Breast Cancer: Strategies for the 1990s 1
0039-6109/90 $0.00
+ .20
Epidemiology and Natural History of Breast Cancer Implications for the Body WeightBreast Cancer Controversy
Alfredo Morabia, MD, PhD, MPH, * and Ernst L. Wynder, MDt
Breast cancer was the principal cause of cancer death in women from 1973 to 1985 in the United States. 50 In 1985, breast cancer was surpassed as a cause of cancer death only by leukemia in girls under the age of 15 and only by colorectal cancer in women past the age of 74. 50 Evolution of our understanding of the differentiation of the normal breast, as well as progress in molecular and cellular biology, has brought new insights into the possible pathways of breast carcinogenesis. These concepts have also led to reconsideration of the risk factors identified or suspected heretofore. In the present paper, we review these new concepts and evaluate their explanatory potential in the body weight-breast cancer controversy. EPIDEMIOLOGY OF BREAST CANCER From 1950 to 1985, the overall age-adjusted mortality rate of breast cancer rose from 22.2 to 23.2 per 100,000 US women. 49 In 1985, overall US age-adjusted mortality rates were 23.3 per 100,000 in whites and 25.3 per 100,000 in blacks,49 but age-specific death rates from 1973 through 1981 show a crossover at age 50 between whites and blacks, with premenopausal black females having higher rates. In 1985, the age-adjusted incidence of *Senior Epidemiologist, American Health Foundation, and Adjunct Assistant Professor, Division of Epidemiology, Columbia University School of Public Health, New York, New York tPresident, American Health Foundation, and Adjunct Member, Memorial Hospital for Cancer and Allied Diseases, New York, and Adjunct Professor of Public Health Practices, Columbia University, New York, New York
Surgical Clinics of North America-Vol. 70, No.4, August 1990
739
740
ALFREDO MORABIA AND ERNST
L.
WYNDER
breast cancer in the us was 104.2 per 100,000 white women and 87.7 per 100,000 black women. 49 Before age 35, black females have higher incidence .rates than white females (Fig. 1). The public health importance of breast cancer contrasts with the lack of identified causes amenable to prevention. This situation warrants further study of factors, such as body weight or dietary fat, whose role in breast cancer etiology is still controversial but that are potentially amenable to manipulation. Elucidation of the mechanisms that relate these factors to breast cancer may contribute to a more thorough understanding of the etiologic basis of the disease and offer a point for intervention.
GENESIS AND NATURAL HISTORY OF BREAST CANCER Differentiation of the Mammary Gland Most human organs are differentiated at the end of fetal life, but most of the development and differentiation of the mammary gland occur after puberty. The mammary gland at birth is formed by primary mammary ducts, which elongate and branch during childhood. At premenarche, the duct epithelium proliferates. At menarche, under the influence of progesterone and, probably, estrogen, terminal duct lobular units, also called acini, are formed that are the dynamic structure of the mammary gland. The number of these acini increases at each menstruation, with new budding of structures until approximately age 35 years. 60 With pregnancy, there is a florid increase in the number of acini, resulting in a full differentiation of their morphology and function. The definitive structure of the duct is essentially settled by the end of the first half of pregnancy. However, just before and during parturition, there is a new wave of mitotic activity in the gland. 60 Thus, differentiation of acini either is progressively fostered during the luteal phases of the menstrual cycle or occurs with a full-term pregnancy. However, when pregnancy does not supervene, full differentiation of the breast may never be attained. At menopause, acini as well as interlobular fibrous tissues undergo atrophy, but the large and intermediate duct system persists. With further involution of the glandular epithelium, the breast becomes composed mainly of large ducts and increased amounts of connective tissue and fat. Breast Carcinogenesis Breast cancer is a disease of the glandular epithelium and is thought to result from a mutation during the differentiation of stem cells into mature epithelial cells. 61 The lineage of the mammary epithelial cells achieves a rapid renewal by a relatively few divisions of immortal stem cells.1O The number of mutations is proportional to the rate and number of stem cell divisions. Factors that accelerate cell division can have a carcinogenic effect, because they increase the likelihood of a mutation with a survival advantage that may be carried in an expanding clone of cancerous cells. Hormones are likely to influence the susceptibility of the breast epithelium to environmental carcinogens such as the heterocyclic amine 1-amino-2-
~
'""'
25 to 29
'I
/
I
30
I
/
, /,
35 to 39 40 to 44
While -----.
