In 1982, Hendrickson et al (1982)
identified UPSC as a clinically aggressive and morphologically distinct variant of endometrial adenocarcinoma. Clear cell endometrial carcinoma was first described in the English literature in 1957. Both cell types have a predilection for distant spread and recurrence (Kay, 1957
; Silverberg and De Giorgi, 1973
; Kurman and Scully, 1976
; Abeler et al, 1996
). Over the last two decades, these histologic types have been grouped together with grade 3 endometrioid cancers as high-risk tumours. However, debate remains whether there is a significant difference in prognosis between these high-risk subtypes and more importantly, if these cell types should be treated as separate disease entities (Munkarah, 2004
This is one of the largest series that compares the clinico-pathologic prognostic factors and outcomes of patients with UPSC and CC vs
G3EC. Previous reports have shown that UPSC is an uncommon uterine cancer but accounts for a disproportionate number of endometrial cancer deaths (Hendrickson et al, 1982
; Carcangiu and Chambers, 1995
). Similarly, in our series, patients with UPSC comprised of only 10% of corpus cancers in our study but accounted for 39% of endometrial cancer deaths. When compared to women diagnosed with G3EC, patients with UPSC accounted for 5% less cases but 12% more deaths. The patients with UPSC and CC were older with median ages at 70 and 68 years compared to 66 years in patients with G3EC. Women with UPSC and/or CC have been described as type II corpus cancers that are associated with older age (Bokhman, 1983
). Although prior reports have consistently demonstrated older age in patients with UPSC and CC compared to endometrioid uterine cancers, this current study is one of first series with sufficient numbers of patients to show a statistically significant increase in age between the high-risk histologies (Abeler and Kjorstad, 1991
; Carcangiu and Chambers, 1995
; Matthews et al, 1997
In this current series, the percent of patients with disease spread beyond the uterus (stage II–IV) at the time of diagnosis was 64, 50, and 40% for UPSC, CC, and G3EC, respectively (P
<0.001). These findings suggest that the poorer prognosis associated with these histologies may be due to advanced disease at the time of diagnosis. In a single institution report of 136 patients, Boruta et al (2004)
also found that patients with UPSC were more likely than those with G3EC to present with extrauterine disease (OR=2.2; 95% CI=1.1–4.5). Additionally, Cirisano et al
, also reported a higher likelihood of upstaging and finding gross metastases among UPSC and CC tumours compared to G3EC controlled for comprehensiveness of surgical staging (Cirisano et al, 1999
). Similar to our series, these reports point towards the more aggressive behaviour of UPSC and CC compared to G3EC. In this current report, we showed that the poorer prognosis of UPSC and CC is not completely explained by the more advanced stage of disease at presentation because the worse prognosis of these cell types holds even after controlling for stage at presentation ().
Other prognostic factors that may impact survival include complete surgical staging and extent of lymph node dissection (Creasman et al, 1987
; Chan et al, 2003
; Huh et al, 2003
). All patients in this study underwent surgical staging based on FIGO criteria and the majority of patients with early stage disease had a lymph node assessment. Previous studies have demonstrated a survival benefit associated with a thorough lymphadenectomy, we proposed to determine if the extent of lymph node dissection may have contributed to the better survival of those with G3EC. Our data showed that the median number of nodes resected between UPSC, CC, and G3EC patients were not statistically different, but UPSC patients continued to have a poorer prognosis. Therefore, using node count as a surrogate, it does not appear that stage migration or an inequality in the extent of surgery contributed to the poorer prognosis of UPSC or CC. Most importantly, higher-risk histologies (UPSC and CC) remained as independent prognostic factors for poorer survival in multivariate analysis.
