The Epidemiology of Endometrial Cancer
William E. Lucas
Table Of Contents
William E. Lucas, MD
INCIDENCE DATA AND GEOGRAPHIC CONSIDERATIONS
PERSONAL (NONDEMOGRAPHIC) RISK FACTORS
THE ENDOCRINE HYPOTHESIS
THE “UNOPPOSED” ESTROGEN HYPOTHESIS
OTHER RISK FACTORS
CHANGING PATTERNS OF OCCURRENCE OF ENDOMETRIAL CANCER
IS ENDOMETRIAL CANCER MORE THAN ONE DISEASE?
EARLY DIAGNOSIS AND PREVENTION BASED ON EPIDEMIOLOGIC FINDINGS
Many case-control studies since 1975 have shown an increased risk of endometrial carcinoma among patients receiving estrogen replacement therapy for menopausal symptoms (Table 1). Although the risk ratio among estrogen users in these reports varies from 2 to 11, the most common estimate of excess risk is between 3 and 8. Looked at another way, the cumulative risk starting at age 50 of developing endometrial carcinoma has been calculated to be 1.0% after 5 years and 3.6% at 10 years for estrogen users, compared with 0.3% at 5 years and 0.8% at 10 years for nonusers.
The case-control studies cited above have tended to obscure the fact that 60% to 85% of women who developed endometrial cancer during the periods covered were not exposed to exogenous estrogens. Although there is little doubt that an increased incidence of endometrial cancer in North America has paralleled the use of estrogens by postmenopausal women, it is important not to lose sight of the fact that this cancer afflicted many women long before estrogen therapy became available and popular. In addition, significant differences in the biologic and clinical behavior of endometrial carcinoma among estrogen users and nonusers have been described. Cancers associated with estrogen use include a preponderance of well-differentiated, localized tumors, with a low incidence of myometrial invasion and metastases, and a 5-year survival rate of well over 90%.10,11 Obesity is not a risk factor among, estrogen users but is a significant risk factor among nonusers. Finally, mortality due to endometrial cancer did not rise during a period in which incidence was rising.
It is appropriate, therefore, to consider risk factors other than exogenous estrogen therapy that may play a causal role in the development of endometrial carcinoma. The clinical dogma that the typical patient with endometrial carcinoma is an obese, postmenopausal woman with a history of relative infertility, who is hypertensive and has impaired carbohydrate tolerance, has long suggested that an underlying endocrinopathy exists among susceptible women. The epidemiologic evidence for this thesis deserves critical analysis. In addition, the demography of endometrial cancer merits attention as a potential source of valuable information in pinpointing risk factors.
|INCIDENCE DATA AND GEOGRAPHIC CONSIDERATIONS|
A significant increase in the incidence of endometrial cancer occurred in North America between 1960 and 1975.32 American Cancer Society data indicate that in excess of 30,000 new cases of endometrial cancer are being diagnosed each year in the United States, compared with approximately 20,000 cases of invasive cervical cancer, a reversal of the ratio of corpus to cervix cancer during the past 2 decades. This increased incidence of endometrial cancer has been widely interpreted to be a result of the marked increase in exogenous estrogen use. During the decade between 1960 and 1970, the sale of oral estrogen preparations tripled in the United States, reaching an estimated peak of 17 million prescriptions a year. It has been estimated that between 1971 and 1975 an excess of 15,000 cases of endometrial cancer occurred due to estrogen use.12 The ensuing 5 years witnessed a decrease in estrogen use and possibly a decrease in the incidence of endometrial cancer.4
When incidence data from other parts of the word for endometrial cancer are studied, a number of interesting observations can be made.13 In general, the incidence of endometrial cancer is much lower in South America, Asia, the Orient, and Africa than in North America and Europe, a reversal of incidence data for cervical cancer in the same areas. In general, endometrial cancer is associated with low parity and cervical cancer with high parity. Cancer of the breast and cancer of the ovary have incidence patterns which parallel those for endometrial cancer.