+-- ..
Black
_-...- .....
55
.""".
to 34
to 64
to 49
Age
60
45
59
to
65
to 69
60
10
70
64
74
to
to 79
75
1~1---r--.---~--r--'--~---r--'---~--r--'r--'
4
7
10
40
70
100
',.",'-
-
... -.-...--~.-- ... - ..
Figure 1. Average annualized breast cancer incidence rates among US black and white women, all SEER areas excluding Puerto Rico, 1982-1984, (From White E: Rising incidence of breast cancer [response to letter]. INCI8O:2, 1988,)
U .5
'U
•
U C
Q.
0
..... • ...• IE: " •
0
0 0 0
400
742
ALFREDO MORABIA AND ERNST
L.
WYNDER
methylimidazo[4,5-f]quinoline 67, 70 or cholesterol epoxides54 because hormones control the differentiation of the mammary gland epithelium and thus regulate the rate of stem cell division. Current knowledge of the mechanisms of differentiation of the mammary gland suggests that both estrogens and progestins may be involved in breast carcinogenesis, This hypothesis is supported by the predominance of the disease among females (breast cancer is 100 times more frequent in females than in males) and by the epidemiologic association of breast cancer with the events of reproductive life. As early as 1945, it was observed that the rise (on the logarithmic scale) in the age-specific incidence of breast cancer was much steeper before ages 40 to 45 years than after and that there was a trough in the age curve at the 50- to 54-year age group that could reflect the effect of hormonal changes related to menopause. 13 Later, it was shown that patients with breast cancer have a reproductive history substantially different from that of controls.40, 44. 77 In particular, patients were younger at menarche and older at first live birth, had delayed menopause, and were less likely to have had premenopausal oophorectomy. As reported in several reviews, 36, 72, 73 these epidemiologic characteristics of breast cancer patients have been found in different populations, in different countries, and using different study designs. Reproductive factors thus form the hard core of the epidemiologic evidence for the role of hormonal factors in the etiology of breast cancer. In 1973, MacMahon and coworkers 45 raised the hypothesis, at that time unsupported by direct evidence in humans, that women who developed breast cancer were more exposed to estradiol and estrone, two estrogen fractions, and had a lower ratio of estriol to estradiol plus estrone than women who did not develop breast cancer. However, evidence from subsequent studies does not support the MacMahon hypothesis. 7. 9. 23, 31. 74 Moreover, Thomas 73 concluded from a review of studies of endogenous estrogen fractions that there was no support for an independent effect of these hormonal fractions on the risk of breast cancer. Estrogen levels in the blood, urine, and breast fluids, as well as the concentration of estrogen receptors on the mammary epithelial cells, have been measured. 7, 9, 23, 31, 45, 74 These studies are limited by the extreme difficulty in obtaining an accurate representation of a premenopausal woman's estrogenic exposure. Not only do estrogen levels vary conSiderably during the menstrual cycle, but it has not been possible until very recently to distinguish the levels of total serum estrogens from those of the free (biologically active) fractions, which more closely reflect the true cellular exposure. 73 Thus, even though there is no support at present for an independent contribution of estrogen fractions to breast cancer etiology, such an effect cannot be ruled out, as the absence of supporting evidence may be attributable to the measurement inaccuracy. The role of progesterone or of its urinary metabolite, pregnanediol, in breast carcinogenesis has been less studied than that of estrogens. Animal studies have shown that progesterone can be a cocarcinogen,48, 55 but human studies have yielded contradictory results and so are inconclusive. 73 A better understanding of the physiology of the mammary gland and the postulated
