Breast CancerResearchand Treatment33: 75-82, 1994. © 1994KluwerAcademic Publishers. Printedin the Netherlands. Report
Early adult body weight, body mass index, and premenopausal bilateral breast cancer: data from a case-control study
Oiske Ursin, 1'2Annlia Paganini-Hill, 3 Jack Siemiatycki,4 W. Douglas Thompson 5 and Robert W. Haile l
1Department of Epidemiology, School of Public Health, Center for the Health Sciences, University of California, Los Angeles, CA 90024, USA; 2 Current address: Department of Preventive Medicine, University of Southern California, Los Angeles, CA, USA; ~ Department of Preventive Medicine, University of Southern California, Los Angeles, CA, USA; ~Department of Epidemiology, Institut Armand Frappier, Quebec, Canada; 5Applied Medical Sciences, University of Southern Maine, Portland, ME, USA
Key words: body mass, obesity, premenopausal bilateral breast neoplasms Abstract Previous studies using current or recent adult body weight and body mass index are inconclusive as to a possible effect of increased body mass on premenopausal breast cancer incidence. Only five studies have presented data on early adult body mass, and no study has reported these data for premenopausal bilateral breast cancer. Because premenopausal bilateral breast cancer is assumed to be partly genetic and partly environmental in origin, it is crucial to identify possible modifiable risk factors for this cancer. We present data on early adult body weight and body mass (Quetelet Index, Q1) from a case-control study of 142 premenopausal bilateral breast cancer cases from Los Angeles County, California, Connecticut, and Quebec, Canada, and 229 sister controls. The odds ratio (and 95% confidence interval) of premenopausal breast cancer adjusted for age, education, alcohol consumption, and oral contraceptive use was 0.7 (0.3-1.4) for women in the highest tertile of QI at age 18. The results do not suggest that elevated body mass index at a young age increases the risk of premenopausal bilateral breast cancer, but lend only weak support to the hypothesis of an inverse association between body mass index and premenopausal breast cancer.
Introduction Data on the association between body weight or body mass and premenopausal breast cancer have been inconsistent. Some studies suggest an increased risk, while other studies suggest a decreased risk Or no effect of body weight or elevated body mass index on premenopausal breast cancer. Most of these studies used weight at diagnosis or study interview, which may be influenced by early disease. Early adult (ages 18-21) body weight or
body mass was used as the exposure variable in five studies [1-6]. In addition to being more readily interpretable than current body mass, this measure is also highly correlated with later adult body mass [7]. Both the cohort study [1-2] and the four casecontrol studies [3-6] found a weak to moderate inverse association between body weight or mass and premenopausal breast cancer, but almost all the confidence intervals covered one. Of the three studies assessing weight or body mass in childhood as a risk factor, two suggested a protective effect of in-
Address for offprints: G. Ursin, Department of PreventiveMedicine,USC Schoolof Medicine,1420San Pablo Street, Los Angeles, CA 90033-9987,USA
76
G Ursin et al.
creased weight or body mass [8-9], while the third suggested an increased risk of premenopausal breast cancer with elevated body mass [4]. If there exist several subgroups of breast cancer with somewhat different etiology, then the effect of environmental risk factors might vary among subgroups. Premenopausal bilateral breast cancer represents a subgroup where the probability of genetic susceptibility is considered high, because of the strong family risk associated with this cancer [1012]. However, genetic modelling research suggests that the breast cancer gene (if it exists), acts with reduced penetrance, and that other factors (genetic and/or environmental) are involved in the same pathway [13]. Identification of modifiable environmental factors associated with increased risk of this cancer could be important for family members of patients with bilateral breast cancer. If elevated body weight or mass at a young age is a risk factor, then this could possibly be a modifiable risk factor. We have earlier described an association between oral contraceptive use and risk of premenopausal bilateral breast cancer from this case-control study [14]. We now present results on the association of premenopausal bilateral breast cancer and body weight and body mass at age 18 and menarche.
