Clinicopathological features and outcome of women with endometrioid and serous ovarian adenocarcinoma were compared.
Clinicopathological features and outcome of women with endometrioid and serous ovarian adenocarcinoma were compared.
Between 1984 and 2004, baseline and follow-up data were prospectively recorded on 1545 patients with ovarian cancer. Of these, 270 had pure endometrioid tumors; 659 had pure serous adenocarcinoma of the ovary. Response to platinum-based chemotherapy (PBC) overall survival, stage-for-stage median progression-free survival (PFS), and cause-specific median survival were compared. Independent predictors of survival were examined by using multivariate analyses.
Median age of diagnosis for patients with endometrioid tumors was younger than those with serous adenocarcinoma of the ovary (60 years vs 62 years; P = .013). They presented more often with early disease (stage I and II; 50% vs 17%; P < .001), had less ascites, and had better performance status both overall and for stage II and III disease. More endometrioid tumors were optimally debulked overall (71% vs 45%; P < .001), but there was no difference according to stage. Objective and CA125 PBC response rates were not significantly different, but median PFS was better for patients with endometrioid tumors (24 months vs 13 months; P < .0001) as was overall median survival (48 months vs 22 months; P < .0001). This relation remained for stage II and III disease and for moderately and poorly differentiated tumors. Patients with concurrent endometrioid ovarian and endometrial malignancies had a survival advantage compared with those with ovarian malignancies alone. Independent predictors of survival after PBC were histological type, debulking status, and disease stage.
Despite similar PBC response rates, endometrioid histology is associated with better survival compared with serous adenocarcinoma of the ovary, even with stage III or poorly differentiated tumors. Cancer 2008. © 2008 American Cancer Society.
Epithelial ovarian cancers are a conglomerate of histopathologically defined entities thought to arise from the ovarian surface epithelium. There has been a lack of publications that specifically address differences in the natural history and distribution of these subtypes. In general, studies have shown that compared with serous adenocarcinoma of the ovary, clear cell carcinoma,1 mucinous carcinoma,2 and carcinosarcoma3 of the ovary all have a poorer prognosis and relative platinum insensitivity. Very few studies have looked specifically at endometrioid carcinoma of the ovary.
Endometrioid ovarian cancer is a specific histopathological entity that accounts for 16% to 25% of epithelial ovarian carcinomas. Histologically, this variant was recognized by the World Health Organization (WHO)4 in 1964. Clinicopathological features include a higher proportion of cases with well differentiated (grade 1) histology and early stage disease. There is also an association with concurrent endometrial carcinomas and endometriosis.5–10
Some authors have suggested that endometrioid pathology is associated with better overall survival than serous adenocarcinoma of the ovary.4, 11–16 However, more recently, there has been speculation that this could be because the majority of these patients have early stage disease.7, 17 To our knowledge, there have been no published stage-for-stage comparisons of the 2 histological types to date. Endometrioid and clear cell variants are postulated to arise from the same cell type,18 and although platinum-based chemotherapy (PBC) is now accepted as the best first-line treatment, Sugiyama et al. found response rates to PBC in ovarian clear cell tumors of 11.1% compared with 72.5% for serous adenocarcinoma of the ovary.1 However, there are few published data that specifically examine the PBC response in endometrioid tumors.
Our aims were to compare a sequential set of patients who were treated for pure endometrioid or serous adenocarcinoma of the ovary at the Edinburgh Cancer Center over a period of 20 years with regard to 1) presenting characteristics, 2) PBC response rates, and 3) survival outcomes (overall and stage-for-stage comparisons). To our knowledge, this is the largest published comparison of this nature to date.
Between 1984–2001, 1545 patients with epithelial ovarian adenocarcinoma were treated at the Edinburgh Cancer Center. After central pathology review by experienced gynecological pathologists (authors A.A.N.and A.R.W.), data were prospectively collected on all patients as part of their routine clinical care. Information collected included age at diagnosis, FIGO (International Federation of Gynecology and Obstetrics) disease stage, tumor grade, surgical debulking status, WHO performance status, concurrent endometrial malignancies, chemotherapy regimen, objective response to chemotherapy, CA125 levels, progression-free and overall survival. Individual informed patient consent was not sought, as these data were routinely obtained and used as part of these patients' clinical care. Ethical approval was given for ovarian cancer specimen collection with annotated clinical information.
