Incidence rates of breast cancer are very high in Western societies, where they have increased dramatically since the beginning of the twentieth century (IARC, 1992). There is substantial epidemiological evidence that these high rates may be due to lifestyle factors, including an energy-dense diet, low levels of physical activity, and childbearing patterns characterized by relatively late first full-term pregnancies and low parity (Kaaks, 1996).
Lifestyle-related variations in breast cancer risk are believed to be largely mediated by alterations in hormonal profiles. Hormones that have been implicated most in breast cancer development are oestrogens and progestogens (Bernstein and Ross, 1993). More recently, however, it has been proposed that insulin resistance may also be a risk factor for breast cancer (Stoll, 1996). Insulin increases the biological activity of insulin-like growth factor I (IGF-I) (Kaaks, 1996) by enhancing IGF-I synthesis, and by downregulating some of the IGF-binding proteins (IGFBP-1, IGFBP-2). Both insulin and IGF-I can promote tumour development by stimulating cell proliferation, while inhibiting apoptosis (Kaaks, 1996). In addition both insulin and IGF-I decrease plasma levels of sex hormone-binding globulin (SHBP) (Folsom et al., 1990) and stimulate the ovarian and adrenal synthesis of sex steroids, notably androgens (Poretsky, 1991;Kaaks, 1996), which would increase the levels of bioavailable oestrogens (Kaaks, 1996). A potential association between insulin resistance and breast cancer is supported by the fact that obesity is both a major cause of reduced insulin sensitivity and a well-documented risk factor for breast cancer after menopause (Hunter and Willett, 1993;Smith, 1994;Stoll, 1996). In addition, two case–control studies have demonstrated a positive association between circulating insulin levels and breast cancer risk (Bruning et al., 1992;del Giudice et al., 1998).
Insulin resistance and hyperinsulinaemia have, apart from obesity, been strongly related to hypertension, increased levels of plasma triglycerides, changed cholesterol levels and impaired glucose tolerance. Collectively, this constellation has been referred to as the ‘metabolic syndrome’, ‘insulin resistance syndrome’ or ‘syndrome X’ (Ferrannini, 1993;Kaaks, 1996;Meigs et al., 1997). This condition has repeatedly been related to increased cardiovascular mortality and an elevated risk of developing diabetes (Meigs et al., 1997).
The aim of this prospective cohort study was to investigate the hypothesis that markers of hyperinsulinaemia are associated with increased risk of breast cancer in pre- and peri/postmenopausal women.
Materials and methods
The Malmö Preventive Project (MPP) was initiated in 1974, as a preventive health project for early detection of individuals at increased risks for cardiovascular disease. The programme recruited individuals living in Malmö, inviting entire birth-cohorts identified through the local population registry to a health examination (Berglund et al., 1996). In all, 10 902 women born between 1926 and 1949 were screened between 1977 and 1992. The participation rate in the MPP was about 70%.
A self-administered questionnaire was used to obtain a comprehensive information about lifestyle, medical history, occupation, level of education and use of medications (Berglund et al., 1996). The questionnaire was revised several times. Information on reproductive variables – age at menarche (before or after 12 years of age), nulliparity, current oral contraceptive use (yes/no) and current hormonal replacement therapy (HRT) (yes/no) – was included in the questionnaire for the 8161 women examined from April 1983 and onwards. The definition of menopausal status was made a priori. Women were classified as peri- or postmenopausal: (1) if they affirmed (questionnaire information) that their menstruations had ceased, that they had menopausal symptoms or that they were taking any ‘female hormonal medication’ because of such symptoms; or (2) if questionnaire information was missing and they were above 58.8 years of age at baseline, which is the reported mean age at menopause, plus 1.96 standard deviations (SD), in comparable populations (Danforth and Scott, 1986). Women were considered premenopausal: (1) if they answered ‘no’ to the questions above; or (2) if questionnaire information was missing and they were below 44 years of age (mean age at menopause minus 1.96 SD). This left 1164 women where menstrual status had not been assessed by the questionnaire, and who were between 44 and 58.8 years old; these women were excluded from further analyses. Hence, this study was based on 9738 women, out of whom 3873 were premenopausal at baseline and 5865 were peri/postmenopausal. Mean age at baseline was 49.6 (SD 7.8) years for all women, for premenopausal 42.8 (SD 7.9) years and for peri/postmenopausal women 54.1 (SD 3.0) years.
