Childhood and adolescent ovarian malignant tumors in Israel A nationwide study

Acta Obstet Gynecol Scand 1999; 78: 813 817 Copyright C Acta Obstet Gynecol Scand 1999 Printed in Denmark All rights reserved Acta Obstetricia et Gynecologica Scandinavica ISSN 0001-6349 ORIGINAL ARTICLE Childhood and adolescent ovarian malignant tumors in Israel A nationwide study JOSEPH MENCZER 1, SIEGAL SADETZKI 2, HAVI MURAD 2, GIULIA BARDA 1, HELENA ANDREEV 3 AND MICHA BARCHANA 3 From the 1 Gynecologic Oncology Unit, Department of Obstetrics and Gynecology, Edith Wolfson Medical Center, Holon, the 2 Department of Clinical Epidemiology, Sheba Medical Center, Tel-Hashomer, Sackler Faculty of Medicine, Tel Aviv University, Tel-Aviv and the 3 Israel Cancer Registry, Ministry of Health, Jerusalem, Israel Acta Obstet Gynecol Scand 1999; 78: 813 817. C Acta Obstet Gynecol Scand 1999 Objectives. To determine the incidence of ovarian malignant tumors in childhood and adolescence, to ascertain the frequency distribution of the various tumor types and to assess time trends in Israel on a nationwide basis. Methods. The study group included all Israeli Jewish patients Æ19 years old with histologically confirmed ovarian malignancies, diagnosed in Israel from 1970 to 1994. Data were obtained from the Israel Cancer Registry. The effects of age at diagnosis and period of diagnosis were analyzed using the Poisson regression. Results. Among the 82 patients identified, the most frequent tumors (72.0%) were of germ cell origin and among those about one third were dysgerminomas. Epithelial tumors were diagnosed in 26.6% of the patients and most of these were borderline malignancies. The incidence rate (IR) for the total group of ovarian malignancies in the 0 19 age group was 0.52 and for ages 5 19 it was 0.71 per 100,000. After adjustment for age, a significant linear trend for a decrease of germ cell tumors over time was found, stemming from a decrease of dysgerminomas. A significant trend for increase in the IRs with age was also found. In addition, a steep rise in the age specific IRs of epithelial borderline tumors was noted in the last 5 year period. Conclusions. The IRs of ovarian malignancies in childhood and adolescence in Israel, as in other countries, is very low as compared to adults and the most common tumors are of germ cell origin while malignant epithelial tumors are very rare. A time period effect in the germ cell tumors that resulted from an inexplicable significant decrease in the age specific IRs of dysgerminomas, was observed. A significant increase in borderline tumors was also noted and may be attributed to greater awareness of pathologists to this entity. Key words: adolescence; borderline malignancy; childhood; dysgerminoma; ovarian tumors Submitted 23 June, 1998 Accepted 16 April, 1999 Ovarian carcinoma is the most frequent gynecological malignancy among Israeli women and its incidence in 1994 was 13.8/ 100.000 (1). As in other populations, epithelial neoplasms are the most frequent histological type, while non-epithelial tumors are rare and constitute only about 12% of the total (2). In contrast, in childhood and adoles- Abbreviations: IR: incidence rate; CI: confidence interval. cence the majority of malignant ovarian tumors are non-epithelial and most of them are of germ cell origin. Important advancements with regard to diagnosis, and in the field of pathology and treatment, have been achieved during the last two decades. Many of these tumors produce markers that allow, nowadays, preoperative diagnosis and better follow up monitoring. A grading system of teratomas allows identification of low and high risk groups. With modern cisplatin based combi-

814 J. Menczer et al. nation chemotherapy, cure can be achieved and reproductive function preserved in the majority of young patients with non epithelial malignant ovarian tumors (3). It is well known that ovarian malignancies in childhood and adolescence are uncommon but information concerning the true incidence in this group in various populations is scarce. The following are the results of a nationwide population based study of ovarian malignancies in childhood and adolescent patients in Israel. Its purpose was to determine the incidence of ovarian malignant tumors in childhood and adolescence, to ascertain the frequency distribution of the various tumor types in this age group and to assess these trends over time. Material and methods The study group included all Jewish patients Æ19 years old, with histologically confirmed ovarian malignancies diagnosed in Israel during the 25 year period from 1970 to 1994. Data were obtained from the Israel Cancer