Methods
Eligible cases were premenopausal bilateral breast cancer patients with at least one living sister and ascertained in one of three study areas, Connecticut, Los Angeles County, California, and parts of the Province of Quebec. The Connecticut and Los Angeles series were ascertained through existing cancer registries. The Quebec cases were ascertained through hospitals in Montreal and Quebec City that cover over 95% of all breast cancers in these two areas as well as most of Southern Quebec Province. Subjects under 50 years of age at the time of diagnosis with bilateral breast cancer were ascertained. Because of the rarity of the condition, the ascertainment included both incident and prevalent cases over the period of record review, which was 19701989 for Los Angeles, 1975-1989 for Quebec, and 1935-1989 for Connecticut.
For each case, her unaffected sisters were the matched controls. The sister had to be free of breast cancer (uni- and bilateral), and gynecological cancer (other than cervical cancer) at the age at which the case was diagnosed. Using sisters of cases as a control group provides unbiased odds ratio (OR) estimates on the condition that the exposure-specific risk of breast cancer is constant over time [15]. Initially, 1075 cases were ascertained. Of these, the treating physician refused to give permission to contact 102 of these cases. A total of 544 cases were alive, found, and responded to a family history questionnaire. Of these, 264 had at least one sister who was alive in 1989, and we obtained permission to contact the sister(s). A second questionnaire concerning demographic information, anthropometric data, menstrual and reproductive history, hormone use, smoking, alcohol consumption, and a food frequency questionnaire was sent to all cases who had responded to the initial questionnaire and who had given us permission to contact their sisters. We obtained risk factor questionnaires from 180 complete case-control sets with one case and at least one sister (315 sisters total). Among the nonresponders, neither the case nor the sister(s) returned the questionnaires in 38 pairs, only the case returned the questionnaires in 33 pairs, and only the sister(s) returned the questionnaires in 13 pairs. Thirty-eight sets were excluded because there was no pathological confirmation of the. diagnosis (6 sets), the case was postmenopausal at time of diagnosis (24 sets), the case's weight at age 18 was unknown (2 sets), the only sisters had endometrial cancer at an early age (1 set), the only sister had not yet reached the age at which the case was diagnosed (5 sets). Thirty-one additional sisters were excluded because they had not reached the age at which the case was diagnosed with breast cancer (n = 24) or because they were diagnosed with bilateral breast cancer (n = 3) or an unknown gynecological cancer (n = 3) before they reached the age at which the case was diagnosed with breast cancer, or their weight at age i8 was unknown (n = 1). Sisters who were diagnosed with postmenopausal breast cancer at a later age were included in the analyses as controls. This yielded 142 cases and 229 sister controls.
Early body mass index and breast cancer
Of these case-control sets 51 were from Los Angeles, 61 from Connecticut, and 30 from Quebec. In the questionnaire, subjects were asked to report their weight at age 18, as well as current weight and height. The subgroup from Los Angeles were also asked to report weight and height at time of menarche. Since current weight was likely to be affected by the disease for many cases whose onset was several years earlier, the current analyses focus on weights at menarche and age 18. Quetelet Index (QI), i.e. weight divided by height squared (kg/m2), was used as a measure of body mass. Subjects also provided information on education, alcohol intake, menstrual and reproductive history, and oral contraceptive use. Estimates of odds ratios (with 95% confidence intervals) were derived from conditional logistic regression analyses with variable ratio matching, incorporating relevant covariates such as age in the model, using the software package EGRET. Age, education, and alcohol and oral contraceptive use have been associated with QI [16-18], although neither age, education nor alcohol intake were strong risk factors in our data (Haile et al., submitted). In our study oral contraceptive use for one year or more was associated with an OR of 1.7 [14]. Height is associated with body weight, and we adjusted for it in the analyses of body weight. Age at first full term pregnancy was only a weak risk factor for premenopausal bilateral breast cancer in our data (Haile et al., submitted). Age at first full term pregnancy has been associated with QI at a later age [16], but not, as far as we know, with QI at age 18. Further, controlling for this variable and parity did not change the results. Therefore we are presenting results adjusted for age, education, alcohol intake, and oral contraceptive use.