These data do not answer the question of whether racial-genetic factors or environmental factors are the primary explanation for geographic incidence differences. In this regard, data regarding nisei daughters and granddaughters of Japanese immigrants to the United States are of interest. Breast, endometrial, and ovarian cancer all have a much lower occurrence rate in Japan than among white North Americans. However, there is a rising incidence of breast cancer among American-born Japanese women, which among second generation Japanese immigrants is approaching the white North American incidence. Although the same increase in endometrial cancer among nisei women is not yet evident, this may be because endometrial cancer largely occurs in an older population than does breast cancer, so sufficient time may not have elapsed to make a significant rise in incidence apparent. As noted, there is a clear association between the incidence of breast and endometrial cancer, and the data cited above strongly suggest that environmental factors underlie the increase in breast cancer among nisei women. The most attractive hypothesis at this time is that dietary factors that increase body fat and alter the metabolism of estrogen play a significant role in explaining the rising incidence of endocrine-sensitive tumors among these women (see below).
|PERSONAL (NONDEMOGRAPHIC) RISK FACTORS|
Personal risk factors for endometrial cancer can be defined as those characteristics that distinguish one individual from others of the same demographic group living in the same area at the same time. 14 To identity these factors requires comparison of similar data from affected and unaffected individuals. A summary of the various risk factors to be discussed is contained in Table 2.
Many reports over the past 5 decades have described a high proportion of overweight among patients with endometrial carcinoma. Unfortunately, in relatively few studies has an entirely adequate comparison been made.
In a comparison of the weights of endometrial cancer patients and controls aged 25 to 29, Wynder and associates found that 21% of the cancer patients were 21 to 50 lb overweight and that 9% were more than 50 lb overweight, compared with 8% and 1%, respectively, among the controls.15 These data suggest that women 21 to 50 lb overweight have three times the risk, and those over 50 lb overweight, have almost ten times the risk, of developing endometrial cancer as women of normal or below normal weight.
In a Boston study controlled for age and economic status, it was found that women in the upper third of the weight distribution had 1.8 times the risk and those in the upper 15% of the weight distribution had 2.4 times the risk for developing endometrial cancer compared with women in the lower two-thirds of the weight distribution.26
When considering weight as a risk factor, it is important to remember that no more than half of the patients with endometrial cancer in the Boston study were in the heaviest one-third of the population.16 In other words, screening programs should not ignore the fact that many women who develop endometrial carcinoma are not overweight.
A positive association between nulliparity and endometrial cancer is a long-recognized characteristic of the disease. In the Boston study cited above, the incidence rate for nulliparas was twice as high as for women with one child and more than three times as high as for women with five or more children.
Whether age at first pregnancy, clearly a risk factor for breast cancer, is a risk factor for endometrial cancer is a question that cannot be answered from data available to date.
A late menopause has been described as a risk factor by several investigators.15,17 Elwood and Cole found a 2.4-fold increased risk of endometrial cancer among women whose menopause occurred at age 52 or later, compared with women whose menopause occurred at or prior to age 49. Since this risk holds up among women in the 60 to 69 and over 70 age-groups it is much more likely that the late menopause can be ascribed to continued ovarian function rather than to early neoplastic or preneoplastic endometrial changes.
An increased incidence of impaired carbohydrate tolerance among patients with endometrial cancer has been described for many years, but most studies suffer from having inadequate comparison groups or none at all.18 Because of the positive association between obesity and diabetes mellitus, the increased incidence of diabetes mellitus among endometrial cancer patients might be dependent on obesity rather than on diabetes as an independent variable. However, after controlling for age, body weight, and socioeconomic status, Elwood and Cole found a 2.8 risk ratio associated with diabetes. In a detailed prospective study of endocrine and metabolic aspects of endometrial cancer, in which age and weight were carefully controlled, Lucas and Yen could find no evidence of an increased incidence of impaired carbohydrate tolerance among endometrial cancer patients.19 It would appear that the final answer regarding the role of diabetes as a risk factor will require similar studies in a larger group of patients and suitably matched controls.
Again, unless age and weight are controlled as variables, no meaningful appraisal of hypertension as a risk factor can be made. In Wynder's study, mean blood pressure did not differ between cases and controls. In the Boston study of Elwood and Cole, the excess risk associated with a history of hypertension was 1.5, after controlling for age and weight, but this was not a significant difference from unity.
Polycystic Ovary Syndrome
Although endometrial cancer is relatively uncommon among premenopausal patients, it has been observed for many years that among patients under 45 who develop endometrial cancer there is a high incidence of ovulatory failure.20 Jackson and Dockerty found 16 endometrial carcinomas among 43 patients with the polycystic ovary syndrome.21 In a retrospective study of endometrial hyperplasia in 97 women under age 35, Chamlian and Taylor found that 73% were nulligravidas and 25% had a diagnosis of polycystic ovary syndrome.22 In 14 (14%) the hyperplasia had progressed to carcinoma in a follow-up period of 1 to 14 years.
The obvious association to be made is that unopposed estrogen secretion in this group of patients induces endometrial hyperplasia and substantially increases the risk of endometrial cancer.