Follow-up data up until 2004 were analyzed anonymously for patients with pure endometrioid or pure serous epithelial ovarian cancers. Patients excluded from analysis were those with 1) mixed histology epithelial ovarian cancer (24 patients had mixed serous and endometrioid pathology) and 2) a previous malignancy other than endometrial or basal cell skin carcinoma. Of the 1261 patients for whom ovarian cancer was their first cancer diagnosis (including those with a concurrent endometrial malignancy), 270 (21.4%) had pure endometrioid adenocarcinoma, and 659 (52.2%) patients had pure serous adenocarcinoma of the ovary.
With the exception of early stage disease or where it was technically impossible, surgery included hysterectomy, bilateral salpingoophorectomy, omentectomy, and cytoreductive surgery for those with advanced disease. Lymphadenectomy was not routinely performed, which is normal practice in the United Kingdom.
For patients with postoperatively measurable disease, objective response to chemotherapy was evaluated by computed tomography. WHO and International Union Against Cancer (UICC) response criteria were used. A complete response (CR) was defined as the complete disappearance of all clinically detectable disease confirmed by second evaluation at least 4 weeks later. A partial response (PR) was defined as a >50% decrease in tumor size for at least 4 weeks without an increase in the size of any other known lesion or the appearance of a new lesion. No change was defined as the absence of any significant change in measurable lesions confirmed by a second evaluation at least 4 weeks later. Progressive disease was defined as the appearance of a new lesion or a >25% increase in tumor size. Those patients who had a CR or PR were deemed to have responded to treatment. For those who had a raised CA125 postoperatively, where possible the CA125 response was analyzed according to Rustin's criteria.19, 20
Progression-free survival was defined as the interval from the date of primary surgery until the date of documented recurrence on the basis of clinical examination, ultrasound, or computed tomography. Duration of survival was the date of primary surgery to death or the last follow-up contact. For those who did not have definitive surgery, then these intervals were calculated from the date of diagnosis. Patients who died of a nonmalignant cause or a new malignancy and had no active ovarian cancer were censored in the survival analysis.
Associations between the histological groups, demographic characteristics, and clinical characteristics were assessed by using the Mann-Whitney U test for continuous variables, and the chi-square test or the Fisher exact test was used for categorical factors. Analyses of survival data were performed by using the log-rank test and the Cox regression model, and survival curves were computed by using the Kaplan-Meier method. A 2-sided P value of .05 was considered statistically significant. The statistical software used was SPSS 12.0 for Windows, Release 12.0.1 (SPSS, Chicago, Ill).
The median age of diagnosis for patients with endometrioid tumors was slightly but statistically significantly younger than those with serous adenocarcinoma of the ovary. The majority of tumors were poorly differentiated in both groups, but the proportion was less in the endometrioid group. Those in the endometrioid group were more likely to present with early stage disease. Overall, significantly fewer endometrioid patients presented with ascites. In case this was because endometrioid tumors present with earlier stage disease, stage-for-stage comparisons were performed and showed this was still the case even for stage II and III disease. More advanced disease stages were associated with more poorly differentiated tumors for both endometrioid and serous adenocarcinoma groups (P < .001 for both, data not shown). The WHO performance status (PS) was significantly better in the endometrioid group. Again, in case this was because endometrioid tumors present more frequently with early stage disease and have potentially less comorbidity, performance status was analyzed on a stage-for-stage basis. The relation remained for patients with stage II and III disease.1
|No. (%)||No. (%)|
|Total patients||270 (29)||659 (71)|
|Median age [range]||60 [32–90]||62 [23–88]||.013*|
|I||92 (34)||62 (9)|
|II||43 (16)||53 (8)|
