Endometrial carcinoma (EC) is the most common genital malignancy in women occurring with a cumulative lifetime risk of approximately 1.5–3%.1, 2 Predominantly a disease of postmenopausal women, EC presents at a median age at diagnosis of 63 years.3 Risk factors associated with EC include unopposed estrogen use, obesity, late menopause, early menarche, nulliparity, hypertension, diabetes mellitus and hereditary tumor syndromes.4, 5 An estimated 5% of all cases of EC is associated with a hereditary cause.6
Hereditary nonpolyposis colorectal cancer (HNPCC), an autosomal dominant cancer syndrome, is caused by a germline mutation in DNA mismatch repair (MMR) genes, most frequently in MLH1, MSH2 and MSH6.7 These mutations predispose to cancer of the colorectum and endometrium, most commonly, and to cancers of various other locations such as the stomach, upper urinary tract, ovaries and small bowel.8 The cumulative incidence of HNPCC-associated EC is approximately 20% by age 70.2 Some MMR mutations in female carriers increase the risk of EC to 40–50%.9, 10, 11, 12
Improved survival has been reported in HNPCC-associated colorectal cancer (CRC) as compared to sporadic CRC of an equivalent stage.13 The 5-year cumulative survival rate of CRC associated with MLH1-mutations is 65% in comparison to 44% in sporadic CRC.14 It is unknown, however, whether the survival rate of other HNPCC-associated tumors is improved compared to that of the sporadic occurrence. The aim of our study was to compare the survival rates of HNPCC-associated EC with apparently sporadic EC.
MATERIAL AND METHODS
Sixty-six patients with EC from 46 families were identified for the survival analysis. These cases were derived from 184 families who were registered in the Netherlands Foundation for the Detection of Hereditary Tumors, a registry developed in 1987 to promote the surveillance in families with HNPCC. The methods used by the HNPCC registry have been reported previously.15 The 66 identified EC cases were from families found to harbor a germline HNPCC mutation or met the previously defined Amsterdam Criteria II.16 The cancer diagnosis was verified by medical record or pathology report.17 Of the 66 patients, 16 patients were excluded from the survival analysis. The FIGO stage was unknown in 13 of the 16 patients, and 2 patients with FIGO stage IIIB and one patient with FIGO stage IIIC could not be stage-matched with controls. The remaining 50 patients comprised the study group for the survival analysis. The time period of diagnoses of EC in the study group ranged from 1957–96. Forty-two of the 50 (84%) cases were diagnosed after 1970.
The clinical information, age at diagnosis, stage, histologic subtype, as well as survival data was collected for patients in the study group.
Patients from the study group were matched with a control group of patients with EC from the Eindhoven Cancer Registry (IKZ), a registry covering an area of approximately one million inhabitants in the Southeast of the Netherlands. Because FIGO stage was not sufficiently registered before 1993, we used only data from patients registered later than 1993. A database of 440 patients with EC registered at The Eindhoven Registry from 1993–97 and lacking other malignancies was identified. Two matched controls were found for each study case. Matching was based on age at diagnosis and the FIGO stage. Complete data on the histologic subtype of the tumor also were collected for each case in the control group for comparison.
Germline mutations in study group
Germline mutations were confirmed in 38 of the 50 cases that comprised the study group. Nineteen patients harbored an MLH1 mutation, 16 an MSH2 mutation and 3 an MSH6 mutation. The germline mutation status in the remaining cases was unknown.
The observation time of the survival analysis was from the date of diagnosis until death or the end of the study in December 2000. In the analysis, patients who died of CRC in the study group were withdrawn alive and censored on their date of death. Differences in survival were tested using the Log-rank test. The Fischer's Exact test was used to detect the differences in the distribution of tumor histologic subtypes between groups. Results with p < 0.05 were considered to be significant.
Clinical characteristics of study and control groups
The clinical characteristics including age at diagnosis, stage and tumor histologic subtype were collected and evaluated. The stage distribution of the cases in the study group were 78% Stage I and 22% Stage III compared to 79% Stage I, 8% Stage II, 8% Stage III and 5% Stage IV observed in the total population of EC cases in the Eindhoven Registry in the period 1993–97. Table I presents the cumulative data for both the study and control groups.