Cancers of Other Sites
The fact that cancers of the endometrium and breast, and to a lesser extent of the endometrium and ovary, occur in the same women with greater frequency than can be explained by chance has been alluded to already.14,23 An association between endometrial and colon cancers has also been reported.24
Although there is some evidence for a familial aggregation of endometrial cancer, the association is not as strong as for breast cancer and is probably not sufficient to consider this a major high-risk factor.25,26
There is a significant risk of endometrial cancer occurring many years after pelvic irradiation for both benign and malignant conditions. In a study of 5749 women who had received pelvic irradiation, Wagoner and Connelly found 16 endometrial cancers compared with an expected occurrence of 1.8 on the basis of age and time-specific rates in the general population.27 A positive correlation between radiation dose and subsequent cancer was noted.
It should be noted that a high proportion of corpus neoplasms after radiation exposure are sarcomas and mixed carcinosarcomas.
Granulosa-Theca Cell Tumors
The relationship between feminizing mesenchymal ovarian tumors, endometrial hyperplasia, and endometrial cancer has generated interest and controversy among gynecologists and pathologists since the first report of such an association by Schroeder in 1922.28 Diddle collected reports on 1189 granulosa-theca cell tumors and found a 35% incidence of endometrial hyperplasia and a 6% incidence of endometrial cancer,29 and Larson found a 10.3% incidence of endometrial cancer in a collected series of postmenopausal patients with granulosa-theca cell tumors.30
Reports of endometrial cancer in several women under age 40 with granulosa cell tumors are noteworthy,31 as is the observation that a preponderance of the tumors associated with endometrial cancer are thecomas or mixed granulosa-theca cell tumors rather than pure granulosa cell tumors. This is cited as morphologic evidence of chronic estrogen stimulation of the endometrium, the theca cell being the probable predominant site of estrogen synthesis.
|THE ENDOCRINE HYPOTHESIS|
Probably the earliest evidence for an endocrine role in the genesis of endometrial adenocarcinoma was the demonstration that unopposed estrogens, either endogenous or exogenous, can produce endometrial hyperplasia.20 The observation that there is a strong association between endometrial hyperplasia and endometrial cancer was first made by Cullen in 190032 and has been reconfirmed by numerous other investigators since then.20 Gusberg first introduced the term adenomatous hyperplasia in 1947, and in subsequent studies involving serial biopsies in patients with hyperplasia, he found that 10% to 20% of these patients, followed for up to 10 years or more, developed endometrial adenocarcinoma.33,34 As already noted, many of these patients had histories suggesting anovulation or oligo-ovulation, supporting the concept of unopposed estrogen stimulation of the endometrium as a common denominator.
This concept must be reconciled with the fact that the average age at diagnosis of endometrial cancer is 59 to 60, or a decade after normal cessation of ovarian estrogen production. The fact is that estrogen is produced in appreciable amounts by postmenopausal women, largely as a result of the conversion of adrenal androstenedione to estrone in peripheral tissues.35 Of particular interest are the observations that obese women not only convert a higher percentage of androstenedione to estrone than nonobese women but also have lower serum sex hormone-binding globulin capacity and therefore higher serum levels of free estrone and estradiol.36
These findings provide evidence that endogenous estrogens may be a significant epidemiologic link between obesity and endometrial cancer in postmenopausal women. The major endocrine interrelationships are depicted schematically in Figure 1.
To assume, on the basis of available evidence, that estrogens cause endometrial cancer would be simplistic and would ignore the fact that some endometrial cancers occur in nonobese, nondiabetic, normally fertile women. Estrogen, both endogenous and exogenous, undoubtedly plays a role in promoting the development of many endometrial cancers. However, the fundamental reason for this induction of cancer remains to be defined.
Endometrial cancer continues to be the most common genital malignancy afflicting women in North America and Europe, with over 30,000 new cases a year occurring in the United States alone.37 Although the annual mortality due to endometrial cancer is less than half that caused by cancer of the uterine cervix, and less than one third the mortality due to ovarian cancer, the morbidity and economic implications continue to be major concerns.