|III||108 (40)||411 (62)|
|IV||21 (8)||128 (19)|
|Unknown||6 (2)||5 (1)|
|Well (grade 1)||25 (9)||43 (7)|
|Moderate (grade 2)||72 (27)||148 (22)|
|Poor (grade 3)||152 (56)||419 (64)|
|Overall||122 (47)||475 (77)||<.001|
|FIGO stage I||28 (32)||24 (42)||.217|
|FIGO stage II||10 (24)||28 (56)||.003|
|FIGO stage III||67 (66)||327 (85)||<.001|
|FIGO stage IV||15 (71)||95 (79)||.571|
|WHO performance status (overall)||<.001|
|0||88 (56)||166 (40)|
|1||52 (33)||156 (37)|
|2||14 (9)||62 (15)|
|3||2 (1)||30 (7)|
|4||0 (0)||2 (1)|
|WHO performance status (FIGO stage II)||.025|
|0||16 (55)||7 (84)|
|1||11 (38)||4 (13)|
|2||1 (3)||1 (3)|
|3||1 (3)||0 (0)|
|WHO performance status (FIGO stage III)||.018|
|0||36 (50)||97 (36)|
|1||26 (36)||115 (43)|
|2||9 (13)||35 (13)|
|3||1 (1)||20 (8)|
|4||0 (0)||1 (0.4)|
|Debulking status (overall)||<.001|
|<2 cm||181 (71)||273 (45)|
|>2 cm||72 (29)||328 (55)|
|FIGO stage I debulked||na|
|<2 cm||91 (100)||61 (100)|
|>2 cm||0 (0)||0 (0)|
|FIGO stage II debulked||.779|
|<2 cm||37 (86)||42 (82)|
|>2 cm||6 (14)||9 (17)|
|FIGO stage III debulked||.061|
|<2 cm||47 (49)||139 (38)|
|>2 cm||49 (51)||227 (62)|
|FIGO stage IV debulked||.406|
|<2 cm||3 (15)||30 (25)|
|>2 cm||17 (85)||90 (75)|
Overall, more patients with endometrioid tumors were debulked to less than 2 cm residual disease. When analyzed on a stage-for-stage basis, there were no statistically significant differences, but there was still a trend for a greater proportion of stage III endometrioid tumors to be debulked compared to serous tumors.
Of patients with endometrioid ovarian tumors, 10% (26 of 270) had a concurrent endometrial malignancy compared with only 0.2% (1 of 659) of patients with serous adenocarcinoma (P < .001). Within the endometrioid group, those patients who also had a concurrent endometrial malignancy had significantly better survival than those who had only an ovarian tumor (median survival not reached vs 46 months; P = .003; Fig. 1). Because the number of patients was small, it was not possible to control for stage in this analysis.
Chemotherapy was administered to 81% (754 of 929) of patients. A significantly smaller proportion of patients with endometrioid tumors had chemotherapy compared with those in the serous adenocarcinoma group; 75% (202 of 270) versus 84% (552 of 659), P = .002, presumably because of earlier stage disease presentation. Platinum-based chemotherapy accounted for 84% (636 of 754) of the total chemotherapy given to both groups.2
|No. (%)||No. (%)|
|Nonplatinum-based chemotherapy||32 (16)||86 (16)|
|Any platinum-based chemotherapy||170 (84)||466 (84)|
|Single agent platinum||122 (72)||294 (63)|
|Platinum+taxane||16 (9)||44 (9)|
|Other combination platinum||32 (19)||128 (28)|
There was no difference between endometrioid and serous adenocarcinoma with regard to the proportion of patients who received single-agent platinum, platinum + taxane, or other combination-platinum chemotherapy (Table 2; P = .077). The other platinum combinations that patients received included cisplatin and prednimustine (67 patients); cisplatin and intraperitoneal alpha-interferon (26 patients); cisplatin and topotecan (25 patients); carboplatin and hexalen (22 patients); cisplatin, 5-fluorouracil, hexalen, and prednimustine (17 patients); cisplatin, doxorubicin, and cyclophosphamide (1 patient); cisplatin and cyclophosphamide (1 patient); cisplatin, 5-fluorouracil, and mitomycin C (1 patient).
Of those patients who received PBC, 272 of 636 (43%) had measurable disease after surgery (126 of 416 received single-agent platinum, 39 of 60 received platinum + taxane, and 107 of 160 received other platinum combinations). The overall objective response rate to PBC was 59%. Upon examining the different PBC regimens, we found that platinum + taxane had a significantly better overall response rate than single-agent or other platinum combinations, (77%, 52%, and 59%, respectively; P = .025).3
|Treatment||Objective response rate|
|Any platinum chemotherapy||30 of 50||60%||129 of 222||58%|
|Single-agent platinum||12 of 22||55%||54 of 104||52%|
|Platinum+taxane||7 of 9||78%||23 of 30||77%|
|Other combination platinum||11 of 19||58%||52 of 88||59%|
Comparison of endometrioid and serous adenocarcinoma groups revealed no statistically significant differences between overall objective response rates to PBC or any of the response rates to the 3 PBC regimens.
