Germline mutations in DNA mismatch repair genes cause hereditary nonpolyposis colorectal cancer (HNPCC), which is characterized by the early occurrence of colorectal cancer and some extracolonic cancers of which endometrial carcinoma (EC) is the most common.1 The lifetime risk of EC has varied between 32% and 60% in different estimates.2, 3, 4 MSH2 mutation carriers seem to be at higher risk for EC than MLH1 carriers, and MSH6 carriers may be at even greater risk.5, 6 The high risk for EC suggests that prophylactic surveillance may benefit patients with HNPCC.
The survival rate of HNPCC-associated EC is favorable. In a group of 125 women fulfilling the Amsterdam criteria, Vasen et al. reported that only 12% died of endometrial cancer.7 Compared with matched sporadic EC cases, Boks et al. found no difference in the survival rate among 50 EC patients from HNPCC families; the overall cumulative 5-year survival rates were 88% and 82% respectively.8 Thus, the question arises whether the surveillance for EC is at all necessary in HNPCC.
Regular surveillance for EC by annual or biannual transvaginal ultrasound (TVUS) scan or endometrial aspiration biopsy has been recommended beginning at age 30–35 years for the early detection of EC or premalignant lesion.9, 10, 11 However, the effectiveness of gynecological surveillance remains to be shown. A prospective trial with 269 women from HNPCC or HNPCC-like families attending annual or biannual EC surveillance with TVUS failed to demonstrate clear benefits. Surveillance detected no cases of EC, but 2 cases of interval cancer did occur. Ultrasound may be insufficient to detect early EC, especially among premenopausal women.12 Ultrasound surveillance of 41 mutation positive women also failed to detect EC cases but guided in finding premalignant hyperplasia in few cases.13 So far, no published data exist on EC surveillance by intrauterine sampling in HNPCC. In sporadic settings, endometrial Pipelle sampling seems highly accurate in detecting EC or atypical hyperplasia.14
This present report describes the results of endometrial cancer screening in the Finnish HNPCC registry with 385 mutation-positive women from 103 HNPCC families over a 10-year period.
CAH, complex atypical hyperplasia; CH, complex hyperplasia; EC, endometrial carcinoma; FIGO, international federation of gynecology and obstetrics; HNPCC, hereditary nonpolyposis colorectal cancer syndrome; PAPA, Papanicolau test; SAH, simple atypical hyperplasia; SH, simple hyperplasia; TATI, tumor-associated trypsin inhibitor; TVUS, transvaginal ultrasound.
Material and methods
Mutation testing has been available in Finland since 1996 and all those tested as mutation carriers have been advised to visit their local hospitals for regular cancer surveillance. Some members of the families fulfilling the Amsterdam criteria began the gynecological surveillance protocol before actual mutation testing. Surveillance guidelines for endometrial cancer in HNPCC in Finland suggest ultrasound and intrauterine sampling biannually or with a 3-year interval after age 30–35.
The data consist of all 385 Finnish female mutation carriers tested from 1996 through May 2005. They represent 103 families with 32 different mutations: 333 patients had the mutation in MLH1, 32 in MSH2 and 20 in MSH6. The uterus had been removed previously, either for benign or malignant reasons, in 138 cases, so there was no need for further surveillance for endometrial cancer. For other reasons, 72 women withdrew from surveillance (Fig. 1). Thus, of the 385 women, 175 actually entered the gynecological surveillance program.
We gathered their previous gynecological information, surveillance data and pathology reports from original patient records from 28 different hospitals and the Finnish Cancer Registry. A single gynecological pathologist reevaluated the pathology reports of EC cases and the specimens of the premalignant lesions in the surveillance group. Survival data were checked at the Population Registration Center of Finland.
Altogether, 503 surveillance visits were arranged and 759 surveillance years passed from the time of enrollment until hysterectomy, death or the end of the study period (September 2005). The median follow up time since the first surveillance visit was 3.7 years (range 0–13 years). Fifty-three women had attended only one surveillance visit by the end of this study.
The methods used in surveillance varied among hospitals and clinics. Clinical examination was performed at every visit and TVUS at 94% of the visits (476/503 visits). Intrauterine biopsies were taken 382 times (74%) according to the method available at that hospital (e.g., Pipelle, Vabra and Endobrush). s-CA12-5 values were obtained at 144 surveillance visits (28%) and s-TATI values at 69 visits (14%).