Height was measured to the nearest centimetre and weight was recorded at intervals of 0.1 kg. Body mass index (BMI) was calculated as weight/height 2 (kg/m 2 ). Blood pressure (BP) was assessed supine in the right arm after a 10-min rest, and was noted to the nearest 5 mmHg. Heart rate per minute was registered in even numbers.
In the morning, after an overnight fast, all subjects gave a blood sample, which was immediately analysed in respect of some standard parameters. Serum cholesterol and triglycerides were assessed by an enzymatic method routinely used by the hospital clinical biochemistry laboratory in Malmö (Fernlund et al., 1991). Blood glucose was analysed with a hexokinase method (Fernlund et al., 1991). The MPP invited women over a period of 15 years. The priorities changed from time to time and some information was not collected in all subjects. One such parameter was the assessment of glucose metabolism by means of an oral glucose tolerance test (OGTT). This test was performed according to one of two different ways of administering glucose, and only in non-diabetic subjects. A 120-min blood glucose value in the OGTT was available in 3070 women, mean age at baseline: 54.6 (SD 4.2) years, who had been given a glucose load of 30 g/m 2 body surface and in 1715 women, mean age 54.2 (SD 2.7) years, all tested after October 1986, who were administered a standard dose of 75 g, as recommended by the WHO (WHO, 1980). An OGTT had been performed in 4090 out of 5865 peri/postmenopausal women, and in 695 out of 3873 premenopausal subjects. Because of the small percentage of premenopausal women for whom OGTT data were available, the association between 120-min glucose and breast cancer risk was only investigated in peri/postmenopausal women.
The present analysis is based on follow-up of breast cancer incidence until 31 December 1997. Each woman was followed until that date, or until she got breast cancer, or died. The national Swedish Causes of Death Registry was used to establish date and cause of death. This registry was updated until 31 December 1996. Women who were alive at this point have been counted in the statistical analyses as alive at 31 December 1997, which was the ultimate follow-up date. The average time from baseline to diagnosis in premenopausal women was 9.6 (SD 5.0) years and in peri/postmenopausal women 6.6 (SD 3.4) years.
Follow-up on cancer and statistical analyses was restricted to invasive breast cancer. Three databases were used for case retrieval. The Swedish Cancer Registry was used for identification of cases of invasive breast cancer diagnosed before 31 December 1995; in all, there were 409 cases among the 9738 women in the cohort. Sixty-six additional cases occurring after 31 December 1995 and before 31 December 1997 were identified by a registry at the Department of Pathology at Malmö University Hospital. Finally, one more case was identified through the Regional Cancer Registry. Out of this total of 476 cases, 150 had been diagnosed prior to the health examination, and these prevalent cases were excluded from further analysis. Hence there were in all 326 cases of incident invasive breast carcinoma during an average follow-up period of 13.1 years. Their mean age at diagnosis was 57.8 (SD 6.2) years. Of these, 112 occurred in women who were classified as premenopausal at baseline, mean age at diagnosis 52.6 (SD 5.8) years, and 157 in peri/postmenopausal women, mean age 60.5 (SD 4.1) years. The TNM system was used for classification of stage at diagnosis (American Joint Committee on Cancer, 1988). Up until the end of 1991 this information was available from a research registry set up at the Department of Surgery, Malmö University Hospital (Garne, 1996). For the period 1992–1997, data on tumour stage were collected from clinical notes in the hospital records.