Results
Over ninety percent of the women in this study were white. The median age at the time of the cases' diagnosis was 41 for both cases and controls (sisters). Median year of diagnosis was 1976. Table 1 shows matched odds ratios for current height, weight, and QI at age 18 and at menarche,
77
analyzed as continuous variables. None of the variables was a strong risk factor for premenopausal bilateral breast cancer. The matched odds ratios for the association between body weight, QI at age 18 and at menarche and premenopausal bilateral breast cancer are displayed in Table 2. When body weight was divided into quartiles, the highest OR (1.3) was observed for weight of 50-54.4 kg, but the 95% CI included 1.0. For tertiles of body weight, the OR of premenopausal bilateral breast cancer decreased with increasing weight, but this was not statistically significant (p for trend = 0.28). Women in the upper tertile had an odds ratio of 0.7 (0.3-1.4). The strongest positive association between QI at age 18 and breast cancer was found in the lowermiddle QI group (QI of 19-20). Depending on the categories and cutpoints, the ORs tended to be less than 1.0 for categories of high body weight or QI. The OR of premenopausal bilateral breast cancer was 0.9 (0.4-2.0) for the highest quartile and 0.7 (0.3-1.4) for the highest tertile. Because we had low power for some of the analyses, only results using tertiles of body weight or Q1 are described in the following. As with body weight and QI at age 18, increasing tertiles of body weight or QI at menarche was assoTable 1. Odds ratios (OR) and 95 % confidence intervals (CI) for the association between height, body weight, Quetelet Index (QI), and premenopausal bilateral breast cancer, using continuous variables
Time of measurement
Cases N
Current height (cm) 142 At age 18 weight (kg) 142 QI (weight/ height 2) 142 At menarche weight 47 OI 47
Controls Unadj. Adjusted N OR 1 OR 2 95% CI
229
1.023
0.91
0.64-1.30
229
0.99
0.99
0.95-1.02
229
0.97
0.98
0.89-1.08
67 67
0.98 0.92
0.99 0.96
0.96-1.02 0.81-1.12
1 OR from conditional logistic regression. 2 OR from conditional logistic regression, adjusted for age, age squared, education, alcohol intake, and oral contraceptive use. 3 OR associated with one unit increase.
78
G Ursin et al.
Table 2. Unadjusted and adjusted odds ratios (OR) and 95% confidence intervals (CI) for the association between Quetelet Index (QI), body weight at age 18 and at menarche, and premenopausal bilateral breast cancer
Cases Controls Unadj. Adj. N
N
OR I
95% CI
OR s
Body weight at age 18, quartiles (kg) 3
(p = 0.48) 4 1.05 1.0s 1.8 1.3 1.1 0.8 1.0 0.8 Body weight at age 18, tertiles (kg) 3 (p = 0.28) _<51 48 71 1.0 1.0 51.1-57 53 82 0.9 0.9 > 57 41 76 0.7 0.7 QI at age 18, quartiles (p -- 0.43) < 18.9 32 56 1.0 1.0 19-20.3 44 49 1.7 1.5 20.4-21.8 35 61 1.1 1.0 _>21.9 31 63 0.9 0.9 QI at age 18, tertiles (p = 0.14) < 19.5 50 70 1.0 1.0 19.6-21.1 52 73 1.0 0.9 >_21.2 40 86 0.6 0.7 Body weight at menarche, tertiles (kg) 3'6 (p = 0.14) < 43.5 21 21 1.0 1.0 43.5-50 16 36 0.4 0.4 > 50 13 25 0.4 0.6 QI at menarche, tertiles 6 (p = 0.11) < 17.8 21 22 1.0 1.0 17.9-19.8 14 31 0.6 0.8 _>19.9 15 28 0.4 0.5 < 50 50-54.4 54.5-59 > 59
29 42 36 35
54 46 63 66
0.6-2.9 0.3-1.8 0.3-1.9
0.5-1.8 0.3-1.4
0.7-3.2 0.5-2.2 0.4-2.0
0.5-1.7 0.4-1.4
0.1-1.5 0.1-2.2
0.3-2.4 0.1-1.7
Weight increase from menarche to age 18, tertiles (kg) 3'6'7 -< 5 13 30 1.0 5.1-10 18 28 1.5 > 10 19 24 1.7 QI increase from menarche to age 18,
(p = 0.40) 1.0 1.4 1.3
0.4-4.5 0.3-5.2
tertiles 6'7
(p = 0.45) 1.0 2.0 1.6
0.6-7.0 0.4-6.1
_<0.3 0.4-2.3 > 2.3
14 16 20
28 24 26
1.0 1.4 1.5
O R from conditional logistic regression. O R from conditional logistic regression, adjusted for age, age
squared, education, alcohol intake, and oral contraceptive use. 3 In addition adjusted for current height or height at menarche. 4 All p-values are for Mantel test for trend.