The EC cases detected in the surveillance program were compared with those detected by symptoms. Altogether 83 symptomatic mutation positive EC patients from the same families had been diagnosed and treated between the years 1963–2004; their median age was 50 years (27–85 years). The treatment of EC patients diagnosed in surveillance took place between 1996 and 2004 and their median age was 52 years (range 36–71 years). The difference between the FIGO-stage distribution of EC-cases in the 2 groups was tested using the chi-squared test. EC related survival curves were constructed using Kaplan–Meier analysis. The significance of differences between the 2 groups was tested using the log-rank test. We analyzed the different methods used in surveillance.
Among the 175 women attending surveillance there were 14 cases (8%) of EC (adenocarcinoma in all). In 11 of these cases the diagnosis of EC was based on the screening examination, 2 others were interval cancers based on symptoms manifesting 3 and 31 months after the surveillance visit. One woman underwent a prophylactic hysterectomy (with no prior endometrial sampling) in addition to a colectomy for cancer and an occult EC was found. Of the 11 EC cases found by true screening, 9 were detected or suspected by endometrial samples, 8 as EC and 1 as complex hyperplasia (CH). Altogether 4 had a suspicious finding in TVUS, and 2 had a suspicious PAPA smear. An increased tumour marker value (CA 12-5) was observed in only 1 case (Table I).
Table I. Endometrial Carcinoma Cases in Surveillance Group
Surveilled time (months)
Age at operation (years)
Screening method at time of diagnosis
Endometrium in TVUS
s-CA 12-5 (kU/l)
NA, data not available; E, endometrioid; EC, endometrial carcinoma; C, clear cell; CH, complex hyperplasia; Sq, with squamous differentiation; susp npl, suspicion of neoplasia.
The distribution of tumour stage tended to be more favourable in the group attending surveillance than in the 83 mutation-positive EC patients who had not attended surveillance, but the difference was insignificant (Pearson chi-square 0.67; Table II). None of the EC-patients died of cancer in the mean followup time of 5 years (1–9 years) compared with 6 deaths due to EC in the nonsurveilled group within the median follow up of 13.7 years (0–42 years). The difference in the survival curves (100 vs. 92% at 10 years), however, was not significant (p = 0.4, log rank; Fig. 2).
Table II. FIGO Stage Distribution of EC Cases Diagnosed by Surveillance or with Symptoms
EC cases among surveilled
Symptomatic EC cases
Altogether 6 patients in the surveilled group exhibited elevated s-CA12-5 values, half of whom were transient; 1 had EC. Six patients had elevated s-TATI values, 1 of whom was discovered to have ovarian cancer on the basis of pain symptoms 5 months later; others showed no sign of neoplasia.
In addition to EC, surveillance of 175 women with IU-samples detected endometrial hyperplasia in 14 (8.0%) women (Table III), 11 of whom underwent a prophylactic hysterectomy and the remaining three, showed normal findings at 6, 16 and 45 months. The total frequency of suspicious IU-samples in 503 surveillance visits was 25 (5.0%), and with EC alone, 11 (2.2%). One additional case of occult CH was discovered in a prophylactic hysterectomy 6 months after a normal surveillance visit, with TVUS only.
Table III. Premalignant Findings in Surveillance Group
Of the 175 women surveilled, ovarian cancer of endometrioid type was detected in 4 (2.3%), none of whom were diagnosed at surveillance, 2 were diagnosed by symptoms 2 and 5 months after normal surveillance visits at the ages of 41 and 45 years (stages III and I). The other had exhibited elevated s-TATI values repeatedly for several years, but all other examinations had given normal results. Two other cases were diagnosed incidentally during an operation for EC or CH at the ages of 42 and 50 years (each of stage I). All are well and alive today.
Of the 175 women in the surveillance group 59 (34%) underwent a hysterectomy in the followup period. Eight patients underwent a hysterectomy because of adenocarcinoma in the IU-sample, 2 patients because of symptomatic interval EC, 5 patients on suspicion of ovarian mass or tumour, and 1 woman because of massive bleeding. The remaining 43 patients underwent a prophylactic hysterectomy because of a premalignant lesion found in surveillance, laparotomy for some other reason or solely for mere prophylaxis.