EPIDEMIOLOGY AND NATURAL HISTORY OF BREAST CANCER
743
mechanisms of carcinogenesis may offer clues as to whether progesterone is protective or harmful in breast carcinogenesis. As explained previously, the period of highest proliferative activity of the mammary epithelium stem cells is during each ovulatory cycle between puberty and either the first full-term pregnancy or menopause among nulliparous women. It thus is important to know whether the proliferation occurs mostly in the first part of the menstrual cycle (the follicular phase), during which estrogens prime, or in the second part (the luteal phase), which is under the influence of both estrogens and progesterone. The answer to this crucial question is still controversial. Vogel and coworkers 76 reported, without quantitative details, that mitotic activity was highest during the follicular phase and thus that estrogen had a proliferative effect on the mammary gland epithelium, whereas progesterone acted as an ahtiestrogen in inhibiting proliferation. These observations Were consistent with Korenman's "open window" theory38 stating that breast cancer is related to a defective luteal phase with low or absent progesterone secretion: women with anovulatory cycles were predicted to be at higher risk because the proliferative effect of estrogens was unopposed by progesterone. Vogel and coworkers based their conclusions on morphologic changes identified in the mammary stroma and epithelium during the phases of the menstrual cycle. However, their approach does not allow quantitation of differences between epithelial and stromal cell growth. 46 Other workers were able to assess ductal epithelium growth quantitatively by autoradiographic analysis. 27, 43, 56 With this technique, ductal cell DNA is radiolabel ed, and the number of mitoses can be counted during the follicular and luteal phases. In contrast to the results of Vogel and coworkers,76 the majority of these studies show that human and animal mammary glands proliferate mostly during the luteal phase of the cycle. During the estrogenpriming follicular phase, terminal ductules are few, and there is no mitotic activity. During the luteal phase, because of the increased progesterone levels, perhaps on the background of estrogen priming or as a result of synergism between the two hormones, there is enhanced mitotic activity. These studies contradict the "open window theory," because, if progesterone were acting as an antiestrogen, proliferation of the mammary epithelium would be less extensive during the luteal than during the follicular phase. Thus, it can be postulated that the risk of carcinogenic mutations is proportional to the time required for all stem cells to undergo differentiation. The risk would then be proportional to the duration of the interval separatihg menarche from the first full-term birth and to the number of ovulatory cycles during this period of time. Indeed, animal studies have shown that the events of reproductive life may inhibit or enhance carcinogenesis according to whether they shorten or extend the period between menarche and full differentiation of the mammary gland. Successful tumor induction occurs when the carcinogenic agent is administered to young virgin animals as long as they are postpubertal. Conversely, tumor incidence decreases with aging of the host, and administration of carcinogens to parous animals almost completely fails to induce carcinomas. 60 In human studies, a similar phenomenon is observed in the increased risk associated with young age at menarche, later age at menopause, or
744
ALFREDO MORABIA AND ERNST
L.