Cox's proportional hazards analysis was used to estimate relative risks (RR), with 95% confidence intervals (CI). Obesity is associated with an increased breast cancer risk in postmenopausal women, but, if anything, with a decreased risk in premenopausal women (Hunter and Willett, 1993;Ursin et al., 1995). This indicates that the associations between anthropometric/metabolic factors and breast cancer incidence may be modified by menopausal status. Because of this, all analyses were performed separately for women who were pre- and peri/postmenopausal at baseline. Relative risks were estimated for quartiles of anthropometric indices, blood pressure, pulse, blood lipids and fasting glucose. Quartile cut-off points were defined on cases, so as to optimize the distribution of cases over the four exposure categories (there are many more non-cases in the cohort) and to maximize the statistical power of tests for association of disease risk with exposure levels. The cut-off points were determined separately for premenopausal women, and peri/postmenopausal women. For the 120-min glucose value in the OGTT, quartile cut-off points were defined on peri/postmenopausal cases plus non-cases together, and separately for the two types of OGTT. It was assumed that either type of OGTT would result in the same expected relative ranking of individuals by quartile levels, even though absolute glucose levels at 120 min were slightly higher in the method with a fixed dose of 75 g of glucose. Tests for linear trend were performed with the mean of the quartiles as weighting factor, except for glucose at 120 min, where instead the quartiles were given scores of 1, 2, 3 and 4.
The analyses were adjusted for age, age at menarche, nulliparity, current oral contraceptive use and current HRT (Kelsey, 1993;Collaborative Group on Hormonal Factors in Breast Cancer, 1997) and some lifestyle variables, such as smoking (never/current/ex) (Ambrosone et al., 1996) and alcohol consumption (yes/no) (Smith-Warner et al., 1998). In a final model, the variables height and weight were also added (Kelsey, 1993). In order to maximize the number of subjects used to analyse the factors of interest in this study, missing values for the reproductive factors included in the model were coded as a separate category.
To investigate if the relationship between breast cancer risk and anthropometric and metabolic factors varied for women from different ages, the analyses were repeated for different age-bands, (i.e. using the numbers of incident cases and the accumulated numbers of person-years either before, or after, a given age limit).
Insulin levels related to anthropometry, blood pressure and metabolic factors
Additional measurements related to the OGTT were assessed in a subsample of women, mainly during the two periods 1977–1978 and 1989–1990. Fasting plasma insulin was measured in 269 women, mean age at baseline 42.8 (SD 10.1) years (after exclusion of women with undefined menopausal status), and 196 women, mean age 45.2 (SD 10.5) years, had information on insulin levels at 120 min in the OGTT. This allowed an estimation of associations of selected indicators of the metabolic syndrome with plasma insulin levels. These associations were evaluated by calculation of Spearman's correlation coefficients. In the 196 subjects where insulin at 120 min was recorded, 193 had been given a glucose load of 30 g/m 2 body surface.
The incidence of breast cancer in the whole material was 226 per 100 000 person-years. Among women classified as premenopausal at baseline it was 193 per 100 000 person-years, in peri/postmenopausal women the incidence was 258 per 100 000 person-years (Table 1
Among premenopausal women, there was no association between breast cancer risk during follow-up and height, weight or BMI (Table 2
). Peri/postmenopausal women had an increased age-adjusted risk for the tallest quartile as compared with the shortest (RR 1.78, 95% CI 1.14–2.77), with a significant trend over quartiles (Table 2), but also in these women, breast cancer risk was not associated with weight or BMI. Blood pressure or pulse showed no association with risk of breast cancer in premenopausal or peri/postmenopausal women (Table 3
). Total cholesterol, triglycerides and fasting glucose did not affect the relative risk of breast cancer in women who were premenopausal at baseline (Table 4
). In peri/postmenopausal women, the relative risk of breast cancer was increased over quartiles of cholesterol (P -value for trend = 0.05) (Table 4). In these women, however, the levels of triglycerides, fasting glucose and glucose at 120 min were not related to different breast cancer risk.
Adjustments in the Cox's regression analysis, in addition to age, for age at menarche, nulliparity, current oral contraceptive use and current HRT, smoking, alcohol consumption, height and weight did not change any of these results. The analyses were repeated in different age-bands focusing on incident cases and person-years of follow-up either before age 55 (in the premenopausal subcohort), or on cases and person years after age 55 (in the peri/postmenopausal subcohort). However, also in the age-band analyses no statistically significant associations were found.
