ACTA REVIEW
Endometrial and ovarian malignancies: epidemiology, etiology and prognostic factors AMIN P. H. MAKARAND CUES G. TROPE From the Department of Gynecologic Oncology, The Norwegian Radium Hospital, Oslo, Norway
Actu Obstet Gynecol Scund 1YY2; 71: 331-336
Epidemiology and etiology Endometrial and ovarian cancer share many similarities in their epidemiology and etiology (1). The geographical patterns of distribution are similar for both cancers, with highest incidence rates reported in white females in Western Europe and North America ( 2 ) . In the Nordic countries, the highest incidence rates for both cancers are observed near Copenhagen and Stockholm, and in Southern Sweden ( 2 ) . The Scandinavian countries, especially Sweden, have the highest cumulative incidence rates of ovarian cancer worldwide (2, 3 ) . The high incidence of ovarian and corpus cancer in the industrialized countries have stimulated the hypothesis that environmental agents may be important etiologic factors. This hypothesis has also been supported by migrant studies which showed that genetic differences are of less significance as etiologic factors than environmental ones (4, 5). This has been manifested in the early studies of Haenszel and Kurihara (4) who observed an increase in the incidence of ovarian and endometrial cancer among Japanese migrants to the U.S.A (who moved from a country with very low risk for both cancers to one with a very high risk). More recent data (5) based on incidence rates in Hawaii (1973-1977) showed that ovarian cancer incidence among first and second generation Japanese migrants was 7 per 100,000 compared to 3.4 per 100,000 in Japan and 9.4 per 100,000 among Hawaii whites. A similar pattern of increase in the incidence of endometrial cancer among first and second generations of Japanese migrants (15.5 and 20.3 per 100,000 respectively) was observed. The corresponding figures for endometrial cancer were 2 per 100,000 (in Japan) and 34.8 per 100,000 (in Hawaii whites).
Among environmental factors suspected to play a r61e in the pathogenesis of ovarian and endometrial malignancies, great importance is attached to fat consumption, obesity and patterns of fat distribution (6-8).A strong relationship between a high Quetelet index (defined as weight (kg) divided by height (m) squared) and risk for ovarian and endometrial cancer has been reported by many authors (6-8). Farrow et al. (6) found that a high Quetelet index significantly increased the risk for ovarian cancer, especially those of serous and endometroid histologic types. Austin et al. (8) also found a statistically significant excess risk for endometrial cancer with central obesity and this excess risk could not be explained by increased estrogen levels. The r61e of pregnancy, breast feeding and hormonal therapy on the pathogenesis of ovarian and endometrial cancers has been widely discussed in the literature (9-15). Regarding ovarian cancer, the most consistent finding among case control studies has been a relative deficit of pregnancies among cases. Several case-control studies have found a decrease in risk of the epithelial ovarian cancer associated with pregnancy, breast feeding and use of oral contraceptives ( 0 0 )(9-14). This prompted the hypothesis of incessant ovulation which holds that factors which suppress ovulation may reduce the risk of developing ovarian cancer (9-14). Gwin et al. (11) showed in a multicenter population-based, case-control study, that the estimated relative risks of epithelial ovarian cancer were 0.6 (95% confidence interval 0.5-0.8) for women who have ever been pregnant, 0.6 (95% confidence interval 0.5-0.8) for women who have ever breast fed, and 0.5 (95% confidence interval 0.5-0.7) for women who have ever used oral contraceptives. The risk of ovarian cancer seemed to deActa Obstet Gynecol Scund 71 (1992)
332
A . P. Makar and C. G. Trope'