s Reference value. 6 Not adjusted for age squared. 7 In addition adjusted for weight or QI at menarche.
ciated with ORs less than one, but the confidence intervals were wide. Increase in weight or QI from menarche to age 18 was, however, associated with odds ratios greater than one, but again the confi•dence intervals were wide. As can be seen from Table 2, adjusting for age, education, and alcohol and oral contraceptive use did not change the results much. When age at first full term pregnancy and parity were added to the model, the results were also essentially unchanged. This study did not have the power to detect significant effect modification. However, in order to address issues such as detection bias or survival bias, as well as a possible role of anovulatory cycles, several subgroup analyses were performed. The results of these exploratory analyses are displayed in Table 3. N o n e of these interactions were statistically significant, and the confidence intervals around all the interaction parameters were wide (results not shown). No strong increased risks associated with elevated QI were observed in any of these subgroups of women, all confidence intervals covered one, and most were wide. In subgroups of women where cancer cases had been diagnosed at an early stage (defined by small tumor size or no lymph node metastasis), ORs less than one were observed for the highest tertile of QI, while this was not observed in subgroups where the cancer cases had been diagnosed with late stage disease. However, as the confidence intervals for early and late stage disease both were wide, no firm conclusions could be drawn. Similarly, elevated QI was associated with a point estimate less than one only in women who did not have a regular menstrual period within one year after menarche, but again the confidence intervals were wide. The O R estimates did not differ between sets where the case was diagnosed within 4 years of data collection compared to those where the case was diagnosed 5 or more years prior to data collection. Finally, the decreasing ORs with increasing tertile of QI were only observed in sets where the case did not have a first degree history of breast cancer. The results for body weight at age 18 were similar to those of QI (data not shown). QI and body weight at menarche gave similar results to those of QI and body weight at age 18, but the confidence intervals were wider.
Early body mass index and breast cancer Discussion This study does not suggest an increased risk of elevated body weight or mass on premenopausal bilat-
79
eral breast cancer, but the results do not provide strong support for a protective effect either. Our findings are similar to those reported for early adult body mass index and premenopausal breast cancer
Table 3. Odds ratios (OR) and 95% confidence intervals (CI) for the association between tertiles of Quetelet Index (QI) at age 18 and p r e m e n o p a u s a l bilateral breast cancer, by levels of selected covariates Covariate
Level
Tertile of QI 1
Tumor size
< 2 cm
1 2 3
18 15 11
1 2 3
Tumor size
L y m p h node metastasis 3
L y m p h node metastasis 3
Menstrual period
> 2 cm
none
present
regular
Cases N
Controls N
OR 2
95% CI
24 13 29
1.0 1.1 0.3
0.3-3.9 0.1-1.2
8 9 8
12 11 12
1.0 1.1 1.2
0.2-7.3 0.2-7.2
1 2 3
26 29 13
33 35 33
1.0 1,2 0.6
0.5-2.8 0.2-1.5
1 2 3
9 13 17
17 14 24
1.0 1.3 1.2
0.3-5.8 0.3-5.2
1 2 3
44 45 34
52 67 74
1.0 0.9 1.1
0.4-1.8 0.5-2.4
Menstrual period
regular
cont. 4
123
193
1.1
1.0-1.2
Menstrual period
irregular
cont.
18
36
0.65
0.1-2.7
Time since diagnosis
_<4 years
cont.
13
20
1.0
0.6-1.8
Time since diagnosis
> 4 years
cont.
