Relationship Between Steroid Excretion Patterns and Breast Cancer Incidence in Israeli Women of Various Origins1. 2 J. Gross, M.D.,3 B. Modan, M.D.,4 B. Bertini, M.D.,5 O. Spira, Ph.D.,3 F. de Waard, M.D.,6 J.H.H. Thijssen, Ph.D.,6 and P. Vestergaard, Ph.D.7 The strata taken into consideration were ethnic origin (European, North African, Middle-Eastern, and Yemenite) and period of immigration (prior to 1953, and since 1954). Originally 75 women were selected in each ethnic category, but for technical reasons the hormone studies had to be curtailed after 213 samples had been analyzed for estrogens and 244 for the other hormones. There was no selective bias in this process, and therefore the women studied may be considered as a representative sample of the population.
ABSTRACT-A nationwide study of the steroid excretion pat· terns in postmenopausal Israeli migrant women demonstrated differences between high· and low·risk groups for breast cancer in the following variables: age at first parturition, number of pregnancies, number of live births, height, and weight. The direction of the differences was in line with those observed for breast cancer patients. The groups also differed in the excretion of estriol, 17-ketosteroids, and allotetrahydrocortisol. Multiple regression analysis revealed that the excretion of estriol was significantly lower in population groups in whom breast cancer incidence was high. Possibly this trend-which has also been observed in adolescent and premenopausal women-reflected environmental influences on peripheral estrogen metabo· lism.-J Natl Cancer Inst 59: 7-11,1977.
TABLE I.-Annual age-adjusted incidence rates of breast cancer in Jewish women in Israel, 50-69 years of age (1960-66), according to country or continent of origin a
Experimental and clinical evidence suggests that hormonal factors play a role in the genesis of breast cancer. However, wide variations in the reported incidence of breast cancer (1-6) provide arguments for the existence of an environmental influence. An interplay between environmental and hormonal factors might therefore be considered (7, 8). MacMahon et al. (9-11) and Lilienfeld et al. (12) showed that an important factor in the etiology of breast cancer is the age at which the woman first gives birth. Although it has been suggested that a critical induction period is between menarche and first pregnancy, a second critical induction period may occur around the time of menopause (11). The latter hypothesis, which implies the presence of at least two disease entities, is supported by the fact that the age-specific incidence curve of breast cancer differs in high- and low-risk populations (13, 14). Thus in Asia (e.g., Japan), Africa, and Latin America the age-specific incidence curves are different from those in the United States and other "Western" countries (1). The rise at premenopausal age is similar, though at a lower level, but after age 50 there is a leveling off to a plateau (3). "Westernization" affects incidence rates primarily in the older age groups. In Israel the incidence of breast cancer differs widely between the various ethnic groups (15); the highest rates occur among European-born and the lowest, among Yemenite women. The age-specific incidence curve of the former resembles the pattern observed in high-risk populations, whereas that of the Oriental group follows the low-risk pattern (16). The present study explores a possible correlation between the marked variation in the incidence of breast cancer among the major Israeli ethnic groups (table 1) and selected personal and hormonal parameters in postmenopausal women.
a
b
Yemen Africa Asia b Europe
27.2 52.5 48.1 124.9
Direct method; total Israeli female population used as standard. Excluding Yemen and Israel.
ABBREVIATIONS USED: El = estrone; E3 = estriol; ll·DOKS = ll-deoxy·17· ketosteroids.
I Received May 20,1976; revised December 27,1976; accepted January 5,1977. 2 Supported in part by Public health Service contract NIH·NCI·6-71· 2016 from the national Cancer Institute, a grant from the Queen Wilhelmina Fund (The Netherlands), and an Edith and Esmond Lando Research Fellowship. 3 Experimental Medicine and Cancer Research, Hebrew University- Hadassah Medical School, Jerusalem, Israel. 4 Department of Clinical Epidemiology, Chaim Sheba Medical Center and Tel·Aviv University Medical School, Tel Hashomer, Israel. Established Investigator of the Chief-Scientist Bureau, Israel Ministry of Health. 5 Endocrine and Cytology Laboratory, General Sick Fund, Tel·Aviv, Israel. 6 Institute for Social Medicine, University of Utrecht, Utrecht, The Netherlands. 7 Rockland Research Institute, Orangeburg, N. Y. 10962.
Random sampling technique. -A random stratified sample of 300 Israeli women, 50-59 years of age, was drawn.