A review of advances in our understanding of the epidemiology of endometrial cancer is important for a number of reasons. Among these are the further clarification of risk factors and of actual and potential risk modifiers, as well as the ability to provide better methods for prevention and early diagnosis and new insights into possible basic causes.
|THE “UNOPPOSED” ESTROGEN HYPOTHESIS|
A wealth of epidemiologic data have confirmed that estrogen unopposed by progesterone will induce, in time, an increasingly atypical degree of endometrial hyperplasia in a substantial number of women so exposed.38,39 After 18 months of unopposed estrogen use, Whitehead found a 30% incidence of endometrial hyperplasia. Of these two thirds were cystic hyperplasias and one third were atypical hyperplasias.40 After 15 years of unopposed estrogen use, the women's risk of developing endometrial cancer was 7-fold greater than for nonestrogen users. Peterson and co-workers have calculated that the absolute risk of developing endometrial cancer is increased from 1: 1000 in unexposed women to 4/1000/yr among estrogen users.41 Of considerable interest is the finding that estrogen users who never took a combination type oral contraceptive have a 10-fold greater risk of developing endometrial cancer than unexposed women.39
Obesity, long recognized as an important risk factor, has been shown to increase exposure of the endometrium to endogenous estrogen for a number of reasons. These include up to a 5-fold increase in the rate of peripheral conversion of estrogen precursors to active estrogen, increased 16-α-hydroxylation of estrone (a biologically more active form), and increased serum levels of free estrogen due to decreased levels of serum sex being globulin in obese women.42,43 The obese postmenopausal woman is at increased risk at least in part because of increased peripheral conversion of δ-4-androstenedione of adrenal origin to estrogen, just as the obese anovulatory or oligo-ovulatory premenopausal woman is at increased risk because of higher blood levels of estrogen and because of a relative lack of the modulating effect of progesterone on a chronically estrogen stimulated endometrium.44
Both endogenous progesterone and synthetic progestogens oppose the trophic effects of estrogen on DNA synthesis, decrease mitotic activity in endometrial epithelium, and produce mechanical sloughing, which decreases the number of proliferating cells.45 Progesterone also decreases the number of estrogen receptors available and promotes the action of estradiol dehydrogenase, which increases the rate of conversion of estradiol to estrone, a much less biologically active estrogen.
Much additional epidemiologic evidence supports the fact that estrogen is a promotor of endometrial cancer. The risk of endometrial cancer is lowered for women who are not obese, just as risk increases in proportion to the degree of obesity.46 The use of a combination oral contraceptive reduces the risk of developing endometrial cancer in proportion to the duration of oral contraceptive use.46 Pike47 has developed a mathematical model that suggests that 5 years of oral contraceptive use will halve the risk of developing endometrial cancer.
Of importance in regard to the prevention of endometrial hyperplasia among users of estrogen is the effect of the addition of a progestogen to the regimen. Whereas up to 30% of unopposed estrogen users will develop hyperplasia (one third of these will be atypical) after 18 months, the addition of a progestogen for 7, 10, or 12 days each month has been found to reduce the incidence of hyperplasia to 4, 2 and 0% respectively.48,49,50 The use of estrogen and a progestogen continuously appears to be effective after several months in producing amenorrhea and endometrial atrophy.51
Also of interest is the apparent protective effect of cigarette smoking in reducing the risk of developing endometrial cancer.52,53,54,55 Among postmenopausal women who smoke 25 or more cigarettes a day, the risk is reduced as much as 50%.56 There is a decreased risk for endometrial cancer among postmenopausal women even while continuing to use estrogen. This reduced risk is associated with decreased serum and urinary estrogen levels. There is evidence that the antiestrogenic effect of smoking may, in part, result from increased metabolism of estrogen to the relatively inactive 2-hydroxylated form.57 It should be emphasized, however, that the protection conferred by smoking is far outweighed by the many serious and well-documented health risks associated with smoking.
|OTHER RISK FACTORS|
Obesity and smoking have been cited above as risk factors that appear to promote their divergent effects on risk via modulation of estrogen metabolism. The same applies to involuntary infertility, where the increased risk is often associated with disturbances of ovulation, such as the polycystic ovary syndrome, where prolonged exposure of the endometrium to unopposed estrogen is the rule.
Autonomous estrogen-secreting ovarian tumors have long been recognized as a relatively rare cause of endometrial hyperplasia and carcinoma. Diabetes mellitus and hypertension are probably not risk factors per se, independent of obesity.