CA125 response was evaluable in 373 of 637 (59%) of those who received PBC (251 of 416 received a single agent, 44 of 60 platinum + taxane, and 78 of 160 had other combination-platinum regimens). Overall, the CA125 response rate to PBC was 63%. Platinum + taxane had a significantly better overall CA125 response rate than single-agent or other combinations of platinum (77% vs 63% and 54%, respectively; P = .037).4
|Treatment||CA125 response rate|
|Any platinum chemotherapy||57 of 86||66%||176 of 287||61%|
|Single-agent platinum||42 of 60||70%||115 of 191||60%|
|Platinum+taxane||6 of 9||67%||28 of 35||80%|
|Other combination platinum||9 of 17||53%||33 of 61||54%|
Comparison between the endometrioid and serous adenocarcinoma groups yielded no statistically significant differences between overall PBC CA125 response rates or any of the different PBC regimens.
Median follow-up was 24.5 months (range, 1 month to 229 months). Overall median progression-free survival was significantly longer for those with endometrioid histology (24 months vs 13 months; log-rank P < .0001). One-year, 3-year, and 5-year progression-free survivals are shown in Table 5. Due to the predominance of early stage disease in the endometrioid group, further analyses were performed on a stage-for-stage basis. Median progression-free survival was significantly longer for the endometrioid groups with stage II or III disease (stage II, 47 months vs 26 months; P = .022) (stage III, 15 months vs 13 months; P = .038). However, there were no differences between histologies for those with stage I or IV disease. Comparisons within the stage III endometrioid group showed that those who had debulking surgery to <2 cm residual disease had a significantly longer progression-free survival than those who did not (21 months vs 9 months; log-rank P = .0001). The same was also true for the serous adenocarcinoma group (16 months vs 11months; log-rank P < .0001). Comparison of stage III patients who had debulking surgery to <2 cm residual disease revealed that there was no difference in progression-free survival between the endometrioid and serous adenocarcinoma groups (21 months vs 16 months; log-rank P = .219). Similarly, there were no differences between histologies for those who had >2 cm residual disease (9 months vs 11 months, respectively; log-rank P = .461).
|% Progression-free survival||% Progression-free survival||Log-rank P|
|Debulk <2 cm||38||82||38||22||116||71||28||21||.219|
|Debulk >2 cm||36||32||19||19||153||44||3||2||.461|
Overall median survival was substantially better for patients with endometrioid histology than those with serous adenocarcinoma (48 months vs 22 months; log-rank P < .0001; Fig. 2). One-year, 3-year, and 5-year survivals are shown in Table 6. Shorter survival was associated with increasing disease stage for both groups (log-rank P < .0001). Because of the predominance of early stage disease in the endometrioid group, survival was further analyzed on a stage-for-stage basis. For stages II and III, median survival was significantly better in the endometrioid group (stage II, 81 months vs 46 months; log-rank P = .029; Fig. 3b) (stage III, 33 months vs 22 months; log-rank P = .002; Fig. 3c). However, there were no significant differences between the 2 histologies for stage I or IV disease (Figs. 3a and 3d).
|% Survival||% Survival||Log-rank P|
|Debulk <2 cm||37||95||65||43||114||89||46||28||.144|
|Debulk >2 cm||36||58||30||19||153||71||14||4||.053|
For stage III endometrioid cancer, those who had debulking surgery to <2 cm residual disease survived longer than those who did not (50 months vs 16 months; log-rank P = .004; Fig. 4a). The same was true for the serous adenocarcinoma group (34 months vs 18 months; log-rank P < .001; Fig. 4b).