Previous studies have also investigated the prognostic difference between the high-risk histological types and found similar results. In a series of 139 patients with early stage disease, those with UPSC had a progression-free and overall survival that were significantly worse than those with G3EC (35 vs
=0.03 for progression-free and 43 vs
=0.02 for overall survival) (Boruta et al, 2004
). Similarly, Cirisano et al
, evaluated survival analyses in stage I–II patients and demonstrated a worse 5-year survival in those with UPSC and CC compared to endometrioid carcinoma (56 vs
=0.11). On the other hand, Alektiar et al (2002
) did not find a survival difference between these histologic cell types; however, only 28% of patients in his series underwent a comprehensive surgical staging procedure. More recently, Creasman et al
, performed a large population-based analysis of 523 stage I patients with high-risk cell types from the FIGO database. Although these authors did not find a survival difference in those with stage IA and IB disease, the survival of women with stage IC was poorer in those with UPSC compared with CC or G3EC (Creasman et al, 2004
). In our series, we showed a worsened survival in those with stage IB and IC UPSC compared to G3EC. However, similar to Creasman's series, we were unable to demonstrate a statistically significant difference between those with stage IA disease. summarises prior studies that have evaluated the 5-year survival of high-risk histologies.
Studies comparing the survival of women diagnosed with high-risk corpus cancers
As with other large population-based series, our report was limited by a lack of central pathology review. To determine if there are significant discrepancies between registry and referral pathologists, Piver et al (2000)
, reviewed slides from a large cancer registry and found only 1% of cases had major differences with regard to either site of origin or histopathologic type. Similarly, Tyler et al (1991)
performed slide reviews on 477 women diagnosed with ovarian cancer and compared the diagnoses of pathologists contributing to tumour registries affiliated with the SEER program to an expert panel of three gynecologic pathologists and found and overall agreement of 97%.
Another shortcoming of this study is the lack of information regarding adjuvant chemotherapy. Nevertheless, even in smaller single institution reports, the treatment of patients with UPSC and CC differed significantly. Recently, McMeekin et al (2005)
performed an analysis of 1203 patients diagnosed with advanced or recurrent UPSC, CC, and endometrioid uterine cancers from the Gynecologic Oncology Group, and found that the responses to chemotherapy between these histologic cell types did not differ. However, histology remained as an important prognostic factor. More specifically, the relative hazard ratio for UPSC and CC was 1.20 (1.02–1.40; P
=0.03) and 1.51 (1.1–2.1; P
=0.01), respectively. Thus, even though we were unable to obtain information regarding adjuvant chemotherapy in our study, it appears that the use of chemotherapy cannot completely explain the survival differences observed in the histologic groups. Furthermore, the survival differences between these histologic cell types persist even in those with early stage disease where adjuvant chemotherapy is typically not employed. Another potential concern is that women with UPSC in our study received less adjuvant radiotherapy compared to CC and G3EC patients. Although this difference in radiotherapy may partially explain the poorer outcome of UPSC patients, we showed that radiotherapy did not impact the survival of our patients in multivariate analysis. Moreover, it is possible that many of the UPSC patients underwent chemotherapy rather than radiotherapy given that UPSC has a predilection for distant metastases (Cirisano et al, 1999
; Huh et al, 2003
; Hamilton et al, 2005
). Despite the limited and incomplete information available on adjuvant treatment, our data reflect the lack of community standard which perpetuates inconsistency of approach and variation both within and between treatment centres.
The data from this population-based study allows one to generate interesting hypotheses. The strength in the large number of patients in this series may overcome potential limitations such as the lack of central pathology review and selection biases associated with large population-based analyses. Nevertheless, the demographic and clinico-pathologic data obtained from this report are a true reflection of the trends and outcomes of US women diagnosed with poor histologic uterine carcinoma who receive medical care from community hospitals based on diagnoses from contributing pathologists rather than from gynecologic pathologists in academic centres.
Our data suggests that UPSC and CC are histologically distinct tumours with aggressive tumour biology. Although current treatment modalities for these high-risk cell types have resulted in similar response rates compared to endometrioid tumours, UPSC and CC are discrete histologic subtypes that should be segregated from the more common endometrioid corpus cancers. In this manner, we can effectively design tailored therapies that may improve the outcomes of women diagnosed with these aggressive cancers.