WYNDER
longer duration of ovulatory activity, that is, the years between menarche and menopause not spent in the pregnant or lactating states, 30. 63 as well as in the protective effect of a young age at first live birth,44 as a full-term birth shifts the mammary gland from a proliferative state with high susceptibility to carcinogens to a low-proliferative, low-risk state. Breast Cancer Promotion Factors incriminated in the development of breast cancer once the carcinogenic mutation has taken place are estrogen receptors,73 social class,3 tobacco use,17 diet,33 and weight. Several studies have shown that overweight is inversely related to survival among breast cancer patients. 6, 17, 19, 22. 28. 29, 51. 71, 80 In a study by Daniell, 17 the finding of four or more tumorous lymph nodes was more common in patients with obesity (weight 30 per cent or more above the average ideal for height with medium habitus according to the 1983 Metropolitan Life Insurance data) among women older than 50 years but not in younger women. The distinction between initiating and promoting risk factors is closely related to the issue of whether premenopausal and postmenopausal breast cancer are two different diseases. Several studies have found that different sets of risk factors seem to determine breast cancer risk before and after menopause. 8. 16. 41. 53, 66, 69 However, it is unknown whether carcinogenesis occurs mostly during the reproductive years, postmenopausal breast cancer risk factors being related to promotion, or whether initiation can still occur during postmenopausal years but as a result of a process distinct from that of premenopausal years. 2 Body weight is one of the factors that seem to be related differently to breast cancer in premenopausal and postmenopausal women. It will serve us to illustrate the implications of the concepts of differentiation and carcinogenesis presented above for our understanding of the genesis and growth of breast cancer. BODY WEIGHT AND BREAST CANCER There is a wealth of epidemiologic reports on the relation of body weight to breast cancer. 37, 58 However, the studies are not easy to compare, as the authors used different indices of overweight such as body weight or Quetelet index (weight for height squared) and set different cut-offs for computing the relative risk estimates. We have reviewed the studies that reported the association of excess body weight and breast cancer separately for premenopausal and postmenopausal women (or that used 50 years of age as the cut-off for menopausal status because they lacked precise information about onset of menopause). Findings from these studies l , 12. 20, 21. 36. 42, 47, 53, 65. 78 (Taioli et aI, American Health Foundation, unpublished data, January 1990) are presented in Tables 1 to 4. It is important to note that the cut-offs most frequently chosen reflect moderate overweight rather than obesity, as 58 or 64 kg is not obese for a woman of average height. A single study36 reported a statistically significant reduction in the risk
745
EPIDEMIOLOGY AND NATURAL HISTORY OF BREAST CANCER
Table 1. Odds Ratios for Breast Cancer Related to Weight Among Premenopausal Women WEIGHT (KG) STUDY
Lin et al42 * Mirra et al47 * de Waard et al21 t Staszewski65 * Wynder et al78 * Kelsey et al36 * Taioli et aI (unpublished)1I
GROUP DEFINITION
0039-6109/90 $0.00
+ .20
Epidemiology and Natural History of Breast Cancer Implications for the Body WeightBreast Cancer Controversy
Alfredo Morabia, MD, PhD, MPH, * and Ernst L. Wynder, MDt
Breast cancer was the principal cause of cancer death in women from 1973 to 1985 in the United States. 50 In 1985, breast cancer was surpassed as a cause of cancer death only by leukemia in girls under the age of 15 and only by colorectal cancer in women past the age of 74. 50 Evolution of our understanding of the differentiation of the normal breast, as well as progress in molecular and cellular biology, has brought new insights into the possible pathways of breast carcinogenesis. These concepts have also led to reconsideration of the risk factors identified or suspected heretofore. In the present paper, we review these new concepts and evaluate their explanatory potential in the body weight-breast cancer controversy. EPIDEMIOLOGY OF BREAST CANCER From 1950 to 1985, the overall age-adjusted mortality rate of breast cancer rose from 22.2 to 23.2 per 100,000 US women. 49 In 1985, overall US age-adjusted mortality rates were 23.3 per 100,000 in whites and 25.3 per 100,000 in blacks,49 but age-specific death rates from 1973 through 1981 show a crossover at age 50 between whites and blacks, with premenopausal black females having higher rates. In 1985, the age-adjusted incidence of *Senior Epidemiologist, American Health Foundation, and Adjunct Assistant Professor, Division of Epidemiology, Columbia University School of Public Health, New York, New York tPresident, American Health Foundation, and Adjunct Member, Memorial Hospital for Cancer and Allied Diseases, New York, and Adjunct Professor of Public Health Practices, Columbia University, New York, New York
Surgical Clinics of North America-Vol. 70, No.4, August 1990
739
740
ALFREDO MORABIA AND ERNST
L.