129
209
1.0
0.9-1.1
First degree family history of breast cancer
Yes
1 2 3
14 15 11
22 25 24
1.0 0.8 0.9
0.3-2.6 0.3-3.1
1 2 3
36 37 29
48 48 62
1.0 0.8 0.6
0.4-1.7 0.2-1.3
1 2 3
5 10 8
10 9 17
1.0 2.4 1.]
0.4-13.6 0.2-6.0
1 2 3
45 42 32
60 64 69
1.0 0.9 0.7
0.4-1.7 0.3-1.4
First degree family history of breast cancer
Maternal history of breast cancer
Maternal history of breast cancer
No
Yes
No
Tertile values of Quetelet Index as in Table 1, _<19.5, 19.6-21.1, > 2/.1. 2 O R from conditional logistic regression, adjusted for age, age squared, education, average alcohol intake, and oral contraceptive use. 3 Only adjusted for age, age squared. 4 O R per unit increase. 5 Crude odds ratios.
80
G Ursin et aL
overall [1-6]. For instance, in the Nurses' Health Study Cohort, relative risks (and 95% C.I.) of 0.7 (0.6-1.0), 0.7 (0.6-1.0), and 0.6 (0.5-0.8) were reported for the three quintiles with QI above 20 kg/m2 (21-22.0 kg/m2, 23-24.9 kg/m2, and > 25 kg/m2 respectively), compared to the quintile with QI less than 20 kg/m2 at age 18 [2]. All four case-control studies found an odds ratio less than one for the uppermost category, although in two of these studies the confidence intervals covered one [4, 5], and in a third study, the CASH study [6], the upper category was much higher (>__30 kg/m3) than in our study. Paffenbarger et al. [3], however, reported a significant OR of 0.65 for QI >_22.0 compared to QI < 19 at age 20. All the above-mentioned studies included all breast cancers. Because of the rarity of bilateral breast cancer, however, only a small proportion of the cases in each of these studies was likely to be bilateral. The fact that our findings are similar to those in these studies suggest that the etiology of premenopausal bilateral breast cancer may be similar to that of unilateral cancer. In our study, ORs below unity associated with the highest tertile of QI were only observed in subgroups of women where the case had been diagnosed with early stage disease, suggesting some detection and/or survival bias. The two cohort studies that addressed the issue, however, found detection bias an unlikely explanation for the inverse association they observed between body mass and premenopausal breast cancer [1-2, 19]. The combination of mammography and clinical breast exams by physicians has been estimated to advance the diagnosis of breast cancer by approximately 1.7 years [20]. However, the two cohort studies found that the diagnosis would have to be advanced by more than 2 years, and possibly as much as 5 years [19], for this to explain the inverse association in these studies. Detection bias is probably not as likely in our study as in other observational studies, since all the controls in our study have a family history of breast cancer, and are probably more likely to have their breasts examined regularly than other women in the population. Survival tends to be poorer in obese than in slim breast cancer patients overall [21-23], but it is controversial how strong a prognostic factor obesity is
among premenopausal women [22, 24-27]. Although the ORs for breast cancer did not differ between sets diagnosed 5 or more years before and those diagnosed within 4 years of data collection, the confidence intervals for both estimates were wide. Further, as we examined only living cases and controls, and included prevalent as well as incident cases, we cannot exclude the possibility that the observed ORs below unity may be due to survival bias. Theoretically, obesity should be associated with increased risk of postmenopausal breast cancer, but possibly with a reduced risk of premenopausal breast cancer. In postmenopausal women, obesity is associated with increased conversion of androstenedione to estrogens (mainly estrone) [28]. The increased bioavailability of estrogen probably increases the risk of postmenopausal breast cancer [27]. In premenopausal women, the importance of this extraglandular conversion of androstenedione to estrone as an estrogen source has been questioned [30], and it has been suggested that the excess androgens and estrogens produced