Downloaded from https://academic.oup.com/jnci/article-abstract/59/1/7/888487 by Universite Paris Descartes user on 07 March 2018
Breast cancer incidence/lOO,OOO population
Urine collection. - Each woman was approached by a specially trained nurse who explained the nature of the study and obtained a full personal history, including marital status, outcome of pregnancies, and menstrual data. She then instructed the woman to collect urine for 24 hours on a day of her choice. The first morning urine was collected into a separate container and served for hormonal cytology. After the first sample had been voided, the 24-hour collection was started. Urine collections were made over a 2year period starting in mid -1970. Immediately following collection, urine samples, inside
MATERIALS AND METHODS
VOL. 59, NO.1, JULY 1977
Origin
7
J NATL CANCER INST
8
GROSS ET AL.
special coolers, were shipped by cab to the study center in Jerusalem. In the center, all urines were treated identically: The urine volume was measured and, if necessary, the sample was diluted to 1,000 ml with distilled H 2 0. The pH was adjusted to 7.4. Four aliquots of 200 ml were prepared in 250-ml polypropylene bottles with leakproof lids (Nalge; Nalgene Co., Rochester, N.Y.). The urine was then sterilized by boiling the bottles in a water bath for 30 minutes. Two samples of each urine were shipped by air to Dr. Vestergaard's laboratory at Rockland State Hospital, Orangeburg, New York, for determinations of androgens and corticosteroids. The other two samples were shipped by cab to Dr. Bertini's laboratory in Tel-Aviv for estrogen measurements. Estrogens. -Each sample sent to the laboratory in TelAviv was extracted, after addition of labeled estrogens, according to a modified procedure of Brown et al. (17). The extracts of E. and E3 were sent to Utrecht, The Netherlands, and quantitated by gas-liquid chromatography, according to the method of Thijssen and Veeman (18). Estradiol was not measured because its excretion in postmenopausal women is low. The 50-59 age group was selected for two reasons: 1) to avoid hormonal fluctuations present in premenopausal women, and 2) because the so-called "Westernization effect" has been noted primarily in postmenopausal women (13). Androgens and corticosteroids. - Total urinary 17oxosteroids were assayed by the method ofVestergaard (19), and 17-ketogenic steroids were estimated by the method of Vestergaard and Sayegh (20). Individual steroids were determined by capillary column liquid chromatography, using multi column techniques (21-24), with quantitation by automated, computerized spectrophotometry (25). Hormone cytology. - The urinary sediment method of Lencioni (26) was used. Slides were read by two of the investigators (B.B. and F.d.W.) and classified into atrophic, intermediate, and karyopyknotic cell types (27). Data source and processing. - The distribution of urinary hormone values was checked using the GTNOR subroutine of the computer programs of the International Mathematical and Statistical Libraries Inc., Houston, Texas. The excretion of 11-hydroxyandrosterone was normally distributed and all other hormones were log-normally distributed. Multiple regression analysis was applied to the data, by use of the Northwestern University (Evanston, Ill.) SPSS package (Statistical Package for the Social Sciences), version 5.8. Incidence rates of breast cancer in the population groups under consideration were derived from a nationwide screening of all general hospitals in Israel for the 1960-66 period (Modan B: Unpublished data), supplemented by information available at the central cancer registry.