WYNDER
breast cancer in the us was 104.2 per 100,000 white women and 87.7 per 100,000 black women. 49 Before age 35, black females have higher incidence .rates than white females (Fig. 1). The public health importance of breast cancer contrasts with the lack of identified causes amenable to prevention. This situation warrants further study of factors, such as body weight or dietary fat, whose role in breast cancer etiology is still controversial but that are potentially amenable to manipulation. Elucidation of the mechanisms that relate these factors to breast cancer may contribute to a more thorough understanding of the etiologic basis of the disease and offer a point for intervention.
GENESIS AND NATURAL HISTORY OF BREAST CANCER Differentiation of the Mammary Gland Most human organs are differentiated at the end of fetal life, but most of the development and differentiation of the mammary gland occur after puberty. The mammary gland at birth is formed by primary mammary ducts, which elongate and branch during childhood. At premenarche, the duct epithelium proliferates. At menarche, under the influence of progesterone and, probably, estrogen, terminal duct lobular units, also called acini, are formed that are the dynamic structure of the mammary gland. The number of these acini increases at each menstruation, with new budding of structures until approximately age 35 years. 60 With pregnancy, there is a florid increase in the number of acini, resulting in a full differentiation of their morphology and function. The definitive structure of the duct is essentially settled by the end of the first half of pregnancy. However, just before and during parturition, there is a new wave of mitotic activity in the gland. 60 Thus, differentiation of acini either is progressively fostered during the luteal phases of the menstrual cycle or occurs with a full-term pregnancy. However, when pregnancy does not supervene, full differentiation of the breast may never be attained. At menopause, acini as well as interlobular fibrous tissues undergo atrophy, but the large and intermediate duct system persists. With further involution of the glandular epithelium, the breast becomes composed mainly of large ducts and increased amounts of connective tissue and fat. Breast Carcinogenesis Breast cancer is a disease of the glandular epithelium and is thought to result from a mutation during the differentiation of stem cells into mature epithelial cells. 61 The lineage of the mammary epithelial cells achieves a rapid renewal by a relatively few divisions of immortal stem cells.1O The number of mutations is proportional to the rate and number of stem cell divisions. Factors that accelerate cell division can have a carcinogenic effect, because they increase the likelihood of a mutation with a survival advantage that may be carried in an expanding clone of cancerous cells. Hormones are likely to influence the susceptibility of the breast epithelium to environmental carcinogens such as the heterocyclic amine 1-amino-2-
~
'""'
25 to 29
'I
/
I
30
I
/
, /,
35 to 39 40 to 44
While -----.
+-- ..
Black
_-...- .....
55
.""".
to 34
to 64
to 49
Age
60
45
59
to
65
to 69
60
10
70
64
74
to
to 79
75
1~1---r--.---~--r--'--~---r--'---~--r--'r--'
4
7
10
40
70
100
',.",'-
-
... -.-...--~.-- ... - ..
Figure 1. Average annualized breast cancer incidence rates among US black and white women, all SEER areas excluding Puerto Rico, 1982-1984, (From White E: Rising incidence of breast cancer [response to letter]. INCI8O:2, 1988,)
U .5
'U
•
U C
Q.
0
..... • ...• IE: " •
0
0 0 0
400
742
ALFREDO MORABIA AND ERNST
L.