Registry that by law is notified of all new cancer patients. The pathologic categorization of histologic subtypes is based on the World Health Organization classification (4) according to the original pathology report and additional data on hospital discharge summaries, when available. For the purpose of analysis, the total study period was divided into five-year periods. The incidence rates (IRs) of the tumors were calculated for Table I. Distribution of patients according to detailed hystologic type No. % Hystologic type Germ cell tumors 59 72.0 Dysgerminoma 19 23.2 Immature 17 20.7 Teratoma ª ª Mixed germ cell 15 18.3 Endodermal sinus 7 8.6 Undefined 1 1.2 Sex-cord stromal 2 2.4 Granulosa cell 1 1.2 Sertoli Leydig cell 1 1.2 Epithelial 21 25.6 Serous 13 15.9 Borderline 9 11.0 Invasive 4 4.9 Mucinous 6 7.3 Borderline 4 4.9 Invasive 2 2.4 Small cell 2 2.4 Total 82 100.0 the five designated periods and three age group categories. Age-specific IRs as well as total IRs for all ages combined were computed using the number of new patients in the appropriate combinations of age group and period as the nominator and the respective populations as denominator. These denominators were taken from the Statistical Abstract of Israel of the appropriate years. The effects of age at diagnosis and periods of diagnosis were analyzed using the Poisson regression with the age group as a categorical main effect and the period as a continuous one. The final analysis does not include an interaction between the age group and the period since it was not found to be significant. The analysis was done for the total group of germ cell tumors as well as for the two subcategories i.e. dysgerminoma and other germ cell tumors, separately. The epithelial tumors Poisson regression analysis was done only for the borderline subgroup because the invasive neoplasms comprised only eight cases. The significance of the effects and their approximate 95% confidence intervals (CI) in subgroups analyses were evaluated using Bonefferoni s correction for multiple comparisons. Differences were considered significant when p was 0.05. In the 0 4 year age group a single two year old patient was diagnosed. Since ovarian tumors are extremely rare in this age group, IRs were calculated for ages 0 19 as well as for ages 5 19. This case was subsequently excluded from further analysis. During the entire 25 years study period, only nine non Jewish (Arab and Druse) patients with ovarian malignancies were diagnosed. Because of the small number and different epidemiological pattern in the non Jewish population, they were not included in this report. Results During the study period, 82 Jewish patients Æ19 years old with ovarian malignancies were recorded in the Israel Cancer Registry. These comprised 1.3% of the total number of ovarian malignancies diagnosed during that period. The IR for the total group of ovarian malignancies in the 0 19 years age group was 0.66 and for ages 5 19 it was 0.71 per 100,000. The distribution of the patients according to the histologic type and subtype is presented in Table I. The most frequent tumors were of germ cell origin (72.0%) and about one third of these (19 of 59) were dysgerminomas. Among the 15 patients with mixed germ cell tumors, five also contained dysgerminomatous elements. Epithelial tumors were di-

Childhood and adolescent ovarian malignancies 815 agnosed in 25.6% of the patients and most of these (13 of 21 61.9%) were borderline malignancies. The median age of the patients with germ cell tumors was 14 years. The youngest was 2 years old and had an immature teratoma. The median age of patients with epithelial tumors was 17 years and the youngest was 9 years old and had invasive serous cystadenocarcinoma. In addition, only two cases had sex cord tumors, both were in the 15 19 year age group and were diagnosed in the 1990 1994 period. Age specific IRs by time period for germ cell tumors are shown in Table II. The total IR of germ cell tumors for the entire study period is 0.47 for ages 0 19 and 0.5 for ages 5 19 per 100,000. These IRs show a marked decrease over time from 0.69 to 0.32 (from 1970 1974 to 1990 1994). The decrease seems to have occurred since 1980. This decrease is prominent in dysgerminomas, while the IRs of the other germ cell tumors seem to remain stable over time. Among the germ cell tumors the relative frequency of dysgerminoma decreased from 7/13 (53.8%) during the first five year period to only 1/ 9 (11.1%) during the last five year period. However, this difference did not reach statistical significance (pω0.074). In the youngest age groups, nondysgerminomatous neoplasms predominated. The