A comparison of post-PBC survival on the basis of tumor grade showed that patients with moderately differentiated (grade 2) and poorly differentiated (grade 3) endometrioid tumors had significantly longer survival than those with serous adenocarcinoma (moderately differentiated, 51 months vs 27 months; log-rank P = .05) (poorly differentiated, 40 months vs 21 months; log-rank P < .001). There was no difference for well differentiated (grade 1) tumors (median survival not reached vs 88 months; log-rank P = .075).
The univariate analysis showed histology, debulking status, performance status, tumor grade, disease stage, and age were all significant prognostic factors. After these factors were entered into the multivariate analysis, histological type (which is surprising and novel), debulking status, and disease stage were independent predictors of survival.7
|Variables||Hazard ratio||95% Confidence interval||P|
To our knowledge, this is the largest prospectively collected pure ovarian endometrioid carcinoma data set. We believe it is the first to present a detailed comparison with pure serous adenocarcinoma of the ovary regarding both objective and CA125 response rates to PBC as well as stage-for-stage survival comparisons.
In summary, there was no difference in the objective or the CA125 response rate to PBC between patients with endometrioid and serous adenocarcinoma. Despite this, survival was better for patients with endometrioid tumors than for patients with serous adenocarcinoma, even with stage III disease or poorly differentiated tumors. After receiving PBC, histological type was an independent predictor of survival favoring endometrioid adenocarcinoma along with debulking status and FIGO stage.
We did not find any significant differences between endometrioid and serous adenocarcinoma groups regarding either objective or CA125 response rates to PBC. We could locate only 1 other study that made reference specifically to PBC response rates in patients who had ovarian endometrioid tumors.7 They found a 72% response rate, although this was not according to WHO/UICC criteria, so a comparison must be undertaken cautiously. In the current study, 29% of the endometrioid group and 48% of the serous adenocarcinoma group had objectively measurable disease after surgery. These proportions are relatively small but comparable to those presented by Sugiyama et al.,1 who found clear cell tumors had only an 11% response rate to PBC. Sugiyama et al. did not evaluate CA125 response, but in our sample, a greater proportion were evaluable by this method, 51% endometrioid and 62% serous adenocarcinoma.
In common with other studies, our patients with endometrioid tumors had a much better overall survival than those with serous adenocarcinoma.4, 6, 11, 12, 21 It has been suggested that the endometrioid survival advantage may be a reflection of an earlier stage of presentation rather than an effect of histological type itself.7, 17 However, this survival advantage was still present for stage II and III diseases in our sample. It is noteworthy that this was despite WHO performance status being better in the serous adenocarcinoma group for stage II disease. The pattern was the same for progression-free survival. Grosso et al.6 compared 106 endometrioid cases with 355 cases of serous adenocarcinoma of the ovary and found 5-year survival was better for the endometrioid group (56% vs 41%); however, the Grosso study did not compare survival on a stage-for-stage basis. Bjorkholm et al.21 compared a number of histological ovarian cancer variants and found endometrioid tumors had a more favorable survival than serous adenocarcinoma, but they compared only within stages IA and IB. Zwart et al.22 compared 21 patients matched for age, stage, grade, and level of cytoreduction and found no significant 5-year survival differences between their endometrioid and serous adenocarcinoma histology groups. With the exception of Grosso et al., none of these reports included patients who received PBC.
Other authors have stated that the majority of endometrioid tumors are well differentiated or moderately differentiated.7 In our series, we found that the majority (56%) of endometrioid tumors were, in fact, poorly differentiated. However, this was still a smaller proportion than the serous adenocarcinoma group (64%). Post-PBC survival comparisons on the basis of tumor grade showed that patients with moderately or poorly differentiated endometrioid tumors still had a survival advantage when compared with serous adenocarcinoma patients. Malkasian et al.15 performed multivariate analysis on sample of 1938 women with epithelial ovarian cancers, among which 958 were serous adenocarcinoma of the ovary and 75 were endometrioid. They grouped histological types on the basis of stage and grade combinations. Generally, there was no difference in survival between histological subtypes, with the exception that patients with endometrioid tumors did better than patients with serous adenocarcinoma in stage IA, grade 2; stage IB/C, grade 2; and stage III, grade 3 groupings. However, Malkasian et al. do advise cautious interpretation of these results, as their number of patients was very small, and there were multiple statistical comparisons.