RESULTS A list of history and of anthropometric characteristics of the total group is given in table 2. The variables showing significant differences between the four main ethnic groups are given in table 3. Significant differences were found for the following variables: age at first parturition, number of J NATL CANCER INST Downloaded from https://academic.oup.com/jnci/article-abstract/59/1/7/888487 by Universite Paris Descartes user on 07 March 2018
TABLE 2. -Selected characteristics of the study population a ~-----
Characteristic
No. of women
Mean
± SD ~-~
Menarche Age at first parturition Age at last parturition No. of pregnancies No. of live births Natural abortions Induced abortions Age at menopause Age as of 1973 Years in Israel Height, em Weight, kg Urine volume, ml Creatinine, g/24 hr Percent smokers
237 124 125 245 245 245 90 241 242 225 231 229 245 245 240
13.2 22.5 35.5 6.3 4.7 0.9 1.6 47.5 56.1 26.3 156.5 64.4 1,113.0 0.65 16.9
-
1.8 6.1 5.8 3.9 3.2 1.7 1.9 4.6 3.7 12.3 6.7 12.9 483.0 0.3 ~---
-
Information on certain characteristics was unavailable in some of the cases. a
pregnancies, number of live births, height, and body weight. These differences are primarily due to the extreme values of age at first parturition, height and weight in the Yemenite, and number of pregnancies and of live births in the European-born women. The steroid excretion patterns of women in the four study groups were examined by an analysis of variance. The results are shown in table 4. Significant differences were noted for the 24-hour excretion of E3, allotetrahydrocortisone, allotetrahydrocortisol, and allocortol. The excretion patterns were then normalized to 1 g of creatinine to avoid possible bias due to collection errors. These data (table 5) reveal significant differences between the ethnic groups in the excretion of E 3, 17-ketosteroids, and allotetrahydrocortisol per gram of creatinine. To establish the relationship between the variables reported in tables 2-5 and the differential incidence of breast cancer, each woman 50-69 years old was assigned the age-specific incidence rate of her country of origin. Correlation coefficients were computed. The variables correlating with the assigned age-specific incidence rates are given in table 6. Since several of these factors are interdependent, multiple regression analysis was applied to the data. A significantly negative correlation between the E3 excretion and the incidence of breast cancer and an apparently positive correlation of E. excretion and the incidence of breast cancer, of borderline significance, were then established.
DISCUSSION The present study reconfirmed the difference in such parameters as age at first delivery, number of pregnancies, height, and body weight, between high- and low-risk groups for breast cancer. The data also revealed significant differences in the hormone profiles of women belonging to high- and low-risk groups. Still, the main finding is the negative correlation between the excretion of E3 and disease incidence, confirming previous observations that estrogen profiles differ in high- and low-risk groups (28-30). Cole and MacMahon (28) and MacMahon et al. (29) noted that Chinese and Japanese women excrete a higher proportion of VOL. 59, NO.1, JULY 1977
9
STEROID EXCRETION PATTERNS AND BREAST CANCER IN ISRAEL
TABLE 3. -Significantly different variables regarding history and anthropometry Country or continent of origin
Age at first parturition
Yemen Africa Asia Europe Analysis of variance, F Level of significance, P
No. 25 38 30 31
Mean 18.8 23.2 25.7 25.4
No. of pregnancies No. Mean 58 63 59 65
No. of live births 58 63 59 65
Weight No. Mean
57 58 55 61
56 57.3 57 67.7 65 65.9 61 65.5
Height
No. Mean
7.6 7.3 6.9 3.5
No. Mean
6.3 5.6 5.2 1.9
152.9 157.6 158.1 157.2
6.57
17.74
32.10
8.20
8.48
ABSTRACT-A nationwide study of the steroid excretion pat· terns in postmenopausal Israeli migrant women demonstrated differences between high· and low·risk groups for breast cancer in the following variables: age at first parturition, number of pregnancies, number of live births, height, and weight. The direction of the differences was in line with those observed for breast cancer patients. The groups also differed in the excretion of estriol, 17-ketosteroids, and allotetrahydrocortisol. Multiple regression analysis revealed that the excretion of estriol was significantly lower in population groups in whom breast cancer incidence was high. Possibly this trend-which has also been observed in adolescent and premenopausal women-reflected environmental influences on peripheral estrogen metabo· lism.-J Natl Cancer Inst 59: 7-11,1977.
TABLE I.-Annual age-adjusted incidence rates of breast cancer in Jewish women in Israel, 50-69 years of age (1960-66), according to country or continent of origin a
Experimental and clinical evidence suggests that hormonal factors play a role in the genesis of breast cancer. However, wide variations in the reported incidence of breast cancer (1-6) provide arguments for the existence of an environmental influence. An interplay between environmental and hormonal factors might therefore be considered (7, 8). MacMahon et al. (9-11) and Lilienfeld et al. (12) showed that an important factor in the etiology of breast cancer is the age at which the woman first gives birth. Although it has been suggested that a critical induction period is between menarche and first pregnancy, a second critical induction period may occur around the time of menopause (11). The latter hypothesis, which implies the presence of at least two disease entities, is supported by the fact that the age-specific incidence curve of breast cancer differs in high- and low-risk populations (13, 14). Thus in Asia (e.g., Japan), Africa, and Latin America the age-specific incidence curves are different from those in the United States and other "Western" countries (1). The rise at premenopausal age is similar, though at a lower level, but after age 50 there is a leveling off to a plateau (3). "Westernization" affects incidence rates primarily in the older age groups. In Israel the incidence of breast cancer differs widely between the various ethnic groups (15); the highest rates occur among European-born and the lowest, among Yemenite women. The age-specific incidence curve of the former resembles the pattern observed in high-risk populations, whereas that of the Oriental group follows the low-risk pattern (16). The present study explores a possible correlation between the marked variation in the incidence of breast cancer among the major Israeli ethnic groups (table 1) and selected personal and hormonal parameters in postmenopausal women.