through this conversion may interfere with normal cyclicity by a complex mechanism [31]. Sherman et al. reported that women who were heavier at age 18 had greater menstrual cycle variability and length than leaner women [32], indicating more frequent anovulatory cycles in the heavy women. Since anovulatory cycles are believed to be associated with decreased risk of breast cancer [33], it has been hypothesized that elevated body mass index has a protective effect on premenopausal breast cancer through more frequent anovulatory cycles [29]. However, the level of obesity necessary to induce anovulatory cycles to such a degree that it reduces women's risk of breast cancer is unclear. Interestingly, limiting the analyses in our study to women reporting regular menstrual periods within a year after menarche yielded no association between body weight/body mass index and breast cancer. While varying degrees of detection and survival bias could explain the discrepancy of results among studies of body weight/body mass and premenopausal breast cancer, the use of widely different category cutpoints and reference values could also explain the inconsistencies if the association between body mass/body weight and log odds ratio of breast
Early body mass index and breast cancer cancer is not linear on the full range of exposure. The women in our study were slimmer than those in the other studies [2-3, 5-6], except for the study from Utah [4]. For instance, in the Nurses' Health Study [2], 14.6% had a QI above 25; in our study only 5% had a QI that high. Although premenopausal bilateral breast cancer is more likely to be hereditary than unilateral breast cancer, both genetic and environmental factors are probably involved in the same pathway [13]. In our subgroup analyses, the inverse association with QI was observed only for sets where the case had no family history, although the coefficient for the interaction term had a wide confidence interval and was not statistically significant. This could suggest that the effect of body mass may be limited to those women who are less likely to carry the breast cancer gene(s). In conclusion, this study does not suggest an increased risk of elevated body weight or body mass index on premenopausal bilateral breast cancer, and it lends only weak support to the hypothesis of a protective effect of elevated body mass. The ORs we observed that were less than one may be due to detection and/or survival bias, although we cannot exclude the possibility that they may also reflect a protective effect of anovulatory cycles associated with obesity. However, the associations between anovulatory cycles, obesity and premenopausal (uni- and bilateral) breast cancer are not well understood, and require further research.
Acknowledgements This publication was supported by grant number CA 36386 from the National Cancer Institute. G. Ursin was a research fellow of the Norwegian Research Council for Science and the Humanities (NAVF) during the time of this study.
References Willett WC, Browne ML, Bain C et aL: Relative weight and risk of breast cancer among premenopausal women. Am J Epidemio1122: 731-740, 1985
81
2. London SJ, Colditz GA, Stampfer MJ, Willett WC, Rosner B, Speizer F: Prospective study of relative weight, height, and risk of breast cancer. JAMA 262: 2853-2858, 1989 3. Paffenbarger RS, Kampert JB, Chang H-G: Characteristics that predict risk of breast cancer before and after menopause. Am J Epidemio1112: 258-268, 1980 4. Pryor M, Slattery ML, Robison LM, Egger M: Adolescent diet and breast cancer in Utah. Cancer Res 49: 2161-2167, 1989 5. Lund E, Adami H-O, BergstrCm R, Meirik O: Anthropometric measures and breast cancer in young women. Cancer Causes and Control 1: 169-172, 1990 6. Chu SY, Lee NC, Wingo PA, Senie RT, Greenberg RS, Peterson HB: The relationship between body mass and breast cancer among women enrolled in the cancer and steroid hormone study. J Clin Epidemiol 44: 1197-1206, 1991 7. Casey VA, Dwyer JT, Coleman KA, Valadian I: Body mass index from childhood to middle age: a 50-y follow-up. Am J Clin Nutr 56: 14-18, 