektiar KM, McKee A, Lin O, et al. Is there a difference in outcome between stage I-II endometrial cancer
of papillary serous/clear cell
and endometrioid FIGO Grade 3 cancer? Int J Radiat Oncol Biol Phys
7. Cirisano FD Jr, Robboy SJ, Dodge RK, et al. The outcome of stage I-II clinically and surgically staged papillary serous and clear cell
endometrial cancers when compared with endometrioid carcinoma. Gynecol Oncol
8. Soslow RA, Bissonnette JP, Wilton A, et al. Clinicopathologic analysis of 187 high-grade endometrial carcinomas of different histologic subtypes: similar outcomes belie distinctive biologic differences. Am J Surg Pathol
9. Alvarez T, Miller E, Duska L, et al. Molecular profile of grade 3 endometrioid endometrial carcinoma: is it a type I or type II endometrial carcinoma? Am J Surg Pathol
10. Alkushi A, Kobel M, Kalloger SE, et al. High-grade endometrial carcinoma: serous and grade 3 endometrioid carcinomas have different immunophenotypes and outcomes. Int J Gynecol Pathol
11. Boruta DM 2nd, Gehrig PA, Groben PA, et al. Uterine serous and grade 3 endometrioid carcinomas: is there a survival difference? Cancer
12. Halperin R, Zehavi S, Habler L, et al. Comparative immunohistochemical study of endometrioid and serous papillary carcinoma of endometrium. Eur J Gynaecol Oncol
13. Abeler VM, Kjorstad KE. Clear cell
carcinoma of the endometrium: a histopathological and clinical study of 97 cases. Gynecol Oncol
14. Carcangiu ML, Chambers JT. Early pathologic stage clear cell
carcinoma and uterine papillary serous
carcinoma of the endometrium: comparison of clinicopathologic features and survival. Int J Gynecol Pathol
15. Matthews RP, Hutchinson-Colas J, Maiman M, et al. Papillary serous and clear cell
type lead to poor prognosis of endometrial carcinoma in black women. Gynecol Oncol
16. Pereira EB, De B, Kolev V, et al. Survey of current practice patterns in the treatment of early-stage endometrial cancer
. Int J Gynecol Cancer
Endometrial cancer; Histology; Uterine papillary serous; Clear cell; NCDB
Supplemental Digital Content
© 2017 by the International Gynecologic Cancer Society and the European Society of Gynaecological Oncology.