The present surveillance program of 175 mutation-positive women at risk for EC yielded 11 asymptomatic cancer cases and 14 others with a premalignant lesion in 503 surveillance visits. The frequency of significant findings was 14.3% of persons surveilled, or 5.0% of screening visits within the median surveillance period of 3.7 years. The stage distribution of the EC tumours in the group under surveillance tended to be more favourable than that of the mutation-positive EC patients of the same families who had no surveillance. Furthermore, none of the surveilled EC patients died of EC compared with 6 in the nonsurveilled patients. No statistical significant difference, however, aroused in stage distribution or in survival, between the 2 groups. Thus, the most important advantage of the surveillance may lie in the frequent detection of premalignant lesions, which enables prophylactic hysterectomy in due time and potentially reduces EC incidence.
In accordance with 2 previous reports on surveillance for EC using TVUS, this method proved relatively insensitive in the present series as well. Adenocarcinoma was diagnosed by IU-sample in several patients with normal endometrium in TVUS. Dove–Edwin et al. found no presymptomatic EC in a group of 269 women belonging to HNPCC families; neither did Rijcken et al. in a series of 41 mutation-positive women.12, 13 This latter study, however, revealed suspicious TVUS findings in 17 of the 179 examinations, which led to endometrial sampling, and the discovery of premalignant complex atypical hyperplasia in 3 cases. In terms of patient years at risk, the combination of endometrial biopsy and TVUS in our series resulted in more than twice as many malignant or premalignant findings as in the series of Rijcken et al. This suggests that endometrial biopsies increase the efficacy of endometrial screening.
Interpretation of the clinical significance of premalignant lesions may be difficult. Endometrial hyperplasia may be simple or complex, and the occurrence of atypia represents the greatest risk for progression.15, 16, 17 There were altogether 14 cases with endometrial hyperplasia in the present IU samples: 4 had CAH, 8 had CH and 2 had SH. Eleven of 14 patients underwent a prophylactic hysterectomy and no EC was found in the removed endometrium. Hysterectomy for premalignant lesions in these 11 cases very likely prevented the development of cancer, at least for some.
Whether surveillance for EC can improve survival in women with HNPCC predisposition is difficult to prove because the prognosis can be very good even in symptomatic cases. In a study of 50 Dutch women with EC from HNPCC families the 5-year survival rate was 88%, and in this present series of 83 symptomatic nonsurveilled women, it was 92% at 10 years. Our figure, however, is based solely on mutation-positive members of the families, and thus excluded several historical cases who had died of EC before genetic testing was possible. This may have resulted in a more favorable estimate of prognosis than in reality. No deaths occurred among the surveilled persons in this present series. It must also be noted that the size of the group in surveillance is still too small and the followup period too short for final evaluation of prognostic advantage.
The use of tumor markers have no place in screening recommendations by the Finnish HNPCC registry. Markers were, however, obtained in 144 screening visits (28%), mostly s-CA 12-5 and s-TATI. Only 1 EC patient exhibited an elevated CA 12-5 level while 5 other patients exhibited such elevated levels for no clear reason. Such was also the case with s-TATI in 5 other cases. One woman repeatedly showed an elevated s-TATI value with no other pathological findings. Eventually she was diagnosed with symptomatic ovarian cancer. In conclusion, tumor marker determination for screening purposes may cause more harm and unnecessary worry than benefit, and is discouraged as a surveillance modality. Unfortunately, surveillance for ovarian cancer seems unsuccessful as TVUS detected none of the present cases. Prophylactic ovarectomy may remain the only effective option to reduce mortality because of ovarian cancer, but it may be hard to accept by persons at risk.
The proportion of prophylactic hysterectomies was surprisingly high in our series, nearly a fourth of those initially enrolled. Fourteen women of the altogether 43 cases with prophylactic hysterectomy showed a premalignant lesion in the prior surveillance samples. Prophylactic surgery certainly reduces or abolishes the risk of EC,18 but can nevertheless be considered major surgery with morbidity and, seldomly, even mortality.19 Therefore, surveillance using endometrial sampling may present a less invasive but still effective alternative to prophylactic surgery in the prevention or early detection of EC in mutation carriers of mismatch repair genes.
In conclusion, endometrial cancer surveillance in HNPCC seems more effective with the use of endometrial biopsies in addition to TVUS alone. Tumors in the surveilled patients tended to be of an earlier stage than in symptomatic cases, and no deaths from EC have yet been observed while under surveillance. A longer followup period and a larger group of surveilled patients is, however, necessary to demonstrate a definite survival benefit. There is potential gain in the possibility to detect premalignant lesions and treat cancer in their early stages without adjuvant external radiotherapy.
The authors thank Ms. Tuula Lehtinen and Ms. Kirsi Pylvänäinen for their technical assistance.