A history of breast cancer is associated with a modest increase in risk (1.72) of developing endometrial cancer according to a study of 60,065 breast cancer patients in Sweden.58 In light of the increased risk of developing breast cancer and endometrial cancer as has been experienced by Japanese women who have immigrated to the United States, it would appear that environmental, rather than genetic factors, are the causal link between breast cancer and endometrial cancer. Studies of the effect of dietary changes on estrogen metabolism are of interest in this regard. For example, in volunteer human subjects the degree of hydroxylation of estrogen at C-2 was found to be significantly reduced, while on a low-fat, high-protein diet.59 By inference, this would decrease the availability of biologically active estrogen. Whether or not long-term modifications of eating habits among American women would decrease the risk of developing endometrial cancer or breast cancer remains to be explored.
|CHANGING PATTERNS OF OCCURRENCE OF ENDOMETRIAL CANCER|
Changes in the incidence and distribution of the morphologic varieties of endometrial cancer appear to be occurring, since the first reports of an association between postmenopausal estrogen use and endometrial cancer appeared in 1975. Several studies have shown a decrease in the incidence of endometrial cancer after a decline in prescribing estrogen in the late 1970s and a subsequent upsurge of progestogen use in combination with estrogen.60,61,62 Still unresolved is the time required for the risk to decrease to that of nonestrogen users after discontinuing estrogen. A decreased risk as early as 6 months to 2 years has been reported. However, a hospital-base data-collection system continues to report an increased risk up to 10 years after cessation of estrogen use.63
|IS ENDOMETRIAL CANCER MORE THAN ONE DISEASE?|
Endometrial cancer, occurring after many years of estrogen therapy, is often in an earlier stage and of lower virulence, with a correspondingly better survival rate than endometrial cancer occurring de novo. As pointed out by Kohorn,64 the incidence of estrogen-associated endometrial cancers is becoming less common because of the use of progestogens. At the same time, with increasing longevity, more postmenopausal women who have undergone normal childbearing, a normal menopause, and who are significantly obese, are being seen after the age of 60 years with a grade 2 or 3 carcinoma associated with an otherwise atrophic endometrium.
It has been proposed that the presence or absence of endometrial hyperplasia defines two groups of patients with endometrial cancer.65 In a clinicopathologic study reported by Deligdisch, patients with endometrial cancer and associated hyperplasia were younger (average age 59 years versus 65 years for patients without hyperplasia), the cancers were generally well differentiated and less invasive, and progesterone receptors were present to a significant degree. Patients without hyperplasia were older, had less well differentiated and more deeply invasive cancers, many of which were morphologic variants generally regarded as more virulent (clear-cell, papillary, and anaplastic), and the progesterone receptor protein content was minimal or absent.
The disquieting facet of these observations is that, aside from postmenopausal bleeding, there are no other symptoms or findings on which to base screening recommendations for early detection. Also, at least for premenopausal patients with endometrial cancer, there appears to be no correlation between patient delay in reporting symptoms and the stage of their disease.66
|EARLY DIAGNOSIS AND PREVENTION BASED ON EPIDEMIOLOGIC FINDINGS|
On the basis of the data reviewed above, several observations seem justified:
From the epidemiologic data reviewed here, it is reasonable to conclude that significant future progress in determining the basic cause/s of endometrial cancer will have to come from basic research in cell biology rather than from epidemiologic studies, although the latter will continue to be of great interest. Of special note is research into the role of oncogene protein expression in endometrial cancer.64 It is to be hoped that basic research of this type will soon provide improved methods to prevent, detect, and treat endometrial cancer.
11. Silverberg SG, Mullen D, Faraci JA et al: Endometrial carcinoma: Clinical pathological comparison of cases in postmenopausal women receiving and not receiving exogenous estrogens. Cancer 45: 3018, 1980
35. MacDonald PC, Edmon CD, Hemsall DL et al: Effect of obesity on conversion of plasma androstenedione to estrone in postmenopausal women with and without endometrial cancer. Am J Obstet Gynecol 130: 448, 1978
36. Nisker JA, Hammond GL, Davidson BJ et al: Serum sex hormone-binding globulin capacity and the percentage of free estradiol in postmenopausal women with and without endometrial carcinoma. Am J Obstet Gynecol 138:637
45. Key TJA, Pike MC: The Dose-effect relationship between “unopposed” oestrogens and endometrial mitotic rate: Its central role in explaining and predicting endometrial cancer risk. Br J Cancer 57 (2): 205, 1988
57. Michnovicz JJ, Herschkopf RJ, Naganuma H et al: Increased 2-hydroxylation of estradiol and possible mechanism for the anti-estrogenic effect of cigarette smoking. N Engl J Med 315: 1305, 1986
59. Anderson KE, Kappos A, Conney AH et al: The influence of dietary protein and carbohydrate on the principal oxidative biotransformation of estradiol in normal subjects. J Clin Endocrinol Metab 59: 103, 1984