crease with increasing duration of use of OCS and the protective effect of OCs was even noticed in ex-users (at least 15 years). Similarly, the risk of ovarian cancer was significantly decreased in women reporting life-long menstrual irregularities (13). However, the explanation of such correlation cannot be merely hormonal suppression. Risch et al. (14) used a logistic regression method to test the hypothesis that equal periods of anovulation, regardless of cause, produced the same reduction in ovarian cancer risk. The authors (14) found that the amounts of anovulatory time attributable to different exposures did not completely account for their protective effects. As anovulation in general represents a protective mechanism against ovarian cancer, we have wondered whether recent advances in the field of infertility, and the use of different hormonal stimulation regimen, could create new high risk groups. Regarding endometrial cancer, anything that increases exposure of the endometrium to unopposed estrogen increases the risk of endometrial cancer for the rest of a woman's life by increasing the frequency of mitosis and subsequent copying errors in the endometrium (10, 15). Reproductive factors must be evaluated in this framework, but their r81e in this regard has received far less attention than the influence of exogenous estrogen and obesity. Van Leeuwen and Rookus (12), in a literature review, showed that while the risk of endometrial cancer among women who ever used combination type OCs was half that of those who never used OC, there was a twofold risk of endometrial cancer in women using sequential OCs (with only 5 days of progesterone addition to each cycle) or receiving estrogen replacement therapy without progesterone. This difference in risk was explained by the hypothesis of unopposed estrogen effect, as combined OCs shorten the period of endometrial exposure to unopposed estrogen from the 14 days of a normal follicular phase to 7 days per 28 day cycle.
Common prognostic factors Ovarian and endometrial cancers share more or less common prognostic factors. These factors are:
1 ) Hormonal receptors The prognostic value of hormonal receptor status in ovarian malignancies remains questionable. Harding et al. (16) showed that the incidence of progesterone (PR) positive tumors correlated inversely with the FIGO stage. Survival of patients with advanced ovarian cancer was significantly prolonged by optiAcra Ohstet Gynecol Scand 71 (1992)
mal initial cytoreductive surgery, platinum therapy, and tumor expression of PR. Similar results were obtained by Sevelda et al. (17). However, the results of Rose et al. (18) were in disagreement. Regarding endometrial cancer, a consistent relationship exists between well differentiated histological types, receptor positivity, and prolonged survival (19, 20). Kleine et al. (19) found that 57% of endometrial carcinomas were estrogen receptor (ER) and progesterone receptor (PR) positive and 24% negative for both receptors. Receptor status correlated with clinical stage and histologic grade, but not with myometrial invasion. Five year survival rate (stage I) and median survival time (stage 11-IV, recurrences) for patients with ER+/PR+ and ER-/PR+ endometrial cancer were significantly better than for ER-IPR- and ER+/PR- patients. A multivariate analysis demonstrated progesterone receptor as a significant prognostic factor next to clinical stage. ER had no significant prognostic relevance. Carcangiu et al. (20) presented a series of 183 primary endometrial carcinomas which were immunohistochemically stained for estrogen receptor (ER) and progesterone receptor (PR) using formalin fixed paraffin embedded sections. Endometroid type adenocarcinoma had the highest degree of positivity for both receptors, followed by adenosquamous carcinoma, serous carcinoma, and clear cell carcinoma. The positivity for ER and PR of the malignant component was statistically correlated with FIGO stage, FIGO grade, and nuclear grade. There was no association between the depth of myometrial tumor invasion and either receptor status of malignant component. There was a significant association between the presence of lymph vessel invasion and the positivity for PR. In addition the authors showed that immunohistochemical analysis of sex steroid receptor status on formalin-fixed, paraffin-embedded tissues offers an excellent alternative to the standard biochemical methods. In conclusion: the prognostic significance of hormonal receptor status is well established in case of endometrial malignancies but it remains questionable in case of ovarian malignancies.