WYNDER
methylimidazo[4,5-f]quinoline 67, 70 or cholesterol epoxides54 because hormones control the differentiation of the mammary gland epithelium and thus regulate the rate of stem cell division. Current knowledge of the mechanisms of differentiation of the mammary gland suggests that both estrogens and progestins may be involved in breast carcinogenesis, This hypothesis is supported by the predominance of the disease among females (breast cancer is 100 times more frequent in females than in males) and by the epidemiologic association of breast cancer with the events of reproductive life. As early as 1945, it was observed that the rise (on the logarithmic scale) in the age-specific incidence of breast cancer was much steeper before ages 40 to 45 years than after and that there was a trough in the age curve at the 50- to 54-year age group that could reflect the effect of hormonal changes related to menopause. 13 Later, it was shown that patients with breast cancer have a reproductive history substantially different from that of controls.40, 44. 77 In particular, patients were younger at menarche and older at first live birth, had delayed menopause, and were less likely to have had premenopausal oophorectomy. As reported in several reviews, 36, 72, 73 these epidemiologic characteristics of breast cancer patients have been found in different populations, in different countries, and using different study designs. Reproductive factors thus form the hard core of the epidemiologic evidence for the role of hormonal factors in the etiology of breast cancer. In 1973, MacMahon and coworkers 45 raised the hypothesis, at that time unsupported by direct evidence in humans, that women who developed breast cancer were more exposed to estradiol and estrone, two estrogen fractions, and had a lower ratio of estriol to estradiol plus estrone than women who did not develop breast cancer. However, evidence from subsequent studies does not support the MacMahon hypothesis. 7. 9. 23, 31. 74 Moreover, Thomas 73 concluded from a review of studies of endogenous estrogen fractions that there was no support for an independent effect of these hormonal fractions on the risk of breast cancer. Estrogen levels in the blood, urine, and breast fluids, as well as the concentration of estrogen receptors on the mammary epithelial cells, have been measured. 7, 9, 23, 31, 45, 74 These studies are limited by the extreme difficulty in obtaining an accurate representation of a premenopausal woman's estrogenic exposure. Not only do estrogen levels vary conSiderably during the menstrual cycle, but it has not been possible until very recently to distinguish the levels of total serum estrogens from those of the free (biologically active) fractions, which more closely reflect the true cellular exposure. 73 Thus, even though there is no support at present for an independent contribution of estrogen fractions to breast cancer etiology, such an effect cannot be ruled out, as the absence of supporting evidence may be attributable to the measurement inaccuracy. The role of progesterone or of its urinary metabolite, pregnanediol, in breast carcinogenesis has been less studied than that of estrogens. Animal studies have shown that progesterone can be a cocarcinogen,48, 55 but human studies have yielded contradictory results and so are inconclusive. 73 A better understanding of the physiology of the mammary gland and the postulated
EPIDEMIOLOGY AND NATURAL HISTORY OF BREAST CANCER
743
mechanisms of carcinogenesis may offer clues as to whether progesterone is protective or harmful in breast carcinogenesis. As explained previously, the period of highest proliferative activity of the mammary epithelium stem cells is during each ovulatory cycle between puberty and either the first full-term pregnancy or menopause among nulliparous women. It thus is important to know whether the proliferation occurs mostly in the first part of the menstrual cycle (the follicular phase), during which estrogens prime, or in the second part (the luteal phase), which is under the influence of both estrogens and progesterone. The answer to this crucial question is still controversial. Vogel and coworkers 76 reported, without quantitative details, that mitotic activity was highest during the follicular phase and thus that estrogen had a proliferative effect on the mammary gland epithelium, whereas progesterone acted as an ahtiestrogen in inhibiting proliferation. These observations Were consistent with Korenman's "open window" theory38 stating that breast cancer is related to a defective luteal phase with low or absent progesterone secretion: women with anovulatory cycles were predicted to be at higher risk because the proliferative effect of estrogens was unopposed by progesterone. Vogel and coworkers based their conclusions on morphologic changes identified in the mammary stroma and epithelium during the phases of the menstrual cycle. However, their approach does not allow quantitation of differences between epithelial and stromal cell growth. 46 Other workers were able to assess ductal epithelium growth quantitatively by autoradiographic analysis. 27, 43, 56 With this technique, ductal cell DNA is radiolabel ed, and the number of mitoses can be counted during the follicular and luteal phases. In contrast to the results of Vogel and coworkers,76 the majority of these studies show that human and animal mammary glands proliferate mostly during the luteal phase of the cycle. During the estrogenpriming follicular phase, terminal ductules are few, and there is no mitotic activity. During the luteal phase, because of the increased progesterone levels, perhaps on the background of estrogen priming or as a result of synergism between the two hormones, there is enhanced mitotic activity. These studies contradict the "open window theory," because, if progesterone were acting as an antiestrogen, proliferation of the mammary epithelium would be less extensive during the luteal than during the follicular phase. Thus, it can be postulated that the risk of carcinogenic mutations is proportional to the time required for all stem cells to undergo differentiation. The risk would then be proportional to the duration of the interval separatihg menarche from the first full-term birth and to the number of ovulatory cycles during this period of time. Indeed, animal studies have shown that the events of reproductive life may inhibit or enhance carcinogenesis according to whether they shorten or extend the period between menarche and full differentiation of the mammary gland. Successful tumor induction occurs when the carcinogenic agent is administered to young virgin animals as long as they are postpubertal. Conversely, tumor incidence decreases with aging of the host, and administration of carcinogens to parous animals almost completely fails to induce carcinomas. 60 In human studies, a similar phenomenon is observed in the increased risk associated with young age at menarche, later age at menopause, or
744
ALFREDO MORABIA AND ERNST
L.