Insulin levels related to anthropometry, blood pressure and metabolic factors
In women who were premenopausal at baseline, all correlations with insulin levels were below 0.2, except measurements of plasma insulin and glucose at 120 min in the oral glucose tolerance test (r = 0.70) (Table 5
). In peri/postmenopausal women, significant correlations were seen between plasma insulin and glucose at 120 min (r = 0.39), and between fasting insulin and BMI (r = 0.52), triglycerides (r = 0.35) and fasting glucose levels (r = 0.35).
We investigated the relation between markers of the metabolic syndrome and risk of breast cancer in a prospective cohort study in Malmö, southern Sweden. Apart from weak associations of breast cancer risk with height and total serum cholesterol in peri/postmenopausal women, no significant differences in breast cancer risk were found for women from different quartiles of body mass index, blood pressure, serum triglycerides or blood glucose levels.
The lack of association of breast cancer risk with BMI contrasts with the findings from previous studies. Taken together, the evidence from previous studies shows a positive association of obesity with breast cancer risk in postmenopausal women. In prospective studies, however, this positive association generally was less apparent, and tended to be significant only for the risk of breast cancers diagnosed well after menopause (i.e. after age 60–65) (Hunter and Willett, 1993). By contrast, studies on breast cancer in premenopausal women have generally shown no strong association of risk with obesity, as in our study, or an inverse association (Ursin et al., 1995). Besides obesity (BMI), body height has repeatedly been shown to be associated with increased breast cancer risk, both before and after menopause (Hunter and Willett, 1993).
As for the other metabolic factors, blood pressure, blood lipids and plasma glucose, most previous studies have also shown only weak or inconsistent associations with breast cancer risk. Some studies have shown a positive association of breast cancer risk with blood pressure (de Waard et al., 1960;Tornberg et al., 1988), whereas other have demonstrated no association at all (Adami and Rimsten, 1978;Franceschi et al., 1990Land et al., 1994;Moseson et al., 1997) or even a negative relationship in some subgroups (Thompson et al., 1989). Only one of these studies was a prospective cohort study, in which blood pressure was measured directly (Tornberg et al., 1988). One prospective cohort study on pulse rate and breast cancer risk (Steenland et al., 1995) showed no significant association. Regarding blood lipids, some studies have demonstrated an increased risk of breast cancer in women with low plasma cholesterol (Tornberg et al., 1988Vatten and Foss, 1990;Tulinius et al., 1997) or high plasma triglycerides (Goodwin et al., 1997), at least in some subgroups, but most studies have found no overall association (Tornberg et al., 1988Vatten and Foss, 1990;Hoyer and Engholm, 1992;Gaard et al., 1994Steenland et al., 1995). High-density lipoprotein cholesterol level has been related to both increased (Moorman et al., 1998) and decreased (Hoyer and Engholm, 1992;Moorman et al., 1998) risk. Finally, three early case–control studies, all conducted before 1974, showed a positive association of breast cancer risk with measures of glucose intolerance (Glicksman et al., 1956;de Waard et al., 1960;Muck et al., 1975).
This study, making use of existing data and biochemical measurements, was motivated by the relatively recent theory that breast cancer risk may be directly related to chronic hyperinsulinaemia (Kaaks, 1996). In a subgroup of this cohort study, however, we found that most of the investigated parameters (blood glucose levels, serum triglycerides and increased blood pressure), although generally related to the insulin resistance syndrome, were only weakly correlated to plasma insulin levels. This suggests that measurements of blood glucose, triglycerides and blood pressure may not be the strongest markers of the insulin resistance syndrome, and hence may not be the ideal markers to study a possible relationship of this syndrome with breast cancer risk. Nevertheless, as this subgroup was considerably younger than the rest of the cohort, the estimated correlations of blood glucose and blood pressure with insulin levels may not have been entirely representative for such correlations in the rest of the cohort, and especially for the older subjects. Factors other than insulin levels that would have been valuable to study are insulin growth factors (IGF) and IGF-binding proteins (IGFBP). However, we only used prospectively collected information and IGF and IGFBP had not been assessed at baseline. Unfortunately, it would be difficult to collect such data (e.g. in a nested case–control study), as many of the participants have since died and a long time has elapsed since diagnosis in cases still alive.