2) Titmor marker CA 125 CA 125 is an antigenic determinant on a high molecular weight glycoprotein recognized by the monoclonal antibody OC 125 which was raised against an ovarian cancer cell line (21). Immunohistochemical studies showed CA 125 to be expressed by more than 80% of non-mucinous ovarian cancers (21). CA 125 was also found on the surface of epithelial cells of Miillerian origin as that of the fallopian tube, endometrium, endocervix, peritoneum, pleura, per-
Endometrial and ovarian malignancies icardium, and bronchus (22, 23). The O C 125 antibody was used to develop an immunoradiometric assay for quantitative measurement of the CA 125 determinant, and in human serum a reference value of 35 U/mL has been accepted as an upper normal level (21). Although elevated serum CA 125 levels are found in more than 80% of cases of invasive epithelial ovarian malignancies, the use of serum CA 125 as a single screening test for ovarian cancer has been limited by rather low predictive values. Elevated serum CA 125 levels occur in a number of benign conditions (24, 25) such as endometriosis, pelvic inflammatory disease, liver cirrhosis and during pregnancy, and a normal serum CA 125 level (< = 35 U/mL) does not exclude the presence of malignancy (24). There is, however, increasing evidence to the prognostic significance of serum CA 125 measurements in patients with epithelial ovarian malignancies. CA 125 level has been found to correlate directly to tumor load, F I G 0 stage and tumor grade (24). Mobus et al. (26) showed by univariate analysis that survival was significantly better for patients with postoperative serum CA 125 level < = 65 U/ml. In a study including 678 patients with invasive epithelial ovarian malignancies managed at The Norwegian Radium Hospital, Makar et al. (personal communication) showed by multivariate analysis that the postoperative CA 125 level is of independent prognostic significance. In patients with residual disease after primary operation those with normal or slightly elevated CA 125 level (up to 65 U/ml) had a far better prognosis than those with higher levels. In patients without residual disease after primary surgery, serum CA 125 with 35 U/mL cut off level was a significant factor for survival, and the 30 months probability of survival for patients with serum CA 125 levels < = 35 U/mL was 0.900 compared to 0.66 for patients with higher values. Changes in the serum CA 125 level during chemotherapy have also been found to have prognostic significance (27-31). Hunter et al. (28) showed that CA 125 half-life of< 20 days was associated with prolonged survival. In patients who eventually were found to be disease free at surgical surveillance procedures, normalization of serum CA 125 levels to
Endometrial and ovarian malignancies: epidemiology, etiology and prognostic factors AMIN P. H. MAKARAND CUES G. TROPE From the Department of Gynecologic Oncology, The Norwegian Radium Hospital, Oslo, Norway
Actu Obstet Gynecol Scund 1YY2; 71: 331-336
Epidemiology and etiology Endometrial and ovarian cancer share many similarities in their epidemiology and etiology (1). The geographical patterns of distribution are similar for both cancers, with highest incidence rates reported in white females in Western Europe and North America ( 2 ) . In the Nordic countries, the highest incidence rates for both cancers are observed near Copenhagen and Stockholm, and in Southern Sweden ( 2 ) . The Scandinavian countries, especially Sweden, have the highest cumulative incidence rates of ovarian cancer worldwide (2, 3 ) . The high incidence of ovarian and corpus cancer in the industrialized countries have stimulated the hypothesis that environmental agents may be important etiologic factors. This hypothesis has also been supported by migrant studies which showed that genetic differences are of less significance as etiologic factors than environmental ones (4, 5). This has been manifested in the early studies of Haenszel and Kurihara (4) who observed an increase in the incidence of ovarian and endometrial cancer among Japanese migrants to the U.S.A (who moved from a country with very low risk for both cancers to one with a very high risk). More recent data (5) based on incidence rates in Hawaii (1973-1977) showed that ovarian cancer incidence among first and second generation Japanese migrants was 7 per 100,000 compared to 3.4 per 100,000 in Japan and 9.4 per 100,000 among Hawaii whites. A similar pattern of increase in the incidence of endometrial cancer among first and second generations of Japanese migrants (15.5 and 20.3 per 100,000 respectively) was observed. The corresponding figures for endometrial cancer were 2 per 100,000 (in Japan) and 34.8 per 100,000 (in Hawaii whites).