a
b
Yemen Africa Asia b Europe
27.2 52.5 48.1 124.9
Direct method; total Israeli female population used as standard. Excluding Yemen and Israel.
ABBREVIATIONS USED: El = estrone; E3 = estriol; ll·DOKS = ll-deoxy·17· ketosteroids.
I Received May 20,1976; revised December 27,1976; accepted January 5,1977. 2 Supported in part by Public health Service contract NIH·NCI·6-71· 2016 from the national Cancer Institute, a grant from the Queen Wilhelmina Fund (The Netherlands), and an Edith and Esmond Lando Research Fellowship. 3 Experimental Medicine and Cancer Research, Hebrew University- Hadassah Medical School, Jerusalem, Israel. 4 Department of Clinical Epidemiology, Chaim Sheba Medical Center and Tel·Aviv University Medical School, Tel Hashomer, Israel. Established Investigator of the Chief-Scientist Bureau, Israel Ministry of Health. 5 Endocrine and Cytology Laboratory, General Sick Fund, Tel·Aviv, Israel. 6 Institute for Social Medicine, University of Utrecht, Utrecht, The Netherlands. 7 Rockland Research Institute, Orangeburg, N. Y. 10962.
Random sampling technique. -A random stratified sample of 300 Israeli women, 50-59 years of age, was drawn.
Downloaded from https://academic.oup.com/jnci/article-abstract/59/1/7/888487 by Universite Paris Descartes user on 07 March 2018
Breast cancer incidence/lOO,OOO population
Urine collection. - Each woman was approached by a specially trained nurse who explained the nature of the study and obtained a full personal history, including marital status, outcome of pregnancies, and menstrual data. She then instructed the woman to collect urine for 24 hours on a day of her choice. The first morning urine was collected into a separate container and served for hormonal cytology. After the first sample had been voided, the 24-hour collection was started. Urine collections were made over a 2year period starting in mid -1970. Immediately following collection, urine samples, inside
MATERIALS AND METHODS
VOL. 59, NO.1, JULY 1977
Origin
7
J NATL CANCER INST
8
GROSS ET AL.
special coolers, were shipped by cab to the study center in Jerusalem. In the center, all urines were treated identically: The urine volume was measured and, if necessary, the sample was diluted to 1,000 ml with distilled H 2 0. The pH was adjusted to 7.4. Four aliquots of 200 ml were prepared in 250-ml polypropylene bottles with leakproof lids (Nalge; Nalgene Co., Rochester, N.Y.). The urine was then sterilized by boiling the bottles in a water bath for 30 minutes. Two samples of each urine were shipped by air to Dr. Vestergaard's laboratory at Rockland State Hospital, Orangeburg, New York, for determinations of androgens and corticosteroids. The other two samples were shipped by cab to Dr. Bertini's laboratory in Tel-Aviv for estrogen measurements. Estrogens. -Each sample sent to the laboratory in TelAviv was extracted, after addition of labeled estrogens, according to a modified procedure of Brown et al. (17). The extracts of E. and E3 were sent to Utrecht, The Netherlands, and quantitated by gas-liquid chromatography, according to the method of Thijssen and Veeman (18). Estradiol was not measured because its excretion in postmenopausal women is low. The 50-59 age group was selected for two reasons: 1) to avoid hormonal fluctuations present in premenopausal women, and 2) because the so-called "Westernization effect" has been noted primarily in postmenopausal women (13). Androgens and corticosteroids. - Total urinary 17oxosteroids were assayed by the method ofVestergaard (19), and 17-ketogenic steroids were estimated by the method of Vestergaard and Sayegh (20). Individual steroids were determined by capillary column liquid chromatography, using multi column techniques (21-24), with quantitation by automated, computerized spectrophotometry (25). Hormone cytology. - The urinary sediment method of Lencioni (26) was used. Slides were read by two of the investigators (B.B. and F.d.W.) and classified into atrophic, intermediate, and karyopyknotic cell types (27). Data source and processing. - The distribution of urinary hormone values was checked using the GTNOR subroutine of the computer programs of the International Mathematical and Statistical Libraries Inc., Houston, Texas. The excretion of 11-hydroxyandrosterone was normally distributed and all other hormones were log-normally distributed. Multiple regression analysis was applied to the data, by use of the Northwestern University (Evanston, Ill.) SPSS package (Statistical Package for the Social Sciences), version 5.8. Incidence rates of breast cancer in the population groups under consideration were derived from a nationwide screening of all general hospitals in Israel for the 1960-66 period (Modan B: Unpublished data), supplemented by information available at the central cancer registry.