age specific IRs of epithelial tumors are shown in Table III. The IR for the total group of epithelial tumors was 0.18 and for invasive tumors only 0.07. No trend was observed in the IRs of the total epithelial group over time. However a steep rise is noticed in the last five year period, reaching to 0.36 per 100,000, that probably reflects the IR increase of borderline tumors. The lowest age specific IR was observed in the youngest age group (0.27 and 0.02 for germ cell tumors and epithelial tumors respectively). Table IV shows that after adjustment for age, there is a significant (p 0.05) linear trend of decrease in the IRs of germ cell tumors over time, with a rate ratio of 0.83. This trend stems especially from the dysgerminomas where the rate ratio is 0.62, i.e. every five years the IRs decrease by an estimated 38% and are significantly different from 1 even after correction for multiple comparisons (p 0.025). This period effect was not found in the other germ cell group, where the IRs seem to be stable over the periods after adjustment for agegroup. Table IV also shows a significant trend of increase in the age specific IRs of germ cell tumors with age. The age specific IRs in the 15 19 age group are 2.7 times higher than in the 5 9 age group. This trend of age effect is seen in both germ cell tumor subtypes but it reaches significance only in the dysgerminomas (rate ratioω6.68; 95% CIΩ 1.21 37.4). The Poisson regression analysis of the borderline subtype was conducted only for the 15 19 age Table II. Age specific incidence rates of total germ cell tumors and main subtypes per 100,000 in the female Jewish population in Israel (aged 5 19 years) Period 70 74 75 79 80 84 85 89 90 94 Total No. IR No. IR No. IR No. IR No. IR No. IR *Total germ-cell Tumors 5 9 1 0.16 1 0.13 5 0.57 3 0.36 1 0.10 11 0.27 10 14 3 0.49 7 1.08 1 0.13 7 0.83 3 0.33 21 0.55 15 19 9 1.40 6 0.94 4 0.61 2 0.27 5 0.55 26 0.72 Total 13 0.69 14 0.69 10 0.43 12 0.49 9 0.32 58 0.50 Dysgerminoma 5 9 0 0.00 0 0.00 1 0.11 1 0.12 0 0.00 2 0.05 10 14 1 0.16 2 0.31 0 0.00 2 0.24 0 0.00 5 0.13 15 19 6 0.93 3 0.47 2 0.30 0 0.00 1 0.11 12 0.33 Total 7 0.37 5 0.24 3 0.13 3 0.12 1 0.04 19 0.17 *Other germ Cell Tumors 5 9 1 0.16 1 0.13 4 0.45 2 0.24 1 0.10 9 0.22 10 14 2 0.32 5 0.77 1 0.13 5 0.59 3 0.33 16 0.42 15 19 3 0.47 3 0.47 2 0.30 2 0.27 4 0.44 14 0.39 Total 6 0.32 9 0.44 7 0.30 9 0.37 8 0.28 39 0.34 * One two year old girl diagnosed in the period 1990 94, is excluded from the Table.

816 J. Menczer et al. Table III. Age specific incidence rates of total epithelial tumors and main subtypes per 100,000 in the female Jewish population in Israel (aged 5 19 years) Period 70 74 75 79 80 84 85 89 90 94 Total No. IR No. IR No. IR No. IR No. IR No. IR Total epithelial 5 9 0 0.00 1 0.13 0 0.00 0 0.00 0 0.00 1 0.02 10 14 0 0.00 0 0.00 1 0.13 0 0.00 1 0.11 2 0.05 15 19 2 0.31 3 0.47 2 0.30 2 0.27 9 0.99 18 0.50 Total 2 0.11 4 0.20 3 0.13 2 0.08 10 0.36 21 0.18 Borderline 5 14 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00 15 19 0 0.00 3 0.47 1 0.15 0 0.00 9 0.99 13 0.36 Total 0 0.00 3 0.15 1 0.04 0 0.00 9 0.32 13 0.11 Invasive 5 9 0 0.00 1 0.13 0 0.00 0 0.00 0 0.00 1 0.02 10 14 0 0.00 0 0.00 1 0.13 0 0.00 1 0.11 2 0.05 15 19 2 0.31 0 0.00 1 0.15 2 0.27 0 0.00 5 0.14 Total 2 0.11 1 0.05 2 0.09 2 0.08 1 0.04 8 0.07 group since all the cases belonged to this group. This analysis shows an increase over time with an estimated rate ratio of 1.76 (95% CIΩ[1.08 2.87]) for the period effect. Discussion As in other countries (5 7) the majority of ovarian malignancies in childhood and adolescence in Israel, in a previous (8) and the present report, were of germ cell origin. The total IRs of ovarian malignancies in general as well as of germ cell tumors in this age group were very low (0.71 and 0.5 per 100,000 respectively). Our data seem to indicate an age specific IR change over time in the germ cell tumors, that resulted from the significant decrease in the IR of dysgerminomas. The etiology of dysgerminoma is unknown and, therefore, the reason for the significant incidence decline of this tumor is obscure. In previous studies the estimated overall notifi- Table IV. Estimates of rate ratio and 95% CI of germ cell incidence rates by age group and period (Poisson regression) All germ cell Dysgerminoma Other germ cell Variable RR (95% CI) RR (95% CI) RR (95% CI) Age 5 9 1.00 1.00 1.00 10 14 2.07 (1.00 4.29)* 2.71 (0.41 17.65) 1.92 (0.75 4.89) 15 19 2.68 (1.32 5.42)* 6.68 (1.21 37.04)** 1.77 (0.68 4.62) Period 0.83 (0.70 1.00)* 0.62 (0.42 0.92)** 0.95 (0.74 1.23) *p 0.05. **p 0.025. cation failure to the Israel Cancer Registry was found