It seems unlikely that lack of significant association of breast cancer risk with obesity and obesity-related metabolic factors was due to confounding by unknown risk factors, as adjustments for height, weight or other potential confounding factors (age at menarche, parity, use of oral contraceptives, use of hormone replacement therapy, smoking, and alcohol consumption) did not change any of our findings. Furthermore, it is unlikely that associations of breast cancer risk with obesity or related metabolic parameters have been obscured by effects of essential statistical interaction by subgroups, as appears to be the case for instance for obesity, which is positively associated with breast cancer risk in elderly women, while showing a weak inverse association with breast cancer risk before menopause. Our examinations of associations by different age-bands showed no evidence whatsoever for such effect modification by age or menopausal status (at baseline), for the associations of breast cancer risk with markers of the insulin resistance syndrome.
One factor that may have caused a bias, however, was a possible relative underdiagnosis of breast tumours in obese women, who had more advanced cancers at diagnosis. In the highest BMI quartile, 48% of the women had a stage II, III or IV tumour at diagnosis, as compared with only 24% in women belonging to the lowest BMI quartile, and this difference was seen in both pre- and peri/postmenopausal women. A delay in diagnosis in obese women may be due to the increased difficulty of detecting small tumours by palpation/self-examination. Alternatively, as both obesity (Stunkard and Sørensen, 1993) and low participation in mammographic screening (Hurley et al., 1994) have been associated with low socio-economic status, obese women may have attended the population-based screening programme for breast cancer in Malmö less often. A relative underdiagnosis of tumours in obese women would lead to a negative bias in relative risk estimates for breast cancer in relation to obesity or to obesity-associated conditions such as hypertension, dyslipidaemia or impaired glucose tolerance.
Exposure misclassification caused by random measurement errors, due to variations over time in weight, blood pressure or metabolic parameters, may have biased the relative risk estimates towards null. In spite of such variations, however, other prospective studies have shown that body weight, blood pressure and blood lipids were relatively stable in middle-aged subjects over periods of several years (Rabkin et al., 1982;Gillum et al., 1982;Rissanen et al., 1988;Andersen and Haraldsdóttir, 1993), and in previous studies from the Malmö Preventive Project single measurements of these characteristics have been associated with increased risk of impaired glucose tolerance, diabetes and cardiovascular disease (Berglund et al., 1996;Larsson, 1999). Thus, although variations over time in these parameters may have attenuated their associations with breast cancer risk, the existence of associations with other disease endpoints in spite of such attenuation suggests that a potential relation of blood pressure, plasma triglycerides or plasma glucose with breast cancer risk must be relatively weak.
In conclusion, this prospective cohort study did not show any positive association of breast cancer risk with markers of the metabolic syndrome, contrary to our expectations in view of the insulin–breast cancer hypothesis. However, the markers considered appeared to be only weakly related to plasma insulin levels. A stronger test of the insulin–breast cancer hypothesis would be based on a prospective study, using direct measurements of fasting plasma insulin.
This study was carried out at the International Agency for Research on Cancer, Unit of Nutrition and Cancer, Lyon, France, where Dr Manjer was supported by a Special Training Award. Financial support was also received from the Ernhold Lundström Foundation, Malmö, Sweden.
1. Adami H-O , Rimsten Å ( 1978 ). Prevalence of hypertension and diabetes in breast cancer
: a case–control study in 179 patients and age-matched, non-hospitalized controls . Clin Oncol 4 : 243 – 249 .
2. Ambrosone CB , Freudenheim JL , Graham S et al. ( 1996 ). Cigarette smoking, N
-acetyltransferase 2 genetic polymorphisms, and breast cancer
risk . JAMA 276 : 1494 – 1501 .
3. American Joint Committee on Cancer. ( 1988 ). Manual for Staging of Cancer , 3rd edn , JB Lippincott , Philadelphia .
4. Andersen LB , Haraldsdóttir J ( 1993 ). Tracking of cardiovascular disease risk factors including maximal oxygen uptake and physical activity from late teenage to adulthood. An 8-year follow-up study . J Intern Med 234 : 309 – 315 .
5. Berglund G , Eriksson KF , Israelsson B et al. ( 1996 ). Cardiovascular risk group and mortality in an urban Swedish male population: the Malmö preventive project . J Intern Med 239 : 489 – 497 .