RESULTS A list of history and of anthropometric characteristics of the total group is given in table 2. The variables showing significant differences between the four main ethnic groups are given in table 3. Significant differences were found for the following variables: age at first parturition, number of J NATL CANCER INST Downloaded from https://academic.oup.com/jnci/article-abstract/59/1/7/888487 by Universite Paris Descartes user on 07 March 2018
TABLE 2. -Selected characteristics of the study population a ~-----
Characteristic
No. of women
Mean
± SD ~-~
Menarche Age at first parturition Age at last parturition No. of pregnancies No. of live births Natural abortions Induced abortions Age at menopause Age as of 1973 Years in Israel Height, em Weight, kg Urine volume, ml Creatinine, g/24 hr Percent smokers
237 124 125 245 245 245 90 241 242 225 231 229 245 245 240
13.2 22.5 35.5 6.3 4.7 0.9 1.6 47.5 56.1 26.3 156.5 64.4 1,113.0 0.65 16.9
-
1.8 6.1 5.8 3.9 3.2 1.7 1.9 4.6 3.7 12.3 6.7 12.9 483.0 0.3 ~---
-
Information on certain characteristics was unavailable in some of the cases. a
pregnancies, number of live births, height, and body weight. These differences are primarily due to the extreme values of age at first parturition, height and weight in the Yemenite, and number of pregnancies and of live births in the European-born women. The steroid excretion patterns of women in the four study groups were examined by an analysis of variance. The results are shown in table 4. Significant differences were noted for the 24-hour excretion of E3, allotetrahydrocortisone, allotetrahydrocortisol, and allocortol. The excretion patterns were then normalized to 1 g of creatinine to avoid possible bias due to collection errors. These data (table 5) reveal significant differences between the ethnic groups in the excretion of E 3, 17-ketosteroids, and allotetrahydrocortisol per gram of creatinine. To establish the relationship between the variables reported in tables 2-5 and the differential incidence of breast cancer, each woman 50-69 years old was assigned the age-specific incidence rate of her country of origin. Correlation coefficients were computed. The variables correlating with the assigned age-specific incidence rates are given in table 6. Since several of these factors are interdependent, multiple regression analysis was applied to the data. A significantly negative correlation between the E3 excretion and the incidence of breast cancer and an apparently positive correlation of E. excretion and the incidence of breast cancer, of borderline significance, were then established.
DISCUSSION The present study reconfirmed the difference in such parameters as age at first delivery, number of pregnancies, height, and body weight, between high- and low-risk groups for breast cancer. The data also revealed significant differences in the hormone profiles of women belonging to high- and low-risk groups. Still, the main finding is the negative correlation between the excretion of E3 and disease incidence, confirming previous observations that estrogen profiles differ in high- and low-risk groups (28-30). Cole and MacMahon (28) and MacMahon et al. (29) noted that Chinese and Japanese women excrete a higher proportion of VOL. 59, NO.1, JULY 1977
9
STEROID EXCRETION PATTERNS AND BREAST CANCER IN ISRAEL
TABLE 3. -Significantly different variables regarding history and anthropometry Country or continent of origin
Age at first parturition
Yemen Africa Asia Europe Analysis of variance, F Level of significance, P
No. 25 38 30 31
Mean 18.8 23.2 25.7 25.4
No. of pregnancies No. Mean 58 63 59 65
No. of live births 58 63 59 65
Weight No. Mean
57 58 55 61
56 57.3 57 67.7 65 65.9 61 65.5
Height
No. Mean
7.6 7.3 6.9 3.5
No. Mean
6.3 5.6 5.2 1.9
152.9 157.6 158.1 157.2
6.57
17.74
32.10
8.20
8.48