to be only about 5% in all age groups (9, 10). It is unreasonable to assume that the age specific IR decline of dysgerminoma is due to less accurate notification of the Registry with time, of this tumor alone. A possible explanation may be a change in the population such as the mass immigration from the former Soviet Union during the last 5 year study period. The young new immigrants may have a lower incidence of such tumors. Obviously, a true age specific incidence decline cannot be ruled out. Malignant epithelial ovarian tumors account only for a minority of ovarian tumors in the young. In a compilation of reports published from 1965 to 1991, Deprest et al. (11) found only 97 patients under the age of 20 with epithelial malignancies and the youngest reported cases were 4 years old (12, 13). Our data are in line with these findings. In the compilation by Deprest et al. (11), only 30% were of borderline malignancy, while in the present study most epithelial tumors (61.9%) were of borderline malignancy. This discrepancy may be due to differences between populations or because of variability in histological interpretation. It is noteworthy that the proportion of borderline epithelial ovarian tumors in adults is only about 10% (14). Our data show a significant trend of age specific IRs increase with time of borderline tumors. These results may be attributed to greater awareness of pathologists to this histological entity. However, due to the small number of cases and the steep

Childhood and adolescent ovarian malignancies 817 increase that occurred in the latest period, these results should be regarded with caution. Finally, even though in our study only eight invasive epithelial ovarian malignancies (four serous and two each mucinous and small cell) were observed during the entire 25 year study period, one should bear in mind this rare possibility during the management of ovarian cysts even in children and adolescents. The etiology of most ovarian epithelial neoplasms is attributed to incessant ovulation and environmental factors. However, it is doubtful that these factors also apply in the very young. Recently mutations in the BRCA1/2 genes have been implicated in the etiology of young onset ovarian cancer, especially in women with a history of familial ovarian cancer (15). Regretfully the family history and the gene status of our very young patients with epithelial ovarian tumors are unknown. References 1. Barchana M, Andreev H, Alon R. Israel Cancer Registry. Cancer in Israel, 1994. Jerusalem, 1997. 2. Mor-Yosef S, Abraham R, Schenker JG. Ovarian cancer in Israel 1960 1989. Harefuah 1997; 132: 153 5. 3. Serov SF, Scully RE, Sobin LJ. Histologic typing of ovarian tumors. In: World Health Organization. International histological classification of tumors. Geneva: The World Health Organization, 1973. 4. Gershenson DM. Update on malignant ovarian germ cell tumors. Cancer 1993; 71: 1581 90. 5. Norris HJ, Jensen RD. Relative frequency of ovarian neoplasms in children and adolescents. Cancer 1972; 30: 713 19. 6. La Vecchia C, Morris HB, Draper GJ. Malignant ovarian tumors in childhood in Britain. 1962 78. Br J Cancer 1983; 48: 363 74. 7. Vani R, Kuntal R, Rao PL. Ovarian tumors in the second decade of life. Asia Oceania J Obstet Gynecol 1991; 17: 231 5. 8. Anteby SO, Mor-Yosef S, Schenker JG. Ovarian cancer in the young. Eur J Gynaecol Oncol 1985; 6: 41 4. 9. Tulchinsky D, Modan B. Epidemiological aspects of cancer of the stomach in Israel. Cancer 1967; 20: 1311 17. 10. Menczer J, Modan B, Oelsner G, Sharon Z, Sampson S. Adenocarcinoma of the uterine cervix. A distinct epidemiologic entity. Cancer 1978; 41: 2464 7. 11. Deprest J, Moerman P, Corneillie P, Ide P. Ovarian borderline mucinous tumor in a premenarchal girl: A review on ovarian epithelial cancer in young girls. Gynecol Oncol 1992; 45: 219 24. 12. Hong SJ, Lurain JR, Tsukada Y, Piver MS, Humbert JR, Freeman AI. Cystadenocarcinoma of the ovary in a 4-year old: Benign transformation during therapy. Cancer 1980; 45: 2227 30. 13. Blom GP, Torkildsen EM. Ovarian cystadenocarcinoma in a 4-year-old girl: Report of a case and review of the literature. Gynecol Oncol 1982; 13: 242 6. 14. Younessi M, Crickard K, Celik C, Yoonessi S. Borderline epithelial tumors of the ovary: Ovarian intraepithelial neoplasia. Obstet Gynecol Surv 1988; 43: 435 44. 15. Boyd J, Rubin SC. Hereditary ovarian cancer: Molecular genetics and clinical applications. Gynecol Oncol 1997; 64: 196 206. Address for correspondence: Prof. Joseph Menczer, M.D. Director Gynecologic Oncology Unit Department of Obstetrics and Gynecology Edith Wolfson Medical Center Holon 58100, Israel