6. Bernstein L , Ross K ( 1993 ). Endogenous hormones and breast cancer
risk . Epidemiol Rev 5 : 48 – 65 .
7. Bruning PF , Bonfrèr JMG , van Noord PAH et al. ( 1992 ). Insulin resistance and breast-cancer risk . Int J Cancer 52 : 511 – 516 .
8. Collaborative Group on Hormonal Factors in Breast Cancer
. ( 1997 ). Breast cancer
and hormone replacement therapy: collaborative reanalysis of data from 51 epidemiological studies of 52,702 women with breast cancer
and 108,411 women without breast cancer
. Lancet 350 : 1047 – 1059 .
9. Danforth DN , Scott JR ( 1986 ). Obstetrics and Gynecology JB Lippincott , Philadelphia .
10. de Waard F , de Laive JWJ , Baanders-van Halewijn EA ( 1960 ). On the bimodal age distribution of mammary carcinoma . Br J Cancer 14 : 437 – 448 .
11. del Giudice ME , Fantus IG , Ezzat S et al. ( 1998 ). Insulin and related factors in premenopausal breast cancer
risk . Breast Cancer
Res Treat 47 : 111 – 120 .
12. Fernlund P, Fex G, Hanson A, Stenflo J, Lundh B (1991). Laurell's Clinical Chemistry
(in Swedish). Studentlitteratur, Lund, Sweden.
13. Ferrannini E ( 1993 ). Syndrome X . Horm Res 39 ( 3 ): 107 – 111 .
14. Folsom AR , Kaye SA , Princeas RJ et al. ( 1990 ). Increased incidence of carcinoma of the breast associated with abdominal adiposity in postmenopausal women . Am J Epidemiol 131 : 794 – 803 .
15. Franceschi S , La Vecchia C , Negri E , Parazzini F , Boyle P ( 1990 ). Breast cancer
risk and history of selected medical conditions linked with female hormones . Eur J Cancer 26 : 781 – 785 .
16. Gaard M , Tretli S , Urdal P ( 1994 ). Risk of breast cancer
in relation to blood lipids: a prospective study of 31,209 Norwegian women . Cancer Causes Control 5 : 501 – 509 .
17. Garne JP (1996). Invasive Breast Cancer in Malmö 1961–1992 – An epidemiological study
. Thesis. Lund University, Malmö, Sweden.
18. Gillum RF , Taylor HL , Brozek J , Anderson J , Blackburn H ( 1982 ). Blood lipids in young men followed 32 years . J Chronic Dis 35 : 635 – 641 .
19. Glicksman AS , Myers WPL , Rawson RW ( 1956 ). Diabetes mellitus and carbohydrate metabolism in patients with cancer . Med Clin N Am 40 : 887 – 900 .
20. Goodwin PJ , Boyd NF , Hanna W et al. ( 1997 ). Elevated levels of plasma triglycerides
are associated with histologically defined premenopausal breast cancer
risk . Nutr Cancer 27 : 284 – 292 .
21. Hoyer AP , Engholm G ( 1992 ). Serum lipids and breast cancer
risk: a cohort study of 5.207 Danish women . Cancer Causes Control 3 : 403 – 408 .
22. Hunter DJ , Willett WC ( 1993 ). Diet, body size, and breast cancer
. Epidemiol Rev 5 : 145 – 156 .
23. Hurley SF , Huggins RM , Jolley DJ , Reading D ( 1994 ). Recruitment activities and sociodemographic factors that predict attendance at a mammographic screening program . Am J Public Health 84 : 1655 – 1658 .
24. IARC (1992). Cancer Incidence in Five Continents
, Vol. VI. International Agency for Research on Cancer, Lyon.
25. Kaaks R ( 1996 ). Nutrition, hormones, and breast cancer
: is insulin the missing link? . Cancer Causes Control 7 : 605 – 625 .
26. Kelsey JL ( 1993 ). Breast cancer
epidemiology: summary and future directions . Epidemiol Rev 5 : 256 – 263 .