WYNDER
longer duration of ovulatory activity, that is, the years between menarche and menopause not spent in the pregnant or lactating states, 30. 63 as well as in the protective effect of a young age at first live birth,44 as a full-term birth shifts the mammary gland from a proliferative state with high susceptibility to carcinogens to a low-proliferative, low-risk state. Breast Cancer Promotion Factors incriminated in the development of breast cancer once the carcinogenic mutation has taken place are estrogen receptors,73 social class,3 tobacco use,17 diet,33 and weight. Several studies have shown that overweight is inversely related to survival among breast cancer patients. 6, 17, 19, 22. 28. 29, 51. 71, 80 In a study by Daniell, 17 the finding of four or more tumorous lymph nodes was more common in patients with obesity (weight 30 per cent or more above the average ideal for height with medium habitus according to the 1983 Metropolitan Life Insurance data) among women older than 50 years but not in younger women. The distinction between initiating and promoting risk factors is closely related to the issue of whether premenopausal and postmenopausal breast cancer are two different diseases. Several studies have found that different sets of risk factors seem to determine breast cancer risk before and after menopause. 8. 16. 41. 53, 66, 69 However, it is unknown whether carcinogenesis occurs mostly during the reproductive years, postmenopausal breast cancer risk factors being related to promotion, or whether initiation can still occur during postmenopausal years but as a result of a process distinct from that of premenopausal years. 2 Body weight is one of the factors that seem to be related differently to breast cancer in premenopausal and postmenopausal women. It will serve us to illustrate the implications of the concepts of differentiation and carcinogenesis presented above for our understanding of the genesis and growth of breast cancer. BODY WEIGHT AND BREAST CANCER There is a wealth of epidemiologic reports on the relation of body weight to breast cancer. 37, 58 However, the studies are not easy to compare, as the authors used different indices of overweight such as body weight or Quetelet index (weight for height squared) and set different cut-offs for computing the relative risk estimates. We have reviewed the studies that reported the association of excess body weight and breast cancer separately for premenopausal and postmenopausal women (or that used 50 years of age as the cut-off for menopausal status because they lacked precise information about onset of menopause). Findings from these studies l , 12. 20, 21. 36. 42, 47, 53, 65. 78 (Taioli et aI, American Health Foundation, unpublished data, January 1990) are presented in Tables 1 to 4. It is important to note that the cut-offs most frequently chosen reflect moderate overweight rather than obesity, as 58 or 64 kg is not obese for a woman of average height. A single study36 reported a statistically significant reduction in the risk
745
EPIDEMIOLOGY AND NATURAL HISTORY OF BREAST CANCER
Table 1. Odds Ratios for Breast Cancer Related to Weight Among Premenopausal Women WEIGHT (KG) STUDY
Lin et al42 * Mirra et al47 * de Waard et al21 t Staszewski65 * Wynder et al78 * Kelsey et al36 * Taioli et aI (unpublished)1I
GROUP DEFINITION