Among environmental factors suspected to play a r61e in the pathogenesis of ovarian and endometrial malignancies, great importance is attached to fat consumption, obesity and patterns of fat distribution (6-8).A strong relationship between a high Quetelet index (defined as weight (kg) divided by height (m) squared) and risk for ovarian and endometrial cancer has been reported by many authors (6-8). Farrow et al. (6) found that a high Quetelet index significantly increased the risk for ovarian cancer, especially those of serous and endometroid histologic types. Austin et al. (8) also found a statistically significant excess risk for endometrial cancer with central obesity and this excess risk could not be explained by increased estrogen levels. The r61e of pregnancy, breast feeding and hormonal therapy on the pathogenesis of ovarian and endometrial cancers has been widely discussed in the literature (9-15). Regarding ovarian cancer, the most consistent finding among case control studies has been a relative deficit of pregnancies among cases. Several case-control studies have found a decrease in risk of the epithelial ovarian cancer associated with pregnancy, breast feeding and use of oral contraceptives ( 0 0 )(9-14). This prompted the hypothesis of incessant ovulation which holds that factors which suppress ovulation may reduce the risk of developing ovarian cancer (9-14). Gwin et al. (11) showed in a multicenter population-based, case-control study, that the estimated relative risks of epithelial ovarian cancer were 0.6 (95% confidence interval 0.5-0.8) for women who have ever been pregnant, 0.6 (95% confidence interval 0.5-0.8) for women who have ever breast fed, and 0.5 (95% confidence interval 0.5-0.7) for women who have ever used oral contraceptives. The risk of ovarian cancer seemed to deActa Obstet Gynecol Scund 71 (1992)
332
A . P. Makar and C. G. Trope'
crease with increasing duration of use of OCS and the protective effect of OCs was even noticed in ex-users (at least 15 years). Similarly, the risk of ovarian cancer was significantly decreased in women reporting life-long menstrual irregularities (13). However, the explanation of such correlation cannot be merely hormonal suppression. Risch et al. (14) used a logistic regression method to test the hypothesis that equal periods of anovulation, regardless of cause, produced the same reduction in ovarian cancer risk. The authors (14) found that the amounts of anovulatory time attributable to different exposures did not completely account for their protective effects. As anovulation in general represents a protective mechanism against ovarian cancer, we have wondered whether recent advances in the field of infertility, and the use of different hormonal stimulation regimen, could create new high risk groups. Regarding endometrial cancer, anything that increases exposure of the endometrium to unopposed estrogen increases the risk of endometrial cancer for the rest of a woman's life by increasing the frequency of mitosis and subsequent copying errors in the endometrium (10, 15). Reproductive factors must be evaluated in this framework, but their r81e in this regard has received far less attention than the influence of exogenous estrogen and obesity. Van Leeuwen and Rookus (12), in a literature review, showed that while the risk of endometrial cancer among women who ever used combination type OCs was half that of those who never used OC, there was a twofold risk of endometrial cancer in women using sequential OCs (with only 5 days of progesterone addition to each cycle) or receiving estrogen replacement therapy without progesterone. This difference in risk was explained by the hypothesis of unopposed estrogen effect, as combined OCs shorten the period of endometrial exposure to unopposed estrogen from the 14 days of a normal follicular phase to 7 days per 28 day cycle.