27. Land CE , Hayakawa N , Machado SG et al. ( 1994 ). A case–control interview study of breast cancer
among Japanese a-bomb survivors. I. Main effects . Cancer Causes Control 5 : 157 – 165 .
28. Larsson H (1999). Characterization of Impaired Glucose Tolerance – Insulin Sensitivity, Islet Function and Metabolic Risk Factors
. Thesis. Lund University, Malmö, Sweden.
29. Meigs JB , D'Agostino RB , Wilson PWF et al. ( 1997 ). Risk variable clustering in the insulin resistance syndrome – the Framingham offspring study . Diabetes 46 : 1594 – 1600 .
30. Moorman PG , Hulka BS , Hiatt RA et al. ( 1998 ). Association between high-density lipoprotein cholesterol
and breast cancer
varies by menopausal status . Cancer Epidemiol Biomarkers Prev 7 : 483 – 488 .
31. Moseson M , Koenig KL , Shore RE , Paternack BS ( 1997 ). The influence of medical conditions associated with hormones on the risk of breast cancer
. Int J Epidemiol 26 : 484 – 490 .
32. Muck BR , Trotnow S , Hommel G ( 1975 ). Cancer of the breast, diabetes and pathological glucose tolerance . Arch Gynak 220 : 73 – 81 .
33. Poretsky L ( 1991 ). On the paradox of insulin-induced hyperandrogenism in insulin-resistant states . Endocr Rev 12 : 3 – 13 .
34. Rabkin SW , Mathewson FA , Tate RB ( 1982 ). Relationship of blood pressure
in 20–39-year-old men to subsequent blood pressure
and incidence of hypertension over a 30-year observation period . Circulation 65 : 291 – 300 .
35. Rissanen A , Heliovaara M , Aromaa A ( 1988 ). Overweight and anthropometric changes in adulthood: a prospective study of 17.000 Finns . Int J Obes 12 : 391 – 401 .
36. Smith U ( 1994 ). Carbohydrates, fat, and insulin action . Am J Clin Nutr 59 ( Suppl 3 ): 686 – 689 .
37. Smith-Warner SA , Spiegelman D , Yaun S-S et al. ( 1998 ). Alcohol and breast cancer
in women: a pooled analysis of cohort studies . JAMA 297 : 535 – 540 .
38. Steenland K , Nowlin S , Palu S ( 1995 ). Cancer incidence in the national health and nutrition survey in follow-up data: diabetes, cholesterol
, pulse, and physical activity . Cancer Epidemiol Biomarkers Prev 4 : 807 – 811 .
39. Stoll AB ( 1996 ). Nutrition and breast cancer
risk: can an effect via insulin resistance be demonstrated? . Breast Cancer
Res Treat 38 : 239 – 246 .
40. Stunkard AJ , Sørensen TIA ( 1993 ). Obesity and socioeconomic status – a complex relation . N Engl J Med 329 : 1036 – 1037 .
41. Thompson WD , Jacobson HI , Negrini B , Janerich DT ( 1989 ). Hypertension, pregnancy, and risk of breast cancer
. J Natl Cancer Inst 81 : 1571 – 1574 .
42. Tornberg SA , Holm LE , Carstensen JM ( 1988 ). Breast cancer
risk in relation to serum cholesterol
, serum beta-lipoprotein, height, weight, and blood pressure
. Acta Oncol 27 : 31 – 37 .
43. Tulinius H , Sigfússon N , Sigvaldason H , Bjarnadóttir K , Tryggvadóttir L ( 1997 ). Risk factors for malignant diseases: a cohort study on a population of 22,946 Icelanders . Cancer Epidemiol Biomarkers Prev 6 : 863 – 873 .
44. Ursin G , Longnecker MP , Haile RW , Greenland S ( 1995 ). A meta-analysis of body mass
index and risk of premenopausal breast cancer
. Epidemiology 6 : 137 – 141 .
45. Vatten LJ , Foss OP ( 1990 ). Total serum cholesterol
and risk of breast cancer
: a prospective study of 24.329 Norwegian women . Cancer Res 50 : 2341 – 2346 .
46. WHO (1980). WHO Expert Committee on Diabetes Mellitus: Second Report
. WHO, Geneva.