Table I. Characteristics of Patients with HNPCC-Associated Endometrial Carcinoma versus Age- and Stage-Matched Control Patients with Endometrial Carcinoma1
The overall mean age at diagnosis of EC in the study group (n = 50) was 49.9 years (range 31–69) and 53.7 years (range 30–72) in the control group (n = 100) (p = 0.004). Seventy-eight percent of the study and control groups presented with FIGO Stage I, and 22% presented with Stage III. There was no significant difference in the distribution of tumor histologic subtypes in the study and control groups (p = 0.55). Study group patients in Stage II were not found in the registry. The mean age at diagnosis of Stage IA, IB and IC EC in the study group (n = 39) was 50.4 (range 33–65) and 52.2 (range 36–69) in the control group (n = 78) (p = 0.13). In Stage IIIA and IIIC EC, the mean age at diagnosis was 48.0 years (range 31–69) and 59.0 years (range 30–72) in the study group (n = 11) and control group (n = 22), respectively (p = 0.009). Similar to the total analysis, the distribution of tumor histologic subtypes in the study and control groups in Stage I or Stage III was not significantly different (p = 0.28 for the cases and controls with Stage IA, IB and IC, and p = 0.71 for the cases and controls with Stage IIIA and IIIC).
Of the 50 patients with HNPCC-associated EC, 32 (64%) patients were still alive at the time of the last follow-up. Eighteen (36%) patients died at a calculated mean age of death of 61.1 years (range 43–87). Seven patients died of complications associated with EC, five patients with CRC, four with another cancer, one due to a non-tumor associated cause, and one of an unknown cause. The survival rates for both the study and control groups were not significantly different (Fig. 1). The overall cumulative 5-year survival rate was 88% and 82% for patients with HNPCC-associated and sporadic EC, respectively (p = 0.59). The survival rate was 92% in Stage IA, IB and IC patients with HNPCC-associated EC and 91% for patients with sporadic EC (p = 0.90). The survival rates of patients with Stage IIIA and IIIC EC were also not significantly different, 72% and 50% respectively (p = 0.38). No difference in survival was observed between patients in the study group with a mutation identified and those without an mutation.
Results of our study show that the survival rate of patients with EC from HNPCC families is not significantly different from that of patients with sporadic EC. The distribution of histologic subtypes of EC is also similar in these 2 groups.
In CRC associated with HNPCC, there is a survival benefit due to differences in the biological behavior of the HNPCC-related tumor compared to sporadic CRC.14 CRC in HNPCC behaves differently with regard to frequency of multiple lesions, location and reduced metastatic tendency.18 A possible explanation for the better prognosis of patients with CRC associated with HNPCC might be that CRC in HNPCC is more immunogenic than its sporadic counterpart due to the microsatellite instability (MSI) caused by MMR defects. A survival benefit is not found in EC associated with HNPCC suggesting a likeness in the biological behavior of both the HNPCC-associated and sporadic form. The similarity in histologic subtype distribution further implies that these forms are biologically similar. Parc et al.19 showed that EC with MSI is associated with high FIGO stage and grade, cribriform growth pattern, mucinous differentiation and necrosis. We also noted a relatively high number (22%) of FIGO Stage III patients in the study group, which presented at a relatively early age, yet this did not translate into decreased survival. Due to this high number of young Stage III patients, we were unable to successfully age match this group that inadvertently affected the age matching of the total study population. Subdivided into Stage I and III groups (there were no Stage II patients in the study group), the majority of the patients in our study presented with Stage I, and their age matching is shown to be correct. Because the Stage III patients that contributed to the mismatching were younger than their controls, and young age is associated with improved survival,20 this would result in a bias toward an improved survival. Survival would be expected to be improved due to improper age matching, yet this was clearly not the case. The EC patients in the control group represent a relatively young group of sporadic EC cases. Because HNPCC-associated EC is diagnosed approximately 15 years earlier than in the general population,21 it is possible that the control group may have included unknown cases with HNPCC MMR mutations, thus obscuring a difference in survival between the 2 groups. The effect would have been minimal, however, because previous studies have indicated that <5% of cases of EC diagnosed in the general population at a young age (<55 years or <60 years) is due to a hereditary cause.22, 23
The cases in the study group were diagnosed over a long period of time (1957–96). Because the treatment methods may have been substantially changed over time, the outcome of the patients in the study group might have been biased to a worse survival. Data from the Eindhoven Cancer Registry,24 however, indicated that the survival did not change in the period 1970–97: the relative 5-year survival rate of patients diagnosed between 1970–78 was 84% compared to 86% for cases diagnosed between 1988–93 and 84% for cases diagnosed between 1994–97. Moreover, none of the cases from the study group diagnosed before 1970 died from EC.
With the initial typical presenting symptom of irregular uterine bleeding, the diagnosis of EC is usually in its early stages and thus, has a relatively high survival rate. Partially attributable to the effective methods of detection and treatment, EC is the most curable of the 10 most common cancers in women.5 With such similarly high survival rates, the power of the study would need to be very high to detect a slight difference. Importantly, findings of our study exclude a large difference in survival rates. The Netherlands Foundation for Detection of Hereditary Tumors is the largest registry of HNPCC families and therefore, represents the largest number of standardized cases on which a study could be carried out. To detect a small significant difference, a study that combined several HNPCC registries would need to be carried out.
We would like to thank L. van der Heijden for his assistance with the automated selection of matched controls.