Common prognostic factors Ovarian and endometrial cancers share more or less common prognostic factors. These factors are:
1 ) Hormonal receptors The prognostic value of hormonal receptor status in ovarian malignancies remains questionable. Harding et al. (16) showed that the incidence of progesterone (PR) positive tumors correlated inversely with the FIGO stage. Survival of patients with advanced ovarian cancer was significantly prolonged by optiAcra Ohstet Gynecol Scand 71 (1992)
mal initial cytoreductive surgery, platinum therapy, and tumor expression of PR. Similar results were obtained by Sevelda et al. (17). However, the results of Rose et al. (18) were in disagreement. Regarding endometrial cancer, a consistent relationship exists between well differentiated histological types, receptor positivity, and prolonged survival (19, 20). Kleine et al. (19) found that 57% of endometrial carcinomas were estrogen receptor (ER) and progesterone receptor (PR) positive and 24% negative for both receptors. Receptor status correlated with clinical stage and histologic grade, but not with myometrial invasion. Five year survival rate (stage I) and median survival time (stage 11-IV, recurrences) for patients with ER+/PR+ and ER-/PR+ endometrial cancer were significantly better than for ER-IPR- and ER+/PR- patients. A multivariate analysis demonstrated progesterone receptor as a significant prognostic factor next to clinical stage. ER had no significant prognostic relevance. Carcangiu et al. (20) presented a series of 183 primary endometrial carcinomas which were immunohistochemically stained for estrogen receptor (ER) and progesterone receptor (PR) using formalin fixed paraffin embedded sections. Endometroid type adenocarcinoma had the highest degree of positivity for both receptors, followed by adenosquamous carcinoma, serous carcinoma, and clear cell carcinoma. The positivity for ER and PR of the malignant component was statistically correlated with FIGO stage, FIGO grade, and nuclear grade. There was no association between the depth of myometrial tumor invasion and either receptor status of malignant component. There was a significant association between the presence of lymph vessel invasion and the positivity for PR. In addition the authors showed that immunohistochemical analysis of sex steroid receptor status on formalin-fixed, paraffin-embedded tissues offers an excellent alternative to the standard biochemical methods. In conclusion: the prognostic significance of hormonal receptor status is well established in case of endometrial malignancies but it remains questionable in case of ovarian malignancies.
2) Titmor marker CA 125 CA 125 is an antigenic determinant on a high molecular weight glycoprotein recognized by the monoclonal antibody OC 125 which was raised against an ovarian cancer cell line (21). Immunohistochemical studies showed CA 125 to be expressed by more than 80% of non-mucinous ovarian cancers (21). CA 125 was also found on the surface of epithelial cells of Miillerian origin as that of the fallopian tube, endometrium, endocervix, peritoneum, pleura, per-
Endometrial and ovarian malignancies icardium, and bronchus (22, 23). The O C 125 antibody was used to develop an immunoradiometric assay for quantitative measurement of the CA 125 determinant, and in human serum a reference value of 35 U/mL has been accepted as an upper normal level (21). Although elevated serum CA 125 levels are found in more than 80% of cases of invasive epithelial ovarian malignancies, the use of serum CA 125 as a single screening test for ovarian cancer has been limited by rather low predictive values. Elevated serum CA 125 levels occur in a number of benign conditions (24, 25) such as endometriosis, pelvic inflammatory disease, liver cirrhosis and during pregnancy, and a normal serum CA 125 level (< = 35 U/mL) does not exclude the presence of malignancy (24). There is, however, increasing evidence to the prognostic significance of serum CA 125 measurements in patients with epithelial ovarian malignancies. CA 125 level has been found to correlate directly to tumor load, F I G 0 stage and tumor grade (24). Mobus et al. (26) showed by univariate analysis that survival was significantly better for patients with postoperative serum CA 125 level < = 65 U/ml. In a study including 678 patients with invasive epithelial ovarian malignancies managed at The Norwegian Radium Hospital, Makar et al. (personal communication) showed by multivariate analysis that the postoperative CA 125 level is of independent prognostic significance. In patients with residual disease after primary operation those with normal or slightly elevated CA 125 level (up to 65 U/ml) had a far better prognosis than those with higher levels. In patients without residual disease after primary surgery, serum CA 125 with 35 U/mL cut off level was a significant factor for survival, and the 30 months probability of survival for patients with serum CA 125 levels < = 35 U/mL was 0.900 compared to 0.66 for patients with higher values. Changes in the serum CA 125 level during chemotherapy have also been found to have prognostic significance (27-31). Hunter et al. (28) showed that CA 125 half-life of< 20 days was associated with prolonged survival. In patients who eventually were found to be disease free at surgical surveillance procedures, normalization of serum CA 125 levels to