1992 8. Le Marchand L, Kolonel LN, Earle ME, Mi M-P: Body size at different periods of life and breast cancer risk. Am J Epidemio1128: 137-152, 1988 9. Hislop TG, Coldman AJ, Elwood JM, Brauer G, Kan L: Childhood and recent eating patterns and risk of breast cancer. Cancer Detect Prev 9: 47-58, 1986 10. Khoury MJ, Beaty TH, Liang K-Y: Can familial aggregation of disease be explained by familial aggregation of environmental risk factors? Am J Epidemio1127: 674-683, 1988 11. Anderson DE: Some characteristics of familial breast cancer. Cancer 28: 1500-1504, 1971 12. Anderson DE: Genetic study of breast cancer: Identification of a high risk group. Cancer 34: 1090-1097, 1974 13. Newman B, Austin MA, Lee M, King M-C: Inheritance of human breast cancer: Evidence for autosomal dominant transmission in high-risk families. Proc Natl Acad Sci 85: 3044-3048, 1988 14. Ursin G, Aragaki C, Paganini-Hill A, Siemiatycki J, Thompson WD, Halle RW: Oral contraceptives and premenopausal bilateral breast cancer, a case-control study. Epidemiology 3: 414-419, 1992 15. Goldstein AM, Hodge SE, Haile RWC: Selection bias in case-control studies using relatives as the controls. Int J Epidemio118: 985-989, 1989 16. Swanson CA, Jones DY, Schatzkin A, Brinton LA, Ziegler RG: Breast cancer risk assessed by anthropometry in the NHANES I Epidemiological Follow-up Study. Cancer Res 48: 5363-5367, 1988 17. Sobal J, Stunkard A J: Socioeconomic status and obesity: A review of the literature. Psycho1105: 260-275, 1989 18. Wallace RB, Heiss G, Burrows B, Graves K: Contrasting diet and body mass among users and nonusers of oral contraceptives and exogenous estrogens: The Lipid Research Clinics Program Prevalence Study. Am J Epidemiol 125: 854-859,1987 19. Tretli S: Height and weight in relation to breast cancer mor-
82
20.
21.
22.
23.
24.
25.
•
26.
G Ursin et al. bidity and mortality. A prospective study of 570,000 women in Norway. Int J Cancer 44: 23-30, 1989 Walter SD, Day NE: Estimation of the duration of a preclinical disease state using screening data. Am J Epidemio1118: 865-886, 1983 Boyd NF, Campbell JE, Germanson T, Thomson DB, Sutherland D J, Meakin JW: Body weight and prognosis in breast cancer. J Natl Cancer Inst 67: 785-789,1981 Tretli S, Haldorsen T, Ottestad L: The effect of premorbid height and weight on the survival of breast cancer patients. Brit J Cancer 62: 299-303, 1990 Senie RT, Rosen PP, Rhodes P, Lesser ML, Kinne DW: Obesity at diagnosis of breast carcinoma influences duration of disease-free survival. Ann Int Med 116: 26-32, 1992 Goodwin PJ, Boyd NF: Body size and breast cancer prognosis: a critical review of the evidence. Breast Cancer Res Treat 16: 205-214, 1990 Ewertz M, Gillanders S, Meyer L, Zedeler K: Survival of breast cancer patients in relation to factors which affect the risk of developing breast cancer. Int J Cancer 49: 526-530, 1991 Schapira DV, Kumar NB, Lyman GH, Cox CE: Obesity and body fat distribution and breast cancer prognosis. Cancer 67: 523-528, 1991
27. Mohle-Boetani JC, Grosser S, Whittemore AS, Malec M, Kampert JB, Paffenbarger RS: Body size, reproductive factors and breast cancer survival. Prev Med 17: 634-642, 1988 28. Grodin JM, Siiteri PK, MacDonald PC: Source of estrogen production in premenopausal women. J Clin Endocrin Metab 36: 207-214, 1973 29. Pike MC: Reducing cancer risk in women through lifestylemediated changes in hormone levels. Cancer Detection and Prevention 14: 595-607, 1990 30. Zumoff B: Relationship of obesity to blood estrogens. Cancer Res 42(suppl): 3289s-3294s, 1982 31. Yen SSC: Chronic anovulation caused by peripheral endocrine disorders. In: Yen SSC, Jaffe RB (eds) Reproductive Endocrinology; Physiology, Pathophysiology and Clinical Management. Second edition. W.B. Saunders, Philadelphia 1986 32. Sherman B, Wallace R, Bean J, Schlabaugh L: Relationship of body weight to menarcheal and menopausal age: Implications for breast cancer risk. J Clin Endocrin Metab 52: 488493, 1981 33. Henderson BE, Ross RK, Judd HL, Krailo MD, Pike MC: Do regular ovulatory cycles increase breast cancer risk? Cancer 56: 1206-1208, 1985