The prevalence of endometrioid tumors in our sample was comparable to other published studies. In common with other reports,7 our finding was that pure endometrioid tumors were the second most common epithelial ovarian cancer (21.4%), surpassed only by pure serous adenocarcinoma of the ovary (52.2%). Our findings confirmed that, in comparison to the serous adenocarcinoma group, the endometrioid group had a greater proportion of stages I and II disease5, 7, 17, 23 and better performance status overall. A smaller proportion of the endometrioid group than the serous adenocarcinoma group had ascites associated with stages II and III disease, although, to our knowledge, there are no other published data for comparison.
In our data set, 10% of endometrioid cases were associated with a concurrent uterine endometrial malignancy. This was comparable to other studies6, 9 (11% to 23%), and, in common with other published data, we found this group had a survival advantage over those who had an ovarian malignancy only.8, 10 However, other authors report conflicting findings.5–7, 9
Consistent with previously published studies, for both groups, patients who had debulking surgery to <2 cm residual disease before PBC survived longer and had a longer time to disease progression than those patients who did not have debulking surgery.
We found histological type, debulking status, and stage were independent predictors of survival in patients who received platinum-based chemotherapy. To our knowledge, this is the first study that is large enough to have allowed investigation of the effect of histological type in this chemotherapy group. Kline et al.7 found significant factors associated with survival were stage, grade, and extent of debulking; however, these were not put into a multivariate statistical model. They proposed that histological type could be a prognostic factor. Grosso et al.6 used univariate analysis and found that FIGO stage, grade, residual disease, lymph-node status, and platinum chemotherapy were all prognostic factors; however, only stage was significant in their multivariate model. Voest24 performed a meta-analysis of 3443 patients with epithelial ovarian cancer and found that significant independent prognostic factors were cisplatin chemotherapy, debulking to <2 cm, FIGO stage, and performance status.
Endometrioid carcinomas of the ovary have distinct molecular features that differentiate them from other types of ovarian cancer. PTEN is a tumor-suppressor gene that negatively regulates the PI3K pathway, which is involved in cell growth and survival. Endometriod ovarian tumors have been found to have frequent loss of heterozygosity (45% to 75%) or mutations25–27 (≈ 20%) of PTEN suggesting frequent deletion or mutational inactivation of this tumor-suppressor gene. Activating mutations of PIK3CA, which codes for part of PI3K, have also been found more commonly in endometrioid ovarian tumors than in histological serous subtypes (20% vs 2.3%, respectively).28 Catenin B alterations have been described in endometrioid ovarian cancers.29 p53 is a transcription factor, which regulates the cell cycle via transactivation of several genes involved in arresting the cell cycle or in promoting apoptosis. The efficacy of cisplatin is thought to be dependent on the normal functioning of p53. Endometrioid tumors generally have normal p53 expression, and serous tumors are more commonly associated with mutant p53 and poorer survival.30, 31 In our sample, the response rates to platinum for endometrioid and serous adenocarcinoma histologies were similar, suggesting that p53 gene status may not be an important determinant of platinum response. It was beyond the scope of this study to analyze individual tumors by these and other molecular features; however, we plan to undertake this analysis once a tissue microarray linked to these data has been generated.
In conclusion, endometrioid histology is associated with a better survival compared with serous adenocarcinoma, regardless of response to platinum-based chemotherapy or disease stage, and is an independent predictor of survival. Recent considerations for trial design in ovarian cancer globally have begun to focus on histology. Several trials based on treatment by histological type are now being designed or have commenced (Japanese Gynecologic Oncology Group/Intergroup Clear Cell Carcinoma study and National Cancer Research Institute's Mucinous Epithelial Ovarian Cancer Trial [MEOC] and Scottish Gynecological Cancer Trials Group's Carcinosarcoma studies), although these approaches remain controversial. On the basis of our findings from this study, we conclude that the chemotherapeutic response of endometrioid ovarian cancer does not differ sufficiently to warrant this approach, and we would not recommend that these patients are treated differently with chemotherapy from serous adenocarcinoma patients. However, with clarity emerging, molecular lesions for endometrioid ovarian cancer demonstrate significant heterogeneity within endometrioid tumors32 and significant differences compared with serous adenocarcinoma of the ovary. There may nevertheless be a future role for separating these tumors for different approaches when integrating molecularly targeted therapy with chemotherapy.